Connor Structural Analisys

JER O M E J. C O NNO R, Sc.D ., M assachusetts In stitu te of Technology, is P rofessor of Civil E ngineering a t M assa chusetts In stitu te of Technology.
H e has been active in teaching and research in structural analysis and m echanics at the U.S. A rm y M aterials and M echanics R esearch Agency and for som e years at M .I.T . H is p rim ary in ter est is in com puter based analysis m ethods, and his cu rren t research is concerned w ith the dynam ic analysis of pre stressed concrete reacto r vessels an d the developm ent of finite elem ent m odels for fluid flow problem s. D r. C onnor is one of the original developers of IC E S -S T R U D L , and has published extensively in the stru ctu ral field.
ANALYSIS OF STRUCTURAL MEMBER SYSTEMS

JEROME J. CONNOR
M assach u setts Institute of T echn ology
THE RONALD PRESS C O M P A N Y NEW YORK
Copyright 1976 by
T he R onald P
ress
C ompany
A l l R ig h ts R e s e r v e d
No part of this book may be reproduced in any form without permission in writing from the publisher.
Library of Congress Catalog Card Number: 74-22535

PRINTED IN THE UNITED STATES OF AMERICA
Preface
W ith the developm ent over the past decade of com puter-based analysis m ethods, the teaching of stru ctu ral analysis subjects has been revolutionized. The traditio n al division betw een structural analysis and stru ctu ral m echanics becam e no longer necessary, and instead of teaching a p rep o n d eran ce of solu tion details it is now possible to focus on the underlying theory. W hat has been done here is to integrate analysis and m echanics in a sys tem atic presentation w hich includes the m echanics of a m em ber, the m atrix form ulation of the equations for a system of m em bers, and solution techniques. T he three fundam ental steps in form ulating a problem in solid m echanics enforcing equilibrium , relating deform ations and displacem ents, and relating forces and deform ations form the basis of the developm ent, and the central them e is to establish the equations for each step and then discuss how the com plete set of equations is solved. In this way, a read er obtains a m o re unified view of a problem , sees m ore clearly w here the various sim plifying assum ptions are introdu ced , and is b etter p rep ared to extend th e theory. T he chapters of P a rt I contain the relevant topics for an essential b ack ground in linear algebra, differential geom etry, and m atrix transform ations. Collecting this m aterial in the first p art of the book is convenient for the con tinuity of the m athem atics presentation as well as for the continuity in the following developm ent. . P a rt II treats the analysis of an ideal truss. T h e governing equations for small strain but arb itrary displacem ent are established and then cast into m atrix form . N ext, we deduce the principles of virtual displacem ents and virtual forces by m anipulating the governing equations, introduce a criterion for evaluating the stability of an equilibrium position, and in terp ret the gov erning equations as stationary requirem ents for certain variational principles. These concepts are essential for an appreciation of the solution schem es de scribed in the following two chapters. P art III is concerned with the behavior of an isolated m em ber. F o r com pleteness, first are presented the governing equations for a deform able elastic solid allow ing for arb itrary displacem ents, the continuous form of the p rin ci ples of virtual displacem ents and virtual forces, and the stability criterion. U nrestrained torsion-flexure of a prism atic m em ber is exam ined in detail and then an ap proxim ate engineering theory is developed. W e m ove on to re strained torsion-flexure of a prism atic m em ber, discussing various approaches for including w arping restrain t and illustrating its influence fo r thin-w alled jii
iv
PREFACE
open and d o sed sections. The concluding chapters treat the behavior of planar and arbitrary carved members. H ow one assembles and solves the governing equations for a m em ber sys tem is discussed in P art IV. First, the direct stiffness m ethod is outlined; then a general form ulation of the governing equations is described. G eo m etrically nonlinear behavior is considered in the last chapter, which dis cusses m em ber force-displacem ent relations, including torsional-flexural coupling, solution schemes, and linearized stability analysis. The objective has been a text suitable for the teaching of m odern structural m em ber system analysis, and w hat is offered is an outgrow th of lecture notes developed in recent years at the M assachusetts Institute of Technology. To the m any students who have provided the occasion of that developm ent, I am deeply appreciative. P articular thanks go to M rs. Jane M alinofsky for h er patience in typing the m anuscript, and to Professor Charles M iller for his encouragem ent. .
J e r o m e J. C o n n o r
Cam bridge, M ass. January, 1976 .
Contents
I MATHEMATICAL PRELIMINARIES 1 Introduction to Matrix Algebra

1-1 1-2 1-3 1-4 1-5 1-6 1-7 1-8 1-9 1-10 1-11 1-12 1-13 Definition of a Matrix Equality, Addition, and Subtraction of Matrices Matrix Multiplication Transpose of a Matrix Special Square Matrices Operations on Partitioned Matrices Definition and Properties of a Determinant Cofactor Expansion Formula Cramers Rule Adjoint and Inverse Matrices Elementary Operations on a Matrix Rank of a Matrix Solvability of Linear Algebraic Equations ^ 3 5 5 8 10 12 16 19 21 22 24 27 30
Characteristic-Value Problems and Quadratic Forms

2-1 2-2 2-3 2-4 2-5 Introduction Second-Order Characteristic-Value Problem Similarity and Orthogonal Transformations The nth-Order Symmetrical Characteristic-Value Problem Quadratic Forms 46 48 52 55 57
46
Relative Extrema for a Function

3-1 3-2 3-3 Relative Extrema for a Function of One Variable Relative Extrema for a Function of n Independent Variables Lagrange Multipliers 66 71 75
66
Differential Geometry of a Member Element

4-1 4-2 Parametric Representation of a Space Curve Arc Length v 81 82
81
vi 4-3 4-4 4-5 4-6 4-7 4-8
CONTENTS
Unit Tangent Vector Principal Normal and Binormal Vectors Curvature, Torsion, and the Frenet Equations Summary of the Geometrical Relations for a Space Curve Local Reference Frame for a Member Element Curvilinear Coordinates for a Member Element
85 86 88 91 92 94
Matrix Transformations for a Member Element

5-1 5-2 5-3 Rotation Transformation Three-Dimensional Force Transformations Three-Dimensional Displacement Transformations 100 103 109
II ANALYSIS OF AN IDEAL TRUSS Governing Equations for an Ideal Truss

6-1 6-2 6-3 6-4 6-5 6-6 6-7 6-8 6-9 General Elongation Joint Displacement Relation for a Bar General Elongation Joint Displacement Relation Force-Elongation Relation for a Bar General Bar Force Joint Displacement Relation Joint Force-Equilibrium Equations Introduction of Displacement Restraints; Governing Equations Arbitrary Restraint Direction Initial Instability
115 116 120 125 130 130 132 134 137
Variational Principles for an Ideal Truss

7-1 7-2 7-3 7-4 7-5 7-6 General Principle of Virtual Displacements Principle of Virtual Forces Strain Energy; Principle of Stationary Potential Energy Complementary Energy; Principle of Stationary Complementary Energy Stability Criteria 152 153 159 162 165 169
Displacement Method Ideal Truss

8-1 8-2 8-3 General Operation on the Partitioned Equations The Direct Stiffness Method 178 178 180
178
C O N TEN TS
vii
8-4 8-5
Incremental Formulation; Classical Stability Criterion Linearized Stability Analysis
191 200
Force Method Ideal Truss

9-1 9-2 9-3 9-4 General Governing Equations Algebraic Approach Governing Equations Variational Approach Comparison of the Force and Mesh Methods 210 211 216 217
210
III ANALYSIS OF A MEMBER ELEMENT 10 Governing Equations for a Deformable Solid

10-1 10-2 10-3 10-4 10-5 10-6 General Summation Convention; Cartesian Tensors Analysis of Deformation; Cartesian Strains Analysis of Stress Elastic Stress-Strain Relations Principle of Virtual Displacements; Principle of Stationary Potential Energy; Classical Stability Criteria 10-7 Principle of Virtual Forces; Principle of Stationary Complementary Energy 229 230 232 240 248
229
253 257
11
St. Venant Theory of Torsion-Flexure of Prismatic Members

11-1 11-2 11-3 11-4 11-5 11-6 11-7 Introduction and Notation The Pure-Torsion Problem Approximate Solution of the Torsion Problem for Thin-Walled Open Cross Sections Approximate Solution of the Torsion Problem for Thin-Walled Closed Cross Sections Torsion-Flexure with Unrestrained Warping Exact Flexural Shear Stress Distribution for a Rectangular Cross Section Engineering Theory of Flexural Shear Stress Distribution in Thin-Walled Cross Sections 271 273 281 286 293 303 306
271
12
Engineering Theory of Prismatic Members

12-1 12-2 Introduction Force-Equilibrium Equations 330 331
330
viii 12-3 12-4 12-5 12-6
CONTENTS
Force-Displacement Relations; Principle of Virtual Forces Summary of the Governing Equations Displacement Method of Solution Prismatic Member Force Method of Solution
333 339 340 349
Restrained Torsion-Flexure of a Prismatic Member

13-1 13-2 13-3 13-4 13-5 13-6 13-7 13-8 13-9 Introduction 371 Displacement Expansions; Equilibrium Equations 372 Force-Displacement Relations Displacement Model 375 Solution for Restrained Torsion Displacement Model 379 Force-Displacement Relations Mixed Formulation 383 Solution for Restrained Torsion Mixed Formulation 389 Application to Thin-Walled Open Cross-Sections '395 Application to Thin-Walled Closed Cross Sections 405 Governing Equations Geometrically Nonlinear Restrained Torsion 414
Planar Deformation of a Planar Member

14-1 14-2 14-3 14-4 Introduction; Geometrical Relations Force-Equilibrium Equations Force-Displacement Relations; Principle of Virtual Forces Force-Displacement Relations Displacement Expansion Approach; Principle of Virtual Displacements Cartesian Formulation Displacement Method of Solution Circular Member Force Method of Solution Numerical Integration Procedures 425 427 429
14-5 14-6 14-7 14-8
435 445 449 458 473
15
Engineering Theory of an Arbitrary Member

15-1 15-2 15-3 15-4 15-5 15-6 15-7 15-8 Introduction; Geometrical Relations Force-Equilibrium Equations Force-Displacement Relations-Negligible Warping Restraint; Principle of Virtual Forces Displacement Method Circular Planar Member Force Method Examples Restrained Warping Formulation Member Force-Displacement Relations Complete End Restraint Generation of Member Matrices 485 488 490 493 499 507 511 517
485
CONTENTS
ix
15-9 15-10 15-11 15-12
Member Matrices Prismatic Member Member Matrices-Thin Planar Circular Member Flexibility Matrix Circular Helix Member Force-Displacement Relations Partial End Restraint
520 524 531 535
IV ANALYSIS OF A MEMBER SYSTEM 16 Direct Stiffness Method Linear System

16-1 16-2 16-3 16-4 Introduction Member Force-Displacement Relations System Equilibrium Equations Introduction of Joint Displacement Restraints 545 546 547 548
545
17
General Formulation Linear System

Introduction Member Equations System Force-Displacement Relations System Equilibrium Equations Introduction of Joint Displacement Restraints; Governing Equations 17-6 Network Formulation 17-7 Displacement Method v 17-8 Force Method 17-9 Variational Principles 17-10 Introduction of Member Deformation Constraints 17-1 17-2 17-3 17-4 17-5 554 555 557 559 560 562 565 567 570 573
554
18
Analysis of Geometrically Nonlinear Systems

18-1 18-2 18-3 18-4 Introduction Member Equations Planar Deformation Member Equations Arbitrary Deformation Solution Techniques; Stability Analysis 585 585 591 597
585
Index
605
Part I MATHEMATICAL PRELIMINARIES
Introduction to Matrix Algebra

1-1. DEFINITION OF A MATRIX
An o rd ered set o f q u an tities m ay be a o n e-d im en sio n al array , such as
' * .1
o r a tw o-dim ensional array, such as

a ll> a t 2 > 1 a ln
^ m21* j
In a tw o-dim ensional array , the first su b scrip t defines th e row lo catio n o f an elem ent an d th e second su b scrip t its co lu m n locatio n . A tw o-dim ensional arra y having m row s a n d n colum ns is called a m atrix o f o rd e r m by n if certain arith m etic o p eratio n s (ad d itio n , su b tractio n , m u lti plication) associated w ith it are defined. T he arra y is usually enclosed in sq u are brackets a n d w ritten as* a il a 21 a,ia a i2 a 22
^m2
a ln 0-2n C lm n
II s
' '
il ri G C __________________________________ 1 1
a 21, a 2 2 > > a 2 n
(1
N o te th a t th e first term in th e o rd er p ertain s to th e n u m b e r o f row s an d th e second term to the n u m b er o f colum ns. F o r convenience, we refer to th e o rd er of a m a trix as sim ply m x n ra th e r th a n of o rd e r m b y n . .
* In print, a matrix is represented by a boldfaced letter. 3
IN TR O D UC TIO N TO M A T R IX A L G E B R A
C H AP. 1
A m atrix having only o n e row is called a row m atrix. Sim ilarly, a m atrix having only one colum n is called a co lu m n m atrix or colum n vector.* Braces instead o f b rack ets are com m only used to d en o te a colum n m atrix an d the colum n su b scrip t is elim inated. Also, th e elem ents are arran g ed h o rizo n tally instead of vertically, to save space. T he v ario u s colu m n -m atrix n o ta tio n s are: C11 C21 Cn\
V
= <s
( 1-2)
CnV V If th e n u m b er of row s an d the n u m b er o f colum ns are equal, th e m atrix is said to be square. (Special types o f sq u are m atrices are discussed in a later section.) F inally, if all the elem ents are zero, th e m atrix is called a null m atrix, an d is represented by 0 (boldface, as in th e p revious case).
Example 1-1 ---------------------------------------------------------------------------------------3 x 4 Matrix 4 3 2 1 x 3 Row Matrix [3 4 2] 3 x 1 Column Matrix Pi W kl 2 x 2 Square Matrix *8 5' _2 7_ 2 x 2 Null Matrix or ~ 3~ 4 or {3, 4, 2} 2_ 2 -7 4 -1 I -3 2 -8 1
"0 O' 0 0
* This is the mathematical definition of a vector. In mechanics, a vector is defined as a quantity having both magnitude and direction. We will denote a mechanics vector quantity, such as force or moment, by means of an italic letter topped by an arrow, e.g., F. A knowledge of vector algebra is assumed in this text. For a review, see Ref. 2 (at end of chapter, preceding Problems).
SEC. 1 -3 .
M A T R IX M U L T IP L IC A T IO N
1 -2 .
EQUALITY, ADDITION, AND SUBTRACTION OF MATRICES
T w o m atrices, a an d b, are equal if they are of th e sam e o rd er an d if c o r respondin g elem ents are equal: a = b w hen % = bu (1 -3 )
If a is of o rd er m x n, th e m atrix eq u atio n a = b co rresp o n d s to mn eq u atio n s: i = 1, 2, . . . , m aij = bfj j = 1 ,2 ,..., n
A dditio n and su b tractio n o p eratio n s are defined only for m atrices o f th e sam e order. T h e sum of tw o m x n m atrices, a an d b, is defined to be the m x n m atrix [ atj + bi}~]: (1~4) Sim ilarly, M F o r exam ple, if a = then a + b = and a - b = 1 3 2 1 1 O' -1 2 0 1 -1 0 3 -1 1 -1 0 ~ M = l atj - h i ] (1 -5 )
-2
-1
-1
It is obvious from the exam ple th a t a d d itio n is com m u tativ e and associative: a + b = b + a a + (b + c) = (a + b) 4- c (1- 6) (1 -7 )
1 -3 .
MATRIX MULTIPLICATION
T he p ro d u ct of a scalar k an d a m atrix a is defined to be the m atrix in w hich each elem ent o f a is m ultiplied by k. F o r exam ple, if k = 5 then /<a = -1 0 an d a =
-2 7
2 1
+ 35'
10
INTRODUCTION TO M ATR IX A LG E BR A
CHAP. 1
Scalar m ultiplication is com m utative. T h at is,

fca a/c [/caiy] (1 -8 )
T o establish the definition of a m atrix m ultiplied by a colum n m atrix, we consider a system of m linear algebraic equations in n unknow ns, .xt , x 2, . . . , x : a u x ! + a i2x 2 + + a lnx = c x
a 2lx l + a 2Zx 2 + ' + &2nx n ~ C2
d -9 )
+ am2x 2 + + amnx = cm This set can be w ritten as n Z aikx k = C t= i i = 1, 2, . . . , m (a)
w here k is a dum m y index. U sing colum n m atrix notation, (1 -9 ) takes the form = = {c;} N ow, we write (1 -9 ) as a m atrix p ro d u ct: [a ,/I {X j } = {c,>
J =
i = 1, 2, . . . , m
(1- 10)
1, 2 , . . . , n
(1-11)
Since (1-10) an d (1 -1 1 ) m ust be equivalent, it follows th at the definition equation for a m atrix m ultiplied by a colum n m atrix is ax = [a.j] {xj} = | E
i = 1, 2, . . . , m
( 1- 12)
This product is defined only when the colum n o rder of a is equal to the row o rd er of x. T he result is a colum n m atrix, the row o rder of which is equal to th a t of a. In general, if a is of o rder r x 5 , and x of o rder s x 1, the product ax is of order r x 1. Example 1 -2 a= 1 -1 8 -4 0 3 x =
(1)(2) + ( - l ) ( 3 ) |
ax = < (8)(2) I- ( 4)(3) - =

(0)(2) + (3)(3)
SEC. 1 -3 .
M ATR IX M ULTIPLIC ATIO N
We consider next the product of tw o matrices. This p roduct is associated with a linear transform ation of variables. Suppose th at the n original variables x u x 2, . . . , x n in (1 -9 ) are expressed as a linear com bination of s new variables .Vis >2, * >
5
*k =
Z V/JO' j=t
( 1- 13)
Substituting for x k in (1-10), I Z au f E = (ci) i = 1, 2, . . . , m (a)
Interchanging the o rd er of sum m ation, and letting Pij = L

v ; i = 1, 2 , . . . , m
k= 1
I, Z, . . ?S
(1-14)
the transform ed equations take the form ( l Pijy, j = i = 1, 2 , . . m (1 -1 5 )
N oting (1-12), we can write (1-15) as py C
(1-16)
where p is m x s an d y is s x 1. Now, we also express (1-13), which defines the transform ation of variables, in m atrix form, x = by where b is n x s. S ubstituting for x in (1 11), aby = c (1-18) (1-17)
and requiring (1-16) and (1-18) to be equivalent, results in the following definition equation for the product, ab: i 1, 2, . . . , m ab - [a,-*.] [bkj] = [ Pij]
n
k = 1 ,2 ,..., n j = 1, 2, . . . , s
(1-19)
Pij
Z U fc=l
This product is defined only when the colum n o rd er of a is equal to the row order of b. In general, if a is of order r x n, and b of o rder n x q, the p ro d u ct ab is o f order r x q. The elem ent at the ith row and yth colum n of the product is obtained by m ultiplying corresponding elements in the ith row of the first m atrix and the ;th colum n of the second matrix.
INTRODUCTION TO M ATR IX A LG E B R A
CHAP. 1
Example 1 -3
' a 1 -1 0 (1)(1) + (0)(0) I ab = O' 1 2 '1 0 1 1 - 1 0 -1" 3 (1)( 1) + (0)(3) - 1 ) ( _ 1 ) + (1X3) ( 0 ) ( - l ) + (2)(3)
b=
(1X1) + (0)(1) I
(i)(o) + (o > (- 1 ) ;
- 1)(1) + (1)(0) ( - l ) ( i ) + (l)(l) (0)(1) + (2)(0) (0)(1) + (2)(1)

ab =
- 1X0) + ( i ) ( ~ n (0)(0) + (2)( - 1 j

i
+1 -i 0
- r +4
0 - 1
+2
-2
+6
If th e p ro d u ct ab is defined, a and b are said to be conformable in the o rd er stated. O ne should n o te th a t a an d b will be conform able in either o rd er only w hen a is m x n an d b is n x m. In the previous exam ple, a and b are co n form able b u t b and a are n o t since the p ro d u ct ba is n o t defined. W hen the relevant products are defined, m ultiplication of m atrices is as sociative, a(bc) = (ab)c (1-20) and distributive, a(b + c) = ab + ac (b -f c)a ba + ca but, in general, not com m utative, ab # ba (1-2 2 ) (1- 21)
Therefore, in m ultiplying b by a, one should distinguish premultiplication, ab, from postmultiplication ba. F o r exam ple, if a and b are sq u are m atrices o f o rd er 2, the p roducts are an a2i ~bn _p21 a l2 bxi b i2 &22_ a12 a22_ O n b n + a l2b2i + ^ 22^21 ij + b L2a21 J?2la ll + b22a2i a llLb12 + < ^12^22 a 21^12 + 0-22^12 ^ U a 12 + b 12a22 &2la l 2 + b22a22_
a22_ J>21 bi2 a li
^ 22_ _a2l
W hen ab = ba, the m atrices are said to com m ute o r to be perm utable.
1 -4 .
TRANSPOSE OF A MATRIX
The transpose o f a [ a ^ ] is defined as the m atrix ob tain ed from a by interchanging row s an d colum ns. W e shall indicate the tran sp o se of a by
SEC. 1 -4
TRANSPOSE OF A MATRIX
aT = [fl0:
a ll a - [fly ] =
(w X n)
a 12 a22 m2
' '
^ln 2h (1 -2 3 ) G -m n
a 21
a Xl a r - [fl/J] =
( x m )
.2i a 22 a2n
aml ' ' &m2 & m n
0-12 a ln
T h e elem ent, afj, at the zth row and /th colum n o f a T, w here now i varies from 1 to n an d j from 1 to m, is given bv aij = < *ji (1 -2 4 )
w here a,- is the elem ent a t the ; t h row an d /th co lu m n of a. F o r exam ple, "3 a 7 5 2" 1 4 ar = "3 2 7 1 5 4_
Since the tran sp o se of a colum n m atrix is a row m atrix, an alte rn a te n o ta tio n for a row m atrix is [ a u a 2, ----- a] = { ^ } r (1 -2 5 )
W e consider next the tran sp o se m atrix asso ciated w ith the p ro d u c t o f two m atrices. Let p = ab (a) w here a is m x n an d b is n x s. T h e p ro d u ct, p, is m x s an d the elem ent, Ptj, is Pij =
k= 1
V h I aikbkJ
i = 1 ,2 ,..., m
./ A, 4 , , S
(b)
T h e tran sp o se of p will be o f o rd e r s x m an d th e typical elem ent is Pij = Pji (c)
w here now i 1, 2 , . . .,5 and j = 1, 2 , . . . ,m. U sing (1 -2 4 ) an d (b), we can w rite (c) as " " 7 =1 2 5 P l j = I aJkbk i = b fo l j t _ (d) k ~ j k= i j i, z , ... ,m I t follows from (d) th a t ' p r - (ab)r = b r a T (1 -2 6 )
E quation (1-26) states that the transpose of a product is the product of the
10
CHAP. 1
transposed m atrices in reversed order. This rule is also applicable to m ultiple products. F o r exam ple, the transpose of abc is (abc)T = c r(ab)r = c rbra r Example 1 -4
3 a = 7 5
(1 27)
2'
1 4 b=
{n I- j
r 13 6]
ab = Alternatively, ar =
13 V
(ab)r = [4
3 2
7 1
5 4 1] '3 bT = [2 7 51 -1 ]
(ab)r = brar = [2
14
= [4
13
6]
1 -5 .
SPECIAL SQUARE MATRICES
If the num bers o f rows and o f colum ns are equal, the m atrix is said to be square and o f o rder n, w here n is the num ber of rows. The elem ents an (i = 1 , 2 , . . . ,n) lie on th e principal diagonal. If all the elem ents except th e principal-diagonal elements are zero, the m atrix is called a diagonal matrix. We will use d for diagonal m atrices. If the elem ents of a diagonal m atrix are all unity, the diagonal m atrix is referred to as a unit matrix. A unit m atrix is usually indicated by I, w here n is the o rd er of the m atrix. Example 1 -5 ------------------------------------------------------------------------------------------Square M atrix, Order 2
~1
3
7'
2
Diagonal M atrix, Order 3

~2
0"
0 0
5 0
0 3_
Unit Matrix, Order 2

1 01
SEC. 1 -5 .
SPECIAL SQUARE MATRICES
11
W e in tro d u ce th e K ro n eck er d elta n o ta tio n : i * J i = j (1-2 8 )
< 5i} = + 1
W ith this n o tatio n , th e unit m atrix can be w ritten as = [<5y] Uj = 1, 2 , . . . , (1 -2 9 )
Also, the diagonal m atrix, d, takes th e form

d = [didij]
(1 -3 0 )
w here d 1, d 2, . . . , d are the principal elem ents. If the p rin cip al diagonal elem ents are all equal to k, the m atrix reduces to [kSij] = k[5ij] = kl (1 -3 1 )
and is called a scalar matrix. Let a be of o rd er m x n. O ne can easily show th a t m u ltip licatio n o f a by a conform able unit m atrix does n o t change a :
a l = a I,a = a
(1 -3 2 )
A unit m atrix is co m m u tativ e w ith any square m atrix o f th e sam e order. Sim ilarly, tw o diagonal m atrices of o rd er n are co m m u tativ e an d the p ro d u ct is a diag o n al m atrix of o rd er n. P rem u ltip licatio n of a by a conform able diagonal m atrix d m ultiplies the z'th row o f a by dt and p o stm u ltip licatio n m ultiplies th e j t h colum n by dj.
E x a m p le 1 - 6 ------------------------------------------------------------------------------------------------------------
O ' -t
3 O ' 0 5
O '
0
'
-1
6 -2
6
'2 0
3 2
O ' "3 1' 2 7 -1 1'

7
2" 7
-1
2 0
0 -1
-7
A square m atrix a for w hich a , y = aji is called symmetrical an d has the p ro p erty th a t a = a r. I f a i;- = a jiii = j) an d th e principal d iagonal elem ents all equal zero, the m atrix is said to be skew -sym m etrical. In this case, a r = a. Any sq u are m atrix can be reduced to the sum of a sym m etrical m atrix an d a skew -sym m etrical m a trix :
a b 4- c
b(j
7 (< 2(j )" fljj)
C ij
2Iîj
& ji)
(1-33)
12
CHAP. 1
The p ro d u ct of tw o sym m etrical m atrices is sym m etrical only w hen th e m atrices are com m utative.* Finally, one can easily show th a t p roducts o f the type (ara) (aar) (arba)
where a is an arb itrary m atrix and b a sym m etrical m atrix, result in sym m etrical matrices. A square m atrix having zero elem ents to the left (right) of the principal diagonal is called an upper (lower) trian g u lar m atrix. Exam ples are: Upper Triangular Matrix 3 0 0 Lower Triangular Matrix 3 5 2 0 7 1 0 0 4 5 7 0 2 1 4
T riangular m atrices are encountered in m any of the com pu tatio n al procedures developed for linear systems. Som e im p o rtan t properties of triangular m atrices are: 1. 2. The transpose o f an upper trian g u lar m atrix is a lower trian g u lar m atrix and vice versa. The product of two triangular m atrices of like stru ctu re is a triangular m atrix of the sam e structure. flu *21 1 -6 . 0
"
bn b21
i + a22^2l
a22_
b22_
! ^ 22^22
OPERATIONS ON PARTITIONED MATRICES
O perations on a m atrix of high o rder can be simplified by considering the m atrix to be divided into sm aller m atrices, called submatrices or cells. The partitioning is usually indicated by dashed lines. A m atrix can be p artitio n ed in a num ber of ways. F o r example,
fli
i
a l2 22 a 32
13 23 a 33 -
a i1
2 1 31
a 12 a 22
a 32
| j
(hi &2S
=
Cl\ 1
21
a =
a 2i
f l 31
#12 I ' ----- 1 <?22 1 | f l 32
a l3 a 23
f l 33
f l 33
O il
N ote th at the p artitio n lines are always straight an d extend across the entire m atrix. T o reduce the am o u n t o f writing, the subm atrices are represented by
SEC. 1 -6 .
OPERATIONS ON PARTITIONED M ATRICES
13
a single symbol. W e will use upper case letters to denote the subm atrices whenever possible and om it the p artition lines.
E x a m p le 1 - 7 ---------------------------------------------------------------------------------------------------------
We represent
"tfji a
as
a21
12 a 22
ai3
a 23
_a3i ~ Al2 a = A _A21 < where

A, t
al 1 a 12 Q-zi
~all a= A, v > b -i
a 23
or
A-u
^21
*21 [a3l
If two m atrices of the same o rder are identically partitioned, the rules of m atrix addition are applicable to the subm atrices. Let < 1
a -
A-12 A 2 2_
| a3z] b = < 1 1 ---------1
A 12
A-22 - [a33]
1
.
12 22_ W b o 1
_A2J
(1-34)
^21
where By and Ay are o f the same order. The sum is

a + b =
A n + Bn I A j 2 + _A 21 + b 21 + B22_
(1-35)
The rules of m atrix m ultiplication are applicable to partitioned m atrices provided th a t the partitioned m atrices are conform able for m ultiplication. In general, two p artitioned m atrices are conform able for m ultiplication if the partitioning of the rows of the second m atrix is identical to the partitioning of the columns of the first matrix. This restriction allows us to treat the various subm atrices as single elements provided th at we preserve the order o f m ul tiplication. Let a and b be two partitioned m atrices: i = 1, 2 , . . . , N
a - [Ay]
b = [B;] W e can write the p ro d u ct as
j = 1, 2, . . . , M j = 1,2,..., M k = 1, 2 , . . . , S
(1-36)
c = ab - [C a] M Qk X! 7=1 i 1,2 k ^ 1 ,2 ,..., S (1-37)
when the row partitions of b are consistent with the colum n partitio n s of a.
14
IN TR O D U C TIO N TO M A T R IX A L G E B R A
CHAP. 1
As an illu stratio n , we co n sid er th e p ro d u ct

ch i a 1 2 a 1 3 V
a b
a 2 y
2 2
2 3
<
b 2
a 3 1
a 3 2
3 3
b 3
S uppose we p a rtitio n a w ith a vertical p a rtitio n betw een th e second an d th ird colum ns.
f l u a 1 2 a
13
a 2 i
a 2 2
2 3
t i A
1 2 ]
3 1
Clj 2
F o r the rules o f m atrix m u ltip licatio n to be ap p licab le to th e su b m atrices o f we m ust p a rtitio n w ith a h o rizo n tal p a rtitio n betw een th e seco n d an d th ird rows. T ak in g
a , b b < b 2
b3
the p ro d u c t has th e form

a b = [ A i i A 1 2 ]
A u B u + A i 2B 2i
T h e con fo rm ab ility o f tw o p a rtitio n e d m atrices does n o t d ep en d o n the h o rizo n tal p a rtitio n in g o f th e first m a trix o r th e vertical p artitio n in g of the second m atrix. T o show this, we co n sid er th e p ro d u c t
* 11
a 2l
a l2
a 22
a 3 2
1 3
^11 ^21
b 3!
^12 b 22
b32
a 2 3
a 3l
3 3
Suppose we p a rtitio n th ird row s:
w ith a h o rizo n tal p a rtitio n betw een th e second and

a \ 1 1 2 a \3 " A n "
2 l
2 2
2 3 _ a 2 1 _
a 3 1
3 2
a 3 3
Since the co lu m n o rd e r o f A n an d A 2J is eq u al to the row o rd e r o f p artitio n in g o f is required. T he p ro d u c t is

b
b ,
n o
1H 1 C * 1 < 31
^ 3 2
In this case, since the row order of B ai and B 12 is the same as the colum n
SEC. 1 -6 .
O P E R A TIO N S ON P A R TITIO N ED M A T R IC E S
15
ord er of a, n o p artitio n in g o f a is necessary an d th e p ro d u c t h as the form ab = a [ B u B 12] = [a B n j a B 12]
T o tran sp o se a p artitio n e d m atrix, one first interchanges the off-diagonal subm atrices an d th en tran sp o ses each subm atrix. If An > 5 a = A 21 A J2 A 22 A,2
a
' ''
A j
a
Am n (1 38)
A ml t
a
T
then AT,
_
a [2 A[
aL
r2
A p articu la r type o f m atrix en co u n tered frequently is th e quasi-diagonal m atrix. T his is a p a rtitio n ed m atrix w hose d iag o n al su b m atrices arc square o f various o rd ers, an d w hose off-diagonal su b m atrices a re null m atrices. An exam ple is flj! 0 0 a = 0
0
U22
&23
a 33
which can be w ritten in p artitio n ed form as a = w here

A i \_Cly i ]
"A. 0
A, =
a 22 a 32
a 23 a 33
and 0 den o tes a null m atrix. T h e p ro d u ct o f tw o q u asi-d iag o n al m atrices o f like stru c tu re (co rresp o n d in g d iagonal su b m atrices are of th e sam e ord er) is a q u asi-d iag o n al m atrix o f th e sam e structure. A, 0 0 0 A2 0 0 ' 0 A Bx 0 0 0 B2 0 0 0 AiB ! 0 0 0 0 (1 -3 9 )
A 2B 2 0 0 AB
K
Ai 1 A 21 0
w here A,- an d B, a re o f th e sam e order. W e use th e te rm quasi to d istinguish betw een p artitio n e d an d u n p a rtitio n e d m atrices having the same form. F o r exam ple, we call
A 22
(1-40)
a low er q u asi-trian g u la r m atrix.
16
INTRODUCTION TO M ATRIX ALG E BR A
CHAP. 1
1 -7 .
DEFINITION AND PROPERTIES OF A DETERMINANT
T he concept of a determ inant was originally developed in connection with the solution of square systems of linear algebraic equations. T o illustrate how this concept evolved, we consider the simple case of tw o equations:
+ a 21x l a l2x 2 a 22X 2 ~ C1 C2 , s .
(a )
Solving (a) for x t and x 2, we obtain

(^ 11^2 2 /
\ a l l a 22 ~
a 12a 2 \ ) X l ~ C l a 22 ~~ C 2 a 12
a 12a 2 l ) X 2 ~
~ ~ Cl a 2 1 +
C2 a l l
/, ^ (b )
The scalar quantity, ^ 11^22 a21a2U\s defined as the determ inant of the secondorder square array ai} ( i,j 1, 2). T he determ inant of an array (or m atrix) is usually indicated by enclosing the array (or m atrix) with vertical lines:
aw a l2
< 22i
22
We use the term s array and matrix interchangeably, since they are synony mous. Also, we refer to the determ inant o f an nth-order array as an Hth-order determ inant. It should be noted th at determ inants are associated only with square arrays, th a t is, with square matrices. The determ inant of a th ird-order array is defined as au a 21 a3l ai2 a 22 a12 i3 23 a33 + a n a22a33 a l l a23a32 - a 12a2la33 + a l2a23a 3 1
+ f l 1 3 f l 2 1 a 32 ~~ a 1 3 a 2 2 ^ 3 1
(1-42)
This num ber is the coefficient of Xj, x 2, and x 3, obtained when the third-order system ax = c is solved successively for Xj, x2< and x 3. C om paring (1-41) and (1 -42), we see th a t bo th expansions involve products which have the following p ro p erties: 1. Each product contains only one elem ent from any row or colum n and no elem ent occurs twice in the same product. T he products differ only in the colum n subscripts. The sign of a product depends on the order of the colum n subscripts, e.g., + Q n ^ 2 2 fl33 fln fl23fl32>
2.
These properties are associated with the arrangem ent of the colum n subscripts and can be conveniently described using the concept of a perm utation, which is discussed below. A set of distinct integers is considered to be in natural order if each integer is followed only by larger integers. A rearrangem ent of the n atu ral order is called a permutation o f the set. F o r example, (1, 3, 5) is in n atural order and
SEC. 1 -7 .
DEFINITION AND PROPERTIES OF A DETERMINANT
17
(1, 5, 3) is a perm utation of (1, 3, 5). If an integer is followed by a smaller integer, the pair is said to form an inversion. The num ber of inversions for a set is defined as the sum of the inversions for each integer. As an illustration, we consider the set (3, 1, 4, 2). W orking from left to right, the integer inversions are: Integer 3 1 4 2 Inversions (3, I)(3, 2) N one (4, 2) N one Total 2 0 1 0 3 This set has three inversions. A perm utation is classified as even (odd) if the total num ber of inversions for the set is an even (odd) integer. According to this convention, (1, 2, 3) and (3, 1, 2) arc even perm utations and (1, 3, 2) is an odd perm utation. Instead of counting the inversions, we can determ ine the num ber of integer interchanges required to rearrange the set in its n atural order since an even (odd) num ber of interchanges corresponds to an even (odd) num ber of inversions. F o r example, (3, 2, 1) has three inversions and requires one interchange. W orking with interchanges rather than inversions is practical only when the set is small. Referring back to (1 -41) and (1-42), we see th a t each product is a perm utation of the set of colum n subscripts and the sign is negative when the perm utation is odd. The num ber of products is equal to the num ber of possible perm utations of the column subscripts that can be formed. O ne can easily show th at there are n-factorial* possible perm utations for a set of n distinct integers. We let (<*!, atz, . . . , a) be a perm utation of the set ( 1 , 2 , . . . , n) and define ' 1an ^ +1 ^ Ill2 = 1 when (a i} a 2, . . . , ) is an even perm utation (1-43) when (aj., a2, , ) is an odd perm utation Using (1-43), the definition equation for an nth-order determ inant can be w ritten as
u (Xj 2
#2 2
*** ***
1H II
a2l
Clnl
a 2n
r s V J
*afl^ \ a i ^ 2 a 2
(1 -44)
tf 2
' '
where the sum m ation is taken over all possible perm utations of (1, 2, . . . , n).
* Factorial n n! = n(n l)(n 2) (2)(1).
18
IN TR O D U C T IO N TO M A T R IX A L G E B R A
CHAP. 1
Example 1 -8
The permutations for n = 3 are ai = 1 <*i = 1 i = 2 i = 2 ai = 3 ai = 3 a2 = 2 a2 = 3 2 = 1 < X 2 = 3 2 = 1 2 = 2 3 = 3 & 3 2
3 = 3 3 = 1 e 132 e 213 e 231
+ 1
3 = 2 a3 = I
-1 H e312 ~ + i e321 ~ - i
I I
= -1 = -1
Using (1-44), we obtain *ii 12 *13 < ^21 a22 #23

a 31
a 32 a 33
a l l fl22a 33 ~ a l l a 23a 32
~ c>12a 21^33 + 12^23fl31
+ a 13a21a32 'a 13a22a31
This result coincides with (1-42).
T h e follow ing p ro p e rtie s o f d eterm in a n ts can be estab lish ed * from (1 -4 4 ): 1. 2. 3. 4. 5. 6. If all elem ents o f any row (o r colum n) are zero, th e d e te rm in a n t is zero. T h e value o f th e d e te rm in a n t is u n ch an g e d if th e row s a n d co lu m n s are in te rc h a n g e d ; th a t is, jar | = |a|. If tw o successive row s (or tw o successive colum ns) a re in terch an g ed , th e sign o f the d e te rm in a n t is changed. If all elem en ts o f one ro w (or one colum n) are m u ltip lied by a n u m b e r k, th e d e te rm in a n t is m u ltip lied by k. If c o rresp o n d in g elem ents o f tw o row s (or tw o co lu m n s) are eq u al o r in a c o n s ta n t ratio , th en th e d e te rm in a n t is zero. If each elem ent in one ro w (o r one colum n) is expressed as th e su m o f tw o term s, th e n th e d e te rm in a n t is eq u al to th e sum o f tw o d eterm in a n ts, in each o f w hich one o f th e tw o term s is d eleted in each elem ent o f th a t row (o r colum n). . If to th e elem ents of any ro w (colum n) are ad d ed k tim es th e c o r resp o n d in g elem ents o f an y o th e r row (colum n), th e d e te rm in a n t is un ch an g ed .
7.
W e d e m o n stra te these p ro p erties for th e case o f a se c o n d -o rd er m atrix . Let

_
T h e d e te rm in a n t is
I!
_aZi
an .. 1
a22_
|a| C l1^22 ~ al 2a21 P ro p e rtie s 1 an d 2 are obvious. It follow s from p ro p e rty 2 th a t |a r j = |a|. W e
* See Probs. 1-17, 1-18, 1-19.
SEC. 1 -8 .
C O FA C TO R E X PA N S IO N F O R M U L A
19
illu strate th e th ird b y in terch an g in g th e row s of a : a = a 21 _a l l

|a'| = <321^12 H
a 22 a \2_
22
ja[
P ro p e rty 4 is also ob v io u s from (b).T o d e m o n strate the fifth, we tak e d 2\ = T h en aj = a u {ka12) - a 12{kaiX) = 0 N ext, let a n = bn + cn A ccording to p ro p e rty 6, la l = |b| + |c| w here
_ _
^22 k&l2
a l2 ~ bl2 + C12
b \i a2\ a22
ui 'Li1 cn 1 21
_
C12 Q22
T his resu lt can be o b ta in e d by su b stitu tin g for a u a n d a l2 in (b). F inally, to illu strate p ro p e rty 7, we tak e = a i i + k a 21 ^12 = a i2 + ka22 b2\ ci2i b2o = fl-21 Then, |b| = ( a n + k a 2 i)a22 - ( a 12 + k a 22)a2l = ja| 1 -8 . COFACTOR EXPANSION FORMULA
If th e ro w a n d co lu m n c o n tain in g a n elem ent, ai}, in th e sq u are m atrix , a, are deleted, the d e te rm in a n t o f the rem ain in g sq u are a rra y is called th e minor o f dij, an d is d en o ted by M [;-. T h e cofactor o f aiJt d en o ted by A i}, is related to the m in o r o f M i} by A,j = ( - 1Y+JM ,j (1 -4 5 ) As an illu stratio n , w e ta k e 3 2 8' a ~ 1 5 7 3 4 1
T h e values o f M i} an d A V ] asso ciated w ith a 23 an d a 22 are M 23 3 5 3

M 22 =
2 3I 8 1
= -1
A 23
( 1 ) 'M 23 + 1
-3 7
A 2 2 = ( ~ l ) 4M 22 = - 3 7
20
IN TR O D U C TIO N TO M A T R IX A L G E B R A
C H AP. 1
C ofactors occu r n atu ra lly w hen (1 -4 4 ) is expanded* in term s of the elem ents o f a row o r colum n. This leads to the follow ing ex p an sio n form ula, called L ap laces ex p an sio n by cofactors o r sim ply L aplaces ex pansion: t
n n
l3i =
X aikîk =
fc=l
Yj
A=1
(1 -4 6 )
E q u atio n (1 -4 6 ) states th a t the d eterm in a n t is equal to the sum of the p ro d u cts o f the elem ents o f any single row o r co lu m n by th eir cofactors. Since the d eterm in an t is zero if tw o row s o r colum ns are identical, if follows th a t
n
I arkA ik - 0 t =i
r /
k = 1
aksAj
= 0
s ^
T he ab o v e identities a re used to establish C ram ers rule in the follow ing section. Example 1 -9 -------------------------------------------------------------------------------------------(l)
a ll a 21 a 31
We apply ( I 46) to a third-order array and expand with respect to the first row:
&1 2 a 22 a 32
fl13
a 23 a 33
= u (-D 2
a 22 Cl32
a 23
fljl + a l2( I)3 3J
23 + a l 3 ( ~ 1) .U
21
31
22 <*32
a 33
<2]t(<222fl33
a 23&32) +
a n{ ~
a 2 l a 33 +
a 23a 3 i ) + a l3(<22la 32 ~~ a 2 2 a 3 l )
To illustrate (1-47), we take the cofactors for the first row and the elements of the second row:
Y j a 2k^lk
k- 1
a 2 l ( a 2 2 a 3 3 ~ a 2 3 a 3 2 ) + a 2 2(
a 2 1a 3 3 + a 2 3 3 l ) + a 2 3 { a 2 i a 3 2 a 2 2 & 3 l ) ~
(2) Suppose the array is triangular in form, for example, lower triangular. Expanding with respect to the first row, we have
a lt a 2l
0
a 22
0 0 3 3 =
a ls CI22 a 32
(rtl l ) i a 22a 33) a l l a l l a 33

a 33
31
Generalizing this result, we find that the determinant of a triangular matrix is equal to the product of the diagonal elements. This result is quite useful.
* See Probs. 1-20, 1-21. t See Ref. 4, sect. 3-15, for a discussion of the general Laplace expansion method. The expansion in terms of cofactors for a row or a column is a special case of the general method.
SEC. 1 -9 .
CRAMER'S RULE
21
T he ev alu atio n o f a d eterm in an t, using th e definition eq u atio n (1 -4 4 ) o r th e cofactor expansion form ula (1 -4 6 ) is qu ite tedious, p articu la rly w hen th e array is large. A n u m b er o f a ltern ate an d m o re efficient num erical p rocedures for evaluatin g d eterm in a n ts have been developed. T hese p ro ced u res are described in References 9 -1 3 . Suppose a sq u are m atrix, say c, is expressed as th e p ro d u ct o f tw o sq u are m atrices, c = ab (a) an d we w an t |c | . It can be show n* th a t the d eterm in a n t o f the p ro d u c t of tw o sq u are m atrices is eq u al to the p ro d u c t o f th e d eterm in a n ts: |c| = |aj |b| (1 -4 8 )
W h eth er we use (1 -4 8 ) o r first m ultiply a an d b a n d then determ in e |ab| depends on the form an d o rd er of a an d b. If they are d iag o n al o r trian g u lar, (1 -4 8 ) is q u ite efficient.! Example 1 -1 0 (1)
a = |a | = - 4 Alternatively, c 5 11 15 29 1 U JI and |c| = 20 1 I I _! and "1 _3 3" 5_
b=
"2
3'
J
|c| -
4_
-2 0
|b| = 5
a = |a| = 5 Determining c fust, we obtain c = 5 5
|b| = 8
|c| = + 4 0
12' 20
|cl = + 4 0
1 -9 .
CRAMERS RULE
W e consider next a set of n e q u atio n s in n u n k n o w n s:

n
k~ 1
ajkx k = Cj
j = 1, 2, . . . , n
Ol 4_
(2)
* See Ref. 4, section 3-16. t See Prob. 1-25 for an important theoretical application of Eq. 1-48.
o 1
(a)
22
INTR O D UC TIO N TO M A T R IX A LG E B R A
C H AP. 1
M ultiplying b o th sides of (a) by A jr, w here r is an a rb itrary integer from 1 to n, and sum m ing w ith respect to j, we o b tain (after in terchanging the o rd er of sum m ation)
n / n \ / ii
Xi ( Xj QjkAjr I
fc=l\j=l
X/ A jrCj
}=1
(b)
N ow , the inner sum vanishes w hen r # k an d equals |a| w hen r = k. T his follows from (1-47). Then, (b) reduces to
n
|a|*r = Z AJrC j
J= 1
(c)
The expansion on th e right side of (c) differs from the expansion n |a| = X ajrAjr
(d)
j=i
only in th a t th e rth colum n of a is replaced by c E q u atio n (c) leads to C ram ers rule, w hich can be stated as follow s:
A set of n linear algebraic eq u atio n s in n unknow ns, ax c, has a unique solution when |a| + 0. T he expression for xr (r = 1, 2 , . . . , n) is the ratio of tw o d eterm in an ts; the d en o m in ato r is |a| and th e n u m e ra to r is the d eterm in an t of th e m atrix o b tain ed from a by replacing th e rth colum n by c.
If |a| = 0, a is said to be singular. W hether a so lu tio n exists in this case will depend on c. All we can conclude from C ram ers rule is th a t the solution, if it exists, will n o t be unique. S ingular m atrices an d th e q u estio n of solvability are discussed in Sec. 1-13. 1-10. ADJOINT AND INVERSE MATRICES
W e have show n in th e previous section th a t th e so lu tio n to a system of n equ atio n s in n unknow ns, M can be expressed as 1 " xi = n L A a cj (b) {xj} = {c< } U = 1, 2, . . . , n (a)
(note th a t we have tak en r / in Eq. c o f Sec. 1 -9 ). U sing m atrix n o ta tio n , (b) takes the form
la l E q u atio n (c) leads n atu rally to the definition o f ad jo in t an d inverse m atrices.
SEC. 1 -1 0 .
A D JO IN T AN D INVERSE M ATR IC ES
23
We define the adjoint an d inverse matrices for the square m atrix a of o rd er n as

ad join t a Adj a -
[/!, ,] 7
(1 -4 9 ) d -5 0 )
inverse a a - x = Adj a
N o te th a t th e inverse m atrix is defined only for a n o n sin g u lar square m atrix. Example 1-11
We determine the adjoint and inverse matrices for
'1
a = The matrix of cofactors is 5 M Then -1
-7
2 3"
3 1 I 2 0 -1 0 +5 ' 5 -1 0 " +7 -1 _ -1 -7 +5 -1 +7/25' -1 /5 + 1/25
2 4
Adj a = [My] r _
0 -10
-1 /5
-10
+7 + 1 /2 5 + 2 /5 - 7 /2 5
j 7 Adj a =
0 + 2/5
la l
U sing the inverse-m atrix notatio n , we can w rite the solution of (a) as x = a *c Substituting for x in (a) an d c in (d), we see th a t a " 1 h as the p ro p erty th a t a a = a a " 1 = I (1-51) (d)
E q u atio n (1 -5 1 ) is frequently tak en as the definition of the inverse m atrix instead o f (1-50). A pplying (1 -4 8 ) to (1-51), we o b tain
! a_ 1 | | a | = I 1
It follows th a t (1 -5 1 ) is valid only w hen |a| ^ 0. M u ltip licatio n by th e inverse m atrix is analogous to division in o rd in ary algebra. If a is symmetrical,, then a 1 is also sym m etrical. T o show this, we take the transpose of (1-51), and use the fact th a t a - a T: (a - 1a)r = a a 1 1 r I
24
IN TR O D UC TIO N TO M A T R IX A L G E B R A
CH AP. 1
P rem u ltip licatio n by a " 1 results in a I T = a - 1 1 and therefore a -1 is also sym m etrical. O n e can also show * th at, for any no n sin g u la r sq u are m atrix, the inverse and tran sp o se o p eratio n s can be in ter chan g ed : b T - 1 = b " 1- r (1 -5 2 ) W e consider next th e inverse m atrix associated w ith th e p ro d u c t o f tw o square m atrices. Let c = ab w here a an d b are b o th of o rd e r n x n and nonsingular. P rem u ltip licatio n by a - 1 a n d then b 1 results in a _1c = b (b~ Ja _ J)c = I It follows from th e definition o f th e inverse m atrix th a t (ab)-1 = b - l a -1 (1 -5 3 )
In general, the inverse of a m u ltiple m atrix p ro d u c t is eq u al to th e p ro d u c t of the inverse m atrices in reverse o rder. F o r exam ple, (abed)' 1 = d - 1 c - 1 b _1a -1 T he d eterm in a tio n of th e inverse m atrix using th e definition eq u atio n (1 -5 0 ) is to o lab o rio u s w hen the o rd er is large. A n u m b er o f inversion p ro ced u res based on (1 -5 1 ) have been developed. T hese m eth o d s are described in Ref. 9 -1 3 . 1 -1 1 . ELEMENTARY OPERATIONS ON A MATRIX
T h e elem entary o p eratio n s on a m atrix are: 1. 2. 3. T h e in terch an g e o f tw o row s o r of tw o colum ns. T h e m u ltip licatio n o f th e elem ents o f a row o r a co lu m n by a n u m b e r o th e r th a n zero. T h e ad d itio n , to the elem ents o f a row o r colum n, o f k tim es th e c o r resp o n d in g elem ent o f a n o th e r row o r colum n.
These o p eratio n s can be effected by p rem u ltip ly in g (for row o p eratio n ) o r postm u ltip ly in g (for colum n o p eratio n ) th e m atrix by a n a p p ro p ria te m atrix, called an elem entary o p eratio n m atrix. W e consider a m atrix a of o rd er m x n. S uppose th a t we w an t to in terch an g e ro w sj an d k. T hen, we premultiply a by an m x m m atrix o b tain ed by m odifying the m th -o rd er unit m atrix, I m, in th e follow ing w ay: 1. 2. In terch an g e <5^ an d V In terch an g e 8kk an d okj.
* See P rob. 1 -2 8 .
SEC. 1 -1 1 .
E L E M E N T A R Y O P E R A T IO N S ON A M A T R IX
25
F o r exam ple, if a is 3 x 4, p re m u ltip lic a tio n by 0 1 0 1 0 0
interchan g es row s 1 a n d 3 a n d p o stm u ltip lic a tio n by 1 0 0 0
0 0 0 0
0 10 0
0 1 1 0
interchan g es co lu m n s 2 a n d 4. T h is sim ple exam ple sh o w s th a t to in terch an g e row s, we first in terch an g e the row s o f th e co n fo rm a b le u n it m a trix a n d pre multiply. Sim ilarly, to in terch an g e colum ns, we in terch an g e co lu m n s o f th e co n fo rm ab le u n it m a trix a n d p o stm u ltip ly . T h e elem en tary o p e ra tio n m atrices for o p e ra tio n s (2) an d (3) are also o b ta in e d by o p e ra tin g o n th e corresponding co n fo rm a b le u n it m atrix . T h e m a trix w hich m ultiplies ro w /' by a is a n m th o rd e r d ia g o n al m a trix h av in g dt = 1 for i ^ j an d dj = a. S im ilarly, p o stm u ltip lic a tio n by an n th o rd e r d ia g o n a l m a trix having dt - 1 for i = j a n d d} a will m u ltip ly th e /th co lu m n by a. S u p p o se th a t we w ant to ad d a tim es row j to row k. T h en , we in sert a in th e k t h row and yth co lu m n o f Im a n d prem u ltip ly . T o ad d o c tim es co lu m n j to co lu m n k, wc p u t a in th e ; t h row a n d /<th co lu m n o f I a n d p o stm u ltip ly . W e let e d e n o te an elem en tary o p e ra tio n m atrix . T h en , ea rep resen ts th e result o f ap p ly in g a set o f elem en tary o p e ra tio n s to th e row s o f a. S im ilarly, ae rep resen ts th e result o f ap p ly in g a set o f elem en tary o p e ra tio n s to th e co lu m n s of a. In general, we o b ta in e by ap p ly in g th e same o p e ra tio n s to th e co n fo rm a b le unit m atrix . Since we s ta rt w ith a u nit m a trix an d since th e elem en tary o p e ra tions, a t m ost, ch an g e th e v alu e o f th e d e te rm in a n t by a n o n zero scalar factor,* it follow s th a t e will alw ays be n o n sin g u lar. Example 1 -1 2 --------------------------------------------------------------------------------------------Wc illustrate these operations on a third matrix:
a =
1 1/2 3 7
-2 1
1/5
2
5
We first: 1. 2. Add ( 3) times the first row to the second row. Add (2) times the first row to the third row.
* See prop erties o f d eterm in a n ts (p age 18).
26
INTRODUCTION TO MATRIX ALGEBRA
CHAP. 1
These operations are carried out by premultiplying by
1 0 0
-3 1 0 2 0 1 and the result is "1 0 _0 1/2 1/5 11/2 7/5 2 27/5
Continuing, we multiply the second row by 2/11: '1 0 0 0 2/1.1 0 O ' 0 1 1 1/2 V5 ' = 0 11/2 7/5 * > 0 27/5 1 1/2 0 1 0 2 1/5 ' 14/55 27/5_
Next, we add ( 2) times the second row to the third row: '1 0 _0 0 1 -2 " 1 1/2 0' '1 1/2 1/5 0 0 1 14/55 = 0 1 0 2 27/5 _ 0 0 1/5 14/55 269/5 5_
Finally, we multiply the third row by 55/269. The complete set of operations is
p o o 10 1 0
[o 0 55/269
0 -2
'1 0
1 2/11
0'
0
0 0
1 1/2 0 1
1_
0 O ' 3 1 0 2 0 i
1 3 -2
1/5 14/55 = b
0 0
This example illustrates the reduction of a square matrix to a triangular matrix using elementary operations on rows, and is the basis for the Gauss elimination solution scheme (Refs. 9, 11,13). We write the result as ea b where e is the product of the four operation matrices listed above:
'1
e = -6 /1 1 + 1870/2959
0
2/11 -220/2959
0
0 55/269
We obtain e by applying successive operations, starting with a unit matrix. This is more convenient than listing and then multiplying the operation matrices for the various steps. The form of e after each step is listed below: 1 Initial Step 1
'
Step 2
1 0 0 -3 1 0 1 2 0
0 1 0 0 0 1
1 0 -6/11 2/U
2 0
o o
SEC. 1-12.
RANK OF A MATRIX
27
Step 3
Step 4
'
-
1
6/11
0
2/11
0
0
-
1
6/11 + 1870/2959
0
2/11 -220/2959
0
0 55/269 _
. + 34/11
-4 /1 1
0.
Two matrices are said to be equivalent if one can be derived from the other by any finite number of elementary operations. Referring to Example 1-12, the matrices 1 1/2 1/5' l 1/2 1/5 3 7 2 and 0 1 14/55 -2 1 5 0 0 1 are equivalent. In general, a and b are equivalent if b = paq (1-54)
where p and q are nonsingular. This follows from the fact that the elementary operation matrices are nonsingular. 1-12. RANK OF A MATRIX
The rank, r, of a matrix is defined as the order of the largest square array, formed by deleting certain rows and columns, which has a nonvanishing deter minant. The concept of rank is quite im portant since, as we shall see in the next section, the solvability of a set of linear algebraic equations is dependent on the rank of certain matrices associated with the set. Let a be of order m x n . Suppose the rank of a is r . Then a has r rows which are linearly independent, that is, which contain a nonvanishing deter minant of order r, and the remaining m r rows are linear com binations of these r rows. Also, it has n r columns which are linear com binations of r linearly independent columns. To establish this result, we suppose the determ inant associated with the first r rows and columns does not vanish. If a is of rank r, one can always rearrange the rows and columns such that this condition is satisfied. We consider the (r + l)th-order determ inant associated with the first r rows and columns, row p , and column q where r < p ^ m , r < q ^ n .
a u a 2l G 12 '
a ir
a 2r
aH
<*2,
22
'
A r +i -
(1-55)
flr2
a P2
We multiply the elements in row j by Xpj (j = 1 , 2 , . . . ,r) and subtract the result from the last row. This operation will not change the magnitude of Ar+i (see Sec. 1-7). In particular, we determine the constants such that the first r elements
C"
U P1
rr f,pr
C !rq
UP < I
28
CHAP. 1
in the last row vanish:

1 a N i fia ^ pi < AP2 ^ = f't* , i C,P2
.
^ 12
^ 22
* *
^r 2
(1-56)
a ir
a 2r
' ' '
a rr
Equation (1-56) has a unique solution since the coefficient matrix is non singular. Then (1 -55) reduces to
a, 11
21
a i2 22
'
'
a ua 2>
I
1 i 1 a iU pq
(1 -5 7 )
rl
a r2 0
'
a n 0
where
pi
p2
(1 -5 8 )
'P1 '
Applying Laplaces expansion formula to (1-57), we sec that A r + l vanishes when a{ ^ = 0. Now, if a is of rank r, A r + l vanishes for all combinations of p and q. It follows that / AP 1
.
^2<y> * * * rq}
pi
(1 -5 9 )
Combining (1-56) and (1-59), we have

a 2l a ll a 22 ' ' rl' a rl < / '-p\ ^pi > < a p\ a i p r +
1, r + 2, . . . , m
(1 -6 0 )
aia
2n
a rn
. V
Equation (1-60) states that the last m r rows of a are linear combinations of the first r rows. One can also show* that the last n r columns of a are linear combinations of the first r columns. Example 1-13
Consider the 3 x 4 matrix
12
2 1 3
5 7
* See Prob. 1-39.
12 14
SEC. 1-12.
RANK OF A MATRIX
29
We see that a is at least of rank 2 since the determinant associated with the first two rows and columns is finite. Then, the first two rows are linearly independent. We consider the determinant of the third-order array consisting of columns 1, 2, and q : 1 2
Ai Iq
a 2q
2 1
5 7
u iq
Solving the system, Aj -f- 22.2 5

22.i + Aj = 7
we obtain
Xl = 3
A2 = 1
If a is of rank 2,
must vanish. This requires
Aîa + ^2a2q ~
q = 3,4
+ a2
Since a 33 and a34 satisfy this requirement, we conclude that a is of rank 2. The rows are related by (third row) = + 3 (first row) + (second row)
One can show* that the elementary operations do not change the rank of a matrix. This fact can be used to determine the rank o f a matrix. Suppose b defined by (1 -61) is obtained by applying elementary operations to a. We know that b and a have the same rank. It follows that a is of rank p . A matrix having the form of b is called an echelon matrix. When a is largfc, it is more efficient to reduce it to an echelon matrix rather than try to find the largest nonvanishing determ inant: .
(p x p) #11 21
hn 1 a \2
a 22
(px(n-p))
b 2P
In
@2n
12
0 0
L 0
( (m -p ) x p)
(1-61)
m2
t 0
" T oJ
((m - p ) x (
- p ))
Example 1-14
2 a = 3 4' 2 12 1 3 7 12
First, we eliminate a 2li and a 3u using the first row: 1 2 0 0

* See Prob. 1-40.
3 3 3
4 -6 -8
-3 -3
30
CHAP. 1
Next, we eliminate
using the second row: 1

a<2> =
2 -3 0
3 -3 0
4 6 -2
0 0
At this point, we see that r 3. To obtain b, we multiply the second row by 1/3, the third row by 1/2, and interchange the third and fourth columns:
4
b = 2
1 Suppose a is expressed as the p ro d u ct o f tw o rectangular m atrices:

(ftt x s) (m x n) (?i x ,s)
(1-62)
O ne can show* th a t the ran k o f a can n o t be greater than the m inim um value o f r associated w ith b and c: r(a) < m in j>(b), r(c)] As an illustration, consider the p ro d u ct a =
~
(1-63)
1/2
1/2
0"
1 0 1 0 0 1
- 1/2
+ 1/2
Since each m atrix is of ran k 2, the ran k o f a will be < 2. E valuating the product, we ob tain "0 0
It follows th a t a is of ran k 1. 1-13. SOLVABILITY OF LINEAR ALGEBRAIC EQUATIONS
We consider first a system o f two equations in three unknow ns:

Clj i a 21 d\ 2
#22
Cl13
#23
*1 x2 *3
(1-64)
ax = c Suppose a is of ran k 2 and #11

#21
#12
#22
(1-6 5 )
* See Prob. 1-44.
SEC. 1 -1 3 .
SO LV ABILITY OF LINEAR ALG EBR AIC EQUATIONS
31
If a is o f rank 2, we can always renum ber the rows and colum ns such th at (1-65) is satisfied. We partition a and x, a =
an a \i a 22
<*13
[A t
A 2]
( 1-66 )
#21
23
x =
-x 3
= [X ,l
'
lx2]
and write (1-64) as A i X x X2 to the right-hand side:
A2X2 = c. Next, we transfer the term involving A tX t = c - A 2X 2 (1-67)
Since lA^ # 0, it follows from C ra m e rs rule th a t (1-67) has a unique solution for X ^ Finally, we can w rite the solution as X , - A f J(c - A 2X 2)
( 1-68)
Since X 2 is arbitrary, the system does not have a unique solution for a given c. The o rder of X 2 is generally called the defect o f the system. The defect for this system is 1. If a is of ran k 1, the second row is a scalar multiple, say X, of the first row. M ultiplying the second equation in (1-64) by 1/A, we have fluX ! + a12x 2 + a l3x 3 = Ci a u x i + a l2x 2 + a, 3x 3 = c2/A If c2 = Acj, the equations are inconsistent and no solution a is of ran k 1, (1-64) has a solution only if the rows of c are m anner as the rows of a. If this condition is satisfied, the tw o are identical and one can be disregarded. Assuming th a t a u x t = (1/flnK c! - a12x 2 - a l 3x 3) (1-69) exists. Then, when related in the same equations in (1 -69) = 0, the solution is (1-70)
T he defect of this system is 2. T he procedure followed for the simple case o f 2 equations in 3 unknow ns is also applicable to the general case of m equations in n unknow ns:
a n a i\ a 12
0-22 ' '
dl a 2n
<
(
*1 c = ^
X2
c2
(1-71)
<^m2
fl'mn
Xn
Cm
If a is of ran k m, there exists an mth o rder array which has a nonvanishing determ inant. W e rearrange the colum ns such th a t the first m colum ns are
32
CHAP. 1
linearly independent. P artitio n in g a an d x,

"flu
21 a 12 ^22
a lm &2m . n "mm
1 1
^ 1, m 4- 1 a 2,
1
* ' '
Clin 2n = [ A, (m * m)
^2
flml
{Xi 2
| 1 j
(m x (w m))
'
X2
m+ 1
1
ttmn
(1-7 2 )
=
X}
{ x,
(! X 1 )
((n -m )x 1 )
x2 }
we w rite (1 -7 1 ) as Aj Xj c A 2X 2 (1-73)
Since jA^ # 0, (1 -7 3 ) can be solved for X i in term s of c an d X 2. T he defect of the set is n m , th a t is, the solution involves n m arb itrary constants represented by X 2. Suppose a is of ran k r w here r < m . Then, a has r row s w hich co n tain an rth -o rd er array having a nonvanishing determ inant. T he rem aining m r row s are linear com binations of these r rows. F o r (1 -7 1 ) to be consistent, th a t is, have a solution, the relations between the row s o f c m ust be the s a m e as those for a. T he defect for this case is n r. Example 1 - 1 5 ----------------------------------------------------------------------------------------As an illustration, consider the third-order system
a l 1-^1 + a l2x 2 +
^ 1 3 * 3 C1
&2lx l + a22*2 + fl23^3 ~ c2

f l 31-x l +
(a)
a i 2 x 2 + a 33-x 3 c 3
Suppose that r ~ 2 and the rows of a are related by (third row)

= A,
(first row)
4- A2 (second
row)
(b)
For (a) to be consistent, the elements of c must satisfy the requirement,

C3 Aj Cj -j-
(c)
the second by A 2, and add to
To show this, we multiply the first equation in (a) by these equations the third equation. Using (b), we obtain
0 c3 Ai Cj A 2c2
(d)
Unless the right-hand side vanishes, the equations are contradictory or inconsistent and no solution exists. When c = 0, (c) is identically satisfied and we see that (a) has a nontrivial solution (x ^ 0) only when r < 3. The general case is handled in the same manner.*
* See Prob. 1-45.
REFERENCES
33
In general, (1 -7 1 ) can be solved when r < m if the relations between the rows of a an d c are identical. W e define the augm ented m atrix, a, for (1-71) as
a 1j a 12
&I n 02
Cl
a --=
a 21
#22
C2
= [a
c]
(1-74)
Cm
W hen the rows of a and c are related in the same way, the rank of a is equal to the rank o f a. It follows th a t (1-71) has a solution only if the ran k of the a u g -. m ented m atrix is equal to the ran k o f the coefficient m a trix : r() = r(a) (1-75)
N ote that (1 -7 5 ) is always satisfied when r{a) = m for arbitrary c. We can determ ine r(oe) and ?-(a) sim ultaneously using elem entary operations on a provided th a t we do n o t interchange the elem ents in the last column. The reduction can be represented as a = [a c] o (1-76)
a 1
where A( 1 1> is o f ran k r(a). If C 'j1 has a nonvanishing element, r(a) > r(a) and no solution exists. W hen r(a) = r(a), (1-7.1) contains r independent equations involving n un knowns. The rem aining m r equations are linear com binations o f these r equations an d can be disregarded. Thus, the problem reduces to first finding r(a) and then solving a set of r independent equations' in n unknow ns. The com plete problem can be efficiently handled by using the G auss elim ination procedure (Refs. 9, 11, 13).
REFERENCES
1. 2. 3. 4. 5. 6.
7. 8. F r a z e r , R. A., W. J. D u n c a n and A. R. C o l l a r : E le m e n ta r y M a tr ic e s , Cambridge
University Press, London, 1963.

T h o m a s , G. B., J r . ; C a lcu lu s a n d A n a ly tic a l G e o m e tr y , Addison-Wesley Publishing
9.
Co., Reading, Mass., 1953. Bo d e w i g , E . : M a tr ix C a lcu lu s, Interscience Publishers, New York, 1956. H o h n , F. E . : E le m e n ta r y M a tr ix A lg e b r a , Macmillan Co., New York, 1958. H a d l e y , G .: L in e a r A lg e b r a , Addison-Wesley Publishing Co., Reading, Mass., 1961. H o u s e h o l d e r , A. S . : T he T h e o ry o f M a tr ic e s in N u m e ric a l A n a ly s is , Blaisdell, Waltham, Mass., 1964. N o b l e , B . : A p p lie d L in e a r A lg e b r a , Prentice-Hall, New York, 1969. H i l D e b r a n d , F. B. : M e th o d s o f A p p lie d M a th e m a tic s , Prentice-Hall, New York, 1952. Faddeeva, V. N . : C o m p u ta tio n a l M e th o d s o f L in e a r A lg e b ra , Dover Publications, New York, 1959.
34
INTRODUCTION TO M ATR IX A L G E B R A
CHAP. 1
10. 11. 12. 13. 14. 15.
R a l s t o n , A. and H. S. W i l f : M a th e m a tic a l M e th o d s f o r D ig it a l C o m p u te r s , Vol. 1,
Wiley, New York, I960. R a l s t o n , A. and H. S. W i l f : M a th e m a tic a l M e th o d s f o r D ig ita l C o m p u te r s , Vol. 2, Wiley, New York, 1967. B e r e z i n , I. S. and N. P. Z h i d k o v : C o m p u tin g M e th o d s , Vols. 1 and 2 , Addison-Wesley Publishing Co., Reading, Mass., 1965. F o r s y t h e , G. E., and C. B- M a l e r : C o m p u te r S o lu tio n o f L in e a r A lg e b r a ic S y s te m s , Prentice-Hall, New York, 1967. V a r g a , R. S .: M a t r ix I te r a tiv e A n a ly s is , Prentice-Hall, New York, 1962. C o n t e , S. D .: E le m e n ta r y N u m e r ic a l A n a ly s is , McGraw-Hill, New York, 1965.
PROBLEMS
1-1.
(a)
C arry o u t the indicated o p eratio n s: "l 3 5 4 2 1 1 o ~2 1 + 7 0 0 i l 3 1 5 5' 3 6 5" 3 -1 2_
(b)
2
1
6_
(c)
1 2~ -1 r _ 4 2 3" +2 0 3_ _3 4_ 1 5_
1 2
(d)
-3 (e)
- 1 2 (0 4 .2 f 3 I' -3
4
4 2 1 3 1 -3
'-1 2
1-2.
E xpand th e following p ro d u cts: $2? ?C lnJ

' >îj
(a)
(b)
{ a u a 2 , . . , a } [ b u b 2 , , fc]
(c)
ci 0
i o
a ll <*12 ^22
C2_
a 21
P R O B LE M S
35
(d)
11
_CI2 l a 22_
1 0
c2
1 -3 .
S how th a t the p ro d u c t of
3
Sj
+ a 2 + a 3 Y j ak 1
3
S 2 by + b2 -f b3 = can be w ritten as
3 3
k= 1
bk
5 i 52 = Z
akbj
k= \ j = i
G eneralize this result for the sum o f n elem ents. 1 -4 . S uppose th e elem ents o f a an d bare functions o f y. Let da dy U sing (1-1 9 ), show th a t if c = ab then dc db da. d y = * T y + d~y 1 -5 . C o n sid er the triple p ro d u ct, abc. W hen is this p ro d u c t defined? Let p = abc D eterm in e an expression for W h at is the o rd er o f p ? D eterm ine Pij for the case w here c = a r . 1 - 6 . E v alu ate the follow ing p ro d u cts: daik _ dy _ db dy dbik
Jy_
(a)
' 3 -2 (b) (a + b)(a + 4 1. "1 > 2 5. 5 4 f I_ .
b )
w here a is a sq u are m atrix.
1 -7 . Show th a t the p ro d u ct o f tw o sym m etrical m atrices is sym m etrical only w hen they are com m utative. 1 -8 . Show th a t the follow ing p ro d u cts are sym m etrical:
(a)
(b)
aTa a Tba w here b is sym m etrical w here c is sym m etrical
(c)
bTa r cab
36
INTRODUCTION TO M A T R IX A L G E B R A
CH AP. 1
1 -9 . E v alu ate th e following m atrix p ro d u ct, using th e in d icated su b m atrices: ~1 3 | 2 ' 2 1 | 3 3 3 5 1 4 1 r 2 1 -1 0 . L et c = ab. Show th a t th e h o rizo n tal p artitio n s of c co rresp o n d to those o f a an d the vertical p artitio n s o f c co rresp o n d to those o f b. H i n t : See Eq. (1-37). 1 -1 1 . A m atrix is said to be sym m etrically p artitio n ed if th e lo catio n s of the row an d colum n p artitio n s coincide. F o r exam ple, i al2 I a13 I a2t a22 I a23 a 33 a3l a32 is sym m etrically p artitio n ed an d i a 11 i
a21
I
a12 I
a 22 J
a 13
a 23
a3l j ^32 J ^33

is unsym m etrically partitioned. N 1 symmetrical partitions.
a = [A iy]
Suppose we p artitio n a square m atrix with i j = 1, 2 .,N
(a) (b) 1-12.
D educe th a t th e d iag o n al subm atrices are sq u are an d A , A Jr have th e sam e order. If a = a T, deduce th a t . .T A = C onsider the p ro d u ct o f tw o sq u are nth o rd er m atrices. c ab
(a)
If a an d b are sym m etrically p artitio n ed , show th a t C jk, A jk, B ik are of the sam e order. Illu strate for th e case o f one p artitio n , e.g., a = An ^21 A1 2 A 22
Suppose we sym m etrically p a rtitio n c. W h at restrictio n s are placed on th e p artitio n s o f a an d b? D oes it follow th a t we m ust also p artitio n a an d b sym m etrically? H i n t : See P ro b . 1 -1 0 . 1 -1 3 . C onsider th e triple pro d u ct, c = a Tba w here b is a sym m etric rth-order sq u are m atrix and a is o f o rd er r x Suppose we sym m etrically p artitio n c. T he o rd er of th e p artitio n ed m atrices are indicated in parentheses.
J p x p) ( p x q)
(b)
(nx n)
C
Cn C 21
(q x p )
Ci 2 C 22 .
x q)
PR O B LE M S
37
(a)
S how th a t the follow ing p a rtitio n in g o f a is co n sisten t w ith th a t o f c.

(r x n) (ex p1 ! (i- x q)
= [ Ai
A2 ]
(b) E xpress C jk ( j , k 1, 2) in term s o f A i, A 2, a n d b. 1-14 . L et d = [ D ;] be a q u asi-d iag o n al m atrix . S how th a t da = [DjAjfc] bd = [B jfcD,.] w hen th e m atrices a re co n fo rm ab ly p a rtitio n e d . 1 15. D e term in e th e n u m b e r o f in v ersio n s an d in terch an g es for the follow ing sets.
(a)
(b)
(4, 3, 1, 2)
(3, 4. 2, 1) H o w m an y p e rm u ta tio n s d o es (1, 2, 3, 4, 5) h av e? C o n sid er th e term s a lai a2a2 a 3n (a) (b)
1-16. 1 -1 7 .
and
ePiii2p3aPiiaPz^lfl3i
T he first su b scrip ts in (a) are in n a tu ra l o rd er. W e o b ta in (b) by rearran g in g (a) such th a t th e second su b scrip ts are in n a tu ra l o rd er. F o r exam ple, rearran g in g
e 231 a 12 a 23 a 31 a =(2,3,1) $ (3,1,2) .
we o b ta in e312 a 31 ^12 a 23 Show th a t if (act , a 2, a 3) is a n even p e rm u ta tio n , ( $ i, $ 2> $ 3) is also an even p erm u ta tio n . U sing this result, show th a t ^ î3(2a3
^2a2^3a3
and, in general, ^
|a| = |aT|
1-1 8.
C o n sid er th e term s ^ccia2a^laci^2x2^3^3 (a) (b)
S uppose th a t < ^2i T hen, (b) tak es th e form S how th a t ^2 < * 2 ^lot2 ^3a3 ^3aj
(c) = - ( a )
G eneralize this resu lt an d estab lish th a t th e sign o f a d e te rm in a n t is rev ersed w hen tw o ro w s are in terch an g ed . 1-1 9. C o n sid e r th e th ird -o rd e r d e te rm in a n t
| |
^âtigian^lai ^ j n ^ 3 a i
38
CHAP. 1
Suppose the second row is a m ultiple of th e first row :

a 2j = fcau
Show th a t I I = 0 .(H int: e. ). G eneralize th is result an d establish S *I properties 5 an d 7 of Sec. 1 -7 . 1-20. S uppose all th e integers of a set are in n a tu ra l o rder except for one integer, say n, w hich is lo cated a t p o sitio n p. W e can p u t th e set in n a tu ra l o rd er by successively in terch an g in g ad jacen t integers. F o r exam ple, 3 2 1 3 2 1 1 2 3 1 2 3 1 1 2 2 3 3
Show th a t |n p| ad jacen t interchanges (called tran sp o sitio n s) are required. It follows th a t th e sign o f th e resulting set is changed by ( _ 1 ) I - pI 1-21. W e can w rite the ex p an sio n for th e th ird -o rd e r d eterm in a n t as
L
i i
a u \ L L ei ^ a 2j a
\j k
?,k j
i = j k i,j,k = 1 , 2 , 3
U sing th e result o f the previous p roblem ,

eijk = ( - l ) |f- %
= ( - l ) i + l ejk
(b)
(c)
and (a) reduces to i, ( - I Y * l M u * Z i=1 = 1
F ollow ing this ap p ro ach , establish L aplaces cofactor expansion fo rm u la for an n th -o rd er d eterm in an t. 1-22. U se L aplaces ex p an sio n fo rm u la to show th a t
1
0
0 1
i i !
0 0
0 0
0 0
(pxp)
0
b n ^21
(p x n) I 0
0 b l2
^22
' '
0 flu
0 12 a 22
'
'
b ! a ( n x p ) (n x n)
C l. 2 "
a 21
'
'
i.i
hm
'
a ni
Cln2
'
1-23.
C o n sid er the qu asi-d iag o n al m atrix,

(P X P) (P x q)
d = By expressing d as
D, 0
(9 x P)
0 D2
"
(q x q)
d = "D l 0
1 \a
Ip 0
o d 2
P R O B LE M S
39
show th a t M - N
|D ,| . Verify this resu lt for
1 1 0 o'
,
"
2
0
3
0
0
2
0
1
0 G eneralize for
3_
d = [A, Sy] 1 -2 4 . Let

_(p x p) (p x q)_
"G _G 2l
(< ? * p)
0
(q * ?)
"
G 22_
Show th a t Ib M G u I | g 22| G eneralize for a q u asi-trian g u lar m atrix w hose diagonal subm atrices are square, of various orders. 1 -2 5 . S uppose we express a as th e p ro d u ct of a low er trian g u lar m atrix, g, and an u p p er trian g u lar m atrix, b.
1 O
a ii a 2i
12
a m
9 11
=
0 0 0
b i2
'
'
U 22
'
'
a 2n
9 21
922
^22
b 2n
a,a
'
a ,w
_9nl
9m
9nn_
W e in trod u ce sym m etrical p artitio n s after row (and colum n) p and w rite the p ro d u ct as
(pxp) (p X q)_ 12 ^ 22_ (pxp) (p x q) 0 " (pxp) (px?)_ B
i
r An : > K J
1
"O n _ g 21
(<7 X P)
(<f x pi
(? x q)
(q * < T )
rB G 22_ L
b 22
(<? x <?)
(q * p )
N ote th a t the diagonal subm atrices of g an d b are trian g u lar in form. (a) Show th a t G j iB n = A u G u B 12 - A 12
G 2 i B u A21
^ 21^12 + G 22B22 = ^22 (b) Show th a t |A ,,| = | G l a| | B , , | an d la l = |G | |G 2Z| |Bu [ [B22j (c) S uppose we require th a t |g| ^ 0 |b| ^ 0
By taking p = 1, 2 , . . . , n 1, deduce that this requirem ent leads to the
40
CHAP. 1
following n conditions on the elements of a :

a ll a 2i
a 12
a 22
'
' ' <>2j r
0
j = 1 , 2 ,. . .. n
The determ inant of the array contained in the first j rows and columns is called the /th-order discriminant. 1-26. Does the following set of equations have a unique solution? "l
1 3 2 3 7 3~ 5 11
*
3
5V
1-27.
Determine the adjoint matrix for 1 2 a =
Does a 1 exist? 1-28. Show that b ~ l,T bT 1-29. Find the inverse of
(a)
3' 2 (b) 4 5
(c)
2
1
(d)
2
1 Let
" n
a \2 a 22
a = and
! 1
a i3
Ai i
23
A l2 A 2 2_
a i\
A, |
a il ~ ~ 7i
r v
i a where the order of
12
B21
2:
is the same as Ajk. Starting with the condition aa 1 13
PROBLEMS
41
determ ine the four m atrix equations relating B,t and A k (j, k 1, 2). Use this result to find the inverse of _ "1 2 4
2 1 2
1 2 1-31. Find the inverse of A

h
Note th at A is (p x q). 1 -3 2 . Find the inverse of d 1 -3 3 . Di 0 0

d
Use the results of Probs. 1-31 and 1-3 2 to find the inverse of b = Bu _0 B 12 B 22
where B n , B 22 are square and nonsingular. H in t : write b as b = 1 -3 4 . I 0 0 B r /B i : B2 V I I 0 0 B 22
Consider the 3 x 4 m atrix 'l

a
2
3
2
1 2
O32 O a n d f l u = a.22 <333 = + 1.
D eterm ine the elementary row operation m atrix which results in a2i = a3l = 1-35. Let
_ ( P X P)
(n x n l
(P M )_
An _A2i ~(?xp)
Aj2 A 22 _ (qxq)
where |A n | = 0 and |a| ^ 0. Show that the following elementary operations on the partitioned rows of a reduce a to a triangular matrix. D eterm ine A\V. 0 C
h
-A
21
'A fi1 0 L ? _ 0
Ai 1 A 21
A 12 A 22
X
0
A 12 L
A(l)"
C (A22 ~ A 2iA n JA i2) 1 1-36. Suppose we want to rearrange the columns of a in the following way: 1 2
a =
2 3
1 4
col 1 - col 3 col 2 - col 1 col 3 -+ col 2
42
CHAP. 1
(a)
Show th a t postmultiplication byII(w hich is called a p erm u tatio n m atrix) results in the desired colum n rearran g e m en t:
0
1 0
o
0 1
1
0 0
(b)
(c) (d) (e)
N o te th a t we ju st rearran g e the corresponding colum ns o f I 3. Show th a t premultiplication by II r rearranges th e row s of a in the same way. Show that i r n = I 3. G eneralize for the case w here a is n x n. Show th at i l T a n l = la I
1-37. Let a be of o rd er 2 x n , w here n ^ 2 . . Show th a t a is of ran k 1 when the second row is a m ultiple of the first row. Also, show th a t when r = 1, the second, t h i r d , . . . , nt h colum ns are m ultiples o f the first colum n. 1 -3 8 . D eterm ine the ran k of
(a)
l 5 3 (b)
1
3 2 -4 2 4 -2
1 4
-1 0 _
3 6 -3
-- 2
1
2 - 1
1 -3 9 . Let a be of o rd er m x n an d rank r. Show th at a has n r colum ns which are linear com binations of r linear independent colum ns. Verify for 2 a = 1 7 3 3 12 4 2 14
1 -4 0 . U sing p roperties 3, 4, and 7 of d eterm inants (sec Sec. 1-7), deduce th a t th e elem entary o p eratio ns do n o t change the ran k o f a m atrix. F o r co n venience, consider the first r row s an d colum ns to be linearly independent. 1 -4 1 . F in d the ran k of a by reducing it to an echelon m atrix.
2 13
3 " 3 7 4 7 2 9
1 1 2
2
1 7
PROBLEMS
43
1 -4 2 .
Show th a t c is at m ost of rank 1.

ai
a 2
[M 2
a
W hen will r(c) = 0? 1-4 3 . C onsider the product,

a ll a 12
c =
021 C lffii
22
&m 2
b n
&U&12
b u b 22
b ln
bzn
(a)
S u p p o se
b is o f r a n k
1an d
0.
)
T h e n , w e
can
w rite
S,
V = v-k < ,
I h i )
k 2,
[ b 2k \
(b)
Show th at the second, third____ nth colum ns of c are m ultiples o f the first colum n and therefore r(c) <: 1. W hen will /'(c) = 0? Suppose r(a) = 1 and a n # 0. In this case, we can write
[ f l jVa /2] = / L j O ? ! va l 2 ] j = 2,
Show th at the second, th ir d ,. . mth roW vS of c are multiples of the first row and therefore r(c) < 1. W hen will r(c) = 0? . 1 -4 4 . C onsider the product
a ll
f l 12
^ 1s
ft 2 s
b n bzi
b \2 b 22
' '
'
b ln b 2n
c Let
021
a 22
(a)
b sn
a m2
bsi
b s2
Ay
Bk =
\Pjlj2 ' ' aJS]

{ b i k b z k b sk}
(b)
U sing (b), we can w rite (a) as Ax c = A 2 s [ B 1B2 -- - B] (c)
Suppose r(a) = ra, r(b) = rh. F o r convenience, we assume the first ra rows of a and the first rb colum ns of b are linearly independent. Then,
ra
A j ^ A jp \p j v a "f"
15 ^ti
I-
2,
. . . , wi
p-1
44
INTRODUCTION TO M ATRIX ALG E BR A
CHAP. 1
rb
=
(a) (b) (c) (d)
q- 1
a kq^q
k ~
rb
I , l'b + 2 , . . . , 11
Show th a t rows ra + 1, ra + 2 , . . . , m of c are linear com binations of the first ra rows. Show th a t colum ns rb + 1, rb + 2 , . . ., n of c are linear com binations of the first rb columns. F rom (a) and (b), w hat can you conclude about an u pper bou n d on r { c)? To determ ine the actual ran k of c, we m ust find the ran k of
Aj >[BiB2 BJ = _Ar.Bj r t Ar aB 2 K K A ,B ! A 2Bi A iB 2 A 2B 2 AiB,f c A 2Br f c
(e)
Suppose ra ^ rb. W h at can you conclude ab o u t r(c) if A* is orthogonal to B l5 B2, . . . , Brft? U tilize these results to find the ran k of '- 1 / 2 - 1/2 -1 1/2 1/2 1 O "l 1 1 1 _t 0 0 1 f 1 2
(f)
Suppose ra ~ rb = s. Show th at r(c) s. Verify for l 2 3 3* -2 n -i 2 r 4
1 1 1
1-45.
C onsider the m x n system u tin 1 22 ami Aj am2 ' ln ' a2 t'l Cn \
X2
: | CmJ j 1 , 2 , . . . , tti (b)
U sing (b), we w rite (a) as AjX = Cj

r
; = 1, 2, . . . , m
(c)
N ow, suppose a is of ran k r an d the first r rows are linearly independent. Then, Y. "WAP k = r + 1, r + 2, . . . , m p= i Show th a t the system is consistent only if
r
Ak =
(a)
z ^ PCP k r + 1, r + 2 , . . . , m P= t N ote th a t this requirem ent is independent of w hether m < n or m > n.
C k=
PROBLEMS
45
(b)
If m < n and r m, the equations are consistent for an arbibrary c. Is this also true when m > n and r = n l Illustrate for
L -1 1 and
I
J * . ! c i ] * 2 = \r * 3
< ' >
-1 1 1-46.
f-v,)
Cl
^2 >
C onsider the following system of equations: x ( -)2-V1 + X2 X2 4x2
2 .x j +
2 x4
= 4
6
3 x 3 -t~ 2 X 4 ~
3xt +
2x3
+ x4 = 9
7xi + l x 2 + 9.x 3 + 7x4 = 23 (a) (b) D eterm ine w hether the above system is consistent using elementary operations on the augm ented matrix. Find the solution in term s of x4.
Characteristic-Value Problems and Quadratic Forms

2-1. INTRODUCTION Consider the second-order homogeneous system, (an - *)xt -f a , 2x2 = 0 ^ 21^1 + (a 22 % )X2 ~ 0 where X is a scalar. Using matrix notation, we can write (2-1) as ax = Xx or (a Xl2)x = 0 (2-3) The values of X for which nontrivial solutions of (2-1) exist are called the characteristic values o f a. Also, the problem of finding the characteristic values and corresponding nontrivial solutions o f(2 -'l) is referred to as a second-order characteristic-value problem.* The characteristic-value problem occurs naturally in the free-vibration analysis of a linear system. We illustrate for the system shown in Fig. 2-1 . The equations of m otion for the case of no applied forces (the free-vibration case) are dt
(2 -2)
(a)
* A lso called e igenvalue problem in so m e texts. T he term eigenvalue' is a hybrid o f the G erm an term E igenw erte and English value.
SEC. 2 -1 .
INTRODUCTION
47
Assuming a solution of the form V l = A xeu y 2 = A 2eu (b)
and substituting in (a) lead to the following set of algebraic equations relating the frequency, a> , and the amplitudes, A 2: (/<! + k2)Ai k2A 2 = m ^ A x
k 2A x
+ k2A 2 = m2w 2A i
(c)
We can transform (c) to a form similar to that of (2 -1 ) by defining new am plitude measures. X = at2
Ai = Az A 2sf m 2
(d)
and the final equations are

/l 2 A/I ]
(e)
/I
1
* 4 ---- A y X A 2
^2
in.
T he characteristic values and corresponding nontrivial solutions of (e) are related to the natural frequencies and norm al m ode am plitudes by (d). N ote that the coefficient m atrix in (e) is symmetrical. This fact is quite significant, as we shall see in the following sections. ,
m2
*2
yi .
mx
/J S v
Fig. 2 - 1 . A system w ith tw o d e g re e s of fre e d o m .
Although the application to dynamics is quite im portant, our prim ary reason for considering the characteristic-value problem is that results obtained for the characteristic value problem provide the basis for the treatm ent of q uadratic
* See Prob. 2-1.
48
CHARACTERISTIC-VALUE PROBLEMS
CHAP. 2
forms which are encountered in the determination of the relative extrema of a function (Chapter 3), the construction of variational principles (Chapter 7), and stability criteria (Chapters 7,18). This discussion is restricted to the case where a is real. Reference 9 contains a definitive treatment of the underlying theory and computational procedures. 2-2. SECOND-ORDER CHARACTERISTIC-VALUE PROBLEM
We know from Cram ers rule that nontrivial solutions of (a n Ape f + tfi2*2 = 0 02 lXj + (a2 2 - X)x2 = 0 (2-4)
are possible only if the determinant of the coefficient matrix vanishes, that is, when
lll ~ ^ a\2
a 2\ a 22 ~
- 0
(2-5)
Expanding (2-5) results in the following equation (usually called the charac teristic equation) for A: ' X1 (an + 022)X + ( a n a 2z < 221^ 12) = 0 We let fi\ all + a22
$2 ~ a iYa l2 ~ ^12^21 |aj
(2 - 6)
(2-7)
and the characteristic equation reduces to X1 - fa x + ji2 = 0

(2 -8)
The roots of (2-8) are the characteristic values of a. Denoting the roots by Aj, X->, the solution is ______
^ i = (0i / t f ~ 4 j 0 2)/2 (2-9)
When a is symmetrical, a 12 = a2\, and fi\ 4/?2 = (au C I2 2 ) 2 + 4(12)z Since this quantity is never negative, it follows that the characteristic values for a symmetrical second-order matrix are always real. Example 2-1 --------------------------------(1)
'2
a =
2 5_ ^= 2+ 5= 7
f t = (2)(5) - (2)(2) = 6 The characteristic equation for this matrix is
a2
IX + 6 0
(a)
SEC, 2 -2 .
SECOND-ORDER CHARACTERISTIC-VALUE PROBLEM
49
Solving (a),
A, = +1
(2)
1
a
-2
010
/) 2 = + J
J ~L 02 ~ +1
where ; = ^ / ~ 1
By definition, nontrivial solutions of (2-4) exist only when X Xx or X% . In what follows, we suppose the characteristic values are real. We consider first the case where / = / , . Equation (2-4) becomes
( fll l ^l)x l + ^12 -^ 2 ~ 0
(a)
a21Xl + (a22 ~~ XV )X2 0 The second equation is related to the first by second eq. ( I times the first eq. \ an ~ XiJ This follows from the fact that the coefficient matrix is singular. (a n Ai)(o22 ~ A ) ~ a i 2 an 0 (c) (b)
Since only one equation is independent and there are two unknowns, the solu tion is not unique. We define x\1', x( 2 u as the solution for X = Xx. Assuming* that i2 / 0, the solution of the first equation is ~ c, X?> = a l2 where c* is an arbitrary constant. Continuing, we let
x (1)
(d)
{x( ,n, X # }
(e)
and take cx such th a t (x(1))r x U) ~ 1. This operation is called normalization, and the resulting column matrix, denoted by Q j, is referred to as the charac teristic vector for . Qi = c , < i + i ,
a 12 j
(2 -io )
2
v1 = 1, + a H Aj _ a 12 //
By definition, Q iQ i = 1
(2 -1 1 )
* If a 12 0, w e w ork w ith the second equation.
50
C H AR AC TER ISTIC -VA LU E PR O BLEM S
CHAP. 2
Since Q i is a solution of (2 -4 ) for X Xy, we see th a t a Q i - A iQ i F ollow ing the sam e procedure for X X2, we o b tain Q 2 = c2 i + 1 , ------ (2 -1 3 ) I2 ) an - h _ a >2 _ (2-14)
Q 2
(2-12)
I w here (* Y w Also, Q [Q 2 = 1 aQ 2 ^ <+
It rem ains to discuss the case w here Xt = X2 If a is sym m etrical, the ch ar acteristic values will be equal only w hen o n = 'fl22 a n d a 12 = ci21 = 0- E q u a tion (2 -4 ) takes the form (a u - X)x1 + (0)x2 = 0 (O)^! + (a u X)x2 0 These equations are linearly independent, an d the two independent solutions ^
x(2) = {0, c 2} T he corresponding characteristic vectors are Q i = { + 1,0} Q 2 = {0, + 1 } If a is not sym m etrical, there is only one independent nontrivial solution w hen the characteristic values are equal. It is of interest to exam ine the pro d u ct, Q T Q 2- F ro m (2 -1 0 ) an d (2-13), we have Q 1Q2 =
- C iC 2
_ 15j
1 + -------------- 2 --------------V t2 }
(a)
N ow, when a is symmetrical, the rig h t-h an d term vanishes since

2
a \\ X2 = {< *22 ^ 1) -------- - flu ~ ^1
(b)
and we see th at Q [ Q 2 = 0. This result is also valid when the ro o ts are equal. In general, Q [ Q 2 ^ 0 when a is unsymmetrical. Tw o n th o rd er colum n vectors U , V having the p roperty th a t U r V = V rU = 0 (2-1 6 )
are said to be o rthogonal. U sing this term inology, Q x an d Q 2 are o rth o g o n al for the sym m etrical case.
SEC. 2 -2 .
SECOND-ORDER CHAR AC TER ISTIC-VALU E PROBLEM
51
Example 2 -2 ---------------------------------------(1)
2
2
2
5_
A2 = +1
A, =
T h e e q u a tio n s fo r A = Aj = + 6 are
-I- 6
4.x:,
2xz =
2x, - x 2 = 0
W e s e e t h a t t h e s e c o n d e q u a t i o n is w e o b ta in j tim e s th e fir s t e q u a t io n . S o lv in g th e fir s t e q u a t io n ,
x( iu = ci
Then,
,^n = 2 x \ i] = 2c j
x'1 1 = c ,{ l, 2}
a n d t h e n o r m a l i z e d s o l u t i o n is
R e p e a tin g fo r A =
A2 =
+ 1 , w e fin d
x (2> =
and
c 2 {\,
- * }
O n e c a n e a s ily v e r ify th a t aQ j = and A jQ

j
1 ,2
Q[Q 2 - QlQ i - 0
( )
1 8
a = I 3 T h e c h a r a c t e r is t ic v a lu e s a n d c o r r e s p o n d in g n o r m a liz e d s o lu t io n s fo r th is m a t r ix a r e
Aj =
+5
A2 =
-1
Ql = 7 ! {2 +1} Q 2 ~ ~ f y f {4 - 1} v
W e s e e t h a t Q [ Q 2 = 0 . A c tu a lly ,
Q IQ 2 =
(3)
ys5 1 -2 -t
i
1
Ai W e have in c lu d e d th is e x a m p le t o + i A2
illu s tr a te th e c a s e w h e r e th e c h a r a c t e r is t ic v a lu e s a r e
52
C H A R A C T E R IS T IC -V A LU E P R O B LE M S
C H AP. 2
complex. The equations corresponding to X /.j are
(1 - r)Xi 2xx = 0 Xj (1 + i)x2 = 0

N ote that the second equation is (1 i) times the first equation. The general solution is
x a) = d j l , ^
Repeating for X = X2, we find
When the roots are complex, X2 is the complex conjugate of XL. Now, we take c2 cx. Then, x (2) is the complcx conjugate of x(1). We determine Ci such that
(Xu))rxc2, = j
Finally, the characteristic values and characteristic vectors are
Xi,2 = i
In general, the characteristic values are complex conjugate quantities when the elements of a are real. Also, the corresponding characteristic vectors are complex conjugates.
2-3.
SIMILARITY AND ORTHOGONAL TRANSFORMATIONS
T h e characteristic vectors for th e seco n d -o rd er system satisfy the follow ing relatio n s:
a Q i = A i Q j
(a)
3Q2 ^ Q 2 W e can w rite (a) as 0

a [ Q i Q
2 ]
[ Q
2 J
(b)
N ow , we let q = [Q i "Ai A . = 0 Q2] 0
(2 -1 7 )
C olu m n j of q co n tain s th e norm alized so lu tio n for Xj. W e call q th e normalized modal matrix * for a. W ith this n o ta tio n , (b) takes th e form aq = q i (2 -1 8 )
* This terminology has developed from dynamics, where the characteristic vectors define the normal modes of vibration for a discrete system.
SEC. 2 - 3 .
S IM IL A R IT Y A N D O R T H O G O N A L T R A N S F O R M A T IO N S
53
W e have show n th a t th e c h ara cteristic v ecto rs are alw ays linearly in d e p e n d e n t w hen a is sy m m etrical. T h ey are also in d e p en d en t w hen a is unsym m etrical, p ro v id ed th a t ^ A2. T hen, |q| ^ 0 except for th e case w here a is un sy m m etrical a n d th e c h ara cteristic values are eq u al. If jq| ^ 0, q -1 exists an d we can express (2 -1 8 ) as (2 -1 9 ) T h e m a trix o p e ra tio n , p *( )p w h ere p is a rb itra ry , is called a similarity trans formation. E q u a tio n (2 -1 9 ) states th a t th e sim ilarity tra n sfo rm a tio n , q _ 1()q? reduces a to a d ia g o n al m a trix w hose elem en ts are th e c h ara cteristic values o f a. If a is sym m etrical, th e n o rm alize d c h a ra c te ristic vectors are o rth o g o n a l, th a t is,
Q[Q2 = Q2Q1 = o
Also, by definition,
Q[Qi = Q2Q2 = l
U sin g these p ro p erties, we see th a t q rq = an d it follow s th a t q -1
Qf
[Qi Q2] -
'1 0
O' 1
(2-20)
A sq u a re m atrix , say p, h av in g th e p ro p e rty th a t p r = p -*1 is called an orthogonal m a trix a n d th e tra n sfo rm a tio n , p T( )p, is called a n o rth o g o n a l tra n sfo rm a tio n . N o te th a t an o rth o g o n a l tra n s fo rm a tio n is also a sim ilarity tran sfo rm a tio n . T h en , th e m o d a l m a trix for a sy m m etrical m a trix is o rth o g o n al a n d we can w rite q r aq = X (2 -2 1 ) E x a m p le 2 - 3
(1)
a =
2 2
2 5
Ai +6
Qi r= (t,2 }
X=
'l 0
"
+6 0
o' + 1_ '1 2
a2_
q - [ Q . QJ-^75 2
-1
54
CHAP. 2
We verify that qr = q 1 and q7aq X: 1 2 1 2 T 2 -1 qq = 5 2 -1 1 2 2 1 2 aq = 2 5 2 1 75 2 6 0 1 1 0 1 75 2 -1 6 2 2 1 1 q aq = 12 - 1 2 -1 (2) a =

/i = +5
1 5 5 _0
0" 5_
1 0 ' I O 2" -1 . 1
1 " 6
7 5 _12 = qX
6 O'
= Jl
1 8
1 3
= - i 3i =
5 .0 O'
q2
-=
7 i7
{ 4 ,-1 }
4 /7 l7~ -1 /7 T 7 .
q =
2 / 7 5 .1/75"
Since a is not symmetrical, qT ^ q \ 1 ( V
Actually, -4 /7 I f 2/V5
'5 O '
.
'- 1 / 7 T ? 6 ) .-1 /7 5
'7 5 7 6 _ 7 l7 /6
275/3" -7 n /3 _
One can easily verify that q ~ aq 0 -1

= 3.
(3) a 3 1= 1 -2
1 -1 . +/ 0
0 -i
q = 7 2 /3
1 1 1 f 1 + 2 2
In this case, q involves complex elements. Since the characteristic vectors are complex conjugates, they are linearly independent and q -1 exists. We find q ~ \ using the defi nition equation for the inverse (Equation (1-50)): 1- i 2 q 1 = jjjj A4j q = 7 V 2 + i 1
SEC. 2 - 4
THE n th-O R D E R S Y M M E T R IC A L C H A R A C T E R IS T IC S
55
One can easily verify that q aq =

i + I 0
= I
2 -4 .
THE nth-ORDER SYMMETRICAL CHARACTERISTIC-VALUE PROBLEM
T h e n th o rd e r sym m etrical ch aracteristic-v alu e p ro b lem involves d e te r m ining th e ch ara cteristic values an d c o rresp o n d in g n o n triv ial so lu tio n s for
a i{X1 +
a12X -2 +
' ' +
a lnX n ~
^22-^2 4 " " ' ^ln-^1 + a2nX2 W e can w rite (2 -2 2 ) as '
& 2nX-n X x2 I annXn
(2 - 22 )
a x = Ax (a X\n)x = 0
(2 -2 3 )
In w hat follows, we su p p o se a is real. F o r (2 -2 3 ) to have a n o n triv ial so lu tio n , th e coefficient m atrix m u st be singular. ja - ALj = 0 (2 -2 4 ) T he ex p an sio n o f th e d e te rm in a n t is (-i)" (a b - i h ^ L + w here Pi = a ll + a 22 + ' ' + ann
P n = |a|
- + ( - 1 m ) = 0 (2 -2 5 )
an d (lj is th e sum o f all th e j t h o rd e r m in o rs th a t can be fo rm ed o n th e d ia g onal.* L ettin g Xlt A2, >K d en o te th e ro o ts, a n d expressing th e ch aracteristic e q u atio n in facto red form , we see th a t Pi = Ai + A2 + 4- A
P2 X\X2 + AjA3 + + / _ ]/ P X\X2 )-
(2 -2 6 )
W e su m m arize below th e th eo retical results for th e real symmetrical case. T he proofs a re to o d etailed to be in clu d ed h ere (see References 1 a n d 9): 1. 2. T h e c h ara cteristic values A1? X2, . . . , Xn, are all real. T h e n o rm alized ch aracteristic v ecto rs Q l , Q 2, ; Qn, ar e orthogonal: Q lQ j = < 5 0i , j = 1 ,2 ,.
* M in o r s h a v in g a d ia g o n a l p iv o t (e.g., d e le te th e k t h r o w a n d c o lu m n ). T h e y are g e n e r a lly c a lle d p rin cip a l m in o r s.
56
CHAP. 2
3.
a is d i a g o n a l i z e d by the orthogonal transform ation involvin g the nor m alized m odal matrix. qraq = X where
q = [ Q
1 Q
n]
A . ~ Example 2 -4 (1)
5 - 2 a -2
yj
0 3 -1
-1
1
[}
Since a is symmetrical, its characteristic values are all real. We first determine /?2, using (2-25): , . /), = 5 + 3 + 1 = + 9 j f?2 +11 + 5 + 2 = +18 0 3 = 5(2) ( 2)(2) = + 6 The characteristic equation is
/(A) = A3 - 9 A2 + 18A -
6 = 0
and the approximate roots are Aj +0.42 A2 w +2.30 A3 ^ +6.28 To determine the characteristic solutions, we expand a x = Ax, (5 - A ).X j ( 2)x j =
2xz
- x 3 = - ( 3 - A)x3 (1 - A)x3 = x2
Solving the first and third equations for x, and x 3 in terms of x 2, the general solution is
yU) x 1 _
S - lj
-xV>
j =
1 xy> = -.-1 1 - I-.

Finally, the modal matrix (to 2-place accuracy) is > 0 .2 2
1 , 2, 3
+0.51 +0.68 -0 .5 2
-0 .8 4 +0.54 -0 .1 0
q = [Q 1Q 2 Q 3J + 0.50
+0.85
(2)
a -
'1 2 0 2 1 0 0 .0 3
SEC. 2 -5 .
QUADRATIC FORMS
57
The expansion of |a AI3] = 0 is (3 - A)(( 1 - A}2 - 4) = 0 and the roots are

Xx = 3
A2 = 3
A3 = - 1
Writing out ax = Ax, we have
(1 - /.)x,
2.Vj When A = 3, (a) reduces to
+ 2x2 , +(1 - X)x2
= 0 = 0
(a)
(3 - /.)x3 = 0
2 x i 2xi + 2 x2 = 0 2 x 2 = 0
(b)
(0}.x3 = 0 We see from (b) that (a - AI3) is of rank 1 when X = 3. The general solution of (b) is *1 = Ci
X 2 = C, X i = c2
By specializing the constants, we can obtain two linearly independent solutions for the repeated root. Finally, the characteristic vectors for A, = X2 * 3 are
Q2 = {0, 0, 1 }
When X X^ 1, (a) reduccs to 2.x j + 2.x2 * 0 2x t + 2x 2 = 0 4.x3 = 0 The general solution and characteristic vector for A3 are x< 1 3) = x( 23) and = 0
This example illustrates the case of a symmetrical matrix having two equal characteristic values. The characteristic vectors corresponding to the repeated roots are linearly inde pendent. This follows from the fact that a AI3 is of rank 1 for the repeated roots.
2-5.
QUADRATIC FORMS
The hom ogeneous second-degree function F + 2^ 12^ 1X2 + S22x 2
is called a quadratic form in x 1} x 2. U sing m atrix notation, we can express F as
58
C H AP. 2
In general, the function F = I Z aJkx Jx k = * Ta x fc=i j= i (2 -2 7 )
w here aJk = a kj , for j ^ k, is said to be a q u ad ratic form in x lt x 2, . . . , x. If F = x Tax is n o nnegative ( ^ 0) for all x an d zero only w hen x 0, we call F a positive definite q u ad ratic form . Also, we say th a t a is a positive definite m atrix. If F ^ 0 for all x b u t is zero for som e x ^ 0, we say th a t F is positive semidefinite. W e define negative definite an d negative sem idefinite q u ad ratic form s in a sim ilar m anner. A q u a d ra tic form is negative definite if F sS 0 for all x an d F = 0 only w hen x = 0. T h e qu estio n as to w h eth er a q u a d ra tic form is positive definite is qu ite im p o rtan t. F o r exam ple, we will show th a t an equilibrium p o sitio n for a discrete system is stable w hen a certain q u a d ra tic form is positive definite. C on sid er the q u a d ra tic form F = *1 = [ x vx 2 x] 0 0 < -*2 (2 -2 8 )
W hen F involves only squares o f th e variables, it is said to be in canonical form . A ccording to th e definition in tro d u ced above, F is positive definite w hen &! > 0 It is positive sem idefinite when b, 0 b2 2* 0 bn > 0 b 2 > 0, / > > 0
an d a t least one o f the elem ents is zero. N ow , to establish w hether x r ax is positive definite, we first reduce a to a diagonal m atrix by applying th e tran sfo rm atio n , q _1( )q, w here q is th e o rth o g onal n orm alized m o d al m atrix for a. W e w rite x r ax = (xTq )(q ~ laq)(q *x) = (x r q) [ Then, letting y = q 'x (a) reduces to a canonical form in y: F = x Tax = y r [Ai< 5 Jj]y (2 -3 0 ) x = qy (2 -2 9 ) (q r x)
(a)
It follows th a t F is positive definite w ith respect to y w hen all th e ch aracteristic values of a are positive. B ut y is uniquely related to x a n d y = 0 only w hen x = 0. T herefore, F is also positive definite w ith respect to x. T h e p ro b lem of establishing w hether x r a x is positive definite consists in determ in in g w h eth er all th e characteristic values o f a are positive.
SEC. 2 - 5 .
Q U A D R A T IC FO R M S
59
W e consider first the seco n d -o rd er sym m etric m atrix a = an

_a 1 2
a 12
a 22_
U sing (2-26), th e ch aracteristic values are related by Al + A2

= fil = a ll
a 22
Al/-2 ~ P i Ct\\Cl22 ~~ &12 = |*| W e see from (a) th a t the conditions Pi > 0 are equivalent to
A\
> 0 A2 ^ 0
(a)
p2 > 0
(b) (c)
Suppose we specify th a t an > 0 |aj = = a u 22 ~ a 1 2 > 0 Since a n > 0 , it follows from th e second req u irem en t in (d) th a t Therefore, (d) is equivalent to (b). W e let A, = A2 = T hen, a is positive definite w hen fix > 0 or
Aj > 0 A2 > 0
a l l a 12 a l2<*12 a 22
(d) > 0.
= a
(2 -3 1 )
02 > 0 (2 -3 2 )
T he quantities fjj an d Aj are called the invariants an d discriminants of a. T he above criteria also apply for th e n th -o rd er case. T h a t is, one can show th a t a is positive definite w hen all its inv arian ts are greater th a n zero. > 0 p2 > 0 > 0 (2 -3 3 )
w here jS7 - is the sum o f all the yth -o rd er principal m inors. E quivalent con d itio n s can be expressed in term s o f the discrim inants. L et Aj rep resen t th e d e te r m in an t o f the array consisting o f th e first j row s and colum ns. 11
a l2
^12
a 22
a\j
a 2j
(2-34)
Qij T he conditions.
A: > 0
' a2j

aj j A > 0 (2 -3 5 )
A2 > 0
are sufficient for a to be positive definite.*

* See Ref. 1 fo r a d e ta ile d p roof. A ls o se e P ro b . 2 - 1 5 .
60
CHAP. 2
Example 2 - 5 -------------------------- ------------------------------------------------------------1 1 1 1 2 2 1' 2 3
The discriminants are
= +1
A2 = 2 - 1 = + 1
A3 1(6 - 4) 1(3 - 2) + 1(2 - 2) = + 1
Since all the discriminants are positive, this matrix is positive definite. The corresponding invariants are
/?] = 1 + 2 + 3 = + 6
P 2 = {2 -
1) + (3 1) + (6 4) = + 5
/?3 = A3 = +1 (2)
*1 1 _1 1 1"
-2
2
2
3
Since A2 is negative (A2 = 3), this matrix is n o t positive definite.
Suppose b is o b tain ed from a by an o rth o g o n al tran sfo rm atio n : b prap = p _1ap If a is sym m etrical, b is also sym m etrical: bT = pra Jp = pTap N ow , b an d a have the same characteristic values.* T his follows from |b - Xln\ = |p_1(a - AI)p| = ja - U n\ (2 -3 8 ) (2 -3 7 ) (2 -3 6 )
Then, if a is positive definite, b is also positive definite. In general, the positive definite ch aracter of a m atrix is preserved u nder an o rth o g o n al transform ation.
REFERENCES
1.
H i l d e b r a n d , F. B .: M e t h o d s o f A p p l i e d M a t h e m a t i c s , Prentice-Hall, New York,
1952. 2.
3.
B o d e w ig ,
E .: M a t r i x
C a l c u l u s , Interscience Publishers, New York,
1956.
S m ir n o v , V. I . : L i n e a r A l g e b r a , Addison-Wesley Publishing Co., Reading, Mass.,
1964. 4. 5.
T u r n b u l l , H. W., and A . C. A i t k e n : A n I n t r o d u c ti o n to th e T h e o r y o f C a n o n i c a l M a t r i c e s , Dover Publications, New York. H a d l e y , G .: L i n e a r A l g e b r a , Addison-W esley Publishing Co., Reading, Mass.,
1961.
* See Prob. 2 - 5 .
P R O B LE M S
61
6. 7. 8.
9.
C r a n d a l l , S. H . : E n g in e e r i n g A n a l y s i s , M cGraw-Hill, N ew York, 1956. N o b l e , B .: A p p l i e d L i n e a r A l g e b r a , Prentice-Hall, New York, 1969. F r a z e r , R. A ., W. J. D u n c a n and A. R. C o l l a r : E l e m e n t a r y M a t r i c e s , Cambridge
University Press, London, 1963.

W i l k i n s o n , J. H . ; T h e A l g e b r a i c E ig e n v a lu e P r o b l e m , Oxford University Press,
10. 11. 12. 13.
London, 1965. F a d d e v a , V. N . : C o m p u t a t i o n a l M e t h o d s o f L i n e a r A l g e b r a , Dover Publications, New York, 1953. R a ls to n , A., and H. S. W ilf: M a t h e m a t i c a l M e t h o d s f o r D i g i t a l C o m p u t e r s , Vol. 2, Wiley, N ew York, 1967. F o r s y t h e , G. E., and C. B. M a l e r : C o m p u t e r S o l u t i o n o f L i n e a r A l g e b r a i c S y s t e m s , Prentice-Hall, New York, 1967. P e t e r s , G ., and J . H. W i l k i n s o n : Eigenvalues o f AX ABX with Band Symmetric A and B , C o m p u t . J ., 12, 398-404, 1969.
PROBLEMS
2 -1 . C o n sid er th e system Ay = ABy
(a)
w here A a n d B arc sym m etrical rcth-order m atrices a n d A is a scalar. S u p p o se B can be expressed as (see P ro b . 1-2 5 ) B = bTb w here b is non sin g u lar. R educe (a) to th e form ax = Ax w here x by. D eterm in e th e expression for a in term s o f A a n d b. 2 -2 . Let x 1; x 2 be tw o rcth-order co lu m n m atrices o r co lu m n vectors a n d let ci, c2 be a rb itra ry scalars. If c jx i + c2x 2 = 0 only w hen c l = c2 = 0, x x an d x 2 are said to be linearly in d ep en d en t. It follows th a t Xj an d x 2 are linearly d ep en d en t w hen one is a scalar m u ltip le o f th e oth er. U sing (2 -1 0 ) a n d (2 -1 3 ), show th a t Q i a n d Q 2 a re linearly in d ep en d en t w hen Xi # X2. 2 3. D eterm in e th e ch ara cteristic values a n d th e m o d al m a trix for 3 2 7 (b)
(a)
2
(b) 0 3
0
5 0
3
0 2
2 - 4 . F o llo w in g th e p ro ced u re o u tlin ed in P ro b . 2 - 1 , d eterm in e th e c h a ra c teristic values a n d m o d a l m atrix for 12yi + 12y2 4 Xy^ 12y y + 63y2 = 9 Xy2
62
CHAP. 2
2 -5 .
Suppose th a t b is derived from a by a sim ilarity tran sfo rm atio n . h = p 'a p
Then, |b - Xln\ = |a - XIn\ an d it follows th a t b an d a have the same ch aracteristic eq uation, (a) D ed u ce th a t X{b) = a[0) ! i2 ii D em o n strate for a = 1 _i -2 -1 _ P = "1 2 r 3 k = 1,2, * , n
(b) 2 -6 .
$2? . , /? are invariant u n d er fo rm atio n is qu ite useful, Show th a t Q[b) = p - W W hen a is sym m etrical, we can w rite qraq X
Express a -1 in term s of q an d X- 1 . U se this result to find th e inverse of a = 2 -7 . 3 2 2~ 7
P ositive integral pow ers o f a sq u are m atrix, say a, arc defined as a = aa
a 3 = aa2 ? L r aar 1 If |a| ^ 0, a - 1 exists, an d it follows from the definition th a t a - 1 a'' = ar_1 (a) (b) S how th a t ar is sym m etrical w hen a is sym m etrical. L et It be a characteristic value o f a. Show th a t A- is a ch aracteristic value o f a r a n d Q , is th e co rresp o n d in g ch aracteristic vector. a rQi = Xr i Qi H i n t : S ta rt w ith aQ ( - an d p rem ultiply by a. 2 - 8 . A linear co m b in atio n o f n o n n eg ativ e in teg ral pow ers o f a is called a polynom ial function o f a an d w ritten as P( a). F o r exam ple, th e th ird o rd er polynom ial has th e form P(a) = c0a3 + Cj a 2 + c2a + c 3I N o te th a t P( a) is sym m etrical w hen a is sym m etrical.
P R O B LE M S
63
Let F(A) = 0 be th e ch ara cteristic e q u atio n for a. W h en the ch aracteristic values o f a are distinct, one can show th a t (see Ref. 1) F{ a) = 0 w here 0 is an n th -o rd e r n ull m atrix . T h a t is, a satisfies its ow n ch aracteristic eq u atio n . T h is resu lt is k n o w n as the C ay ley -H am ilto n T heorem . (a) Verify this th e o re m for 2 1' a = 1 2 Note: F(a) = a 2 Show th a t 1 = 7 r (a Pa (c) 2 -9 . ~ + (hh) for n 3
(b)
E stab lish a g en eral ex p ressio n for a 1 using (2-2 5 ). D eterm in e w h eth er the follow ing q u a d ra tic form s are positive definite. F = 2 x \ + 4 x 1x 2 + 3%2 F = 3x1 + 5 x f + 6 x f 4 x t x 2 + 6 x jx 3 8 x 2x 3 Show th a t a necessary b u t n o t sufficient c o n d itio n for a to be positive flu > 0, a22 > 0 , ----- a > 0
(a) (b)
2 - 10.
definite is (Hint:
T ak e x ( # 0 a n d x } = 0 for j # i , j = 1 , 2 , , n)
2 -1 1 . If |a| = 0, ax = 0 h as a n o n triv ial so lu tio n , say x ^ W h a t is the value o f x ] a x i ? N o te th a t X = 0 is a ch ara cteristic value o f a w hen a is singular. 2 -1 2 . L et C be a sq u are m atrix. S how th a t C TC is positive definite w hen |C| 0 a n d positive sem idefinite w hen |C| = 0. (Hint: S ta rt w ith F = x T( C TC)x an d let y = Cx. By definition, F can equal zero only w hen x = 0 in o rd e r for th e form to be positive definite.) 2 -1 3 . C o n sid er th e p ro d u c t C TaC , w here a is positive definite an d C is square. S how th a t C 7a C is positive definite w hen |C| 0 an d positive sem idefinite w hen jC[ = 0. G en eralize this resu lt for the m u ltip le p ro d u ct, C c ^ - c r a c ^ ' - c ^ c , , 2 -1 4 . L et a be a n m th -o rd er positive definite m atrix an d let C be o f o rd e r m x n. C o n sid er the p ro d u ct, b = C TaC Show th a t b is positive definite only when th e ra n k o f C is eq u al to n. W h a t can we say a b o u t b w hen r(C) < n l 2 -1 5 . C o n sid er th e q u a d ra tic form
al 1 Cl I 2 a 22
' '
'
a ln &2n
F = [ x t x 2 x]
al 2
<
x 2
2 n
'
@ nn
Vx
64
CHAP. 2
We partition a symmetrically, - [x rx ri ,(P*P) An

A j2
" ( < ?p)
(p*3) A 12
A 22 22
(4 x q)
| Xj | ( X2 J
X 1)
(f* 1)
where q n p. The expansion of F X aX has the form JF = X f A u X j + 2X [A 12A2 + X [A 22X 2 Now, we take X 2 0 and denote the result by Fp: Fp - X 'fA uX i F o r Fp > 0 for arbitrary X ls Au m ust be positive definite. Since )A1x\ is equal to the product of the characteristic values of A11? it follows that |A t i| must be positive. (a) By taking p ~ 1, 2 , . . . , n, deduce that Ap = |A U | > 0 p = 1, 2 , . . . , n
are necessary conditions for a to be positive definite. N ote that it remains to show that they are also sufficient conditions. (b) Discuss the case where Ap = 0. 2 -1 6 . Refer to Prob. t 25. Consider a to be symmetrical. (a) Deduce th at one can always express a as the product of nonsingular lower and upper triangular matrices when a is positive definite. (b) Suppose we take /jn - b22 = = bnn - + 1 Show that a is positive definite when
gS j >0 j = 1,2,..., n
and positive semi-definite when

9n ^ 0 j 1 , 2 , n
(c)
and at least one of the diagonal elements of g is zero. Suppose we take g = b f. Then,
and Ap - lA n | = bz n b l 2 ' ' ' bpp Show th at the diagonal elements of b will always be real when a is positive definite. 2 -1 7 . If a quasi-diagonal matrix, say a, is symmetrically partitioned, the subm atrix A n is also a quasi-diagonal matrix. Establish that a ~ [A A /] i,j = 1) 2 , ----- N
is positive definite only when A t {i 1, 2 , . . . , N) are positive definite.
PROBLEMS
65
H in t: Use the result of Prob. 1-23. Verify for i 3 0 0 0 o' 0 0 2 1 5 2
2 -1 8 . Suppose we express a as the product of two quasi-triangular matrices, for example, (n x n) a = where p + q n. We take B n lp B 22 = I,,
(p X p)
(PM )
Gn
0 G 22
(9
G 21
\q x p)
Bn 0
22
* q)~
Show th a t the diagonal subm atrices of g are nonsingular for arbitrary p when a is positive definite.
3
Relative Extrema for a Function
3 -1 . RELATIVE EXTREMA FOR A FUNCTION OF ONE VARIABLE
Let /( x ) be a function o f x w hich is defined for th e in terv al x t ^ x < x 2. If f ( x ) /( a ) ^ 0 for all values o f x in the to tal interval x t ^ x ^ x 2, except x = a, we say th e function has an absolute m in im u m at x a. If /( x ) f( a ) > 0 for all values o f x except x = a in the subinterval, ct ^ x ^ fi, co n tain in g jc = a, we say th a t f( a) is a relative m inim um , th a t is, it is a m in im u m w ith respect to all oth er values o f f { x ) for the p artic u la r subinterval. A bsolute an d relative m axim a are defined in a sim ilar m an n er. T h e relative m ax im u m an d m inim um values o f a function are called relative extrem a. O ne sh o u ld n o te th a t/( x ) m ay have a n u m b er o f relative extrem e values in th e to tal interval x t < x ^ x 2. As an illu stratio n , co n sid er the function show n in Fig. 3 -1 . T h e relative extrem a are f(a), f{b), f[c), f(d). U sing th e n o ta tio n in tro d u ced above, we say th a t f( b) is a relative m in im u m for th e in terv al ab < x ^ [ib. T h e ab so lu te m axim um an d m in im u m values o f / occur a t x = a an d x = d , respectively. m
Fig. 3-1. S ta tio n a ry p o in ts a t p o in ts A, B, C, and D.

66
SEC. 3 - 1 . R ELATIV E EXTR E M A FOR A FUNCTION OF ONE V A R IA B LE
67
In general, values o f x a t w hich the slope changes sign co rresp o n d to relative extrem a. T o find the relative extrem a for a co n tin u o u s function, we first d eter m ine the p o in ts a t which th e first derivative vanishes. These po in ts are called stationary points. W e then test each statio n ary p o in t to see if the slope changes sign. If the second derivative is positive (negative) the statio n ary p o in t is a relative m inim um (maximum ). If th e second derivative also vanishes, we m u st consider higher derivatives a t the statio n ary p o in t in o rd er to determ ine w hether the slope actually changes sign. In this case, th e th ird derivative m u st also vanish for the statio n ary p o in t to be a relative extrem um . Example 3 -1 -------------------------------------------------------(I) f ( x ) -jx 3 + 2 x 2 + x + 5 Setting the first derivative equal to zero,
= x 2 + Ax + 1 = 0 dx
df
and solving for x, we obtain
*i,2 = - 2
The second derivative is
d 2 1 r 2x " f" 4 2(x -j- 2)

dx
T hen,* x t = 2 + ^ 3 corresponds to a relative minimum and x = x 2 = 2 ^/3 corresponds to a relative maximum.
(2) f ( x ) = (x - a ) 3 + c
The first two derivatives are
df
dx d2 f
-j' = 3(x - a) = 6 [ x - a)
(a)
Since both derivatives vanish at x = a, we must consider the third derivative:
dx 3 6
The stationary point, x = a, is neither a relative minimum nor a relative maximum since the third derivative is finite. We could have also established this result by considering the expression for the slope. We see from (a) that the slope is positive on b o th sides of x = a. The general shape o f this function is shown in Fig. E 3-1.
68
RELATIVE EXTREMA FOR A FUNCTION
CHAP. 3 Fig. E 3 -1
/CO
The sufficient condition for a stationary value to be a relative extrem um (relative m inim um (maximum) when d2 f ) d x 2 > 0 (< 0)) follows from a con sideration o f the geom etry of th e /(x ) vs. x curve in the vicinity of the stationary point. We can also establish the criteria for a relative extrem um from the Taylor series expansion of f{x). Since this approach can be readily extended to func tions of m ore than one independent variable we will describe it in detail. Suppose we know the value o f/(x ) at x = a and we w ant f ( a + Ax) where Ax is some increm ent in x. If the first n + 1 derivatives o f/(x ) are continuous in the interval, a < x < a + Ax, we can express f ( a + Ax) as f{ a + Ax) - f(a)
L,
} ___ d"+lM ) {n + 1)! dxn+l
where djf(a)/dxj denotes the jth derivative o f/(x ) evaluated at x ~ a, and the rem ainder R is given by
where c, is an unknow n num ber between a and a + Ax. E quation (3 -1 ) is called the Taylor series expansion* of /(x ) about x a. If f ( x ) is an nth-degree polynom ial, the (n 4- l)th derivative vanishes for all x and the expansion will yield the exact value of f ( a + Ax) when n term s are retained. In all other cases, there will be some error, represented by R, due to truncating the series at n terms. Since R n depends on I, we can only establish bounds on R n. The fol lowing example illustrates this point.
* See Ref. 1, Article 16-8.
SEC. 3 -1 . RELATIVE EXTREMA FOR A FUNCTION OF ONE VA R IA B LE
69
Example 3 -2
We expand sin x in a Taylor series about x ~ 0 taking n - 2. Using (3-1) and (3-2), and noting that a 0, we obtain sin Ax = Ax + R z (a)
R2 = ------ cos 6
The bounds on [J?2| are
(Ax)3
c,
0 ^
^ Ax
(b)
|Axi3 . . lAxI3 i - cos Ax < j/?2| < 1
(c)
If we use (a) to find sin (0.2), the upper bound on the truncation error is (0.2)3/6 ~ 0.0013.
If Ax is sm all with respect to unity, the first term on the right-hand side of (3-1) is the d o m inant term in the expansion. Also, the second term is m ore significant th an the third, f o u rth ,. . . , nth terms. W e refer to dj/dx Ax as the first-order increm ent in /(x ) due to the increm ent. Ax. Similarly, we call %d2 f / d x 2{Ax)2 the second-order increment, and so on. N ow , f(a) is a relative m inim um when f(ci + Ax) f{a) > 0 for all points in the neighborhood of x = a, th a t is, for all finite values of Ax in some interval, r\ ^ Ax ^ s, where tj and e are arb itrary small positive num bers. C onsidering Ax to be small, the first-order increm ent dom inates and we can write f ( a + Ax) f(a) df(a) Ax + (second- and higher-order terms) (3-3)
F o r f ( a + Ax) f{a) to be positive for both positive an d negative values of Ax, the first order increm ent m ust vanish, th at is, df{d)jdx m ust vanish. N o te that this is a necessary but not sufficient condition for a relative minimum . If the first-order increm ent vanishes, the second-order increm ent will dom inate: f ( a + Ax) f(a) = - r T ^ ) 2 + (th ird -a n d higher-order terms) 2 ax (3-4)
It follows from (3-4) th a t the sccond-order increm ent m ust be positive for f ( a + Ax) - f(a) > 0 to be satisfied. This requires d2 f( a )/ d x 2 > 0. Finally, the necessary and sufficient conditions for a relative m inim um a t x = a are m
dx
= 0
dx
> 0
(3 -5 )
If the first two derivatives vanish at x = a, the third -o rd er increm ent is now the dom inant term in the expansion. f ( a -f- Ax) + f(a) 7 O uX (Ax)3 + (fourth- and higher-order terms) (3-6)
Since the third-order increment depends on the sign of Ax, it m ust vanish for
70
RELATIVE EXTR E M A FOR A FUNCTION
CHAP. 3
f(a ) to be a relative extrem um . T he sufficient co n d itio n s for this case are as follow s: Relative Minimum = 0 d*f dx4 > 0 (3 -7 ) R elative M aximum
dx3
l
dx3
= 0
J4 / rfx < 0
T he n o ta tio n used in the T ay lo r series expansion o f/(x ) becom es som ew hat cum bersom e for m o re th an one variable. In w hat follows, we in tro d u ce new n o ta tio n which can be readily extended to the case o f n variables. F irst, we define A / to be th e total increm ent in f ( x ) due to the increm ent, Ax. A / = f ( x + Ax) - f ( x ) (3 -8 )
T his increm ent depends on Ax as well as x. N ext, we define the differential o p erato r, d, as. d( ) = d P- Ax dx (3 -9 )
T h e result o f o p eratin g on /( x ) w ith d is called th e first differential an d is den o ted by d f : df ~ ~ dx A x d f( x , Ax) (3 -1 0 )
T h e first differential o f/(x ) is a function o f two in d ep en d en t variables, nam ely, x an d Ax. I f /( x ) = x, then d f /dx 1 and d f = dx Ax (3 -1 1 )
O ne can use dx and Ax in terchangeably; how ever, we will use Ax ra th e r th a n dx. H igher differentials o f f ( x ) are defined by iteration. F o r exam ple, the second differential is given by d (df dx \ d x Since Ax is independent o f x, N Xy Ax (3 -1 2 )
fx
an d d 2 f reduces to d2 f(x) d2f = dx1
(A x ) =
(Ax)2 = d 2 f ( x , Ax)
(3-13)
In form ing the higher differentials, we take d(Ax) 0.
SEC. 3 - 2 .
FU NC TIO N OF n IN D EPEN DEN T V A R IA B L E S
71
U sing differential n o ta tio n , th e T ay lo r series ex p an sio n (3 -1 ) a b o u t x can be w ritten as A / - df + \ d 2 f + .. + ! < / / + R n 2 n! (3 -1 4 )
T he first differential represents the first-o rd er in crem en t in f ( x ) due to the increm ent, Ax. Sim ilarly, the second differential is a m easu re o f th e secondord er increm ent, a n d so on. T h e n ,/(x ) is a sta tio n a ry value w hen d f = 0 for all perm issible values o f Ax. Also, th e sta tio n a ry p o in t is a relative m inim um (m axim um ) w hen d 2 f > 0 ( < 0 ) for all perm issible values o f Ax. T h e above criteria reduce to (3 -5 ) w hen th e differentials are expressed in term s o f the derivatives. Rules for form ing the differential o f th e sum o r p ro d u c t o f functions arc listed below for reference. P ro b lem s 3 - 4 th ro u g h 3 - 7 illu strate their application. / = u(x) + v(x) d f = du + dv d2 f d(df) = d 2u + d 2v f = u(x)v(x) d f = u d v + v du d2 f = u d2v 4 - 2 du dv + v d 2u (3 -1 6 ) (3 -1 5 )
d f = % dy + f / y
(3 -1 7 )
3-2.
RELATIVE EXTREMA FOR A FUNCTION OF n INDEPENDENT VARIABLES
Let f ( x j, x 2, . . . , x j be a co n tin u o u s function o f n in d ep en d en t variables (Xi, x 2, . . . , x j . W e define A / as the to tal in crem en t in / due to increm ents in the in d ep en d en t variables (A xl5 Ax 2, . . . , Ax): Af = f i x i + A x ^ x 2 + A x2, . . . , x + Ax) - f ( x u x 2, . . . , x) (3 -1 8 )
If A / > 0 ( < 0 ) for all p o in ts in th e n e ig h b o rh o o d o f (x l5 x 2, . . . , x), we say that / ( x , , x 2, . . . , x) is a relative minimum (maximum). W e establish criteria for a relative ex trem u m by exp an d in g / in an n -dim ensional T ay lo r series. T he procedure is identical to th a t followed in th e on e-d im en sio n al case. A ctually, we ju s t have to ex ten d the differential n o ta tio n from one to n dim ensions.
72
CHAP. 3
We define the rc-dimensional differential operator as d = A xL + ~ Ax2 + + y - Ax = f . dx2 dx ^ Y Axi vXj (3~ 19)
where the increments (Axj, Ax2 Ax) are independent of (x1? x 2, . . . , x). The result obtained when d is applied to / is called the first differential and w ritten as df. t f - f g j A * , (3-20)
H igher differentials are defined by iteration. For example, the second differen tial has the form d2 f - W ) = i ~ ( i A * ,) Ax, (3-21)
Since Ax, are considered to be independent, (3-21) reduces to " d2 f = Now, we let < > = m
f(2) _
n <?Y Ax Ax* j - i oxjdxk J
(3-22)
_5xj-5xk Ax {Axj}
k = 1, 2 , . . . , n
(3-23)
and the expressions for the first two differentials simplify to d f - AxTfU) d2 f = Axr f(2) Ax ,3 _ 24)
The Taylor series expansion for / ab o u t (xj, x 2, , x), when expressed in term s o f differentials, has the form & f= df + l d 2 f+ --- + ^ d n f+ R 2 u! (3-25)
We say th a t / ( x x, x 2, . . . , x) is stationary when d f 0 for arbitrary Ax. This requirem ent is satisfied only when r 1' = 0 E quation (3-26) represents n scalar equations, namely, ^ 0
OXj
(3-26)
j 1 , 2, , n
(3-27)
The scalar equations corresponding to the stationary requirement are usually
SEC. 3 -2 .
FUNCTION OF n INDEPENDENT VAR IABLES
73
called the Euler equations for / . N ote that the num ber of equations is equal to the num ber of independent variables. A stationary point corresponds to a relative m inim um (maximum) of / when d2 f is positive (negative) definite. It is called a neutral point when d 2 f is either positive or negative semidefinite and a saddle point when d 2 f is indif ferent, i.e., the eigenvalues are bo th positive and negative. This term inology was originally introduced for the two dim ensional case where it has geom etri cal significance. To summarize, the solutions o f the Euler equations correspond to points at which / is stationary. The classification of a stationary point is determ ined by the character (definite, semidefinite, indifferent) of f< 2 > evaluated at the point. We are interested in the extrem um problem since it is closely related to the stability problem . The extrem um problem is also related to certain other p ro b lems of interest, e.g., the characteristic-value problem . In the following exam ples, we illustrate various special forms of / which are encountered in m ember system analysis. ---------- ------------------------------------ ------------------ Example 3 -3 ---------
/ = f ( y .V 2 , > yA yj - yj(x i, x2, . . . , x) 'd f - / " df dy/) df = X, - Axk = Z ( I v ~ I k~ 1 U X -k k- I \/= 1 dyj 0Xk;
Now,
dyj * I ~ A.\t k=l dxy

It follows that
' d f
df = E 7 , *yj
i= i
Repeating leads to
'd f r
d 2y j + X d 2f dym dys
d2 f
= I
j=
j -
o.Vj
m- 1oym dyj
Consider the double sum,
S 5
/ = Z I
UjWJkh\
(a)
j = 1 fc= 1
The first differential (see Prob. 3-9) has the form

s s
df =
j**1 fc- 1
Z (duj wjkvk + Uj d w jkvk + Ujwjk dvk)
(b)
Introducing matrix notation,
U = {;}
and letting
W = |> ;J f = ut wy c /u = {duj}
V = {V kj (c)
(d)
74
RELATIVE EXTREM A FOR A FUNCTION
CHAP. 3
and so forth, we can write d f as

d f = d( uT wv)
d u Tw v + uT d w v + uT w d s One operates on matrix products as if they were scalars, but the order must be preserved. As an illustration, consider / = \ x 'ax x 'c (f) where a, c are constant and a is symmetrical. Noting that d a = dc = 0 and d x e Ax, the first two differentials are d f = AxT(ax c)
o,
Comparing (g) and (3-24), we see that
= Ax a Ax
A T .
(gj
f (l) = ax - c
r> = .
The Euler equations are obtained by setting
f (1)
.
equal to 0:
(h)
ax = c
(i)
The solution of (i) corresponds to a stationary value of (f). If a is positive definite, the stationary point is a relative minimum. One can visualize the problem of solving the system ax = c, where a is symmetrical from the point of view of finding the stationary value of a second-degree polynomial having the form / = -|x3ax x rc.
(3)
Suppose / = u/v. Using the fact that
d
dxj
fu \
1 du
>Pxj + 1 / du
v \d.X:
U
u
/1 \ (a)
<xi W
v dx.-J
we can write
d f = d ( - I - (du I dv) v! V
fu\
(b)
We apply (b) to
x ax XX where a is symmetrical, and obtain (see Prob. 3 -5 )

dX = ~ (ax ax)
(c)
Axr
XX
(d)
d 2X = ~
xx
(Axra Ax - X A x T Ax - 2 dX Axrx)
Setting dX = 0 leads to the Euler equations for (c), ax - Xx 0 which we recognize as the symmetrical characteristic-value problem. (e)
SEC. 3 - 3 .
L A G R A N G E M U LT IP LIE R S
75
q u o tie n t.
The quotient x ra x /x rx, where x is arbitrary and a is symmetrical, is called R a y le i g h 's We have shown that the characteristic values of a are stationary values of Rayleighs quotient. This property can be used to improve an initial estimate for a characteristic value. For a more detailed discussion, see Ref. 6 and Prob. 3-11.
3-3.
LAGRANGE MULTIPLIERS
U p to this p o in t, we have considered only th e case w here the function is expressed in term s o f in d ep en d en t variables. In w h at follows, we discuss how one can m odify th e p ro ced u re to h an d le the case w here som e o f the variables are n o t indep en d en t. T his m odification is conveniently effected using L agrange m ultipliers. Suppose / is expressed in term s of n variables, say x l7 x 2, . . . , x, som e o f which are n o t in dependent. T h e general statio n ary req u irem en t is (3 -2 8 ) for all a rb itra ry differentials o f th e independent variables. W e use dxj instead of A Xj to em phasize th a t som e o f the variables are d ep endent. In o rd er to establish the E u ler eq u atio n s, we m u st express d f in term s of the differentials of the in d ep en d en t variables. N ow , we suppose th ere are r relatio n s betw een the variables, o f th e form gk( x l t x 2, = 0 k = 1,2 , . . . , r (3 -2 9 )
O ne can co n sid er these relatio n s as c o n strain t co n d itio n s o n th e variables. Actually, th ere a re only n r in d ep en d en t variables. W e o b ta in r relatio n s between the n differentials by o p eratin g on (3-29). Since gk 0, it follows that dgk 0. Then, dgk = Z d xj = 0 k = 1, 2 , . . . , r (3 -3 0 )
U sing (3-30), we can express r differentials in term s o f th e rem aining n r differentials. Finally, we reduce (3 -28) to a sum involving th e n r in d ep en dent differentials. E q u atin g th e coefficients to zero leads to a system o f n r equations w hich, to g eth er w ith th e r c o n strain t eq u atio n s, are sufficient to determ ine th e s ta tio n a ry points. Example 3 - 4 --------------------------------------------------------------------------------------------We illustrate the procedure for n 2 and r = 1: / = / t a , x 2) g { x u x 2) = 0 The first variation is
(a)
76
CHAP. 3
Operating on g ( x t, x 2) we have
dg =
------ d x y
dg
H----------- d x 2 = 0
dg
dxl
cx-j
(b)
Now, we suppose d g /d x 2 5= 0. Solving (b) for d x 2 (we replace d x , by Ax 1 to emphasize that x , is the independent variable.) , and substituting in (a), we obtain
df df =
OX
(og
Idg\
(C )
( dg j dg \ d f ' Px ii d x j d x 2
Axj
(d)
Finally, the equations defining the stationary points are

df_ _ ( d g dxi i d \ df_ =
0 (e)
\ d x t j d x 2) o x 2 9 (x j, x 2) = 0
To determine whether a stationary point actually corresponds to a relative extremum, we must investigate the behavior of the sccond differential. The general form of d2f for a function of two variables (which are not necessarily independent) is
d2 f = d ( ~ dx, + j - dx2 \ \d x t dx2 J
^ z d2f 2 df = y v - d x , d x k + ^ r - d 2x J : 1 dxjdxi, J-i dxj
(f)
We reduce (f) to a quadratic form in the independent differential. Axb using (c), and noting < /2x, = 0, , 32 / / flu flu \ 5 / ~d2 f , , J '/ . u2 + I ----+ 11----------(g) d2 f = (Ax,)2 d x l \ d x t d x 2J d x 2 _3xf Ud x v<jx2 where
h
dg I dg
=
d x j dx2
The character of the stationary point is determined from the sign of the bracketed term.
An automatic procedure for handling constraint conditions involves the use of Lagrange multipliers. We first describe this procedure for the case of two variables and then generalize it for n variables and r restraints. The problem consists in determining the stationary values of f { x : , x 2) subject to the con straint condition, g(xy, x 2) = 0. We introduce the function H, defined by H ( x {, x 2, X) - f { x u x 2) + Xg(xu x 2) (3-31)
where X is an unknown parameter, referred to as a Lagrange multiplier. We
SEC. 3 -3 .
LAGRANGE MULTIPLIERS
77
consider x 2 and X to be independent variables, and require H to be sta tionary. The Euler equations for I I are 3H dxt
CX2
d f t ^ dg _ Q dxy cx! =
0X2
+ X~- = Q
0X2
(3-32)
-pj- = g(x i , x 2) = 0 We suppose dg/dxz ^ 0. Then, solving the second equation in (3-32) for X, and substituting in the first equation, we obtain X= - dx2j ox2 and
OX
(3-33)
= . \v CXt j dx2J o x 2 g{x 1; x 2) = 0
(3-34)
Equations (3-34) and (e) of the previous example are identical. We see that the Euler equations for II are the stationary conditions for / including the effect of constraints. Example 3 -5 --------------------------------------------------------------------- ^---------------f r- 3.Xj + 2x2 g = Xi - X 2 = 0
2xj - lx 2
We form H = / + Xg,
H - 3 x f + 2 %l + 2x, + 7x2 +
x 2)
The stationary requirement for H treating x 1; x 2 , and X as independent variables is 6X| -i- 2 4- X 0
4x2 + 7 - X = 0
X! ~ x 2 = 0 Solving this system for Xj, x 2 and X we obtain

X = 4xz + 7
Xj = x 2 = 9/10
This procedure can be readily generalized to the r constraints. The problem consists of determining f ( x u x 2, . . ., x n), subject to the constraints gk(xu k = 1, 2 , . . . , r. There will be r Lagrange multipliers
case of n variables and the stationary values of x 2, . . . , x) = 0, where for this case, and H has
78
R ELATIV E E X TR E M A FOR A FUNCTION
CHAP. 3
the form
r
H = / +
Kdk = # ( x 'i, X2, . , . , x, Ay, a 2, . . . , / r)
(3 -3 5 )
T he E uler eq u atio n s for H are (3 -3 6 ) (3 -3 7 ) W e first solve r equations in (3 -3 6 ) for the r L agrange m ultipliers, an d then determ ine the n co ordinates of the statio n ary points from the rem aining n r equ atio n s in (3-3 6 ) an d the r co n strain t eq u atio n s (3-37). T h e use of L ag rang e m ultipliers to in tro d u ce co n strain t con d itio n s usually reduces the am o u n t o f algebra.
REFERENCES
1.
2. C a lc u lu s a n d A n a l y t i c a l G e o m e t r y , Addison-Wesley Publishing Co., Reading, Mass., 1953. C o u ran t; R., Differential and Integral Calculus, Vol. 1, Blackie, London, 1937. C o u r a n t , R., Differential and Integral Calculus, Vol. 2 , Interscience Publishers, New York, 1936. Hancock, H., T h e o r y o f M a x i m a a n d M i n i m a , Dover Publications, New York, I960. Apostol, T. M., M a t h e m a t i c a l A n a l y s i s , Addison-Wesley Publishing Co., Reading, Mass., 1957. C ra n d a ll, S. H., E n g in e e r in g A n a l y s i s , McGraw-Hill, New York, 1956. H ildebrand, F. B., M e t h o d s o f A p p l i e d M a t h e m a t i c s , Prentice-Hall, New York, 1952.
Thomas, G. B., Jr.,
3. 4. 5. 6. 7.
PROBLEMS
3 - 1 . D eterm ine th e relative extrem a for (a) f ( x ) = 2 x 2 -f Ax + 5 (b) f ( x ) 2 x 2 + 8x + 10 (c) f ( x ) a x 2 + 26.x T c (d) f(x) = x 3 + 2 x 2 + x + 10 (e) f ( x ) = ^ x 3 -f- 2 x 2 + 4x + 15 (0 f ix) = (x - d f -h (x - a)2 (g) f ( x ) = Jax3 + j b x 2 + cx + d 3 - 2 . E x p an d cos x in a T ay lo r series a b o u t x = 0, ta k in g s = 3. D eterm ine the u p p er an d low er b o u n d s o n R 3. 3 - 3 . E x p a n d (l + x )1/2 in a T ay lo r series a b o u t x O takingw = 2. D e te r m ine u p p er an d low er b o u n d s o n R 2. 3 - 4 . F in d d f an d d2 f for (a) / = x 2 + 2x + 5 (b) / = 3x3 + 2 x 2 + 5x + 6 (c) / = x 2 sin x (d) / = cos y w here y x 3
PR O BLEM S
79
3 -5 .
Let / = h(x)/ u(x ). Show th a t d f - (du f dv ) ' v d2 f = - (d 2u f d 2v) 2 v v
3 - 6 . Let til5 u2, u3 be functions of x an d / = f ( u 1; w2, %). D eterm ine df. 3 -7 . Suppose / = u(x)w(y) w here y = y(x). D eterm ine expressions for d f and d2 f . A pply to (a) u x 3 x (b) w = cos y (c) y = x 2 3 -8 . Find th e first tw o differentials for th e following functions: (a) / = x j + 3 x f x j + x y x f (b) / = 3 x f + 6 x xx z + 9 x | + 5x, 4 x ? 3 -9 . C onsider f ud, w here = ub u ,V 2) and yi = Show th a t rf/ (sf(uy) u d v + v du d2 f ud2v + 2 du dv + vd2u * 2) y 2 v'2(-x i' x2) v = u(yj, y 2)
N ote th a t the rule for form ing the differential o f a product is independent of w hether the term s are functions o f the independent variables (x x, x 2) o r of dependent variables. 3 -1 0 . Classify the statio n ary points for the following functions: (a) / - 3x] + 3x2 - 9 x t + I2 x 2 10 (b) / = 3x \ + 6x t x 2 + 2x2 + 2 x 1 + l x 2 (c) / = 3xf + 6x 1x 2 + 3x1 + 2xj + 2x2 (d) / = 3 x | 4- 6x t x 2 + 4x2 + 2 x t + 7 x 2 (e) / = 3x'j + 6x^x2 + 3x2 3x! 3 -1 1 . C onsider R ayleighs quotient, . x T ax xT 'a T x A j Y ' XX XX where x is arb itrary . Since a is sym m etrical, its characteristic vectors are linearly independent an d we can express x as
x = Zi
where Q j (j = 1, 2 , . . . , n) are the norm alized characteristic vectors for a. (a) Show th a t
z
l = J - n -------
j= l
80
R ELATIVE E X TR E M A FOR A FUNCTION
CHAP. 3
(b) (c)
Suppose x differs only slightly from Q k. Then, N N Specialize (a) for this case. H i n t : F a c to r o u t Xk an d c% . U se (b) to obtain an im proved estim ate for X. 3 1
for j ^ k.
x ^ {1, 3} T he exact result is X= 1 x - {1, - 2 } 3 -1 2 . U sing Lagrange m ultipliers, determ ine the statio n ary values for the following constrained functions: (a) / = x f - x \ g = x\ + x, = 0 (b) f = x f + x \ + x i Qi = x L + x 2 + x 3 - 1 = 0 92 x i ~~ x 2 + 2x3 + 2 = 0 3 -1 3 . C onsider the problem of finding the statio n ary values o f / = x Ta x = x Ta Tx subject to the co n strain t condition, x 7x 1. U sing (3 -3 6 ) we w rite H = / + Xg ~ x Tax A(xTx 1) (a) Show th a t the equations defining the statio n ary points o f / are a x = Xx (b) xrx = 1
R elate this problem to the characteristic value problem for a sym m etri cal m atrix. 3 -1 4 . Suppose / = x Tx an d g = I x r a x 0 w here a r = a. Show th a t the E uler equations for H have the form
W e see th a t the L agrange m ultipliers are the reciprocals of the characteristic values of a. H ow are the m ultipliers related to the statio n ary values of / ?
4
T he geom etry of a m em ber elem ent is defined on ce the curve co rresponding to the reference axis an d the p roperties o f the n o rm al cross section (such as area, m om en ts o f inertia, etc.) are specified. In this chapter, we first discuss the differential geom etry o f a space curve in co nsiderable detail an d then extend the results to a m em ber element. O u r prim ary objective is to introduce the concept o f a local reference fram e for a m em ber.
4-1.
PARAMETRIC REPRESENTATION OF A SPACE CURVE
A curve is defined as the locus o f points w hose position vector* is a function of a single p aram eter. W e take an o rth o g o n al cartesian reference fram e having directions X u X 2, and X 3 (see Fig. 4 -1 ). Let f be the p o sitio n vector to a p oint *3
F ig . 4 - 1 . C a rte s ia n re fe re n c e fra m e w ith p o s itio n v e c to r r(y).
* The vector directed from the origin of a fixed reference frame to a point is called the position
vector. A knowledge of vectors is assumed. For a review, see Ref. 1.

81
80
R ELATIVE EXTREM A FOR A FUNCTION
CHAP. 3
(b) (c)
Suppose x differs only slightly from Q fc. Then, |c,| |c*| for j ^ k. Specialize (a) for this case. Hint: F a c to r o u t Xk and c l . U se (b) to o b tain an im proved estim ate for X. a =
X {1, - 3 } T he exact result is A= 1 x = {1, - 2 } 3 -1 2 . U sing L agrange m ultipliers, determ ine the statio n ary values for the following con strain ed functions: (a) f = x j - x j g = xj + x2 = 0 (b) / = x j + x \ + x j g1 = Xj + x2 + x3 - 1 = 0 . , g2 ~ x 1 ~ x 2 + 2 x 3 + 2 = 0 3 -1 3 . C onsider the problem of finding the statio n ary values o f / x r a x = x r a r x subject to th e co n strain t condition, x r x = 1. U sing (3 -3 6 ) we w rite H = / + Xg = x r ax A(xr x 1) (a) Show th a t the equations defining the statio n ary p o in ts o f / are a x Ax (b) xrx = 1
L 1t
~3
1" i
R elate this problem to the characteristic value p roblem for a sym m etri cal m atrix. 3 -1 4 . S u p p o s e / = x r x a n d g = 1 x Tax = 0 w here a 7 = a. Show th a t the Euler eq u atio n s for H have the form 1x ax = -r
A
x r ax = , 1
W e see th a t the L agrange m ultipliers are the reciprocals o f th e characteristic values of a. H ow are the m ultipliers related to th e statio n ary values o f / ?
4
T he g eo m etry o f a m em b er elem ent is defined once th e curve co rresp o n d in g to the reference axis an d th e p ro p erties o f th e n o rm al cross section (such as area, m om en ts o f inertia, etc.) are specified. In this chapter, we first discuss the differential geom etry o f a space curve in co n sid erab le detail a n d then ex ten d the results to a m em b er elem ent. O u r p rim ary objective is to in tro d u ce the concept o f a local reference fram e for a m em ber.
4-1.
PARAMETRIC REPRESENTATION OF A SPACE CURVE
A curve is defined as th e locus o f p o in ts w hose p o sitio n vector* is a function of a single p aram eter. W e tak e an o rth o g o n al cartcsian reference fram e h aving directions X u X 2, an d X 3 (see Fig. 4 -1 ). Let f be the p o sitio n vector to a p o in t
Fig. 4-1. Cartesian reference frame with position vector f(y).

* The vector directed from the origin of a fixed reference frame to a point is called the position A knowledge of vectors is assumed. For a review, see Ref. 1.
81
vector .
82
D IFFER EN TIAL G E O M E TR Y OF A M E M B E R E LE M E N T
CHAP. 4
on the curve having co o rd in ates Xj (j 1, 2, 3) an d let j: be th e param eter. W e can represent the curve by r = r(y) (4 -1 )
3
Since r =
xjip an altern ate rep resen tatio n is xj = xj(y) (j = 1, 2, 3) (4 -2 )
B oth form s are called the p aram etric rep resen tatio n of a space curve.
Example 4-1 --------------------------------------------------------------------------------------(1) Consider a circle in the X i - X 2 plane (Fig. E 4-1A ). We take y as the polar angle and let a |r|. The coordinates are
xj
and
r
a cos y
x 2 = a sin y a c o s y i + a sin y l 2
(2) Consider the curve (Fig. E 4-1B ) defined by = a cos y

x2 = b sin y
(4-3)
x3 = cv
where a, b, c are constants. The projection on the X , - X 2 plane is an ellipse having semiaxes a and b. The position vector for this curve has the form
r = a cos y? j + b sin y i 2 + cyf3
4 -2 .
ARC LENGTH
F igure 4 - 2 show s tw o n eighboring points, P an d Q, co rresp o n d in g to y an d Ay. T h e cartesian co o rd in ates are X j an d x j + Ax j ( j 1, 2, 3) an d the length o f the chord from P to Q is given by
y +
I fe l2 =
j= 1
(A x ,)2
(a)
As Ay -> 0, the chord length \PQ\ ap p ro ach es th e arc length, As. In th e limit,
ds2 Yj dxf
j= i N o tin g th a t
dx; -
dy
dy y
(c)
we can express
ds
as /d x ^ 2 + 'd x ^ 2 + ' dx-xY

dy
1/2
d s " +
dy
(4 -4 )
*2
Fig. E4-1A
*3
Fig. 4-2. Differential segment of a curve.

83
84
D IFFER EN TIAL G E O M E TR Y OF A M E M B E R E LEM EN T
CH AP. 4
Finally, integrating (4 -4 ) leads to > ' f/V fv dx , .\2 \z yo + / J x 2Y { d x 3\ 2 1/2 (4 -5 )
W e have defined <is such th a t 5 increases w ith increasing y. It is cu sto m ary to call th e sense of increasing 5 the positive sense o f the curve. T o simplify the expressions, we let a +
JM
5 f a dy J yo
1/2
(4 -6 )
Then, the previous eq u atio n s reduce to ds a dy (4 -7 )
O ne can visualize a as a scale factor which converts dy in to ds. N o te th a t a > 0. Also, if we take y = s, then a = + 1 .
Example 4 -2
Consider the curve defined by (4-3). Using (4-6), the scale factor is a = [a2 sin2 y + b2 cos2 y + c2] 1/2 We suppose that b > a. One can always orient the axes such that this condition is satisfied. Then, we express a as a = (b2 + c2)112 [1 - k 2 sin2 y ]1/2 where k2 = The arc length is given by s J a dy (b2 + c2)i,z I [1 k2 sin2 v]1/2 dv
Jo
b2 - a2 b 2 + c2
Jo L
The integral for s is called an elliptic integral of the second kind and denoted by {k, y). Then, s = (b2 + c2)112 E(k, y) Tables for E[k, y) as a function of k and y are contained in Ref. 3. When b = a, the curve is called a circular helix and the relations reduce to a = (a2 + c2)1'2 5 = const. s = ay
SEC. 4 - 3 .
U N IT T A N G E N T VECTO R
85
4-3.
UNIT TANGENT VECTOR
W e co n sid er again th e n eig h b o rin g po in ts, P(y) a n d Q{y + Ay), show n in F ig u re 4 - 3 . T h e c o rresp o n d in g p o sitio n vectors are ? ( v), r( y + Ay), and PQ = f(y + Ay) r{y) = A f (a)
As Ay -> 0, PQ a p p ro a c h e s th e ta n g e n t to th e curve a t P. T hen, th e u n it ta n g en t vector a t P is given by* PQ df nv t lim (4 -8 ) A,->0jpg| ds U sing the chain rule, we can express t as dr ds dr dy dy ds 1 dr a dy (4 -9 )
Since o c > 0, t alw ays p o in ts in th e positive d irectio n o f th e curve, th a t is, in th e direction o f increasing s (o r y). It follow s th a t df/ dy is also a ta n g e n t v ecto r an d dr dy fd f d f \ l!2 T y 'T y ) (4 -1 0 ) a
E q u atio n (4 -1 0 ) reduces to (4 -6 ) w hen r is expressed in term s o f cartesian coordinates.
Fig. 4 -3 . Unit tangent vector at P{y).
* See Ref. 1, p. 401.
86
DIFFERENTIAL GEOMETRY OF A MEMBER ELEMENT
CHAP. 4
Example 4 -3 ----------------------------------------------------------------------------------
We determine the tangent vector for the curve defined by (4-3). The position vector is
f = a cos > T( + b sin y i 2 + c y i 3
Differentiating f with respect to y,

dr . _ = asm y i i + b cos y r 2 + c i 3 dy
and using (4-9) and (4-10), we obtain

a + [ a 2 sin2 y + b 1 c o s 2 y + c2] m i =
- [ a sin yTj + b cos yT1 + rt3] a
When a = b, a = [a2 + c2] l/2 = const, and the angle between the tgngent and the X s direction is constant. A space curve having the property that the angle between the tangent and a fixed direction ( X 3 direction for this example) is constant is called a helix.*
4-4.
PRINCIPAL NORMAL AND BINORMAL VECTORS
Differentiating t * I = 1 with respect to y, we have t dt = 0 dy

(a)
It follows from (a) that d t / d y is o r t h o g o n a l to I. The unit vector pointing in the direction of d t / d y is called the p r i n c i p a l n o r m a l v e c t o r and is usually denoted by h. n
1 dl
dl: dy dy (4-11)
where dt _ d_ / I ctf dy d y \ a dy The binormal vector,

b,
is defined by
b = i x h
(4-12)
We see that b is also a unit vector and the three vectors, i, n, b comprise a right handed mutually orthogonal system of unit vectors at a point on the curve. N ote that the vectors are uniquely defined once ? ;{y) is specified. The frame associated with t, b and n is called the moving trihedron and the planes deter mined by ( I, n), (h, b) and (b, t) are referred to as the osculating normal, and rectifying planes (see Fig. 4-4).
* See Ref. 4, Chap. 1.
SEC. 4 -4 .
PRINCIPAL NORMAL AND BINORMAL VECTORS
87
Fig.
4-4. Definition of local planes.
Example 4-4
We determine h and h for the circular helix. Wc have already found that
211/2 [*2 + c2J

and
t = - [ a sin vfj 4- a cos y l 2 + cj3]
Differentiating t with rcspect to y, we obtain

dt dy a
[cos y i i + sin y *x]
Then, 1 dt n -----= cos i>ii sin y i 2 dt d y

dy
The principal normal vector is parallel to the X r X 2 plane and points in the inward radial direction. It follows that the rectifying plane is orthogonal to the X i - X 2 plane. We can determine b using the expansion lor the vector product.
t-
a sin y cos y This reduces to -
a cos y
sin > 0
c , c a b - sin y i! ---- cos y l 2 H f3
The unit vectors are shown in Fig. E4-4.
88
CHAP. 4
Fig. E 4-4
*3
*2
4-5.
CURVATURE, TORSION, AND THE FRENET EQUATIONS
The derivative o f the tangent vcctor with respect to arc length is called the curvature vector, K. di d2r (4-13)
d2r K = ds2 Using (4-11), we can write K
1 di a dy
di - = Kn ds
(4-14)
N ote that K points in the same direction as h since we have taken K > 0. The curvature has the dim ension L ~ 1 and is a m easure of the variation of the tangent vector with arc length. We let R be the reciprocal of the curvature: R = K (4-15)
In the case of a plane curve, R is the radius of the circle passing through three consecutive points* on the curve, and K \d9/ds\ where 0 is the angle between i and To show this, we express t in terms of 6 and then differentiate with respect to 5. F rom Fig. 4 -5 , we have t = cos 0 i y + sin 9 i 2
* See Ref. 4, p. 14, for a d iscu ssion o f the term in ology thcee c o n secu tiv e p oin ts.
SEC. 4 -5 .
CURVATURE, TORSION, AND THE FRENET EQUATIONS
89
Then K - [ sin 61, -f cos 0 i2] and dfj ds n R d6/ds [ - s i n 011 + cos 0i2] \d9/ds\
de
ds
In the case of a space curve, the tangents at two consecutive points, say P and Q, are in the osculating plane at P, that is, the plane determined by I and n at P. We can interpret R as the radius of the osculating circle at P. It should be noted that the osculating plane will generally vary along the curve.
>2
Xi
Fig. 4 -5 . Radius of cu rva ture fo r a plane curve.
The binorm al vector is norm al to both i and ti and therefore is norm al to the osculating plane. A measure of the variation of the osculating plane is given by db/ds. Since b is a unit vector, db/ds is orthogonal to b. To determine whether db/ds involves i, we differentiate the orthogonality condition - 6 = 0, with respect to s. _ db - dt 1 ds ds But dt/ds Kn and b ri 0. Then, db/ds is also orthogonal to t. and involves only h. We express db/ds as db --- = ~xn (4-16)
US
where t is called the torsion and has the dimension, L ~ l .
90
CHAP. 4
It rem ains to develop an expression for x. Now, h is defined by b t x n Differentiating with respect to s, we have db Ts This reduces to db ds ~ f dn ds dt - , r v 'els X n ^ ds
since n x n = 0. Finally, using (4-16), the torsion is given by _ _ dn 1 r dn x = n ' t x b ds a ay (4-17)
N ote th at x can be positive or negative whereas K is always positive, according to our definition. The torsion is zero for a plane curve since the osculating plane coincides with the plane of the curve and b is constant. Example 4 -5 ---------------------------------------------------------------------------------------The unit vectors for a circular helix are
- =1- I r - -vj i + a cos vi2 - + c!3J - i t asm a

n cos
sm yf2
b = - [c sin v?i c cos y l 2 +
where Then,
a = (a 2 + c2)II2
a2 and 1 r dn t = ~o o c rfy
c
a 2 + c2
c
const
= 5 -------- = const a a 4- c
We have developed expressions for the rate of change of the tangent and binorm al vectors. T o complete the discussion, we consider the rate of change of the principal norm al vector with respect to arc length. Since h is a unit vector, dn/ds is orthogonal to h. From (4-17),
SEC. 4 -6 .
GEOMETRICAL RELATIONS FOR A SPACE CURVE
91
To determine the com ponent of dfi/ds in the t direction, we differentiate the orthogonality relation, t h 0. (b) It follows from (a) and (b) that dn ds Kt + rb (4-18)
The differentiation formulas for t, h, and b are called the Frenet equations. 4-6. SUMMARY OF THE GEOMETRICAL RELATIONS FOR A SPACE CURVE
We summarize the geometrical relations for a space curve: Orthogonal Unit Vectors dr 1 dr t = = ----- = tangent vector as a dy . 1 dt . . n = principal norm al vector dt dy dy b = t x n ---- binorm al vector a df dy ds dy
(4-19)
Differentiation Formulas (Frenet Equations) dt ds db ds dn 1 dt a dy = Kn
a dy 1 dn - K t -f zb a dy ds I Jd? = curvature K a jdy 1 - dn . r = - b = torsion a dv
(4-20)
We use the orthogonal unit vectors (t, n, b) to define the local reference frame for a member element. This is discussed in the following sections. The Frenet
92
D IF F E R E N T IA L G E O M E T R Y O F A M E M B E R E L E M E N T
CH AP. 4
eq u atio n s are utilized to establish th e g o v ern in g differential eq u atio n s for a m em ber elem ent.
4-7 .
LOCAL REFERENCE FRAME FOR A MEMBER ELEMENT
T h e reference fram e asso ciated w ith t. P i, an d b a t a p o in t, say P, on a curve is uniquely defined once th e curve is specified, th a t is, it is a p ro p erty o f th e curve. W e refer to this fram e as the natural fram e a t P. T h e co m p o n en ts o f th e u n it v ectors (t, n, b ) are actu ally th e d irectio n cosines for the n a tu ra l fram e w ith respect to th e basic cartesian fram e w hich is defined by th e o r th o g o n al u n it vectors (7l5'i 2, ^3). W e w rite th e rela tio n s betw een th e u n it vectors as t < n > b ^11 ^21 ^31 ^12 ^2 2 ^32 ^13
^2 3 < '2 > ^33
(4 -2 1 )
O ne can express* th e d irectio n cosines in term s of derivatives o f th e cartesian co o rd in a tes (jc1 9 x 2, jc3) by ex p an d in g (4-19). Since (f, n, b) are m u tu ally o r th o g o n al u n it vectors (as well as i ly 72, 73) the d irec tio n cosines are related by ljmC km = 6jk m1 j, k = 1, 2, 3 (4 -2 2 )
E q u a tio n (4 -2 2 ) leads to th e im p o rta n t resu lt [ r = W ( *-23)
an d we see th a t [ i jk] is an orthogonal m a trix .f T h e results p resen ted ab o v e arc ap p licab le to an a rb itra ry c o n tin u o u s curve. N ow , we co n sid er th e cu rv e to be th e reference axis for a m em b er elem ent an d ta k e th e positive ta n g en t d irectio n a n d tw o o rth o g o n a l d irectio n s in th e n o rm al p lan e as the d irectio n s for th e local m em b er fram e. W e d en o te th e directions o f th e local fram e by ( Y*, Y2, Y3) and th e c o rresp o n d in g u n it v ectors by ( t : , ( 2, 13). W e will alw ays tak e th e positive ta n g en t d irec tio n as th e Yt d irec tio n (t ! t) an d we w o rk only w ith rig h t h an d ed system s (it j x t 2 = ? 3). T h is n o ta tio n is show n in Fig. 4 - 6 . W hen th e cen tro id o f th e n o rm al cross-section coincides w ith th e o rig in of the local fram e (p o in t P in Fig. 4 - 6 ) a t every p o in t, the reference axis is called th e cen tro id al axis for th e m em ber. It is co nvenient, in this case, to tak e Yz, Y 3 as th e p rin cip al in e rtia directio n s for th e cross section. In general, we can specify th e o rie n ta tio n o f th e local fram e w ith respect to th e n a tu ra l fram e in term s o f th e angle < />betw een th e p rin cip al n o rm a l d irectio n an d th e Y2 d irectio n . T h e u n it vectors defining th e local an d n a tu ra l fram es
* See Prob. 4 -5 . t See Prob. 4 -6 .
SEC. 4 - 7
LO C A L R EFER ENC E FR AM E FOR A M E M B E R ELE M E N T
93
are related by h = * _ t 2 = cos (fin + sin 0 b t 3 = sin 0 n + cos (j)b C om bining (4 -2 1 ) a n d (4 -2 4 ) an d d en o tin g the p ro d u ct o f the tw o d irectio n cosine m atrices by p, th e relatio n betw een th e u nit vectors for th e local an d basic fram es takes the concise form (4 -2 4 )
t = pi
w here
\ I
(4 - 25)
1
*1 3
*12
1 2 2 COS
p =
*2 1 COS ~ *21
0 + 1 3J sin 0 i si n 0 + COS 0 j ~
0 -f
*52
sin 0 i
* 23 COS ~ *23
*22
s in 0 +
* 3 2 COS
0 J
0 + * 3 3 sin 0 s in 0 + * 3 3 COS 0
N o te th a t th e elem ents of p are the d irectio n cosines for the local fram e w ith respect to the basic frame. fijk = h ' h = cos (Yp X k) (4 -2 6 )
Since b o th fram es are o rth o g o n al, p r = p~ *. W e will utilize (4 -2 5 ) in th e next chapter to establish th e tran sfo rm a tio n law for th e co m p o n en ts of a vector.
Fig. 4-6. D e fin itio n o f lo c a l re fe re n c e fra m e fo r th e n o rm a l c ro s s s e c tio n .
94
D IFFE R E N TIA L G E O M E TR Y OF A M E M B E R E LE M E N T
CHAP. 4
Example 4 - 6 --------------------------------------------------------------------------------------We determine p for the circular helix. The natural frame is related to the basic frame by
t
' a . sin v
a
- cos y
a
a
a
c~
< n '=
b
cos y - sin y
a c ,
sin y cos y
a c
0 < >2 a
J 3,
= [*;*] {*}
Using (4-25) ~ a ---- sin y a . c . , cos y cos q >H sin v sm a
a Sill (j) j a a j - co s (p O L
4 -8 .
CURVILINEAR COORDINATES FOR A MEMBER ELEMENT
W e take as curvilinear co o rd in ates ( yx, y 2, ,y3) for a point, say Q , the p aram eter y Y o f th e reference axis an d th e co o rd in ates (y 2, y$) of 0 w ith respect to the o rth o g o n al d irections ( Y2, Y 3) in the n o rm al cross section (see Fig. 4 -7 ). Let R R { y u y 2, y 3) be th e p o sitio n vector for Q { V i , y 2, y 3) an d r = F(vt ) the position vector for th e reference axis. T hey are related by R = r + y 2i 2 + y 3t 3 w here h = t 2{yi) - cos < f> n -f sin (j> b _ 13 = t^ y * ) = - s i n (j)h + cos (f> b W e consider < j>to be a function o f y,.
y2
(4 -2 7 )
SEC. 4 - 8 .
C U RVILINEAR C O ORDINATES FOR A M EM BER ELEM EN T
95
T he curve th ro u g h p o in t Q corresponding to increasing y x with y2 an d y 3 held co n stan t is called the p aram etric curve (or line) for y y. In general, there are three p aram etric curves th ro u g h a point. We define u} as the unit tangent vector for the y i p aram etric curve through Q. By definition, Uj = 1 dR 9j dR 9j = dyj (4-28)
The differential arc length along the y, curve is related to dy} by dsj = (4 -2 9 )
aVj
} J =
9i
y'j
T his_ n o tatio n is illustrated in Fig. 4 -8 . O n e can consider the vectors Uj (or dR/dyj ) to define a local reference fram e at Q.
Fig. 4-8. Vectors defining the curvilinear directions. O p eratin g on (4-27), th e p artial derivatives of R are
96
DIFFERENTIAL GEOMETRY OF A M EM BER ELEM ENT
CHAP. 4
We see that (4-30) It rem ains to determ ine Now, and g }. df <Vj Also, ^ 2 , dh dip r\ . / db _ c/^>\ *= cos < f>1 + -y-- b + sin 4 > - j----- m V î ^->1 / W i ^ 1/ dt 3 . , / dh r dd>\ . ( db _ d< p\ = - s i n 0 1 + 0 ) + cos 0 i -7 ------h W )( fl'.V i / \ 4 i rf.Vi/ W e use the F renet equations to expand the derivatives of /? and b. Then, dt2 ( d(j)\ _ - = a K cos (ht , f t + 13 dy 1 dy J dt = aX sin (6r , f a z + j 2 V <*37 ~ at = ott j
(a)
(b)
(0
and finally,
a (l K j/ 2) ^ + (a t +
dj
( y 2t 3 -
v , r 2)
(4_ 3J)
3^2 = ^2 C0S ^
y 3 S> n 4 >
W e see from Fig. 4 - 9 th a t y'2 is the coordinate of the point with respect to the principal norm al direction.
Fig. 4 -9 . D e fin itio n of y'2.
Since dR/dyx (and therefore t/ j ) involve t 2 and r 3, the reference fram e defined by wI; u2, u3 will n o t be orthogonal. However, we can reduce it to an orthogonal
REFERENCES
97
system by taking d(f> dy which requires (f) f ax dy J.V O W hen (4-3 2 ) is satisfied, dR dy1 and
u j = z, at
(4-32)
(4-33)
= a(l - Ky'2)t j
(4 -3 4 )
= oc(l - Ky'2) In this case, th e local fram e a t Q coincides with th e fram e at the centroid. O ne should note th a t this sim plification is practical only w hen ax can be readily integrated.
Example 4 - 7 -------------------------------------------------------------------------------------The parameters a and

t
are constant for a circular helix:

a = ( a 2 + c 2) 112
r = Then,
XT =
and integrating (4-33), we obtain
0 = ()' - v0) = a
ts
follows from (4-34).
For this curve, $ varies linearly with y (or arc length). The parameter
ds,
< 3 \ =
dy y
= <*(i - Ky'i)
= a | J - - y'2 or
REFERENCES
1. 2. 3. 4. G. B., Jr.: A n a l y t i c a l G e o m e t r y a n d C a lc u lu s, Addison-Wesley Publishing Co., Inc., Reading, Mass., 1953. Hay, G. E .: V e c t o r a n d T e n s o r A n a ly s i s , Dover Publications, New York, 1953. Jahnke, E., and F. Emde: T a b le s o f F u n ctio n s, Dover Publications, New York, 1943. S tru ik , D. J .: D if f e r e n ti a l G e o m e t r y , Addision-Wesley Publishing Co., Reading, Mass., 1950.
T h om as,
98
D IFFER EN TIAL G E O M E TR Y OF A M E M B E R ELE M E N T
C H AP. 4
PROBLEMS
D eterm in e t, n, b, a, K, %for the follow ing curves: Xj = 3 cos y x 2 = 3 sin y x 3 = 5y Xi = 3 cos y x 2 6 sin y x 3 = 5>' r = y i t + y 2i 2 + >3T3 Xj = aePy cos y x 2 = ae^y sin y x 3 = cy w here a, /?, c are real constants. 4 - 2 . If x ^ = 0, th e curve lies in the X 1- X 2 plane. T hen, t = 0 a n d b + 13. T he sign of b will depend on the relative o rien tatio n o f n w ith respect to t. Suppose th e eq u atio n defining the curve is expressed in th e form *2 = f i x J x3 = 0 (a) 4 -1 . (a) (b) (c) (d)
E q u atio n (a) co rresp o n d s to tak in g x t as the p a ram eter for the curve. (a) D eterm in e the expressions for 1. h, b , a, an d K co rresp o n d in g to this rep resen tatio n . N o te th a t x t = y and r = x, 7, + ,/'(x1)72 + 0 i3. Let
l - r g .r (b) A pply th e results o f (a) to *2 = ~ xj) w here a and b are co n stan ts. T his is the e q u atio n for a p a ra b o la sym m etrical ab o u t Xn = b[ 2. Let 6 be th e angle betw een t an d 7j. cos 0 = (d)
if
(c)
/ ,.
D educe th a t a = sec 0. E xpress L n, b, and K in term s o f 0. Specialize (c) for the case w here 0 Z is negligible w ith rcspcct to unity. T his ap p ro x im atio n leads to sin 6 % ta n 0 ^ 0 cos 0 1
A curve is said to be shallow w hen 02 1. 4 - 3 . Let K = 1/7? and t = 1/ R t. Show th a t (see (4 -2 0 ))
PR O B LE M S
99
4 -4 . or
T he eq u atio n s for an ellipse can be w ritten as Xj = a cos y

Y 2 Y2
x 2 = b sin y
(a)
_*i + f * = 1
a2 b2
(b)
D eterm ine a , t., n for b o th p aram etric representations. T ake x l as th e p aram eter for (b). D oes y have an y geom etrical significance? 4 - 5 . Show th at , 1 dxk *1* ~~T~ a dy fa* ds d t lk dy
f2 k ~
1/2
d2x k dv2 2 f d 2\k I dy2
1 da. d x k a dy dy 1 da d x k\ 2 112 a dy dy J
Ai î2 =
A 3
1 o
A2 - A i O
i
Ai *22 A.*
As ~ A 2
0
A
j
4 -6 . Then,
Let
< / = [ ( r l *J 2 tjl]
[tjk] = U sing (4-22), show th a t
A (2 ^3
[< * ] T = [ - * 4 - 7 . D eterm in e p for P ro b . 4 - la. 4 - 8 . D eterm ine p for P rob. 4 - l b . 4 - 9 . Specialize p for the case w here the reference axis is in th e X 2 plane. N o te th a t b = w here |A u| = 1. W hen the reference axis is a plane curve and < p = 0, we call th e m em ber a p la n a r m em ber. 4 -1 0 . We express th e differentiation form ulas for lj as dt s = at (a) (b) Show th a t a is, in general, skew sym m etric for an o rth o g o n al system o f unit vectors, i'.e., t j * t k = S jk. D eterm ine a. Suppose th e reference axis is a plane curve b u t (j) ^ 0. The m em ber is not p la n a r in this case. D eterm ine a.
5
Matrix Transformations for a Member Element
5-1. ROTATION TRANSFORMATION
Suppose we know the scalar co m p o n en ts of a vector with respect to a ref erence fram e an d we w an t to determ in e the co m p o n en ts of th e vector co rre spo n d in g to a second reference frame. W e can visualize the d eterm in a tio n of the second set of co m p o n en ts from th e p o in t o f view of applying a tran sfo rm a tio n to the colum n m atrix o f initial com ponents. W e refer to this tran sfo rm a tio n as a rotation transformation. Also, we call th e m atrix w hich defines the tran sfo rm atio n a rotation matrix. Let X] , ij (j = 1, 2, 3 and n = 1, 2) be the directions an d co rresp o n d in g unit vectors for reference fram e n. (See Fig. 5 -1 .) We will generally use a su p er script to indicate the reference fram e for directions, u nit vectors, and scalar
Fig. 5-1. Directions for reference frames "1 and 2.

100
SEC. 5 -1 .
ROTATION TR AN SFO R M ATIO N
101
com ponents in this text. W e consider a vector, a. T h e scalar com ponents of a w ith respect to fram e n are a an d we express a as a = a'i?l + an 2l \ + a%?3 = (a")7 'in N ow , a is independent of the reference frame. Then, a = (a 1)7 ^ 1 = (a2)r i 2 () (5 -1 )
T o proceed further, we m u st relate the two reference frames. W e w rite the relations between th e u n it vectors as (5 -2 ) w here /?jfc is the scalar co m ponent of l] w ith respect to I*. T h e tran sfo rm atio n m atrix, p, is nonsingular when th e u n it vectors are linearly independent. S ub stituting for i2 an d eq uating th e coefficients o f i1 leads to
a 1 = p r a 2 ( b ) a 2 = ( P V 1 * 1
Finally, we let
Rt 2 =
R21 - (R 12) * 1 = pr
W ith this n o tatio n , the relations betw een the com p o n en t m atrices take the form (5 -4 )
The o rder of the superscripts on R corresponds to the direction of the tra n s form ation. F o r exam ple, R 12 is the ro tatio n tran sfo rm atio n m atrix co rre sponding to a change from fram e 1 to frame 2. W e see th at the transform ation m atrix for the scalar com ponents of a vector is the inverse transpose o f the transform atio n m atrix governing th e unit vectors for the reference frames.
Example 5-1 ----- --------------------------- ------- - - --------------------- - -----

We consider the two-dimensional case shown in Fig. E 5-1. The relations between the unit vectors are cos 01} 4- sin 0 i 2
} \ sin 07} + cos <j>i\
(a)
We write (a) according to (5-2).

i2 = p i '
(b)
102
M ATR IX TR AN SFO R M ATIO N S FOR A M EM BER ELEM EN T
CHAP. 5
Then,
R!l - t)' cos (j)
(c)
sin (j) sin 9
<d)
cos 6
|PT| - cos 0 cos (j) + sin 0 sin tp and 1 W\ cos < f> sin < j>
- sin 0 cos 0
(e)
When both frames are orthogonal, < fi - 0 and pr = p
Fig. E5-1
The result obtained in the preceding exam ple can be readily extended to the case of two 3-dim ensional o rthogonal reference frames. W hen bo th fram es are orthogonal, the change in reference frames can be visualized as a rigid body ro tatio n of one fram e into the other, (]jk is the direction cosine for X ] with respect to X k, and the ro tatio n transform ation m atrix is an o rth o g o n al m atrix: R 12 - M f t , = cos (Xj, XI) ,5 -5 ) '
In Sec. 4 - 7 , we defined the orien tatio n o f the local frame (fl5 J2>t3) at a point on the reference axis of a m em ber elem ent with respect to the n atu ral frame (f, h, b) at the point. This frame, in turn, was defined w ith respect to a fixed cartesian frame (T1? i2, T 3). In o rd er to distinguish between the three frames, we use superscripts p and p' for the local an d n atu ral frames at p and a superscript 1 for the basic cartesian fram e: = {*! ^21 t p {t, n, h)
i1 = {V
h, h )
(5-6)
SEC. 5 - 2 .
TH R E E -D IM E N S IO N A L FORCE T R A N S F O R M A T IO N S
103
W ith this n o tatio n , the relations betw een the u nit vectors an d th e various ro tatio n m atrices are: tp = RP Ptp' -- R lpi tp' = R ^ i 1 1. 2. F ro m (4-21),
r
1p' = [ y i 0 cos (j) sin (j? 0 sin (j) cos (j)
(5 -7 )
F ro m (4-24),
R pp =
0 0
3.
R lp = | i - R 1"'R p
P defined by (4-25).
5-2.
THREE-DIMENSIONAL FORCE TRANSFORMATIONS
The equilibrium analysis o f a m em ber elem ent involves th e d eterm in atio n of the internal force and m o m en t vectors a t a cross section due to external forces an d m om ents acting on the m em ber. W e shall refer to bo th forces an d m om ents as forces. Also, we speak o f the force an d m o m en t a t a point, say P, as the force system a t P. T he relatio n sh ip betw een the external force system a t P an d the statically equivalent internal force_system at Q has a sim ple form w hen vector n o ta tio n is used. C o n sid er a force F and m om ent M acting at P show n in Fig. 5 -2 . T he statically equivalent force an d m om ent a t Q are F L M cquiv = M + v x F F (5 -8 )
O ne can visualize (5 -8 ) as a force tran sfo rm atio n in which the force system at P is transform ed in to the force system at Q. T his tran sfo rm atio n will be
linear if r is con stan t, th a t is, if the geom etry o f the elem ent does not change appreciably when th e external loads are applied. W e will w rite (5 -8 ) in m atrix form and trea t force tran sfo rm atio n s as m atrix transform ations.
104
MATRIX TRANSFORMATIONS FOR A MEMBER ELEMENT
CHAP. 5
W e develop first the m atrix tran sfo rm atio n associated with the m o m en t of a force ab o u t a point. Let F P be a force vector acting at point P and M Q the m om ent vector at point Q corresponding to F P. We will always indicate the poin t of application of a force or m om ent vector with a subscript. T he relation between M Q and F P is M q = QP x F P (5 -9 ) W e w ork with an o rthogonal reference fram e (frame 1) show n in Fig, 5 - 3 and w rite the com ponent expansions as f p = Z F h 'j = w , = Z M h i = C )Tm p T he scalar com ponents o f QP are xj,j Xqj . E xpanding th e vector cross pro d u ct leads to
M Q l = XpQFp
' XpQ
0
( s _ 10)
(5-11)
~ ( X P3
X Q 3) ~
XQ l)
1 1
F ( X P2 ~ ( XPl
-X Q 2)
+ ( X P3 ~
~(XP2
~
0
[ + (*P i
-
X Ql)
XQl )
x Ql)
N o te th a t is a skew -sym m etric m atrix. O ne can in terp ret it as a forcetran slatio n transform ation m atrix. The force at P is transform ed by X pQ into
'A
Fig. 5 -3 . Notation for orthogonal reference frame.
SEC. 5 - 2
THREE-DIMENSIONAL FORCE TRANSFORMATIONS
105
a m o m en t at Q. N o te th a t the o rd er o f the subscripts for the tran slatio n tra n s form ation m atrix, XpQ, co rresp o n d s to the o rd er o f the tran slatio n (from P to <2). Also, XpQ an d F P m ust be referred to th e same fram e, th a t is, the super scripts m ust be equal. U p to this point, we have considered only one o rth o g o n al reference frame. In general, there will be a local o rth o g o n al reference fram e associated w ith each p o in t on the axis of th e m em ber, an d these fram es will coincide only w hen the m em ber is prism atic. T o han d le the general case we m ust in tro d u ce ro ta tion tran sfo rm atio n s w hich tran sfo rm the co m p o n en ts o f F and M from the local fram es to the basic fram e (fram e 1) an d vice versa. W e use a su p erscrip t p to indicate the local fram e at p o in t P an d the ro tatio n m atrix co rresponding to a tran sfo rm atio n fro m th e local fram e a t P to fram e 1 is d en o ted by R pl. W ith this n o tatio n , Fp = R plF (5 -1 2 ) Mfc = R ^ M q and th e general expression for M& takes the form M & = (R 1 < 7 Xj,2R pl)F (5 -1 3 )
We consider next the to tal force tran sfo rm atio n . T he statically equivalent force an d m o m en t at Q associated w ith a force an d m om ent at P are given by Fq = Fn Mo = + QP x F P
(a)
W hen all the vectors are referred to a com m on frame, say fram e-1, th e m atrix transform atio n is h XpQ We let or i
' PQ ~
'T
! 1
Fl J M i!
(b)
(5 -1 4 )
The 6 x 1 m atrix / S q is called th e force system at Q referred to fram e 1. U sing this n o tatio n , (b) simplifies to (5 -1 5 ) W hen the force system s are referred to local frames, we m u st first transform them to a co m m o n fram e an d then apply (5-15). U tilizing the general ro ta tio n matrix, m pn = 0 i R pn (5 -1 6 )
and applying
2F\. = M p\ F p
(a)
106
M ATR IX TR AN S F O R M A T IO N S FOR A M E M B E R ELE M E N T
CHAP. 5
we o b tain (5-1 7 ) _ C 7 T V i .S C fPP p< Q p
E q u atio n (5-1 7 ) states th a t w hen the m atrix tra n sfo rm a tio n ,^ - ^ is applied to ZFP P we o b tain its statical equivalent at Q. A ctually, we could leave off th e subscripts an d superscripts on 2T w hen we w rite (5-17). H ow ever, if ZT ap p ears alone, we m ust include them . N o te th a t the force tran sfo rm atio n generally involves both tran slatio n and ro tatio n . T he o rd er o f the subscripts co rresp o n d s to the direction of the tran slatio n , e.g., from P to Q. Sim ilarly th e o rd er of the superscripts defines the direction o f the ro ta tio n o r change in reference frames, e.g., from fram e p to fram e q. In general, th e geom etry of a m em b er elem ent is defined w ith respect to a basic reference frame which we take as fram e 1. T o evaluate we m ust determ ine R pl, R a l, an d from the geom etrical relations for the m em ber. W e have already discussed how one determ ines R lp in Secs. 4 - 7 and 5 - f . W hen th e m em ber is p lan ar* an d the geom etry is fairly sim ple (such as a straig h t or circular m em ber), we can take fram e 1 parallel to one o f the local frames. This elim inates one ro tatio n transform ation. F o r exam ple, suppose we take fram e 1 parallel to fram e p. T hen, R )p = 1 an d reduces to
~R1 < ?
1 0 h R lqx P l Q 1 R 1
-RM
"
R ^ X fQ j RM
(5-18)
parallel, = 1 0 1 A R pT 1 ~R ,7i R P1
~RM
x
'
I qR ^ T
(5-1 9 )
W hen b o th p and q are parallel to 1, -r i~PQ reduces to STj>Q.

a r il _ PQ ~ c a rl ^ PQ
(5-20)
By transform ing from P to Q an d back to P, we o b tain

5/T P _ a p -q p Oj-PH dTTP > p -J QP,J pQj/ p
(a)
(5-21)
an d it follows th a t
QP ~
_ ( orp i 1 W PQ)
If the tran sfo rm atio n from P to Q is carried o u t in the o rd er P - S t , -* S 2, . . , S -> Q, w here S t , S 2, are in term ed iate points, the tran sfo rm a tion m atrix, is equal to the p ro d u ct of th e in term ed iate tran sfo rm atio n m atrices. (5-22)
* If the reference axis is a p la n e cu rve a n d the lo c a l fram e c o in c id e s w ith th e n a tu ra l fram e (< j> = 0) w e sa y th e m e m b er is p lanar.
SEC. 5 -2 .
THREE-DIM ENSIONAL FORCE TR AN SFO R M ATIO N S
107
w here s 2, , s N are arbitrary reference frames. It is co n v en ien t to tak e a co m m o n reference fram e for the in term ed iate transform ations.
Example 5 -2
We consider the plane circular member shown in Fig. E 5-2. We take frame 1 parallel to frame p. Then,
Xp ~ X q = {a sin 9, a(l cos 6), 0}
XpQ XpQ
0 0
<7(1 C O S f>) cos Q
a sin 0
a(l cos 6)
a sin 6
- sin 9 0 cos 0 0
RU = RP3 =
sin 9
The transformation matrix has the form

.T ty = R ?Xv rQ =
Rp
__j_ 0_ '
Rp"XfQ j R'"7
where
'0 0
a( 1 - cos 9)
0 0
a sin 9
+< j( 1 - cos ff)

a sin 9
0
F ig. E 5 -2
Example 5 -3
As an illustration of the case where the geometry is defined with respect to a basic cartesian frame, we consider the problem of finding for a circular helix. The general
108
MATRIX TRANSFORMATIONS FOR A MEMBER ELEMENT has the form
C H AP. 5
expansion for
!_ ? _ ___ ' R 1?Xj.QRpl i R l9Rpl_
_____
R ^XpgR7 '1 j Rp9_
The parametric representation for a circular helix is given in Sec. 4 -1 : x* = a cos y

x \ a sin y
*3
= cy
where xj ( j 1,2,3) are the cartesian coordinates with respect to the basic frame (frame 1). Let y P and y Q b e the values of y corresponding to points P and Q. The coordinate matrices for P and Q are Xp = { a cos yP, a sin y P, c y P}
X q {a cos y Q, a sin y Q, cvQ }
Then,
0
Xl PQ =
c ( y P \ q )
- c(>V - Vo) 0
a (cos yP cos v0)
a(sin y P sin y Q) a(cos v P cos vQ)
-a (s in y P sin y Q)
To simplify the algebra, we suppose the local frame coincides with the natural frame at every point along the reference axis, that is, we take (/> = 0. Using the results of Sec. 4 -7 , the rotation matrices reduce to
sin
a
yQ
- cos
a
y0
c
-
R t _
-c o s
c ,
yQ yQ
- sm y < 2
a 0
a
a
~ sm
a
a
- cos
a
yQ
- sin Rpl =
yP
-cos
yP
a
c
sm w
' = (R lf )r
a - cos y P
C L C
sin y P 0
-COS Vp a
a.
where a 2 a 2 + c 2. Evaluating the product, R 1,Rpl, we obtain /a V A :V cos n + W I* / sm i7 a

COS r)
.
- - ( 1 - cos rj)
ac
a~
c
RP = RlRP> =
a - sin n a
. - sin a
y \
(1 - cos If) a
ac
c a
sm ? ;
+ I - I cos ,
SEC. 5 - 3 . T H R E E -D IM E N S IO N A L D IS P LA C E M E N T T R A N S F O R M A T IO N S
w h e r e rj = yP y Q. A ls o ,
109
R ^ X ^ R '1 a c ~ (2 2 c o s n 1 / s i n ri) X ac ( > / c o s t] s i n tj) y. ac2 a r s i n n + - ( 1 cos/)( :i I 1 1 1 ac ac a 1 ------- 0 / c o s i ; - - s i n t ] ) j y- n s i n n + ( I t a a a -!~ -11 1 c2 a2 [ cr\ s i n t) [ ------- 1 ] c o s > i ---------s i n tj i [ 1 % (X . ----------- 1 ----------- I. C2 ) j 1 1 c2 a2 if c o s q + s i n n 1 t 0L 1 1 , c o s n )(a , c~ )
a 2c cs - 2 ;- ( l c o s o ) ------ r n s i n n a a*
N o t e th a t w e c a n s p e c ia liz e th e a b o v e g e n e r a l r e s u lts fo r th e c a s e o f a p la n e c ir c u la r m e m b e r ( E x a m p le 5 - 2 ) b y ta k in g c = 0 and = 0.
5-3.
THREE-DIMENSIONAL DISPLACEMENT TRANSFORMATIONS
Let P and Q be tw o p o ints on a rigid body. S uppose th a t the b o d y experiences a tran slatio n an d a ro tatio n . W e define uP an d < Z > P as the tran slatio n and r o ta tion* vectors for p o in t P. T h e co rresp o n d in g vectors for p o in t Q are given by < Q
COq Up C 'jp
-j* ( O p x
P Q
(5 -2 3 )
E q u atio n (5 -2 3 ) is valid only when \cdP\2 is negligible w ith respect to unity. Since PQ = QP an d o5P x PQ = PQ x a>P, an a ltern ate form for uQ is
u = up + Q P x d>p
(5 -2 4 )
W e define <0/1 -
K !
KJ
h o
(5 -2 5 )
as the displacem ent m atrix for P referred to fram e 1. T h e displacem ent a t Q resulting from th e rigid body displacem ent a t P is given by =
I XpQ r z
(5 -2 6 )
W e consid er next the case w here the local fram es a t P a n d Q do n o t coincide. The general relatio n betw een th e d isplacem ents has the form
pq ^ . I a J 1 X'
L
=
0 | h < 7 XpQRpl~ R l9Rpl | R 1 R14RP1
(5 -2 7 )
O ne can s h o w t th a t a lte rn a te form s o f (5 -2 7 ) are
{srp P \y''Tmp P= (zr^pfWp
(5^28)
* The units of |a>) are radians, t See Prob. 5-7.
110
M A T R IX T R A N S F O R M A T IO N S FOR A M E M B E R E LE M E N T
CHAP. 5
W e see th a t th e displacem ent tran sfo rm atio n m atrix is th e inverse transpose o f the co rresp o n d in g force tran sfo rm a tio n m atrix. T his result is qu ite useful.
REFERENCES
1. 2. 3. 4. 5.
H a l l , A. S., and R. W . W o o d h e a d : F r a m e A n a l y s i s , 2d ed., W ile y , New York, 1967. MoRrcE, P. B.: L i n e a r S t r u c t u r a l A n a ly s i s , Ronald Press. New York, 1959. P estel , E ., and L eck ie , F . : M a t r i x M e t h o d s in E l a s t o m e c h a n i c s , McGraw-Hill, New York, 1963. L ivesley , R. K .: M a t r i x M e t h o d s o f S t r u c t u r a l A n a l y s i s , Pergamon Press, London, 1964. M a rtin , H. C .: I n t r o d u c ti o n to M a t r i x M e t h o d s o f S t r u c t u r a l A n a l y s i s , McGraw-Hill, New York, 1966.
PROBLEMS
5 - 1 . C o n sid er th e tw o-dim ensional cartesian reference fram es show n. If a 1 = {50, 100}, find a 1.
P ro b . 5 - 1
*2
5 - 2 . T h e o rien tatio n o f tw o o rth o g o n al fram es is specified by the direction cosine table listed below.
x\ Xj
X\ 1/2
x\
1/2 1/2
XI
J}12 -V 2 /2 0
1/2
-V '2 / 2
Xi
(a)
JV2
D eterm in e R 12. Verify th a t (R 12)T = (R 12) l .
P R O B LE M S
111
(b) If a 1 = {10, 5, 10}, find a2. , (c) If a 2 = {5, 10, 10}, find a 1. 5 - 3 . C onsider tw o points, P an d Q, having co o rd in ates (6 ,3 ,2 ) an d ( 5 ,1 ,4 ) w ith respect to fram e 1. T he d irection cosine tables for the local reference fram es are listed below.
11 tp 11 tP 2 tP 3 n n n
1/2 1/2 -7 2 /2 V 2 /2 0
7 2 /2
1/2 1/2
- s /2 /2
1/2 1/2
V ? /2 " V 2/2 0 1/2 1/2 x /2 /2
(a) (b) (c) 5 -4 . straight
D eterm in e 3C J> Q and SC'pQ. D eterm ine &~pq. S uppose = (100, 50, 100,20, 40, +60}. C alculate C o n sid er th e p la n a r m em ber consisting of a circu lar segm ent an d a segm ent show n in the sketch below. P o in t P is at the center o f th e circle. Prob. 5-4
(a) D eterm in e by tran sfo rm in g directly from P to Q. Also find ^ qp(b) D eterm in e Z 'P q by tran sfo rm in g from P to S and then from S to Q. (c) F in d . F'fc co rresp o n d in g to = {0, 0 ,1 ,0 , 0, 0}. 5 - 5 . C o n sid er the circu lar helix, 4v r 2 cos v ii + 2 sm y l 2 ^ - T3.
71
(a) (b)
S uppose <p(y) 0. D eterm in e S uppose <p(y) ~ y . D eterm in e
T ak e y P ?r/2, y Q n/4.
112
M A T R IX T R A N S F O R M A T IO N S FOR A M E M B E R E L E M E N T
C H AP. 5
5 -6 . Refer to P ro b le m 5 - 3 . D eterm in e Uq Q co rre sp o n d in g to llp P = {1/2, - 1/4, 1/3, - 1 / 1 0 , 1/10, 0}. Verify th a t = & PJ 5 -7 . Verify th a t (5 -2 7 ) an d (5 -2 8 ) are eq u iv alen t form s. N o te th a t I3 X QP I I 3 = ( a rhr)
1
\T
i h
5 -8 . C o n sid er th e p lan e m em b er show n. T h e reference axis is defined by *2 = f ( x i). *2 *\ Prob. 5-8
Xx
(a)
(b)
D eterm in e N o te th a t th e local fram e at P coincides w ith the basic fram e w hereas th e local fram e at Q coincides w ith the n a tu ra l fram e a t Q. Specialize p a rt (a) for th e case w here * 2 = p ' ( x {b - x?) a n d th e c o o rd in a te o f p o in t Q is eq u al to b/4. U se th e results o f P ro b . 4 - 2 .
Part II ANALYSIS OF AN IDEAL TRUSS
6
Governing Equations for an Ideal Truss
6 -1 . GENERAL
A system o f b ars* c o n n ec ted a t th e ir ends by frictionless hinges to jo in ts an d subjected on ly to forces ap p lied a t th e jo in t centers is called a n ideal tru s s .t T h e b ars a re assu m ed to be w eightless an d so assem b led th a t th e line c o n necting th e jo in t centers a t th e en d s o f each b a r coincidcs w ith th e c e n tro id al axis. Since th e b ars are w eightless a n d th e hinges are frictionless, it follow s th a t each b ar is in a sta te o f d irect stress. T h e re is on ly o n e force u n k n o w n a sso ciated w ith each b ar, nam ely , th e m ag n itu d e o f th e axial force; th e d ire c tio n o f the force coincides w ith th e line co n n ec tin g th e jo in t centers. If th e b a rs lie in one plane, th e system is called a plane o r tw o-dim ensional truss. T h e re are tw o d isp lacem en t c o m p o n e n ts asso ciated w ith each jo in t o f a p la n e truss. Sim ilarly, a g en eral system is called a space o r three-dim ensional tru ss, an d there a re three d isp la cem e n t c o m p o n e n ts asso ciated w ith each jo in t. W e su p p o se th ere are m b a rs (m em bers) a n d / jo in ts. W e define i as i 2 i 3 for a p la n e tru ss for a sp ace tru ss ,, ( 6 - 1)
U sing this n o ta tio n , th ere are ij d isp lacem en t q u an titie s asso ciated w ith th e j joints. In general, som e o f th e jo in t-d isp la c e m e n t co m p o n e n ts a re prescribed. Let r be th e n u m b e r o f p rescrib ed d isp lacem en t c o m p o n en ts (d isp lacem en t restrain ts) an d nd th e to ta l n u m b e r o f u n k n o w n jo in t disp lacem en ts. It follow s th a t nd = ij - r (6 -2 )
C o rresp o n d in g to each jo in t d isp lacem en t re stra in t is a n u n k n o w n jo in t force

* A prismatic member is conventionally referred to as a bar in truss analysis. t See Ref. 1.
115
116
GOVERNING EQUATIONS FOR AN IDEAL TRUSS
CHAP. 6
(reaction). W e let nf be the to tal num ber of force unknow ns. Then, nf = m + r Finally, the to tal num ber of unknow ns, n, for an ideal truss is n nf + nd = ij + m (6-4) (6 -3 )
T he equilibrium equations for the bars have been used to establish the fact th a t the force in each b ar has the direction of the line connecting the jo in t centers at the ends of th e bar. T here rem ains the equilibrium equations for the joints. Since each jo in t is subjected to a concurrent force system, there are ij scalar force-equilibrium equations relating the bar forces, external jo in t loads, and direction cosines for the lines connecting the jo in t centers in the deform ed state. In o rder to solve the problem , th a t is, to determ ine the b a r forces, reac tions, and jo int displacem ents, m additional independent equations are required. These additional equations are referred to as the bar fo rce-jo in t displacem ent relations and are obtained by com bining the b ar fo rce-b ar elongation relation and b ar elo n g atio n -jo in t displacem ent relation for each o f the m bars. In this chapter, we first derive the elongation -joint displacem ent relation for a single b ar and then express the com plete set of m relations as a single m atrix equation. T his'procedure is repeated for the bar force-elongation relations and the jo in t force-equilibrium equations. W e then describe a procedure for in troducing the joint-displacem ent restraints and sum m arize the governing eq u a tions. Finally, we briefly discuss the solvability of the governing equations for the linear case. In this case, the question of initial instability is directly related to the solvability. In C hapter 7, we develop variational principles for an ideal truss. The tw o general procedures for solving the governing equations are described in C hapters 8 and 9. W e refer to these procedures as the displacement and force m ethods. They are also called the stiffness an d flexibility m ethods in some texts. T he basic concepts em ployed in form ulating and solving the governing equations for an ideal truss are applicable, with slight extension, to a m em ber system having m om ent resisting connections. Some auth o rs start w ith the general system and then specialize the equations for the case o f an ideal truss. We prefer to proceed from the truss to the general system since the basic form ulation techniques for the ideal truss can be m ore readily described. To adequately describe the form ulation for a general system requires introducing a considerable am o u n t of notatio n which tends to overpow er the reader.
6-2.
ELONGATION-JOINT DISPLACEMENT RELATION FOR A BAR
W e num ber the jo in ts consecutively from 1 through j. It is convenient to refer the coordinates of a joint, the joint-displacem ent com ponents, and the external jo in t load com ponents to a com m on right-handed cartesian reference frame. Let X j, i} {j = 1, 2, 3) be the axes and corresponding orth o g o n al unit
SEC. 6 -2 .
E LO N G A TIO N -JO IN T DISPLACEMENT RELATION
117
vectors for the basic frame. T he initial coordinates, displacement components, and com ponents of the resultant external force for joint k are denoted by > Pkj U ~ K 2, 3) and the corresponding vectors are written as h = Z Xkfij = * [i j=i uk = u[i
Pk = Pfc'
(6-5)
The coordinates and position vector for jo in t k in the deformed state are
flkj, PkPk =
% =
X j.
Ik ' = h + a k 4- U k
(6 - 6)
Figure 6 -1 illustrates the notation associated with the joints. Xj
Xi
Fig. 6 - 1 . N otation to r jo in ts .
We num ber the bars from 1 through m and consider b ar n to be connected to joints k and s. The centroidal axis of b ar n coincides with the line con necting joints k and s. From Fig. 6 -2 the initial length of b ar n, denoted- by L, is equal to the m agnitude of the vector A r = rs fk : L l Ar Ar (6-7)
118
CHAP. 6
Since th e basic fram e is o rth o g o n al, (6 -7 ) reduces to

3
L l = (x, - x*)r (xs - x t ) =
(xSJ - x kj}2 ;=i
(6 -8 )
Before the o rien tatio n of th e b a r can be specified, a positive sense o r direc tion m ust be selected. W e ta k e the positive sense for b a r n to be fro m jo in t k *3
Fig. 6-2. Undeformed position of Bar n. to jo in t s an d define anj as the direction cosine for the positive sense of b a r n in th e undeformed state w ith respect to the X j directio n :
a j =
T ~ ( A r * ;) =
^n
L'n
(6 9 )
It is convenient to list the d irectio n cosines in a row m atrix, a . = [aia2a3] = - x fc)r (6 -1 0 )
Lun
N o te th a t a a j = 1, due to the o rth o g o n ality o f the reference frame. Finally, we let t be the unit vector associated w ith the positive d irectio n o f b a r n in the undeformed state. By definition, t n = j ~ Ar = ai L'n (6 -1 1 )
T he deform ed position of b ar n is show n in Fig. 6 -3 . T h e length an d direc tion cosines for b ar n are equal to the m ag n itu d e an d d irectio n cosines for the vector, Ap = ps pk . L et L + en be the deform ed length, v the u nit vector
SEC. 6 - 2 .
E L O N G A T IO N -JO IN T D ISP LA C E M E N T RELATION
119
*3
X2
Fig. 6-3. D e fo rm e d p o s itio n o f B a r n. associated w ith the positive direction in the deform ed state, an d P the corre sponding direction cosine m atrix. These q u antities are defined by (L + e)2 = Ap Ap v = Ap = pi
(6 -12)
L + e 1
(6 -1 3 )
L + e
(Ap 7j)
(6 -1 4 )
We consider first (6-12). S ubstituting for Ap, Ap Ar + (us u k) and noting (6-7), (6-11), we obtain, after dividing bo th sides by Lflt ( h
,2 \2 2 1 _ = 1I + + z n(us etJ 1 1 . 1 ,. l + -I- - 2 2 (11 , ^ *) (. - U + j = - 1 uk) fc)
(a)
(6 -1 5 )
The expression for the direction cosine, f}nJ, expands to 1

Pnj =
a iy +
jr^ ( W .v
(6 -1 6 )
1 + Ln
We list the /Fs in a row matrix, p.
120
GOVERNING EQ UATIONS FOR AN IDEAL TRUSS
CHAP. 6
Pn= 1
1 T + 7~ ( , - U fc) n
(6-17)
By definition, en is the change in length of b ar n. Then, e,JLn is the extensional strain which is considerably less than unity for m ost engineering m aterials. F or example, the strain is only 10 3 for steel at a stress level of 3 x 104 ksi. The relations simplify if we introduce the assum ption of small strain, eJL 1 E xpanding the left-hand side of (6-15), and noting (6-18), we obtain e a(Uj - u*) + y - (u.s - u J T(as - uh) The direction cosines for the deform ed orien tatio n reduce to P +
V-
(6-18)
(6-19)
(us - U ;c )r
(6-20)
To simplify the expression for e further, we need to in terp ret the q u ad ratic terms. U sing (6-20), we can write (6-19) as = (.. + \ (P - *)^ (* ~ k) (a)
This form shows th at the second-order term s arc related to the change in orientation of the bar. If the initial geom etry is such th a t the bar cannot ex perience a significant change in orientation, then wc can neglect the nonlinear terms. We use the term linear geometry for this case. The linearized relations are e w a(u, P a (6-21)
We discuss this reduction in greater detail in C hapter 8. Since we are concerned in this chapter with the form ulation o f the governing equations, we will retain the nonlinear ro tatio n terms. However, we will assum e small strain, i.e., we w ork with (6-19), (6-20).
6-3.
GENERAL ELONGATION-JOINT DISPLACEMENT RELATION
W e have derived expressions for the.direction cosines and elongation of a b ar in term s of the initial coordinates and displacem ent com ponents of the jo in ts at the ends of the bar. By considering the truss as a system or netw ork, the geom etric relations for all the bars can be expressed as a single m atrix equation. The relations for b ar n, which is connected to jo in ts s and k (positive direction
SEC. 6 -3 . G ENERAL E L O N G A T IO N -JO IN T D ISPLACEM EN T RELATION
121
from k to s) are sum m arized below for convenience:

L t = (xs x k)T{xs x k)
= 7- (x.v - Xk)T Lrn = y(us - uk)

J n = n + X (u , ^
ZL/ i
Ut ) '
P = <* + y - (us - uO I.,n
U p to this point, we have considered joints < ? and k as coinciding w ith the positive an d negative ends of m em ber n. N ow we introduce new notatio n which is m ore convenient for generalization of th e geom etric relations. Let n+, n_ denote the jo in t num bers for the jo in ts at the positive and negative ends of m ember n. The geom etric relations take the form (we replace s by n + an d k by n _ in (a)):
U = (xB - x_)T(xn+ - x_)

a = * -* n en = Yn(un+ - UH _)
7~ ( x nt
T - x.Y (6-22)
Y =
1 / \T + h - ( u* - U J 1 T
P = n + f-(U ,u
L'n
To proceed further, we m ust relate the bars and jo in ts of the system, th a t is, we m ust specify the connectivity o f the truss. T he connectivity can be defined by a table having m row s an d three colum ns. In the first colum n, we list the b ar num bers in ascending order, and in the other two colum ns the corresponding numbers, n + and of the jo in ts at the positive and negative ends of the m em bers. This table is referred to as the branch-node incidence table in netw ork theory.* F o r stru ctu ral systems, a branch corresponds to a m em ber an d a node to a joint, and we shall refer to this table as the m em ber-joint incidence table or simply as the connectivity tabic. It should be noted th a t the connectivity depends only on the num bering of the bars and joints, th at is, it is independent of the initial geom etry and disto rtio n of the system. Example 6-1 ------------------------------------------------------------------------------------------As an illustration, consider the two-dimensional truss shown. The positive directions of the bars are indicated by arrowheads and the bar numbers are encircled. The connectivity
* See Ref. 8.
122
GOVERNING EQUATIONS FOB AN IDEAL TRUSS
CHAP. 6
t a b le (w e lis t it h o r iz o n t a lly t o s a v e s p a c e ) f o r th is n u m b e r in g s c h e m e t a k e s t h e fo llo w in g fo rm :
B ar, n
to
11
+ J o in t J o in t
("+ ) (n )
1 2
2 3
4 5
5 6
1 4
2 5
3 6
1 5
2 6
4 2
5 3
Fig. E 6-1
W it h th e c o n n e c t iv it y ta b le , th e e v a lu a t io n o f th e in itia l le n g t h a n d d ir e c tio n c o s in e s c a n b e e a s ily a u to m a te d . T h e in itia l d a t a c o n s is ts o f th e j c o o r d in a t e m a t r ic e s , X j, x 2, . . . , X j. th e c o n n e c t iv ity ta b le a n d th e n = 1, 8_ 5, a n d
T o c o m p u t e L n a n d a , w c fir s t d e te r m in e n + a n d n _ fr o m u s e th e fir s t t w o e q u a t io n s o f ( 6 - 2 2 ) .
F o r e x a m p le , fo r b a r 8, 8 +
'8.
x 8_ = x , - x 5
L l = (x i - x 5)r (x x - x.,)
*8 = 7 - ( * i ~ x 5) r
Jua
We define e and
as the system elongation and joint-displacem ent matrices,

e ~ { e l ^2) 1 e m f ( 6 2 1 )
C U = {ul5o2, .. . ,u,-} and express the m elongation-displacem ent relations as a single m atrix equation
u = (6 -2 4 )
where s is o f order m x ij. The elements in the nth row of involve only the elements o f v - Then, partitioning into submatrices, of order 1 x i, where k = 1 , 2 , . . . , m and i 1 , 2 , . . . , j , it follows that the only nonvan ishing submatrices for row n are the two submatrices whose column number corresponds to the joint num ber at the positive or negative end of m ember n,
SEC. 6 -3 . GENERAL E L O N G A TIO N -JO IN T DISPLACEMENT RELATION
123
namely, n + and n~ : & nn+ ~ +Y n = Tn when # + or n_ (6-25)
c/f -- 0
Example 6 -2 ----------------------------------------------------------------------------------------The matrix can be readily established by using the connectivity table. For row n, one puts +y at column n + , y n at column n_, and null matrices at the other locations. The general form of the matrix for the truss treated in Example 6 -1 is listed below. We have also listed the elongations and joint displacement matrices to emphasize the significance of the rows and partitioned columns of
U2 3 4 5
< ? 1
Yf
-Y
0
-Y z
0 0
Y3
0 0
-Y3
0 0 0
-Y 4
^2 *3
0 0 0
Y2
0 0 0
Y6 0
0 0 0 0
Y7 0
0
-Y s
Y*
*5
Ys
0
-Y e
0
0
0 .
*6
< ? 7
0
0
0
0
-Y-7 0
e eg
Ys
0 0
0
0
"Ys
0 0
Y9
y 10
0
-Y 9 0
< ?10 <?n
Yi o
0
-Y u
Yu
The .o/ m atrix depends on both the geom etry and th e topology. It is of interest to express in a. form where these two effects are segregated. T he form of (6-25) suggests that we list the ys in a quasi-diagonal matrix,
y =
Y2
(6-26)
124
C H AP. 6
an d define C as k 1 , 2 , ,m
C = [C w]
t =
1,2,...
(6 -2 7 )
Cnm. = Qtf = 0 Then,
Cnn- ~ ^ t ^ n+ o r n_
(6 -2 8 )
T h e n etw o rk term inology* for C is a ugm ented branch-node incidence m a trix. W e shall refer to it sim ply as th e co nnectivity m atrix.
Example 6 -3 ----------------------------------------------------------------------------------------T h e c o n n e c t iv ity m a tr ix fo r E x a m p le 6 - 1 2 s in c e th e s y s te m is t w o - d i m e n s i o n a l . is l i s t e d b d o w . T h e u n it m a t r ic e s a r e o f o r d e r
J o in t N u m b e r s
1 1 2 3 4 5
B ar
2 -h + I2
+ I2
-I2 + I2
-I2
+ I2 +h + I2 + I2 + Iz
+ I2 -I2 -I2
n o f C th e
-I2
-I2 ~U I2 -h -I2
+ I2 + I2
w ith b a r n. I t fo llo w s is u s u a lly
6 7 8 9 10 11
O n e can th a t c o n s id e r r o w k o f C
t o d e fin e th e tw o jo in ts a s s o c ia te d b a rs a s s o c ia te d w ith jo in t
column
d e fin e s
k.
T h is
a s s o c ia tio n
* See Prob. 6 -6 .
See also Ref. 8.
SEC. 6 - 4 .
F O R C E -E L O N G A T IO N R E L A T IO N FOR A BA R
125
referred to as in c id e n c e . We say that a joint is positive incident on a bar when it is at the p o s i t i v e end o f the bar. Similarly, a bar is positive incident on a joint when its positive end is at the joint. For example, w e see that joints 1 and 4 are incident on bar 5 and bars 3 ,4 , 6, 8, and 11 are incident on joint 5. We will use this property of the connectivity matrix later to generalize the joint force-equilibrium equations.
6 -4.
-
FORCE-ELONGATION RELATION FOR A BAR
By definition, each b a r o f an ideal tru ss is prism atic an d su b jected on ly to axial lo a d a p p lied a t th e cen tro id o f th e end cross sections. It follow s th a t th e only n o n v a n ish in g stress c o m p o n e n t is th e axial stress, a, an d also, a is c o n sta n t th ro u g h o u t th e bar. W e will co n sid er each b a r to be h o m o g e n eo u s b u t we will n o t req u ire th a t all th e b a rs be o f th e sam e m aterial. T h e strain , e, will be c o n sta n t w hen th e b a r is h o m o g e n e o u s an d th e fo rce-elo n g atio n re la tio n will be sim ilar in fo rm to th e u n iax ial stress-strain cu rv e for th e m aterial. A typical a s curve is sh o w n in Fig. 6 - 4 . T h e in itial p o rtio n o f th e cu rv e is essentially s tra ig h t for en g in eerin g m a teria ls su ch as steel an d alu m in u m . A m aterial is said to be elastic w hen th e stress-strain curve is u n iq u e, th a t is, w hen the curves c o rre sp o n d in g to in creasin g an d d ecreasin g a co in cid e (O A B and B A O in Fig. 6 - 4). If th e b eh av io r for d ecreasin g a is different, th e m a teria l is said to be inelastic. F o r d u ctile m aterials, the u n lo a d in g curve (BC) is essen tially p arallel to th e in itial curve.*
0 E la s tic
Fig. 6-4. Stress-strain curves for elastic and inelastic behavior. W e in tro d u c e th e follow ing n o ta tio n : A - cross sectio n al a re a F - axial force, po sitive w h en te n sio n e 0 = in itial elo n g atio n , i.e., e lo n g atio n n o t asso ciated w ith stress
* A detailed discussion of the behavior of engineering materials is given in Chap. 5 of Ref. 2.
126
GOVERNING EQ UATIONS FOR AN ID EAL TRUSS
CHAP. 6
Since th e stress an d strain are co n stan t th ro u g h o u t the bar, F aA e - U

e0 =
(6-29)
W e convert th e cr-e relation for th e m aterial to the force-elongation relation for th e b a r by applying (6-29).
Fig. 6-5. L in e a r e la s tic b e h a v io r. W e consider first the case w here the stress-strain relation is linear, as show n in Fig. 6 -5 . A m aterial having this p ro p erty is called Hookean. T h e initial and transform ed relations are cr = E(s - c0) V AE
F =
(e e0) - k{e - e0) + e0 = f F + e 0
(6-30)
e = ~ F
W e call k, f the stiffness and flexibility factors for the bar. Physically, k is the force required per unit elongation an d / , which is the inverse of fc, is the elon gation due to a unit force. W e consider next the case w here the stress-strain relation is ap p ro x im ated by a series of straight line segments. T he m aterial is said to be piecewise linear. F igure 6 - 6 shows this idealization for tw o segments. A superscript (j) is used to identify the m odulus an d lim iting stress for segm ent j. T he forceelongation relation will still be linear, b u t now we have to determ ine w hat
SEC. 6 -4 .
FORCE-ELONGATION RELATION FOR A BAR

F
127
<=>
Fig. 6 - 6 . P ie cew ise lin e a r a p p ro x im a tio n .
segm ent the deform ation corresponds to and also w hether the strain is in creasing (loading) o r decreasing (unloading). F o r unloading, the curve is as sum ed to be parallel to the initial segment.* T he relations for the various possibilities are listed below. 1. Loading or Unloading Initial Segment F Act = F n) F = k n '(e - $ > ) 2. Loading Second Segment F w < F ^ A a {2) - F {2) F = ki2\e ~ 4 2)) 4 2) - e\}> + ( / (1> - / (2>)F<!) 3. Unloading Second Segment
frW
<
(6-31)
(6-32)
p*
<c jr ( 2 )
F = I6l \e - gj) g* = e* / ' 1' / * O ne can readily generalize these relations for the nth segm ent.t
(6-33)
* We arc neglecting the Bauschinger effect. See Ref. 2, Sec. 5.9. or Ref. 3, Art. 74. t See Prob. 6-8.
128
G O V ER N IN G E Q U A T IO N S FOR AN ID E A L TR U S S
CH AP. 6
Example 6 -4
We consider a bilinear approximation, shown in Fig. E6-4. Fig. E6 4
a (ksi) F (kips)
(1)
Taking
L = 10 ft = 120 in
A = 1 in
we obtain AE (1' kw = j - = 83.3 kips/in. AEW & < 2> = ------ = 41.7 kips/in. Fil) = Aa< 'u = 30 kips
4 21 = etf> + { / (1) - /< 2>)F(1) = 120 e0 - 0.36 in.
/ (1) = l//c(1) = 12 x 10 3 in./kip f (2) 24 x 10 3 in./kip
Segment 1 Segment 2
(2)
F (83.3)(f 120 e0) F = (41.7)(e - e ^ })
Suppose a force of 35 kips is applied and tht: bar is unloaded. The equivalent initial strain is (see Equation 6-33 and Fig. 6-6): e* = e* - f il)F* e* = + (/< 2) ~ / <n)F* = e( o U + 0.06 in.
T h e p ro ced u re described ab o v e utilizes th e segm ent stiffness, w hich can be in terp rete d as an average tan g en t stiffness for th e segm ent. W e hav e to m odify th e stiffness an d eq u iv alen t in itial elo n g atio n only w hen th e lim it of the seg-
SEC. 6 - 4 .
F O R C E -E LO N G A T IO N R E LA T IO N FOR A BAR
129
m ent is reached. A n a lte rn a te p ro c e d u re is b ased on using the in itial linear stiffness for all th e segm ents. In w hat follows, we o u tlin e th e initial stiffness approach.
F
Fig. 6-7. Notation for the initial stiffness approach. C o n sid er Fig. 6 - 7 . W e w rite the forcc-clongation rela tio n for segm ent 2 as F = k a \ e - e( 0u ) - A = k( l\ e - ? 0ieq) w here e0 cq is in terp rete d as th e eq u iv alen t lin ear in itial stra in an d is given by = W + . / (1) A d) A - {k( i ) k i2))(e e\ o (6 -3 5 ) 6 _ 34)
f(D p a ))
T he eq u iv alen t in itial strain , e0>eq, dep en d s o n e, th e actu al strain . Since e in tu rn depends o n F, one has to iterate on < ? 0-cq regardless o f w h eth er th e seg m ent lim it has been exceeded. T h is d isad v an tag e is offset so m ew h at by th e use of /c(1) for all th e segm ents. T h e n o ta tio n in tro d u ced for the piecew ise lin ear case is req u ired in o rd e r to distinguish betw een th e v ario u s segm ents a n d th e tw o m ethods. R ath er th an co n tin u e w ith this d etailed n o ta tio n , w hich is to o cum bersom e, we will d ro p all the a d d itio n al su p erscrip ts an d w rite th e fo rce-d efo rm atio n relatio n s for b ar n in the sim ple lin ear form fn
kn(en e0t )
(6 -3 6 )
en ~ e0, n + JnFn
130
G O V ER N IN G EQ U A T IO N S FOR AN ID E A L TR U S S
C H AP. 6
w here k, / , an d e0 are defined by (6 -3 1 ) th ro u g h (6 -3 5 ) for th e physically n o n lin ear case.
6 -5 .
GENERAL BAR FORCE-JOINT DISPLACEMENT RELATION
T h e force-deform ation a n d d efo rm atio n -d isp lacem en t relatio n s for b a r n are given by (6 -2 2 ) an d (6-36). C o m b in in g these tw o relatio n s leads to an expres sion for th e b a r force in term s o f the displacem ent m atrices for th e jo in ts at the ends of th e b ar. T h e tw o form s a r e : F = K(e - e0,n) = F 0> + kyan+ - knyun_ pQ,n an d t ^n_) ^0, n JnFn (6 -3 8 ) -K&O.n (6 -3 7 )
W e can express th e force-displacem ent relations for th e m b a rs as a single m atrix eq u atio n by defining
F = { F iF 2 Fm]
(6 -3 9 )
k{ k2 k.
= f-i
an d n o tin g (6-24). T h e generalized form s o f (6 -3 7 ) and (6 -3 8 ) are:

F = k(e e0) = F 0 + k r f W
(6 -4 0 ) (6 -4 1 )
an d
M = e 0 + fF
6 -6 .
JOINT FORCE-EQUILIBRIUM EQUATIONS
L et F n be th e axial force v ecto r for b a r n (see Fig. 6 -8 ). T h e force v ecto r has the d irectio n o f th e u n it vector, v, w hich defines the o rie n ta tio n o f th e b ar in the deform ed state. N ow , v = pi. T hen, Fn = F nvn = Fnpi (6 -4 2 )
W hen F n is positive, the sense o f F i s j h e sam e as th e positive sense for the bar. C on tinuing, we define Fmi), an d F n_ as th e forces exerted by b a r n on th e jo in ts a t th e positive an d negative ends o f th e bar. F ro m Fig. 6 - 8 ,
SEC. 6 - 6 .
JO IN T F O R C E -E Q U IL IB R IU M E Q U A TIO N S J o in t
131
F# .+
F = F n vn
J o in t _ w
Fn n _
Fig. 6 8. Notation for barforce. W e co n sid er n ex t jo in t k. T h e ex tern al jo in t lo a d v ecto r is pk, w here Pft*- F r equ ilib riu m , the resu lta n t force v ecto r m u st eq u al zero. T hen, Pk = Z f ii. -
Pk
j+ = k
t-^ k
F_
(a)
T he first su m m atio n involves the b ars w hich are positive incid en t o n jo in t k (positive end a t jo in t k) an d th e second the bars w hich are negative incident. U sing (6 -43), th e m atrix equ ilib riu m e q u a tio n for jo in t k tak es th e form :
P *= Z M P ?) < - )
j +- k
I . ~k
L et 0* be th e general ex tern al jo in t lo a d m a trix : & = {Pi> P2, , Pi] (y x 1) (6 -4 5 )
W e w rite th e com p lete set o f jo in t force-eq u ilib riu m e q u atio n s as: 0> = m (6 -4 6 )
N o te th a t the row s o f p ertain to th e joints a n d the co lu m n s to th e bars. W e p a rtitio n in to su b m atrices o f o rd e r i x 1. m = [ m !k] I = 1,2,.,.,./ and {ij X m) k 1 , 2 , . . . ,m
Since a b a r is incid en t only on two jo in ts, there will be only two elem ents in any co lu m n o f 39. F ro m (6 -4 4 ), we see th a t, for co lu m n n, 38 n+ 0
n
= + PJ (6 -4 8 ) I ^ n+ or
&n n = w hen
T he ffl m atrix can be readily developed using th e co n n ectiv ity table. It will have the sam e form as w ith y rep laced by P. W h en th e geom etry is linear, P = y = a a n d 28 s/T.
132
CHAP. 6
Example 6 - 5 ------------------------------------------------------------------- ----- -------The matrix for the truss of Example 6-1 has the following general form-. Bar Numbers 1
1 2 3 I T T 5 [ V O 1 +PT -P i +P2 -P2
5
+P s r
9
+ PI
10
11
+P 6 + P7 r
J-P^
-pr -P L + P[o
+ 91
-P I +PJ -P I
-p ? -PS -P S -p f
+ Pu
We could have also utilized the connectivity* matrix C to develop It was pointed out in Example 6 -3 that the elements of the /ah column of C define the incidence of the bars on joint k. Using this property, we can write the generalized form of (6-44) as = C r pr F where Pi 0 0 Finally, we have C 7pT = (PC)' 6-7. INTRODUCTION OF DISPLACEMENT RESTRAINTS; GOVERNING EQUATIONS and #/, (6-50) 0 P2 0 ...
0
(a)
0 p*_
(m x im)
(6-49)
We have developed the following equations relating F, e, e = ,.q1H e0 + fF 0> =
(a)
where the elements of U and are the external joint-displacement and external joint-load matrices arranged in ascending order. Also, in our derivation, we have considered the components to be referred to a basic reference frame. Now,
* See Sec. 6 -3 , Eq. 6-2 7 .
SEC. 6 -7 .
INTRODUCTION OF DISPLACEMENT RESTRAINTS
133
when joint displacement restraints are imposed, there will be a reduction in the num ber of jo in t displacement unknowns and a corresponding increase in the num ber of force unknowns. This will require a rearrangem ent of < fU, 0* and i%. Let r be the num ber of displacement restraints and nd the number of displace ment unknowns. There will be nd prescribed joint loads and r unknown joint loads (usually called reactions) corresponding to the nd unknown joint displace ments and the r known joint displacements. We let U l5 0 2 be the column matrices of unknown and prescribed joint displacement com ponents and P 1 ? P 2 the corresponding prescribed and unknown joint load matrices. The re arranged system joint displacement and joint load matrices are written as U, P:
nd 4- r = ij
M S:
rp
(6-51)
We point out that the com ponents contained in U (and P) may be referred to local reference frames at the various joints rather than to the basic frames. This is necessary when the restraint direction at a joint does not coincide with one o f the directions of the basic frame. Finally, wc let A and B be the trans form ation matrices associated with U and P. Then, (a) takes the fo rm : e = P = BF We partition A, B consistent with the partitioning of U, P: [ A[ I A2 ] (m*nd) (ra *r) (nd x in) bA b 2 (r x m) = AU = e0 + fF (b)
B and write (b) in expanded form:
(6-52)
e = A t U t + A 2U 2 = e0 + fF P i = B tF P2 - B2F '
(6-53) (6-54) (6-55)
Equation (6-53) represents m equations relating the m unknown bar forces, the nd unknow n displacements, and the r prescribed displacements. E quation (6-54) represents nd equations involving the m unknown bar forces and the nd prescribed jo in t loads. Lastly, E quation (6-55) represents r equations, for the r reactions in terms of the m bar forces. W hen the geometry is nonlinear, A and B involve the joint displacements. If the geometry is linear, A BT, and
B j = A j"
j =
1, 2
(6-56)
134
G O VER N IN G EQ UATIO N S FOR AN ID EA L TRUSS
CHAP. 6
W e have in tro d u ced th e displacem ent restraints into th e form u latio n by replacing ,s/, w ith A, B. It rem ains to discuss how one determ ines A, B from 5z/, In the follow ing section, we trea t the case of an a rb itrary restrain t direction. W e also describe how one can represent th e in tro d u ctio n of displace m en t restraints as a m atrix tran sfo rm atio n .
6 -8 .
ARBITRARY RESTRAINT DIRECTION
W h en all the restrain t directions are parallel to the d irectio n of the global reference frame, we o b tain U from ^ I by sim ply rearran g in g th e rows o f U such th a t the elem ents in the first nd row s are the u n k n o w n displacem ents an d the last r row s co n tain th e prescribed displacem ents. T o o b ta in A, we perform the sam e o p eratio n s on th e columns of s i . Finally, since P co rresp o n d s to U , we o b tain B by o p eratin g o n the rows o f PA o r alternately, by o p eratin g on the columns of an d then tran sp o sin g th e resulting m atrix. W hen the restrain t a t a jo in t does n o t coincide w ith one o f the directions of the basic frame, it is necessary first to transform the jo in t displacem ent an d external load co m p o n en ts from the basic fram e to a local fram e associated w ith the restraint a t the jo int. Suppose there is a displacem ent restrain t a t jo in t k. Let Y k j (j = 1, 2, 3) be the o rth o g o n al directions for the local reference fram e associated w ith th e displacem ent restrain t at jo in t k. Also, let ukj an d pkj be the corresp o n d in g displacem ent an d external jo in t load com ponents. Finally, let R ok be the ro ta tio n tran sfo rm atio n m atrix for the local fram e at jo in t k with respect to th e basic fram e (frame o). T he co m p o n en ts are related b y : u * k = R V Pk = R^Pk w here
R* -
(6 -5 7 )
[cos { Y l X j)]
(6 -5 8 )
W e have o m itted th e fram e superscript (o) for q u an tities referred to th e basic fram e (ukj, pkj) to simplify the notatio n . W e define UJ, HP* as the system join t-d isp lacem en t an d -force m atrices referred to th e local jo in t reference frames, tfl3 = {itj, Ut, . . . , uj} ^ = {Pi, P i - - . , Pj} R ol M 0J R o2
-
(6 -5 9 )
and 3%oJ as th e system jo in t-ro ta tio n m atrix,
(6-60)
R oj
Then, (a)
SEC. 6 - 8 .
AR BITR AR Y RESTRAINT DIRECTION
135
O perating on the initial equations with (a), (e = s t W ) (& >= m ) leads to

&SJ = 0 t oJ3B
(e = .v/J')UJ) (0>J = 3 8 F)
,<a/' = ,q /(.# oJ)'
(b) (6-61)
The transform ation of to is the sam e as for th e case where the restrain t directions are parallel to the directions of the basic frame, th a t is, it will involve only a rearrangem ent o f the rows o f &S3. Similarly, we o b tain A by rearranging the colum ns o f s-/J. T he steps are A -> [A j 0S} - B Example 6 -6
To obtain the submatrices in column k oi-<a/J we postmultiply the submatrices in column
k of .r/ by R0* - 1. We can perform the same operation on HBr and then transpose the
^2]
b;
B,
resulting matrix or, alternately, vvc can premultiply the submatrices in row k of 38 by R'1 As an illustration, see the &SJ matrix for Example 6 -5 on page 136. The matrix can be determined by transposing SS1 and replacing p by y.
O ne can visualize the in tro d u ctio n of displacem ent restraints as a m atrix transform ation. We represent the operations *U - U as U = DW P = Y>i? and call D the dispiacem ent-restraint tran sfo rm atio n m atrix. W hen the restraint directions are parallel to the directions o f the basic frame, D is a perm u tatio n m atrix which rearranges the rows of U. W e o b tain D by applying the same row rearrangem ent to a unit m atrix of o rd er ij. P o stm u lti plication by D r etTects the sam e rearrangem ents on th e columns. Also,* D r - D " 1. F o r the general case o f arb itrary restraint directions, we first determ ine tfi3 and then U. Now, m 3 = 0 ioJq t (a) The step, -> U, involves only a perm utation of the row s of U J an d can be represented as u = (6 -6 3 ). and 0> -* P (6 -6 2 )
nwJ
where I I is the perm u tatio n m atrix corresponding to the displacem ent restraints.
* See Prob. 1-36 for a discussion of permutation matrices.
136
GOVERNING EQUATIONS FOR AN (DEAL TRUSS
CHAP. 6
SL
h .~ CO. in D
o f- COr> i fid 1
. c pi c CO. ft5
~'.ai CO. C'l U. Ctf
CO. Qd 1
ca . 00 k. as -
COo 05 1
r~
? rCO_ r* > Q*
Kr CO. v h i
SO u.
t-\o eo . ri o OS
QO. * he I
CO. in u, Q o i
5 ID CO. os i
s. GO. < # o os
r~ -+ CO. v O
OS
1
I u. 1
CO. * o cA
f-* ro ecu m !
b 1 r>J C3. rs
OS
hn CO. c o 1
e i o
i i ea ts a; I
1
s r-l n J cx rorn CX cx
'T S
oo cx
SEC. 6 -9 .
IN IT IA L IN S T A B ILIT Y
137
C om bining (a) an d (6-63), we have U = and it follows th a t D = n m oJ (6 -6 4 ) (b)
Since b o th I I and 0 t oj are o rth o g o n al m atrices, D is also an o rth o g o n al m atrix. U sing (6-62),
r A = <s/D r
(c)
and th en substitu tin g for 39, an d D in term s o f the geom etrical, connectivity, local ro ta tio n m atrices lead to B = n # oJ(p c )T
o J \T A = y C ( i m oJ)
(6 ~ 6 5 )
E quation (6 -6 5 ) is o f interest since th e various term s are isolated. H ow ever, one w ould n o t generate A, B w ith it.
6-9.
INITIAL INSTABILITY
The force equilibrium eq u atio n s relating the prescribed external jo in t forces and the (internal) b a r forces has been expressed as (see E q u atio n 6 -5 4 ): P, = BtF (a)
where is (nd x t) an d F is (m x 1). W hen the geom etry is nonlinear, Bj depends on the jo in t displacem ents as well as on the initial geom etry an d restraint directions. In this section, we are concerned with the b ehavior u nder an infinitesim al loading. Since th e n o n lin ear term s depend on the load intensity, they will be negligible in co m p ariso n to the linear term s for this case, i.e., we take B] as constant. Then, (a) represents nd linear equations in m unknow ns. If these equatio n s are inconsistent for an a rb itra ry infinitesim al loading, we say the system is initially unstable. W hen the geom etry is linear, B i is in d ep en d en t o f the loading an d th e initial stability criterion is also applicable for a finite loading. This is n o t tru e for a nonlinear system. W e treat stability u nder a finite loading in C h ap ter 7. C onsider a set o f j linear algebraic eq u atio n s in k unknow ns. ax = c (b)
In general, (b) can be solved only if a an d [a Jc ] have the sam e rank.* It follows that the equ atio n s are consistent for an a rb itra ry rig h t-h an d side only w hen the rank o f a is eq u al to j, the to tal n u m b er o f equations. A pplying this condition
* See Sec. 1-13; see also Prob. 1-45.
138
GOVERNING EQ U A TIO N S FOR AN ID EA L TRUSS
CHAP. 6
to (a), we see th a t the truss is initially unstable w hen the rank o f is less th a n nd. F o r the truss to be initially stable u nder an a rb itra ry loading, m u st be of ran k rtd. T his requires m > nd. T h a t is, the n u m b er o f bars m u st be at least equal to th e nu m b er of unknow n displacem ent com ponents. Since the ran k m ay still be less than nd, this con d itio n is necessary b u t n o t sufficient for initial stability. In o rder to determ ine w hether a truss is initially stable, o n e m ust actually find th e ran k of B j. T he following exam ples illustrate various cases o f initial instability. Example 6 - 7 --------------------------------------------- ------------------------ ---T he force-equilibrium eq u atio n s for the accom panying sketch are: _ Fig. E6-7
F Ft
Pi I
F2
F3
-1 + 1
l til
Pi 2 P21 P22
P31
+ 1
+ 1 + 1
R ow 3 is ( - 1 ) times row 1. T h e equations are consistent only if p21 - p n . Since m < nd, we know the system is unstable for an arb itrary loading w ithout actually finding ^(BJ.
SEC. 6 -9 .
IN IT IA L IN S T A B IL IT Y
139
Example 6 -8 ---------------------------------------------------------------------------------------W e fir s t d e v e lo p th e
.4? m
a tr ix
fo r th e
tr u ss s h o w n
in
Fig.
E 6 -8 A
and
th e n s p e c ia liz e
it f o r v a r i o u s r e s t r a i n t c o n d i t i o n s .
Fig. E6-8A
c _-
X 2
M 4 ,
\ < 1 --1
y \e
F
'
3
0
f
F 3
F ,
P lL
- 1
cos 6
0 )
P l2
+ 1
s in 9
(5 )
P21
+ 1 iA!
cos 0
SP (D
P22
s in 9
P3)
+ 1
cos 6
P32
- 1
s in 0
Pa. 1
- 1
cos 0
P 42
- 1
s in 6
T h e r e a r e th r e e r e la tio n s b e tw e e n th e r o w s o f
(1) (2 ) (3 )
ro w ro w (s in 0 ) ( r o w ( 1 )
( i) (2 ) +
+ +
ro w ro w
(J)
ro w ro w
(5 )
= =
ro w ro w
(7) (8 ) )
(4 ) +
ro w
(3 ))
co s 0 (ro w
(4 )) =
co s 6 (ro w
140
G O VERNING EQ UATIO N S FOR AN ID E A L TRUSS
C H AP. 6
The first two relations correspond to the scalar force equilibrium conditions for the external joint loads:
4. X
Pkl = Pll +
Pll +
P31 +
p 41 =
k= 1
d
k~l
(a)
E Ph-2 = Pi 2 + P i 2 + P32 + ?42 0
The third relation corresponds to the scalar moment equilibrium condition:

4 4.
Z Mk = Z
It = 1
( ~ Xk2Pkl + Xk lp k2) = 0
(b)
where M k is the moment of the external force vector acting at joint k with respect to point O , the origin of the basic frame. We obtain relation (3) by taking O at joint 4. Equation (b) reduces to ~ d ( p u + p 2 l ) + b { p 22 + p i 2 ) = 0 (c) Using d = L sin 9
l .= L i cos a b 0
^'
we can write (c) as cos 6 p 31 = sin 0 { p l x + p n ) ~ cos 0 p 22 (e)
which is relation 3. We see that rows 2 and 5 are independent. The remaining set (rows 1, 3, 4, 6, 7, 8) contains only three independent rows. Now, we obtain from by first taking a linear combination of the rows (when the restraints are not parallel to the basic frame) and then deleting the rows corresponding to the joint forces associated with the prescribed joint displacements. Since 0ft has th r e e linear dependent rows, it follows that we must introduce at least th r e e restraints. Initial instability will occur if 1.
2.
An insufficient number o f restraints are introduced (nd > 5). A sufficient number of restraints are introduced (na = 5) but the rows of I*! are not linearly independent. We say the restraints are n o t independent in this case. These cases are illustrated below.
Case I Fig. E6-8B 1

2
SEC. 6 - 9 .
IN IT IA L IN S T A B IL IT Y
141
We obtain B, by deleting rows 6 and 8 (corresponding to /?32 and p 42). The system is stable only when the applied joint loads satisfy the condition
P it + P n + Pm = ~P*i
Case 2 Fig. E6-8C
We delete rows 4,6, and 8. The number of restraints is sufficient (nd = 5) but the restraints are not independent since /-(BJ < 5. Actually, r(B5) = 4. T o make the system stable, at least one horizontal restraint must be introduced.
In E xam ple 6 - 8 , we show ed th a t there are three relations betw een the row s o f .# for a tw o-dim ensional truss. These relatio n s co rresp o n d to the force- and m om ent-equilib riu m co n d itio n s for th e com plete truss. T o establish th e relatio n s for th e th ree-d im en sio n al case, we sta rt w ith the equilibrium eq u atio n s,
<i*l)
Pc =
j
0
(2i-3)x 1
(a)
0 (b)
I Mi {= l
where M ( is th e m o m en t o f p( w ith respect to an a rb itra ry m o m en t center, 0. F or convenience, we tak e 0 a t th e origin o f th e basic reference frame. P a r titioning 39, 'a t' & > 3S}
(6 -66)
where 39k is of order (i x m) and using the m atrix notation introduced in
142
G O VER N IN G EQ U A TIO N S FOR AN ID E A L TR U S S
CHAP. 6
Sec. 5 - 2 for the m om ent,* the equ ilib riu m eq u atio n s tak e the form
fa ,= o
t- 1 (- I
(6-67)
( 6 - 68)
t x ro< = 0
E q u atio n (6 -6 7 ) represents i relatio n s betw een the row s o f M , row q + row (q + i) + + row [i(j 2) + q] row [/(/' 1) + <?] q = 1, 2 ,. .. i (6 -6 9 ) a n d (6 -6 8 ) co rresp o n d s to (2 i 3) relations. W e have show n th a t there are a t least 3(i 1) relations betw een th e cows o f 33. N ow , we o b ta in B by co m b in in g an d rearran g in g th e row s o f ffl. It follows th a t B will also have a t least 3(i 1) relatio n s betw een its row s. Finally, we o b ta in B, by deleting the row s co rresp o n d in g to th e restrain ts. F o r the system to be initially stable, we m ust in tro d u ce a t least 3 (i 1) restrain ts: r = no. o f restrain ts ^ 3(/ 1) (6 -7 0 )
N o te th a t this req u irem en t is in d ep en d en t o f the n u m b e r o f bars. Also, it is a necessary b u t n o t sufficient co n d itio n for initial stability. T h e n u m b er o f restrain ts m ust also satisfy the necessary co n d itio n nd < in. This requires r = W ~ nd) > (ij - m) (6-7 1) B oth (6 -7 0 ) an d (6 -7 1 ) m ust be satisfied. depending on th e arran g em en t o f th e bars. E ith er co n d itio n m ay co n tro l r,
REFERENCES
1. 2. 3. 4. 5. 6.
N o r r is ,
C. H., and J. B. W i l b u r :
E l e m e n t a r y S t r u c t u r a l A n a l y s i s , McGraw-Hill,
7. 8.
New York, 1960. C r a n d a l l , S. H., and N. C . D a h l : A n I n t r o d u c ti o n to th e M e c h a n i c s o f S o lid s, M cGraw-Hill, New York, 1959. Timoshenko, S .: S t r e n g t h o f M a t e r i a l s , Part 2, Van Nostrand, New York, 1941. T im o shen k o , S., and D. H. Y o u n g : T h e o r y o f S t r u c t u r e s , McGraw-Hill, New York, 1945. M c M i n n , S. J .: M a t r i c e s f o r S t r u c t u r a l A n a ly s i s , Wiley, New York, 1962. M a r tin , H. C .: I n t r o d u c ti o n to M a t r i x M e t h o d s o f S t r u c t u r a l A n a l y s i s , McGrawHill, New York, 1966. L ivesley , R. K .: M a t r i x M e t h o d s o f S t r u c t u r a l A n a l y s i s , Pergamon Press, London, 1964. F enves , S. J., and F. H. B r a n i n : Network-Topological Formulation o f Structural Analysis, J. S tr u c t. D i v . , ASCE, Vol. 89, N o. ST4, pp. 483-514, 1963.
* See Eq. 5 -1 1 .
PR O B LE M S
143
PROBLEMS
6 -1 . D eterm in e m ,j, r, an d ty for the follow ing p lan e trusses: Prob. 6-1
6 -2 .
Suppose b ar n is connected to jo in ts s an d k w here x* = {1, 1, 0} (ft) xs {5, - 5 , - 2 } (ft) .
(a) (b)
T ak e the positive d irectio n of b ar n from k to s. D eterm ine L, a, and t. S uppose uk = {1/10,1/20, 1/10} (inches) us = {1/20, - 1 / 1 0 , - 1 /3 0 } (inches)
F in d % and p k. N o te th a t the units o f x an d u m u st be consistent. D eterm in e en an d P, using th e exact expressions (E quations 6 -1 5 , 6 -1 7 ), th e expressions specialized for th e case of sm all strain (E q u a tions 6 -1 9 , an d 6 -2 0 ), an d the expressions for the linear geom etric case (E q u atio n 6-2 1 ). C o m p are the results for the three cases. 6 - 3 . D iscuss w hen the linear geom etric relatio n s are valid an d develop the a p p ro p ria te n o n lin ear elongation-displacem ent relations for th e trusses shown. Assume no su p p o rt m ovem ents. 6 -4 . C o n sid er th e truss show n: (a) E stablish th e connectivity table. (b) List the initial direction cosines. D o we have to include non lin ear geom etric term s for this truss? (c) L o cate the n o n zero subm atrices in j&, using the connectivity table. D eterm in e th e com plete form of
144
G O V ER N IN G E Q U A T IO N S FOR AN ID E A L TRUSS
C H AP. 6
Prob. 6-3
*2
Xl
Prob. 6-4
xx
(d) D eterm in e C. (e) Verify th a t aC . 6 - 5 . D eterm in e stf for th e th ree-d im en sio n al tru ss show n. 6 - 6 . C o n sid er the d-c n etw o rk show n. T he ju n c tio n s are generally called nodes, an d th e line co n n ectin g tw o n o d es is called a b ran ch . T h e encircled n u m b e rs refer to th e b ran ch es a n d th e arro w h ead s indicate the positive sense (of the cu rren t) for each b ranch.
P R O B LE M S
145
Let Vj (j = 1, 2 , . . . , 5) d en o te the p o te n tial a t n o d e j. Also, let n+ an d n_ denote th e n o d es at th e positive a n d negative ends o f b ran ch n. T h e p o te n tial Prob. 6-5
Prob. 6-6
drop for b ran ch n, in d icated by e, is given by
W e define v an d e as v = {ul5 v 2, . . . , v 5} general n o d e p o te n tial m atrix e ( e 1, e 2, . . . , e7j = general b ran ch p o te n tial difference m atrix an d w rite th e system o f b ran ch p o te n tia l d iffe re n c e -n o d e p o te n tia l relatio n s as e .c/v D e term in e using th e b ra n c h -n o d e co nnectivity table. D iscuss how the truss p ro b lem differs from the electrical n etw o rk p ro b lem w ith respect to the. form o f 0. H o w m an y in d e p en d en t co lu m n s does h av e? In n etw o rk theory, is called th e au g m en ted b ra n c h n o d e incidence m atrix.
146
G O VER N IN G EQ U A TIO N S FOR AN ID EA L TRUSS
C H AP. 6
6 - 7 . T ak e L 20 ft, A 2 in 2, an d the a-e curve show n. (a) D evelop the piecew ise linear force-elongation relations. (b) Suppose a force of + 60 kips is applied an d then rem oved. D eterm ine the force-elongation relatio n for the inelastic case. (c) Suppose the b ar experiences a tem p eratu re increase o f 100 F. D eter m ine the initial elongation. C o n sid er the m a teria l to be alum inum .
a (ksi)
Prob. 6-7
6 -8 .
G eneralize E q u atio n 6 - 3 2 for segm ent j. S tart w ith ,? = e tf + f U)F
an d express in term s of q u an tities associated w ith segm ent ( j 1). 6 - 9 . G eneralize E q u atio n 6 -3 5 for segm ent j. 6 -1 0 . Suppose the stress-strain rela tio n for initial loading is ap p ro x im ated , as in th e sketch, by <r E(s be 3) Prob. 6-10
P R O B LE M S
147
D eterm in e expressions for EJ an d E \ th e secan t a n d ta n g en t m oduli. D eterm in e expressions for ks a n d k r. S uppose the m aterial behaves inelastically for decreasing |cr|. C o n sid er th e u n lo ad in g curve to be parallel to th e initial tangent. D eterm in e the force-elongation relatio n for AB. 6 -1 1 . R epeat P ro b . 6 10, using th e stress-strain relatio n = ^ (ff + c|<r|") w here E, c, an d n are constants. 6 -1 2 . F o r the acco m p an y in g sketch:
(a) (b) (c)
Xi
Prob. 6-12
Xi
15'
15'
(a) L o ca te the n onzero subm atrices in fB. (b) A ssem ble (91 for the linear geom etric case. 6 - 1 3 . R epeat P ro b . 6 - 1 2 for the th ree-d im en sio n al tru ss show n. Prob. 6-13
2
10' *2
5'
6 -1 4 . C o n sid er th e electrical n etw o rk o f P ro b . 6 -6 . (a) L et i be th e cu rre n t in b ran ch n. T h e positive sense o f i is from, n o d e n_ to n o d e n +. N ow , the to ta l cu rren t flow ing in to a n o d e m ust equal th e to tal cu rren t flow ing o u t o f th e node. T his req u irem en t leads to one eq u atio n for each n o d e involving the b ran ch cu rren ts incid en t o n
148
G O V ER N IN G EQ U A TIO N S FOR AN ID EA L TRUSS
CHAP. 6
th e node. L et 1 = {h> >h } ~ general b ran ch -cu rre n t m atrix
Show th a t th e com plete system o f n o d e eq u atio n s can be w ritten as

(5x1)
s#Ti = (b) (c)
(a)
w here s i is given in P rob. 6 - 6. H o w m any in d ep en d en t eq u atio n s does (a) rep resen t? (H int: has only four in d ep en d en t columns). W hen the resistance is linear, the cu rren t an d p o ten tial d ro p for a b ran ch are related by en = e0t + Ri (b)
w here e0 is the b ran ch em f an d R is th e b ran ch resistance. An altern ate form is ' * ii R>i 0, n)
N o te the sim ilarity betw een (b) and the linear elastic m em ber forceelo n g atio n relation. Show th a t the com plete system o f b ran ch curr e n t- n o d e p o ten tial relations can be w ritten as
e = ,.c/v = e0 + Ri i = R~\e - e0) = R~K p/
v
~ R _1e0
E q u atio n s (a) an d (c) are the governing u n p a rtitio n e d eq u atio n s for a iinear-resistance d-c n etw ork. T he p artitio n ed eq u atio n s are developed in P ro b . 6 -2 3 . It should be n o ted th a t th e n etw o rk p ro b lem is o n e dim ensional, th a t is, it does n o t involve geom etry. T he m atrix depends only on the topology (connectivity) o f th e system . A ctually, s i co rresp o n d s to th e C m atrix used in Sec. 6 -3 with i 1. 6-15. Refer to P ro b . 6 -1 2 . Suppose u xu u^z, u52 are prescribed. Identify Bj, an d B 2. 6-16. Refer to P ro b . 6 -1 2 . (a) D evelop th e general form o f 48J. (b) Suppose u2 l, u42, W 52 are prescribed. T h e o rien tatio n o f the local fram e a t jo in t 5 is show n in th e sketch. D eterm in e a n d B 2.
Prob. 6 -1 6
*2
P R O B LE M S
149
6 -1 7 . Refer to P ro b . 6 -1 3 (a) D evelop the general form o f BJ. (b) D eterm in e Bi an d B2 co rresp o n d in g to the follow ing prescribed dis placem ents:
U 1 1, U 12> u 3l> u 33> u 2 3> U 13
T he local fram e a t jo in t 2 is defined by the follow ing d irectio n cosine table.
h. Yi Y\ n
-1A /2 1/2 1/2
*2 1A/2 1/2 1/2
*3 0 - 1A/2 1/V2
6-18 . C o n sid er the tw o-dim ensional truss show n. T he bars are o f equal length an d 0 is th e center o f the circum scribed circle. T he restrain t direction is a degrees counterclockw ise from th e tan g en t at each jo in t. Investigate the initial stability o f this system . R ep eat for the case o f four bars.
Prob. 6-18
6 -1 9 . S uppose nd = m. T hen, B,^ is o f o rd er m x m. T h e equilibrium equ atio n s for P i = 0 are

(m x in) (m x 1)
i*l
Bi
(a)
If (a) has a nontrivial so lu tio n , the ra n k o f B i is less th a n m an d th e system is initially unstab le (see P ro b . 1-45). R ath er th a n o p erate on B l5 to determ ine K B J, we can proceed as follows: (1) W e tak e the force in som e bar, say b a r k, equal to C: Fk = C
150
GOVERNING EQ UATIO N S FOR AN ID EAL TRUSS
CHAP. 6
(2) U sing the jo in t force-equilibrium equations, we express th e rem aining b ar forces in term s of C. (3) T he last equilibrium eq u atio n leads to an expression for F k in terms o f C. If this reduces to an identity, /'(Bj) < nd since a n o ntrivial solution for F exists. This procedure is called the zero load test. (a) Apply this procedure to P rob. 6 -1 8 . T ake F j = C an d determ ine F2 > F 3, an d then F t using the equilibrium co n d itio n (sum m ation of forces norm al to r m ust equal zero) for jo in ts 1, 2, 3. (b) W hen nd = m and the geom etry is linear, the truss is said to be statically determ inate. In this case, we can determ ine F, using only the equations o f static equilibrium , since the system, P t = B t F, is square. D o initial elongations and su p p o rt settlem ents introduce forces in the bars of a statically d eterm inate truss? 6-20. M odify the zero load test for the case w here nd < m. N o te th a t the general solution o f B jF = 0 involves m r(Bt) a rb itrary constants. 6-21. Investigate the initial stability of the tw o-dim ensional truss shown. Use the zero load test.
Prob. 6-21
T a
a
1
5
6 -22. Investigate the initial stability o f the system shown. T he restraint directions are indicated by th e slashed lines.
P rob . 6 - 2 2
P R O B LE M S
151
6 -2 3 . W e generalize th e results o f P ro b s. 6 - 6 a n d 6 - 1 4 for a n etw o rk having b bran ch es an d n nodes. Let e = b ran ch p o te n tial diff. m atrix = {elf e 2, . . , eb} i b ran ch cu rren t m a trix = {it , i2, . . . , ib} v = n o d e p o ten tial m atrix = v2, , v}
T he general relatio n s are (1) n o d e e q u atio n s (n eq u atio n s)

( nxb) (ftxl) (xl)
(a)
and (2) b ran ch eq u atio n s (b equations) e = .s&v = e0 + Ri (b)
N ow , s i/T has only n 1 in d ep en d en t row s. O n e can easily show th a t the row s o f ,<s/T are related by . n-1 row n Yj row k k= 1 (c)
It follows th a t (a) represents only n 1 in d ep en d en t eq u atio n s, an d one eq u atio n m ust be disregarded. S uppose we delete th e last eq u atio n . T his co rresp o n d s to deleting the last colum n o f ,?f (last row o f M r \ W e p a rtitio n
(b x n) b x (n - 1) 6x1
sA and let
= [
-c'/ 2]
(d)
A. T he reduced system o f n o d e eq u atio n s has the form Ari = 0 (e)
N o te th a t A T co rresp o n d s to B[ for the truss problem . E q u atio n (e) represents (n 1) eq u atio n s. Since v is o f o rd er n, one o f th e node p oten tials m u st be specified. T h a t is, we can only determ in e the p o ten tial difference for th e nodes w ith respect to an a rb itra ry node. W e have deleted the last colum n o f $ w hich co rresp o n d s to n o d e n. T herefore, we take v as the reference poten tial. (a) Let
, '" V2 (n - X 1)x , 1 = N ~ L
Show th a t
1-
AV (b) S um m arize th e governing eq u atio n s for th e netw ork. T he o p eratio n si >A v V
co rresp o n d s to in tro d u cin g displacem ent restrain ts in the tru ss p ro blem . C o m p a re the necessary n u m b e r o f restrain ts req u ired for the n etw o rk an d truss problem s.
7 Variational Principles for an Ideal Truss

7-1. GENERAL
The form ulation of the governing equations for an ideal truss described in C hapter 6 involved three steps: 1. 2. 3. The elongation of a bar was related to the translations of the joints at the end of the bar. Next, the bar force was expressed in terms of the elongation and then in term s of the jo in t translations. Finally, the equilibrium conditions for the joints were enforced, re sulting in equations relating the external jo in t loads and internal b ar forces.
T he system equations were obtained by generalizing the m ember forcedisplacement and joint force equilibrium equations and required defining only two additional transform ation matrices (jtf, 38). Later, in C hapter 10, we shall follow essentially the same approach to establish the governing equations for an elastic solid. In this chapter, we develop two variational principles and illustrate their application to an ideal truss. The principle of virtual displacements is treated first. This principle is ju st an alternate statem ent of force equilibrium. Next, we discuss the principle of virtual forces and show that it is basically a geo metrical com patibility relation. Both principles are then identified as the sta tionary requirements for certain functions. F or this step, we utilize the m aterial presented in C hapter 3, which treats relative extremas of a function. Finally, we discuss the question of stability of an elastic system and develop the stability criterion for an ideal truss. Why bother with variational principles when the derivation of the governing equations for an ideal truss is straightforw ard? O ur objective in discussing them at this time is prim arily to expose the reader to this point of view. Also, we can illustrate these principles quite easily with the truss. Later, we shall
152
SEC. 7 -2 .
PRINCIPLE OF VIRTUAL DISPLACEMENTS
153
use these principles, particularly the principle of virtual forces, to construct approxim ate formulations for a member.
7-2.
The principle of virtual displacements is basically an alternate statem ent of force equilibrium. We will establish its form by treating first a single particle and then extending the result to a system of particles interconnected with internal restraints. The principle utilizes the concept of incremental work and, for completeness, we review briefly the definition of work before starting with the derivation. Let v be the displacement of the point of application of a force F in the direction of F. The work done by F (see Fig. 7 -1 ) is defined as W = W0 + f F dv = W(v) Jvo (7-1)
where r0 is an arbitrary reference displacement. Since is a function of v, the increment in W due to an increment Ay can be expressed in terms of the differentials of W when F is a continuous function of u .t
a w = d w + y 2 w + dW dW rAt> = F Av dv dF d 2W d(dW) = ~ ( A v ) 2 dv
1 .
We refer to dW as the first-order work. Similarly, we call d z W the secondorder work. If dFjdv is discontinuous, as in inelastic behavior, we must use the value of dFjdv corresponding to the sense of Av. This is illustrated in
F
w o
"
Fig. 7 -1 . W ork in te g ra l fo r the o ne -d im e nsio na l fo rc e -d is p la c e m e n t relation.

t D ifferential n otation is introduced in Sec. 3 - 1 .
154
VARIATIONAL PRINCIPLES FOR AN IDEAL TRUSS
CHAP. 7
Fig. 7 -2 . We use dF/dv = for Au > 0, and dF/dv = - k 2 for Au < 0. N ote th at W is not a single-valued function of v when there is a reversal in the F-v curve.
F
Ffg. 7-2. W ork integral for d ire c tio n -d e p e n d e n t force.
We consider_first a single mass particle subjected to a system o f forces (see Fig. 7-3). Let R be the resultant force vector. By definition, the particle is in equilibrium when R 0. We visualize the particle experiencing a displacement increment Ah from the initial position. The first-order work is t d W = R Aw (7-3) If the initial position is an equilibrium position, dW - 0 for arbitrary M since R 0. Therefore, an alternate statem ent of the equilibrium requirement is: The first-order work is zero for an arbitrary displacement of a particle from an equilibrium position. The incremental displacement Au is called a virtual displacement-, this state ment is the definition of the principle of virtual displacements.
One can readily generalize (7-4) for the case of S particles. Let dWq be the first-order work associated with the forces acting on particle q and Auq the
t W e consider the forces to be co n tin u o u s fun ctions o f AS.
SEC. 7 -2 .
155
corresponding virtual-displacement vector. If particle q is in equilibrium, dWq = 0 for arbitrary Auq. It follows that the scalar force-equilibrium equations for the system are equivalent to the general requirement, s . dW = J] dWq = 0 for arbitrary Aiiq 1-1 q = 1, 2 , . . . , S
(7-5)
Equation (7-5) is the definition of the principle of virtual work for a system of particles. In general, some of the forces acting on the particles will be due to internal restraints. We define dWE as the first-order work done by the external forces and dW, as the w ork done by the internal restraint forces acting on the particles. Substituting for dW , (7-5) becomes dWE + dW} 0 for arbitrary Aur/ q = 1,2 , . . . , S ta)
Now. let dWD be the work done by the internal restraint forces acting on the restraints . We use the subscript D for this term since it involves the
A lii I ------- I
p / (Deformed)
AU y I H
p
4 --- h -
A / V W ------ 1
J i ---------\/\/V ---- \------ 1

(Initial)
O --------------------- O
(Deformed) ^
V /-------- * Fi A Ml I _____ . i
^
-*
F>
-O I ____ I
A 2
Fig. 7 -4 . W ork done on the m ass p a rticle s and inte rn a l restraints.
deformation of the restraints. The restraint force acting on a particle is equal in magnitude, but opposite in sense, to the reaction of the particle on the restraint. Since the points of application coincide, it follows that ' dWD = - dWj (b)
As an illustration, consider the simple system shown in Fig. 7-4. F or this case, we have dWD F i A U i + F ) A u 2 dW[ =
F
Aui
F,
A u2
156
V A R IA T IO N A L PR IN C IPLES FOR AN ID E A L TRUSS
CHAP. 7
U sing (b), we can w rite (a) a s : dWE = dWD 9=1,2 for a rb itra ry Auq (7 6
Also, the general principle o f v irtual displacem ents can be expressed as follows: T he first-o rd er w ork done by th e extern al forces is equal to th e first-o rd er w ork d o n e by the in tern al forces acting on th e restrain ts for any a rb itra ry v irtual displacem ent o f a system of particles from an equilibrium position. W e em phasize again th a t (7 -6 ) is ju s t an a ltern ate statem en t o f the force equilibrium co n d itio n s for th e system . Som e a u th o rs refer to (7 -6 ) as the w ork eq uation. T o apply the principle o f v irtual displacem ents to a n ideal truss, we co n sider th e jo in ts to be m ass po in ts an d th e b ars to be in tern al restrain ts. W e have defined 0* a n d as the colu m n m atrices o f external jo in t lo ad s an d corresp o n d in g jo in t displacem ents. T hen, dW E = 0>T M t (a) w here A<7/ co n tain s the v irtual jo in t displacem ents. T h e first-o rd er w ork d o n e by the restrain t forces acting o n b ar n due to the virtu al displacem ents is f (dW D )n = F n den G eneralizing (b), wc have dW D = F T de F inally, th e w ork e q u atio n for an ideal tru ss has th e form tsfll = F r de for a rb itra ry A m (7 -7 ) (c) (b)
T h e scalar force-equilibrium eq u atio n s are o b ta in e d by su b stitu tin g for de in term s o f ISPU. It is convenient to first establish th e expression for th e differential elo n g atio n o f an individual b a r an d th en assem ble de. O p eratin g on en, e = n(u+ - u_) + - (u+ - u J T(u+ - u J an d n o tin g th e definition o f P (see (6 -2 2 )), we o b ta in den = [a + y - ( u + - u J T] (Au+ - Au_) = P(Au+ - Au_) (7 -8 )
t % = F de Wd(e). We must use the rules for forming the differentials of a compound function since en depends on the joint displacements. Using (3-17), we can write
SEC. 7 - 2 .
PR IN C IP LE OF V IR T U A L D IS P LA C E M E N TS
157
T he assem bled form follows from (6-25). W e ju st have to replace y w ith P: de - t # 1 A4 ?/ S u b stitu tin g for de in (7 -7 ), 2Pr m = F 'W (a) (7 -9 )
and requirin g (a) to be satisfied for arb itra ry tsPll results in th e jo in t force force equilibrium equations. F o r the geom etrically lin ear case, e = w here s tf is c o n s t a n t an d de = A ll follows directly from e. W e have treated the geom etrically non lin ear case here to show th a t th e principle o f virtu al displacem ents leads to forceequilibrium eq u atio n s w hich are co nsistent w ith the geom etrical assum ptions associated w ith th e d efo rm atio n -d isp lacem en t relations. Example 7 - 1 ------------------------------------------------------------------------------------
We consider a rigid member subjected to a prescribed force, P, and reactions R h R 2, as in the diagram. There is no internal work since the body is rigid. Introducing the virtual
Fig. E7-1
-------- 1 ' '
Ait]
An | Ali
*2
---
----------
displacements shown above, and evaluating the first-order work.

d W = dW K =
Rj (
&u1 +
R 2 Aw2 _ P Aiij,
(a)
Now, A up is not independent: A u . = A u j I I ----- 1 + A n ,
d\
p
Then,
d W = Au1 \ r x -
V
P \1 -
LJ
VL
(b)
0 | + Au 2 | p 2 - P f 0 j = 0
(c)
Requiring (c) to be satisfied for arbitrary A u u A a 2 leads to
ds
(A )
R2 = P -
which are the force and moment equilibrium equations, in that order.
158
V A R IA TIO N A L PRINCIPLES FOR AN IDEAL TRUSS
CHAP. 7
Example 7 -2
We consider the outside bars to be rigid (see sketch). To obtain the force equilibrium equation relating P and the internal bar forces F l; F 2, we introduce a virtual displacement, Au L, of the point of application of P :
dW E = P dW D = F 1 d e t + F2 dez
(a)
The first-order increments in the elongations are

d e t = Aux cos 0 d e 2 = Au2 cos 6 = A u y cos 6
where 0 defines the initial position. Then, equating d W E and d W ^

dWE = dW D
(b)
for arbitrary A (c)

cos
V
P = ( F t F 2) 6
The force in bar 3 does not appear explicitly in the equilibrium equation, (c). It is possible
Aw2
Aw i
Fig. E7-2
to include F3 even though bar 3 is rigid by treating it as a Lagrange multiplier.f We consider Am 2 as independent in the work equation:
P A
- (Fx cos 0 ) A uj + (F, cos 0)Ai/2 = 0
(d)
Now,
Ai/ j Au2 = 0
(e)
Multiplying the constraint relation by A, adding the result to (d), and collecting terms, we obtain A u ^ P F, cos 6 A) + A u 2(F 2 cos 0 + A) = 0 (f) Finally, we require (f) to be satisfied for arbitrary Awt and Aw2- The equilibrium equations
See Sec. 3-3.
SEC. 7 -3 .
PRINCIPLE OF VIR TU AL FORCES
159
are
P F3 cos 0 + A
F2 c o s 6 + I
and we recognize /. as the force in bar 3.
(g)
7-3.
PRINCIPLE OF VIRTUAL FORCES
T he principle o f virtual forces is basically an altern ate statem ent of geo m etrical com patibility. W e develop it here by o p eratin g on the e lo n g a tio n jo in t displacem ent relations. L ater, in C h ap ter 10, wc generalize the principle for a three-dim ensional solid an d describe an altern ate derivation. W e restrict this discussion to geometric linearity. T he governing equations are 0* = # F (a) e = (b)
N ow , we visualize a set of b a r forces AF, and jo in t loads, A ^9, w hich satisfy the forcc-equilibrium eq u atio n s: A ,# = AF
(c)
A force system which satisfies the equations o f static equilibrium is said to be statically permissible. E q u atio n (b) relates the actual elongations an d jo in t d is placem ents. If we m ultiply the eq u atio n for ek by AF*, sum over the bars, and note (c), we o b tain the result A F r e = A F T{ m T()n ) = t\0> Tm which is the definition of the principle of virtual forces: T he actual elongations an d jo in t displacem ents satisfy the condition A FTe = 0 (7-1 0 )
(d)
for any statically perm issible system of b a r forces and jo in t loads. T he principle o f virtual forces is independent of m aterial behavior but is restricted to the geom etrically linear case. T he statically perm issible system (AF, A # ) is called a virtual-force system. T o illustrate th e ap p lication of this principle, we express U and A ^ in p artitio n ed form, A P, Hi A ^ => AP = (a) u 2 A P2 where U 2 co n tain s th e prescribed su p p o rt m ovem ents. U sing (a), (7 -1 0 ) takes the form : A Fr e - A P [ 0 2 - A P [U X (b)
160
V A R IA T IO N A L P R IN C IP LE S FOR AN ID E A L TR USS
C H AP. 7
If th e elo n g atio n s are know n, we can d eterm in e th e u n k n o w n disp lacem en ts by specializing A P ^ T o determ in e a p a rtic u la r d isp lacem en t co m p o n en t, say ukj, we g en erate a force system co nsisting o f a u n it value o f pkj an d a set o f b a r forces an d reactio n s w hich eq u ilib rate pkj = 1. A1 = F U , - (7 -1 1 )
T h e in tern al b a r forces an d reactio n s a re o b ta in from an eq u ilib riu m analysis o f a statically d eterm in a te stru ctu re. Since only o n e elem ent o f A P i is finite, A P f U 1 => (l)ufcj and (b) reduces to
% = eTF|li, l- U / P 2|,w (7-12)
(c)
T h e p rin cip le o f v irtu al forces is also used to estab lish g eo m etric co m p a ti bility rela tio n s req u ired in the force m e th o d w hich is discussed in C h ap ters 9 and 17. W e o u tlin e th e ap p ro ac h h ere for com pleteness. O n e w o rk s w ith self-equilibrating virtual-force system s, i.e., statically perm issible force system s w hich involve only b a r forces an d reactions. By definition, a self-equilibrating force system F * , P * satisfies
B jF * = P* = 0
P f = b 2f *
F o r this case, (b) reduces to er AF U I A P2 = 0 (7 -1 4 )
E q u a tio n (7 -1 4 ) represents a re stric tio n on th e elo n g atio n s an d is called a geometric compatibility eq u ation.
Example 7 -3 ---------------------------------------------------------------------------------------The truss shown (Fig. E 7-3A ) has support movements (Hu 32. lh ) and is subjected to a loading which results in elongations ( e u e 2) in the diagonal bars. We are considering the outside bars to be rigid.
Fig. E 7 - 3 A
SEC. 7 - 3 .
PR IN C IP LE OF V IR T U A L FORCES
161
To determine the translation, u, we select a statically determinate force system consisting of a unit force in the direction o f u and a set o f bar forces and reactions required to equilibrate the force. O ne possible choice is shown in Fig. E 7-3B . Evaluating (7-12) leads to
u = - + t7,
cos 6
- tan 0(713 - (i2 )
This truss is statically indeterminate to the first degree. A convenient choice of force redundant is one of the diagonal bar forces, say F 2. The equation which determines F 2 is
Fig. E7-3B
derived from the geometric compatibility relation, which, in turn, is obtained by taking a self-equilibrating force system consisting of F 2 = + 1 and a set of bar forces and reactions required for equilibrium. The forces are shown in Fig. E 7-3C .
cos 0
Fig. E7-3C
O
Evaluating (7-14), we obtain
e y + ez 0
0
(a)
To show that (a) represents a geometrical compatibility requirement, we note that the elongation-displacement relations for the diagonal bars are
e x ' u cos 0 ez ~ - u cosO
(bj
Specifying c, determines it and also e 2 We could have arrived at Equation (a) starting from Equation (b) rather than (7-14). However, (7-14) is more convenient since it does not involve any algebraic manipulation. We discuss this topic in depth later in Chapter 9.
162 7-4.
V A R IATIO N A L PRINCIPLES FOR AN IDEAL TRUSS
CHAP. 7
STRAIN ENERGY; PRINCIPLE OF STATIONARY POTENTIAL ENERGY
In this section, we specialize the principle of virtual displacem ents for elastic behavior an d establish from it a variational principle for the jo in t displacem ents. We start with the general form developed in Sec. 7 -2 , F r de = g>T Am for arb itrary A ?/ (a)
If we consider all th e elements o f U to be arbitrary, i.e., unrestrained, de = m (b)
and (a) leads to the com plete set of forcc-equilibrium equations in unp artitio n ed form,
(c)
W e can obtain the equation for P , by rearranging (c) o r by startin g with the p artitio n ed form of 0>T 0>T A<9/ => P 7 AU - P [ A U i + P I A 0 2 and noting th at A U 2 = 0 since U 2 is prescribed. T he reduced form is F r cie P f AU! = 0 w here now de - B[ A U j + B l A U 2 ^ B[ A U t In w hat follows, we will w ork with (7-15). O u r objective is to interpret (7-15) as the stationary requirem ent for a function of U t. W e consider F to be a function of e, w here e = e(U,). The form of F F(e) depends on the m aterial behavior, t W e could express F in term s o f U j b u t it is m ore convenient to consider F as a com pound function of e. T he essential step involves defining a function, VT Fr (e), according to F 7 dc = W ith this definition, an d letting
dCj
(d)
for arb itrary AUj
(7-15)
= dVT
(7-16)
nP= vTwe can write (7-15) as tillp = 0
p [u t = n F(Uj
for arb itrary A U i
(7-17) (7-18)
We call VT the to tal strain energy function and F1P the total potential energy. O ne should note th a t VT exists only when F is a continuous single-valued function of e. This requirem ent is satisfied w hen the m aterial is elastic. E quation (7 - 18) states th a t the jo in t force-equilibrium equations (Pi BjF) expressed in term s of the unknow n displacem ents are the E uler equations for the
f See Secs. 6 - 4 , 6 - 5 .
SEC. 7 -4 .
STRAIN ENERGY
163
total potential energy. It follows th a t the actual displacem ents, i.e., the dis placem ents which satisfy the equilibrium equations, correspond to a stationary value of n r It rem ains to discuss how one generates the strain-energy function. By definition, d V r-^d V j (a) and dVj = Fj dej (b) w here Vj is the strain energy for b a r j . Since we are considering pound function o f E quation (b) is equivalent to F M - f y
Vj
to be a com
(7-19)
T h at is, the strain energy function for a b ar has the property that its derivative with respect to the elongation is the b ar force expressed in term s of the elonga tion. Finally, we can express Vj as (7-20) *5 - L / ' * ' where e0 is the initial elongation, i.e., the elongation n o t associated w ith the force. Actually, the low er limit can be taken arbitrarily. This choice corresponds to taking Vj as the area between the F-e curve and the e axis, as shown in Fig.7 -5 . F
Fig. 7-5. G ra p h ic a l re p re s e n ta tio n of s tra in e n e rg y and c o m p le m e n ta ry e ne rg y, W e consider the linearly elastic case. U sing (6-30),
Fj =
k M j
e o ,j)
(a)
Then . Vj = h H ei ~ eo,j)2 (7-21) The total strain energy is obtained by sum m ing over the bars. W e can express VT as
U 17 1
Vt -
I Vj = i( e - e o ^ e - c0) ;= i
(7-22)
164
V A R IA T IO N A L PR IN C IPLES FOR AN ID EAL TRUSS
CHAP. 7
Finally, we su b stitu te for e in term s of U l5 U 2, using e - A jU i + A 20 2 (7 -2 3 )
W hen the geom etry is linear, A 1; A2 are co n stan t an d VT is a q u ad ratic function. If the geom etry is nonlinear, VT is a fourth degree function of the displacem ents. U p to this point, we have show n th a t the displacem ents defining an equilib rium position correspond to a statio n ary value o f the potential energy function. T o determ ine the ch aracter (relative m axim um , relative m inim um , indifferent, neutral) o f the statio n ary point, we m ust exam ine the behavior of the second differential, d2TLp, in the n eighborhood o f the statio n ary point. O p eratin g on d n p, and noting th a t AP x 0 leads to d2Tlp = d{dll p) = d 2 VT (7 -2 4 ) dz VT Y, (dF) de.j + Fjd2ej) j T h e next step involves expressing d 2 VT as a q u ad ratic form in A U ^ We restrict this discussion to linear behavior (both physical and geom etrical). T he general non lin ear case is discussed in Sec. 17.6 W hen the geom etry is linear, we can operate directly on (7-23) to generate the differentials o f e, de = A j AUj d2e 0 since A j is constant. W hen the m aterial is linear, dV = k de (b) ( .
w here k is a diagonal m atrix containing the stiffness factors (AE/L) for the bars. Then, d2 VT reduces to d2VT - d F T de = deTk de (1-25) = A U ftA lk A J A U , If de ^ 0 for all nontrivial A U j, d2Vr is positive definite and th e statio n ary p o in t is a relative m inim um . This criterion is satisfied when the system is initially stable, since de = 0 for A t^ # 0 w ould require that A x AU! 0 (m eq u atio n s in nd unknow ns) (a)
have a nontrivial solution. But a n o ntrivial solution of (a) is possible only w hen r(A t ) < nd. H owever, A! = for the geom etrically linear case an d r(Bt) = nd w hen the system is initially stable. Therefore, it follows th a t the displacem ents defining the equilibrium position for a stable linear system correspond to an absolute minimum value o f the potential energy. ----------------- - ---------- - -----Example 7 -4 - -------------------- -
W e e s t a b lis h t h e t o t a l p o t e n t ia l e n e r g y f u n c t i o n fo r t h e tr u s s c o n s id e r e d in E x a m p le 7 - 2 . F or c o n v e n ie n c e , w e assu m e no in itia l e lo n g a tio n or su p p o rt m o v e m e n t. T h e s tr a in
SEC. 7 - 5 .
e n e r g y is
C O M P LE M E N T A R Y ENERGY
165
VT =
+ k2e\)
(a)
S u b s t it u t in g f o r t h e e lo n g a t i o n s in t e r m s o f t h e d is p la c e m e n t , e y = r e s u lt s in VT = a n d fin a lly n ,, = T h e fir s t d iffe r e n t ia l o f is j(k i + k 2) u j c o s 2 6 (c) Wj c o s 0 e2 u2 cos 0 cos 0 (b )
i (ky
k 2)uj
co s2 0 -
(d )
d n p = {[(/<!
4-
k2)c o s 2 (?>!
P ^ A u j
(e)
R e q u ir in g T lp t o b e s t a t io n a r y le a d s t o th e E u le r e q u a t io n , P i w h ic h is j u s t t h e f o r c e - e q u i l i b r i u m = [(^ i + k 2) c o s 2 0 ] u , (f)
e q u a tio n
(F i
P 2) c o s 0
(g )
w ith th e b a r fo r c e s e x p r e s s e d in te r m s o f th e d is p la c e m e n t u s in g
F j = k tet ku{ c o s 0
T h e s e c o n d d i f f e r e n t i a l o f n p is
f 2 = k2c2
fe2ut
cos
(h )
< 1 zn p =
a n d w e se e th a t th e s o lu tio n ,
[(/c j
+ k 7)c o s 2
0 ](A i)2
( i)
u, =
(/ct + c o r r e s p o n d s to a n a b s o lu te m in im u m w hen 0 = 0.
Pi
k 2) c o s z 0
(j)
0 . T h e t r u s s is i n i t i a l l y u n s t a b l e
v a lu e o f n p w h e n 6 #
7 -5 .
COMPLEMENTARY ENERGY; PRINCIPLE OF STATIONARY COMPLEMENTARY ENERGY
The principle o f v irtual forces can be transform ed to a v ariatio n al principle for the force red u n d an ts. W e describe in this section how one effects the tra n s form ation an d utilize the principle later in C h a p te r 9. T his discussion is restricted to linear geom etry. We sta rt w ith E q u atio n s (7 -1 3 ) an d (7-14), w hich we list below for co n venience : eT A F - U l A P 2 = 0 (a) where AF, A P 2 represent a self-equilibrating force- system , i.e., they satisfy the the following co n strain t relatio n s: AF = 0 A P2 - B 2 AF (b)
(c)
166
VA R IATIO N A L PRINCIPLES FOR AN IDEAL TRUSS
CHAP. 7
O u r objective is to establish a function of F, whose Euler equations are (a) and (b). We cannot w ork only with (a) since F is n o t arbitrary b u t is constrained by the force-equilibrium equations, P ! = B jF (nd equations in rn variables) (d)
W e interpret e 7 AF as the first differential of a function V* K*(F), dV$ = eT AF - d V f (7-26)
and call V f the com plem entary energy function for bar j. By definition,
e/Fj) = -Jr v!
(7-27)
T h at is, the com plem entary energy function for a bar has the property th a t its derivative with respect to the bar force is the elongation expressed in term s of the force. We express V f as vt = f e>dF> <7 - 28)
This definition corresponds to taking V f as the area bounded by the F-e curve and the F axis as shown in Fig. 7 -5 . Also, the strain and com plem entary energy functions are related by V} + V f = Fjej W hen the m aterial is linear elastic, (7-29)
o.j + f j F j Vj = *o j F j + i f j f } VT = e j F + j F TfF
ej =
(7-30)
Next, we define I I f as: IT. - V f - V I P 2 = v * - u b 2f (7-31)
We call IT^F) the total com plem entary energy function. W ith these definitions, E quations (a), (b), and (c) can be interpreted as rfllc - 0 subject to the constraint condition d(P, - B ^ - 0 (f) (e)
We can com bine (e) and (f) into a single equation by introducing Lagrange multipliers. Follow ing the procedure described in Sec. 3 -3 , we add to (7-31) the jo in t force equilibrium equations and write the result a s :
n; - n c + (Pt
- B xF )r a
(7-32)
where a = {al5 a 2, . . . , oc d} contains the Lagrange multipliers. T he Euler equa
SEC. 7 -5 .
COM PLEM ENTARY ENERGY
167
tions for IT treating F and a as independent variables are

d n 'c 0
for AF, Aa arbitrary (7-33)
1 1 e(F) = B [a + B jU 2 B iF = P ,
We recognize the first equation in (7-33) as the m em ber force-displacement relation, and it follows th a t a = An alternate approach involves first solving the force-equilibrium equation, (d). There are nd equations in m variables. Since Bi is of rank nd when the system is initially stable, we can solve for nd bar forces in term s of P t and the remaining (m ud) bar forces. O ne can also w ork with a com bination of bar forces and reactions as force unknowns. We let q m nd = num ber of redundant forces X = { X l7 X 2, . . . , X q} m atrix o f force redundants and write the solution of the force-equilibrium equations as F = F 0 + FxX ? ! ~ 1*2. 0 + 1*2, *X The force system corresponding to X j is self-equilibrating, i.e., (B jF JX = 0 for arb itrary X (7-36) (7-35) (7 _ 34)
We substitute for F in (7-31) and transform n c(F) to ITC (X). Then, rfIIt(X) - eT AF - C 2 r A P2 = (er F , - U 2 r P 2>,)AX and the Euler equations are er F , - U 2 r P 2. , - 0 (7-37) 8
N ote that (7-37) is ju st a reduced form of (7-33). Also, we could have obtained this result by substituting directly in (a). ' U p to this point, we have shown that the force redundants which satisfy the geom etric com patibility equations correspond to a stationary value of the total com plem entary energy. To investigate the character of the stationary point, we evaluate the second differential. O perating on (g), d2U c = de F* AX = d2V$ (h)
If d2V% is positive definite with regard to AX, the stationary point is a relative minimum. This requirem ent is satisfied for the linear elastic case. To show this, we note that
de ~ f AF = fF x AX
'
d2V$ = ! _ / , - ( = AXT(F jfF ,)A X
(ll
168
VAR IATIO N AL PRINCIPLES FOR AN IDEAL TRUSS
CHAP. 7
Since f contains only positive elements, d V f is positive definite with regard to AX provided that there does n o t exist a nontrivial solution of F , AX = AF = 0 (J)
F o r (j) to have a nontrivial solution, there m ust be at least one relation between the colum ns of F x. But this w ould correspond to taking force redundants which are n o t independent, and the solution scheme w ould degenerate. Therefore, we can state th a t the actual force redundants correspond to an absolute m ini m um value of n c for the linear elastic case.
Example 7 -5 -----------------------------------------------------------------------------------We consider the truss treated in Example 7 -3 . It is statically indeterminate to the first degree with respect to the bars (statically determinate with respect to the reactions) and we
The force influence matrices defined by (7-35) follow from the force results listed on the sketches: F 0 P {l/cos 0; 0; 0; 0; tan 0; 0} F* = ( + 1; 4-1; cos 0; - s i n 0; sin 0 ; cos 0} f*2,o = P{ 1; tan 0 ; + tan 0} P2,.v = 0 Assuming a bar is rigid is equivalent to setting / = 0 for (he bar. Then, the comple mentary energy is due only to the diagonal bars:
V*r = V* + V*
= ^0. 1^1 + ^0.2^2 + l i f l F l + fzFl)

We convert V * to a function of X by substituting
(C)
F2 =
+X
(d)
Finally, n c(AT) has the form

2
Differentiating (e) leads to <iric - 1 *0, 1 + e o. 2 + / i jTgJ + (A + A ) ^ | a x d 2n c = (A +

m a x
(f)
(g)
)2
SEC. 7 -6 .
STA BILITY CRITERIA
169
The Euler equation follows from (f):
Comparing (h) with (a) of Example 7 -3 , we see that the Euler equation for n c(X) is the geometric compatibility equation expressed in terms of the force redundant.
7-6.
STABILITY CRITERIA
Section 6 - 9 dealt with initial stability, i.e., stability of a system under in finitesimal load. We showed there th a t initial stability is related to rigid body m otion. A system is said to be initially unstable when the displacem ent restraints are insufficient to prevent rigid body m otion. In this section, we develop criteria for stability of a system under finite loading. If a linear system is initially stable, it is also stable under a finite loading. However, a nonlinear (either physical or geom etrical) system can bccom c unstable under a finite load. We consider first a single mass particle subjected to a system of forces which are in equilibrium . Let u be the displacem ent vector defining the equilibrium position. W e introduce a differential displacem ent A it, and let AW be the work done by the forces during the displacem ent Aw. If A W > 0, the particle energy is increased and m otion w ould ensue. It follows th at the equilibrium position (u) is stable only when A W < 0 for arb itrary Ail We consider next a system of particles interconnected by internal restraints. Let A WE be the increm ental w ork done by the external forces and A W, the increm ental w ork done by the internal restraint forces acting on the particles. The total w ork, A W , is given by A W = AW e + AH',
(a)
The system is stable when A W < 0 for all arb itrary perm issible displacem ent increm ents, th a t is, for arb itrary increm ents of the variable displacem ents. Now, we let AIô he the w ork done by the internal restraint forces acting on the restraints. Since A WD = A W h we can express the stability requirem ent as A I^d - A WE > 0 (7-38)
One can interpret A WDas the w ork required to deform the system to the alternate position and AW E as the actual w ork done on the system. W hen the behavior is continuous, we can express A WD an d AWE as T aylor series expansions in term s o f the displacem ent increm ents (see (7 -2 )): A WE = dWE + \ d 2WE + AW = dWD + - y 2WD +
(b)
We have show n th at the first-order w ork is zero at an equilibrium position: dWD - dWE = 0
(c)
170
VARIATIONAL PRINCIPLES FOR AN IDEAL TRUSS
CHAP. 7
If we re ta in only th e first tw o te rm s in (b), th e general stab ility c o n d itio n re duces to d 2WD - d 2WE > 0 for all a rb itra ry perm issib le d isp lacem en t increm ents (7 -3 9 ) E q u a tio n (7 -3 9 ) is called th e classical stab ility criterio n . R etain in g only th e first tw o differentials co rre sp o n d s to co n sid erin g only infinitesimal d isp lace m e n t in crem ents. If (7 -3 9 ) is satisfied, th e eq u ilib riu m p o sitio n is stab le w ith respect to an infinitesim al d istu rb a n ce. In o rd e r to d eterm in e w h eth er it is stab le w ith respect to a finite d istu rb a n ce, one m u st use (7 -3 8 ). If d 2WD = d 2WE (7 -4 0 ) for a p a rtic u la r set o f d isp lacem en t in crem ents, th e eq u ilib riu m p o sitio n is said to be neutral, a n d th e re exists an a lte rn a te eq u ilib riu m p o sitio n infinitesim ally disp laced from th e first p o sitio n . O n e can in te rp re t (7 -4 0 ) as th e necessary c o n d itio n for a b ifu rca tio n o f eq u ilib riu m positions. T o show this, su p p o se U an d U rep resen t th e d isp lacem en t co m p o n en ts for th e tw o possible eq u ilib riu m p o sitio n s o f a system w here U = U + AU (a)
Also, let R an d R rep resen t th e resu ltan t forces co rre sp o n d in g to U a n d Cl. W e can express R as R = R + d R + i d 2R + - (b) N ow , th e seco n d -o rd er w o rk for th e in itial eq u ilib riu m p o sitio n is given by d 2 W = d 2 WE d 2 WD = A U r d R (c)
If d 2W 0 fo r so m e finite AU, it follow s th a t dR = R fj A U = 0 T h e c o n d itio n IM = 0 (e) (d)
is eq u iv alen t to (7 -4 0 ). F inally, if we co n sid er A U to be infinitesim al,
R = R + dR (f) a n d (7 -4 0 ) im plies R = 0. T o ap p ly th e classical stab ility crite rio n to a n ideal truss, we n o te th a t the first-o rd er w o rk term s have the form
dWE ~ P [ A U , dW D = Y jF j dej w here V 2, P i are prescribed. O p e ra tin g o n (a) yields
(7-41)
d 2WD = l [ F j d 2ej + dFj deJ ]
SEC. 7 - 6 .
STABILITY CRITERIA
171
an d the stab ility criterio n reduces to stab le n e u tra l u n stab le d 2WD > 0 d 2WD = 0 d 2WD < 0 for a rb itra ry n o n triv ial AU! for a p a rtic u la r n o n triv ial for a p a rtic u la r n o n triv ial A U t (7 -4 2 )
w here d2WD is a q u a d ra tic form in A U ^ W e p o stp o n e discussing how one transform s (7 -4 1 ) to a q u a d ra tic form in A U j u n til th e n ex t chapter. W hen th e m a teria l is elastic, we can identify (7 -3 9 ) as th e req u irem en t th a t n p be a relative m inim um . By definition, d l l p = dVT - dWE F o r elastic b ehavior, dVT = dW D and it follows th a t d z WD - d 2WF r = d zn l l .p (7 -4 3 ) Finally, we can state: An equ ilib riu m p o sitio n for a n elastic system is stab le (neutral, u n stable) if it co rresp o n d s to a relative m in im u m (neutral, indifferent) sta tio n a ry p o in t o f th e to ta l p o te n tial energy. (a)
-------Example 7 -6 ------------------------------------------------------------------------------- :
The system shown in Fig. E7-6A consists of a rigid bar restrained by a linear elastic spring which can translate freely in the x 2 direction. Points A and A denote the initial and deformed positions. We will first employ the principle of virtual displacements to establish the equilibrium relations and then investigate the stability of the system. Fig. E7-6A
U 2 Pi - A'
The first-order w ork terms are
dWD = F de dWF = P , du2
(a)
172
V A R IA T IO N A L PRINCIPLES FOR AN ID EA L TRUSS
CHAP. 7
where F , e are the spring force and extension. Since the bar is rigid, the system has only one degree of freedom, i.e., only one displacement measure is required to define the configura tion. It is convenient to take 0 as the displacement measure. The deformation-displacement relations follow from the sketch:
e =
= L(sin 8 sin B0)

d 0 cos 0 )
t u 2 = L(cos
Then,
F k e /cL(sin 0 sin 9 0)
(c)
and
d e = (cos 0 )L A d d u 2 (sin 9 )L AO
(d)
Using (a) and (d), the principle of virtual displacements takes the form
d W D d W E = { F cos 9 P 2 sin 0 } (L AO) =
for arbitrary A 9
(e)
Finally, (e) leads to the equilibrium relation,

F cos 0 = P 2 sin 0
(f)
which is just the moment equilibrium condition with respect to point O . We transform (f) to an equation for 0 by substituting for F using (c). The result is
/ p \
sin 0 - ( - - ) tan 6 = sin 0 o
\KLj
(g)
Since the system is elastic,

dW n dW E = d llp
and (e) is equivalent to rfllp = 0 for arbitrary AO
<h)
(0
The potential energy function for this system has the form n P = %ke2 P 2 u 2 \ k L 2(sin 0 sin 9 0 ) 2 P 2 ^ (c o s 0 o cos 9) and (g) can be interpreted as
d Y lP
(J)
= 0 => Eq. (g)
(k)
Curves o f (P 2/ k L ) vs. 8 for various values of 0 O are plotted in Fig. E 7-6B . The result for 0 o = 0 consists of two curves, defined by 0 = 0 cos 0 = P 2/ k l for arbitrary P 2/k L for (P 2j k h ) < 1 (1)
To investigate the stability of an equilibrium position, we have to evaluate the secondorder work at the position. After some algebraic manipulation, we obtain
d 2 Tlp = d 2 W0 d 2 WE = k (L AO ) 2
cos 9 P 2 /k L cos 6
(m )
Let 9* represent a solution o f (g). Applying (m) to 9* results in the following classification:
REFERENCES
3 FI cosJ 0* >
173
stable neutral unstable One can show that (n) is equivalent to stable neutral unstable
kL
cos3 0*
3 flt
Pi
kL
P i
(n)
cosJ 8 * <
kL
dP2
He > 0
dP2
dQ
= 0
(o)
< o dO
A transition from stable to unstable equilibrium occurs at point A, the peak of the loaddeflection curve. The solution for 0 o = 0 is different in that its stable segment is the lin e a r solution and the neutral equilibrium point [ P 2 k L ) corresponds to a branch point. Both the linear and nonlinear branches are unstable.
Fig. E7-6B Il
kL
REFERENCES
1. 2.
3. 4. W a n g , C. T .: A p p li e d E la s tic ity , McGraw-Hill, New York, 1953. L a n g h a a r , H. L .: E n e r g y M e th o d s in A p p lie d M e c h a n ic s , Wiley, New York, 1962. R e s s n e r , E . : On a Variational Theorem in Elasticity, J. M a th . P h y s ., Vol. 29, pages 9 0 - 9 5 , 1 9 5 0 . A r g y r i s , J. H., a n d S. K e l s e y : E n e r g y T h e o re m s a n d S tr u c tu r a l A n a ly s is , Butterworths, London, I960. . C h a r l t o n , T. M .: E n e r g y P r in c ip le s in A p p lie d S ta tic s , Blackie, London, 1959. H o f f , N. J .: T h e A n a ly s is o f S tr u c tu r e s , Wiley, & New York, 1956. W a S H I Z U ,K .: V a r ia tio n a l M e th o d s in E la s tic ity a n d P la s t ic i ty , Pergamon Press, 1968.
5. 6.
7.
174
V A R IA T IO N A L PR IN C IPLES FOR AN ID EAL TRUSS
CHAP. 7
PROBLEMS
7 - 1 . C onsider the tw o-dim ensional sym m etrical truss show n. A ssum e u2 = u 3 = 0. (a) D eterm ine the first two differentials o f f , and e2 by o p eratin g on the expanded expression (equation 6 19) for e. (b) W hen a b , we can neglect th e n o n lin ear term involving u 12 in the expressions for e and p. Specialize (a) for this case. (c) W hen a b, we can neglect th e n o n lin ear term involving ux t in the expressions for e and p. Specialize (a) for this case. Prob. 7-1
7 - 2 . Refer to the figure o f P ro b . 7 -1 . A ssum e ii2 = u3 = 0 an d a b. U sing the principle o f virtual displacem ents, determ ine the scalar force-equili brium eq u atio n s for jo in t 1. 7 - 3 . Suppose a force F is expressed in term s o f e, F - CjlC + K 2*3 w here e is related to the independent variable u by e = u + hu2 (a) (b) (a)
D eterm ine th e first tw o differentials o f th e w ork function, W = W{u), defined by W = r Fde Jo
(b)
S uppose (a) applies for increasing e and F = C i(e - eg) (c)
for e decreasing from e*. D eterm ine d 2W a t e = e*. 7 - 4 . Refer to P rob. 6 -2 3 . T he n I independent n o d e eq u atio n s relating the b ranch currents are represented by A 7i = 0 (a)
N ow , the b ranch potential differences, e, are related to the n 1 independent node potentials, V, by e = AV (b)
P R O BLEM S
175
D educe th a t th e requirem ent,

i T de 0
for a rb itra ry AV
(c)
is equivalent to (a). C o m p are this principle with the principle o f virtu al dis placem ents for an ideal truss. 7 -5 . C onsider the tw o-dim ensional truss shown. A ssum e u 2 u* 0. (a) U sing (7-14), o b tain a relation betw een the elongations an d u32. T ake the virtual-force system as A F 2 and th e necessary b ar forces an d re actions required to eq u ilib rate A F 2. (b) U sing (7-12), express i<n , ui2 in term s o f e u e3. N o te th a t b ar 2 is n o t needed. O ne should always w ork w ith a statically determ inate system w hen applying (7-12).
Prob. 7 - 5
7 -6 . Refer to P ro b . 6 -2 3 . O ne can develop a v ariatio n al principle sim ilar to the principle of v irtual forces by o p eratin g on th e b ranch potential differ e n c e -n o d e p o ten tial relations. Show th a t
A ir e = 0 (a)
for any permissible set o f cu rren t increm ents. N o te th a t the currents m u st satisfy the node equations
A ri = 0
D educe K irchhoffs law (the sum o f the voltage d ro p s aro u n d a closed lo o p m ust equal zero) by suitably specializing Ai in (a). Illu strate for the circuit shown in P rob. 6 - 6, using branches 1, 2, 4, an d 6 . 7 -7 . By definition, the first differential of the strain-energy function due to an increm ent in U j has the form
m m
dVT = dVn = F n den n= 1 n= 1
(a)
We w ork with VT expressed as a co m p o u n d function of e = e(U) since it is m ore convenient th a n expressing V directly in term s o f U t . O ne can also
176
V A R IA T IO N A L PR IN C IP LES FOR AN ID EA L TRUSS
C H AP. 7
w rite dVT as dVT = VT^ M J x (a) (b) U sing (b), show th a t the system of ij jo in t force-equilibrium eq u atio n s expressed in term s o f the jo in t displacem ents can be w ritten as: dVT _ daa (b) ^ lk t 1, 2, . . . ,7 k = 1, 2, . . . , i
E q u atio n c . is called C astig lian o s principle, p a rt I. Show th a t an altern ate form o f (c) is Pik = m de Y F" = 1 ou(k (d)
N o te th a t (d) is ju st the expansion o f (c). R ew ork P ro b . 7 - 2 , using (d). 7 - 8 . D eterm in e V(e), dV, an d d 2V for th e case w here th e stress-strain relatio n has the form (sec P ro b . 6 -1 0 ) a = (e b3) 7 - 9 . D eterm ine V*(F), dV*, an d d 2V* for the case w here the stress-strain relatio n has the form ' 7 -1 0 .
= -(< ?
1/ h
C O 3)
3\
Show th a t (7 -1 2 ) can be w ritten as
u k j = pkj n t
w here n c - n c(P j) is defined by (7-31). This resu lt specialized for U 2 = 0 is called C astig lian o s principle, p a rt II. A pply it to P ro b . 7 -5 , p a rt b. A ssum e linear elastic m aterial and f l = j ' 2 = f 2 = f . 7 -1 1 . T he c u rren t an d p o ten tial d ro p for a linear resistance are related by ej = eo ,j + R jh In v ertin g (a), we can express ij as a function o f ej. ij = R j ' ^ e j - e0 J ) (a) S uppose we define a function, Wfej), w hich has th e p ro p erty th a t dWj dej D eterm ine W) correspo n d in g to (b). (b) L et W = Y j Wj w here b to tal n u m b e r of branches. C on sid erin g j mi the b ranch p o ten tial d ro p s to be functions o f the n o d e p o ten tials, deduce th a t th e actu al n o d e p o te n tials V co rresp o n d to a statio n ary value of W. Use the results o f P ro b . 7 -4 . T he E uler e q u atio n s for ij(ej) (c) (b) (a)
P R O B LE M S
177
(c)
W W (V ) are th e node cu rren t equ ilib riu m eq u atio n s expressed in term s o f the n o d e potentials. S uppose we define a function w hich has the p ro p erty th a t
D eterm in e W f co rresp o nd in g to (a).

b
(d)
L et W * =
W J.
Show th a t the E uler e q u atio n s for W * = ir (AV) W * =
n
(e)
= iTe -
n(i,V)
(e)
are the governing eq u atio n s for a d-c netw ork. Show th a t th e actu al cu rren ts co rresp o n d to a statio n ary value o f W*. O n e can either in tro d u ce the co n strain t co n d itio n , A r i = 0, in (e) o r use the result o f P ro b . 7 -6 . 7 -1 2 . In v estig ate the stability of the system show n below. T a k e / c = a L 2ks
P
Prob. 7-12
L in e a r tra n s la tio n a l re s tra in t R ig id rod
k r (L in e a r ro ta tio n a l re s tra in t)
//////// and consid er a to ran g e from 0 to 6.
8 Displacement Method Ideal Truss

8 -1 . GENERAL
T h e basic eq u atio n s defining th e b eh av io r o f an ideal truss consist o f forceequilibrium eq u atio n s an d force-displacem ent relations. O n e can reduce the system to a set o f eq u atio n s involving only th e u n k n o w n jo in t displacem ents by su b stitu tin g th e force-displacem ent relatio n s into the forcc-equilibrium equations. This p artic u la r m eth o d o f so lu tio n is called the displacement or stiffness m ethod. A lternatively, one can, by elim in atin g th e displacem ents, reduce th e governing eq u atio n s to a set of eq u a tio n s involving certain b ar forces. T h e la tte r p ro ced u re is referred to as th e force o r flexibility m ethod. W e em phasize th a t these tw o m eth o d s are just alte rn a te p ro ced u res for solving the sam e basic eq u atio n s. T h e displacem ent m e th o d is easier to au to m a te th a n the force m eth o d an d has a w ider ran g e o f ap p licatio n . H ow ever, it is a co m p u ter-b ased m ethod, i.e., it is not suited for h an d c o m p u tatio n . In contrast, th e force m e th o d is m o re suited to h a n d c o m p u ta tio n th a n to m achine com p u tatio n . In w h at follows, we first develop th e e q u atio n s for th e d isp lacem en t m ethod by o p eratin g on th e governing eq u atio n s expressed in p a rtitio n ed form . W e then describe a p ro ced u re for assem bling th e necessary system m atrices using only the connectivity table. T his p ro ced u re follows n atu ra lly if one first op erates on the u n p a rtitio n e d eq u atio n s an d then in tro d u ces th e d isp lacem en t restraints. T he rem ain in g p o rtio n of th e c h ap ter is devoted to the tre a tm e n t o f n o n lin ear behavior. W e o u tlin e an in crem en tal analysis p ro ced u re, ap p ly the classical stability criterio n , an d finally, discuss linearized stab ility analysis.
8-2.
OPERATION ON THE PARTITIONED EQUATIONS
T h e gov ern in g p artitio n ed e q u atio n s for an ideal tru ss a re developed in Sec. 6 - 7 . F o r convenience, w e sum m arize these e q u atio n s below.
178
SEC. 8 - 2 .
O P E R A T IO N ON THE P A R TITIO N E D E Q U A T IO N S
179
P i = B jF P 2 = B 2F F = F { 4- k A 1U l F ; = k( e 0 + A 20 2)
(nd eqs.) (r eqs.) (m eq s.)
(a) (b) (c) (d)
T h e u n k n o w n s are th e m b a r forces (F), th e r reac tio n s (P 2), a n d th e nd jo in t d isp lacem en ts ( U : ). O n e can co n sid er F, to rep resen t the in itial b a r forces, th a t is, th e b a r forces d u e to th e initial elo n g atio n s an d s u p p o rt m o v em en ts w ith U j = 0. T h e te r m k A tU j rep resen ts the b a r forces due to U t . W h en th e m aterial is linear elastic, k an d e 0 a re co n stan t. A lso, Aj = B j w h en th e g eo m etry is linear. W e o b ta in a set o f nd e q u a tio n s rela tin g th e nd d isp lacem en t u n k n o w n s, U j, by su b stitu tin g for F in (a). T h e resu ltin g m a trix e q u a tio n h as th e form (B ik A iJU j = P , - B ^ , (8 -1 )
W e solve (8 -1 ) for L^, d e term in e F from (c), a n d P 2 from (b). T h e coefficient m atrix for is called th e system stiffness m atrix a n d w ritten as K u - B ,k A , (8 -2 )
O n e c a n in te rp re t B 1F i as rep resen tin g th e in itial jo in t forces d u e to th e in itial elo n g atio n s a n d s u p p o rt m o v e m en ts w ith U ] = 0 . T h en P j - B jF ; rep resen ts the net u n b alan ced jo in t forces. W h en the g eo m etry is lin ear, K n reduces to K u = B jk B [ = A^kAi (8 -3 )
If th e m a te ria l is linear, k is c o n s ta n t a n d po sitiv e definite for real m aterials. T hen, th e stiffness m a trix for th e lin ear case is positive definite w hen th e system is initially stable, th a t is, w hen r ( B J = M d. t C onversely, if it is n o t positive definite, the system is in itially u nstable. If th e m a te ria l is n o n lin e a r, k a n d e0 d ep en d o n e. W e h av e em p lo y ed a piecew ise lin ear re p re se n ta tio n for th e force-elo n g atio n cu rv e w hich resu lts in lin ear relatio n s. H ow ever, o n e h as to ite ra te w hen th e lim itin g elo n g atio n for a segm ent is exceeded. T h e g eo m etrically n o n lin e a r case is m o re difficult since b o th A a n d B d ep en d on U j. O n e can iterate o n (8 -1 ), b u t th is req u ire s solving a n o n sy m m e trica l system o f eq u atio n s. It is m o re efficient to tra n sfo rm (8 -1 ) to a sy m m etrical system by tran sferrin g so m e n o n lin e a r term s to th e rig h t-h a n d side. N o n lin e a r analysis p ro ced u res are tre a te d in Sec. 8 - 4 . Even w hen th e b eh av io r is co m pletely linear, th e p ro ced u re o u tlin e d ab o v e for g en era tin g th e system m a trice s is n o t efficient for a larg e stru c tu re , since
t See Prob. 2-14.
180
DISPLACEM ENT M ETHOD: IDEAL TRUSS
CHAP. 8
it requires the m ultiplication o flarg e sparse matrices. F o r example, one obtains the system stiffness m atrix by evaluating the triple m atrix product, K n = ' A fkA i (a)
O ne can take account of sym m etry and the fact th a t k is diagonal, but is generally quite sparse. Therefore, w hat is needed is a m ethod o f generating K which does n o t involve m ultiplication of large sparse matrices. A m ethod which has proven to be extremely efficient is described in the next section.
8-3.
THE DIRECT STIFFNESS METHOD
Wc start w ith (6-37), the force-displacem ent relation for bar n : F n = Fo.n + V /nV ~ F0,n ~ (a)
where w+, n_ denote the joints at the positive and negative ends of bar n. One can consider F0 as the bar force due to the initial elongation with the ends fixed (u,l4. = u = 0). N ow , we let p +1p be the external joint force matrices required to equilibrate the action of F. N oting (6-43), we see th at P*+ = Pn- = ~Pn + S ubstituting for F, (8 -4 ) expands to Pn+ = PTÔ.n + /cnP,TYnU+ ~ Pn~ Pni ^ (8-4)
O ne can interpret (b) as end a c tio n -jo in t displacem ent relations since the elem ents of FP,y are the com ponents of the b ar force with respect to the basic frame. C ontinuing, we let k = K f ny (8 -5 ) N o te th at k is o f order i x i where i = 2 or 3 for a two or three-dim ensional truss, respectively. W hen the geom etry is linear, p = y = oc and k is sym m etrical. W ith this n o tation, (b) takes a m ore com pact form,
P+ = KFo.n + ku, - ku_ pn. = - pr/'o ,n - kU,1 + + kun.
(S-6)
W e refer to k as the b ar stiffness matrix. E q u atio n (8 - 6) defines the jo in t forces required for b ar n. T he to tal jo in t forces required are obtained by sum m ing over the bars.
SEC. 8 -3 .
THE DIRECT STIFFNESS METHOD
181
W e have defined
= { p i , p 2, . . ,P ;}
% (ij x
I)
= {ut, u 2, . . . , u,}
(y x 1)
as the general external jo in t force an d jo in t displacem ent matrices. N ow , we write the com plete system of i j jo in t force-equilibrium equations, expressed in terms of the displacem ents, as (8-7) We refer to y f , which is of o rder ij x ij, as the u n r e s t r a i n e d system stiffness matrix. T he elem ents of a re the required jo in t forces due to the initial elongations and C/Ctfl represents the required jo in t forces due to the jo in t displacem ents. W e assem ble '/T and . ^ 0 in p artitioned form, w orking with successive m em bers. T he contributions for m em ber n follow directly from (8- 6).
0 P O( P a r t i t i o n e d F o r m I s j x
1)
+ F 0, npi y ~ / ' o, , K
J
( P a r titio n e d F o rm Is j
in row
n+
in row n_
X j)
+ k k,t k + kn
in row in row
n +,
colum n
n+
in row n+, colum n n_ colum n n+ in row row nn_, column n nn_ _ in _, colum
(8 -9 )
Example 8 -1 ------- - --------------------------- ------ ----------
The connectivity table and general form of -Y f and f P 0 for the numbering shown in Fig. E8-1 are presented below:
Fig. E8-1
182
D IS P LA C E M E N T M E T H O D : ID EA L TRUSS
C H AP. 8
Bar + join t -j o in t
1 1 4
2 2 1
3 2 3
4 4 3
5 2 4
u3
u4
Pi P2 P3 P4
kj - k 2
k2
^2 k2 + ~k3 k3 + k5 ~ k 3 k3 + - k 4 k4
- k i k5 k4 kj + k4 + k5
-k ,
~"k5
1 < Po. 2 Po. 3 Po, 4

Po,
-Fo. iur - Fn. 2 P2 + ^ 0, ^0, - ^ 0 . 301 - F o , 4PI - ^ 0 . iPT + F o , 4 PI
Fo.sP? >
Fo.
slS
Example 8 -2
The external force matrix, p ;, involves u, and the displacement matrices for those joints connected to joint j by bars. Now, p, corresponds to row j and Uj to column j of X . By suitably numbering the jo in ts , one can restrict the finite elements o f to a zone about the diagonal. This is quite desirable from a computational point of view.
Fig. E 8 - 2
S ect. 1
N
T -------------------- * ---- 7
I 3
\k
NT ! \ ! ! A J
1 1 i 1 :> ------ 1 -------------* \ 4
\
I 7 ----- 1s5> ------1 ---------------- *-----HO -------------*

f
H ---------------- * L -i
\ 1
Consider the structure shown. We group the vertical joints into sections. The equi librium equations for section k involve only the joints in section k and the adjacent sections. For example, the equations for section 3 (which correspond to p5, p6) will involve only the displacement matrices for sections 2, 3, 4. This suggests that we number the joints by se c tio n . The unpartitioned stiffness matrix corresponding to the above numbering scheme
SEC. 8 - 3 .
THE D IR EC T STIFFN ESS M ETH O D
18 3
is listed below. N ote that ,iC has the form of a quasi-tridiagonal band matrix when it is partitioned according to s e c tio n s rather than individual joints. The submatrices for this truss are of order 4 x 4 .
Ul Pi Pz k, + k 2 - k i -ki u3 k 2 -k 4 u4 5 u u7 Us
ki + k 3 1 - k 3 + k4 j j k2+ k3
P3
~ k2
-k 3
-k 5
-k*
k 7
+k5 +k7 '+ k 6 i J* 1 P4 k4 + k 5 j +k. 1 Ps -k 6 1 k 7 j 0 J* 1 k 6 -t-k9 kg -ks { U " >
+ k l0 Pe -k . 1 -k 9 kg+k7 + k9 + k i j j + k l2 P7 kj o k u 1 j k 10 + k u ~ k i3 k u ~ k 12
j + k 13 Ps k l2 | k 13 k i 2 + ki 3
T h e in tro d u ctio n of d isp lacem en t restrain ts involves first tran sfo rm in g th e p artitio n e d elem ents o f an d .3T to local fram es associated w ith th e restrain ts, p erm u tin g th e actu al row s, an d finally p artitio n in g the actu al row s. T h e steps are indicated below.
W e w rite the system o f jo in t force-equilibrium e q u atio n s referred to th e local jo in t fram es as #>J = t f W * + (8 - 10) T h e tra n sfo rm a tio n law s for th e su b m atrices of J T an d SP0 follow from (6-5 7 ). &%,n = R 0nP o ,n 3HTin = R 0C J*r(nR 0n' T
t,n =
1, 2 , . . . J
(8 -1 1 )
184
D IS P LA C E M E N T M ETH O D : ID EA L TRUSS
C H AP. 8
T he step, 0*3 -+ P, involves only a rearran g em en t of the row s of SP}. W e o b tain th e corresponding stiffness m atrix, K, by perform ing th e s a m e o p eratio n s on b o th the rows an d c o l u m n s of ;XTJ. T he rearran g ed system o f eq u atio n s is w ritten as P = KU + P0 (8 -1 2 ) Finally, we express (8 -1 2 ) in p artitio n ed form : ? ! = K 11U I + K 12U 2 + P 0;1 P 2 = K 21U j + k 22u 2 + P 0. 2 T he first eq u atio n in (8-1 3 ) is identical to (8-1). Example 8 -3 ---------------------------------------------------------It is of interest to express the partitioned elements of K in terms of the geometrical, connectivity, and displacement transformation matrices. We start with the general un partitioned equations (6-28), (6-40), and (6-44), (6-50):
0* = 3 8 F = C rpTF
(8 -1 3 )
(a)
(b)
F = F 0 + k s / % = F 0 + kyO ff Then, substituting for F in (a) and equating the result to (8-7) leads to JT = C r(pTky)C ^ 0 = CT(pTFo)
(c)
(d)
The matrix, Prky, is a quasi-diagonal matrix of order im. The diagonal submatrices are of order i, and the submatrix at location n has the form, We have defined this product as k. Then, if we let
X kB
*r k,
/mPIyi M 2Y2 /( Rr y
(e)
we can express JjT as
jr - cT kR c
(0
Carrying out (8-9) for n = 1 , 2 , . . . , m is the same as evaluating the triple matrix product. Obviously, (8-9) is more efficient than (f). The introduction o f displacement restraints can be represented as P = D&>
Pj =
P2 = D 2& and
(g)
*U = D rU = D fU t + D [U 2
(h)
SEC. 8 -3 .
THE DIRECT STIFFNESS M ETHOD
185
Substituting (g) and (h) in (8-7) and equating the result to (8-13), we obtain Kst = D p T D ( r = D sC rpTkyCD,T Po., = = D ,C rpke0
s,t - 1,2
(i)
In o rd er to o b ta in (8-13), we m u st rearran g e the row s an d colum ns of 2fC and then partitio n . This o p eratio n is quite tim e-consum ing. Also, it leads to rectangular subm atrices. In w hat follows, we describe a pro ced u re for in tro ducing displacem ent restrain ts which avoids these difficulties. We s ta rt w ith th e com plete system o f eq u atio n s referred to the basic frame, :/fU = - <3% = < 4 > n (8 -1 4 )
We assem ble an d jT , using (8 - 8) and (8 -9 ). Then, we ad d to ??0 the external force m atrices for those jo in ts which are unrestrained. It rem ains to modify the row s an d colum ns corresp o n d in g to jo in ts w hich are either fully o r partially restrained.
Case A : Full Restraint
S uppose U q Uq. T h en pq is unknow n. W e replace the eq uation for pfl by u = u, This involves the follow ing o p eratio n s on the subm atrices o f X an d 0*^.
1. O n y f . Set off diagonal m atrix elem ents in row q an d colum n q equal to 0 and the d iagonal m atrix elem ent equal to I ;.
(8 -1 5 )
2.
On
A dd term s in iPN d u e to uq: & n, r =

P 7 > N,q =
r fiq
(8 - 16)
t* q
I = 1,2 , . . . J
Case B : Partial Restraint Local Frame
We suppose the rth element in u^ is prescribed.

t^,. = T P qt prescribed p qr = unknow n
186
D IS P LA C E M E N T M E T H O D : ID E A L TRUSS
CHAP. 8
W e have to delete the eq u atio n co rresp o n d in g to p^r an d replace it w ith
Step I Assemblage o f Basic M atrices W e assem ble E g, G g, u*, p* acco rd in g to the follow ing: 1. E q an d G 9, W e s ta rt with E - I; and we set En. 0 Grr + 1 G = 0;
2. u*. W e start w ith an ith -o rd e r co lu m n v ecto r having zero elem ents and we set the elem ent in the rth row equal to uq qr. N o te th a t this m atrix involves only the prescribed displacements (local frame) in th eir n atu ra l locations. 3. p*. W e s ta rt w ith an /th -o rd er colum n vector h aving zero elem ents and we insert th e values o f the prescribed jo in t forces (local frame) in their n a tu ra l locations. N o te th a t the elem ents co rresp o n d in g to th e reactio n s are zero. W hen th e jo in t is fully restrained, E = 0, G = I,
u* - u
p* - 0j
Suppose jo in t 5 is partially restrained. T he d a ta consist of: (a) (b) T h e ro ta tio n m atrix, R 05, defining th e d irectio n of th e local fram e a t 5 w ith respect to th e basic frame. T h e d irectio n (or directions) of th e displacem ent re stra in t an d the value (or values) o f th e prescribed displacem ent. d irectio n r, iuf,. (c) T h e values o f th e p rescrib ed jo in t fo rces: Plj j = r j 1 ,..., i
As an illu stratio n , suppose r = 2. T hen, in (b), we read in r 2 In (c), we read in Pi 1 T h e four basic m atrices are (for r 2) 'l E5 = 0 0 0 0 0 P53 u \2
o"
0 1 g 5 =
0 0 0
0 1 0
o 0 0
SEC. 8 -3 .
TH E D IR EC T STIFFN ESS M ETH O D 0
187
l *52
P i; Ps = < 0
P 53
I I
Step 2 Operation on XT and (PN 1. P rem u ltip ly row q o f an d 0 > N by E^R04.

t =
= E gR0^ P Stq - e , r ^ , 9 2. P o stm u ltip ly co lu m n g o f
P N,c = & * , ' ~ * = i , 2, . . . , ; 3. P o stm u ltip ly colum n g of MT<q = ^ 4. A dd G g to . i f ^

VI ~ <tq +
O ?(E9R0)T
1 , 2 , . . . , . /
(a)
by R 0< J' Tu* a n d ad d to R ?' r <
N-
(b)
by (E9R0q)r ^ = 1 ,2 ,... ,j
(c)
(d)
5.
A dd + p* an d + u * to ^ N , q
N .q
0 *N ,q
U q
P*
(e)
T he o p e ra tio n on row q an d colum n q are su m m arized below. On
jsr
j r = EqR.jrqe -? fC q = ^ q ( E gR0g)T J T = (E,,R(V ^ ( E 9R0,)r + G q C^ q I = 1, 2, . . . , . / (8 -1 7 )
= 0>n j - x r lq R09, r u* I 1,2, . . . , j
(8 -1 8 )
* Y , - (ER0' ) ^ < J - (E9R 0qp r MR ^ Tu* + P* + < W hen is sym m etrical (this will be the case w hen th e system is geom etrically linear), we can w o rk only with the subm atrices on an d above the diagonal. T h e co n tra cted o p eratio n s for th e sym m etrical case are threefold: & v . t = & n, i - ^ R 9' r u* W t q = - I ftq(FJqR Q iq )T I = 1, 2, . . . , q 1
(8-19)
188
D IS P LA C E M E N T M E T H O D : ID E A L TR USS
CHAP. 8
& N, q = (E ,R 0q)3% :q - (EqR V qqR q' T< s r qq = (EqR 0qy r q(l(Equ 0qy + G q
& N .t
= 0 > N , t t f q T <R0q T
+ P* + <
(8 -20 )
V*
(8 -21)
yirql = (E ,R 09) j r af t q + 1. q -\- 2 , . . . . j
T he o p eratio n s o u tlin ed ab o v e are carried o u t for each restrain ed jo in t. N o te th a t th e m odifications for jo in t 4 involve only row q an d co lu m n q. We d en o te th e m odified system o f e q u atio n s by
( 8 - 22 )
E q u atio n (8 -2 2 ) represents ij eq u atio n s. T h e coefficient m atrix J T* will be nonsingular w hen K u is n o n sin g u lar. T o show this, we s ta rt w ith th e first eq u atio n in (8 -1 3 ) a n d an a d d itio n a l set o f r- du m m y eq u atio n s:
E q u a tio n (a) represents ij eq u atio n s. T his system is tran sfo rm ed to (8 -2 2 ) w hen we p erm u te U , P/v to 0 J N. T hey are related by (sec (6-63)) U = m /J 0>} = I T P w here I I is a p erm u ta tio n m atrix . It follow s th a t (b)
(c)
= n Tp $ and, since I I is an o rth o g o n a l m atrix, |j r * | = |k u | (8-2 3)
It is m o re co n v en ien t to w ork w ith (8 -2 2 ) ra th e r th a n (a) since th e solution o f (8 - 22) yields the jo in t d isp lacem en t m atrices listed in th e ir n a tu ra l order, th a t is, acco rd in g to increasing jo in t nu m b er. O nce Ql3 is kn o w n , we convert th e jo in t d isp lacem en t m atrices to the basic fram e, using
T h e b a r forces are d eterm in ed from F = F 0t + kyn(ufI+ - u_) N ext, we calcu late F n p I an d assem ble in p a rtitio n e d form by su m m in g the
SEC. 8 -3 .
THE DIRECT STIFFN ESS M ETH O D
189
co n trib u tio n for each m em ber. F o r m em ber n, we p u t (see (8 -4 )) in row n+ in row n O nce is know n, we co n v ert the force m atrix for each p artially restrain ed jo in t to the local jo in t reference fram e, using
= r0 < ,P ?
T he final resu lt is req u ired to eq u ilib rate the b a r forces. T his o p eratio n provides a static check o n the so lu tio n in ad d itio n to furnishing the reactions. W hen th e p ro b lem is geom etrically n o n lin ear, y, an d k depend o n the jo in t displacem ents. In this case, it is generally m o re efficient to ap p ly an increm ental fo rm u latio n ra th e r th a n iterate on (8 - 22).
Example 8 -4 ---------------------------------------------------------------------------------------W e illu strate these o p eratio n s for th e tru ss show n in Fig. E 8 -4 .
c
*2
\ /
\ X
N
/ / / \
Fig. E8-4
.....
Xi i r/
' /
/V s
1.
M em ber-Joint C onnectivity Table

Bar (n)
+Joint (n+)
1 1 2
2 3 ]
3 1 4
4 3 2
5 4 2
6 3 4
7 5 3
8 3 6
9 5 4
10 6 4
11 5
6
joint (n_)
2.
A ssem blage o f
We consider the geometry to be linear. Then, p = a and k = fcaja. Applying (8-9) results in :'Jf listed below.
190
CHAP. 8
\ 1 2 3
1 k! + k 2 + k 3 - k j -k , -k 2
2 -k 2
3 k3
0 k! + k 4 + k 5 - k 4 -k 4 k 2 4* k4 + kg + k 7-t-k8 4 k5 -k 6 k 3 + k , + k6 - k 9 1 *kg + k 1 0 5 0 6 "ks k 10 k u + kg + kio + k j, -k 7 kg + k7 + kg + k ,l k n " kjo _ k$

- K
k 7
- k8
N ote that 3.
is symmetrical and quasi-tridiagonal, with submatrices of order 4 x 4 .

I n tr o d u c tio n o f J o in t D is p la c e m e n t R e s tr a in ts
The original equations are
y c m = 0> - P J > = & > s

where 2P contains the e x te r n a l joint forces. We start with If joint q is un restrained, we put p, in row q of &>N. If joint q is fully restrained, we modify and & y according to (8-15) and (8-16). Finally, if joint q is partially restrained, we use (8-19) through (8-21). Since .J f is symmetrical, we have to list only the submatriccs on and above the diagonal. It is convenient to work with successive joint numbers. For this system, joint 2 is fully restrained and joints 4, 6 are partially restrained. The basic matrices for joints 4,6 and the initial and final forms of J4T, are listed below. Note that this procedure does not destroy the banding of the stiffness matrix.
J o in t 4
(w42 is prescribed)
R04 = 12
0 0 1
r _.
1 O' _0 0 u* = {0, n42}

J o in t 6
G* =
L 1 P* = {p4.110}
(uf, 2 is prescribed)
SEC. 8 4.
IN C R E M E N T A L FO R M U LA T IO N
I n itia l m a tr ic e s ( J f and
(l) (l) (3) 13 ^23
^33
191
SP^)
(s) K )
K )
14
11
y f \2 X~21
*1*
34
n . 1 ^ 0. 2
^36 ^46
0,3 0.4
X 'u J C 45
S3 'm
X 55
% 56
'^ 6 6
0. 5
- n .,
F in a l m a tr ic e s (J T * and .^jj)
1 (l) ( u 2.) 0 (u 3) (114) ( u 5) K )
3C i d E *
0 ......... 0 0
-^0. l+ p .
3 T , 21l 2 - J r t 4 U j
1,
-* '3 3
jfj.E *
^35
jr ',6(E6R06)7 '
.4 * 0
J^~3 aV* Tu t +
P3
- j r 3(iR 06'
4Cft
44E4
E 4 X 45
E4-jr46(E6R06)T
+ p! +
0 . 4 - ^ 24h 2) u*
+ G4
E^.tf" 44U4 - E4.y/"4ftR0S w .
Sy n.
5s
. ^ 56(E6RnT
+ p5- . ^ 56R06 ru,t (E6R06).#"6f,(E{1 R0f,)r + G(;

e 6r 0#(
- ^ 0. 6 - J T j 6uJ
- j r 6#R0-T uj)+p s+u ?
8 -4 .
INCREMENTAL FORMULATION; CLASSICAL STABILITY CRITERION
E q u atio n s (8-13), (8 -2 2 ) arc valid for b o th linear an d n o n lin ear behavior. H ow ever, it is m o re efficient with respect to c o m p u tatio n al effort to em ploy an increm ental fo rm u latio n w hen the system is n onlinear. W ith an increm ental form ulation, one applies the load in increm ents an d determ ines the co rre sp onding incremental displacem ents. T he to tal displacem ent is o b tain ed by sum m ing the displacem ent increm ents. A n increm ental toading pro ced u re can also be used w ith (8 -1 3 ) but, in this case, one is w orking w ith to tal displacem ent ra th e r th a n w ith increm ental displacem ent. In this section, we develop a set of eq u atio n s relatin g th e increm ental external load an d the resulting increm ental displacem ents. These eq u atio n s are also no n lin ear, but if one w orks w ith sm all load increm ents, the eq u atio n s can be linearized. O u r ap p ro ach will be sim ilar to th a t followed previously. W e first establish increm ental m em ber force-displacem ent relatio n s a n d then apply th e direct stiffness m ethod to
192
CHAP. 8
generate the increm ental system equations. W e com plete the section w ith a discussion of th e classical stability criterion. W e s ta rt w ith (8-4), which defines the external jo in t forces required to equ ilib rate the actio n of the force for b ar n, P+ = F p J p_ = - p n+ (a)
E q u atio n s (a) are satisfied a t an equilibrium position. W e suppose an in crem ental external load AP is applied an d define AU as th e resulting increm ental displacem ent for th e new eq u ilibrium position. Since F an d p depend on U, their values will change. L etting AF, AP be the to tal increm ents in F, P due to AU, an d requiring (a) to be satisfied at b o th positions, leads to the following increm ental force-equilibrium eq u atio n s: Ap+ = F API + AFpJ + AF Ap^ Ap_ = Apn+ (8-2 4 )
T o proceed further, we need to evaluate the increm ents in e an d p. T h e exact relations are given by (6 - 22): en = a(un+ - u_) + i(p - a n)(u+ - u.) (i. i L/n - u_)r (a)
T o allow for the possibility of retaining only certain n o n lin ear term s, we w rite (a) as /} - a = (un+ - u_)Tgn e = a(un+ - u J + i{u+ - unJ Tg(u+ - u .) = If all th e n o n lin ear term s are retained, 2n Ij (8 -2 5 )
T o neglect a p a rticu la r displacem ent com ponent, we delete the corresponding elem ent in . F o r geom etrically linear behavior, g 0. O p eratin g on (8-25), we o b tain Ap = dP = (Au+ - Au_)Tgn (8-2 6 ) and Aen = de + \ d 2en den = P(Au+ - Au_) d2e - c/p(Aun+ - Au_) It rem ains to evaluate A F. W e allow for a piecewise linear m aterial and em ploy the re la tio n s! developed in Sec. 6 -4 . F o r convenience, we d ro p all the
f See (6-31), (6-32), and (6-33).
(8-27)
SEC. 8 - 4 .
IN C R EM EN TAL F O R M U LA TIO N
193
n o tatio n pertaining to a segm ent an d w rite the generalized increm ental expression in the sim ple form A F = k(Ae - Ae0) (8 -2 8 )
where k, Ae0 are co n stan t for a segm ent. T hey have to be changed if the lim it of the segm ent is exceeded o r th e b a r is unloading. Since Ae is unknow n, one has to iterate, taking th e values of k, Ae0 corresp o n d in g to the initial equilibrium position as the first estim ate. T his is equivalent to using the tangent stiffness. The initial elongation, Ae0, is included to allow for an increm ental tem p eratu re change. S ubstituting for Ae, (8 -2 8 ) takes th e form A F n = A F q> + k de + \ k n d2en A F0, = - K Ae0i (8 -2 9 )
Finally, we substitute for AF, Ap in (8 -2 4 ) an d g ro u p the term s as follows: Apn+ = k (t(Au+ - Au_) + Ap0, + Apff, Ap. = Ap+ where k.n = F ngn + /C ^Pn Ap0, = Ap,n = k ( - A e 0. n d ffi 4- { d 2enp,T + den d $ + h d 2en d p j) We interpret k, as the tangent stiffness m atrix. T he vector, Apg, contains linear, quadratic, and cubic term s in Au. W e have included th e subscript g to indicate that it is a n o n lin ear geom etric term. We w rite the to tal set of increm ental jo in t equilibrium equations as A9> M l + A ^ o + A0>g (8 -3 2 ) (8 -3 1 ) (8 -3 0 )
where ,%~t is assem bled using (8 -9 ) and A0>o + A0>g w ith (8 - 8). N o te that is sym m etrical. Finally, we introduce the displacem ent restrain ts by a p plying (H 19) (8 21). T he m odified eq u atio n s are J T f* A ilJ = A0>* - A0>% - A0>* (8 -3 3 )
It is convenient to include the prescribed increm ental su p p o rt displacem ent terms in A8P* so th a t A3P* involves only the increm ental tem perature and A t he variable displacem ent increm ents. T he co n tracted equations are Kr> ii A U i A Pj APo, i AP3i i Kf_i2 A U 2 (8 34)
where Kt| n is sym m etrical. We can n o t solve (8-3 3 ) directly for A 6 ?/ since A c o n t a i n s q u ad ratic an d cubic term s in A1 ?/. T here are a num b er o f techniques for solving n o n lin ear algebraic equations, t W e describe here the m eth o d o f successive substitutions,
t See Ref. 12.
194
D IS P LA C E M E N T M E T H O D : ID E A L TR USS
CHAP. 8
w hich is th e easiest to im p lem en t, b u t its convergence rate is slow er in com pariso n to m ost o f the o th e r m ethods, t F irst, we n o te th a t J lT * A P/> *, an d A 3 P % are in d ep en d en t of A W l. They de pen d only on th e initial eq u ilibrium position an d the in crem en tal loading. We com bine A.SP* an d A .ô an d w rite (8 -3 3 ) as W ? A m J = A .^v - A .^* N ow , we let A U {n) represent th e wth estim ate for (n -f- l) th estim ate by solving jT *
A U A r f/J
(8-35) an d determ in e the
("+ =
A@>%
A<P*
[A *
(8 -3 6 )
T h e iteratio n involves only evalu atio n o f A&>* an d b ack -su b stitu tio n once J f * is transform ed to a trian g u lar m atrix. T he facto r m eth o d is p articu larly co n venient since JT,* is sym m etrical. W ith this m ethod, J f * => S r S w here S is an u p p er trian g u lar m atrix. W e rcplace (8 -3 6 ) with S A * - > - Q S Q = In linearized in crem en tal analysis, we delete A iP * in (8 -3 5 ) an d tak e the solu tio n of # * A W J = A & > % (8 -3 9 ) as the actu al displacem ent increm ent. O n e can in terp ret this schem e as one cycle successive substitu tio n . T he so lu tio n degenerates w hen th e ta n g en t stiffness m atrix becom es singular. T o investigate the beh av io r in the n eig h b o rh o o d of this po in t, we apply the classical stability criterio n developed in Sec. 7 -6 . T he a p p ro p ria te form for a truss is given by (7 -4 1 ): m . d 2WD = (F n d2e + dFn de ) > 0 n- 1 for a rb itra ry A U Xw ith A U 2 = 0 (a) (8 -3 7 )
W e have alread y evaluated the above term s. U sing (8 -2 6 ), (8 -2 7 ), an d (8 -2 9 ) w ith A e 0 = 0, . F d2en + dFn den =>(AuI1 + - A u .)r k(,(Aun+ - Au_) (b)
a n d (a) can be w ritten as d 2W D = A U fK ft u A U i > 0 for a rb itra ry A U X (8 -4 0 )
I t follows th a t K t n m ust be positive definite for a stable equ ilib riu m position.
f Iterative techniques are discussed in greater detail in Secs. 18-7, 18-8, 18-9.
SEC. 8 - 4 .
IN C R E M E N T A L FO R M U LA T IO N
195
But K (i u and CfCf are related by
:/f? = rT
Km, 0
0 L
(8 -4 1 )
w here II is a n o n sin g u lar p erm u tatio n m atrix, w hich rearran g es the elem ents of ls.%1 according to
au = n m J
(8-42)
Then, J f * an d Kfi u have th e same definiteness p ro p e rty .! Finally, we can classify the stability o f an equilibrium p o sitio n in term s o f th e d eterm in an t of the tangen t stiffness m a trix : J D = \.yrt\ = |K r. u D > 0 D = 0 D <
stable neutral unstable
(8-43)
Example 8 -5
We illustrate the application of both the total (8-13) and incremental (8-34) formulations to the truss shown in Fig. E8-5A. To simplify the analysis, we suppose the material is linearly elastic, = k2 k, and there are no initial elongations or support movement. b d Fig. E8-5A
The initial direction cosines for the bars are = { -0 ~b] 2 = j [ ~ ^ - fc] (a)
The deformed geometric measures are defined by (8-25). They reduce to

P =
e =
+ ufg.
Y n iii n = 1, 2
(b)
y = a + |u[g for this example. Since b d, we can neglect the nonlinear terms due toMu , i.e., we can take _ _ I 0 0 (c) gn * L 0 1
f See Prob. 2 -1 3 for a proof, j See Sec. 2 -5 .
196
U s i n g (c),
DISPLACEM ENT M ETHOD: IDEAL TRUSS
CHAP. 8
Yi [< *
'!
+ m12]
Pi * -*
[d
- b
(d)
72 - r [ - rf I - b + & 1 2 ]
P2 ~
! b + h12]
C o n t in u in g , th e b a r fo r c e - d is p la c e m e n t r e la tio n s a r e
F = ke = /cyu,
F in a lly , th e fo r c e - e q u ilib r iu m e q u a tio n
n = 1, 2
(e)
fo r j o in t 1 fo llo w s b y a p p ly in g ( 8 - 6 ) to b o t h b a r s .
Pi =
Om
+ k 2)U i =
/c (P [Y i +
P l Y i ) ! '!
(f)
E q u a t i o n s (e) a n d ( f) e x p a n d t o
P1
P 12
and
2k [ j )
0 .
jui 1
(g)
I ~ j(b - uu )(b - \ u 12)
i 2k
Fi y ( ^ " i i (b 2uiz)uii)
k
(h) *12)^
u , L in th e t h a t ( g ) is
Fl = X (_c/Wl1
T h e d ia g o n a l fo rm
o f t h e c o e f f i c i e n t m a t r i x is d u e t o t h e f a c t t h a t w c n e g le c t e d p. T h is a p p r o x im a tio n u n c o u p le s th e e q u a tio n s . N o te
e x p r e s s io n s fo r y a n d
t h e f i r s t e q u a t i o n i n ( 8 - 1 3 ) w i t h U 2 a n d P 0> ( s e t t o 0 . S o l v i n g t h e f i r s t e q u a t i o n t in (g ), w e o b t a i n
(i)
T h e c o r r e s p o n d in g b a r fo rc e s a re
P11
O')
F2 = - f i
T h is r e s u lt is a c t u a lly th e s o lu t io n fo r th e lin e a r g e o m e tr ic c a s e .
T h e e x p r e s s io n fo r u , 2 a n d th e c o r r e s p o n d in g b a r fo r c e s fo llo w fro m th e s e c o n d e q u a tio n in (g ).
P 12
2k 2 (b L
u l2 )(b -
j u 12)
t Equation (g) is (8 i) with F i = 0.
SEC. 8 -4 .
i?
INCREM ENTAL FORMULATION

f
k tu (6 \ ^ J Pi2 J u 12) i2 = - r --------
197
F2 =
2 b u 12
(1 )
We can write (k) as

Pi;
*1 2
L
2 k (b u l2 )(b -
i 12)
(m)
and solve (m) by iteration. Alternatively, one can specify 12 and evaluate p 1 2 from (k). The latter approach works only when there is one variable. The solution is plotted in Fig. E8-5B.
12
Fig. E8-5B
2 vO
We describe next the generation of the incremental equations which follow from (8 -2 6 )(8-32). Applying (8-26), (8-27) to (b)-(d) results in
AP., = Wn =
Ati i2 M
L
+ u l z )A u n
de = p Au, = ocni A u u + y ( ~ b
d 2 en = Aufg,, Au, = -t-(Au12)2
(n)
A
= +
A
~
n = 1, 2
We arc assuming no initial elongation. Then, AF = k A e n = k(den + \ d 2 e n) (o)
The tangent stiffness matrix and incremental geometric load term are defined by (8-31). Using (n), we obtain
kf xrt\ K,n =
Sym
t -~
-{t-b + u ^ 2 + ~
(P)
2L
/An V AU12/
A u A k 12 + -> M u i i V 2 ( j J . ( b + , ( Au, !(12) 4- j A u j 2 1 - j )
(q)
Tgg
D ISP LA C E M E N T M E TH O D : ID EAL TRUSS
CHAP. 8
Finally, we assemble the incremental equilibrium equations for joint 1 using (8-30). Apj = (k(, ! + k,i2)Aui + Ap^, i + A p,.2 (r)
4
fd \2 2k 0
Au,, I
+ F 2)
, 1 /F . ( b + u i2)~ + ~^(F i
2k
< Au J
|_____________ 0_ ______________ | + l ^ + Au^ \
(s)
Apn ' A puJ
Note that (s) is (8-34). Also, the incremental equations are uncoupled. We restrict the analysis to only /;I2 loading. Setting F l = F 2 in (s) results in
j - ( f + ^ - ( - 6 + 12)2^
Au l 2 = A p i 2
+ w,W
(0
where F is determined from (e). The coefficient of Au 1 2 is the tangent stiffness with rcspcct to u l 2 . tip , 2 2 k ,
, v + - - ( - b + ul2)2
dut2
I \
(u)
Applying the classical stability criterion (8-43) to (t), we see that > 0
d p 12 d u 12
stable neutral unstable
- 0 < 0
(v)
Points A , B are stability transition points and the segment A -B is unstable. If k = 0, the truss is neutral with respect to An . Now there is a discontinuity in k at F F ebl the pin-ended Euler load, when the material is linearly elastic:
1*1 < F",

F = -F
cb
L
k = 0
(w)
7t2 / 3 ~ lF
To determine whether the members buckle before point A is reached, we compare F A with F eb. Using (u),
AE Fa = A Eb2
~ i ( ~ b + wi2
L-
3L
(x)
Then, for system instability rather than member instability to occur, b must satisfy
( l n 2h \ 112 b< k - r )
where
p
r(y)
is the radius of gyration of the section.
SEC. 8 - 4 .
IN C R E M E N T A L FO R M U LATIO N
199
Lastly, we outline how one applies the method of successive substitution to (t). For convenience, we drop the subscripts and write (t) as
kt A u = A p A p
(z)
In the first step, we take Ap = 0.

kt
(aa)
The second estimate is determined from Au(2) = (Ap - Ap'1') Generalizing (bb), Au('I+1) = 1 (A p - Ap" > ) (cc)
(bb)
K
The convergence is illustrated in Fig. E 8-5C . Case (b) shows how the scheme diverges
Fig. E8-5C
Ap
Ap
ia)
(b) in the vicinity of a neutral point (k, 0). Convergence generally degenerates as k t >0 and one has to resort to an alternate method.
< *
200
CHAP. 8
8 -5 .
LINEARIZED STABILITY ANALYSIS
In the previous section, we illu strated the behavior of a geom etrically n o n linear system. T he analysis involves first solving the nonlinear equilibrium equatio n s for the displacem ents and then applying the classical stability crite rion to determ in e the stability o f a p artic u la r eq u ilibrium p o sitio n . O n ce the n o n lin ear eq u ilibrium eq u atio n s are solved, the stability can be readily d e te r m ined. N ow , if a geom etrically nonlinear system is loaded in such a way th a t it behaves as if it w ere geom etrically linear , we can neglect the displacem ent term s in K(| n , th a t is, we can tak e p = a in the expression for k,. T his a p pro x im atio n is qu ite convenient since we have only to solve the linear problem in o rd er to ap p ly the stability criterion. W e refer to this p ro ced u re as linearized stability analysis. A ccording to (8-40), an equ ilib riu m position is stable (neutral, unstable) w hen th e tan g en t stiffness m atrix is positive definite (positive sem i-definite, in different). W e generate tran sfo rm to Ctif* and test CfC* for positive defi niteness. W e have show n th a t <#'* a n d Kt, u have the sam e definiteness p ro p erty, i.e., Kt> n is positive definite if cfT* is positive definite. W o rk in g w ith 3C? ra th e r th a n K,, u avoids having to p erm u te the row s an d colum ns. In linearized stab ility analysis, we ap p ro x im ate k t with k,.
=
k a ja
F g
(8 -4 4 )
T he first term is the linear stiffness m atrix. W e in terp ret th e second term as a geometric stiffness. T he b ar forces are determ in ed from a linear analysis of the truss. If the lo ad in g is defined in term s o f a single lo ad p aram eter, a , we can w rite (8 -4 4 ) as
F
"
)F
"
(8 -4 5 )
II f S.liq , ,f
T he tan g en t stiffness m atrix is generated by applying the D irect Stiffness M eth o d to each term in (8-45). W e express th e actual an d modified m atrices as K(, n = K , . u + KSiU and r / f f = :/?* + u r * (8 -4 7 ) w here is the system stiffness m atrix for linear behavior. It is sym m etrical an d positive definite w hen the system is initially stable. T h e g eom etrical stiff ness, K g, is also sym m etrical but it m ay not be positive definite. E q u atio n (8 -4 6 ) show s th a t the ta n g en t stiffness m atrix varies linearly w ith the load param eter. If th e system is initially stable, Kr, H is positive definite for X = 0. As X is increased, a tran sitio n from stable to n eu tral eq u ilibrium m ay occur a t som e lo ad level, say Acr. T o d eterm in e Acr, we n o te th a t n eu tral equilibrium (see (8 -4 3 )) co rresp o n d s to |K ,> n | = 0 w hich, in tu rn , can be in terp reted as th e existence o f a n o n -triv ial so lu tio n of K u iA U ^ O (a) (8 -4 6 )
SEC. 8 - 5 .
LIN E A R IZ E D S T A B IL IT Y A N A L Y S IS
201
Substituting for K r, u tran sfo rm s (a) to a ch aracteristic value problem , K f, n A U i = - A K j . h A U ! and ?iC T is the sm allest eig en v alu e! o f (8-48). Since |K t, u | |3T*|, we can w ork w ith .yr* m 3 = 0 V = -x ,y f* a m J (8- 49) (8 -4 8 )
instead o f (8 -48). B oth eq u atio n s lead to th e sam e value o f Xcr. H ow ever, (8-49) has r a d d itio n al characteristic values equal to 1 since we have ad d ed r dum m y eq u atio n s. T o show this, we su b stitu te for CriT* using (8 -4 1 ) an d note (8-42).
IT
"Kf, ii 0
0 Ir
Dl
X . 11
(A U 2J ^
fA U ^ |A 0 2 l
(a)
P rem ultiplyin g by H - ( n r ) ~ \ (a) becom es K(. n A Ui = AU2 T he solution o f (c) is

Al &2 = K -1 r > 0 Y 0 >+ c 2 + + c, < i
0J
V .0 J
This so lu tio n m u st be disregarded since A U 2 is actually a null m atrix.
Example 8 -6 ------------------------------------------------------------------------------------
Consider the system shown. We suppose the bars are identical, the material is linearly elastic, and there is no support movement. The geometry change is negligible under a vertical load and we can use the linearized stability criterion. Working with the undcformcd geometry, we have
Fl ~ Fl ~ 2 b X
(a)
t Matrix iteration (Ref. 1) is a convenient computational scheme for determining A cr. We apply it to .
(K9 i ,)AU1 = ( y W n AU,

which satisfies the restrictions on the method.
202
D ISPLACEM EN T M ETH O D : IDEAL TRUSS
CHAP. 8
Fig. E8-6
-Xi
W e le t k t =
k2
k.
T h e s y ste m
s tiffn e s s m a t r ic e s f o llo w
fro m
(8 -4 4 ) a n d
(8 -4 5 ).
K,, u = kU + k;,2 = Ka Wl + a 22) '1 Y Lj o

(b )
2k
and
(rj
L
K9, n kS i i + kg 2 j r (8 i + 82)
I t r e m a in s to d e te r m in e g , a n d g 2 w h ic h a r e d e fin e d b y P = W e n e g l e c t u l 2 in ctn + u [g T h is is r e a s o n a b l e w h e n d b.
(c)
(d ) The
th e g e n e r a l e x p r e s s io n
f o r ji .
a p p r o x im a t e e x p r e s s io n s fo r
p and
are
pt * j - ( y + M n
-fc]
P2 * [-</ +
'1 O' 0. 1 0" 0 0
ii
- 6 j
(e)
Ri 82 - F in a lly ,
Ks, n =
and
(f)
K , u
K ,,u
;J K t , n
2k
r w w 0
A
f b V
0
(g)
17 I w
0 0
SEC. 8 - 5
LIN EAR IZE D S T A B ILIT Y AN A LY S IS
203
Neutral equilibrium (Kti u is semidefinite/ occurs at / d\ 2

2A Eb ( d\ 2
" m (z)
l)
(h)
Note that (g) has only o n e eigenvalue instead of tw o . This is a consequence of our using approximate expressions for Equations (e) instead of the exact expressions. At ?. A cr, the system is neutral with respect to Awj x, i.e., the buckling mode is antisymmetric. Neutral equilibrium also occurs when the bars either buckle or yield. The value of X for Euler buckling of the bars is
_ 2b
^cr, eb
_ 2b
rb ~ ~
L ""
n jE f 1?
2 A E b ( np^ 2
~l t I t
(i)
Comparing (h) and (i), we see that Euler buckling of the bars controls when
d >
np
(J)
The exact expression for g is
Q n
If we work with (k),
I
*Jn
li
(k)
11 = r 12
and
(1 )
1 T 0
b X
fd \ 2
K(,H = K ,.u + AK,iU = 2 k o

_
(b V w _
1
h
(m)
In this case, there are two characteristic values and therefore two critical values of L Ac, i = 2 kb
'd V
(n)
X
The second root corresponds to neutral equilibrium with respect to Aw12- For this example,
d h, and the first root defines the critical load.
It is of interest to compare Acr, 2 with the buckling load found in Example 8-5. There we considered d b and followed the nonlinear behavior up to the point at which the slope of the p 1 2 u l 2 curve vanished (neutral with respect to Ai/(2):
K t = dPi2 - = n 0 => p i 2 du i2
(o)
The linearized result is significantly higher than the true buckling load. In general, the linear buckling load is an upper bound. How close it is to the actual value will depend on the geometry and loading. When d b , it is quite close, while it considerably overestimates the true load for d b.
204
CHAP. 8
REFERENCES
1. 2. 3. 4. 5. 6. 7.
N o r r is , C. H., and J. B. W ilb u r : E le m e n ta r y S tr u c tu r a l A n a ly s i s , McGraw-Hill, N e w York, 1960. H a l l , A. S., and R. W. W o o d h e a d : F ra m e A n a ly s is , Wiley, New York, 1967. A r g y r is , J. H. and S. K elsey: E n e r g y T h e o r e m s a n d S tr u c tu r a l A n a ly s is , Butterworths,
8.
9. 10. 11. 12.
London, 1960. L iv e s le y , R. K.: M a t r i x M e th o d s o f S tr u c tu r a l A n a ly s is , Pergamon Press, 1964. d e V eu b e k e , B. F., M a t r i x M e t h o d s o f S tr u c tu r a l A n a ly sis., Pergamon Press, 1964. M a r t in , H. C .: I n tr o d u c tio n to M a t r i x M e th o d s o f S tr u c tu r a l A n a ly s i s , McGraw-Hill, New York, 1965. A r g y s is , J. H .: R e c e n t A d v a n c e s in M a t r i x M e th o d s o f S tr u c tu r a l A n a ly s i s , Pergamon Press, 1964. R u b in s te in , M. F . : M a t r i c C o m p u te r A n a ly s is o f S tr u c tu r e s , Prcntice-Hall, 1966. P r z e m ie n ie c k i, J. S.: T h e o r y o f M a t r i x S tr u c tu r a l A n a ly s is , M cGraw-Hill, 1968. T h o m p so n , J. M. T . and A. C. W a l k e r : T h e Nonlinear Perturbation Analysis of Discrete Structural Systems, In t. J. S o lid s S tr u c tu r e s . Vol. 4, 1968, pp. 757-768. R u b in s te in , M. F.: S tr u c tu r a l S y s t e m s S ta tic s , D y n a m ic s , a n d S ta b i li ty , PrenticeHail, 1970. R a l s t o n , A .: A F ir s t C o u rse in N u m e r ic a l A n a ly s is , McGraw-Hill, 1965.
PROBLEMS
8 - 1 . C o n sid er U 2 an d P , to be prescribed an d the b eh av io r to be physically linear. (a) Express VT U t in term s o f U 1; U 2. U se
T 1 P=
Pf
2
vt = (b)
Z
j= i
kM j ~ e0 J )2 = i( e - e0)r k(e - e0)
Show th a t (8 -1 ) arc th e E uler eq u atio n s for n ^ U j ) . N o te th a t dVT F T de de = B f A U t
E xpress d 2Tlp as a q u a d ra tic form in A U ^ H i n t : O b ta in d 2e by o p eratin g o n (7-8). 8 - 2 . F o r the stru ctu re sk etch ed : (a) D eterm in e K u . (b) D eterm in e ut an d F due to a te m p eratu re increase o f 100F for all the bars. A ssum e no su p p o rt m ovem ents at jo in ts 2, 3, 4. 8 - 3 . F o r th e stru ctu re sk etch ed : D eterm in e th e displacem ents, b a r forces, an d reactions. 8 - 4 . Refer to E xam ple 8 - 2 . S u p p o se we n u m b e r th e jo in ts as show n. D evelop th e general form o f J T an d co m p are w ith the result o f E x am p le 8 - 2 f 5 \ 8 - 5 . F o r th e stru ctu re sketched, d eterm in e f ) jf '* . VE a J 8 - 6 . F o r th e stru ctu re sk etch ed : D evelop th e general form o f J f *. In d ica te how you w o u ld o b ta in K u .
(c)
P R O B LE M S
205
Prob. 8 - 2
*2
= 3 X 10 4 ksi B ar areas = 3 i n . 3 C o e ffic ie n t o f th e rm a l e x p a n s io n 6 X 1 0 ~ 6/ F
Prob. 8-3 2010 '
in. to th e left
Prob. 8-4
206
CHAP. 8
-2 0 '
Prob. 8-5
E c o n s ta n t f o r all bars
Bar A re a
3a
2
3
4a
3a
4a 2 .5 a
4 5
Prob. 8-6
8 - 7 . D eterm in e the lo ad-deflection rela tio n for the system show n. C o n sid er the m aterial to be linearly elastic an d th e bars to be identical. A ssum e no in itial elo n g atio n o r s u p p o rt m ovem ent. 8 - 8 . In v estig ate the elastic stab ility o f th e system show n. A ssum e the m aterial is linearly elastic an d no s u p p o rt m ovem ents. U se th e linearized stab ility criterio n a n d w o rk w ith th e exact expression for g. R ew o rk the p roblem , co n sid erin g d b a n d using th e co rresp o n d in g a p p ro x im a te expres sion for g. 8 - 9 . D eterm in e th e low est critical lo ad for th e truss show n. A ssum e th e m aterial is linearly elastic an d all b ars have the sam e stiffness.
Prob. 8 - 7
d b
i-j - A 2 k -M-
Prob. 8-8
Prob. 8-9
208
DISPLACEM ENT METHOD: IDEAL TRUSS
CHAP. 8
8 - 1 0 . The governing equations for geom etrically nonlinear behavior of a linearly elastic discrete system such as a truss are nonlinear algebraic equations containing up to third-degree displacem ent terms. We have expressed them as ^ & >0 + y e w - & >0 + (a)
where stf, 88 contain linear displacem ent terms. This form is dictated by our choice of m atrix notation. In o rder to expand (a), we m ust shift from m atrix to indicial notation. F o r convenience, we em ploy the sum m ation convention. If a subscript is repeated in a term, it is understood the term is sum m ed over the range of the repeated subscript. An exam ple is Oijbj s n X aijbj y=i (./ = 1, 2 , . . . , n ) (b)
We write the r'th equilibrium equation for the system as (this representation is suggested in Ref. 8-10): (Ki} + K ijkUk + K iJktUkUf )Uj = XP, - P0, , where i, j, k, I range over the total num ber of unknow ns, Uj is the total value of the ; th displacem ent unknow n, X is a load param eter, Pt defines the load distribution, and the K 's are constants which can be interpreted as second-, third, and fourth-order tensors. The second-order tensor, K tj, is the linear stiffness matrix. (a) We generate the system tensors by superim posing the contribution of each bar. The first step involves converting the m atrix expressions (d) where Fn Kfin + ^ 0. n = y(u, - u/(_)
(c)
7 =
(u+ - uJ r
over to indicial form. We drop the n subscript, define p and u as

(f)
and write (d) in the form

Pi ( h j
ktjku k
k i j k t u ku f ) u j
Po.i
(g)
PRO BLEM S
209
Show th a t kjj > < ijk

K jk(
kccs< y .rcsjcrj 7 [ C s i C r J + Crtr)] k 2L

(h)
where c is defined by u, - u
P ! -I]
|U )u
= cu
(0
D iscuss how you would locate the ap p ro p riate addresses for th e bar stiffness tensors in the system tensors. W hat sym m etry properties do the k's exhibit? D o these properties also apply for the system tensors? (b) D evelop the increm ental equations relating A h, AA and com pare with (8-30). (c) Specialize the increm ental equations for linearized stability analysis. 8 -1 1 . F o r the structure sk etch ed : Prob. 8-11
(a)
(b)
D eterm ine the nonlinear increm ental equilibrium equations at the equi librium position corresponding to 0, p2 p2, C n the linearized critical load. T ake A = 0 and solve for Ap2 as a function o f A u {. C om m ent on how the system behaves when a sm all horizontal load, p l = e p 2, is applied in addition to p2.
9 Force Method Ideal Truss

9 -1 . GENERAL
T h e basic eq u atio n s for th e linear geometric ease have the form P i - B 3F e - B fU i + B7 2U 2 = e 0 + fF P 2 = B 2F (a) (b)
(c)
w here the elem ents o f B t and B 2 are constants. E q u atio n (a) represents nd linear equ atio n s relating the nd prescribed jo in t forces a n d the m unk n o w n b ar forces. F o r the system to be initially stable, r(Bi) nd, th a t is, the row s of B t m ust be linearly independent. This requires m > nd. In w hat follows, we consider only stable systems. If m = nd, the system is said to be statically determ in ate since one can find the b ar forces and reactions using only the eq u atio n s of statics. T he defect o f (a) is equal to m nd = q, and is called the degree o f indeterminacy. O n e can solve (a) for nd b ar forces in term s of th e applied forces and q b ar forces. W e refer to the system defined by th e nd bars as the p rim ary stru ctu re and th e q unk n o w n forces as force red u n d an ts. In o rd er to determ ine F, q ad d itio n al eq u atio n s relating the b a r forces are required. These eq u atio n s are called com patibility co n d itio n s an d are o b tained by o p eratin g on (b) which represents m relatio n s betw een the nd unk n o w n displacem ents an d the b a r forces. T he general pro ced u re o u tlined ab o v e is called the force o r flexibility m ethod. This p ro ced u re is applicable only w hen the geom etry is linear. In w h at follows, we first develop th e governing eq u atio n s for th e force m e th o d by o p eratin g on (a)-(c). W e then show how one can establish th e com patibility equ atio n s using th e principle o f v irtual forces an d discuss the extrem al ch aracter of th e force red u n d an ts. Finally, we co m p are the force m e th o d for a truss w ith th e m esh m e th o d for an electrical netw ork.
210
SEC. 9 - 2 .
G O V ER N IN G EQ U A TIO N S A L G E B R A IC A P P R O A C H
211
9-2.
GOVERNING EQUATIONS ALGEBRAIC APPROACH
We consid er th e first nd colum ns o f Bi to be linearly in dependent (if the system is initially stable, one can alw ays ren u m b er the b ars such th a t this condition is satisfied) an d p a rtitio n B2 an d F as follows: Bi = [ B n (d x m l ( < i x nd) 1 b I2 i (n < ?Xq)
B 2 - [ B 2I ! b 22] (r x m) (r x nd) 1 (rxg) ('M X 1) (mx 1) f I-\ i| b = 1 (q l?2 X1) J
(9-1)
The b ars co rresp o n d in g to F i com prise th e primary structure an d F 2 co n tain s the q red u n d a n t b a r forces. U sing (9 -1 ), the force-equilibrium eq u atio n s ((a) and (c)) tak e the form = P , - B 12F 2 P 2 = B 21F 1 + B 22F 2 (nd cqs.) (re q s.) (9 -2 ) (9 -3 )
Since |B n | ^ 0, we can solve (9 -2 ) for F j, co n sidering P t, B 12 as righthand sides. T h e com plete set o f q + 1 so lu tio n s is w ritten as F x = F j . o + F 1>jr2F 2 w here F t . 0 an d F x(F2 satisfy B l l F l . 0 = P1 B u F x. f , = - B 12 (9 -4 )
(9 o
1
N o te th a t the k t h colum n o f F 1iFj co n tain s the b ar forces in th e p rim ary stru ctu re due to a u n it value o f the fcth elem ent in F 2- Also, F I>0 co n tain s the b a r forces in th e p rim ary stru ctu re due to th e ap p lied jo in t loads, P l5 w ith F 2 = 0. T he reactio n s follow from (9 -3 ): 1*2 1*2, 0 + P 2.F Z I, 2
P 2. 0 = B21F li0 P 2 .Fi = B 2 i F i , f 2 + B 22

W e consider next (b). P a rtitio n in g e, e 0, an d f,
(9 -6 )
212
FORCE M ETH O D : IDEAL TRUSS
CHAP. 9
an d using (9-1), the force-displacem ent relations expand to B f iU t + B L 0 2 = e x = e i . 0 + f j F j BF2U 1 + B 22U 2 = 62 = ^2,0 + f2F 2 (iy eqs.) (tf eqs.) (9-8) (9 9)
O nce ej is know n, (9 -8 ) can be solved for U i. We o b tain the eq u atio n for F 2 by elim inating U | in (9-9). F irst (see (9 -5 )) we n o te th at B h = - ( B n F liF2)r -V lrM i (a)
Then, prem ultiplying (9 -8 ) by F f, f2, adding the result to (9 -9 ), and using (a), (9 -6 ) leads to P2 r , f 2t }2 = e 2 + F [. f 2ei e 2, 0 + f 2p 2 + F [ ,F 2(6l.O + fIF11)
(9 -1 0 )
(9-11)
T he first form, (9-10), shows th at the eq u atio n s are actually restrictions on the elongations. O ne can in terp ret (9 -1 0 ) as a com patibility condition, i.e., it m ust be satisfied in o rd er for the bars to f i t in the deformed stru ctu re defined by U t. T he second form, (9-11), follows when we express the elongations in term s of the b ar forces. Finally, we su b stitu te for F 1 and w rite the result as f22F2 = d 2 w here
^ 22
(9 -1 2 )
= f2 + If,
(9 -1 3 )
d 2 2,0 ~ I 1f . r 2(e 1. 0 + f i Fi . o) + P l , F i 0 2 T he coefficient m atrix, f 22, is called the flexibility m atrix for F 2. O ne can show th a t f22 is positive definite w hen the b ar flexibility factors in f2 are all positive.t If the m aterial is physically nonlinear, f an d e0. depend on F. Iteration is m inim ized by applying the loading in increm ents an d ap p ro x im atin g the force-elongation relation with a piecewise linear representation. T he incre mental eq u atio n s are sim ilar in form to the total equations.^ W e ju st have to replace the force, displacem ent, an d elongation term s with their increm ental values and interpret f as a segm ental (tangent) flexibility. At this point, we sum m arize the steps involved in the force m ethod. I. Determination o f F i, 0, P 2.o> F j. y2, and P 2 F2
We select a stable prim ary stru ctu re F t an d determ ine the b ar forces and reactions due to p ! and a unit value of each force red u n d an t. This step involves q + 1 force analyses on the primary structure. N o te th a t we o b tain the primary structure by deleting q = m nd bars. T he selection of a prim ary structure and solution o f the force equilibrium eq u atio n s can be com pletely automated.'*
f S ee P rob . 9 - 1 . t See P rob . 9 - 4 . 5 W e red u ce to an echeJon m atrix.
S e e (1 -6 1 ).
SEC. 9 -2 .
G O V ER N IN G EQ U A T IO N S A L G E B R A IC A P P R O A C H
213
2.
Determination o f F 2. F . , a n d P 2
W e assem ble f 22, d>, and solve f 22F 2 = d 2 for F ,. T h en , we d eterm in e F t and P 2 by co m b in in g the q + 1 basic solutions. F i = F ^ o + F i i fj F2 f*2, 0 + P 2, F2F 2 3. Determination o f U l .
O nce F j is know n, we can ev aluate e l5 e i = e i, 0 + f j F i and then solve (9 -8 ), BT ,U i = c x - B h V 2 for U (. If only a lim ited n u m b e r o f displacem ent co m p o n en ts are desired, one can determ ine these co m p o n en ts w ith o u t actually solving (9-8). T o show this, we write U ! as U i = (B u 1)7^ - (B21B u ') t 0 2 (a) We see from (9 -1 5 ) th a t th e k th colum n o f BJV co n tain s th e b ar forces in the primary stru ctu re due to a unit value o f th e Ath elem ent in Also, it follows from (9 -6 ) th a t the kth colu m n o f B 2 lB ]7 co n tain s th e reactions due to a u nit value of the /<th elem ent in P ^ N ow , we o b ta in th e kth elem ent in U t (which corresponds to the /<th elem ent in P ,) by m ultiplying the kth co lu m n o f B f / by e{, the /<th colum n o f B 21B7j1 by O j. an d ad d in g th e tw o scalars. T hen, letting F i.p ;k = F i P 2, PiK P 2 due to an unit value of pjk with F 2 0 due to an u n it value o f pJk w ith F 2 = 0 (9 -1 4 )
we can w rite the expression for 1ijk as = Ff.B e , P J .W 0 , (9 -1 5 )
Note th at one w orks w ith the statically d eterm in an t primary stru ctu re to d eter mine the displacem ents. Example 9 - 1 ------------------------------------------------------------- -- - - ------Step 1 ; D eterm ination o f F 1 0 , P 2 0, Fj F,, and P 2 n
For the truss shown in Fig. E 9-1 A,

nd = 2 m = 3 q =
We take
as the redundant ba'r force:
F, = {Fi,/r 2}
F2 = {F3}
The primary structure consists or bars 1 and 2. N ote that all force analyses are performed on the p r im a r y structure. The forces and reactions corresponding to P, and F 3 - + 1 can be readily obtained using the method of joints. The results are shown in Fig. E 9-1B .
214
FORCE M E T H O D : ID EA L TRUSS
CHAP. 9 F ig. E 9 - 1 A
(1) A { = 1.0 in.2

(2 )
A 2 0.5 in.2
A 3 = 0.5 in.2
10 4 ksi f o r all bars.
M a te ria l is lin e a rly elastic. E = 3
(3) ô,i 1/16 in.

(4) U3 2 + l / 1 0 i n .
X ^0,2 = eo,3 = 0-
z< 4 i = 1/15 in. Fig. E 9 - 1 B
We could have obtained the above results for F , by solving B 11 F 1 = Pi - B 12F2 which, for this system, has the form
~-. 6 .- 8
Step 2:
+ .6' + .8_ w
0 {2Oj X
Determination of f22, d2, F 1 ? and F2
Since only u 32 and u4, are finite, we can contract 0 2 and P2,
P i I P32' /;41 }
0 2 = 41 I
and write
P ^ O , = (P2,FJ 'r 0'2
SEC. 9 -2 .
GOVERNING EQUATIONS ALGEBRAIC APPROACH
215
The force matrices follow from step 1:

F i.o = { -2 0 .8 3 , -4 .1 7 }
(kips) (kips) (kips)
Fi.,a { - * - * } Pi. Fj = {+ 1. +1}

Also, we are given that
el,0 ~ {e0,
e 2. 0 =
e0, 2} = {lV) 0}
3} = 0
(inches) (inches)
{eO,
C 2 = {+1^5, - tt}
It remains to assemble f , , f 2 and evaluate The flexibility factors are (in./kip)
f 2 2
and
d 2 .
12(25) X
Then,
12(25)
12(20)
fi =
J2 -
1.5 X 104 "i

2
T T T io5
f _ 7 i (T I1 _o h .
f 2 = [ / 3 ] =
O' (2 x 1 0 '2) 1
0.8(2 x 1 0 - 2)
Evaluating the various products in (9-13), (9-12) reduces to 1.38F3 = -7 .3 1 Solving (a), we obtain F 2 = {F3} = - 5 .2 7 kips
(a)
F 1 - F . . . + F IiftF1
0.87 kips]
Equation (a) actually represents a restriction on the elongations. The original form of (a) follows from (9-10).
e3 t ei ~ 8e2
(b)
Equation (b) reduces to (a) when we substitute for the elongations in terms of the bar forces.
Step 3 :
Suppose only
Determination o f the Displacements

mu
is desired. Using (9-15),

uh F f,
Pllei ( P 2. p , , ) 7 t - f 2
(c)
Now,
U 2 = My. Is}
ei =
e i . o + f i F t = { -24, .0 1 8 }
We apply a unit load at joint 1 in the A", direction and determine the bar forces in the p r im a r y structure and the reactions ( p 32, P 4 1 ) corresponding to the nonvanishing prescribed displacements:
F i , pu = {>!) 1*2. Pll = {0, -* }
216
FORCE M E T H O D : ID EA L TRUSS
CHAP. 9
Substituting in (c), we obtain

u l i + .18 5 .033 == + .1 5 in
If both displacement components are desired, we apply (9-15) twice. This is equivalent to solving (9-8).
9-3.
GOVERNING EQUATIONS VARIATIONAL APPROACH
W e o b tain ed the elongation co m p atib ility eq u atio n s (9 -1 0 ) by o p eratin g on the elongation-displacem ent equations. A lternatively, one can use th e principle o f virtual forces developed in Sec. 7 -3 . It is show n th ere (see E q u atio n (7 -1 4 )) th a t th e tru e elo ngations satisfy the co n d itio n , A F Te - A P f 0 2 = 0 (a)
for any statically perm issible system o f virtu al b ar forces and reactio n s which satisfy the co n strain t condition, B, AF - A P : = 0 (b)
E q u atio n (b) states th a t th e virtual b a r forces c an n o t lead to increm ents in the prescribed jo in t loads, i.e., they m ust be self-equilibrating. N ow , using (9-4), (9-5), we can w rite
F _ m _ f j ,.j + [0 j
(.I'iJ w here
Fb J l h
F,
(c)
B, { ^ j = P , an d B, T hen AF = 4 = 0
(d,
(e)
jf
| AF2
(f)
satisfies (b) for a rb itrary A F 2. T he reactions due to A F 2 are o b tain ed from (9 -6 ): A P 2 B2 A F P 2( f 2 A F 2 (g) S u b stitu tin g for A F an d A P 2, (a) expands to A F 2(Ff, F :iQx + e 2 P 2, F jU 2) = 0 E q u atio n (h) m u st be satisfied for a rb itra ry A F 2. Finally, it follows th a t F f.j^ ei + e2 P 2, F2 & 2 ~ 0 (i) (h)
E q u atio n (i) is identical to (9-10). N o te th a t the elo n g atio n com patibility
SEC. 9 - 4 .
C O M P A R IS O N OF THE FORCE A N D M ESH M ETH O D S
217
equations are independent o f th e m aterial behavior. If th e m aterial is physically linear, (i) leads to a set o f q linear eq u atio n s in F 2 w hen we su b stitu te for the elongations in term s of th e b a r forces. W e d eterm in e the displacem ents by app ly in g the general form o f th e principle of virtual forces (see (7 -1 0 ))
AFTe - AP r t ) 2 = A P f U j
w here the virtu al forces satisfy the force-equilibrium equations,
(j)
APi - Bj AF
A P 2 = B 2 AF Since only F 2 is req u ired to eq u ilib rate P l5 we can take A F t = F liPl A P i
af2
= o
(J) (m) (n)
and (j) leads to
Ui = F [ , Pte Piei - P PlU T?T, l ~ r 2, Pi ^ 22

N o te th a t I?
O n e can in terp ret th e com p atib ility eq u atio n s expressed in term s of F 2 as the E uler eq u atio n s for the to ta l co m p lem en tary energy function, n cv f ~ p i g 2 - n c( F 2) (o)
This ap p ro ac h is discussed in sec. 7 -5 . W e take X = F 2 in (7-35). Then, F v= X A and (7-3 7 ) coincides w ith (i). W e have w ritten the ex p an d ed form o f (i) as f 22F 2 = d 2 Since (i) are th e E uler e q u atio n s for n c, dTlc = A F 2(f22F 2 - d 2) and it follows th a t n c - i F 2f22F 2 - F 2d2 (s) (r) (q)
p,
P 2. .r P2.F2
(P)
for the linearly elastic case. O n e can show th a t the statio n ary p o in t co rresp o n d s to a relative m in im u m value o f I I C w hen th e tan g en t flexibility factors for the re d u n d an t b ars are all positive, t
9-4.
COMPARISON OF THE FORCE AND MESH METHODS
It is of in terest to co m p are the force m eth o d for a tru ss w ith th e p ro ced u re followed to find the cu rren ts in an electrical netw ork. T h e la tte r involves the
f See Prob. 9-8.
218
FORCE M E T H O D : ID E A L TR USS
CHAP. 9
ap p licatio n o f K irch h o ffs law s a n d is called th e m esh m ethod. V ario u s phases of th e electrical n etw o rk fo rm u latio n are discussed in P ro b s. .6 - 6 ,6 - 1 4 , an d the governing e q u atio n s for a linear resistance d-c n etw o rk are developed in P robs. 6 -1 4 , 6 -2 3 . W e list th e n o ta tio n a n d governing e q u atio n s for convenience: b n N M Vj k+, k _ ik n u m b e r o f bran ch es n u m b e r o f nodes n - 1 b N = b n. + 1 p o ten tial at n o d e j w ith respect to the reference p o te n tial, o. n o d es at positive an d negative ends of b ran ch k c u rren t in b ran ch /<, positive w hen directed from n o d e /c_ to node k+ ek = p o te n tial d ro p for b ran ch k + e0, k = em f for b ran ch k R k = resistance for b ran ch k = = = = = =
T h e governing e q u atio n s expressed in m atrix n o ta tio n are (see P ro b . 6 -2 3 ): Ari - 0 e = AV = e0 + R i w here i { 'i, <2, , ib} (9 -1 8 ) V = {Vlf (N eqs.) (b eqs.) (9 -1 6 ) (9 -1 7 )
%
R=
*2
a n d A is o b ta in ed by deleting th e last co lu m n of the b ra n c h -n o d e co n n ec tivity m atrix s i . N o te th a t s i has only tw o entries, in any row . F o r row k (k = b),
l,2,...,
$kk- +1 stfkk+ = 1 s./kj = 0 j ^ k+ o r /<_ j = 1 ,2 ,..., N (9 -1 9 )
A ctually, s d is ju s t the m atrix eq u iv alen t o f the b ran ch -n o d e co nnectivity table.
Example 9 -2 ---------------------------------------------------------------------------------------A network can be represented by a line drawing consisting of curves interconnected at various points. The curves and intersection points are conventionally called branches and nodes respectively. Each branch is terminated at two different nodes and no two branches have a point in common which is not a node. Also, two nodes are connected by at least one path. A collection of nodes and branches satisfying the above restrictions is called a linear connected graph. If each branch is assigned a direction, the graph is said to be
SEC. 9 - 4 .
C O M P A R IS O N OF THE FORCE AN D M ESH M ETH O D S
219
oriented. The connectivity relations for a network are topological properties of the cor responding oriented graph. Consider the oriented graph shown. We list the branch numbers vertically and the node numbers horizontally. We assemble s i working with successive branches. Finally, we obtain A by deleting the last column (col 4} o f ,s/.
Fig. E9-2
Branch 1 2 3 4 5 6 -1 -1
2 + 1 + 1 + 1
-1
-1 + 1 N -1
N ow , A has N linearly in d e p en d en t columns. T herefore, it is possible to solve (9 -1 6 ) for N b ra n c h cu rren ts in term s o f b - N = M b ran ch currents. W e su p p o se th e b ran ch es are n u m b e re d such th a t th e first N row s o f A co n tain a n o n v an ish in g d e te rm in a n t o f o rd e r N a n d p a rtitio n A, i after row N. (hxN) A = (N*N) _ _A i _
~{M x\ )
A^
(N* U
(9 -20)
220
FORCE M E T H O D : ID E A L TRUSS
C H AP. 9
In tro d u cin g (9 -2 0 ) in (9 -1 6 ) leads to A f ii = A2i 2 (9-2 1 )
Since }Aj | ^ 0, we can solve for ij in term s o f i 2. W e w rite th e so lu tio n o f the node eq u atio n s as
(9 -2 2 ) M < ,
N o te th a t C i is o f o rd e r N by M an d is related to A i, A 2 by c x = - ( A 2A r x)T (9 -2 3 )
It rem ains to d eterm in e a set of M e q u atio n s for i2. O n e can express (9 -1 7 ) in p a rtitio n ed form and th en elim inate V, o r alte rn a tively, one can use the v ariatio n al principle developed in P ro b . 7 - 6 . U sing th e first ap p ro ach , we w rite (9 -1 7 ) as e* = A jV = e (t o + R *ii e 2 = A 2V = e2. o -f R 2i2 O nce i t is know n, we can find V from A |V e t e i.o + R iij (9 -2 5 ) (N eqs) (M eqs) (9 -2 4 )
E lim inating V from the second e q u atio n in (9 -2 4 ) an d using (9-2 3 ), we o b tain e 2 + C fe i 0 (9 -2 6 )
E q u a tio n (9 -2 6 ) represents M eq u atio n s relatin g th e b ran ch p o ten tial differ ences (voltages). F inally, su b stitu tin g for e;- in term s of ij leads to
(R2 + C t R jC O h e2,o ~ C {e1> 0
(9 -2 7 )
T h e coefficient m atrix for i2 is positive definite w hen th e b ran ch resistances are positive. This will be th e case for a real system. T h e essential step in the so lu tio n involves solving (9-2 1 ), th a t is, finding Q . N o te th a t C j co rresp o n d s to F 1iFz for th e truss problem . Also, th e branches com prising A! (and iO co rresp o n d to th e primary stru ctu re. A lthough the e q u atio n s for th e truss an d electrical n etw o rk are sim ilar in form , it sh o u ld be noted th a t the n etw o rk p ro b lem is one d im en sio n al w hereas the tru ss p ro b lem involves the geometry as well as the connectivity o f th e system . O n e can as sem ble Cj. using only the top o lo g ical p ro p erties o f the o rien ted g rap h w hich represents the netw ork. T o find the co rresp o n d in g m atrices ( F 1>0 an d F i >f2) for a truss, one m ust solve a system o f lin ear eq u atio n s. In w h at follows, we describe a p ro ced u re for assem bling C j directly from the o rien ted g raph. A closed p ath co n tain in g only one rep eated node th a t begins an d ends at th a t node is called a mesh. O n e can rep resen t a m esh by listing sequentially the branches traversed. A tree is defined as a co n n ected g rap h hav in g no
SEC. 9 - 4 .
C O M P A R IS O N O F TH E FORCE AN D M ESH M ETH O D S
221
m eshes. L et bT be th e n u m b e r o f b ran ch es in a tree co n n ectin g n nodes. O ne can easily show th a t bT = n - 1 - N (9 -2 8 ) W e reduce a g ra p h to a tree by rem oving a sufficient n u m b e r o f branches such th a t no m eshes rem ain. T h e b ran ch es rem oved are generally called chords. T h e req u ire d n u m b e r o f ch o rd s is equal to b bT = b N = M . N ow , we associate the b ran ch es com p risin g a tree w ith the row s o f Ai. Selecting a tree is eq u iv alen t to selecting N linearly in d ep en d en t row s in A. T he M ch o rd s co rresp o n d to th e re d u n d a n t b ran ch es, th a t is, th e row s o f A2. N o te th a t one can alw ays n u m b e r th e b ran ch es such th a t the first N b ran ch es define a tree. C h o rd j a n d the u n iq u e p a th (in th e tree) co n n ectin g th e term in als o f ch o rd j define a m esh, say m esh j. W e ta k e the positive d irectio n o f m esh j (clockw ise or counterclockw ise) such th a t the m esh d irectio n coincides w ith the positive direction for ch o rd j. N ow , th e cu rren t is constant in a m esh. S uppose b ran ch r is c o n tain ed in m esh j. T h en , the c u rre n t in b ran ch r d u e to a u n it value of ij is eq u al to + 1 ( 1) if the positive directio n s o f b ra n c h r a n d m esh / coincide (are o p p o site in sense). W e have expressed th e so lu tio n o f the n o d e eq u a tio n s as
(N x
1)
(N x M )
I)
ii
C,
i2
N ow , we tak e th e elem ents o f i2 as th e chord (mesh) cu rren ts. T h en ij represents the required b ran ch cu rren ts in the tree. W e assem ble C j w o rk in g w ith the colum ns. T h e co lu m n co rresp o n d in g to i,- involves only th o se b ranches o f the tree w hich are co n tain ed in m esh j. W e en ter ( + 1, - - 1 ,0 ) in row k of this colum n if b ran ch k is (positively, negatively, no t) included in m esh j. Exam ple 9 -3 ------------------------------------------------------------- -------- ......... ........ F o r th e grap h in e x a m p le 9 -2 , N = n 1 = 3 and b = 6. T hen M = b 6 N = 3
a n d w e m u s t r e m o v e 3 b r a n c h e s to o b t a in a tre e . ch o rd s. T h e r e s u lt in g tr e e is s h o w n in F ig . 9 - 3 .
W e ta k e b r a n d ie s
4 , 5 ,
and
a s th e
W e i n d i c a t e t h e c h o r d s b y d a s h e d lin e s .
222
For this selection of a tree,
FORCE METHOD: IDEAL TRUSS
CHAP. 9
i = {h, < 2- h]
*2 = {'4, < 5, 's}
The meshes associated with the chords follow directly from the sketch: mesh 4 mesh 5 mesh 6 - (I), +(2) (5), ( | ) , +(2) (3), +(T) (a)
To assemble we list the branches of the tree vertically and the chord numbers hori zontally. We work with successive columns, that is, successive chords. The resulting matrix is listed below. Note that C 1 is just the matrix equivalent of (a). < ----------- Chords -------4 1 Branches of the tree i I 2 3 -r +
i
5 0 -f1 -1
6 4 1 0 -1
The matrices, A, and A2, follow from Example 9 -2 : -1 0

.
+ 1 4 ~1 + 1 0 0 0
0 0 -1 0 -1 -1
0 0
-I .+ 1 One can readily verify that
Ci = (A2Ai 1)r T he m atrix, C = {Ct , Im}, is called the branch-m esh incidence m atrix. Using (9-23), we see that A and C have the property (/VxAf) ArC = 0 (9-29) Also, we can express the com patibility equations, (9-26), as (Mx 1) Cre 0
(9-30)
The rows of C T define the incidence of the meshes on the branches. Equation (9-30) states th at the sum o f the potential drops aro u n d each m esh m ust be zero and is ju st KirchhofFs voltage law expressed in m atrix form. T he matrix
PR O BLEM S
223
form ulation o f th e netw ork problem leads to the same system of equations that one w ould o b tain by applying K irchhofFs cu rren t and voltage laws to the various nodes an d meshes. This, of course, also applies to the truss problem . The tw o approaches differ only with respect to the assem blage o f the governing equations. In th e conventional approach, one assem bles the equations in dividually. This involves repeated application of the basic laws. W hen the equations are expressed in m atrix form, the steps reduce to a sequence of m atrix m ultiplications.
REFERENCES
1. 2. 3. 4. 5. 6. 7. 8.
N orris , C. H., and J. B. W il b u r : E le m e n ta r y S tr u c tu r a l A n a ly s is , McGraw-Hill, New York, 1960. H a ll , A. S., and R. W. W o o d h e a d : F ra m e A n a ly sis, Wiley. New York, 1967. M orice , P. B .: L in e a r S tr u c tu r a l A n a ly sis, Ronald Press, New York, 1969. R u b in st e in , M. F.: M a t r ix C o m p u te r A n a ly s is o f S tr u c tu r e s , Prentice-Hall, 1966. P rzem ieniecki , J. S.: T h e o ry o f M a t r ix S tr u c tu r a l A n a ly s is , McGraw-Hill, 1968. R u b in st e in , M. F .: Structural Systems Statics, Dynamics, and Stability, PrenticeHall, 1970. Di M a g g io , F. D., and W. R. S piller s : Network Analysis o f Structures/ E ng. M e c h . D id ., A.S.C.E., Vol. 91, No. EM 3, June 1965. F envi -s , S. J.. and F. H. B r a n in , Jr.: Network-Topological Formulation o f Struc tural Analysis, J. S tr u c tu r e s D iv ., A.S.C.E., Vol. 89, No. ST4, August 1963.
PROBLEMS
9 - 1 . Show th a t the coefficient m atrix f 22 is positive definite for arb itrary rank of F 1>f, when f2 is positive definite. Use the ap p ro ach suggested in Problem s 2 -1 2 th ro u g h 2 -1 4 . 9 -2 . Solve th e following system using the procedure outlined in Sec. 9 -2 . Take Xj = { x h x 2) xi 2 1 2 ~2 2 3 1 f 2 4 2_ W Uj j 2 3 4 1
2
r 2 0
2 5
*2
0 0 0
0 0
*3 I X4 0 0 *1 0 0 < -^2 1 0 *3 0 2
3" 3
1
>
2/
9 -3 . C onsider a system o f m equations in n unknow ns, ax c, where m > n. Suppose r(a) = n and the first n row s of a are linearly independent. Let q m n. (a) Show th a t the consistency requirem ent for th e system leads to q relations betw een the elem ents of c. (b) In terp re t (9 -1 0 ) from this p o in t of view.
224
FORCE M E T H O D : iD E A L TR U S S
C H AP. 9
9 -4 .
D ev elo p a n in crem en tal force fo rm u latio n startin g w ith A P, = B , AF A P2 = B 2 AF Ae = B [ A U 2 + B j A 0 2 = Ae'0 + V A F
w here / ' , Ae'0 rep resen t the flexibility facto r an d in crem en tal initial elo n g atio n for the segm ent co rresp o n d in g to th e in itial value o f F. O n e has to m odify b o th / ' an d Aer 0 if th e lim it of th e segm ent is exceeded (see sec. 6 - 4 for a detailed treatm en t). C o n sid er th e case w here the lo ad in g d istrib u tio n is c o n stan t, i.e., w here only th e m a g n itu d e is increased. Let P [ = A \ f/ w here X is th e load p a ram eter an d if/ defines the lo ad in g d istrib u tio n . D iscuss how you w ould o rganize the co m p u ta tio n a l schem e. Also discuss how you w o u ld acc o u n t for eith er yielding o r buckling o f a bar. D istinguish betw een a re d u n d a n t b ar an d a b ar in the p rim a ry structure. 9 - 5 . Solve P ro b . 8 - 3 w ith th e force m e th o d : T ak e F 3 as the force re d u n d an t. 9 - 6 . A ssem ble the eq u atio n s for F 2 = {F s, F 9, F l0, F, J for th e truss show n. Prob. 9-6
Xi
(1 ) (2 ) (3 )
M a te ria l is lin e a r elastic a n d th e f le x ib ilit y fa c to rs are equal. O n ly K 42 is fin ite . Take =
{42 } - 4 2 -
O n ly in itia l e lo n g a tio n fo r b ar 4 .
9 - 7 . F o r the tru ss s h o w n : (a) U sing (9-1 0 ), d eterm in e the elo n g atio n -co m p atib ility relations. T ake bars (6), ( P as the red u n d a n t bars. (b) Express u52 in term s of th e elo n g atio n s an d s u p p o rt m ovem ents. 9 - 8 . By definition (see (7 -2 6 ) an d (7 -3 1 )) d n c = AFre - A P J 2
P R O B LE M S
225 Prob. 9 - 7
T hen d 2n c = d ( d n e) = A F t de E xpress d 2n c as a q u a d ra tic form in A F 2. C o n sid er the m aterial to be n o n lin ear elastic an d establish criteria for the statio n ary p o in t to be a relative m inim um . 9 - 9 . C o n sid er th e o rien ted linear g rap h show n.
Prob. 9 - 9
(a) (b) (c)
D eterm in e A. D eterm in e C. Verify th a t ATC = 0
Part III ANALYSIS MEMBER
OF A ELEMENT
10
Governing Equations for a Deformable Solid

1 0 -1. GENERAL T h e fo rm u latio n o f th e governing e q u atio n s for th e b eh av io r o f a deform able solid involves th e follow ing three steps: 1. Study o f deformation. W e analyze the ch an g e in shape o f a differential volum e elem ent d u e to displacem ent o f the body. T he q u an tities re q u ired to specify the d efo rm atio n (change in shape) are conventionally called strains. T his step leads to a set o f e q u atio n s relatin g th e strain s a n d derivatives of th e d isplacem ent co m p o n en ts a t a point. N o te th a t the analysis o f strain is purely a geometrical problem . Study of f er ee s . W e visualize th e b o d y to consist o f a set o f differential volum e elem ents. T h e forces d u e to th e in teractio n s of ad jacen t volum e elem ents are called internal forces. Also, the in tern al force per u n it area actin g on a differential area, say d Aj , is defined as the stress vector, 5j. In this step, we analyze the state o f stress at a p o in t, th a t is, we investigate how th e stress vector varies w ith o rie n ta tio n o f the a re a elem ent. W e also apply th e co n d itio n s o f static equ ilib riu m to the volum e elem ents. T his leads to a set o f differential eq u atio n s (called stress eq u ilib riu m eq u atio n s) w hich m u st be satisfied a t each p o in t in th e in terio r o f th e b o d y an d a set of algeb raic eq u a tio n s (called stress b o u n d a ry conditions) w hich m ust be satisfied at each p o in t on the surface o f the body. N o te th a t th e stu d y o fferees is purely a n equilibrium problem . Relate forces and displacements. In this step, we first relate the stress and stra in co m p o n en ts a t a point. T h e form o f these e q u atio n s depends o n the m aterial b eh av io r (linear elastic, n o n lin ear elastic, inelastic, etc.). S u b stitu tio n o f th e strain -d isp lacem en t relatio n s in th e stress-strain relatio n s leads to a set of eq u atio n s relatin g th e stress co m p o n en ts a n d derivatives o f th e displacem ent co m p o n en ts. W e refer to this system as the stress-displacem ent relations.
229
2.
3.
230
GOVERNING EQUATIONS FOR A DEFORMABLE SOLID
CHAP. 10
T h e gov ern in g e q u a tio n s for a d efo rm ab le solid co n sist o f th e stress equilib riu m eq u atio n s, stress-d isp lacem en t relatio n s, an d th e stress a n d d isp lacem en t b o u n d a ry co n ditions. In this ch ap ter, we develop th e g o v ern in g eq u a tio n s for a lin early elastic solid follow ing th e steps o u tlin ed above. W e also extend th e v a ria tio n a l principles developed in C h a p te r 7 for an ideal tru ss to a th ree -d im e n sio n al solid. In C h a p te r 11, we p resen t St. V en an ts th e o ry of to rsio n -flex u re o f p rism atic m em bers an d ap p ly th e th e o ry to som e sim ple cross sections. St. V en an ts th eo ry provides us w ith co n sid erab le in sight as to th e n a tu re o f th e b ehavior an d also as to h ow we can sim plify th e co rre sp o n d in g m a th e m a tic a l p ro b lem by in tro d u cin g certain assu m p tio n s. T he co n v en tio n al en g in eerin g th e o ry of p rism atic m em b ers is d ev elo p ed in C h a p te r 12 an d a m o re refined theory for th in w alled p rism atic m em b ers w hich includes th e cffect o f w arp in g o f the cross section is discussed in C h a p te r 13. In C h a p te r 14. we d ev elo p th e engi neerin g th e o ry for an a rb itra ry p la n a r m em ber. F inally, in C h a p te r 15, we p resen t th e engineering th e o ry for a n a rb itra ry space m em ber.
10-2.
SUMMATION CONVENTION; CARTESIAN TENSORS
L et a an d b .rep resen t n th -o rd e r co lu m n m atrices: 3

j ^n}
b - {b u b2, . . . , b n
T h eir scalar (inner) p ro d u c t is defined as
( 10-1)
aTb = bra = a yb i + a 2b 2 + - + ab -
n a {bi i- 1
(a)
T o av o id h av in g to w rite th e su m m a tio n sign, we in tro d u c e th e co n v en tio n th a t w hen an index is rep eated in a term , it is u n d e rsto o d th e te rm is su m m ed over th e range o f th e index. A cco rd in g to this c o n v en tio n
Y
i1
ci'bi = ckbi
(i = 1, 2 , . . . , n)
(10-2)
an d we w rite the scalar p ro d u c t as
arb = a{bi
(1 0 -3 )
T h e su m m a tio n co n v en tio n allow s us to rep resen t o p e ra tio n s o n m u lti dim en sio n al array s in c o m p act form . It is p artic u la rly co n v en ien t fo r fo rm u lation, i.e., estab lish in g th e g o v ern in g eq u atio n s. Wre illu strate its ap p lic a tio n below.
Example 10-1
1. Consider the prod uct of a rectangular matrix, a, and a colum n vector, x. c = ax a is
m x n
(a)
SEC. 1 0 -2 .
S U M M A T IO N C O N V E N T IO N ; C A R T E S IA N T E N S O R S
231
The typical term is

Ci
n
X J=1 Cl i j Xj EE Cl i j Xj (b )
2.
Let a, b be square matrices, x a column vector, and f . g scalars defined by

/ = x r ax
g = x r bx
The matrix form o f the product, fg, is
(c)
fg = (xr ax)(xr bx)

One could expand (d) but it is more convenient to utilize (b) and write (c) as
(d)
/ = atjXfXj
g = b k(x kx c
(e)
Then,
f g = a i p krx iXjXkx c
= DijkeXiXjXyXe
3. We return to part 1. The inner product of c is a scalar, / / ,
(f )
H c 7e = x T(aTa)x
U s in g (b),
H = CjCj = a ika i(x kx r
(g)
(h)
The outer product is a second-order array, d, d = c c T axxTaT and can be expressed as

d jj q c j Qikcij[XkX{
(i)
= A iJkfx kx , According to the summation convention,

da = d u
(j)
+ d 22 + trace of
(k)
T h e n , w e c a n w rite (h ) as
H = d a = A iikCx kx (
(1)
4. Let (Tip e ij represent square second-order arrays. The inner product is defined as the sum o f the products of corresponding elements:
I n n e r p r o d u c t (au, ei}) = a ueu
* j = +
^22^22 +
' ' ' +
&12e l 2 +
<J2 l e 21 +
' ' ' ifij
In order to represent this product as a matrix product, we must convert one-dim ensional arrays.
e i} over to
L et \ b( 2 \ M,1 1 rep resen t a o n e-d im en sio n al set o f elem ents asso ciated w ith an o rth o g o n a l reference fram e h av in g d irec tio n s X ( P , X ^ K X ^ \ If th e
232
G O V E R N IN G E Q U A T IO N S FOR A D E F O R M A B L E S O LID
C H A P . 10
corresp o n d in g set for a second reference fram e is rela ted to the first set by
b f ] = a jk
b[1]
(1 0 -4 )
y.jk =
cos { X f \ X \ t }) J, k
1, 2, 3
we say th a t th e elem ents of b co m p rise a first-o rd er cartesian tensor. N o tin g (5-5), we can w rite (1 0 -4 ) as b(2) = R l2b(1) (1 0 -5 ) and it follows th a t the set of o rth o g o n a l co m p o n en ts o f a vector are a first-order cartesian tensor. W e know th a t th e m a g n itu d e o f a v ecto r is in v arian t. Then, th e sum o f th e sq u ares of the elem ents o f a first-o rd er te n so r is in v arian t.
2)
( 10- 6)
A seco n d -o rd er cartesian te n so r is defined as a set o f d o u b ly su b scrip ted elem ents w hich tran sfo rm acco rd in g to h(2) -y l/mn /)(l) n ,jnv/ jk n /. k, n. )J
= i. 2, 3
(10 -7 )
A n a ltern ate form is b<2) = R 12b(1)(R 12)T

( 10 -8)
T he tran sfo rm a tio n (1 0 -8 ) is o rth o g o n a l an d th e trace, sum o f th e p rin cip al seco n d -o rd er m inors, an d th e d eterm in a n t are in v a ria n t.t
(1 0 -9 ) w here Pi - bjj Pi = jb| b ii Pi = bn b\z bZ2 + &22 b-il bi3 ^33 +

b u b 13
^3i
b}i
In the cases we en co u n ter, b will be sym m etrical.
10-3.
ANALYSIS OF DEFORMATION; CARTESIAN STRAINS
Let P d en o te an a rb itra ry p o in t in th e u n d efo rm cd sta te of a b o d y an d r th e positio n v ecto r for P w ith respect to 0, the o rigin o f an o rth o g o n a l cartesian reference fram e. T h e co rresp o n d in g p o in t an d p o sitio n v ecto r in th e deform ed state are tak en as P'; p and th e m o v em en t from P to P' is rep resen ted by the displacem en t vector, u. By definition, p = r + u T his n o ta tio n is show n in Fig. 1 0 -1 .
t See Prob. 2-5.
(10 10)
SEC. 1 0 -3 .
A N A LY S IS OF D E F O R M A T IO N ; C A R T E S IA N S T R A IN S
233
E xcluding rigid body m o tio n , th e d isp lacem en t from th e initial undeform ed positio n will be sm all for a solid, an d it is reaso n ab le to tak e th e initial cartesian co o rd in ates (xj) as th e in d e p en d en t variables. T his is know n as th e L agrange A3
*2
ap p ro ac h . Also, to sim plify the deriv atio n , we w o rk w ith cartesian co m p o n en ts for u. T hen, a = u{xj) = ujlj
P =
( 10 - 11)
P (*;)
W e consid er a differential line elem ent a t P rep resen ted by th e v ecto r dr. (See Fig. 10-1). T h e in itial len g th and d irectio n cosines are ds an d ay. W e are using the su b scrip t n o ta tio n for p a rtia l differentiation.
J.j = J OXj dr = r j dxj = dxfi; ds(zfij) T h e c o rresp o n d in g line elem en t in the deform ed sta te is dp. follow ing th e L agrange ap p ro a c h , p p ( x j ) , an d we can w rite dp - p ' j d x j = ds(xjp j) The extensional strain ,
e,
( 10- 12)
Since we are
(1 0 -1 3 )
is defined as the relative ch an g e in length with respect
234
C H AP. 10
to th e initial le n g th .t \dp\ = (1 + e)\dr\ U sin g th e d o t p ro d u c t, (1 0 -1 4 ) becom es (1 + e)2 = - r j {dp dp) = ajakp i j p. k F inally, we w rite (a) as e(l + is) = a jakejk
e jk = i ( p . j ' p,k ~ Sjk)
(1 0 -1 4 )
(a)
(1 0 -1 5 )
O n e can read ily estab lish th at (e jk) is a se c o n d -o rd er sy m m etrical cartesian ten so r.J T ak in g th e line elem ent to be initially p arallel to the X , d irectio n an d lettin g i rep resen t th e exten sio n al strain , we see th a t e,(l + k ) = eu (no sum ) it n (1 0 -1 6 ) = z i f K i ' P. i ~ I) T o in terp ret th e off-diagonal term s, ejk, we co n sid er 2 initially o rth o g o n a l line elem ents rep resen ted by dr \ , dr'2 (see Fig. 1 0 -2 ) a n d having d irectio n cosines
Fig. 10-2. N o ta tio n fo r s h e a rin g s tra in .

f T h is is th e d e fin itio n o f L a g r a n g ia n str a in . In th e E u le ria n a p p r o a c h , th e c a r te s ia n c o o r d in a te s (?/,) fo r th e d e fo r m e d s ta te are ta k e n as th e in d e p e n d e n t v a r ia b le s,
U j= U jih)
a n d th e str a in is d e fin e d as , ,
Xj = x f a )
,
\dr\ = (I - e)\dp\
+ S e e P ro b . 10 4. It is k n o w n as G r e e n s str a in te n so r . T h e e le m e n ts, e j h are a ls o c a lle d th e c o m p o n e n ts o f fin ite stra in . T h e y r ela te th e d iffe re n c e b e tw e e n th e s q u a r e o f th e in itia l a n d d e fo r m e d le n g th s o f th e lin e e le m e n t, i.e., a n a lte r n a te d e fin itio n o f ejk is
|<ip|2 - ds2 = 2ejk d x j dxk
SEC . 1 0 -3 .
A N A L Y S IS OF D E F O R M A T IO N ; C A R T E S IA N S T R A IN S
235
dyj, a2j. W e define f y \ 2 as the angle betw een th e lines in th e deform ed state. T h e ex p ressio n for 712, w hich is called th e sh earin g strain , follow s by ta k in g th e d o t p ro d u c t o f th e d efo rm ed vectors.
(n , \ - , dP\ ' dP2 c o s ^ - v , J j = s , 7 , 2 = 1- - . ^

S u b stitu tin g for dp), U --I ,, ,, , \dpj\ - (1 + e'jWj
, , (a)
dPj ~ i& jkP. k)dSj
(sum on k only)
(b)
a n d n o tin g th a t th e lines are in itially o rth o g o n a l, Bj&kt = (a) takes th e form (1 + Sj)(l + s ^ s in > i 2 = 'al fx2keJk (1 0 -1 7 ) (c)
S pecializing (1 0 -1 7 ) for lines p arallel to X h X } show s th a t eu is rela ted to th e sh earin g strain , y;j. (1 + e;)(i + j)sin yu = 2e,j = p t i - f>j (1 0 -1 8 )
E q u a tio n s (1 0 -1 5 ) an d (1 0 -1 7 ) are actu ally tra n sfo rm a tio n law s fo r extensional and sh earin g strain . T h e state o f strain is com p letely defined once th e strain te n so r is specified for a p a rtic u la r set o f d irectio n s. T o g en eralize these expressions, we co n sid er tw o o rth o g o n a l fram es defined by th e u n it vectors tj an d tj (see Fig. 1 0 -3 ), ta k e th e initial fram e p arallel to the g lo b al fram e (tj Tj), an d let otjk = tj ik. W ith th is n o ta tio n :
jj
!X ik< X jf6 kf
(1 0 -1 9 )
K1 + ie'f) = e'u ^ , M n0 su m ) (1 + ,-)(l + j)sin yij 2e;jj
T h e stra in m easu res (e, y) a re sm all w ith resp ect to u n ity for en g in eerin g m a teria ls such as m etals a n d concrete. F o r exam ple, e % 0(1 0 3) for steel. T herefore, it is q u ite re a so n a b le (aside from th e fact th a t it sim plifies th e expressions) to assu m e s , y 1 in th e stra in expressions. T h e rela tio n s for sm a ll stra in are: e.- eit ( 10- 20) }'ij ~ I t rem ain s to ex p an d ejk. N ow , p = r 4- u = (xm + iim)T, (a)
D ifferentiating p with respect to x n
236
G O VER N IN G EQ U A TIO N S FOR A D E F O R M A B L E S O LID
CH AP. 10
an d su b stitu tin g in to th e definition o f ejk (E q u atio n (1 0 -1 5 )) leads to ejk ~ H uj,k + Uk.j) 4- i n m.jU,ik (sum on m only)
( 10-
21)
In o rd er to simplify (10-21), we m u st establish the geom etrical significance of the v ario u s term s. *3
*2
Fig. 10-3. U n it v e c to rs d e fin in g tra n s fo rm a tio n of o rth o g o n a l d ire c tio n s . W ith this objective, we co n sid er a line elem ent initially parallel to th e X x axis. F igure 10 - 4 show s the initial and deform ed positions, a n d th e angles 0 i 3 which define the ro ta tio n o f the line to w a rd th e X 2, X 3 directions. T h e geom etrical relations o f in terest to us are sin 0 t3 = sin 0i 2 (1 + i)2 Also, by definition, i(l + i e ,) = en -- It] t i + j ( l l f i j + U2, i + W3> l )
ll3. 1
1 + fil
_____ 2M ____
(a)
(b)
(1 + e,)co s 0 13 (1 + Hi. i)2 + m|, 1 + 3. 1
(c)
(d)
W e solve (a), (b) for u2, i and u3_t ,

3, i == (1 + ei)sin 0i3 2, i = (1 + i)sin 0 i 2 cos 0 13
(1 0 -2 2 )
SEC. 1 0 -3 .
A N A LY S IS O F D E F O R M A T IO N ; C AR TES IAN STR AINS
237
an d then solve (c) for u l t l . u i. i = (1 + i) {1 ~ A } ilz 1 A sin 2 0 13 + co s2 0 13 sin 2 0 12 A pplying th e binom ial expansion, (1
-
(10-23)
x )i/2
1X(] +
J.x
(10-24)
to (1 A Y 12, we can w rite (10 -2 3 ) as .-I / ,' 1' 1 = S i f _ t ( î-4 + )} ~j{ i , 4 (10-25)
In w hat follows, we assum e small strain an d express the derivatives an d extensional strain (see E q u atio n (d)) as
w 3. i
= O(0l3)
U2 . i
0(0*2, 0?3)
UI, 1 = 0(6] j ^12i ^ 13) ~ e ll u l, l( l + 0(fil, 012, 0?3)) + 0(0 1 2 1 $ 13) T he various ap p ro x im ate theories are o b tain ed by specializing (f). ^3>u3
(e)
(0
Fig. 10-4. In itia l a nd d e fo rm e d p o s itio n s o f a lin e e le m e n t. In the linear geometric case, the ro tatio n s are neglected w ith respect to strain. F orm ally, one sets 0 ]2 = 0i3 = 0 in (f) an d the result is a linear relation between strain and displacem ent, ~ elx ul f l (g) N o te that, according to this ap p ro x im atio n , the deformed o rien tatio n coincides
238
G O VERNING E Q U A TIO N S FOR A D E F O R M A B L E S O LID
CHAP. 10
w ith th e initial orien tatio n . T he general relations for th e linear geom etric case (small strain and infinitesim al ro tatio n ) are ei = eii = uiti
Vij j ~
(no sum )
(10-26)
U j ' j ~f~ Uj ,
T he next level o f a p p ro x im atio n is to consider 92 to be o f th e sam e o rd er as strain. 02 = 0(e) 1 sin 6 & d cos 6 m I W e can neglect u lr x with respect to 1 in (f), but we m ust retain uf, i a d since they are o f O(02). ex m e i i ui, i + i(2. i + w i x) i (10-27)
(h)
The com plete set of strain -d isp lacem en t relations for sm all strain and smallfinite ro ta tio n are listed below for reference. fi; = eH = uu + Huj2 ; + u l i )
}'ij = 2eu = w,. j 4- it :, i + u ki ,-M fc. j
(no sum )
(10-2 8)
i = j
W e utilize these expressions to develop a geom etrically n o n lin ear fo rm u latio n for a m em ber in C h ap ter 18. Lastly, if no restrictio n s are im posed on the m ag n itu d e of the ro tatio n s, one m u st use (10-21). T he relatio n s for finite ro ta tio n an d sm all strain are , = eu = uif i +
] O' tfj, j ( l T
, + uj:, + /?.;)
U[ i ) -j- U j ;(1 Uj j )
(no sum)
-|fU^j
([() ^
i i= - j = k N o te th a t th e tru ss fo rm u latio n presented in C h a p te r 6 allow s for arb itra ry m ag n itu d e of the ro tations. W e have show n th at linear strain -d isp laccm en t relations are based on the follow ing restrictio n s: 1. 2. T h e strains are negligible w ith respect to unity, and P ro d u cts of the ro ta tio n s are negligible w ith respect to th e strains.
T he first co n d itio n will alw ays be satisfied for engineering m aterials such as m etals, concrete, etc. W h eth er the second restrictio n is satisfied d epends on th e co n figuration o f th e b o d y an d th e ap p lied loading. If the body is m assive in all th ree directions, the ro ta tio n s are negligible w ith respect to th e strains for an arbitrary loading. W e have to include th e n o n lin ear ro ta tio n term s in th e strain displacem ent relatio n s only if th e body is th in (e.g., a th in p late or slender m em ber) a n d the applied lo ad in g results in a significant change in the geom etry. As an illu stratio n , consider the sim ply su p p o rted m em b er show n
SEC. 1 0 -3 .
A N A LY S IS OF D E F O R M A T IO N ; C A R TE S IA N STR AIN S
239
in Fig. 10 -5 . W e can neglect the change in geom etry if only a transverse loading is applied (case 1). H ow ever, if b o th axial and transverse loads are applied (case 2), th e change in geom etry is no longer negligible an d we m ust include the n o n lin ear ro ta tio n term s in the strain-displacem ent relations.
/////
Case 2 ( Q f ) Case 1 2 )
F ig . 1 0 - 5 . E x a m p le o f lin e a r and g e o m e tric a lly n o n lin e a r b e h a v io r.
To tre a t a geom etrically n o n lin ear problem , we m ust w ork w ith th e deform ed geom etry ra th e r th a n the initial geom etry. This can be defined by track in g the m ovem ent o f a tria d of line elem ents initially parallel to th e global directions. We let dfj be the initial set a n d dpj th e deform ed set (see Fig. 1.0 -6 ). By definition, dT'j = dxfij dpj = p t j dxj \dPj\ = (1 + j)dxj T he unit vector p o in tin g in th e d irectio n of dpj is d en o ted by Vj. U sing (a), wc can w rite f' f T e , f - 1 ^ p_ j for sm all strain (no sum) (no sum)
(b)
Finally, we express Vj in term s o f the unit vectors for the initial frame.
Vj ~ fij'J k
a 1 Pjk - Y T T -
jk
, 1 1 k-^ for sm all strain
(1 0 -3 0 )
SJk + ukj
W e will utilize (10 -3 0 ) in the next section to establish th e stress equilibrium eq u atio n s for th e geom etrically n on lin ear case. E q u atio n s (1 0 -3 0 ) reduce to v,. % 7j (10-31) for the geometrically linear case an d to Vj ij + PjkTk + Pjfi t j jt k I (no sum ) { }
for the case of small strain and small-finite rotations.
240
GOVERNING EQ UATIO N S FOR A D E FO R M A B LE SO LID
CHAP. 10
Fig. 10-6. In itia l and d e fo rm e d g e o m e trie s .
10-4.
ANALYSIS OF STRESS
T h e effects of the surroundings on a body such as co n tact pressure, g rav ita tional attractio n , etc., result in internal forces. In this section, wc establish the equilibrium conditions for the internal forces in a body. This step is generally called the analysis of stress. C onsider a body subjected to some effect w hich results in internal forces. We pass a cutting plane th ro u g h the deform ed body and separate the tw o segm ents as show n in Fig. 10-7. W c let m denote the outward n o rm al direction for the internal face of body and refer to this face as the + m face. In general, the subscript, m, is used for quantities associated w ith the + m face. N ow , we consider a differential area elem ent A A m, and let A F, be the resultant internal force vector acting on this element. T he stress vector, g,. is defined as < ?m lim
6Am - 0
(10-33)
N o te th a t 5m has the units of force/area. Also, it depends o n the orientation o f the area element, i.e., on the direction of the o u tw ard norm al. W e do not allow for the possibility of the existence of a m om ent acting on a differential area element. O ne can include this effect by defining a couple-stress v ecto rf in ad d itio n to a stress vector.
t See Ref. 6. p. 68.
SEC. 1 0 -4 .
AN ALYSIS OF STRESS
241
W e consider next the corresponding area elem ent in the m face. From N ew tons law, A F _ m = - A Fm (a) and it follows th at <7-m = ~ O m (10-34) The stress vector has the sam e m agnitude an d line of action b u t its sense is reversed.
Note: Deformed state
Fig. 10-7. N otation fo r in te rn a l fo rc e . In o rder to analyze the state of stress at a point, say Q , we need an expression for the stress vector associated with an arb itrary plane through Q. W ith this objective, we consider the tetrah ed ro n shown in Fig. 10 -8 . T he o rien tatio n of the arbitrary plane is defined by q, the outw ard no rm al direction. T he outw ard norm als for the o th er three faces are parallel to the reference axes ( X y , j = I. 2, 3). To simplify the notatio n , we use a subscript /' for quantities associated with the X j face, that is, the face whose outw ard n o rm al points in the + X j direction. F o r exam ple, we write
Xj = C - X }, = A A Xj ffj d-j = AAj -C j
(10-35)
etc. The force vectors acting at the centroids of the faces are show n in Fig. 10-8. The term A<r() represents the change in 6() due to tran slatio n from Q to the centroid. F o r equilibrium , the resultant force and m om ent vectors m ust vanish. In the limit (as P -* Q), the force system is concurrent an d therefore we have to
242
G O VERNING EQ U A TIO N S FOR A D E F O R M A B LE S O LID
CHAP. 10
consider only the force equilibrium condition. F ro m Fig. 1 0 -8 , we have 5q + A 5 q = AA 1 (dj + A (ij)
l\A q
(a)
N ow , A A ! is the projection o f A A q on the X 2- X 3 plane. N o tin g th a t the projec tio n of A A q o n a plane is eq u al to A A q times the scalar p ro d u c t o f i q an d the unit n o rm al vector for the plane, an d letting aqj be the d irectio n cosine for the q direction w ith respect to th e X j d irectio n , we can w rite AA -j-J- = aqJ = cos(g, X j) = i 1, Finally, in th e limit, E q u atio n (a) reduces to
Oq = Vqjfj
(10-37)
(10-3 6 )
O nce the stresvs vectors for three o rth o g o n al planes at Q are know n, we can deter m ine the stress vector for an a rb itra ry plane th ro u g h Q w ith (10-37). *2
Fig. 10-8. D iffe re n tia l te tra h e d ra l e le m e n t. E q u atio n (1 0 -3 7 ) is the tran sfo rm a tio n law for th e stress vector. T h e co m po n en t o f d q in a p articu lar d irectio n is equal to the scalar p ro d u c t o f d q an d a u nit vector p o in tin g in th e . desired direction. N ow , we express the stress vectors in term s o f th eir com p o n en ts w ith respect to th e co o rd in a te axes X j (j = 1, 2, 3).
SEC. 1 0 -4 .
A N A L Y S IS OF STRESS
243
N o te th a t the first su b scrip t on a stress co m p o n en t alw ays refers to th e face, an d the second to th e direction. F o r exam ple, a l2 acts on th e X x face and p o in ts in th e X 2 d irectio n . T h e positive sense of the co m p o n en ts for a negative face is reversed since <?_, = dj. T h e n o rm al ( < jj}) and in -p lan e (ajk) co m p o n en ts are generally called n o rm al an d sh earin g stresses. T his n o ta tio n is illu strated in Fig. 1 0 -9 .
Xi
Fig. 1 0 - 9 . N o ta tio n fo r s tre s s c o m p o n e n ts .
S u b stitu tin g for th e stress vcctors in (1 0 -3 7 ) results in = v-qfjk (1 0 -3 9 )
T he c o m p o n en t o f 5 q w ith respect to an a rb itra ry d irectio n , m, is d eterm in e d from

Gqm Gq
h i
1*0 (b)
L etting * = XmlJk an d n o tin g (10-38), (a) ex p an d s to = xqp mk j k (c)
W e generalize (c) for tw o o rth o g o n a l fram es specified by th e u n it vectors t h i'j (see Fig. 1 0 -3 ) w here C J = lj t j = <Xjkh (1 0 -4 0 )
D efining g \j as th e c o m p o n en t acting on th e V, face in the t] d irectio n and identifying f r ' - w ith l q, Jm, (c) tak es th e form . < j = <*ik*jrkt T his result show s th a t th e set (<7;j) is a seco n d -o rd er cartesian tensor. (1 0 -4 1 )
244
GOVERNING EQ UATIONS FOR A D E FO R M A B LE SO LID
CHAP. 10
It rem ains to establish the equilibrium equations for a differential volum e element. The equilibrium equations relate to the deform ed state, i.e., we must consider a differential elem ent on the deformed body. Since we have defined the stress com ponents w ith respect to the global cartesian directions, it is n atural to w ork w ith a rectangular parallelepiped having sides parallel to the global directions. This is show n in Fig. 10-10. P o in t 0 is at the centroid of the element. h
Fig. 10-10. D iffe re n tia l v o lu m e e le m e n t in E u le ria n re p re s e n ta tio n . The stress vectors are considered to be functions of the d efo rm cd t coordinates (rji). We obtain the forces acting on the faces by expanding the stress vectors ab o u t 0 and retaining only the first two te rm s .| L etting b denote the external force per unit volum e and enforcing the equilibrium conditions leads to YF = 0 ^ ^
c ,li
+ b = 6
(10-42)
and M 0 = 0 = * 7j x dj = 0 (10-43)
T he scalar force equilibrium equations are o b tain ed by expanding the vector equations using (10-38). F orce equilibrium
. . .
drjj
+ bk = 0
k = 1 ,2 ,3 A 7^ I
(10-44)
M om ent equilibrium
ajk ak]
- j ^
(10-45)
M o m en t equilibrium requires the shearing stress com ponents to be symmetrical. T hen, th e stress ten so r is sym m etrical and there are only six independent stress m easures for the three-dim ensional case and three for the tw o-dim ensional case.
f We are following the Eulerian approach here. Later we will shift back to the Lagrange approach. I Second- and higher-order terms will vanish in the limit, i.e.. when the element is shrunk to a point.
SEC, 1 0 -4 .
ANALYSIS OF STRESS
245
E quations (10-44) m ust be satisfied at each point in the interior of the body. Also, at the boundary, the stress com ponents m ust equilibrate the applied surface forces. We define v as the outward norm al vector at a point on the deformed surface and write t v = P r f j (10-46) The external force per unit deformed surfacc area is denoted by p . Pn = Pnjh (10-47)
Applying (10-37) leads to the stress-boundary force-equilibrium relations: Pn ~ Pnk&k V Pnj $nk@ kj
(10-48) j 112, 3
W hen pj is prescribed, i.e., pnj = pnj, (10-48) represent boundary conditions on the stress com ponents. If iij is prescribed, pnJ is a reaction. O ur derivation of strain-displacem ent relations em ployed the Lagrange approach, i.e., we considered the displacem ents (and strains) to be functions of the initial coordinates (xt). The analysis of stress described above is based on the Eulerian approach, where the deformed coordinates are taken as the independent variables. This poses a problem since the strain an d stress measures are referred to different volum e elements. Figure 10-11 shows the initial and
Fig. 1 0 -1 1 . C o m p a riso n of E u le ria n and L a g ra n g ia n re p re s e n ta tio n s fo r a v o lu m e element. tSee Prob. 10-12.
246
G O V E R N IN G E Q U A T IO N S FOR A D E F O R M A B L E S O L ID
C H A P . 10
deform ed a re a elem ents c o rresp o n d in g to th e tw o view points. T o be consistent w ith the L ag ran g e strain s, we m u st w o rk w ith a n o n o rth o g o n a l p arallelepiped w hose sides are p arallel to th e d efo rm ed line elem ents in th e an aly sis of stress. C onversely, to be co n sisten t w ith th e E u lerian stresses, w e h av e to refer the strain m easu res to n o n o rth o g o n a l d irectio n s in th e in itial state. In the lin ear g eo m etric case, we assu m e sm all strain an d neglect th e change in o rie n ta tio n due to ro tatio n . T h e tw o ap p ro ac h es coalesce a n d we ju s t have to replace w ith Xj a n d jjllk w ith ak w here y.llk is the d irectio n co sin e for the initial d irec tio n of th e ex terio r n o rm al. T h e linear eq u ilib riu m eq u a tio n s are: da f 4 + K = o xi i cx Pnj = ank~kj
(10-49)
F o r th e geo m etrically n o n lin e a r case, we w ork w ith stress m easu res referred to the d efo rm ed d irectio n s (sec Fig. 1 0 -6 ) defined by th e u n it v ectors, vr We define a) as the stress vecto r p er u n it initial a re a actin g on th e face which initially is n o rm ai to th e d irectio n , b* as the force per u n it initial volum e, and p* as the force p er u n it initial surface area. F ig u re 1 0 -1 2 show s this n o ta tio n for th e tw o -d im en sio n al case. T h e stress an d force v ectors are co n sid ered to be functions o f th e in itial co o rd in a tes (x ;). T h e eq u ilib riu m eq u a tio n s at a n in te rio r p o in t are fl a) b * = 0 pxj (1 + r,j)vj x a) = 0 W e express th e b o d y force an d stress vectors as b* = h f l . k ' h 'k Oj = (Tji{\ + Ci)Vi (10-5 1 ) (10-5 0 )
T h e set, of,-, is called the K irch h o ff stress ten so r. S u b stitu tin g for vh using (10-30), resu lts in th e follow ing scalar eq u atio n s, w hich c o rre sp o n d to (1 0 -4 4 ) an d (1 0 -4 5 ): ~ (cr)( + r f p t ' ,) + b? - 0 OXj (. 1, 2, 3 (1 0 -5 2 )
4. - 4,
T h e b o u n d a ry eq u ilib riu m eq u a tio n s a re o b ta in e d by ex p an d in g Pn = = P*J1J
00-53,
(1 0 -5 4 )
and have the form

Pnj = at( h + OflUj,i)
(10-55)
SEC. 1 0 -4 .
A N A L Y S IS OF STR ESS
247
T hese e q u a tio n s ap p ly for a rb itra ry strain an d finite ro ta tio n . F o r sm all strain, we n eglect th e ch an g e in d im en sio n s a n d sh ap e o f th e v o lu m e elem ent. This assu m p tio n is in tro d u c e d by ta k in g b* as b p* as p a) % rfiV; (1 0 -5 6 )
Since th e d efo rm ed u n it v ectors are o rth o g o n a l (to s 1), th e K irch h o ff stresses ffjj n o w co m p rise a se c o n d -o rd er cartesian te n so r an d they tra n sfo rm acco rd in g
(I + dx 2
)dX2
b dx\ dx2
---------------------------------------- ---
Fig. 10-12. D e fin itio n o f s tre s s c o m p o n e n ts in L a g ra n g ia n r e p re s e n ta tio n . to (10-41). T h e eq u a tio n s sim plify so m ew h at if we assu m e small-finite ro ta tio n .! F o r infinitesim al ro ta tio n (linear geom etry), < jk j( ss aj r Vj & ij, an d th e e q u a tio n s red u ce to (1 0 -4 9 ), (10-50). In w h at follow s, we will w o rk w ith th e K irch h o ff stress c o m p o n en ts to keep th e tre a tm e n t general. H ow ever, we will assu m e sm all strain.
+ See Prob. 10-16.
248
GOVERNING EQ UATIO N S FOR A D E F O R M A B LE SO LID
CHAP. 10
10-5.
ELASTIC STRESS-STRAIN RELATIONS
A body is said to be elastic if it retu rn s to its initial dim ensions an d shape w hen the applied forces are rem oved. T he w ork done d u rin g the d efo rm atio n process is independent of the o rd er in w hich th e body is deform ed. W e treat first an a rb itrary elastic m aterial an d then specialize the results for a linearly elastic m aterial. O u r startin g p o in t is the first law of th erm odynam ics: 5 W = d V T + 6Q (a)
w here 8 W = first-order w ork done by th e forces acting on the body 5V t = first-order change in th e to ta l strain energy (also called internal energy) 5Q first-order change in the to tal h eat content. W hen the d efo rm atio n process is isotherm al or' adiabatic, dQ 0, an d (a) re duces to S W = SVT (b) N ow , we apply (b) to a differential volum e clem ent in the deform ed state (see, e.g., Fig. 10-12). W e define V as the strain energy p er unit initial volume. In general, V is a function of the d efo rm atio n m easures. V V(ei}) = ......... Vj2 ) (10-57)
T h e m aterial is said to be hypcrelastic (G reen-type) when V is a co n tin u o u s function. This requires d 2V c 2V (10-58) dek( dei} ceu ffekr By definition, 5VT S V( dx j d x 2 d x 3) PV SV ^~~5eu ceL j (10-59)
w here de^ is the first-order ch a n g e f in etj due to an increm ental displacem ent, Aw. Also, one can show th a t th e first o rd er w ork d o n e by the force vectors acting on the elem ent is J < 5W = ( { o \ ' A u \ , + {d\ A ),2 + {d\ Au);3 + b A u ) d x l d x 2 d x 3 = K / ^ ij) dxl dx2 ^ 3 E quatin g SVT and b W leads to the general form of the stress-strain relatio n for a G reen-type m aterial, 4- = c.'el7
f See Prob. 10 J1. t See Prob. 10-18. The forces are in equilibrium, ie. they satisfy (10-50).
(10_ 6Q)
(10-61)
SEC. 1 0 -5 .
ELASTIC STRESS-STRAIN R ELATIO N S
249
This definition applies for arb itrary strain. O nce V is specified, we can obtain expressions for th e stresses in term s of the strains by differentiating V. Since V is continuous, the stress-strain relations m ust satisfy (10-58), which requires (10-62)
oenm
dCij
In w hat follows, we restrict the discussion to small strain and a linearly elas tic m aterial, i.e., w here the stress-strain relations are linear. W e also shift from indicial n o ta tio n to m atrix notation, which is m ore convenient for this phase. We list the stress and strain com ponents in colum n m atrices an d d ro p the superscript k on the K irchhoff stress co m p o n en ts:
I ^ I I j a 22> a 33> c> 1 2 ' n 2 3 3 1 I
= {e x i ^22, ^33, 2e12, 2e23, 2e 31j
(10-63)
{fiie23)l2>23>3l} <5V = cfj 5etj => g t <5e (10-64)
W ith this no tatio n , The m atrix tran sfo rm atio n laws a r e f c = 1>
e'
= Te.
(10-65)
Since S V is invariant u nder a transform ation of reference frames, the tran s form ation m atrices are related by (Tff) % = I The total strain, e, is expressed as
E - c + Act
(10-66)
(10-67)
where s co n tain s the initial strains n o t associated w ith stress, e.g., strain due to a tem p eratu re increm ent, and A is called the m aterial com pliance m atrix. We w rite th e inverted relations as
< t D (e e )
(10-68)
where D = A -1 is the m aterial rigidity m atrix. E q u atio n (10-6 2 ) requires D (and A) to be sym m etrical. T he elem ents of A are determ ined from m aterial tests, and D is generated by inverting A. S u bstituting for <r in (10-64), we obtain the form of the strain energy density for the linear case, v = \ ( Z - )TD( - ) (10-69)
Since V > 0 for a rb itra ry (e ), D and A are positive definite matrices. There are 21 m aterial co n stan ts for a linearly elastic G reen-type m aterial. The num ber of independent co n stan ts is reduced if the m aterial structure
t See Prob. 10-6, 10-13.
250
C H A P . 10
exhibits sy m m etry .f In w h at follows, we d escribe the tra n sitio n from a n a n iso tro p ic m a te ria l to a n iso tro p ic m aterial. A m a te ria l w hose stru c tu re has th ree o rth o g o n a l axes o f sy m m etry is called o rth o tro p ic . T h e stru ctu re o f an o rth o tro p ic m a te ria l ap p e a rs id en tical after a 180 ro ta tio n a b o u t a sy m m etry axis. T o d eterm in e the n u m b e r o f in d e p e n d e n t co n sta n ts for this case, we su p p o se X lt X z , X z are axes o f sy m m etry an d co n sid er a 180 ro ta tio n a b o u t X z . W e use a p rim e su p erscrip t to in d icate the ro ta te d axes. F ro m Fig. 1 0 -1 3 , X\ = - X x X3 X3 (a)
X'2 = X2
T h e stress and d efo rm a tio n q u an titie s are related by (we rep lace 1 by 1 an d 3 by - 3 in th e sh ear term s) < =
1 2
*if ~ ff12
e.' = c<
^23 ~ &23 3 = ~ )23
i=
1 ,2 ,3
(b)
"l3 ";3
V1 2 = ~Jl2
y 13 = > ' 13
N o w , th e stress-strain rela tio n s m u st be id en tical in form . W e ex p an d f. = Atr, z 1 A<r', an d s u b stitu te for <r' using (b). E q u a tin g th e ex p ressio n s for et
X2
Fig. 10-13. Rotation of a x e s for sym m etry with re sp e ct to the X2-X3 p lane,
t A material whose structure has no symmetry is said to be anisotropic.
SEC. 1 0 -5 .
E L A S T IC S T R E S S -S T R A IN R E L A T IO N S
251
a n d leads to th e follow ing relatio n s betw een th e elem ents o f A,

a l4-&l2 a 24'l2 + + r t 1 5 a 23 = a 2 5 a 23 ~ a 3S&23 = ~ a l 4 (Tl 2 ~ a 24-i 2 ~ ~ ^15^23 a 2 5 a 23 ( C)
a 3 i G 12 +
~ ~ a 3 4 <712 a 3 5 2 3
F o r (c) to be satisfied, th e coefficients m u st vanish identically. This requires

a l4 = a 24 = 34 ~ a l5 = 0 0 0
2 5 a 35 =
W e co n sid er n ex t th e ex p an sio n s for y'j . T he sym m etry co n d itio n s req u ire a 46 = a 56 = 0. By ro ta tin g 180 a b o u t X j, we find
a 16 ~ a 26 a 3 6 ~ a A5 ~ 0 (e )
A ro ta tio n a b o u t th e X 2 axis will n o t resu lt in an y ad d itio n a l conditions. F inally, w hen the stra in s are referred to the structural sy m m etry axes, the stress-strain relatio n s for an o rth o tro p ic m a teria l reduce to
-I F '2 S3 V12 y 23 >=
al 1
a l2 a !3
f a l2 a 22 #23 13 a 23 ^33 j !
0
\ < * ii &12 &33
|
j 4
0 0 0
< <?12 <?23 <731
Vn
! 0 i o
flss
0
a 6b
W e see th a t A is q u asi-d iag o n al an d involves 9 in d e p en d en t co n stan ts. T h ere is no in teractio n betw een ex ten sio n an d shear. Also, th e sh earin g effect is u n coupled , i.e., crJ2 leads on ly to y l2 . An a lte rn a te form o f th e o rth o tro p ic stress-strain rela tio n s is
1
J l
+ l +
V 21 1 (T22
V31 33
'
<1
E 2
3 V3 2 a 33
C2 =
l*-2
AT AT
*12 r22 ' *11
El 1
3
E~,
1 13
'
V23
<*11
fi3 ~
M3
<*33
"
Ei
' X
<y2 2
'/12 r
^12
c \2
y 23 ,
'- '2 3
7
'-r 31
3i
w here E t are ex ten sio n al m o d u li, G (j are sh ear m o d u li, v jk are co u p lin g coeffi cients, an d A T is th e te m p e ra tu re increm ent. T h e co u p lin g term s are related by
252
G O VERNING EQ U A TIO N S FOR A D E F O R M A B LE S O LID
CH AP. 10
I t is relatively straig h tfo rw ard to invert these rela tio n s.! O n e sh o u ld n o te th a t (10-7 1 ) apply only when X t coincide w ith the m aterial sym m etry directions, t If the stress-strain relations a re in v arian t for arb itra ry d irections in a plane, the m aterial is said to be transversely o rth o tro p ic o r iso tro p ic w ith respect to the plane. W e consider the case w here th e X x d irectio n is th e preferred d irec tion, i.e., w here th e m aterial is iso tro p ic with respect to the X 2- X 3 plane. By definition, A is in v a rian t w hen we tran sfo rm from X l - X 2- X i to X ^ X ^ - X ' ^ . This requires
V;
E2 = E 3 ~ E V32 E? '3 V23 E *-2 v
G 12 1 G2 3
G 31 = G
e
'
(1 0 -7 3 )
2(1 _+ v)
A *2
= M
and th e relations reduce to ex = iiy A T + ^ r ( 7 u ft 1 1 H & T + -=r(c22 ~ e2 kit 1 (cr22 t 1 cr33) + ~ ver3.0 voZ2) 2(1 + v)
> 23
(10-74)
e3 = n A T + - (<r33 - vIcfu 1
7l 2 r
1
12
Gi
> 3 ,t
<>1
7 7 - O'.U
7 ,------- ^ 2 3
T here are five in d ep en d en t co n stan ts (, v, j , v-j, G {). Lastly, th e m aterial is called iso tro p ic w hen th e stress-strain relatio n s are in v arian t for a rb itrary directions, X \ - X ' 2-X'3. F o r this ease, A = A' for a rb i trary X ^ - X j - X ' ^ . T he relatio n s are o b tain ed by specializing (1 0 -7 4 ): G i = n AT +
2(1 + v)
----
~ ( 7 i -
v { o Li
<rk k ) )
(10-7 5 )
!
N o te th a t now there are only tw o in d ep en d en t co n stan ts (, v). T he coupling coefficient, v, is called P o isso n s ratio. The inverted form of (10 -7 5) is w ritten as a a = cr + (A + 2 G)e,i + /.{Ej + %) a ;j - Gyu a 0 = - ( 3 a + 2G)j.i A T
t See Prob. 10-19 for the inverted form of (10-71). I See Prob. 10-21
See Prob. 10-22.
(10-76)
SEC. 10 6.
P R IN C IP LE OF V IR T U A L D IS P LA C E M E N T S
253
w here / . G are called L am e co n stan ts an d are related to E, v by E G = sh ear m odulus = X = vE (1 + v)(l - 2v) 2(1 + v) (10-77)
Since D m u st be positive definite, v is restricted to 1 < v < 1/2. T he lim iting case w here v = + 1/2 is discussed in P ro b lem 1 0-24.
10-6.
PRINCIPLE OF VIRTUAL DISPLACEMENTS; PRINCIPLE OF STATIONARY POTENTIAL ENERGY; CLASSICAL STABILITY CRITERIA
C h a p te r 7 dealt w ith v aria tio n a l principles for an ideal truss. F o r co m pleteness, we derive here th e 3-dim ensional form of th e principle of v irtual displacem ents, principle of statio n ary p o te n tia l energy, an d th e classical stability criterion. T h e p rin cip le o f virtu al forces an d sta tio n a ry co m p lem en tary energy are tre a te d in th e next section. T h e principle of v irtu al displacem ents states th a t th e first-order w o rk d o n e by the extern al forccs is equal to th e first o rd e r w o rk d o n e by th e in tern al forces (SW D) acting on the restrain ts for an a rb itra ry virtual displacem ent of the body from an eq u ilib riu m position, f In th e co n tin u o u s case, the external load in g consists of b o d y {b) an d surface (p) loads an d th e in tern al forces are rep resen ted by th e stress vectors. W e follow th e L agrange ap p ro ach , J i.e., we w ork w ith L agrange finite strain co m p o n en ts (ejk), K irchhoff stresses (a*), an d external force m easures per u n it initial volum e o r area (b*, p*). T his is consistent w ith o u r d eriv atio n o f the equilib riu m equations. Let A u d en o te th e v irtu al displacem ent. T he firsto rd er ex tern al w o rk is Swf . = Lfj^* ' ^ d x i d x 2 clxi + jjP* Am dCl = [j fb f Au; d x 1 d x 2 dx$ + j[p*( A ui dQ, (1 0 -7 8 )
w here Q is th e in itial surface area. T h e to ta l in tern al d efo rm atio n w o rk is o b ta in ed by sum m ing the first-o rd er w ork d o n e by th e stress vectors acting on a differential volum e elem ent.
5
w d
Mi*? Au
j d x i d x2 d x 3
(10-79)
3 e tJ d x y d x 2 d x 3
E q u atin g (a) an d (b), we o b ta in the 3-dim ensional form o f the p rin cip le o f
t See Sec. 7-2. See Fig. 10-12. 8 See (10-60).
254
GOVERNING EQ UATIONS FOR A D E F O R M A B LE SO LID
CHAP. 10
virtual displacem ents, 5W d = SW E JJjffJ * AU j d x i d x 2 d x 3 = * Au d x t d x 2 d x 3 + jjf tf Aw c/fi (10-80)
I! ' .. JJJcr^ de{] d x x d x 2 d x 3 j j \b* Aw, d x { d x 2 d x 3 + j j p* A u, dQ R equiring (10-80) to be satisfied for arb itrary (continuous) AH is equivalent to enforcing the equilibrium equations. T o show this, we w ork with the vector form and utilize the following inte gration by parts fo rm u la : |
'yr
f ~ ij
dxy dx2 dx3 =
f g< xnj dQ -
g - d x , dx2 dx3 ( :.X:
(10-81)
w here txnj is the direction cosine for the initial outward n o rm al () w ith respect to the X j direction. O p eratin g on the left-hand term and eq uating coefficients o f Au in the volum e and surface integrals leads directly to (10 -5 0 ) and (10-54). The principle o f virtual displacem ents applies for a rb itrary loading (static or dynam ic) and m aterial behavior. W hen the behavior is elastic an d the loading is independent o f time, it can be interpreted as a v ariatio n al principle for the displacem ents. T he essential steps required for the truss fo rm u latio n are de scribed in Sec. 7 -4 . Their extension to a co n tin u o u s body is straightforw ard. W hen the behavior is elastic, .* - i v ffy = - r <V,j (a )
L etting ^ d e n o t e the to tal strain energy, the left-hand side o f (.10-80) reduces to 8W d = Jjjrxf,' Setj d x , d x 2 d x 3 -> fJJ < 5V d x , d x 2 d x 3 SVT We consider th e surface area to consist of 2 zones as show n in Fig. 10-14. Q 4- Q,, (b)
w here displacem ents are prescribed on Qd, u; = U ; on Q d (10-82)
and surface force intensities are prescribed on Pm = Pm on
T he displacem ent variation, Auh is adm issible if it is co n tin u o u s and satisfies Au, = 0 o n fid (10-83)
W e also consider the surface and body forces to be independent of the displace ments. W ith these definitions, the principle of virtual displacem ents is trans-
i See Prob.
10-25.
SEC. 1 0 -6 .
PR IN C IP LE OF V IR TU A L D IS P LA C E M E N TS
255
form ed to c>nj; = 0
Tl = VT -
for a rb itrary adm issible Aw; fJjSfii/ d x t

dx2 dx3 -
(10-84)
ff p * Hi dQ .
w here n p is the to tal p o ten tial energy functional. A ccording to (10-84), the displacem ents defining an equilibrium position co rresp o n d to a statio n ary value of the total p o te n tial energy functional. N o te th a t this result applies for a rb itrary strain and finite ro tatio n s. T he only restrictions are elastic behavior and static loading.
Fig. 10-14. C la s s ific a tio n of b o u n d a ry zo n e s. Exam ple 1 0 - 2 ------------------------------------------------------------------------------------------Direct methods of variational calculus such as Rayleigh-Ritz, Galerkin, weighted resid uals, and others are applied to n p to determine approximate solutions for the displacements. In the Rayleigh-Ritz method, one expresses the displacements in terms of unknown param eters, q, and proscribed functions, </>(xj, x 2, x 3),
/ =
u? +
i=i
on Q,j
(a )
where
up
<jtf = 0
1 for; = 1 , 2 , . . . , N ]
(b) v
The displacement boundary conditions on Qj are called essential boundary conditions. Substituting for u-t transforms n p to a function of the q 's. When the material is linearly elastic, V is a quadratic function of the strains. Then, V will involve up to fourth-degree terms for the geometrically nonlinear case. If the behavior is completely lin ea r, U p reduces to: n = Const. + qrQ + iq rKq
q = { q W Qn V i2) l*3)} (-^v x i)

K is symmetrical
^c)
256
G O V E R N IN G E Q U A T IO N S FOR A D E F O R M A B L E S O L ID
C H A P . 10
Finally, requiring n to be stationary for arbitrary <5q leads (for linear behavior) to Kq = Q (d)
The strains are evaluated by operating on (a) and the stresses are determined from the stress-strain relations. Polynom ials and trigonometric functions are generally used to construct the spatial distribution functions. The mathematical basis for direct methods is treated in numerous texts (see Refs. 9, 10).
T he classical stab ility criterio n for a stab le eq u ilib riu m p o sitio n is f

c )2Wd
52WE > 0
for a rb itra ry Au
w here 5 2WE 5(5 WF ) is th e seco n d -o rd er w o rk d o n e by the ex tern al forces d u rin g th e in crem en tal d isplacem ent, A u , a n d ' d 2WD 3{oWn) is th e secondo rd e r w o rk d o n e by the in tern al forces actin g o n the re stra in ts d u rin g the in crem en tal d efo rm a tio n resulting from Au. T h e form o f th e w ork term s for a co n tin u o u s b o d y are o b ta in ed by o p e ra tin g on (1 0 -7 8 ) a n d (1 0 -7 9 ) S 2We J J | 3b* Au d x y d x 2 d x 3 + [j dp* An dQ
= JJJ 8b? A U j d x x d x 2 d x 3 + JJ dp? Auf dQ

S 2Wd = J |j 36 k ' Au j d x t d x 2 d x 3 = ,fi l i a ' l l dei} + o$jd2eu ) d x x d x 2 d x , If < 5 2Wd = d2 We for a p a rtic u la r Au, th e eq u ilib riu m p o sitio n is n eu tral. T h e positio n is u n stab le if 5 2 WD < S 2WE. N o te th a t 6b, 5p are null vectors w hen th e forces are prescribed. F o r elastic b eh av io r, the in crem en tal d efo rm a tio n w o rk is eq u al to th e increm en t in stra in energy ( bWD = 5V,), an d (1 0 -8 4 ) can be w ritten as S 2n p = 5(<5np) > 0 for a rb itra ry A ft (1 0 -8 6 )
I t follow s th a t a stab le eq u ilib riu m p o sitio n c o rresp o n d s to a relative m in im u m value o f the to ta l p o te n tia l energy. B ifurcation (n eu tral eq u ilib riu m ) occurs w hen < 5 2n p = 0 for som e Au, say A iiB. If th e lo ad in g is p rescribed, c>2ITp = 3 2 Vr , a n d 3 2Vt = 0 a t b ifu rcatio n . T he governing e q u atio n s for b ifu rca tio n can be o b ta in e d by ex p an d in g 8 2Wd = 32We. T h is involves tran sfo rm in g th e in teg ran d o f 6 2Wn by ap p ly in g (10-81). Since b ifu rca tio n co rresp o n d s to th e existence o f a n a lte rn a te eq u ili b riu m p o sitio n , it is m o re co n v en ien t to form th e in c re m en tal eq u a tio n s directly. T he e q u atio n s for th e case o f linearly elastic m a teria l a n d p rescrib ed extern al forces are listed below.
t See Sec. 7 -6 for a derivation of the classical stability criterion. + See Probs. 10-11, 10-18.
SEC. 1 0 -7 .
P R IN C IP LE OF V IR T U A L FORCES
257
1.
Equilibrium Equation in the Interior (<5< j)t + uti i So-ji + a)i A u ( ,) = 0

OXj
I = 1, 2, 3
2.
Stress-B oundary Force Equations on Znri&O/j + Uj.i&tfi + <$i Au/.j) = 0 Stress-Strain Relations
< 5 rr = D < 5 fi
J =
2. 3
(1 0 -8 7 )
Strain-D isplacem ent Relations 5eu = + Av Aw,- = 0 4- um%iAum_j + um ; A,.;) on Qd
10-7.
PRINCIPLE OF VIRTUAL FORCES; PRINCIPLE OF STATIONARY COMPLEMENTARY ENERGY
L et U j be th e actu al d isp lacem en ts in a b o d y d u e to som e lo a d in g a n d ei} the geom etrically lin ear stra in m easu res co rre sp o n d in g to wf. T h e stra in an d d isp lacem en t m easu res are rela ted by e.7 = i K j + uj.i) Also, Uj = u on (b) (a)
O nce the strain s are know n, we can find the disp lacem en ts by solving (a) an d enforcing (b). T h e p rin cip le o f v irtu a l forces is basically a p ro c e d u re for d e te r m in in g disp lacem en ts w ith o u t h av in g to o p e ra te o n (a). It applies only for linear geom etry. W e developed its form for an ideal tru ss in Sec. 7 -3 . W e will follow the sam e a p p ro a c h h ere to estab lish the th ree-d im en sio n al form. T he essential step involves selecting a statically perm issible force system , i.e., a force system w hich satisfies th e lin ear eq u ilib riu m eq u atio n s. F o r th e co n tin u o u s case, th e force system co n sists o f stresses, A a u ; surface forces, Apnh on an d reactions, Apni, o n Qd. S tatic perm issib ility req u ires AOji, j ^ A pni aj A(Tji Apni - anj A<jji
o n Qa on i l j
(1 0 -8 8 )
If we m u ltip ly eu by A a,,, in teg rate ov er th e v o lu m e using (1 0 -8 1 ), a n d n o te th e static relatio n s, we o b ta in th e follow ing id en tity , t jjj'e.j A oij d x 1 d x 2 d x 3 j j u; AJ>ni dQ, +
t See Prob. 10-26.
j U; A pni dQ.
(1 0 -8 9 )
258
C H A P . 10
w hich is referred to as th e p rin cip le o f v irtu al forces (o r stresses). T his resu lt is ap p licab le for a rb itra ry m a te ria l behavior. H ow ever, th e g eo m etry m u st be linear. S uppose th e tra n s la tio n a t a p o in t Q o n Q a in th e d irectio n defined by ig is desired (see Fig. 10-15). L et dQ be th e disp lacem en t. W e ap p ly a u n it force at Q in th e \q d irectio n a n d g en erate a statically perm issible stress field. (1) iq a t p o in t Q => A a \ f an d Ap\,f
T h e in teg ral on 2 reduces to (1 )dQ, a n d it follow s th a t dQ = A< j( tf d.\i d x 2 dx$ JJu , Ap^f dQ a, (1 0 -9 0 )
A second ap p lic a tio n is in th e force m eth o d , w here o n e reduces th e g o v ern ing e q u a tio n s (stress eq u ilib riu m a n d stress disp lacem en t) to a set o f eq u a tio n s
involving o n ly force u n know ns. W e s ta rt by expressing th e stress field in term s o f a p rescrib ed d istrib u tio n (a0) an d a co rrectiv e field ((Tc),
a,j = of; +
(1 0 -9 1 )
w here <r is a p a rtic u la r so lu tio n o f th e eq u ilib riu m e q u a tio n s w hich satisfies th e b o u n d a ry c o n d itio n s on j + bi - 0 II %nj&ji a n d Gij satisfies 4. j = 0 U -njO 'ji = 0 II an XnjVji on n . on Q d (1 0 -9 3 ) on (1 0 -9 2 )
Stress fields satisfying (1 0 -9 3 ) are called self-equilibrating stress fields. F o r th e ideal truss, a 0 co rresp o n d s to th e forces in th e p rim a ry s tru c tu re d u e to the p rescrib ed lo a d in g an d < rc rep resen ts th e c o n trib u tio n o f th e force red u n d an ts.
SEC. 1 0 -7 .
P R IN C IP LE OF V IR T U A L FORCES
259
T he governing e q u atio n s for th e force re d u n d a n ts w ere o b ta in ed by enforcing geom etric co m p atib ility , i.e., th e b a r elo n g atio n s a re co n strain ed by th e re q u ire m ent th a t the deform ed b a r lengths f i t in th e assem bled stru ctu re. G eo m etric co m p atib ility for a co n tin u u m requires th e strain s to lead to continuous displacem ents. O n e can estab lish the strain co m p atib ility e q u atio n s by o p eratin g on th e strain - d isp lacem en t relations. T his ap p ro a c h is d escribed in P ro b . 1 0-10. O n e can also o b ta in these e q u atio n s w ith th e principle of v irtu al forces by ta k in g a self-equilibrating force system . L ettin g A<xc, Apc d en o te the v irtu al stress system , (10 -89) reduces to JJJ'eij A o ^ d.Xj d x 2 d x 3 = (T , A pc ni d il a (1 0 -9 4 )
T he com p atib ility eq u atio n s are d eterm in ed by expressing a-,- in term s o f stress functions an d in teg ratin g th e left-h an d term by p arts. W e illu strate its ap p lica tion to the plane stress pro b lem . Exam ple 1 0 -3 -------------------------------------------------------------------------------------------If the stress components associated with the normal direction to a plane are zero, the stress state is called planar. We consider the case where cr13 = cr23 = a 33 = 0. The equilibrium equations and stress-boundary force relations reduce to
^ 11,1 + 21.
hi
a l2.1
a 22.2 +
b2 =
P i l l ~
Pn2 1
+ *2021
12 + C C n2(T22 ^ ^
The stress field, a[h must satisfy (a) with bs = h2 0 and also pnl p2 = 0 on Q. We can satisfy the equilibrium equations by expressing <7^ in terms of a function, ijj, as follows .f aU ~ 12 a 22 = r P. u
*12 = Oil = <J'.l2 < C )
The boundary forces corresponding to a\j are o

P U ^ 'I'.Z
d
Pn2=
as
us
(d)
where s is the arc length on the boundary (sense is from A, -+ X 2). Substituting for ac, pc in terms of i//, (10-94) expands to
j ( fit A i/y
22
e2
11
12
M f . \ i)d x \
dx2
(e)
I i Aipt 2 I I A ll/, 2 ~ 2 JSd V

0 011 S.
1 ) I
) dS 0
T h e r e i s n o l o s s i n g e n e r a l i t y b y t a k i n g A(/> =
T h e n , i n t e g r a t i n g (e ) b y p a r t s ,
JK ei, 22 + 2 , 1 1 -
> 12, 12) ^
rfx j d x 2 = 0
(f)
t See Prob. 10-14.
260
G O VER N IN G EQ U A TIO N S FOR A D E F O R M A B L E S O LID
CHAP. 10
and requiring (f) to be satisfied for arbitrary Aip results in the strain compatibility equation, l. 22 + E 2, 11 7li, 12 = 0 which is actually a continuity requirement W 1. 122 + U2. 211 (U1,212 + I(2, 112 ) = 0 We express (g) in terms of ij/ by substituting for the strains in terms of the stresses.f (h) (g)
T he principle o f virtual forces is also em ployed to generate ap p ro x im ate so lutions for th e stresses. It is convenient to shift over to m atrix n o ta tio n for this discussion, a n d we w rite (1 0 -9 4 ) as JjjV / A<rc d x t d x 2 d x 3 = | f u r Apr dQ n', W e express th e stress m atrix in term s of prescribed stress states a n d u n k n o w n param eters, ah
= = o + 4!<pt 4- c/2 9 2 + , 4- a ri]),. (10~ 95)
w here a0 satisfies (10 -9 2 ) and < f) (/' = 1, 2 , . . . , r) are self-equilibrating stress states, i.e., they satisfy the h o m o g en o u s eq u ilibrium eq u atio n s an d b o u n d ary co n d itio n s on Q a. T he co rresp o n d in g surface forces arc
4-
-1-
+ - + iifi,
P = P 0, = 6
1 (i= ] . 2 , . . . , r ) j
n 0nQ -
(10 -9 6)
T aking virtual-force system s co rresp o n d in g to Aa, (t = 1, 2 , . . . , r) results in r eq u atio n s for the p aram eters. C|j,7(|)I - d x i d x 2 d x 3 Jj1 u76, dQ. ' a, / ~ 1,2,...,/(10-97)
In o rd er to proceed, we need to in tro d u ce th e m aterial p roperties. W hen the m aterial is linearly elastic, e = 4 A a = s 4 A<7 + cijAfyj and th e eq u atio n s expand to fjOj = di fj = U = i, j = 1 , 2 , . . . . r dxi d x 2 d x i (s + AfTWx, d x 2 d x 3 (10-98) (a)
dt j j uT0, dQ l'J 0>
O ne sh o u ld n o te th a t (1 0 -9 7 ) a.re weighted co m p atib ility co n ditions. The tru e stresses m u st satisfy b o th equilibrium and co m p atib ility th ro u g h o u t the
f See Prob. 10-27.
SEC. 1 0 -7 .
PR IN C IPLE OF V IR TU A L FORCES
261
dom ain. W e call gc the corrective stress field since it is required to co rrect the com patibility e rro r due to <r. F o r com pleteness, we describe here how one establishes a v ariational principle for < 7 C ij. O u r startin g p o in t is (10 -9 4 ) restricted to elastic behavior. W e define V* = y * according to SV* = A(t( 7 - r A<t (10-99)
an d call V * the complementary energy density. The form of V* for a linearly elastic m aterial is V* = 0 t b + (10-100) By definition, V* com plem ents V, i.e., V + V* = a i}el} Then, letting V f = JfJ V* d x t d x 2 d x 3 we can w rite (1 0 -9 4 ) as 5nc = 0 for a rb itrary Acf.(10-102) (10-101)
nc= n
[j % P m dn =
a.
nc (ay
(10' I03)
T his form is called the principle of statio n ary com plem entary energy and show s th a t the true stresses co rresp o n d to a statio n ary value o f ric. Since pni is linear in the second v ariatio n o f n f reduces to 5 2U C - 52V f = fjf Setj A ac u d x , dx2 dx, We shift over to m atrix n o ta tio n an d express < 5 e as = A ,A a (10-105) (10-104)
w here A, represents the tan g en t com pliance m atrix. N ow , A, m u st be positive definite in o rder for the m aterial to be stab le.f Then, (52n t > 0 for arb itrary Acrc and we see th a t the so lu tio n actually co rresp o n d s to a relative minimum value o f Tlc. T he app ro x im ate m ethod described earlier can be applied to fTt.. S ubstituting for <r given by (10-95) converts n c to a function of the stress param eters (fl1; a 2, , flr). W hen the m aterial is linearly elastic, n c = i a rfa - a rd + const (10-106)
The eq u atio n s for the stress p aram eters follow by requiring IT to be statio n ary for arb itrary A a,: <5nc = A aT(fa - d) = 0 (i fa = d 62V (10-107)
f The classical stability criterion specialized for elastic materia] and linear geometry requires r- (jf;7T), Se > 0 for arbitrary < S which, in turn, requires D, to be positive definite. Since A, - D f 1, it follows that A, must be positive definite for a stable material.
2 62
GOVERNING EQUATIONS FOR A DEFORMABLE SOLID
CHAP. 10
O perating on 5Xlc,
S2n c
=
A a Tf Aa
(10-108)
a n d n o tin g th a t < 52n c > 0, we co n clu d e th a t f is positive definite.
REFERENCES
1. C r a n d a l l , S. J., a n d N . C. D a h l : A n I n tr o d u c tio n to th e M e c h a n ic s o f S o lid s ,
2. 3.
4. 5.
6. 7.
8.
9. 10.
McGraw-Hill, New York, 1959. B is p l in g h o f f , R. L., M a r ., J. W.. and T. H. H. Pian: S ta t ic s o f D e f o r m a b le S o lid s , Addison-W esley, Reading, M a s s ., 1965. W a n g , C . T .: A p p l i e d E la s t ic i ty , M cGraw-Hill, New York, 1953. T im o s h e n k o , S. J ., and J. N . G o o d ie r : T h e o r y o f E la s t ic i ty , 3d ed., McGraw-Hill, New York, 1970. S o k o l n ik o f f , I. S.: M a th e m a tic a l T h e o r y o f E la s t ic i ty , 2d ed., M cGraw-Hill, New York, 1956. . F u n g , Y. C . : F o u n d a tio n s o f S o l i d M e c h a n ic s , Prenfice-Hall, 1965. L e k h n it s k ii , S. G .: T h e o r y o f E la s t ic i ty o f a n A n is o tr o p ic E la s tic B o d y , H olden-Day, San Francisco, 1963. W a s h iz u , K . ; V a r ia tio n a l M e t h o d s in E la s t ic i ty a n d P l a s t i c i t y , Pergamon Press, 1968. H il d e b r a n d , F. B.: M e t h o d s o f A p p li e d M a t h e m a t ic s , Prenlicc-Hall, 1965. C r a n d a l l , S. J .: E n g in e e r in g A n a ly s is , M cGraw-Hill. N ew York, 1956.
PROBLEMS
1 0 -1 . W rite o u t th e ex p an d ed form o f th e follow ing p ro d u cts. C o n sid er the rep eated indices to ran g e from 1 to 2. (a) aUkXjXk (b) itTij(uit j + uj'j) w here a;j = a# (c ) (< 5m j 4- Um,j){dmk + U lt k) ~ < \jk 1 0 -2 . L et / be a c o n tin u o u s fu n ctio n o f x 2, E stab lish th e tra n s fo rm atio n law s fo r df/dxj and d2fjdxj c x k. 1 0 -3 . E stab lish th e tra n sfo rm a tio n law for a{jbk w here ciih bk are cartesian tensors. 1 0 -4 . P ro v e th a t
& j k ~~ :l(P , j ' P , k
* 5 jk)
is a seco n d -o rd er cartesian ten so r.
H i n t : E x p an d dp dp
1 0 -5 . E q u a tio n s (1 0 -1 9 ) a re the stra in tra n sfo rm a tio n laws. Since eu is a sym m etrical seco n d -o rd er cartesian ten so r, th ere exists a p a rtic u la r set of directions, say fo r w hich efj is a d ia g o n al array . W h a t are th e stra in co m p o n e n ts for th e X p fram e? C o n sid er a re c ta n g u la r p arallelep ip ed hav in g sides d X J in th e u n d efo rm ed state. W h a t is its d efo rm ed sh ap e a n d relative change in volum e, w ith respect to its initial v o lu m e? Specialize th e ex p ressio n for s% for small strain. T h en d eterm in e ev for th e in itial (X^) d irectio n s a n d small strain. F inally, show th a t ev is in v arian t.
P R O B LE M S
263
10-6.
(a) (b)
s'i, y'ij
Specialize (1 0 -1 9 ) for sm all stra in an d w rite o u t th e ex pressions for in term s o f e2, . . . 3. L et e = {&u e2, e3, > i2 , y23, y3 i}- W e can express th e stra in tra n s fo rm a tio n (sm all strain ) as E' = T sS
(c)
D ev elo p th e form o f T u sing th e results o f p a rt a. E v alu ate T E in term s o f cos 0 , sin 6 for th e ro ta tio n sh o w n below. C o m m en t on th e tra n s fo rm a tio n law for th e o u t-o f-p la n e sh ear strain s 732-
A2
Prob. 10-6
1 0 -7 . In th e E u lerian a p p ro a c h , the cartesian c o o rd in a te s (rjt) for the defo rm ed sta te are ta k e n to b e th e in d e p en d en t variables, i.e., Uj = Wjfak) A lm ansis stra in te n so r is defined as \dp\2 - (ds)2 = 2 Ejk drjj drjk D ete rm in e th e expression for Ejk in te rm s o f th e d isplacem ents. C o m p a re th e result w ith (10-21). 1 0 -8 . C o n sid er the case o f tw o -d im en sio n al d efo rm a tio n in th e X r X 2 p lan e (e3 y 13 = y23 = 0). L et e, eb, ec b e th e extensions in th e a, b, c d irec tions defined below an d let = (ea, e.b, ec}. W e can w rite ;V = B (a) (b) (c) (d) D eterm in e th e g en eral form o f B. D eterm in e B 1 for 6a 0, 9b = 45, 8C = 90. D eterm in e B _1 for 9a = 0, dh = 60, 0C = 120. E x ten d (a) to th e th ree -d im e n sio n al case. C o n sid er six d irectio n s having d irec tio n cosines a ^ , ccj2, a ;3 w ith resp ect to X x, X 2, X 3. C an we select th e six d irectio n s a rb itra rily ? x j = XjU]k)
264
G O VER N IN G EQ UATIO N S FOR A D E F O R M A B L E S O LID
CHAP. 10
Prob. 10-8 *2
10-9.
F o r sm all strain, the volum etric strain is

Sv =
fij +
-2 +
3 =
e l 1 +
e 22
^.13
R ath er th an w ork w ith e-tj, one can express it as the sum o f tw o tensors, p.. tj c 1 J
-J-
pW *lj
w here e\f is called the spherical strain tensor, (s> - ij

1
vjjCn
and e\ f is th e d ev iato r strain tensor. (a) W rite o u t the expanded form for e\f an d e\f. (b) D eterm ine the first in v arian t o f e$f, ejf and co m p are with the in v arian t of ei}. 10-10, This question concerns strain com patibility equations. (a) Show th a t c 2elnk d zek J } 2emf _ d 2en! cxm cx{ cx c x k 8xm d x k cx dx, w here & nk & kn 1 ( cu hi) cX Î ^ + 2 \cxk ox
(b)
an d k, I, in, 11 range from 1 to 3. This expression leads to six in d ep en d en t conditions, called geometric compatibility relations, on the strain m easures. Show th a t for tw o-dim ensional d efo rm atio n in the X x- X 2 plane (e3 e13 = e23 = 0; this called plane strain) there is only one co m p atibility equ atio n , an d it has the follow ing fo rm :
G j, 2 2 +
82 . 11
) l 2 . 12
Is th e follow ing strain state perm issible? gj = k(x f + x j ) e2 = k x 2 }' 12 = 2 k x l x 2 k = co n stan t
P R O B LE M S
265
1 0 -1 1 . E q u a tio n (1 0 -2 1 ) defines the strain m easures d u e to displacem ents, M i- T o analyze geom etrically n o n lin ear behavior, o n e can em ploy an incre m ental form ulation. L et A u, rep resen t the displacem ent in crem en t an d Aejk the increm en tal strain. W e w rite Aejk = 5ejk 4- W e jk w here bejk co n tain s linear term s (A,-) an d b 2ejk involves q u a d ra tic term s. T he rj-symbol denotes th e first-o rd er change in a functional an d is called th e v a ria tio n al o p e ra to r (see Ref. 8). W e refer to be as th e first v a ria tio n of e. D eterm in e the expressions for be, b 2e. 1 0 -1 2 . Let i be th e u nit vector defining the in itial o rien tatio n o f the differential line elem ent dr a t a p o in t. dr = dsi in = anjij
T he un it vector defining th e o rien tatio n in the deform ed state is v. dp = (1 + F.)dsv v =
D eterm ine the general expression for pni. T hen specialize it for sm all strain. 1 0 -1 3 . T he several p a rts o f this q u estio n concerns stress tran sfo rm atio n . (a) S tartin g with (10-41). w rite o u t the expressions for o'u, o\j in term s of ^ll* 22> >^13(b) Let or = {(Tu, <722. <733, $ \ 2 , ^ 23, <r31} stress m atrix. W e express the stress tran sfo rm a tio n as a m atrix pro d u ct.
(7' -
(c)
D evelop th e form o f T,, using th e results of p a rt a. E v alu ate T in term s o f cos 0, sin 0 for the axes show n.
(d)
P lan e stress refers to the case w here = < t2 3 C33 = 0. W e w ork w ith reduced stress an d strain m atrices,
,
G {(71 1, C J2 2 , @1 2 }
8 =
{ f i j , 8 2 , >] 2 }
a n d w rite the tran sfo rm a tio n s in the sam e form as the three-dim ensional case: o' - T,<r Z' = T,
266
G O VERNING EQ U A TIO N S FOR A D E F O R M A B L E S O LID
C H AP. 10
E v alu ate
from p a rt c above and T e from P ro b . 1 0 -6 . T jT e - I 3
Verify th a t
10- 14. T his qu estio n develops a p ro ced u re for g en eratin g self-equilibrating stress fields. (a) E x p an d th e linear eq u ilibrium eq u atio n s, (1 0 -4 9 ) an d (10-50). (b) Specialize the eq u ilibrium eq u atio n s for plane stress (< 7i3 o 23
C 33
0
).
(c)
Suppose we express the tw o-dim ensional stress co m p o n en ts in term s o f a function 1 // = x 2). as follows: Gii
^22
22
11
J. v 2 ^ 2
dx j
dx2
(j 12 ~ @ 21 ~
Â 12
The n o ta tio n for body an d surface forces is defined in th e following sketch.

Prob. 1 0-1 4
*2
---------------- ----------
Xi
Verify th a t this definition satisfies the equ ilib riu m eq u atio n s in the interior. Show th a t th e expressions for Pi an J p 2 in term s o f derivatives w ith respect to x l5 x 2, an d s are Pi = P2 = 1 0 -1 5 .
-
'^ 2
cs
ani L b x d x {
J
d
cs
<A >i -
a ,,2 J ,2 h 2 d x 2
T he m ean stress, crm > is defined as am ~ + ^ 33)
R ath er th a n w ork w ith Oij, we can express it as th e sum o f tw o tensors,

Oij = o \ f + o \ f
P R O B LE M S
267
w here a \f is called the spherical stress tensor,

ojf -
c r j f is th e dev iato r stress tensor. (a) W rite o u t the expanded form s for a-f and c r j f . (b) D eterm ine the first in v ariant o f a - f , a j f . 1 0 -1 6 . E stablish th e stress-equilibrium eq u atio n s for smail-jinite ro ta tio n and sm all strain. 1 0 -1 7 . S tartin g with (10-52), (1 0 -5 5 ) specialized for sm all strain, establish the increm ental equilibrium eq u atio n s in term s of A a k, Au, Ab*, and Ap*. G ro u p acco rd in g to linear an d q u a d ra tic term s. Specialize these eq u atio n s for the case w here the initial po sitio n is geom etrically linear, i.e., w here we can ap p ro x im ate fijt( with y.jk in the increm ental equations. 1 0-18. P ro v e (10-60). Hint:
and
deJk - i (p _ j bp. k + p . k Sp,j)
5p,j = Ailj
10--19. Verify th a t the in verted form o f (10 -7 1 ) is
< y = D(e - c)
w here ^11 = E l / C 3 D l2 (C 2 /'Cl )Dl i O 13 C J > t i D 2 2 = 2/C , + ( C 2 / C l )D 1 2 ^23 = ^'32^2/Ci + (C 2/ C , ) D u
O 33 = E t , + Vm D xs 4 - V32D 23
and C i I ~ vUEz/E*) Cz v2l + V3lV32(E'2/E3,) c C4 = v3] -4 Specialize for plane strain (3 = >'13 = y2i = 0) 1 0 -2 0 . C o n sid er 2 sets of o rth o g o n al d irections defined by the u n it vectors tj and i'j. The stress-strain relatio n s for the two fram es are e = g + A c t .
e'
= (8y + A V
Express A' in term s of A an d T. Also determ ine D '. . 1 0 -2 1 . C o n sid er th e three-dim ensional stress-strain relatio n s defined by (10-71). (a) Specialize for p lan e stress (a 33 = c l3 a 2 3 = 0).
268
GOVERNING EQ UATIO N S FOR A D E F O R M A B LE SO LID
CHAP. 10
(b)
Let < y
{ci i, O22, C12}

{ C l, B2 , >'1 2 }
= 0 4-Arc Verify th a t D has the follow ing form : v21 0 D = El 1 - HV21 1 V 2J 0 _ w here n = n 0 0 G , y r (1 (1 - m i 1) 1
(C)
E2 A ssum ing X ^ X 2 in the sketch are m aterial sym m etry directions, determ ine D ' for th e X'r X'2 frame. U se th e results o f P rob. 10-13, 10-20. W h at relations betw een ihe pro p erties are required in order for D ' to be identical to D ? Prob. 10-21
X 3 directions. Then in tro d u ce a ro tatio n ab o u t the X t axis an d consider the expression for > 23. Iso tro p y in the X 2- X 3 p lane requires 1
y 23
f723
1 0 -2 3 . Verify th a t the d irections of principal stress an d strain coincide for an isotropic m aterial. Is this also true for an o rth o tro p ic m aterial? 1 0 - 2 4 . E q u atio n s (10 -7 6 ) can be w ritten as
a ij
u5ij
Aev
< 5^ + 2Ge,,
w here e is th e volum etric strain. U sing the n o ta tio n in tro d u ced in P ro b s. 10 -9 an d 1 0 -1 5 (a) Show th a t
<7m K c r + a
PR O B LE M S
269
(b) (c)
w here K is the b u lk m o d u lu s = (/3(1 - 2v)). D iscuss th e case w here v = b Show th a t a i f = 2 Ge\f Verify th a t th e strain-energy density can be w ritten as
2Oij{eij
<
5 ijeft)
= 4 * ^ ~ ?) + - w ? i/(J) 4 . j/(rf) D eterm in e F (s) an d K< d) for th e isotropic case. W hen v = -j, e, = e. W e m ust w ork w ith 7 stress m easures (er^-, erm) an d th e m ean stress has to be determ ined from an equilibrium co n sideration. S um m arize the governing eq u atio n s for the incom pressible case. 1 0 -2 5 . P ro v e (1 0 -8 1 ) for the tw o-dim ensional case. Is this fo rm u la re stricted to a specific d irectio n of in teg ratio n on the b o u n d a ry ? D oes it apply for a m ulti-connected region, such as show n in th e figure below ? (d) Prob. 10-25
1 0 -2 6 . 1 0 -2 7 . m aterial to 1 0 -2 8 .
Verify E q u a tio n (10-89). Refer to E xam ple 1 0 -3 . Express (g) in term s o f \j/. C o n sid er the be o rth o tro p ic. Verify th a t th e statio n ary req u irem en t <mR = 0 for a rb itra ry A;ih A o f h A pni , n K=
-
w here V* Tifui) d x l d x 2 d x 3
If p n iU id Q |'J p mU<i Ttj)dCl
h n, <jfj = KirchhofT stress
eu = L ag ran g e strain = i(u ,. } + uh , -I- ut ^
j)
V* = co m p lem en tary energy density (initial volum e) b*, p* p rescrib ed force m easures (initial dim ensions) leads to the com plete set of, governing eq u atio n s for an elastic solid, i.e., 1. stress equ ilib riu m eq u atio n s 2. stress-displacem ent relatio n s 3. stress b o u n d ary co n d itio n s on Q 4. displacem ent b o u n d a ry co n d itio n s on Qd 5. expressions for the reactio n surface forces on
270
GO VER N IN G EQ U A TIO N S FOR A D E F O R M A B L E S O LID
CHAP. 10
T his v ariatio n al statem en t is called R eissners principle (see Ref. 8). (a) T ran sfo rm to I I P by req u irin g the stresses to satisfy th e stress displacem ent relations. H i n t : N o te (10-101). (b) T ran sfo rm n R to n c by restricting th e geom etry to be lin ear (ak = c an d eu (uiyj + uL f)/2) an d req u irin g the stresses to satisfy th e stress equilibrium eq u atio n s an d stress b o u n d ary co n d itio n s o n Hint: In teg ra te cr;7 -e,;- by parts, using (10-81). 10- 29. In te rp re t (10-90) as
where P Q is a force applied at Q in the direction of the displacem ent measure, dQ.
11

11-1. INTRODUCTION AND NOTATION
A body w hose cross-sectional dim ensions are sm all in co m p ariso n w ith its axial dim ension is called a member. If the cen tro id al axis is straig h t a n d the shape an d o rie n ta tio n of the n o rm al cross section are c o n s ta n t,! the m em ber is said to be prismatic. W e define the m em ber geom etry w ith respcct to a global reference fram e ( X U X 2, X 3), as show n in Fig. 1 1 -1 . T he X axis is taken to coincide w ith the cen tro id al axis and X 2, X 3 are tak en as the p rin cip al in ertia directions. W e em ploy the follow ing n o ta tio n for th e cross-sectional p ro p e rtie s ; A = [j dx 2 dx 3 = j J dA 11 = j f e ) 2 dA h = j f e ) 2 dA Since X 2, X 3 pass th ro u g h the cen tro id an d are p rin cip al in ertia directions, th e cen tro id al co o rd in a tes and p ro d u c t of in ertia vanish: (H-l)
( 11 -2 )
I n = j> 2 * 3 dA = 0 O n e can w o rk w ith an a rb itra ry o rien tatio n of the reference axes, b u t this will com plicate th e derivation. St. V en an ts th eo ry of torsion-flexure is restricted to linear behavior. It is an exact linear fo rm u latio n for a p rism atic m em ber subjected to a prescribed
t T h e c a se w h er e th e c r o s s-se c tio n a l sh a p e is c o n s ta n t bu t the o r ie n ta tio n v aries a lo n g th e c e n tr o id a l a x is is treated in C h a p te r 15.
271
270
G O VERNING EQ UATIO N S FOR A D E F O R M A B LE SO LID
C H AP. 10
T his v ariatio n al statem en t is called R eissners principle (sec Ref. 8). (a) T ran sfo rm I I R to n p by req u irin g the stresses to satisfy th e stress displacem ent relations. Hint: N o te (10-101). (b) T ransform n R to n c by restricting the geom etry to be linear {< rk a an d eu = (wf,j 4- Ujti)/2) and req u irin g the stresses to satisfy the stress equilibrium equations an d stress b o u n d ary con d itio n s on Q?. Hint: In teg rate a ^ j by parts, using (10-81). 1 0 -2 9 . In te rp re t (10-90) as
where PQ is a force applied at Q in the direction of the displacement measure, dQ.
11

11-1. INTRODUCTION AND NOTATION
A body w hose cross-sectional dim ensions are small in com parison with its axial dim ension is called a member. If the cen tro id al axis is straig h t and the shape and o rien tatio n of the norm al cross section arc c o n sta n t,! th e m em ber is said to be prismatic. W e define the m em ber geom etry with respect to a global reference fram e ( X x, X 2, A'3), as show n in Fig. I l - l . T he X t axis is taken to coincide w ith the centroidal axis and X z, X 3 are taken as th e principal inertia directions. W e em ploy the following n o ta tio n for the cross-sectional p ro p e rtie s: A dx2 dx3 = dA h = H (x3) dA
h = jj(x2)2 dA ( 11- 1)
Since X 2, X 3 pass th ro u g h the centroid and are principal inertia directions, the centroidal co o rd in ates an d p ro d u ct of inertia vanish: x 2 dA = 0
1 23 *3
c~
Y
*3
A JJ
dA = 0
dA = 0
O ne can w ork w ith an arb itrary orien tatio n of the reference axes, b u t this will com plicate the derivation. St. V enants th eo ry of to rsio n :flexure is restricted to linear behavior. It is an exact, linear form ulation for a prism atic m em ber subjected to a prescribed
t T h e case w h ere th e c r o s s-se c tio n a l sh ap e is c o n sta n t b u t th e o r ie n ta tio n varies a lo n g th e c en tro id a l a x is is treated in C h ap ter 15.
271
272
TO R SIO N -FLE XU R E O F P R IS M A T IC M E M B E R S
CHAP. 11
d istrib u tio n of surface forces applied on th e end cross sections. Later, in C h ap ter 13, we m odify the St. V enant theory to acco u n t for displacem ent restrain t a t the ends an d for geom etric noniinearitv. *2
F ig. 1 1 -1 . N o ta tio n fo r p ris m a tic m e m b e r.
T he d istrib u tio n o f surface forces on a cross section is specified in term s of its statically equivalent force system at the centroid. F ig u re 1 1 -1 show s the stress co m p o n en ts on a positive face. W e define F+, M+ as the force an d m om ent vectors acting at the centroid w hich are statically equivalent to the d istrib u tio n o f stresses over the section. T he co m p o n en ts o f /*+. M+ are called stress resultants an d stress couples, respectively, an d th eir definition eq u atio n s are F i = J K i dA F 2 = . O l 2 dA F 3 - JJo-13 dA M x = JJ(x2ff13 - x 3ari2)dA M 2 = j W n clA M 3 = - i j W u dA T he intern al force and m o m en t vectors acting on the negative face are d en o ted by F _ , M _ . Since F_ ~F+ M - = M + (1 1 -4 )
it follows th a t th e positive sense of th e stress resu ltan ts an d couples for the negative face is o p posite to th a t show n in Fig. 1 1 -1 . W e discuss next the p u re-to rsio n case, i.e., w here the end forces are statically equivalent to only M v. W e then extend th e fo rm u latio n to acco u n t for flexure
SEC. 1 1 -2 .
THE P U R E-TO R SIO N PR O B LE M
273
and tre a t torsional-flexural coupling. Finally, we describe an ap p ro x im ate procedure for determ ining th e flexural sh ear stress d istrib u tio n in thin-w alled sections.
11-2.
THE PURE-TORSION PROBLEM
C onsider the prism atic m em b er show n in Fig. 11 -2 . T here are no b o u n d ary forces acting on the cylindrical surface. T he b o u n d ary forces acting on the end cross sections are statically eq u iv alen t to ju s t a tw isting m o m en t M l . Also, there is no restrain t w ith respect to axial (out-of-plane) displacem ent at th e ends. The analysis o f this m em b er presents the pure-torsion p roblem . In w hat follows, we establish the governing eq u atio n s for p u re to rsio n , using the approach originally suggested by St. V enant. *0
i/i
Fig. 11-2. Prismatic member in pure torsion. R ath er th a n a tte m p t to solve th e th ree-d im en sio n al p ro b lem directly, we im pose the follow ing co n d itio n s on th e b eh av io r an d then d eterm in e w hat problem these co n d itio n s co rresp o n d to. 1. 2. E ach cross section is rigid w ith respect to d efo rm atio n in its plane, i.e., e2 = e3 = >'23 = 0 E ach cross section experiences a ro ta tio n a> y a b o u t th e Xy a x is t and an out-of-p lan e d isp lacem en t iiy.
These co n d itio n s lead to th e follow ing expansions for the in-place d isplace m ents: ll2 C l)1X3 (1 1 -5 )
1/3 =
+CQy X 2
The co rresp o n d in g lin ear strain s are

&2 .
3 =
J 23
1 1 1. 1
( 1 1 - 6)
= 1,3 + -X2 l , l
}'l2 = W l. 2 + U2. I = Ul. 2 ~ -*3W1, 1

Vl3 = Ul,3 + 3 ,1
t Problem 11 1 treats the general case where the cross section rotates about an arbitrary point.
274
T O R S IO N -F LE X U R E OF P R IS M A T IC M E M B E R S
C H A P . 11
N o w , th e strain s m u st be in d e p en d en t o f x l since each cross sectio n is sub jected to th e sam e m o m en t. T his req u ires w i,i = ki W| = U ,(x2,-X3) (1 1 7)
W e co n sid er th e left en d to be fixed w ith respect to r o ta tio n an d express coj, as

=
k l X } '
( 11- 8)
i = w here (j)t = (j)t{ x2, x 3) defines th e o u t-o f-p la n e d isp lacem en t (w arping) o f a cross section. T h e strain s a n d stresses c o rre sp o n d in g to th is p o s tu la te d disp lacem ent b eh a v io r are l 2 = 3 = ^23 y t 2 = kA4>t,2 }'l3 = k l(<Pt,3 + X 2) and
"ll = 2 2 ^33 = cr23 ~ 0
' 0 (n-9)
crl 2
= G y l2 = G kxi fa ' 2 -
x 3)
(T1 2 ( x 2 , x 3 )
(1 1 -1 0 )
cr13 = Gc713 = G l c ^ f 3 + x 2) = a 13(x2, x 3) W e are assu m in g th a t th e m a terial is is o tr o p ic t an d th e re a re n o in itial strains. O n e step rem ain s, nam ely, to satisfy th e stress-eq u ilib riu m e q u a tio n s an d stress b o u n d a ry c o n d itio n s on the cy lin d rical surface. T h e co m p lete system of lin ear stress-eq u ilib riu m eq u atio n s, (1 0 -4 9 ), rcduces to 2 \ , 2 + 31,3 0 (1 1 -1 1 )
S u b stitu tin g for th e shearing stresses an d n o tin g th a t G /q is c o n s ta n t lead to th e differential e q u atio n / p2 \ h -2 + J ' ! W>r = V 2r/;r = 0
J x l d x l
(1 1 -1 2 )
w hich m u st be satisfied a t all p o in ts in th e cro ss section. T h e ex terio r n o rm a l n for th e cylindrical surface is perpendicular to th e X 1 directio n . T h e n a nl = 0, a n d th e stress b o u n d a ry co n d itio n s, (1 0 -4 9 ), red u ce to
Pni =
a2^2i + an3ff3i = 0
(1 1 -1 3 )
U sing (1 1 -1 0 ), th e b o u n d a ry c o n d itio n for < j> t is

<X2 ( 4>t , 2 ^ 3) + 3 + * 2) = 0
V d* ' / 2x C. 3 - < = y .n3x 2
(11-14)
(on S)
t P ro b lem
1 1 - 3 treats th e o r th o tro p ic case.
SEC. 1 1 -2 .
TH E P U R E -T O R S IO N P R O B LE M
275
T h e p u re -to rs io n p ro b le m involves solving V 2<, 0 subject to (11-14). O n ce < p, is know n, we d eterm in e th e d istrib u tio n o f tran sv erse sh earin g stresses from (11-10). N o te th a t (j> t d ep en d s on ly o n th e sh ap e of th e cross section. T he shearin g stress d istrib u tio n m u st lead to no sh earin g stress resu ltan ts:
^2
T his requires <50, dx7 dA
j y^ 1 2
o
dA = 0
S'3 = jj'tf 13 d A = 0
(a)
(b) W e sta rt w ith (1 1 -1 5 )
T o proceed fu rth er, we need certain in teg ratio n form ulas. dA - A fctnj dS
J J
CXi
w hich is ju s t a special case of (1 0 -8 1 ). A pplying (1 1 -1 5 ) to jf V 2tp dA leads to G re e n s theorem , c\j/ V 7 dA = O a n2 - + a,l3 dS T ' O x2 t X 3J dn If i// is a h a rm o n ic fu n ctio n (i.e., V 2ij/ = 0), G re e n s th e o re m requires O JL dS = 0 on
(1 1 -1 6 )
(c)
N ow , < j> t is a h a rm o n ic fu nction. F o r the fo rm u latio n to be co n sisten t, (1 1 -1 4 ) m u st satisfy (c). U sin g (U -1 5 ), (c) tran sfo rm s to
>(an2x 3 - ccnix 2)dS =
.d v ,
x, X i) d /4 - 0 ' ox.
(d)
Since 3(jytjdn is specified o n th e b o u n d ary , we c a n n o t ap p ly (11 -1 5 ) d irectly to (b). In this case, we use th e fact th a t 0 a n d w rite d(j)t
dxj dx: : * ' ) * *
X;
(j = 2, 3)
(e)
In te g ra tin g (e), d A = (t> .v. dS J on U = 2, 3)
(f)
a n d th e n su b stitu tin g for th e n o rm al derivative, verifies (b). T h e c o n s ta n t k x is d eterm in e d from th e rem a in in g b o u n d a ry co n d itio n , M } JJ(x2o "j 3 x i (j12)dA (1 1 -1 7 )
276
TORSION-FLEXURE OF PRISMATIC MEMBERS
CHAP. 11
W e s u b stitu te for th e sh earin g stresses a n d w rite th e re su lt as My = G k y J w here J is a cro ss-sectio n al p ro p erty , = (1 1 -1 8 )
JJ V
(T^
xi + x, + ,
(1 1 -1 9 ) 3
A t this p o in t, w e su m m arize th e results for th e p u re -to rsio n p roblem . 1. Displacements ul = k ^ , u2

A' I-'-3 a>ix2
k lXl GJ 2. Stresses a t2 M j / cV />j J \cX 2

+ .X,
. ( . 'A ,
(1 1 -2 0 )
M i (c4 (T,j = Governing Equations in A: on S:
= *2*3 ~ a 3x 2
It is po ssib le to o b ta in the exact so lu tio n for (j)t for sim ple cro ss sections. T he p ro ced u re ou tlin ed ab o v e is basically a displacement m eth o d . O n e can also use a force ap p ro ac h for th is p ro b lem . W e sta rt by expressing th e sh earin g stresses in term s o f a stress function i//, so th a t th e stress-eq u ilib riu m eq u a tio n (E q u a tio n 1 1 -1 1 ) is id en tically satisfied. A n a p p ro p ria te d efin itio n is "12
dx3 dlP CXn
( 11 -21 )
T he sh earin g stresses for th e X, v d irectio n s, sh o w n in Fig. 1 1 -3 , follow d irectly from th e d efinition e q u a tio n cij/
SEC. 1 1 -2 .
THE PU R E-TO R SION P R O B LE M
277
T aking S 90 co u n terclo ck w ise from th e ex terio r n o rm a l d irectio n , an d n o tin g th a t th e stress b o u n d a ry c o n d itio n is crln = 0, lead to th e b o u n d a ry c o n d itio n for ij/, \b co n st on S (1 1 -2 3 ) W e establish th e differential eq u a tio n for ijj by req u irin g th e w arp in g function < [ > t to be co n tin u o u s. F irst, we e q u a te th e expressions for a in te rm s o f i{/ a n d (j)t : M,
\% ~ V , 3 J ~ W t . 2 x 3)
(a)
A*i ^13 = -<A.2 = ~ j
, ,
+ * 2)
N ow , for co n tin u ity ,

f i t , 23 </>(, 3 2
(b)
O p e ra tin g o n (a), we o b ta in V > It is co nvenien t to express )// as / T h e g o v ern in g e q u a tio n s in term s o f i// a r e t cr12
^13
-2
Mi ~T T
(c)
(1 1 -2 4 )
M[ # J dx3 Mi # J ( in /I ) (on b o u n d a ry S ;) (1 1 -2 6 ) cx7 (U -2 5 )
an d
S u b stitu tin g (1 1 -2 5 ) in th e d efinition eq u a tio n for M x leads to th e follow ing expression fo r J : J = x
+ x O X2 C .V 3y
(a)
A pplying (1 0 -8 1 ) to (a) an d noting']: th a t < |> s Xjdnj dS Ai = a re a enclosed by th e in te rio r b o u n d a ry curve, M s, = Ci = const
t E q u atio n s (11-26) can be in te rp re te d as th e g o v e r n in g e q u a tio n s fo r an in itia lly s tre tch e d
(b)
m e m b r a n e su b je c te d to n o r m a l pressure. S e e R ef. 3.
T h i s i n t e r p r e t a t i o n is c a l l e d t h e m e m b r a n e a n a l o g y .
% T h e S' d i r e c t i o n i s a l w a y s t a k e n s u c h t h a t n S h a s t h e s a m e s e n s e a s X 2 ~ + S d ir e c tio n fo r a n in t e r io r b o u n d a r y is o p p o s it e to th e + S
X 3.
T h e n , the
d ir e c tio n fo r a n e x te rio r b o u n d a r y
s i n c e t h e d i r e c t i o n f o r n is r e v e r s e d . T h i s is t h e r e a s o n f o r t h e n e g a t i v e s i g n o n t h e b o u n d a r y i n t e g r a l .
278
CHAP. 11
we can w rite J = -- 2 jj<// d A + (11 -27) w here ^ = 0 o n the exterior b o u n d a ry . T o d eterm in e th e co n stan ts Q fo r th e m u ltip ly co n n ec ted case, we use the fact th a t (f)t is co n tin u o u s. T his req u ires * * dS = 0 h dS for an a rb itra ry closed curve in th e cross section. *3 (1 1 -28)
*2
Fig. 11-3. Definition of n-s and /-v directions.
^3
Fig. 1 1-4. G raphical representation of sector area.
SEC. 1 1 -2 .
TH E PU R E -T O R S IO N P R O B L E M
279
C o n sid er th e closed cu rv e show n in Fig. 11 -4 . T h e sh earin g stra in yis is given by

JlS ~ a S 2 7 l2 + a S 3 > 'l3 (a )
U sing (1 1 -9 ), w e can w rite (a) as

J 1S = k li*S2<t>,,2 +
3 -
X3&S2
* 2^ 3)
( f
+ p)
(1 1 -2 9 )
w here p is th e p ro je c tio n o f the rad iu s v ecto r o n th e outward n o rm a l.f T h e m a g n itu d e o f p is eq u al to th e p e rp e n d ic u la r d ista n ce from th e origin to th e tan g en t. In te g ra tin g betw een p o in ts P, 0 , we o b ta in JO p w here ApQ \ I P dS = secto r a re a enclosed by th e arc PQ a n d th e rad iu s v ecto rs to P a n d Q.
7 is
dS
4 >t , p
2 /4 pq)
(11 30)
F inally, ta k in g P Q,% yl s dS = 2 k 1A s (11 31) w here /4S d en o tes th e a re a enclosed by the curve. Since a X j = G y ^, we can w rite M, a is dS = 2 G k l A s = 2 A S (1 1 -3 2 )
M
N o te th a t th e + d irec tio n for (11 32) is from X 2 to w ard X 3. Also, this resu lt is independent o f th e lo c a tio n of th e origin. In stead o f using (1 1 -9 ), we co u ld have sta rte d w ith th e fact th a t th e cross section ro ta te s a b o u t th e cen tro id . T h e d isp lacem en t in th e + S d ire c tio n fol low s from Fig. l l - 4 : us ct>1(Ta x f ' i s) = S u b stitu tin g for us in Vis = s. i + i , s and n o tin g th a t u l = ki<j)t lead to (11-29). U sing (1 1 -2 2 ), we can w rite # an M i dtp j--J on ni (1 1 -3 5 ) (1 1 -3 4 )
0
)^
k 1x 1p
(1 1 -3 3 )
t T h i s i n t e r p r e t a t i o n o f p is v a l i d o n l y w h e n S i s d i r e c t e d f r o m fo r th is ca se. \ S ee P ro b . 1 1 - 1 4 fo r an a lte rn a te d e riv a tio n . 5 T h is d e v e lo p m e n t a p p lie s fo r a r b itr a ry ch o ic e o f th e if a r o t a t i o n a b o u t A 'j p r o d u c e s a t r a n s l a t i o n in t h e + S
A ' 2 t o X 3, i . e . , c o u n t e r c l o c k w i s e
d irectio n .
T h e s ig n o f p is p o s it iv e
+ S d irectio n .
E q u a t i o n ( 1 1 - 2 9 ) is u s e d to See P r o b . 11 - 4 .
d e t e r m i n e t h e w a r p i n g d i s t r i b u t i o n o n c e t h e s h e a r i n g s t r e s s d i s t r i b u t i o n is k n o w n .
280
TORSION-FLEXURE OF PRISM ATIC M EM BERS
CHAP. 11
Then, substituting for a is in (11 - 32), we obtain (}) - t (is = 2AS J s cn (U -3 6 )
where n is th e outward norm al, A s is the area enclosed by S, and the + S sense is from X 2 to X 3. This result is valid for an arb itrary closed curve in the cross section. We employ (11-36) to determ ine the values of > ]/ at the interior b o u n d aries of a multiply connected cross section. It is of interest to determ ine the energy functions associated with pure to r sion. W hen the m aterial is linearly elastic and there are no initial strains, the strain and com plem entary energy densities are equal, i.e., V V* We let V = V dA = strain energy per unit length (11-37) The strain energy density is given by
v = y (7x2 + y h )
Substituting for y l2, y l3. y = ~ y ' [Wv ~ x 3)2 + W >i, 3 + x 2)2]
(a)
(b)
and integrating (b) over the cross section, we obtain V = {G Jk{ Since V* = V, and M L G J k l , it follows that (11-39) (11-38)
2G
W j
------------ X } afcoi i + dTx dx i
Fig. 11-5.
D iffe re n tia l e le m e n t fo r d e te r m in a tio n of th e ro tatio n a l w o rk .
Instead of integrating the strain-energy density, we could have determ ined the w ork done by the m om ents acting on a differential element. C onsider the element shown in Fig. 11-5. The boundary forces acting on a face are statically equivalent to ju st a torsional m om ent. Also, the cross sections are rigid in
SEC. 1 1 -3 .
T H IN -W ALLE D OPEN CROSS SECTIONS
281
their plane and ro tate ab o u t X
The relative ro tatio n of th e faces is (a)
coj + - dx, ( a). /<, d x x dx j J
and the first-order w ork done by the external forces due to an increm ent in a>l reduces to 5WE = $ p T Au dS = M i A/<j d x 1 (b) Now, 3W e = oVT = JJJ <51/ d x t d x 2 d x 3 = S V d x , (c) for an elastic body. Then, expanding <5P,
_ dV =
and it follows that
dV
AA'j M l A/ct 1
(d)
(11-40)
11-3.
APPROXIMATE SOLUTION OF THE TORSION PROBLEM FOR THIN-WALLED OPEN CROSS SECTIONS
We consider first the rectangular cross section show n in Fig. 11 6. The exact solution for this problem is contained in num erous texts (e.g., see Art. 5 -3 of Ref. 1) and therefore we will only sum m arize the results obtained.
X2
d/2
f----- 1/2-
-//2 j
Fig. 1 1 - 6 . N otatio n fo r re c ta n g u la r s e c tio n .
282
T O R SIO N -FLE XU R E OF PR IS M A T IC M E M B E R S
C H AP. 11
W hen t ^ the m axim um shearing stiess (points 5, 6). T h e exact expressions are dt3 ; - k , t
occurs a t x 2
i /2, -X 3
(11 -4 1 ) u where K! , 1 192/A _ 1 , , -(- 1 y ------ T ^ ta n h A * 71" V v = 0. 1. .. (2 + l ) 5 i - M-i k ,
K2 2n+l K
2
7T2 n ^ o . 1 ( 2 H +
l ) 2 COSh
/*T
711 7 ,
V alues o f K lf K z for d/t ranging from 1 to 10 are tab u lated below:
djt
1 2 3 4 5 10
Ei
0.422 .687 .789 .843 .873 0.936
El
0.675 .930 .985 .997 .999 1.000
If t d, we say the cross section is thin. T h e ap p ro x im ate so lu tio n for a thin rectangle is J v $ dt h
1 ^ 1 nk l x 2 a 13 ~ 2 -----x 2 2iG J
(11-42)
t ~ ^ 2^3 > A

Y 2 x 2
(We tak e djt = 00 in the exact solution.) T h e shearing stress a 13 varies linearly across the thickness and k | < - ^ r
A view o f th e w arped cross section is show n in Fig. 11-7. Since the stress function ap p ro ach is qu ite co nvenient for the analysis of thin -w alled cross sections, we illu strate its ap p licatio n to a thin rectan g u lar
SEC. 1 1 -3 .
T H IN -W A L LE D OPEN CROSS SE C TIO N S
283
cross section. L ater, we shall extend th e results o b ta in ed for this case to an a rb itra ry thin w alled o p en cross section. T he governing eq u atio n s for a sim ply connected cross section are sum m arized below for convenience (see (11-26), (1 1 -2 7 )): V2!A = - 2 (in A) tp 0 <J\n (on th e b o u n d ary ) Mi # J IS (a)
J = 2 P dA w here the S d irectio n is 90 counterclockw ise from the n d irec tio n .! Since t is sm all an d a l2, the sh earin g stress co m p o n en t in th e thickness d irectio n , m ust
X I. 11I Fig. 11-7.

W a rp in g fu n c tio n fo r a r e c ta n g u la r c ro s s s e c tio n .
vanish on the b o u n d a ry faces, it is reaso n ab le to assum e a l2 = 0 a t all p o in ts in the cross section. T his co rresp o n d s to tak in g if/ in d ep en d en t o f x }. T he eq u atio n s rcduce to dx* = o Solving (b), we o b tain f = -* t + j
J 2d \
a 13
-2
(b) at x 2 : t
_ i(/ d x 2 = 3d t ' J f/2 ~ J M t d$ M, = 2 - X2 J f [/2
(c)
t T h is a p p lie s fo r
co u n terclo ck w ise from
X 2- T h e g e n e r a l r e q u i r e m e n t is t h e n S s e n s e
m u st co in cid e w ith the X 1-X i sense.
284
TO R SIO N -FLE XU R E OF P R IS M A TIC M E M B E R S
CHAP. 11
T he expression for i(j developed above m ust be co rrected n ear the ends (x3 = d/2) since it does n o t satisfy the b o u n d ary condition,
_
\j/ 0
at x 3 = i
d
-
This will lead to < 7j 2 0 near th e ends, b u t will have a negligible effect on J an d a max. A ctually, th e m om ent due to the ap p ro x im ate linear expansion for 0 i3 is equal to only one h alf the applied m o m en t: (e) T he corrective stress system (<r13) carries M J 2. This is reasonable sincc, even th ough o-i2 is sm all in com parison to c;max, its m o m en t arm is large.
Fig. 11-8.
N o ta tio n for th in -w a lle d o p e n c ro s s se c tio n .
W e consider next the a rb itra ry thin-w alled open cross section show n in Fig. 11-8. T he S curve defines the centerline (bisects the thickness) and the n direction is n o rm al to S. We assum e a in = 0 an d take ijj ~ n 2 + t 2/ 4. This corresp o n d s to using the so lu tio n for th e thin rectangle an d is reaso n ab le when S is a smooth curve. T he resulting expressions fo r J and a , s are
( 1 1-4 3)
SEC. 1 1 -3 .
T H IN -W A L LE D OPEN CROSS S E C TIO N S
285
T he results for a single thin rectangle are also applied to a cross section consisting o f thin rectan g u lar elem ents. Let d{, f; d en o te th e length an d thickness o f elem ent i. W e ta k e J as I As an illu stratio n , consider the sym m etrical section show n in Fig. 1 1 -9 . A pply ing (1 1-4 4 ), we o b tain J = i ( 2 b , t } + d wt* w) T he m axim um sh earin g stress in th e center zone o f an elem ent is tak en as m.i = ~ l , - G k ^ (1 1 -4 5 )
In general, there is a stress co n cen tratio n at a re e n tra n t co rn er (e.g., p o in t A in Fig. 11 -9 ) w hich d epends o n the ra tio of fillet rad iu s to thickness. F o r th e case 1 1
> > r-
I '/ t
....... " " "
m
i
z tm
' fw
y /////////z /A
Fig. 11-9.
S y m m e tr ic a l w id e - fla n g e s e c tio n .
o f an angle having equal flange thicknesses, the fo rm u la t ^fillet (1 1 -4 6 )
1+ t 4 it ) 77
w here rf is the fillet rad iu s an d a m is given by (14-45), gives g o o d resu lts for rf jt < 0.3. T he stress increase can be significant for sm all values of rf jt. F o r exam ple, < T fll|e, = 3.5<rm for rf = 0.1?. N u m erical p ro ced u res such as finite differences o r the finite elem ent m ethod^ m u st be reso rted to in o rd er to o b tain exact so lu tio n s for irreg u lar sections. .
t See Ref. 2 and Appendix of Ref. 9. j See Ref. 4.
286
TORSION-FLEXURE OF PRISMATIC M EM BERS
CHAP. 11
1 1 -4 .
APPRO XIM ATE SOLUTION OF THE TORSION PRO BLEM FOR THIN-W ALLED CLOSED CROSS SECTIONS
The stress function m ethod is generally used to analyze thin-walled closed cross sections. F o r convenience, the governing equations are sum m arized below (see (11-26), (11-27), (11-36)): V2< ? = -2 $ = 0 ^ - Ct J = 2ffy dA + 2 C i/ l i a iS d\j/ s t> n We consider first the single cell shown in Fig. 11-10. T he Sc) curve defines the centerline. Since there is an interior boundary, we have lo add a term (in A) (on the exterior boundary) (on the interior boundary, St) (/I; = area enclosed by S<) ( is the outward norm al for S and f S sense from X 2 toward X 3)
Mt d \j/
-
Sn
dS = 2Ac
Fig. 11-10.
S in g le c lo s e d c e ll.
involving C x to the approxim ate expression for ij/ used for the open section. We take as i $ = $ + +2
n2 4
2V
fJ
w h e re
ijrc
re p re s e n ts th e c o n t r i b u t i o n o f th e i n te r io r b o u n d a r y . T h is ex p ressio n
SEC. 1 1 -4 .
T H IN -W A L L E D C L O S E D C R O S S S E C T IO N S
287
satisfies the one-dim ensional com patibility equation and boundary conditions, r,2 <
=
-2
o n 2
0 = 0 (// =
at n ~ + r/2 at n ~ t / 2
and is a reasonable approxim ation when S is a sm ooth curve. Differentiating \j/, . Ci c\jj
= 2 ? ? ----------
(b)
{
ds
and substituting (b) in the expressions for the shearing stress com ponents lead to Om 0
+
_ _0 i _c < X {S -h <Ji5
(11-48)
The tangential shearing stress varies linearly over the thickness and its average value is < 7^,. We let q be the shear stress resultant per unit length along S, positive when pointing in the + S direction,
4 = f1 J - 1/2
and call q the shear flow. Substituting for a is, we find
Q= y Ci = c \ st
(11-49)
(11-50)
The additional shearing stress due to the interior boundary (i.e., closed cell) corresponds to a constant shear flow around the cell. O ne can readily verify! that the distribution, q = const, is statically equivalent to only a torsional m om ent, M \ , given by M \ = 2 q A cl (11-51) The torsional constant is determ ined from J = dA + 2C X A, = M JG ki (a)
Substituting for 1 // using (11-47), we obtain

j
/ - j7 4 + - jJc
(11-52)
J = i f Sdr3dS
r = 2Cx At X
E quation (a) was established by substituting for the shearing stresses in term s of $ in the definition equation for A/j and then transform ing the integrand. We could have arrived at (11-52) by first expressing the total torsional m om ent as
M v = M\ + M\
(
11 53
-
t See Prob. 11-5.
288
C H A P . 11
w here M \ is the open section co n trib u tio n an d M \ is d u e to th e closure. Next, we w rite M , = GkiJ Then, J = J + J e an d it follow s th a t (1 1 -5 6) F inally, using (11-51), we can express J c as r = M \ / ( M \/J) = 2 A cl[ q / ( M t / J ) ] (1 1 -5 7 ) (1 1 -5 5) M l = G k xJ G k v Jc (11-54)
T his result show s th a t we should w ork w ith a m odified sh ear flow, C = q/iM JJ) (1 1 -5 8 )
ra th e r th a n w ith th e actu al sh ear flow. N o te th a t C = Ci for th e single cell. It rem ains to determ in e C, by enforcing c o n tin u ity o f th e w arp in g function on th e centerline curve. A pplying (1 1 -3 2 ) to 5^, we have
I
S u b stitu tin g for o'ls, leads t o t
c ,c 0 Mi Uj s dS = 2 /4cl
(1 1 -5 9 )
= cut = - J -
2A c l
(1 1 -6 0)
O n e sh o u ld n o te th a t C t is a p ro p erty o f th e cross section. O n ce C Y is know n, we can ev alu ate .7 from (11 -5 2 ) an d the sh earin g stress from
Mi ( C,\ ffts = ~ j - y t + J
(1 1 -6 1 )
Example 11-1 --------------------------------------------------------------------------------------Consider the rectangular section shown. The thickness is constant and a, h are centerlinc dimensions. The various cross-sectional properties are
A U Ac\ ~ ab
<) = . t
2{-a + ^
abt C, a + b Jc =
J = i t \ 2 ( a + b)] l a 2b2t a + b
t See Prob. J l -6.
SEC.
1 1 -4 .
T H IN -W A L L E D
CLOSED CROSS
S E C T IO N S
289
We express J as J = Jc 1 + For this section, J Jc We consider a > b. Then, Jc \b 1ft\2( 3 \b b\2 1+ aj J
Tc
The section is said to be thin-walled when t b. In this case, it is reasonable to neglect J vs. J c.
Fig. E11-1
The stress follows from (11-61), M ' c >/ , x l l \ where, for this section, < A t
C,
\bj
If the section is thin-walled, we can neglect the contribution of ps, i.e., we can take ctjs ~ o\s = qjt = M, 2 /lc|
W e consider next th e section show n in Fig. 1 1 - I t . R a th e r th a n w o rk w ith if/, it is m ore co n v en ien t to w o rk w ith th e shear flows for th e segm ents. W e n u m b e r th e closed cells consecutively an d ta k e th e + S sense to coincide w ith th e X 2 - X 1 sense. T h e + S sense for the o p en segm ents is a rb itrary . W e define qj as the sh ear flow for cell j a n d w rite Qj = ci (1 1 -6 2 )
N o te th a t C:, is the value o f \j} o n th e in terio r b o u n d a ry o f cell j an d the sh ear flow is c o n sta n t alo n g a segm ent. T he to tal sh ear flow d istrib u tio n is o b ta in e d
290
CHAP. 11
and open s e g m e n ts ; A , and A? a re c e n te rlin e areas.
by superimposing the individual cell flows. Then, the shear flow in the segment common to cells i and j is the difference between 4 , and qj. The sign depends on the sense of S.
fo rS ,
q = q 2 - qi for S2
(11_ 63)
The shearing stress is assumed to vary linearly over the thickness. For con venience, we drop the subscripts on c t v . s and write the limiting values as
<7 = O 0 + <7C
where
o M , c U M > ( C - ' ''
(11-64)
It remains to determine C j, C2, and J. We have shown (see (11-55)) that J = J + J c and M'
We determine J " from
I
segments
(a)
Jc
(b)
t y ds
(11-65)
SEC. 11-4.
THIN-WALLED CLOSED CROSS SECTIONS
291
Substituting for M f ,,| M\ ~ 2qxA y + 2q2A 2 2 j~ (C 1A 1 + C 2A 2) in (b) leads to J c 2(A1 C l + A 2C 2 ) ( 11 - 66 ) The constants C; are obtained by enforcing continuity of < p t on the centerline of each cell. This can also be interpreted as requiring each cell to have the same twist deformation, ky.% $
s.
(c)
(q/t)clS - 2
M J
Ai
' = 1,2
(11-67)
Substituting for q in terms of C and letting a .1 = cj)

I s. I
ds
0,22 O
dS
(112 Oj j
dS
(11-68)
where a 12 involves the segment common to cells 1 , 2 , the continuity equations take the following form;
au C l + ^ 12^ 2 ~
2A x
(11-69)
&i2Ci + o22C 2 = 2 A 2
We solve this system of equations for C t, C 2, then determine J c with (11-66), and finally evaluate the stresses with (11 -64). We can represent the governing equations in compact form by introducing matrix notation. The form of the equations suggests that we define
CY le v
With this notation, J c = 2A'rC aC - 2A Substituting for A in the expression for J c , J c = C raC
a =
a 12
a l2
(11-70)
a i2
(11-71)
(a)
and noting that J c is positive, we conclude that a must be positive definite. The complementary energy per unit length along the centroidal axis is defined by (11-39), _l_ Mf F* = afs dA 2G 2GJ
( b )
f We apply (11-51) to each cell, t See (11-32).
292
CHAP. 11
Since varies linearly over the thickness, the open and closed stress dis tributions are uncoupled, i.e., we can write where
V* open
1
t?* U +
p * u .</
, rfS
t
(11-72)
K 1 2G
"
al,)
U/w
(a.)
It is reasonable to neglect the open contribution when the section is thin-walled.
Example 1 1 -2 ------------------------------------------------------------------- ---------------------The open-section torsional constant for the section shown is
+ 2{b + d + h)tl + /?f f]
Applying (11-68) to this section, we obtain
A x hd A 2 = hb a u = (h + 2d) + h h al2 = l 2
(b)
q22
I ~
H " 2&) -t-

t2
and the following equations for C 1 ? C2 and J.

1
rf'N f ) C, hj
C2 =
dti (c)
- ( ~ ) c l + (lI . + -
+ 2 ^ ) C 2 = 2btj
\h j
hj
J c = 2/i(Ctrf + C2b) J - J + J c
F in a lly , the shear stress intensities in the various segments are
SEC. 1 1 -5 .
T O R S IO N -F LE X U R E W ITH U N R E STR AIN ED W A R PIN G
293
When cl = b. Cx = C2 =
2bti
(e)
and the section functions as a single cell with respect to shear flow.
1 1 -5.
TORSION-FLEXURE WITH UNRESTRAINED WARPING
Consider the prismatic member shown in Fig. 11-12. There arc no boundary forces acting on the cylindrical surface. The distribution of boundary forces
2l
Fig. 1 1 - 1 2 . P ris m a tic m e m b e r in s h e a r lo a d in g .
on the cross section at x x L is statically equivalent to a single force P 2h , acting at the centroid. Also, the end cross sections are not restrained against warping, i.e., out-of-plane displacement. In what follows, we describe St. Venants torsion-flexure formulation for this problem. Later, in Chapter 13, we shall modify the theory to include restraint against warping.
294
CHAP. 11
We start by postulating expansions for the stresses. The stress resultants and couples required for equilibrium at are
x :
F x = F 3 = Aft = M 2 =
F2 = P2
M 3 = P 2(L - x x)
(a)
Introducing (a) in the definition equations for the stress resultants and couples leads to the following conditions on the stresses:
JJ(7n clA =
dA = 0
11X2(7! i dA = P 2( L - .YjJ
JJ(Ti2 i//l = P 2
(b)
j j cr1 3 - 0
j x ia l2)dA 0
The expansion,
~ ~ ~ r
x 2
{ L
x , ) x , ,
(c )
'3
'3
satisfies the first three conditions (i.e., F\ , M 2, M 3) identically since
JJx2
- JJx2x 3
/ 3
= 0
JJx| <4 =
The last three conditions (i.e., F 2, F 3 , Afi) require <t12 ct, 3 to be independent of Xj. This suggests that we consider the following postulated stress behavior:
M3 P2 , ffn = _ X2 = _ . . (L - x,)x2
<3
*3
ffi 2 = ^ 1 2 (^ 2 , x 3) ^13 xi)

@ 22 <^33 = &23 = 0
(11-73)
Introducing (11-73) in the stress-equilibrium equations and stress boundary conditions for the cylindrical surface leads to
P2
a 21, 2 + ^31, 3 +
13
r~ X 2 ~
0
0
0n
(11-/4)
% n2&2 l + < X n 3&3l ~
(on S)
At this point, we can either introduce a stress function or express (11-74) in terms of a warping function. We will describe the latter approach first. The displacements can be found by integrating the stress-displacement rela tions. We suppose the material is linearly elastic, isotropic with respect to the X 2-X 3 plane, and orthotropic with respect to the axial direction. This is a convenient way of keeping track of the coupling between axial and in-plane
SEC. 1 1 -5 .
TO R SIO N -FLE XU R E WITH UNRESTRAINED W A R PIN G
295
deformation.
Substituting for the stresses in (10-74), we obtain Gi = Hi, i = -=r
1
Vi
= - t t t " CL - xi)x 2
V1 P 2
P2 /
= 2 , 2 = - p-ffn =
~ x 'i ) v 2
(a)
3 ,3
Vl
.
-r - ( L
v / , 2 /r
bt
3
X ^ A '2
V 12
1,2 + w -2
= ~ ( ? i 2 function o f x 2, x 3 = ;- c i = function o f x 2, x 3
7i3 u t,
72 3 =
u 2, 3 +
u 3, 2 =
Integrating the first three equations leads to

p2
u l ~ T ~ J~ (LXl i-Xl).X2 + fi(X2, X3)

= 2e T ^ ~ ~X + x3) (b)
U2
l ^X 2
V\P 2
3 = x i ) x 3x 3 + 7 3 (.X j , : c2 )
The functions / 1 ( / 2, / 3 are determined by substituting (b) in the last three equations. We omit the details and just list the resulting expressions, which involve seven constants:
fi ~ C y + C5X2 + C fiX 3 4- </;(x2, X3)

/2 =
C2
Csî + C4X3 k ix i.x3

(L 2E h
x ^ x l + p (Z, 2E J 3
ix O x f
'C)
fi ~ C 3 C ^ X j C 4X 2 + /c 1 x jX 2
The constants C 1 ( C 2, , C 6 are associated with rigid body motion and fc, is associated with the twist deformation.! We consider the following displacement boundary conditions: 1. The origin is fixed:
Ul =
2. A
u2
= u3 = 0
at (0,0, 0)
line element on the centroidal axis at the origin is fixed :
2,
tS e e F.q. (11-5).
= U3,
a t (>
296
TORSIO N-FLEXURE OF PRISMATIC M EM BERS
CHAP. 11
3.
A line element on the X 2 axis at the origin is fixed with respect to rotation in the X 2-X 3 plane: w2, 3 = 0 at (0 , 0 , 0 )
These conditions correspond to the fixed-end case and are sufficient to eliminate the rigid body terms. The final displacement expressions are
p
ui =
u2 * ^ 2
U3
Eih
| -(Lx* - W i ) * i + 4>(x 2, x 3)
- xf) + ' ~ r 7~\ Lyi\ 2Eyl
,r . . X i ) a 2X 3 - f , k 1X l X 2
M 'i* 3
(11 75)
/3
v1^2 hi ^
One step remains, namely, to satisfy the equilibrium equation and boundary condition. The transverse shearing stresses are given by
1
1
C /1
(7, 2 =
0 ,2
A 'lX 3 +
2.C.A
- - - - - (.\\1 3
Xj)
2\
(11-76)
1 1j X 2 -v' - 2 .X 2-X 3 77-0-13 = 0.3 +

k
Substituting for the stresses in (11-74), we obtain the following differential equation and boundary condition for 0 :
72. _ P i A V !
1
ix 2 A *2 x3 \ ,
(11-77)
= /q (a 2x 3 - a (l3x 2) + - - - - dn
< 5 0
, ,
V j P 2
2EI-,
+ % ,< 3X2X 7 ,
The form of the above equations suggests that we express < fi as 4 > +
r
CJ1 J 3
(0 2 r
2)
+ _p . " ( 0 2 r f +
Li ^
2)
(11-78)
where 0 t is the warping function for pure torsion and 0 2 (. and 0 2 rf are harmonic functions which define the warping due to flexure. Substituting for 0 leads to the following boundary conditions for 0 2r and 0 2,;:
SEC. 1 1 -5 .
297
and therefore the formulation is consistent. Terms involving vx /E are due to in-plane deformation, i.e., deformation in the plane of the cross section, and setting vi/E = 0 corresponds to assuming the cross section is rigid. Then, 4 > 2r defines the flexural warping for a rigid cross section and (j)Z d represents the correction due to in-plane deformation. The shearing stress is obtained by substituting for 0 in (11-76). We write the result as Uij <?1 j,t + Glj,r + 1 j,d (j ~ 2, 3) (11-80) where <rlJi( is the pure-torsion distribution and eru r, a ijtd are flexural distri butions corresponding to < f> 2r and < j> 2d:
Cl 2 , r = -r - (2r, 2 ~ i x l )
3 0 2 r. 3
^13,
P> ~r~
(11-81)
0 \ 2,d = ( ~ ~ |-~ [0 2 d . 2 + H v2 + * 3)] _ / v 1G ! \ P 2
<^13. rf I J y 3 ~ ^2-x 3)
The pure torsion distribution is statically equivalent to only a torsional mo ment, M ltt G ik iJ. One can show thatf r dA P 2 d dA 0 JJci y,r dA 0 j j a j dA 0 (11-82)
Note that the shear stress due to in-plane deformation does not contribute
to P 2.
The total torsional moment consists of a pure torsion term and two flexural terms, = GjfciJ + where
S 2r Xi 2 r,?>)dA
4 ^ S2j j
(11-83)
$2 d~
^.X3 + X2 ( j ) 2 d ,3 ~ ~X3 < t > 2 d ,2 )dA
Since < fi2r and < j> 2d depend only on the shape of the cross section, it follows that S2r and S 2 (/ are properties of the cross section. For convenience, we let x 3 = ^S2r I and (11-83) reduces to My = G jiy J - PzZ 3 (11-85) Now, P 2x 3 is the statically equivalent torsional moment at the centroid due
f See Prob. 11-10.
- S2d^J
(11-84)
298
CHAP. 11
to the flexural shear stress distribution. Then, x3 defines the location of the resultant of the flexural shear stress distribution with respect to the centroid. The twist deformation is determined from
K -J- (M i + P2x 3)
\J \J
(11-86)
where M x is the applied torsional moment with respect to the centroid. If P 2 is applied at the centroid, = 0, and
(a) The cross section will twist unless 3 c3 0. Suppose P2 has an eccentricity e3. In this case (see Fig. 11-13), = ejP 2, and
(*3 -
e3)
(b)
For flexure alone to occur, e3 must equal x3.
Fig. 1 1 -1 3 . N otation fo r e c c e n tric load.
Whether twist occurs depends on the relative eccentricity, e3 x3. Now, to find x3, one must determine S2r and S2d. This involves solving two secondorder partial differential equations. Exact solutions can be obtained for simple cross sections. In the section following, we present the exact solution for a rectangular cross section. If the section is irregular, one must resort to such numerical procedures as finite differences to solve the equations. In Sec. 1 1 - 7 , we describe an approximate procedure for determining the flexural shear stress distribution in thin walled cross sections. Suppose the cross section is symmetrical with respect to the X 2 axis. Then, < x2 is an even function of x3 and an3 is an odd function of x3. The form of the boundary conditions ( 1 1 - 7 9 ) requires cp2r and < p2d to be even functions of x3
SEC. 11-5.
299
for this case. Finally, it follows thatf S2r = 0 and S2li ~ 0. Generalizing this result, we can state: The resultant of the shear stress distribution due to flexure in the X } direction passes through the centroid when X j is an axis of symmetry for the cross section.
Shear center
\\
Xi
Fig. 11-14 . C oo rd ina te s of the shear center.
We consider next the case where the member is subjected to P2, P3 and M j at the right end (see Fig. 11-14). The governing equations for the P 3 loading can be obtained by transforming the equations for the P 2 case according to
x 2 -+ x 3
u 2 u 3
x 3 - - X 2
M 3 * -u2 d
dx2
&12
cx2
&li
dx3 013
cx2 -0-12
13
Two additional flexural warping functions must be determined. The expres sions defining the flexural shear stress distributions due to P 3 are
^12, r
-J -
P i, 03r. 2 h P3
r {far, 3 -
O i J .r
7X 3 )
(11-87)
~ X 3X 3J
V 1 G
0-12. d = ^ - 7 - 1 0 3 ^ . 2
I2
' Qj
l
^13 .d " y
t
^1^1 ^3 c l/
1 ^(* 2
2
2\
3j
(p.2 is
even in
x3, < f > t3 is
o dd in * 3s and S2n
S2 < t involve
only integrals of odd functions o f x3.
300
T O R S IO N -F LE X U R E OF PR IS M A T IC M E M B E R S
C H AP. 11
where < f> 3r, < f)3d are harmonic functionst satisfying the following boundary conditions:
8<!> 3r
1 2 ~ a3*3 . 2
"
C(j) i rf _
2 N
2
( H -8 8 )
on
^n2-^-2X3
Jt|j3 I
/.x 2 +
"
x3 \
Note that the distribution due to (j)3d leads to no shearing stress resultants. Finally, the total normal stress is given by
= + 1 7 Xi 7
= ~ ^ L ~ Xl) ( r
+ y 2 X2)
( n - 89)
Superimposing the shearing stresses and evaluating the torsional moment, we obtain
Afx = G j M
- P25f3. + P 3x 2
(11-90)
where x 2 defines the location of the resultant of the flexural shear stress distri bution due to P 3. One can interpret x 2, x 3 as the coordinates of a point, called the The required twist follows from (11-90):
s h e a r c e n t e r .
f { M \ + P2X2 , ~ P 3-
2}
(3-)
Since (see Fig. 11-14)

M y + P 2x 3 Pfx2
= the applied moment with respcct to the shear center = M r we can write (a) as ki =
7 7 -7
Mr
(11-92)
To determine the twist deformation (and the resulting torsional stresses), one must work with the torsional moment with respect to the center, not the centroid. For no twist, the applied force must pass through the shear center. In general, the shear center lies on an axis of symmetry. If the cross section is completely symmetrical, the shear center coincides with the centroid. It is of interest to determine the complementary energy associated with torsion-flexure. The only finite stress components are <ru , a 12, arid crJ3. Then V * reduces to
s h e a r
r
y * = L
1
Ly
1 ,
CTy
1
LTy
dA
(a)
The contribution from a l: follows directly by substituting (11 -89) and using the definition equations for / 2, 13.
t The total flexural warping function fo r P 3 is
* m G j l V i r ~ l Xi) + I h V "
+ V 3
SEC. 1 1 -5 .
TO R S IO N -FLE X U R E WITH U N R E S TR A IN E D W A R PIN G
301
= _L m + Ml ) 2Ei \ l 2 I3) N ow , the to tal sh earin g stress is the sum o f three term s: 1. 2. 3. <7 a p u re torsional d istrib u tio n due to M T aFl, the flexural d istrib u tio n due to F 2 o>3, th e flexural d istrib u tio n due to F 3
(11-9 3 ) {
E ach o f th e flexural d istrib u tio n s can be fu rther subdivided in to 1. 2ar. Fp th e d istrib u tio n co rresp o n d in g to a rigid cross section (defined by (pjr) < ?d, Fp th e d istrib u tio n associated w ith in-plane defo rm atio n o f th e cross section (defined by cpjd)
W e com bine th e flexural d istrib u tio n s an d express the to tal stress as

^12 ^13 = = ^ 1 2 ,t + 2,r + + OlZ.d <*13, J
(a) F o r exam ple, (b)
w here the various term s are defined by (1 1 -8 1 ) an d (11-87). Fi /,

o 1 2. r = -J - W 2 r , 2 I3
j
7 x 2l + V ^2
,
3r, 2
T he com p lem en tary energy due to pure to rsio n follows from ( U - 3 8 ) and (11-92): ' Mr (1 1 -9 4 ) (cf 2,( + v 2 n .r)dA W e express <Tiy, r as
0 \2 ,t ~
_ [ -A x a. 7 r3 F >0Z3 r , 7 <p2r.2 +
I3
I2
(11-9 5 )
ffl3 , r 7 0 2 1 %3 + ' 3
7 " 03r. 3 *2
w here 02r = (p2>- ~

13
<]>3r = < t> 3. ~ ^ 3
E x p an d in g (<7l2,r + o"i3, r) an d in tegrating over the cross section, we o b ta in ! F i , 2 F 2F 3 , F 3 (^1 2. r + 0"j3,rM-4 7 - -I- ~ IA-2 A 23 ^3
1
2 + <Pjr,3)dA I /t 2 3 hi
Xj(l)jr dA
j # /< j, k = 2, 3 I
(11-96)
(02r, 2(/>3r, 2 + <l>2r, 3<t>ir, l ) d A
/ 2^3
x d>4>2r d A
t See Prob. 11-11.
302
TO R SIO N-FLEXU R E OF PR IS M A TIC M EM BER S
CHAP. 11
The coupling term , \ f A 23, vanishes when the cross section has an axis of symmetry. We consider next the coupling between a . , and o. r.
jj(& 1 2,t< 7 l2 ,r
Mf
Y+ /%*
a l3 ,[ < ? t3 ,r ) d A
=T
+
W l , 2 X j ) I Y ~ (P 2r 2 +
' '
j.
(f>3r, 2
\ i
3 + * 2) <t) 2 r , 3 +
12
- (f) 3 ,., 3
MT T~
7~ < P2t + ~
'3 7 '3
12
03r
y
\ (4 > i.2 ~
X 3K
2 +
(01. 3 +
* 2 )0.,3
j> t dA = 0 0 3 f J V 2< h J The rem aining term s involve or,* the shearing stress distrib u tio n due to in plane deform ation of the cross section. We will n o t attem p t to expand these term s since we are interested prim arily in the rigid cross section case. Sum m arizing, the com plem entary energy for llexure-torsion w ith unre strained w arping is given by
r 2r +
p* =
1 2 Ei
_ i + 2 - -f (11-98) ^2 ^23 ^3 j 4- term s involving v J E
^ F 2F 3
F V
where M T + jF2x 3 F 3;c2. W e introduce the assu m p tio n of negligible in-plane deform ation by setting v J E = 0. Similarly, we intro d u ce the assum p tion of negligible w arping due to flexure (0 2r ~ 0, < }3r w 0) by setting 1 /4 1 = 1/^2 1/^23 = 0. In Sec. 11-7, we develop an ap proxim ate procedure, called the engineering theory , for determ ining the flexural shear stress distrib u tio n , which is based upon integrating the stress-equilibrium eq uation directly. This ap p ro ach is sim ilar to the torsional stress analysis procedure described in the previous section. Since the shear stress d istrib u tio n is statically in d eterm in ate when the cross section is closed, the force red u n d an ts have to be determ ined by requiring the w arping function to be continuous. F o r pure torsion, continuity requires (see (1 1 -3 2 )) . s & i s , t d S 2Gl k l A$ (a) where the integration is carried o u t in the X 2- X 3 sense aro u n d S, and A s is the area enclosed by S. To establish the continuity con d itio n s for flexure, we operate on (11 -8 1 ) and (11-27). T here are four requirem ents:
< /f ijr => s is, r )fj d S =
0
2,3
SEC. 1 1 -6 .
EXACT F LE XU R AL SHEAR STRESS DISTRIBUTIO N
303
In the engineering theory of flexural shear stress distribution, the cross section is considered to be rigid, i.e., the d istrib u tio n due to in-plane d efo rm atio n is neglected. T he consistent co n tin u ity condition on the flexural shearing stress is j s o l s dS = 0 (11-100) O ne can take the + 5 direction as either clockwise o r counterclockw ise. By definition, the positive sense for o ls coincides with the + 5 direction.
1 1 -6 . E XA C T F LE X U R A L S H EAR STRESS D IS TR IB U TIO N FOR A R E C TA N G U LA R CRO SS SEC TIO N
W e consider the problem o f determ ining the exact sh ear stress d istrib u tio n due to F 2 for the rectangular cross section show n in Fig. 11-15. F o r co n venience, we first list the governing eq u atio n s: *2
, = Ji2 sl h , } h -d t 12
A **dt
Fig. 11-15.
N o ta tio n fo r re c ta n g u la r c ro s s s e c tio n .
1.
W arping functions
304 2.
TORSION-FLEXURE OF PRISMATIC MEMBERS Shearing stresses

&12
CHAP. 11
~jr(<l>2r, 2 \ X\ ) H ---- l . y - - [(/>2d. 2 + H X 2 + X l )]

^3 3
F2
Cl 3 ~ ( $ 2 r, 3 / -i J3
V{G\E2
------ \<P2d,3 *2 *3.) 3
Determination o f (j)2lT he bo u n d ary conditions for < /)2r are 1 AA2 ** * i \ i ) 3 = 0 W e can tak e the solution as < t> 2 r = y 2*2 The corresponding stresses and w arping function are 02r = 4>2r ~ kx 2 ~ gdZX2 pXj (b) d a t l '2 = 2 (a) a t ^3 = 2
ff13,r = 0 O ne can readily show that JJ0"12, r dA = F 2 Finally, we evaluate 1/A2 using (11-96): 1 _ I x 2(j)2y dA = ~ (11 - 102)
4 i * /? Determination o f (j)2d The b o u n d ary conditions for I (d2 4}2d.2 2 ("4 V t

<P2d,3 =
2
are d at x 2 2 y t
a t .X3 = + -
(a)
N ow , the form o f (a) suggests th a t we express (j)2d as (p2 d = 1*2*3 - 6*2 - f ( x 2 , x 3) (b)
SEC. 1 1 -6 .
EXAC T F LE XU R AL SHEAR STRESS D ISTR IBUTIO N
305
w here / is an h arm o n ic function. T he shearing stresses and b o u n d ary co n d i tions expressed in term s of / are <ri2.a m - g - (JES j, 2) vl G l F 2 , '( E 1, and f,
2
V y G
(c)
f,
3)
= x\
d at x 2 = + (d)
f 3 = 0
at x 3 = - f j ~2
It rem ains to solve V2 / = 0 subject to (d). Since the cross section is sym m etrical, / m ust be an even function of x 3 and an odd function o f x 2. W e express / as f = B 0x 2 +
=
v 2,
1, 2 , . . .
( 2 n n x 2\ . , ( 2 n n x 2\ cos ( - I sm n ^ I
(e)
This expansion satisfies V2 / = 0 and the b o u n d ary condition a t x 3 = t j 2. T h e rem aining b o u n d ary co n d itio n requires . , m/uA 2 nnxi 7 c o s h j c o s - = x f t l\ I - - < v3 < + - J
(f)
E xpandin g x l in a F o u rier cosine series an d eq uating coefficients leads to t2 r Bo 12 (g)
s, = 5 ( - ) * - - - d 2 \n n I , nnd y cosh f T he final expressions for the shearing stresses are cosh E J, -)
t Y
^ (-1 )" 2 /m X j
2n n x z \ nnd j t /
I
= 1 , 2, . . .
_ COs - ---11 1
1 \ cosh
ft/
VlGy F 2 ^13, d E h
sinh (i_Y v
n- l \2.. .
2nnx-> nnd sin 2nnx* (11-103)
(-ir
" 2
cosh
T his system is statically equivalent to zero.
306
CHAP. 11
To investigate the error involved in assuming the cross section is rigid, we note that the maximum value of a 12, r occurs at x 2 0: 12,1 Specializing ffi2,d for x 2 0,
(<5'l2,d)x2= 0
(a)
v1G 1 F2 d2 7 r E h 4 4 (-i r ( ,
I 1 cosh A
2 rmx3 C* cos t
(b)
where C = = Now, C decreases rapidly with n. Retaining only the first term in (b) leads to the following error estimate, \a 12,d\
'12. f
V
rind
(c)
2 vl G] E
1 cosh Ttd (d)
Results for a representative range of d/t. and isotropic materia) are listed below. They show th at it is reasonable to neglect the corrective stress system for a rectangular cross section. The erro r decreases as the section becomes thinner, i.e., as d/t becomes large with respect to unity: d/'t 2 1 1 2
1 1 -7 .
0.024 0.092 0.122
ENGINEERING THEORY OF FLEXURAL SHEAR STRESS DISTRIBUTION IN THIN-W ALLED CROSS SECTIONS
The exact solution of the flexure problem involves solving four secondorder partial differential equations. If one assumes the cross section is rigid w ith respect to in-plane deform ation, only two equations have to be solved. Even in this case, solutions can be found for only simple cross sections. W hen the cross section is irregular, one must resort to a numerical procedure such as finite differences or, alternatively, introduce simplifying assum ptions as to the stress distribution. In w hat follows, we describe the latter approach for a thinwalled cross section. The resulting theory is generally called the engineering theory of shear stress. We apply the engineering theory to typical cross sections
SEC. 1 1-7.
ENGINEERING THEORY OF FLEXURAL SHEAR STRESS
307
and also illustrate the determ ination of the shear center and the energy co efficients, IjAj \j -= 2, 3). Figure 11 -1 6 shows a segment defined by cutting planes at x x and x r + d x u Since the cross section is thin-walled, it is reasonable to assume that the norm al stress. < rl u is constant through the thickness and to neglect a ln. Also, we work Xi
/ 1 _d / to, / on ids
x 2
F ig . 1 1 - 1 6 .
D iffe re n tia l th in -w a lle d s e g m e n t.
with the shear flow, q, rather than with < ylS. Integrating the axial force-equilibrium equation, dq (a) dxx dS
with respect to 5, we obtain the following expression for q,

< 3=
<Ia
d fs ctj t dS cx 1 J S a
(11-104)
Equation (11-104) is the starting point for the engineering theory of shear stress distribution. Once the variation of a n over the cross section is known, we can evaluate q. Now, we have shown that the normal stress varies linearly over the cross section when the member is subjected to a constant shear ( f 2, F3
308
TO R S IO N -FLE X U R E OF PR IS M A T IC M E M B E R S
C H AP. 11
con stan t) and the end sections can w arp freely. N o tin g th a t th e m em b er is prism atic, the derivative o f eru for this case is
da1
x3 dM2 J 2 dx i
dM3 (a)
dx,.
I 3 dx j
a n d (11-104) expands to
? = Qa ~
r
7
fs
, f 3 x 2t dS 7-
fS x 3r dS
S
a
(b)
3
1 3
T he integrals rep resen t the m o m en t of th e segm ental area w ith respect to X 2 , X and are generally d en o ted by < 32, Q 3 : q
2
= f* x 3I dS = Q 2(S, Sa) * Sa v Sa
(11-105)
q.3 =
W ith this n o tatio n , (b) simplifies to q = q A - ~ 1 Q3 - ~ - Q ^3 ^2

.2
(U -1 0 6 )
E q u a tio n (11-106) defines the sh ear flow d istrib u tio n for th e case o f negligible restrain t against w arping, i.e., for a linear v ariatio n in n o rm al stress. N o te th a t q is positive w hen p o in tin g in th e + 5 direction. W e co n sid er first the open section show n in Fig. 1 1-17. T h e end faces are unstressed, i.e., < Ia = = 0 (a)
q B
T ak in g the origin for S at A, ( t l - 1 0 6 ) reduces to _ q F2 n /3 F3 I 2 Qz Q2 JO X31 dS
(11-1 0 7 )
6 3 f x 21 dS J0
W e determ ine Q2, Q} and then com bine acco rd in g to (11-107). T he shearing stress d istrib u tio n co rresp o n d in g to F 2, Q= ~ ~
*3
0.3
(a)
satisfies j j<r12 dA F 2 dA = 0 identically. T o show this, we ex p an d q , q = qis = (qs 2 ) h + ( q ^ d h (c) (b)
SEC. 1 1 -7 .
E N G IN E ER IN G TH EO R Y OF FLE X U R A L SH E AR STRESS
309
*3
Fig.
1 1 -1 7 .
F le x u ra l s h e a r flo w o p e n s e g m e n t,
and evaluate the shear stress resu ltan ts: rr o l 2 dA = a l3 d A = E q u atio n (b) requires p | qa S 2 dS qrxs 3 , r r ff 2 aS2S3 dS
(d)
ii F2 dS
77
y-s'iQi d s
I
[o
dxt av - I s
~ L~
(e)
dS 0
N ow , (f)
In teg ra tin g (e) by p arts an d noting th a t X 2, X 3 are prin cip al cen tro id al a x e s,t we o b ta in 0 ^siQ^dS = J o x \ t dS / 3 (g)
JO
s* y-siQids = [ x 3?3]ga - | Sb x 2 x 3t dS = 0
j
T h e sh ear stress d istrib u tio n pred icted by (a) is statically equivalent to a force F 2 ?2 T o determ in e the lo catio n of its line o f action, J we ev aluate the m o m en t w ith respect to a convenient m o m en t center. By applying th e sam e argum ent, one can show th a t the sh ear flow co rresp o n d in g to F } is statically
t See Eq. (11-2). j See Prob. 11-12.
310
TORSIO N-FLEXURE OF PR ISM ATIC M EM BER S
CHAP. 11
equivalent to a force, F 3I 3. T he intersection of the lines of action of the two resultants is the shear center for the cross section (see Fig. 11-17).
Example 11-3 --------- ------------------------------ -------- ------

C o n s i d e r th e th in r e c ta n g u la r s e c tio n s h o w n . + q p o in t s in th e + X W e ta k e + S in t h e + X
2 d ire c tio n .
Then,
2 d ire c tio n
a n d q jt
a 12. T h e v a rio u s te rm s are
T h i s r e s u l t c o i n c i d e s w i t h t h e s o l u t i o n f o r cr1 2 . r o b t a i n e d i n S e e . 1 1 - 6 . n e e r i n g t h e o r y is e x a c t f o r a r i g i d c r o s s s e c t i o n , i.e., f o r v J E 0.
A ctu a lly , th e e n g i
Fig. E11-3
i
2
Example 11-4 -----------------------------------------------------------------------------------W e o f F ig . d eterm in e E I1-4 A . th e d istrib u tio n O n ly tw o o f q co rresp o n d in g A B and to F
2
to
for be
th e
sy m m e tric a l sin ce
se ctio n j <23 [ i s
segm en ts,
B C , have
co n sid ered
sy m m e tric a l.
Segment AB
Q
3 nip h t.fS
A cco rd in g
to
our
d e fin itio n ,
+ q
p o in ts
in
th e
+ S
d ire c tio n
(fro m
to
B ).
S in ce
is
n e g a t i v e f o r t h is s e g m e n t , it a c t u a l l y a c t s i n t h e n e g a t i v e S d i r e c t i o n ( f r o m B t o A ).
SEC. 1 1 -7 .
ENGINEERING THEORY O F FLE XU R AL SHEAR STRESS
311
^2
Fig. E 1 1 -4 A
*3
I Segment B C
W e m ea su re S fro m B to C . Then,
-br
(?3
h bj-tj- + +
X2 ) } t H( h 2 x
q _
2) ]
h
N o t e t h a t t h e a c t u a l s e n s e o f q is f r o m in F ig . E 1 1 - 4 B . It is o f i n t e r e s t t o e v a l u a t e A 2. S p e c ia liz in g ( 1 1 - 9 6 ) fo r a th in - w a lle d se ctio n , C to B . T h e d istrib u tio n a n d sen se o f q a re s h o w n
dS = F |
(o7.s\ r)r2 d A
a n d s u b s titu tin g fo r q y ie ld s 2 dS
A,
(a)
We let
^2
A w Ar A t = -
n J 23 t
a r e a o f th e w e b to ta l fla n g e a r e a kA w = = d t 2 b f tf
(b)
(c)
The resulting expression for A r'is

2 A ..
1 A ..
312
TO R SIO N-FLEXU R E OF PR ISM ATIC M E M B E R S X7
CHAP. 11
Fig. E 11-4B
T h i s f a c t o r is q u i t e c l o s e t o u n i t y .
F o r e x a m p le , ta k in g as ty p ic a l, fo r a w icie -fia n g e s e ctio n .
If = 2 tw
bf w e fin d = y
Af 3AW
k = 0 .9 5
T h e s h e a r i n g s t r e s s c o r r e s p o n d i n g t o F 3 v a r i e s p a r a b o l i c a l l y in t h e f l a n g e s a n d is z e r o in th e w e b . E a c h fla n g e ca rrie s h a lf th e s h e a r a n d
1
/Ij
6 1
5 A f
(c)
5 h tf
Example 1 1-5 Cross-Sectional Properties

T h i s s e c t i o n ( F i g . E l 1 - 5 A ) is s y m m e t r i c a l w i t h r e s p e c t t o X 2- T h e s h i f t in t h e c e n t r o i d f r o m t h e c e n t e r o f t h e w e b d u e t o t h e d i f f e r e n c e in f l a n g e a r e a s is
= b + b 2t2 + d tv
(a)
W e n e g l e c t t h e c o n t r i b u t i o n o f t h e w e b in I 2 s i n c e it i n v o l v e s t l :
l i ~ Ui)i + {h)i Tjihb\ + t2 bl)
(b)
SEC.
1 1 -7 .
E N G IN E E R IN G T H E O R Y O F F L E X U R A L S H E A R S T R E S S
313
Determination o f Q 2
T a k i n g S a s s h o w n in th e s k e tc h , w e h a v e
- j & |i = a
6 2
]*.- =
s i n c e X 2 is a n a x is o f s y m m e t r y .
(c)
Fig. E11-5A
+q~ b\
T*
-1 / / / 1
C e n tro id
,T
+?Xx b2
~h
Distribution o f q Corresponding to F}
T h e s h e a r f l o w c o r r e s p o n d i n g t o F 3 is o b t a i n e d b y a p p l y i n g
q = --lQ
12
(d)
a n d is s h o w n in F i g . E l 1 - 5 B . in e a c h fla n g e .
T h e s h e a r s tre s s v a n is h e s in th e w e b a n d v a r ie s p a r a b o l i c a l l y
I n te g r a tin g th e s h e a r flo w o v e r e a c h fla n g e , w e o b ta in
F3
11
(e)
Then,
th e d is tr ib u tio n
is s t a t i c a l l y e q u i v a l e n t t o .
F 3 73 a c t i n g a t a d i s t a n c e e f r o m
th e left
fla n g e , w h e re
- - ------ =
8 2
-^ R
d -
1 1 2) 2
ff)
Since X 2 is an axis o f sym m etry, the shear center is located at the intersection o f R and X 2,
314
CHAP. 11
*3
Fig. E11-5B
T h e c o o r d in a t e s o f t h e s h e a r c e n t e r w it h r e s p e c t to th e c e n tr o id a r e
x 3 0 t,I
v2 = e (1 + A) -
(g)
= d Torsional Shear Stress
M ih
1H -A
.V
T h e f l e x u r a l s h e a r s t r e s s d i s t r i b u t i o n is s t a t i c a l l y e q u i v a l e n t t o a t o r s i o n a l m o m e n t e q u a l t o F 3j ? 2 w i t h r e s p e c t t o t h e c e n t r o i d . w ith M i respect to th e ce n tro id . W e h a v e d efin ed M , as th e re q u ir e d to r s io n a l m o m e n t m om ent w h ich m u st b e b a la n c e d by to rsio n is
T h e n , th e
F 3x 2 M t , t h e r e q u i r e d t o r s i o n a l m o m e n t w i t h r e s p e c t t o t h e s h e a r c e n t e r . in S e c . 1 1 - 3 , th e m a x i m u m
U sin g
th e a p p r o x im a t e th e o r y d e v e lo p e d s e g m e n t is
t o r s i o n a l s h e a r s t r e s s in a
M r
''maxi/
w here
./
(h)
(i)
We consider next the closed cross section shown in Fig. 11-18. We take the origin for S at som e arbitrary point and apply (11-106) to the segm ent SP-S:
< 7 = /i - 7 - 8 3 - 7 - Q 2
j Sp
I' 2 n
(11-108)
x 3f dS
<2s
rs
Sp
where is the shear flow at P. T he shear flow distribution is statically indeter m inate since qy in unknow n. We have previously shown th a t q q 1 con-
SEC. 1 1 -7 .
315
stant is statically equivalent to only a torsional m om ent equal to 2 ^1y4cj73. The second and th ird term s are statically equivalent to F 2 T2 and F 3 i3. T he co n stan t q t is determ ined by applying the continuity requirem ent to the centerline curve. Since the engineering theory corresponds to assum ing the cross section is rigid with respect to in-plane deform ation, we use (11-100). A'j
Fig. 11-18.
N o ta tio n fo r c lo s e d c e ll.
T he flexural shear stress distribution m ust satisfy I I dS ( ) a is dS O q - - 0 Js ~s ' for an arb itrary closcd cu rv e.t Substituting for q, dS dS F, F-i , O Q3 - - + r h Scl dS
() 0.2
(11-109)
Sci
(a)
and considering separately the distributions corresponding to F 2 and F 3, we obtain q = < -h,. + F F Q2) < Jr2 ~ r ( & 2 ~ 6 3 ) < 7 Ft j C 3
' 3 J2
dS Q3 r () Scl dS
C o
dS
2
( 11-110)
(D t
Each distrib u tio n satisfies (11-109) identically. Also, the distrib u tio n (q)F. is statically equivalent to a force Fjlj, located x k units from the centroid. N ote that q Bj leads only to a torsional m om ent equal to 2BjAcl.
t One can interpret f l ) -1 0 9 ) as requiring the Bexural shear stress distribution to lead to no twist deformation. See Prob. 1 1 -1 4 for the more general expression, which allows for a variable shear modulus.
$16
T O R S IO N -F L E X U R E O F P R IS M A T IC
MEMBERS
CHAP. 11
T he general expression for i / A j follows from (11-96);

d .S F ]
(11-111) j =
2' 3
Substituting for (g)f j ,
v '
. t o V r - tJ
1 1 T = 72 ' (f (/^ 2B' G* + 6*) T * k - j . k = 2>3) Aj / t i 5c l m d noting that 1 dS I az>
(b)
B,-cb - - = 6
Scl
s.,i Sc
Qk
tve obtain =
5c/
dS
(c)
(11-112)
, ,
Qk ~
.Is ., J >d 1
(j * k ; j t k = 2, 3)
*j
Ik li *
which applies for an arb itrary single cell. --------------------Example 1 1 - 6 ------------------------------------------------ -------
W e illu stra te th e d e te r m in a tio n o f (<?)r, f o r t h e s q u a r e s e c t i o n o f F ig . E 1 1 - 6 A . It c o n v e n i e n t t o t a k e P a t t h e m i d p o i n t s i n c e t h e c e n t e r l i n e is s y m m e t r i c a l .
.
is
Fig. E11-6A
Cross-Sectional Properties A ci = a 2 / 2 = 3 r(l2 ) + 2 ( ^ ,4 - = 4^

A =
/ t f 3N i
<?2
Sat
10
SEC. 1 1 -7 .
317
Determination o f Q z
W e s ta r t a t P a n d w o r k c o u n t e r c lo c k w is e a r o u n d th e ce n te rlin e . t i o n a n d a c t u a l s e n s e o f q d u e t o Q 2 a r c s h o w n in F i g . F,1 ] t o a n e g a t i v e i.e., c l o c k w i s e , q . 6B. T h e re su ltin g d is tr ib u N o te th at + Q 2 c o rre sp o n d s
Fig. E11-6B
Evaluation o f B}
B y d e fin itio n ,
^ B, -
dS Ql
i r
i f
3.5 t
ds
is e q u a l t o (2 / 3 ) ( b a s e ) x
U s in g th e a b o v e resu lts, a n d (h eig h t), w e o b ta in
n o tin g th a t th e a r e a o f a p a r a b o la
f
* /2
dS
..
I> & - ;- = + cm
28
Distribution o f Flexural Shear Flow fo r Fz

T h e s h e a r f l o w is g i v e n b y
V2 1
3 a
t)
(+
sen se clo ck w ise). T o
T h e t w o d is tr ib u tio n s a r e p lo tte d in F ig . E l 1 ~ 6 C . m o m e n t s a b o u t t h e m i d p o i n t (O i n
lo c a te th e lin e o f a c tio n o f th e re su lta n t, w e s u m
th e s k e t c h : 19 aF
126 The resultant acts e units to the rig h t o f 0 , where
318
TORSION-FLEXURE OF PRISM ATIC M EM BERS
CHAP. 11
Fig. E11-6C
avn/
F i n a l l y , t h e c o o r d i n a t e s o f t h e s h e a r c e n t e r w i t h r e s p e c t t o t h e c c n t r o i d ( w h i c h is A u n i t s t o th e rig h t o f 0 ) are 16 H ---------- a 315
x 7 =
Torsional Shear Flow

T h e s h e a r f l o w f o r p u r e t o r s i o n is d u e t o M T , t h e t o r s i o n a l m o m e n t w i t h r e s p e c t t o t h e s h e a r center. F o r th is se ctio n ,
Mj
W e a p p ly th e th e o ry
_ . _ Mi x-2 ^ 3 + X3 F 2 M j
in S e c . 11-4 . O n e ju st
16
al' :
re p la ce M , w ith M T in
d e v e lo p e d
has
to
SEC. 1 1 -7 .
E N G IN E E R IN G T H E O R Y O F F L E X U R A L S H E A R S T R E S S
319
Equation (11-61): , ^lsjnuix MT f Cj \t + / V v!) Ci = $at J = a3t + Hat3
Determination o f 1jA i Applying (11-112), we find i i /X s : i i Note that 3at is the total web area.
?ds
r,
L276 f/s\ 1.2 J - h3at
We consider next the analysis of a two-cell section and include open segments for generality. There is one red u n d an t shear flow for each cell. We select a convenient point in each cell and take the shear flow at the point as the redun d an t for the cell. This is illustrated in Fig. 11 19: qj represents the shear flow redundant for cell j and the + S sense coincides with the X 2 ~X 3 sense to be consistent with the purc-torsion analysis. The + S sense for the open segments is inw ard from the free edge. For convenience, we d ro p the CL (centerline) subscript on S and A.
Fig. 1 1 -1 9 . Notation for multi-cell section.
320
T O R S IO N -F LE X U R E O F P R IS M A T IC M E M B E R S
C H A P . 11
T he to ta l sh ear flow is the sum o f q 0> the open cross-section d is trib u tio n (<?i = < ?2 = 0), a n d qR, the d is trib u tio n due to th e re d u n d a n ts: q qo + <7r (1 1 -1 1 3 )
W e d eterm in e q 0 by ap p ly in g (1 1 -1 0 7 ) to the v ario u s segm ents. T h e re d u n d a n t shear-flow d istrib u tio n is th e sam e as for p u re to rsio n (see Fig. 1 1-11). F inally, we o b ta in a system o f e q u a tio n s relatin g q u q 2 to F 2, F 3 by ap p ly in g the co n tin u ity req u ire m en t to each centerline, f dS = n q 0 t (1 1 -1 1 4 )
w here q is positive if it p o in ts in th e + 5 directio n . U sin g th e ajk n o ta tio n defined by (11 -6 8 ), the e q u atio n s ta k e th e follow ing fo r m : a llQl + a 12<?2 a 12^1 + ^22^2 - F>2 f dS = Dj (F 2, F 3) - ^ (In <h I (1 1 -1 1 5 )
D:
T h e sh ear flows ( qu q 2, ,) for pure to rsio n are related by (we m u ltip ly (1 1 -7 1 ) by M t / J a n d n o te (11 - 6 2 ) ) Mj an c Ji,t + ^ 12^ 2, t 2 /li r
a 12c h , t + a 2 2 q 2 ,t 2 A 2
(1 1 -1 1 6 )
Mj __
T h u s, th e co m p lete sh ear stress analysis involves so lv in g aq b for th ree differ en t rig h t-h a n d sides. T h e eq u a tio n s d ev elo p ed ab o v e can be read ily generalized.
Example 11-7 ------------------------------------------------ ---------------------------------------W e d e te r m in e th e fle x u ra l s h e a r stress d is tr ib u tio n show n in F ig . E 1 1 -7 A . W e lo c a te l\ and P 2 at c o r r e s p o n d in g to f 3 fo r th e se ctio n
the
m id p o in ts
to
ta k e
ad van tage
of
sym m etry.
C ross-Sectional Properties
A i 2a z a2 gtf
A 2 A =
/ , - 3 , l+ 2
6a
a
(3 a t ) 4 4a =
7 4 a
c l
a 22
f See Prob. 11 34 for the more general expression, which allows for a variable shear modulus.
SEC. 1 1 -7 .
EN G IN E ER IN G TH EO R Y OF F LE X U R A L SH E AR STRESS
321
*3
Fig. E 11-7A
2a-
D istribution o f qR
T h is system (F ig . E l 1 - 7 B ) is s t a t ic a lly e q u iv a l e n t to a m o m e n t
2 a 1{ 2 q l
q 2 ) .
D istribution o f q 0 Due to Fs
W e a p p ly
q = - --3 - 02
to th e v a r io u s s e g m e n ts s ta r tin g a t p o in ts P , , P 2. T h e r e s u l t i n g d i s t r i b u t i o n is s h o w n in
F ig . E l l - 7 C .
D eterm ination o f q x, q 2
Dt =
s, q
ds
' dS ~ ~ .
i r,
7 ~ 2 F j
>2
T h e e q u a tio n s fo r a n d </2 a r e
Si
t/o - '
+ - - 1 '
e.
*^ 3
7 a
x = 2 f3 - /, + 4q2
7 S o l v i n g (a), w e fin d a
(a)
2 F, T77
161 11 161 a F 3 a
(b)
Vl + T77
T h e t o t a l d i s t r i b u t i o n is o b t a i n e d b y a d d i n g q R a n d q 0 a l g e b r a i c a l l y .
322
TORSION-FLEXURE OF PRISMATIC M EM BER S
CHAP. V,
Fig. E11-7B
< 7 1
i
2
q va q2
I 1 1
( < 7
qâ *i
1
W 2 qia
1
< 7 1
q xa
t
< 7 2
Fig. E11-7C
Location o f the Shear Center

T a k i n g m o m e n t s a b o u t th e m i d p o i n t o f th e le ft w e b , a n d le t tin g e b e th e d is t a n c e t o th e lin e o f a c tio n o f th e re s u lta n t, w e o b ta in
M( +
)> =
2 a 2( 2 q >
92)
(2a)
(3a) ( ~
F 3^)
et\
f 2 f
32\ + 2? /
x 2 =
T h e s h e a r c e n t e r is l o c a t e d o n t h e X 2 a x i s a n d e (2 a A) = + 0 .0 5 5 #
REFERENCES
1.
2.
W ang,
C.
T .:
A p p lie d E la s tic ity , M c G r a w - H i l l , N e w Y o r k . a n d J. N .
1953.
T im o s h e n k o , S. J .,
N e w Y o r k , 1970.
G o o d ie r :
T h e o r y o f E la s tic ity , 3 d e d ., M c G r a w - H i l l ,
PR O BLEM S
3.
4.
323
M a te ria ls, M c G ra w -H ill, N ew Y ork,
D e n H a r t o g , J. P .:
1952.
A dvanced
S tren gth
o f
H errm ann,
L . R . : E la stic T o rs io n a l A n a ly sis o f Irreg u lar S h a p e s , D ecem ber 1965. T heory o f E la s tic ity ,
J.
Eng. M ech .
D iv ., A . S . C . E . , 5. S o k o ln ik o ff, 1956. 6. 7. 8.
I . S . : M a th e m a tic a l
M c G ra w -H ill,
N ew
Y o rk,
im o s h e n k o , S.
J.: S t r e n g t h o f M a t e r i a l s , P a r t 2 , V a n N o s t r a n d , N e w Y o r k , 1 9 4 1 .
Y o r k , 1966. T h i n - w a l l e d G i r d e r s ),
C e r n ic a , J .
D a b ro w sk i,
N . : S tren g th o f M a te ria ls, H o lt, R in e h a rt, N e w R . : G e k r iim m t e d im m v a n d ig e T r a c e r ( C u r v e d
S p r in g e r -V e r la g , B e rlin , 1968. 9.
K ollbrunner,
1969.
C.
F.,
and
K.
B a s i e r : T o r s i o n in S t r u c t u r e s , S p r i n g e r - V e r l a g , B e r l i n ,
10.
V la sov, V. Z . : O d e n , J.
T h in - W a lle d E la s t ic
B ea m s, Israel P ro g ra m
fo r S c ie n tific
T ra n s la
tio n s, J e ru sa le m , 196 1. 11. T . : M e c h a n ic s o f E la s tic S tr u c tu r e s , M c G r a w - H i l l , N e w Y ork, 1967.
PROBLEMS
11 -1 . T he pure-torsion form ulation presented in Sec. 1 1 -2 considers the cross section to ro tate ab o u t the centroid, i.e., it takes u2 = - c o , * , u 3 = +( 0
^ 2
c 1 = k lx l ! = fc,0t
Suppose we consider the cross-section to ro tate about an arb itrary point (x*, A'|). The general form of (a) is u 2 = - w , ( x 3 - x |) 3 (a) (b)
+ C U 1(X 2 X ^)
w , = k lx i + c t
U, = k ,< *
S tarting with E q u atio n (b), derive the expressions for crt2, a i 3 and the governing equations for </>*. W hat form do the equations take if we write (/> * = < j)t + cz - x 2x f + x 3x f
D o the torsional shearing stress distrib u tio n and torsional co nstant J depend on the center of tw ist? 1 1 -2 . Show t h a t ./ can be expressed as
J = Jj[x| + x^ - {(j)lt 2)2 - (< /> ,. 3)2]d/l

= h - i m Hint: . V2< /> dA = 0
.*)2
+ i4>us)i ']iA
C om pare this result w ith the solution for a circular cross section and com m ent on the relative efficiency of circular vs. noncircular cross section for torsion. 1 1 -3 . Derive the governing differential equation and b oundary condition for 4 > t for the case w here the m aterial is o rth o tro p ic and the m aterial sym m etry axes coincide with the X x, X 2, X 3 directions.
324
C H AP. 11
1 1 -4 . T he v aria tio n in th e w arp in g function (f)t alo n g an a rb itra ry curve S is o b ta in ed by in teg ratin g (11-29), (a) w here pct is the p erp en d icu lar d istan ce from th e cen ter o f tw ist to th e tangent. O n e selects a positive sense for S. T he sign o f p is positive w hen a ro ta tio n ab o u t th e cen ter o f tw ist results in tra n sla tio n in th e -f S d irectio n . W e express Vi s as 1 Ixs- g 0^ and (a) reduces to
(c)
J Mv G ~ T iS (b)
D eterm in e the v aria tio n o f (j), alo n g th e centerline for th e tw o thin-w alled open sections show n.
Prob. 1 1 -4
lf
d i -*2
Center o f t w is t
iL
, V ,
t d.
2
C e n te r o f tw ist
bn(a)
b/ 2 (b)
1 1 -5 .
Verify th a t the d istrib u tio n , q const, satisfies ft 2 ~ j j cri2 dA dS = 0
^3 = j V 13 dA = qasz dS = 0 M y j ( f x q iy)dS = 2q A cl for the closed cross section sketched. 1 1 -6 . Refer to P rob. 1 1 -4 . T o apply E q u a tio n (c) to th e centerline of a closed cell, we n o te th a t (see (11 50))
P R O B LE M S
325
Then,
Integrating (b) leads to th e d istrib u tio n of (pi A pply (b) to the section shown. Take < /> , 0 at p o in t P. D iscuss th e case w here a = b.
Prob. 1 1 -6
t i 1
a t *
-
11-7. D eterm ine the to rsio n al shear stress d istrib u tio n an d to rsional constant J for the section shown. Specialize for r a. 11-8. D eterm ine the eq u atio n s for Cj (j = 1, 2, 3) an d J for the section shown. G eneralize for a section consisting of n cells. 1 1 -9 . D eterm ine the d istrib u tio n of torsional shear stress, the torsional constant J, and th e d istrib u tio n of th e w arping function for the section shown. Take < fi, = 0 on the sym m etry axis an d use the results presented in P ro b . 11-6. 11-10. Verify E q u atio n (11-82). U tilize (11-15). 11-11. T he flexural w arping function < j)jr satisfy = Xj in A
326
T O R S IO N -F LE X U R E OF PR IS M A T IC M E M B E R S
CHAP. 11
-'7a-
Prob. 1 1 -8
1
t a
/,
-4 ------------------
Prob. 1 1 -9
+5
+S
T a
i
U tilizing th e follow ing in teg ratio n form ula, ( /i , 2 / 2 , 2 + r f-f d x 3 = q>/ , - 2 rfS J on f i V2 / 2 rfx2
f i ,
3/ 2 , 3 ) ^ 2
w here f 2 are a rb itra ry functions, verify E q u atio n (11 -96). 11-12. Refer to Fig. 11-17. S tartin g w ith (1 1-107), derive th e expressions for the co o rd in ates o f th e sh ear center in term s o f th e cross-sectional param eters.
PR O B LE M S
327
1 1 -1 3 . D eterm in e th e flexural shear flow d istrib u tio n s due to F 2, F 3 an d locate the sh ear center for th e live thin-w alled sections shown.
Prob. 11-13
(a)
2/
2/ t t t t
1
t t\ t\ t
1 t
_t_
----------- a -
(c)
(d)
t h
328
C H AP. 11
I I - 1 4 . W e established th e expression fo r the tw ist d efo rm a tio n (E q u atio n (1 1 -3 1 ) by req u irin g the to rsio n al w arp in g fu n ctio n to be co n tin u o u s. O ne can also o b tain this result by ap p lying the p rin cip le o f virtu al forces to the segm ent show n as p a rt o f th e acco m p an y in g figure.
Prob. 11-14
(a)
i
2G{2F.)
*3
[ _^
a
------------X2
G(E) 2GUE)
|
t
t a -----------(c)
P R O B LE M S
329
T h e general principle states th a t (J jc r Act d.A)dx 1 = ( jj A bTu d A ) d x 1 4- |j j A p7 Au dA\Xu x +dx (a)
for a statically perm issible force system . N ow , we select a force system acting on th e end faces w hich is statically eq u iv alen t to only a to rsio n a l m o m en t M v If we consid er the cross section to be rigid , the rig h t-h an d side o f (a) reduces to AMjCo, ! d x j, an d we can w rite
, 1 <i. i = k i = A M , 4
et
A a dA
(b)
N ext, we selcct an a rb itra ry closed curve, S (p art b of figure), an d co n sid er the region defined by S a n d the differential thickness dn. W e specialize the virtu alstress system su ch th a t Aa = 0 o u tsid e this d o m ain an d only Actis is finite inside th e d o m ain. F inally, using (11 -51), we can w rite dn(A(riS) = s an d E q u a tio n (b) reduces to
(d)
(c)
T he d eriv a tio n s p resen ted in th e text arc b ased on a c o n stan t sh ear m o d u lu s G th ro u g h o u t the section, so we replace (d) w ith Gki = ~ (j)<x,s-</S (e)
If G is a variable, say G f G * (w here f /( .v 2. >; ,)), we have to w o rk w ith G k i = 2 , -ArS < t ( 7 - V i s ,s j Also, we define the to rsio n a l co n sta n t J acco rd in g to G * k lJ = M , (g) (0
C o n sid er a thin -w alled section com p risin g d iscrete elem ents h aving different m aterial p ro p erties. D evelop th e expressions for th e to rsio n al an d flexural shear flow d istrib u tio n s acco u n tin g for variab le G a n d E. D eterm in e th e n o rm al stress d istrib u tio n from the stress-strain relation. A ssum e a linear v aria tio n in extensional strain a n d ev alu ate the coefficients o f th e strain ex p ansion from the d efinition e q u a tio n s for F ,, M 2, and M v A pply y o u r fo rm u latio n to the section show n in p a rt c o f th e figure.
12
Engineering Theory of Prismatic Members

12-1. INTRODUCTION
St. V en an ts th e o ry o f flexure-torsion is restricted to th e case w here 1. 2. T h ere arc no surface forces a p p lied to th e cy lin d rical surface. T he en d cross sections can w arp freely.
T h e w arp in g fu n ctio n 0 consists o f a term d u e to flexure (</>/) a n d a te rm d u e to p u re to rsio n (</>,). Since < /> is independent o f a ,, th e lin ear ex p an sio n
1 1 Mz M$ (7u - H - x 3 -----x 2 A h h
,<-> , \ (12- 1)
is the exact s o lu tio n ! for ctu . T h e to ta l sh earin g stress is given by oxs o t -\- Of (12 2)
w here ot is th e p u re-to rsio n d is trib u tio n (due to < /> ,) an d c f rep resen ts the flexural d istrib u tio n (due to (j)/)- W e g enerally d eterm in e 07 by ap p ly in g the engineering th eo ry o f sh ear stress d istrib u tio n , w hich assum es th a t th e cross section is rigid w ith respect to in -p lan e d efo rm a tio n . U sin g (1 2 -1 ) leads to th e follow ing expression for the flexural sh e a r flow (see (11 -1 0 6 )): Q 3 r0.2 c !b <7/t y ~ I ' 2 ~ ~r~ 3 ~ (1 2 -3 )
T he w arp in g function will d ep en d o n x j if forces are ap p lied to th e cy lin d rical surface o r th e ends are restrain ed w ith resp ect to w arping. A te rm d u e to variable w arp in g m u st be ad d ed to the lin ear ex p an sio n fo r rru . T his leads to an a d d itio n a l term in th e expression for the flexural sh ear flow. Since (1 2 -1 )
t A l i n e a r v a r i a t i o n o f n o r m a l s t r e s s is e x a c t f o r a h o m o g e n e o u s b e a m . san d w ich b eam ) are treated by assu m in g a lin ea r v a r ia tio n
C o m p o s i t e b e a m s (e.g .. a and o b ta in in g
in e x t e n s i o n a l s t r a i n 11 - 1 4 and 12-1.
t h e d i s t r i b u t i o n s o f o-u
fro m th e stress-stra in re la tio n .
See P robs.
330
SEC. 1 2 -2 .
F O R C E -E Q U IL IB R IU M E Q U A T IO N S
331
satisfies th e d efin itio n eq u a tio n s for F t , M 2, M 3 identically, th e n o rm a l stress co rrectio n is self-eq u ilib ratin g ; i.e., it is statically eq u iv alen t to zero. Also, th e sh ear flow co rre c tio n is statically eq u iv alen t to only a to rsio n a l m o m en t since (1 2 -3 ) satisfies th e d efinition e q u atio n s for F 2, F 3 identically. In the en g in eerin g th e o ry o f m em bers, we neglect th e effect o f v ariab le w arp in g on th e n o rm a l an d sh earin g stress; i.e., we use th e stress d istrib u tio n p red icted by the St. V e n a n t th eo ry , w hich is b ased o n constant w arp in g an d no w arp in g re s tra in t a t th e ends. In w h at follows, we dev elo p th e g o v ern in g eq u atio n s for the engineerin g th e o ry a n d illu strate the tw o g en eral so lu tio n p ro ced u res. T h is fo rm u latio n is restric ted to th e linear geometric case. In th e next ch ap ter, wre p resent a m o re refined th e o ry w hich acc o u n ts for w arp in g re stra in t, a n d in vestigate th e e rro r in v olved in th e eng in eerin g th eo ry .
1 2 -2 .
F O R C E -E Q U IL IB R IU M E Q U A T IO N S
In th e en g in eerin g theo ry , we tak e the stress resu ltan ts a n d couples referred to the cen tro id as force q u an tities, an d d eterm in e th e stresses using (1 2 -1 ), (1 2 -3 ), a n d th e p u re -to rsio n a l d istrib u tio n d u e to M T. T o estab lish the forceeq u ilib riu m eq u atio n s, we co n sid er th e differential elem ent show n in Fig. 12 -1 . T h e statically eq u iv alen t external force an d m o m en t v ecto rs p er unit
d x\l2 j------- dx
Fig. 12-1. Differential element for equilibrium analysis.
length a lo n g X x a re d en o ted by b, m. S u m m ing forces an d m o m en ts aboiut 0 leads to th e follow ing v ecto r eq u ilib riu m eq u a tio n s (note th a t F _ = F + , M - - M+):
332
ENGINEERING THEORY OF PRISM ATIC M EM BER S
CHAP. 12
We obtain the scalar equilibrium equations by introducing the component expansions and equating the coefficients of the unit vectors to zero. The re sulting system uncouples into four sets of equations that are associated with stretching, flexure in the X r X 2 plane, flexure in the A ',-X 3 plane, and twist.
Stretching
dF t
dx\
Flexure in X , -X-, Plane
------ t- b 2 0
a xi
d F 2
dM, + m3 + F2 - 0 dx [
Flexure in X r X 3 Plane
d F * + d xi d M j + d xi m 2 F 3 =
0
(12-4)
b,
Twist
This uncoupling is characteristic only of prismatic members; the equilibrium equations for an arbitrary curved member are generally coupled, as we shall show in Chapter 15. The flexure equilibrium equations can be reduccd by solving for the shear force in terms of the bending moment, and then substituting in the remaining equations. We list the results below for future reference.
Flexure in X \-X 2 Plane
dM-.
Fn = d xi m 3
d 2M 3 dx i Flexure in X 1-X3 Plane
dm3
+
dx i
-*>3=0
(12-5)
SEC. 1 2 -3 .
F O R C E-D ISPLAC EM EN T R ELATIO N S
333
Note that the shearing force is known once the bending moment variation is determined. The statically equivalent external force and m om ent components acting on the end cross sections are called end forces. We generally use a bar superscript to indicate an end action in this text. Also, we use A, B to denote the negative and positive end points (see Fig. 12-2) and take the positive sense of an end
* 2
Fig. 12-2.
N o t a t io n a n d p o s i t i v e d ir e c t io n f o r e n d f o r c e s .
force to coincide with the corresponding coordinate axis. The end forces are related to the stress resultants and couples by
F Bj =
M Bj pA j
M
a
C ^ j]si = L
\ M j ] X l,-L (/ = 1? 2 , 3 ) (1 2 - 6 )
= =
[ ^ 7 '] - X i = 0
A minus sign is required at A, since it is a negative face.
12-3.
FORCE-DISPLACEMENT RELATIONS; PRINCIPLE OF VIRTUAL FORCES
We started by selecting the stress resultants and stress couples as force parameters. Applying the equilibrium conditions to a differential element re sults in a set of six differential equations relating the six force parameters. To complete the formulation, we m ust select a set of displacement param eters and relate the force and displacement parameters. These equations are generally called force-displacement relations. Since we have six equilibrium equations, we m ust introduce six displacement param eters in order for the formulation to be consistent. Now, the force param eters are actually the statically equivalent forces and moments acting at the centroid. This suggests that we take as displacement
334
EN G IN E ER IN G TH EO R Y OF P R IS M A T IC M E M B E R S
C H AP. 12
param eters the equivalent rigid body translations and rotations of the cross section at the centroid. We define u and d3 as it = YjU f li ~ equivalent rigid body translation vector at the centroid <y =r equivalent rigid body rotation vector By equivalent displacements, we mean JJ (force intensity) (displacement) dA = F u + M - To
A
(12 8)
N ote th at (12-7) corresponds to a linear distribution of displacem ents over the cross section, whereas the actual distribution is nonlinear, owing to shear de formation. In this approach, we are allowing for an average shear deform a tion determ ined such th at the energy is invariant. We establish the force-displacement relations by applying the principle of virtual forces to the differential element shown in Fig. 12-3. The virtual-force
Fig. 12-3.
S ta tic a lly p e r m is s ib le fo r c e s y s te m .
system is statically permissible; that is, it satisfies the one-dim ensional equili brium equations - - ( A F f) = 6 dx 7 - (AM . I + (T, x A F +) = 0
d x i
(a)
Specializing the principle of virtual forces for the one-dim ensional elastic case, we can write dV* d x x = A.P, (b) where dt represents a displacem ent quantity, and P ( - is the external force q u an tity corresponding to d{. The term dV* is the first-order change in the one dim ensional com plem entary energy density due to increm ents in the stress resultants and couples.
SEC. 1 2 -3 .
FO R C E -D IS P LA C E M E N T R ELATIO N S
335
Evaluating the right-hand side of (b), we have AF du rz + AM ax, dco dx{ dx.
(C)
U sing the second equation in (a), (c) takes the form I d, APj = f du _ \ dco AF . ------ 1 - i j x co I + AM + -
\ d x i J d x i
dxx
(d)
Finally, evaluating the products, we obtain AP; = [ A F i + A F 2(1^2 , i W3) -+ AF 1 + (O2) 4- A M ta)lt 1 + A M 20) 2 .1 I" AM$(o 3 ' j J dx\ C ontinuing, we expand dV*: dP* = V ( .
3
(12-9)
fd ? *
A/,- +
J
cV*
3 M ,
AM
7
( 12 - 10)
T he quantities C; and kj are one-dimensional deformation measures. Equating (12-9) and (12-10) leads to the following relation between the deform ation measures and the displacem ents: dV*
t i = h .
k\ = v 77 _ ~ a)i.
d M ;
CO 3
k 2 ~ a M
2
dV *
dV*
e 2 ~ ^ T f
^ 2
U2. 1
( 12 - 11 )
< ?P*
(?P *
<?3
T F * M 3
^3
We see that-
1 . e-i is the average extensional strain. 2 . e2, e 3 are average transverse shear deformations.
3. 4. /cj is a twist deformation. k 2, /C 3 are average bending deform ation measures (relative rotations of the cross section about X 2, X 3).
O nce the form of V * is specified, we can evaluate the partial derivatives. In w hat follows, we suppose th at the m aterial is linearly elastic. We allow for the possibility of an initial extensional strain, but no initial shear strain. The general expression for V* is F* < j? + < 7 i? + (cr2 i 2 4- erf3) dA (a)
where e? denotes the initial extensional strain. Now, V* for unrestrained torsion-flexure is given by (U -9 8 ). Since we are using the engineering theory
336
ENGINEERING THEORY OF PRISMATIC MEMBERS
CHAP. 12
of shear stress distribution, it is inconsistent to retam terms involving in-plane deformation, i.e., v J E . Adding terms due to a s = F J A , ffiC?, and n e g l e c t i n g the coupling between F 2 , F 3 leads to
(12 - 12 )
where
Mf Ml + jF2 X3 F3X2
We take (12-12) as the definition of the one-dimensional linearly elastic com plementary energy density for the e n g i n e e r i n g theory. One can interpret e?, k , k y as weighted or equivalent initial strain measures. Differentiating (12-12) with respect to the stress resultants and couples, and substituting in ( 12 - 1 1 ), we obtain the following force-displacement relations:
To interpret the coupling between the shear and twist deformations, we note (see Fig. 12-4) that
u2 Uj = x 3 cO} X
(a)
2OJ j
displacements due to a rigid body rotation about the Comparing (a) with ( 1 2 - 1 3 ) , we see that the cross section twists about the s h e a r c e n t e r , not the centroid. This result is a consequence of neglect ing the in-plane deformation terms in V*, i.e., of using ( 12 - 12 ). Instead of working with centroidal quantities (M l; u2, m 3), we could have started with M T and the translations of the s h e a r c e n t e r . This presupposes that the cross section rotates about the shear center. Wc replace w 2, ui (see Fig. 1 2 - 4 ) by
ce n tro id a l sh ea r cen ter. 112 11S 2 + < 0 1 -V3
defines the
(12-14)
u3 =
US3 ~
SEC. 12-3.
FORCE-DISPLACEMENT RELATIONS
337
where uS2, uS3 denote the translations of the shear center. The terms involving F2, F 3, Mj in (12-9) transform to
AM]!
1
+ AF2(US2 i ~
W 3)
A F3(U s3
1 +
CO2)
(a)
Then, taking M T as an independent force parameter, we obtain

Mt GJ
F2
< U l, 1
ga2
F 3
US 2 ,
(12-15)
G /4 ,
% 3, 1
Since the section twists about the shear center, it is more convenient to work with M r and the translations of the shear center. Once hS2, w S3, and a?i are
*2
Fig. 12-4.
T r a n s la t io n s of th e c e n tro id a n d th e s h e a r c e n te r.
known, we can determine u2, V3 from (12-14). We list the uncoupled sets of force-displacement relations below for future reference.
Stretching
0 , 1 e, + -g -,
Flexure in X x-X2 Plane GA:

A h W s2 , 1 ' W 3
h i3
338
EN G INEER IN G TH EO R Y OF P R IS M A T IC M E M B E R S
CHAP. 12 (1 2 -1 6 )
Flexure in X 1- X 3 Plane
F ,
GA :
k% +
11S 3, 1 +
0)2
M-, E l,
0) 2 , 1
Twist A bout the Shear Center Mr

= Ct> i . i
GJ
The developm ent presented above is restricted to an elastic material. Now, the principle of virtual forces applies for an arbitrary m aterial. Instead of first specializing it for the elastic case, we could have started with its general form (see (10-94)), j j e r Aw dA (12-17) fv , where e represents the actu a l strain matrix, and Act denotes a system of statically permissible stresses due to the external force system, APr We express the integral as .
f[ eT Act dA (e;A F j + kj A
M j)
(12 - 18 )
and determ ine ej, kj, using Ac as defined by the engineering theory. F o r example, taking AF y AM 2 aM 3 , , Acrt - 1 x 3 - ----- x 2 (a) A Ii leads to
1
<Z\ c k 2 = -
si
x ^ . d A
(b)
k3 =
X 2
Once the extensional strain distribution is known, we can evaluate (b). Using (12-18), the one-dim ensional principle of virtual forces takes the form
U l f e / A F j + kj
A M j)]d xi = >/, A P t
(12 - 19)
The virtual-force system must satisfy the one-dim ensional equilibrium equations (12-4). One should note that (12-19) is applicable for an a rb itra ry material. W hen the m aterial is elastic, the bracketed term is equal to d P * t and we can write it as JXl d P * d x , = Y d t A P, (1 2 - 2 0 )
SEC. 12-4.
S U M M A R Y O F T H E G O V E R N IN G
E Q U A T IO N S
339
The expanded form for the linearly elastic case is
[e
A F
( <
) A
f i
( c a ~ )
A F j
A M
( 12 - 21 )
+ ( k + i f ) A M 1 + i 14 + i f ) A M s
dXl = </, A/5,
We use (12-21) in the force m ethod discussed in Sec. 12-6. 12-4. SUMMARY OF THE GOVERNING EQUATIONS
At this point, we sum m arize the governing equations for the linear engineering theory of prism atic members. We list the equations according to the different modes of deform ation (stretching, flexure, etc.). The boundary conditions reduce to either a force or the corresponding displacem ent is prescribed at each end.
Stretching (F\, u {) F u ! + by =
, + 0
AE ~ >1
( , 2 ~ 22)
F v or u y prescribed at Yt = 0 , L
Flexure in X r X 2 Plane (F2, M 3, u2, 3) F 2 ! + b2 0 M 3t i + m 3 + F2 = 0 _ ~r~7~ ~
0>4 2
2,
i ~ <^3
(12-23)
+ l< = W 3- '
or F 2 prescribed at = 0, L M 3 or co3 prescribed at .x, = 0, L

u 2
Flexure in the
X ,
-X 3 Plane (F3, + b3 = 0 M2 i F ni2

F *
M 2, w3 ,
o j2)
F3
= u 3. I + OJ2
j.
(12-24)
+ k2 = a i>2 , i
E I2 u 3 or F 3 prescribed at x x = 0 , L ft)2 or M
2
prescribed at jc, 0, L
340
CH AP. 12
Twist A bout the Shear Center ( M

M lf M j T i { } T l y
0
T ,
, u 2, u 3)
M r or o)y prescribed at x 1 = 0, L mT m l + b 2 x 3 b 3 x z u 2 x 3 co1

u 3 X
(12-25)
2CO 1
12-5.
DISPLACEMENT METHOD OF SOLUTION-PRISMATIC MEMBER
The displacement m ethod involves integrating the governing differential equations and leads to expressions for the force and displacem ent param eters as functions of jxt . W hen the applied external loads are independent of the displacements, we can integrate the force-equilibrium equations dircctly and then find the displacements from the force-displacemcnt relations. If the applied load depends on the displacements (e.g., a beam on an elastic foundation), we m ust first express the equilibrium equations in terms of the displacem ent param eters. This problem is m ore difficult, since it requires solving a differential equation rather than ju st successive integration. The following examples illus trate the application of the displacem ent m ethod to a prism atic m ember. Example 12-1 ------------------------------------------------------------------------------- --
W e c o n s id e r th e c a s e w h e r e b 2 c o n s t ( F i g . E 1 .2 - 1 ) . T h is lo a d i n g w ill p r o d u c e fle x u r e
in t h e X l - X 2 p l a n e a n d a l s o t w i s t a b o u t t h e s h e a r c e n t e r if t h e s h e a r c e n t e r d o e s n o t lie o n th e X 2 a x is . a p p ly W e s o lv e th e tw o u n c o u p le d p r o b le m s , s u p e r im p o s e th e r e s u lts , a n d th e n
th e b o u n d a r y
c o n d itio n s .
Flexure in X x- X 2 Plane
W e s ta r t w ith th e fo r c e - e q u ilib r iu m e q u a tio n s . F
2.
b 2 ~ F 2
(a)
(b )
t = I n t e g r a t in g (a ), a n d n o t i n g t h a t b 2 =
c o n st, w e h a v e
F 2 )jc, = 0
b 2X i = 0, L :
F o r c o n v e n i e n c e , w e u s e s u b s c r i p t s A , B f o r q u a n t i t i e s a s s o c i a t e d w i t h .v ,
(d)
M 3 =
~ X 1F A2 + j b 2x i
(f)
SEC. 1 2 -5 .
D IS P LA C E M E N T M ETH O D OF SO LU TIO N
341
W e c o n s id e r n e x t th e fo r c e - d is p la c e m e n t r e la tio n s , M 3 <^3, 1 /, F j_ (g )
'2. I =
CO, +
(h)
GA7
I n te g r a tin g (g ) a n d
th e n (h ), w e o b ta in
"3
0>A3 + - = r ( X l M -13 jb l*1

6
lA ^ A 2 + s M l )
( A ^ , l ?2 X i (
2
U2 UA2 + * 1 ^ , 4 3 +
X 1
A
1 2
( i) A 2\ G A 2 E l J 4 3 \ 2 E I 3J V G A 2 E l J
T h e g e n e r a l fle x u a l s o lu tio n (fo r b 2 =
c o n s t ) is g iv e n b y (e ), ( f) , a n d ( i).
Fig. E12-1
* 2
r
X i
Twist About the Shear Center

T h e a p p lie d t o r s i o n a l m o m e n t w i t h r e s p e c t t o t h e s h e a r c e n t e r is mr = b 2x 3
(J )
S u b s t i t u t i n g f o r m T in t h e g o v e r n i n g e q u a t i o n s ,
M j ^
fUjM T
(k)
G J
a n d in te g r a tin g , w e o b ta in M r = M
a t
(1 )
i = +
7 ; 7
(x
at
T h e a d d itio n a l c e n t r o id a l d is p la c e m e n ts d u e to tw is t a r e
l l 2 X^QJi
(m)
342
E N G IN E ER IN G TH EO R Y OF P R IS M A T IC M E M B E R S
C H A P . 12
Cantilever Case
W e s u p p o s e t h a t t h e le ft e n d is f ix e d , a n d t h e r i g h t e n d is fr e e . T h e b o u n d a r y c o n d itio n s
are
UA 2 0) A3 ( 0 A
F B2 M B3 M
bt
(n)
S p e c ia liz in g th e g e n e r a l s o lu tio n fo r th e s e b o u n d a r y c o n d it io n s r e q u ir e s
P A2
^2^
?b2L2
jVjf a n d th e fin a l e x p r e s s io n s r e d u c e to
jt
(o)
b ^ X 'iL t
F 2 =
b 2( L
X j)
M 3 =
b2 (
L x y + \ x \
b 2x 3 ( L
x ,)
1^2
u3 = - x 20 )i
b 2 ' 2 f x [ L 2
(
c a , x 2L 2 + 3 (L ix ,)
i / 2
j l _______ i
\ (p )
i m
x i
) + l 2 b L2
) 2E h )
\ l2 E ~ h
G A 2J
m = e T3 v~2
b 2 x 3x j (Oi = G J
It
is
o f in te r e s t
to =
co m p are
th e
d e fle c tio n s
due
to
b e n d in g
and
sh ear
d e fo r m a tio n .
E v a lu a tin g u 2 a t x ,
L , w e have
^ 8 2 Ibendin
8
E IS
\ h ,L 2
^B2 [shear dcformniion
'y
2
GA2
? * > .
(q)
S
6ft
Cr L,2 A
c o n s id e r v = 0 .3 ( d = d e p th ):
r e c ta n g u la r
cross
s e c tio n
and
is o tr o p ic
m a te r ia l w ith
E
=
2 .6
13
A 2
d2
10
(r)
5 A
!-(
B y d e f i n i t i o n , d j L is s m a l l w i t h
0 2
r e s p e c t t o u n i t y f o r a m e m b e r e le m e n t a n d , t h e r e f o r e , i t is
SEC. 1 2 -5 .
r e a s o n a b le to n e g le c t
D IS P LA C E M E N T M E T H O D OF SO LU T IO N
tr a n s v e r se s h e a r d e fo r m a tio n 0. w ith re s p e c t to
343
b e n d in g d e fo r m a tio n
fo r th e is o t r o p ic c a s e .f F o r m a lly , o n e s e ts l / A 2 =
Fixed-End Case
W e c o n s id e r n e x t th e c a s e w h e r e b o th e n d s a r e fix e d . T h e b o u n d a r y c o n d itio n s a re
UA 2 = OJa3
C O A1
= 0
1 1 b i = (Ob 3 = (Obi = 0
S p e c i a liz in g (h ), (i), a n d ( k ) f o r th is c a s e , w e o b t a in
(s)
b2L F
A2
2
bh2iL 2
A3
12
b 2x 3L
(t)
M j 17- = T h e fin a l e x p r e s s io n s a r e ( L F-,
b2 ( V
X,
Lx i
M 3 = b2 ( ~
\ 1 2
xf
+ T
- 1 - -F
~2~
M r
/?2-^3
\2
+
7
(u)
24E li
a J !X 3
( L x 1 xf) H -------- ( L 2x f 2 L x i + x f )
2G A 2
W3
%2W l
b2 ( L 2
,
E h V 12
Xi - XT H------X ?
4
6 1
,,
= 2 G l (LX l
x .)
Example 12-2 ------------ -----------------------------------------------------------------W e c o n s id e r a fo r c e s a p p lie d T h e and m e m b e r ( F ig . E 1 2 - 2 ) W e r e s tr a in e d at th e le ft e n d , a n d s u b je c te d o n ly to a t th e r ig h t e n d . a llo w and fo r th e p o s s ib ility o f s u p p o r t m o v e m e n t a t A . r o ta tio n s a t B m em ber in te r m s o f th e e n d a c tio n s a t B W e can
e x p r e s s io n s su p p o rt
fo r th e tr a n s la tio n s at A are
m ovem ent
c a lle d
fo r c e - d is p la c e m e n t
r e la tio n s .
o b ta in th e s e r e la tio n s fo r a p r is m a tic m e m b e r b y d ir e c t in te g r a t io n o f th e fo r c e - d is p la c e m e n t
t F o r s h e a r d e fo r m a tio n to b e s ig n ific a n t w ith re s p e c t to b e n d in g d e fo r m a tio n , G / E m u s t b e o f t h e s a m e o r d e r a s l j A sL 2 w h e r e A s is t h e s h e a r a r e a . H o w e v e r , it m a y b e s a tis fie d fo r a s a n d w ic h b e a m T h i s is n o t p o s s i b l e f o r t h e i s o t r o p i c c a s e . See P ro b . 1 2 -1 .
h a v in g a so ft c o re .
344
r e la tio n s . In
EN G INEER IN G TH EO R Y OF P R IS M A TIC M E M B E R S
th e n e x t s e c tio n , w e illu s tr a te an a lte r n a tiv e ap p roach , w h ic h
CHAP. 12
u tiliz e s th e
p r in c ip le o f v ir tu a l fo r c e s .f
Fig. E12-2
* 2
T h e b o u n d a r y c o n d itio n s a t
L are = fi
=
I n te g r a tin g th e fo r c e - e q u ilib r iu m
^B y-
(a)
e q u a tio n s a n d a p p ly in g (a) le a d t o th e f o llo w in g e x p r e s
s io n s fo r th e s tre s s r e s u lta n ts a n d c o u p le s :
Fj =
M r ~
?*j
M m M U1 -
(/' = 1.2,3)
(L -
M 2 -
Xl)FB 3
Mi = M b> ,+
(L
x{)F!)2
U s in g (b ), th e fo r c e - d is p la c e m e n t r e la tio n s ta k e th e fo r m
0>3, I
~ Ljl 3
[^ B 3
(L
u 2. i
0*3
Fb2
jC j yrz M b t
( c)
GA2
ct>2. 1 = ~ r ~ ^ 2 (L
GJ
X i) F fl3]
CjI 2
u 3, i
o j2
- p r r F B3 C r/i 3
-pryM gr iy j
t See Prob. 12-11.
SE C . 1 2 -5 .
D IS P L A C E M E N T M E T H O D O F S O L U T IO N
L , w e o b ta in
345
I n t e g r a t in g (c) a n d s e t t in g x x =
M /ll +
F Bi A L L _ L 1
+ r r y ^ b2
0i B-$ - oj a
3 +
El3
T
+
M b3
2El3
L 2 _
+
L * 3
L
M j 3T
L 3
f l2
UB2 ~
W A2
-L<UA 3 +
^T~ M b 3 2z i 3
L 2 _
0J
,g 3 T
3 i3
L ~ <U^ 2
(d)
+ ,-^ 7 - W b 2 k l 2
~ .~ T ^ B 3 t b l 2 _
,
Mfi 3 l<^3 ~ i^ /12
2
L 2 ii/ 2
L .v 2
- 7 - M ( i
M
G v/
st
+
'C / 4 3 3 /2 /
B3
(Og 1 = co/11
- z r ~ M BT
GJ
F i n a l ly , w e r e p la c e A / BT b y
M a r =
M m
Xs^bz
x 2 & 133
(e)
a n d w r it e th e e q u a t io n s in m a t r ix fo r m :
L fU A E
L 3 3E ll
'8 2
GA~2 +
L x 3x 2
~
g
L x 3 G J 2E h
B2
+ L x j/ G J
7 ~
L +
L s 3 / 2 L x A2 G J I? 2E U
'/S3
LX 3X2 ~
g
G /I3
B3
+ L x jjG J
X3L C Bl
X 2ls G J
/, G J
GJ
L 2 ( B 2 2E l 2
L M <32 E h
L 0>B3 2E J, E I
B3
{ i #,41 > UA2 +
L (0 A3, M,43 L(DA2, C t)A1, C O A2, (1 ) 43}

is c a ll e d th e m e m b e r fle x ib ility m a tr ix and
( 0
T h e
c o e ffic ie n t m a t r ix
is g e n e r a lly d e n o t e d
by fe . W e o b t a i n e x p r e s s i o n s f o r t h e e n d f o r c e s in t e r m s o f t h e e n d d i s p l a c e m e n t s b y i n v e r t i n g f. T h e fin a l r e la t io n s a r e lis t e d b e lo w fo r fu tu r e re fe re n c e :
346
ENGINEERING THEORY OF PR IS M A TIC M EM BER S

_ AE x F b 1 " ^ ( B 1 ~ UA\)
r b2 =
1 2
CHAP. 12
/f / ,H B 2
J J ,
, +
12E JJS 3, ----------------------ÛJb1 , ~ ( A t >
T t L 3
H- 4 l ) ------- J 2 ~ ( < V B 3 E li
1 2 /f
fl3
3
12 E
_
l$ x 2 . 0JA l ) (h )
~~L G J
(U b 3 12 E +
W/(3) H ------ ~ 2 ~ ^ W b 2 L l
----------r l ------ ' c0f!:l ' 1/
A ffll
l
1 2
21 2)
(w
bi
/*x 3 r~3 L/ ( UB2
6 / J x 3 u A2) H -------- ^ 2 I* 6
, + ^
3)
1 2 E I& 2 +
E lfx 2
(W ZJ3 ~
u Az) H ------ 2
(yB2 +
2)
/|
Mb2 ~
, 2
r
6E1*x2 ^------ 7 1 '" ((UBl ~~

T 2 v E li a 2) 7
(4 +
, /J ^ 2 ) 7 com
(2
6 /|
A ffl3 = ------- JJ~(Ud2
6/fS3
uA l) 4-------- -2---- V0)B1 ~ WA l)
+ w h ere
(4 +
E l* a 3) a ) b 3 + x
(2
a 3) />
0)^3
12E I2
2
12 / 3 ~ 7 n m G A 2L 2 _ h _
l - n g a 3l z h
1
11 =
a2
1 + 3
tra n sv e rse sh ear d e fo rm a tio n by s e ttin g
W e ct 2 =
in tr o d u c e CI3
0
th e
a s s u m p tio n
o f
n e g lig ib le
. by a A ?al = -W b i = 1 , 2 , 3)
T h e e n d fo r c e s a t A a n d B a r e r e la te d
M a2 M 2 + E F Bi
M A3 M B3 : s . / f.T 2
/ * B2
(i)
W e lis t o n ly th e e x p r e s s io n s fo r M A 1 , E / f,
A^/12 = ~ V T ~ (UB3 UA 3) + ----- ^ 7 ----(w fll ^ l )

7
SEC. 1 2 -5 .
D IS P LA C E M E N T M ETHOD OF SO LU TIO N
347
Example 12-3
W e c o n s id e r n e x t th e c a s e w h ere th e a p p lie d lo a d s depend on th e d is p la c e m e n ts . T o s i m p l i f y t h e d i s c u s s i o n , w e s u p p o s e t h e s h e a r c e n t e r is o n lo a d e d o n ly in t h e t h e X 2 a x i s a n d t h e m e m b e r is
X r X z p la n e .
T h e m e m b e r w ill e x p e r ie n c e o n ly fle x u r e in t h e
X x~Xz
p la n e u n d e r th e s e c o n d itio n s . T h e g o v e r n in g e q u a tio n s a r e g iv e n b y (12 -2 3 ):
F 2, i F z =
bz
= -
(a) fflj (b )
M 3, i
W3 , 1
7.7
L I 3 F 2 ~ ~ G A 2
(c)
... (a )
U 2, 1 ' A n a lte r n a te fo rm o f ( a ) is
W j
O n ce M i
is k n o w n , w e c a n , u s i n g ( b ) , f i n d F 2 .
N o w , w e s o l v e ( d ) f o r a>3 a n d s u b s t i t u t e i n ( c ) :
Fz &>3 x 2.l ~ G A 2
( 0
Wj,) * u2 , u -pr~ ~ 2, 11 + prr~

G A 2 G A 2 Then,
F 2.1
^2
M 3 =
E l 3 ( 1*2 . 1 1
7T T" ^2
(g )
and
Fz
m 3 ~
L I 3 ( z<2 , m V
b2 j G A 2
fe . 1 ] /
(h )
F in a lly , w e s u b s titu te fo r u 2 a n d th e lo a d te r m s : d\>2 +
in (e ) a n d o b t a i n a f o u r t h - o r d e r d i f f e r e n t i a l e q u a t i o n i n v o l v i n g
d2
b2 / h i 2 , \dx 1 /
J a 'i \ G / 4 2 T h e p r o b le m
r e d u c e s to s o lv in g ( i) a n d s a tis fy in g th e b o u n d a r y c o n d it io n s : F 2 o r z<2 p r e s c r i b e d ) at X! M 3 o r o j3 p r e s c r ib e d )
0
, L
(j)
N e g le c tin g
tra n sv e rse
sh e a r d e fo rm a tio n 0)
s im p lifie s
th e e q u a tio n s
s o m e w h a t.
T h e
re
s u ltin g e q u a tio n s a r e (w e s e t l / G A 2 =
348
E N G IN E E R IN G TH E O R Y OF P R IS M A T IC M E M B E R S
CH AP. 12
A s a n illu s tr a tio n , c o n s id e r th e c a s e o f lin e a r r e s tr a in t a g a in s t tr a n s la t io n o f th e c e n t r o id , e .g ., a beam on a lin e a r ly th e e la s tic fo u n d a tio n . T h e and d is tr ib u te d th e lo a d in g to th e c o n s is ts o f tw o
te r m s , o n e d u e to W e w r ite
a p p lie d
e x te r n a l lo a d in g
o th e r d u e
r e s tr a in t fo rc e .
b 2 =
ku2
( 1)
w h e r e q d e n o te s th e e x te r n a l d is tr ib u te d lo a d a n d W e su pp ose m 3 = = k: = 0 , k is c o n s t a n t , a n d
k is t h e s tiffn e s s f a c t o r f o r t h e r e s t r a in t . is n e g li g i b le .
tr a n s v e r se s h e a r d e fo r m a tio n
S p e c ia liz in g (k ) fo r th is c a s e , w e h a v e (0 3 = M
t ,
u 2, l E I 3u 2, h ! k (m )
Fz d 4u 2 ~r~r d xf F
!!
q (n )
---------- u 2 H E I 3 2 )
--------E h 3
2or
u 2 p r e s c r ib e d
a t x !.= M } o r <u 3 p r e s c r i b e d ] T h e g e n e r a l s o l u t i o n o f (n ) is
0, L
(o )
,p
s in
Ixi + C2 c o s
A x t)
+ e*x\C3 s i n
C 4 c o s A .x ,)
(i)
u2 u 2 p + e~Xxi{Ci sin A x s + C2 cos A x t)
L b: u2 = w2, p (q ) +
where u2,P represents the particular solution due to q. Enforcement of the boundary conditions at x 0, L leads to the equations relating the four integration constants. The function e ~ Xx decays with increasing x , whereas e l x increases with increasing x. For A x > a s 3 , e ~ Xx a s 0 . If the member length L is greater than 2 ( 3 / / ) = 2 L b (we interpret L b as the width of the boundary layer), we can approximate the solution by the following: 0 ^ Xi < Lb:
L
b
<
Xj
< <
Lb
X! < L :
u2
w 2 iP +
e Ax' ( C i s in
Axj
C4
cos
A xt)
The constants (Ct, C2) are determined from the boundary conditions at Xj = 0 and (C3, C4) from the conditions at X i = L . Note that C3 and C 4 must be of order e ~ XL since u2 is finite at X! = L.
Application 1
T h e b o u n d a r y c o n d itio n s a t x t = 0 ( F ig . E 1 2 - 3 A ) a r e
u2 = M 3 =
E h u 2, i i =
0 (n ),
S in c e q is c o n s t a n t , t h e p a r t i c u l a r s o l u t i o n f o llo w s d i r e c t ly f r o m
i.P
q fk
T h e complete solution is
u2 = - (I k e ^ c o s A x i)
SEC. 1 2 -6 .
FORCE M ET H O D OF SO LU TIO N
349
Fig. E 1 2 -3 A
q const
7777777/7777/77/777777777/77/77/7Z///777/7///////7/7777 '
* 2
Application 2
T h e b o u n d a r y c o n d itio n s a t X j U2t i = 0 (F ig . E 1 2 - 3 B ) a r e
F2 =
a n d t h e s o l u t i o n is
- / 3 u 2 ,1 1 1
P/2
.
s in A x t )
Pa
2
.
e
.
+
2k
*x ( c o s a x i
T h e fo u r b a s ic fu n c tio n s e n c o u n te r e d f/ / 1 =
are + s in X x )
e ~ ;-t ( c o s X x
yji 2 =
e ~ )x s in A x
--------- ij/\
2/1
( 1 2 -26 )
1^3
e ~ ;x{cos X x
s in A x )
~ tj/'2
ij/4 =
e ~ AX c o s X x
- -
i/ / ' 3
T h e ir v a lu e s
o v e r th e r a n g e fr o m
Xx
0 to
Xx
5 are
p re se n te d
in T a b l e
12 -1.
Fig. E12-3B
P
_________________ I
/ / / / } / m / / / / / / f m Z / / 7 7 / / / / ) j 7 T 7 7 / / / / / / 7 7 7 / / f / / / / ) / / / ~ Xx
* 2
12 6.
FORCE METHOD OF SOLUTION
In the force m ethod, we apply the principle of virtual forces to determ ine the displacem ent at a point and also to establish the equations relating the force redundants for a statically indeterm inate member. We start with the one dim ensional form of the principle of virtual forces developed in Sec. 12-3 (see E quation 12-19):
J . . E M F , + kj
AM j) ] d x | = < f , AP,
(a )
350
CHAP. 12
Table 12-1
N u m e r ic a l V a lu e s o f th e F u n c tio n s
Xx
I't
< A 4
AX
0.0 0.2
0.4
1.000
0.965 0.878 0.763 0.635 0.508 0.390 0.285 0.196 0.123 0.067 0.024 - 0 .0 0 6 - 0 .0 2 5 - 0 .0 3 7 - 0 .0 4 2 - 0 .0 4 3 - 0 .0 4 1 - 0 .0 3 7 - 0 .0 3 1 - 0 .0 2 6 - 0.020 - 0 .0 1 6 - 0.011 - 0 .0 0 8 - 0 .0 0 5
0.000
0.163 0.26 i 0.310 0.322 0.310 0.281 0.243
1.000
0.640 0.356 0.143 - 0 .0 0 9 - 0.111 - 0 .1 7 2 - 0.201 - 0 .2 0 8 - 0 .1 9 9 - 0 .1 7 9 - 0 .1 5 5 - 0 .1 2 8 - 0.102 - 0 .0 7 8 - 0 .0 5 6 - 0 .0 3 8 - 0 .0 2 4 - 0.012 - 0 .0 0 4
1.000
0.802 0.617 0.453 0.313 0.199 0.109 0.042 - 0 .0 0 6 - 0 .0 3 8 - 0 .0 5 6 - 0 .0 6 5 - 0 .0 6 7 - 0 .0 6 4 - 0 .0 5 7 - 0 .0 4 9 - 0 .0 4 1 - 0 .0 3 2 - 0 .0 2 4 - 0 .0 1 8 - 0.012 - 0 .0 0 7 - 0 .0 0 4 - 0.001
0 0.2
0.4
0.6 0.8 1.0 1.2

1.4
0.6 0.8 1.0 1.2

1.4
1.6 1.8 2.0 2.2

2.4
0.202
0.161 0.123 0.090 0.061 0.038
1.6 1.8 2.0 2.2

2.4
2.6 2.8
3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8 5.0
0.020
0.007 - 0.002 - 0 .0 0 9 - 0.012 - 0 .0 1 4 - 0 .0 1 4 - 0 .0 1 3 - 0.012 - 0.010 - 0 .0 0 8 - 0 .0 0 7
2.6 2.8
3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8 5.0
0.002
0.006 0.008 0.009 0.009 0.008
0.001 0.002
where
are the actual one-dimensional deformation measures; i represents a displacement quantity; AP -, is an external virtual force applied in the direction of d
eh kj
The relations between the deformation measures and the internal forces depend on the material properties and the assumed stress expansions. The appropriate relations for the linear elastic engineering theory are given by (12-13). If a displacement is prescribed, the corresponding force is actually a reaction. We use 2k, ARk to denote a prescribed displacement and the corresponding reaction increment, and write (a) as ta C lto
AFJ + kJ A M jfld x y
dk
ARk = I
d,
APi
(12-27)
where d { represents an unknown displacement quantity. To determine the displacement at some point, say Q, in the direction defined by the unit vector t Q, we apply a virtual force APQ iq, and generate the necessary internal forces and reactions required for equilibrium using the one-dimensional force-equilibrium equations. We express the required virtual-force system as
A F i =
AM j
f l q a p q
M jr q A P q AR k = R ktQAPQ
(12-28)
SEC. 12-6.
351
Introducing (12-28) in (12-27) and canceling APQ leads to

dQ
~ lA k , q
+ J *,
+ kfM j'Jjfldxi
(1 2 -2 9 )
This expression is applicable for an arbitrary material, but is restricted to the linear geometric case. Since the only requirement on the virtual force system is that it be statically permissible, one can always work with a statically deter minate virtual force system. The expanded form of (12-29) for the linearly elastic case follows from ( 12 - 21 ):
d ,i ~ Y jR k ,Q (l k +
F i
GA
F 2,Q
i / : V
z.Q +
Mr
+ M
G A i J ~ 3' Q '
( !^3 + r r
GJ
T,Q
(12-30)
+I * + w ,m where
KhJ
I M 3 , <2
El
V 'j C ? U
1
k3 = h
.\2i dA
Finally, we can express (12-29) for the elastic case in terms of V*:
p *
dQ -
.R k (12 -3 1)
This form follows from (12-20) and applies for an arbitrary elastic material.
Example 1 2 - 4 ---------------------------------------------------------------------------------
YV c c o n s id e r th e c h a n n e l m e m b e r s h o w n in F ig . F .1 2 - 4 A . i s l i n e a r l y e l a s t i c a n d t h a t t h e r e is n o s u p p o r t m o v e m e n t . W e s u p p o s e th a t th e m a te r ia l W e w ill d e t e r m in e th e v e r t ic a l
Fig. E12-4A
* 2
Centroid
r iO
X,
Vt
V (u /
m Shear center
j e j
352
ENGINEERING THEORY OF PRISMATIC M EM BERS
CHAP. 12
d is p la c e m e n t o f th e w e b a t p o in t Q d u e t o 1. 2. th e c o n c e n t r a te d fo r c e P a te m p e r a t u r e in c r e a s e A T , g iv e n b y
A T = czjXx + a z X i X 2 + ci^XyX^
Force System Due to

A p p ly in g th e e q u ilib r iu m
P in F i g . E l 2 4 B le a d s to
c o n d it io n s to th e s e g m e n t s h o w n
F2 = M
t
- P
+Pe -P (L x j
0
M 3 =
F , =
(a)
== M 2 =
Fig. E12-4B
L -x,
AU
( a
w hen d o w n w a r d , i.e ., in fo rc e at Q . th e u n it d o w n w a r d The r e q u ir e d
j
'
S h e a r c e n t e r a x is
M
Pe
Virtual-Force System
W e we ta k e dQ p o s itiv e a ~ X 2 d ir e c tio n . in te r n a l T o be c o n s is te n t, fo llo w fro m m u s t a p p ly fo rc e s
F ig . E 1 2 - 4 C :
.a M j,
q
=* = e
'
> ,q
(f
- ,) M 2, j
(b )
F;
O O
I I
F M
j. q j. q
(./ =
1, 2
, 3)
(c )
L _ x __ L
2
II
Fig. E12-4C
L I
(
M T tQ
1
e
Shear cen ter axis /
1
fi/il
SEC. 1 2 -6 .
FORCE M ETHOD OF SOLUTION
353
Initial Deformations
T h e i n i t i a l e x t e n s i o n a l s t r a i n d u e t o t h e t e m p e r a t u r e i n c r e a s e is e? = a A T a { a xX i + a 2x t x 2 + X 1X 3 ) (d )
0 3
T h e e q u iv a le n t o n e - d im e n s io n a l in itia l d e fo r m a t io n s a r e
g 'l dA = afljXi
~A
1
.x 3 c ? r f/ l =
t x a 3x i
(e)
2
x 2e d A a a 2x 1
D e t e r m i n a t i o n o f clQ S u b s t i t u t i n g f o r t h e f o r c e s a n d in it ia l d e f o r m a t i o n s in ( .1 2 - 3 0 ) , w e o b t a i n
[L/2 r p
^ " J o [
g
Pei
a a 2x i + *3 (f) X [) G J +
A 2 +
f L e 2L 5 L 3 1 x a 2L * P \ ------ - f - --------- - |- V 4 [2G A 2 2 GJ 48 F J }j 48
Example 12-5
W hen th e m a te r ia l is n o n l i n e a r , w e m u st u se (1 2 - 29) ra th e r th a n (12 -3 0 ). T o illu s tr a te th e n o n lin e a r e a s e , w e d e te r m in e th e v e r tic a l d is p la c e m e n t d u e to P a t th e r ig h t e n d
Fig. E12-5
* 2
X2
I'
shear c e n te r )
o f th e m e m b e r s h o w n
in
F ig . E 1 2 - 5 .
W e
su p p o se
th a t tr a n s v e r s e s h e a r d e fo r m a tio n
is
n e g lig ib le , a n d ta k e th e r e la tio n
b e tw e e n k 3 an d M 3 a s
k s = a , M 3 + a 3M ]
(a)
354
E N G IN E E R IN G TH E O R Y OF P R IS M A T IC M E M B E R S
q
C H A P . 12
N o tin g th a t o n ly f 2, d.Q r e d u c e s t o
a n d M 3rQ a r e f in it e , a n d l e t t i n g e 2 ^ d Q J o k 3M 3 ig d x 1
0, th e g e n e r a l e x p r e s s io n fo r
(b )
N ow , M 3 M 3, T hen. k3 = - P a ^ L x t) P 3 a 3( L x ,)3 (d )
q
- P ( L (L
X l) X j) (c )
S u b s t it u t in g f o r k 3 in (b ) , w e o b t a in
,
d-Q
J j P cii
, + P a3
L 5
W e describe next the application of the principle of virtual forces in the analysis of a statically indeterm inate m ember. W c suppose th a t the m em ber is statically indeterm inate to the rth degree. T he first step involves selecting r force quantities, Z i, Z 2, . . . , Z , . . These quantities m ay be either internal forces or reactions, and are generally called force redundants. U sing the force-equilibrium equations, we express the internal forces and reactions in term s of the prescribed external forces and the force redundants.
F ) = F ). o + i F j . kZ k fc - i
M j = Mj, o +
k 1
M j 'kZ k
(12-32)
R i
R i, o
R i , kZ k k - 1
The m em ber corresponding to Z t = Z 2 = = Z r = 0 is conventionally called the prim ary structure. N ote th a t all the force analyses are carried ou t on the primary structure. T he set ( F Jt 0, M ;-, 0, R i , o) represents the internal forces and reactions for the prim ary structure due to the prescribed external forces. Also, ( F j t k , M j > k , R i t k ) represents the forces an d reactions for the prim ary structure due to a unit value of Z k. O ne m ust select the force resultants such th a t the resulting prim ary structure is stable. O nce the force redundants are know n, we can find the total forces from (12-32). It rem ains to establish a system of r equations relating the force redundants. W ith this objective, we consider the virtual-forcc system consisting of A Zft and the corresponding internal forces and reactions, A Fj = Fjt k AZ k AM ; = M j i k A Z k
A Rf = R i tk A Z k
SEC. 1 2 -6 .
FO R C E M ETH O D OF SO LU T IO N
355
This system is statically permissible. Substituting (a) in (12-27), and noting th a t APi = 0 , we obtain
1*1
r Z
( ej F j. k +
k j M .uk) \ d x
E tliR i,
(12-33)
Taking k = 1, 2, . . . , r results in a set of r equations relating the actual de form ations. O ne can interpret these equations as compatibility conditions, since they represent restrictions on the deform ations. To proceed further, we m ust express the deform ations in term s of Fj, M y In what follows, we suppose th at the m aterial is linearly elastic. The com patibility conditions for the linearly elastic case are given by
fl? + F > -* +(^) +

21)
+ [k2 + M i -
{ { )
M 3, k
+
d x i
(w ) M t
=
M 2, k
(12-34)
+ \
k = 1, 2,. . . , r A m ore com pact form, which is valid for an arb itrary elastic m aterial, is
dV* <1R-
(k = 1 , 2 , . . . , r)
(12-35)
T he final step involves substituting for Fy M j using (12-32). W e w rite the resulting equations as
^
f kjZ j
A fe
( /c
l , Z . . . , r )
(12-36)
j= i
where
/ = /> =
j
Jxi
+
|
1
* +
6T1F> -,F2k + cklF>F31

[ -g j- M <2 2, j M 2. k + M 3'j M 3 jk
T, j M
Tk
dxl
^ 3 ,0 N
f F
^k
Z d iR i, k
+
'
XI
. *
+
,G /4U ,
3.k
'M r , qN
GJ
3, o '
3, k
dx1
The various term s in (12-36) have geom etrical significance. U sing (12-30), we see th a t f jk is the displacem ent of the prim ary structure in the direction of Z j due to a unit value of Z k. 1 Since f jk f kj, it is also equal to the displacem ent in the direction of Z k due to a unit value o f Zj. G eneralizing this result, we can write (<*,),, . , 1 = ( d , ) Pi, 1 (1 2 -3 7 )
356
C H AP. 12
where i, j are arbitrary points, and Pn corresponds to d, i.e., i has the same direction and sense. E quation (12-37) is called Maxwells law o f reciprocal deflections, and follows directly from (12-30). The term Afc is the actual dis placem ent of the point of application of Z k, minus the displacem ent of the prim ary structure in the direction of Z,c due to support movem ent, initial strain, and the prescribed external forces. If we take Z k as an internal force quantity (stress resultant or stress couple), Afc represents a relative displacem ent (trans lation or rotation) of adjacent cross sections. One can interpret (12-36) as a superposition of the displacem ents due to the various effects. They are generally called superposition equations in elem entary texts.f If the m aterial is physically nonlinear, (12-36) are not applicable, and one m ust start with (12-33). The approach is basically the same as for the linear case. However, the final equations will be nonlinear. The following examples illustrate some of the details involved in applying the force m ethod to statically indeterm inate prism atic members.
Example 12-6
T h is lo a d in g ( F ig . E 1 2 - 6 A ) w ill p rod u cc fle x u r e fin ite . in th e X {- X z p la n e and tw is t about t h e s h e a r c e n t e r ; i.e ., o n l y fir s t d e g r e e .
F2,
and
M Ta re
T h e m e m b e r is i n d e t e r m i n a t e t o t h e
W e w ill ta k e th e r e a c t io n a t B a s th e fo r c e r e d u n d a n t.
Fig. E12-6A
* 2
< 7
A 'i
/ u
\ S h ear
c e n te r
/7777
H
Prim ary Structure
O n e can s e le c t th e p o s it iv e se n s e o f th e r e a c tio n s a r b itr a r ily . (S e e F ig . F J 2 - 6 B .) W e w o r k w ith th e tw is tin g m o m e n t w ith r e s p c c t to th e s h e a r c e n te r. to th e in te r n a l fo rc e s b y T h e r e a c tio n s a r e r e la te d
R ,
Z ,
R 2 ----- [T72] Y, = o
^3 ~ ~
- 0
R4
+ [ M r ] x ,-o
t See, for example, Art. 13-2 in Ref. 3.
SEC. 1 2 -6 .
FORCE M ETH O D OF S O LU TIO N
357 Fig. E 12-6B
X-,
* 2
0 ^ 4 ,^ 4 R i,d 2
. -X l x 3
\ z
I '/
:
Z, = 0
0
I
17 / 7 7
M 7
t *!.?!
Force S ystem Due to Prescribed E xternal Forces (Fj 0, M : 0, R t 0)
Fig. E12
< / ' /l/3 -0 ^
B
T//,o F 2.0 T
\
Shear center ax
qe
- L - x r
f 2 , 0
~ q(L
-
x ,)
* 1.0 ^ 2,0
= =
M r , 0
X j)
tfL & '
(b)
3 ,0
x i)2
R- 3. 0
1.0
- ^ 3 ,0
^ 2 , 0 0
^4, 0
qpL
Force S ystem Due to Z x + 1(FJ- J, M j 1, i?; t)

Fig. E12-6D
M 3J
(
Shear ccnter axis
X Xj
M r,i
F2,1
358
E N G IN E ER IN G T H E O R Y OF P R IS M A T IC M E M B E R S
^ 2,1 M r> 1 m 3>1 f i . i = = = =
+ 1
C H AP. 12
= ^ 2, 1 x t) =
0
+
- 1
e + (L
^3. 1 = 4 .l ==
- L e
(c)
F 3 , 1 == M 2 ; !
Equation f o r Z ,
We suppose that the member is linearly elastic. Specializing (12-36) for this problem,
/ n Z i = A i
f n
GA
i= 1
{ F 2, i ) 2 +
7 7 7
GJ
( M t , 1)2 +
- - { M E h
3,
j)2
d x y
(d )
At = I M i. 1 -
I F i. (j A.2
r ~7
qF
2, 1
1 GJ
r ; ~ M T. o ^ r , 1 +
/ \
( &3 H
Af, 0'\ r~ r ~ ) A / Eh j
dx 1
and then substituting for the forces and evaluating the resulting integrals, we obtain =
J u
L
G A
Le2
2
+ G J +
L3
3 E h
^ ^ 2 /^ ? 3
,
A j =
(I 2
C?^?4 -j- - -----
_L
G/l2 4 GJ + 4E/3
e i _ L
(e )
/ <3
(L
x ^ d x i
The value of Z x for no initial strain or support movement is

4i B / J L + a
Z\ qL 8 *
1
G L l ,L 2 3E ( h G \ A 2L 2
JL 2
( 0
...
JL 2
/Vrtfl/ Forces
The total forces are obtained by superimposing the forces due to the prescribed external system and the redundants:
F2 F2.0 + Z[F2 t1 = q(L Xj) + Z]

M T qe{L x t) eZi
M 3 = ~ \ ( E - Xl)2 + ( L - x i )ZI
R i
^ 2
= =
Z t
q
(g)
z t
* 3
/ Z ,
/ ? 4=
efoL - Z,)
SEC. 1 2 -6 .
FORCE M ETHO D OF SO LUTION
359
Example 1 2 -7
T h is lo a d i n g ( F ig . E 1 2 - 7 A ) w ill p r o d u c e o n ly fle x u r e in th e X r X 2 p la n e . t h e m a t e r i a l is p h y s i c a l l y n o n lin ea r a n d ta k e th e e x p r e s s io n k.^ -f3
W e su pp ose
fo r k 3 a s (a)
ctjM y
T o s im p lify th e a n a ly s is , w e n e g le c t tr a n s v e r s e s h e a r d e fo r m a tio n .
Fig. E 12-7A
x , j ,
S h ea r c e n te r
/////
----------------------------------- L ---------------------------------------
Primary Structure
R{ = Zy R 2 = - ( F 2)X l = o
Ri ~ ~(^3)xi =Q
(b)
Fig. E 12-7B
* 2
Force System Due to Prescribed External Forces (see Example 12 -6 ) F2, o = ~ci(L - *i)
360
ENGINEERING THEORY OF PRISMATIC M EM BERS
CHAP. 12
to Z x + l(see Example 12 6 ) Fz.i = + 1 Ri.t = + 1 R3 , 1 = - L C om patibility Equation

S in c e t h e m a t e r ia l is n o n l in e a r , w e m u s t u s e ( 1 2 - 3 3 ) . N e g le c tin g th e tr a n s v e r se sh e a r
1 = L xt
R2 . 1 =
- 1
(d)
d e f o r m a t i o n t e r m ( e 2) , t h e c o m p a t i b i l i t y c o n d i t i o n r e d u c e s t o
I0
W e s u b s t it u t e fo r k 3 u s in g (a ):
M * 3 . i d x i =
d iR i. i
(e)
(a i M 3 + N ow ,
a 3 M 3) M 3 j d x 1 =
] T ^, -R^ t
k 3 M 3 1 r/.Xj
( 0
M 3 = M 3i0 + Z ,M 3. i
(g) ~ 2 ^ ~ x ^2 + ^ i ~ A' i )
I n t r o d u c i n g (g ) in (f) , w e o b t a i n th e f o l l o w i n g c u b i c e q u a t i o n f o r Z i :
Z?
3 lJ \ 4.
r j + Zi
| /
j
q L A (
1V 3
a 3q 2 L 4 \
28
~
;
~ 'i. fc j(L X jW
jci
= ^ j c/j + J -f- d 1 d 2 L i i 3 V
F o r th e p h y s ic a lly lin e a r c a s e ,
(h )
(0
a n d (h ) r e d u c e s to 3 3E I 3
= q
i t
d2 L d3
/c(L -
v jr fx j
(j)
Example 12-8
T h e m e m b e r s h o w n ( F i g . E 1 2 - 8 A ) is f i x e d a t b o t h e n d s . W e c o n s id e r th e c a s e w h e r e th e W e ta k e m a t e r i a l is l i n e a r l y e la s t ic , a n d t h e r e a r e n o s u p p o r t m o v e m e n t s o r i n it ia l s t r a in s . th e e n d a c tio n s a t B r e fe rr e d to th e s h e a r c e n te r a s th e fo r c e r e d u n d a n ts .
Z j = F nl
Z i =* $ { i 3 M
jb
(a)
Z 3
T h e f o r c e s a c t i n g o n t h e p r i m a r y s t r u c t u r e a r e s h o w n in F ig . E 1 2 - 8 B .
Initial Force System

Et,
0=
M i . o = P{a M T_0 P x 3
X i)
(b)
SEC. 1 2 -6 .
FORCE M ETHOD OF SOLUTION
361
Fig. E12-8A
Fig. E12-8B
* 2
Fig. E12-8C
M Tfi
j,P
Shear c en ter axis / Px
I'la
a x x ------ 4------------- b
362
ENGINEERING THEORY OF PR IS M A TIC M EM BER S
CHAP. 12
= +1
Fig. E12-8D
^ r ,i
t1 \
S h ear c e n te r a x is
- L - X i
f" 2, 1 ~
+ 1
^ M r , j =
3.1 0
.Y| (c)
Z2 = +1
Fig. E12-8E
t ,2
^ 2,2
c l
L X]
(d )
Z 3 = +1
Fig. E12-8F
^ 3 ,3
M r>
J
L - x i
Shear center axis
M r, 3 =
2, 3
,3
(e)
SEC. 1 2 -6 .
FORCE M ETHOD OF SO LUTION
363
Com patibility Equations

T h e c o m p a t ib ilit y e q u a t io n s fo r th is p r o b le m h a v e th e fo rm
X j- i
1
f kJZ j
A k
(k =
1, 2 , 3)
fk j
",
I r / ij
F 2 , j F 2, k + 7T7 M
W f 2. 0 S
t,jM
r . k + -ZT7h !
j ' W3, k
d .X i
( 0
Afc -
G / l;. .
) Fi + (*c /) W r'* + (% f) M s'*.
dxx
S u b s t itu t in g fo r th e v a r io u s fo r c e s a n d e v a lu a t in g th e r e s u ltin g in te g r a ls le a d lo w in g e q u a tio n s : / L I \ G / 4h \ + 3E l J L 2 Z i 2EJ

1
to
th e fo l
f Z l +
L 2 Z 2 = - P
a 1 / a 3 a 2b ------------ 1 ---------- 1 j. G A 2 E h \ 3 2
\ 2 E h L \ ------- Z , V / .,; L
I
(g )
= -
----------------2 e j 3
iz , -
G ./
F in a lly , s o lv in g (g ), w e o b ta in
I +
1
26 L
1
a L G A 2 ----------------- i
1 2
/J/j
L 2GA~2 _ (h )
x Z , a 2b P r r
1
+ C lL G A i Z > = 12E J 3 +
l
P a x, -
2g
Application
Sup pose W e can th e m e m b e r is s u b je c t e d to th e d is tr ib u te d lo a d in g sh ow n in F ig . E 1 2 - 8 G .
d e te r m in e th e fo r c e
re d u n d a n ts b y s u b s titu tin g fo r P , a, a n d P a = b L q dx Xj Xy
b in (h ),
1 ( i)
a n d in te g r a t in g th e r e s u lt in g e x p r e s s io n s .
T h e g e n e r a l s o l u t i o n is
364
w h ere
CHAP.
12
C 1+
A s an illu s tr a tio n , w e
12
E I3
L 2G A 2 c o n s id e r th e case w h ere q is c o n sta n t. T a k in g q = con st in ( j) ,
w e o b ta in
Z\
Z , =
qL
2 (k)
12
x^qL
F ig . E 1 2 - 8 G
f
X -, Xi
REFERENCES
1. 2.
T im o s h e n k o , S.
1946.
J.: A d v a n c e d S t r e n g t h
o f M a ter ia ls, V a n N o s tr a n d , N e w Y o r k , 19 4 1.
H e te n y i, M . : B e a m N o rris ,
C . H .,
s on E la s tic F o u n d a tio n , U n iv e r s ity o f M ic h ig a n P re ss, A n n A r b o r ,
3. 4.
5.
6
and
J.
B. W i l b u r :
E lem en ta ry
S tr u c tu r a l A n a ly sis,
M c G r a w - H ill,
N e w Y o r k , 1960 .
A s p l u n d , S. O . : S t r u c t u r a l
196 6 .
M e c h a n i c s : C la s s ic a l a n d M a t r i x M e t h o d s , P r e n tic e -H a ll,
D en H a rto g , O den, G ere,

N ew J.
J. P . : A d v a n c e d S tre n g th o f M a te r ia ls , M c G r a w - H ill, N e w Y o r k , 19 5 2 . Y o rk , 1967. 1965.
T. :
M e c h a n ic s o f E la s tic S tru c tu r e s, M c G r a w - H ill, N e w
7.
8
J. M . a n d
W e a v e r, W .:
A n a ly sis o f F r a m e d S tr u c tu r e s, V a n
N o s tra n d ,
M a rtin ,
H . C . : I n tr o d u c tio n to M a t r i x M e t h o d s o f S t r u c t u r a l A n a l y s i s , M c G r a w - H ill,
Y o r k , 196 6 .
PROBLEMS 12-1. The accompanying sketch shows a sandwich beam consisting of a core and symmetrical face plates. The distribution of norm al stress over the depth is determined by assuming a linear variation for the extensional strain:
PR O BLEM S
l C n = = ~ x 2k 3 E si
365
(a)
We relate k 3 to M 3 by substituting for erlx in the definition equation for M 3:
M3 = jj x2ffn
(b)
M 3 = {EcI3
c
+ ^ 3 .^ 3
To simplify the notation, we drop the subscript and write (b) as M - ( /)equjv/c3 where ( i)equiv is the equivalent homogeneous flexural rigidity.
Prob. 12-1
* 2
(c)
Face
ff i T ft/2
O'12 A *
i ) 3 l . t .
/
C ore
hi 2 \
Face
tf
The shearing stress distribution is determined by applying the engineering theory developed in Sec. 11-7. Integrating the axial force-cquilibrium equation over the area A* and assuming a 12 is constant over the width, we obtain
H ( ^ ll, 1 + @ 21 , 2 + o 31i 3 )d A =
0
A*
V
id)
b< J12 = Jj- ffu. I dA

A*
Then, substituting for a llt a n = - ( Ek 3 )x 2 =

m
i - E x z)
eq ui v
(e)
and noting that F 2 M 3i j, (d) becomes

F i
^ 1 2
b(EI)t eqmv
JJ
r r
Xt E dA
(f)
366
ENGINEERING THEORY OF P R IS M A TIC M EM BER S
CHAP. 12
(a) (b)
Apply Equations (e) and (f) to the given section. The flange thickness is small with respect to the core depth for a typical beam. Also, the core material is relatively soft, i.e., Ec and Gc are small with respect to E f . Specialize part a for Ec = 0 and tf /h 1. Also determine the equivalent shear rigidity {GAz\ aviv, which is defined as
_ 2 1 c -2
12
2G
aA =
, ,
>
2 (GA 2)0 F oquiv
(c)
The member force-deformation relations are

F i ( G A 2 )cq u iv m
3
(EI) eqtuv
Refer to Example 12-1. Specialize Equation (q) for this section and discuss when transverse shear deformation has to be considered. 12-2. Using the displacement method, determine the complete solution for the problem presented in the accompanying sketch. Comment on the influence of transverse shear deformation.
Prob. 12-2
q = const
12-3. F or the problem sketched, determine the complete solution by the displacement method. 12-4. Determine the solution for the cases sketched. Express the solution in terms of the ij/ functions defined by (12-26).
PROBLEM S
367
Prob. 1 2 -3
*2
Prob. 12-4
Y /777777//777777777/77777777777/7/7//77777
(a)
(I
777777777777777)777777777777777777777777777777777 (b)
77777777777777777777777777777777777777777T7777777 (c) 12-5. The formulation for the beam on an elastic foundation is based on a continuous distribution of stiffness; i.e., we wrote
b2 ku2
(a)
Note that k has units of force/(length)2. We can apply it to the system of discrete restraints diagrammed in part a of the accompanying sketch, provided that restraint spacing c is small in
368
ENGINEERING THEORY OF PRISM ATIC M EM BER S
CHAP. 12
comparison to characteristic length (boundary layer) Lb, which we have taken as

3
U
( k/4E l) .1/4
(b)
A reasonable upper limit on c is c <
hi
15
(c)
Letting kd denote the discrete stiffness, we determine the equivalent distributed stiffness k from k = kd/c (d) Evaluate Lb with (b), and then check c with (c).
Prob. 12-5
I
(a)
/ / / / # / / / / / / / / / / ! '/ / / / / / / / / ) / / / / / / / / / / / ? / / / ? /
7 7 /7
777/
7777
7777
a /2
I i i
I i c ' i
'L.E.r,
a/2
7777
7777
7777
7777
7777
<b) Consider the beam of part b, supported by cross members which are fixed at their ends. Following the approach outlined above, determine the distribu tion of force applied to the cross members due to the concentrated load, P.
PR O BLEM S
369
Evaluate this distribution for a 24 ft L = 64 ft c = 1 ft l t = 1(
12-6. Refer to Example 12-3. The governing equation for a prism atic beam on a linearly elastic foundation with transverse shear deform ation in cluded is obtained by setting b 2 q - ku 2 in (i). F o r convenience, we drop the subscripts: d4u d x4 We let
k
e i
k d2u GA d x 2
k E l1
=
4 4 4
1 /
E I\f
k E C
dm \ dx J
, -
d1 ( 0 dx2 ^
q \ GA)

and (a) takes the form d*u

S ? ~ 4 a
, d2u
+
,,.
" = (e>
N ote that ^ is dimensionless and X has units of 1/length. The homogeneous solution is u e~ax(Ci cos bx + C 2 sin bx) + e +"*(C3 cos bx + C 4 sin bx ) where
a
b
(d)
=
= m
1 + c ) l/2
a
1/2
To specialize (d) for negligible transverse shear deformation, we set = 0. (a) Determine the expression for the boundary layer length (e -3 0). (b) Determine the solution for the loading shown. Assume L large with respect to L h. The boundary conditions at x = 0 are
co =
0
Investigate the variation of M m ax and umax with from 0 to 1 .
Consider to vary
Prob. 12-6
777777777777777777777777)V
______________I
/ // / // / // / // / // / / // / /
370
CHAP. 12
12-7. Refer to the sketch for Prob. 12-3. Determine the reaction R and centroidal displacements at = L/2 due to a concentrated force Pi2 applied to the web at x, = * L/2. Employ the force method. 12-8. Refer to Example 12-7. Assuming Equation (h) is solved for Z ls discuss how you would determine the translation u2 at x x L/ 2 . 12-9. Consider the four-span beam shown. Assume linearly elastic be havior, the shear center coincides with the centroid, and planar loading. (a) Compare the following choices for the force redundants with respect to computational effort: 1 . reactions at the interior supports 2 . bending moments at the interior supports (b) Discuss how you would employ Maxwells law of reciprocal deflections to generate influence lines for the redundants due to a concentrated force moving from left to right.
Prob. 12-9
12-10. Consider a linearly elastic member fixed at both ends and subjected to a temperature increase
Determine the end actions and displacements (translations and rotations) at mid-span. 12-11. Consider a linearly elastic member fixed at the left end (A) and subjected to forces acting at the right end (B) and support movement at A. Determine the expressions for the displacements at B in terms of the support movement at A and end forces at B with the force method. Compare this approach with that followed in Example 12-2.
13

13-1. INTRODUCTION
The engineering theory of prismatic members developed in Chapter 12 is based on the assumption that the effect of variable warping of the cross section on the normal and shearing stresses is negligible, i.e., the stress distributions predicted by the St. Venant theory, which is valid only for constant warping and no warping restraint at the ends, are used. We also assume the cross section is rigid with respect to in-plane deformation. This leads to the result that the cross section twists about the shear center, a fixed point in the cross section. Torsion and flexure are uncoupled when one works with the torsional moment about the shear center rather than the centroid. The complete set of governing equations for the engineering theory are summarized in Sec. 12-4. Variable warpmg or warping restraint at the ends of the member leads to additional normal and shearing stresses. Since the St. Venant normal stress distribution satisfies the definition equations for F\, M 2, Mj, identically, the additional normal stress, ar, must be statically equivalent to zero, i.e., it must satisfy JjVn dA = JJje2< rr u dA - JJx 3< rr n dA = 0 (13-1) The St. Venant flexural shear flow distribution is obtained by applying the engineering theory developed in Sec. 11 -7. This distribution is statically equiva lent to F 2, F 3 acting at the shear centcr. It follows that the additional shear stresses, < y l 2 and a\ 3, due to warping restraint must be statically equivalent to only a torsional moment: fjcr i 2 dA ~ 0 [ R n dA = -- 0 (13-2)
To account for warping restraint, one must modify the torsion relations. We will stili assume the cross section is rigid with respect to in-plane deformation.
371
370
CHAP. 12
12-7. Refer to the sketch for Prob. 12-3. Determine the reaction R and centroidal displacements at = L / 2 due to a concentrated force Piz applied to the web at Xi = L/2. Employ the force method. 12-8. Refer to Example 12-7. Assuming Equation (h) is solved for Z t, discuss how you would determine the translation u2 at Xj - L/2. 12-9. Consider the four-span beam shown. Assume linearly elastic be havior, the shear center coincides with the centroid, and planar loading. (a) Compare the following choices for the force redundants with respect to computational effort: 1 . reactions at the interior supports 2 . bending moments at the interior supports (b) Discuss how you would employ Maxwells law of reciprocal deflections to generate influence lines for the redundants due to a concentrated force moving from left to right.
j---------------------------------------------------------- ;-------------------------------------------1
/ f if ? n % r i9 n r ftn r f t n
P ro b . 1 2 - 9
12-10. Consider a linearly elastic member fixed at both ends and subjected to a temperature increase
T = a t + a 2x z + ci} x $
Determine the end actions and displacements (translations and rotations) at mid-span. 12-11. Consider a linearly clastic member fixed at the left end (/4) and subjected to forces acting at the right end (B ) and support movement at A. Determine the expressions for the displacements at B in terms of the support movement at A and end forces at B with the forcc method. Compare this approach with that followed in Example 12-2.
13

13-1. INTRODUCTION
The engineering theory of prismatic members developed in Chapter 12 is based on the assumption that the effect of variable warping of the cross section on the normal and shearing stresses is negligible, i.e., the stress distributions predicted by the St. Venant theory, which is valid only for constant warping and no warping restraint at the ends, are used. We also assume the cross section is rigid with respect to in-plane deformation. This leads to the result that the cross section twists about the shear center, a fixed point in the cross section. Torsion and flexure are uncoupled when one works with the torsional moment about the shear center rather than the centroid. The complete set of governing equations for the engineering theory are summarized in Sec. 12-4. Variable warping or warping restraint at the ends of the member leads to additional normal and shearing stresses. Since the St. Venant normal stress distribution satisfies the definition equations for F t, M 2 , identically, the additional normal stress, o \ must be statically equivalent to zero, i.e., it must satisfy j'K it dA = |fx 2or n dA = JJxjOu dA - 0 (13-1) The St. Venant flexural shear flow distribution is obtained by applying the engineering theory developed in Sec. 11-7. This distribution is statically equiva lent to F2, F 3 acting at the shear center. It follows that the additional shear stresses, a\ 2 and c ^ , due to warping restraint must be statically equivalent to only a torsional moment:
fj< 7 j2 d A
0
2)
.[JV,3 dA = 0
To account for warping restraint, one must modify the torsion relations. We will still assume the cross section is rigid with respect to in-plane deformation.
371
372
R ES TR A IN ED T O R S IO N -F LE X U R E OF P R IS M A T IC M E M B E R
C H A P . 13
In w hat follows, we develop the governing equations for restrained torsion. We start by introducing displacem ent expansions and apply the principle of virtual displacements to establish the force param eters and force-equilibrium equations for the geometrically linear case. We discuss next two procedures for establishing the force-displacem ent relations. The first m ethod is a puredisplacem ent approach, i.e., it takes the stresses as determ ined from the strain (displacement) expansions. The second m ethod is sim ilar to w hat we em ployed for the engineering theory. We introduce expansions for the stresses in terms o f the force param eters and apply the principle of virtual forces. This cor responds to a mixed form ulation, since we are actually w orking with expansions for both displacements and stresses. Solutions of the governing equations for the linear mixed form ulation are obtained and applied to thin-w alled open and closed cross sections. Finally, we derive the governing equations for geomettrically nonlinear restrained torsion. 13-2. DISPLACEMENT EXPANSIONS; EQUILIBRIUM EQUATIONS
The principle of virtual displacem ents! states that jjjV r <5s d(v ol.) JJjb 7 Au rf(vol.) + j j p r Au ^(surface area) (a)
is identically satisfied for arbitrary displacem ent, Au, when the stresses () are in equilibrium with the applied body (b) and surface (p) forces. We ob tain a system of one-dim ensional force-equilibrium equations by introducing expan sions for the displacem ents over the cross section in term s of one-dim ensional displacem ent param eters. This leads to force quantities consistent with the dis placem ent param eters chosen. We use the same notatio n as in C hapters 1 1, 12. The X t axis coincides with the centroid; X 2, X 3 are principal inertia axes; and x 2, x 3 are the coordinates of the shear center. We assum e the cross section is rigid with respect to in-plane deform ation, w ork with the translations of the shear center , and take the dis placem ent expansions (see Fig. 13-1) as U\ W i + W2-X 3 <^3-^2 + u 2 = us2 - g>,(x3 - 3c3)
3
(13-3)
u s3
( .x 2
x 2)
where < /> is a prescribed function of x 2, x 3, and u i >un2 y us3 are the rigid body translations of the cross section.
2 . a;,, a)2, C 03 are the rigid body ro tatio n s of the cross section ab o u t the
3. shear center and the X 2, X 3 axes. / is a param eter definining the w arping of the cross section. The variation over the cross section is defined by < j).
N ote that all seven param eters are functions only of x x. F o r pure torsion
t See Sec. 10-6.
SEC. 1 3 -2 . D IS P LA C E M E N T E X P A N S IO N S ; E Q U ILIB R IU M EQ UATIO N S
373
(i.e., the St. Venant theory developed in C hapter 11), one sets / = a>li t = const and < f > = (j)t. F o r unrestrained variable torsion (i.e., the engineering theory developed in C hapter 12), one sets / 0. Since there are seven displacem ent param eters, application of the principle of virtual displacements will result in seven equilibrium equations.
*3
C e n tro id
Fig. 13-1.
N o t a t io n f o r d i s p l a c e m e n t m e a s u r e s .
T he strain expansions! corresponding to (13-3) are

l
fi2 =
u l, J +
3 =
w 2. 1 * 3
0
& 3. 1*2
+ /
10
}>23
712 Us 2 . 1 CO3 W i. t(X 3 .? 3) + f ( j ) ' 2

V l3 = Us3,
1 +
+ Q>i, 1(^2 * 2) + f4>, 3
(13-4)
Using (13-4), the left-hand side of (a) expands to HJ0T d(vol.) = J.rt[ / r, Am1(1 + F 2(^ s2,i - Aeu3)
-I- F 3 ( A u 4 3- j + M f
1 4
- A +
oj2
M +
2 A o?2 , 1
A a )3 . 1
(b )
J V f^ A / , 1
where the two additional force param eters are defined by

= J .k n ^ dA
M r J j ( c r 1 2 <^ > 2 + 0 l3 (p ,3 )d A
(13-5)
N ote that quantity
has units of (force) (length )2 and M R has units of moment. The is called the bimoment.
T h e n o n lin e a r s tr a in e x p a n s io n s a r e d e r iv e d
r T h i s d e r i v a t i o n is r e s t r i c t e d t o l i n e a r g e o m e t r y . in S e c . 1 3 - 9 .
374
RESTRAINED TORSIO N-FLEXURE OF PRISM ATIC M EM BER
CHAP. 13
To reduce the right-hand side of (a), we refer the transverse loading to the shear center. The additional load terms are dS distributed bimoment dA = external bim oment at an end section (xj = 0, L) Then
jjJ b T A u
(1 3 -6 )
d(vol.) F j J p T A u ^(surface area) = j Xl[>, A! + b2 Aus2 + b 3 Au s3 + mT Acu, + m 2 A< x > 2

+
m 3 Acl>3 + lsf~ \d x i
[ F * A w ^ -F F 2 A u s 2
(c)
+ F 3 AuS3 + M j Aai + M 2 Ao>2 + M 3 Aco3 + M $ A/'j.t:i =o ,t The definitions of bj, ntj, mT, F}, M }, M T are the same as for the engineering theory. Finally, we equate (b), (c) and require the relation to be satisfied for arbitrary variations of the displacement parameters. This step involves first integrating (b) by parts to eliminate the derivatives and then equating the coefficients of the displacement parameters. The resulting equilibrium equations and bound ary conditions are as follows:
Equilibrium Equations
F \, i + b i =
0
F2. i + b 2 = 0 F3 > t + b3 = 0
M r> ^ F yyij
0
M 2> i M 3t
i +
F3
-F
tn 2
F 2 - F W I3 = 0 Mq, 1 M j 5 + m,j, 0
Boundary Conditions a t x x - 0
(13-7)
F, =
ux = l
Ks3
= U .,2 = Us 3
or or or or or or or
F2 = - f 2 f 3 MT m2 m3
= -f 3 = -M ; = -M;
(Oi = W 1 (2 = 0)2 03 3 =
= M;
Af* -
Boundary Conditions at x x L
These are the same as for x j = 0 with the minus sign replaced with a plus sign. F o r example: / = 7 or
SEC. 1 3 -3 .
D ISP LA C E M E N T M ODEL
375
We recognize the first six equations as the governing equations for the engineering theory. The additional equation,
i M
r
<
X j
<
/ = /
or
- + M*
x t = 0, L
(d)
is due to warping restraint. Also, we see that one specifies either f or the bim om ent at the ends of the member. The condition f J applies when the end cross section is restrained with respect to warping. If the end cross section is free to warp, the boundary condition is = M4 , ( + for x t = L). To interpret the equation relating M R and the bimoment, we consider the definition for M R, M r = j ^ a l 2< f> .2 + Integrating (e) by parts leads to M r = Pi<J) d S ff^(C i 2,2 + Cij,3)<i/4 Utilizing the axial stress equilibrium equation, Ol 2 , 2 + f f 13. 3 + O 'll. l = 0 we can write
M
r
3)<^4
(e)
(f)
(g)
^
= =
P m#
4> d S +
J J c / x tu , 1 d A
M j,' 1
We see that (h) corresponds to the axial equilibrium equations weighted with respect to (f> ,
J J C c " 1 2 , 2 + < 7 1 3 , 3 + C i i , i)(p d A + $ (/? i a 2 < ? i2 d S 0
V
1
(i)
M
r
4-
w ij,
In m ost cases, there is no surface loading on S, i.e., p, = 0 on the cylindrical boundary. We will discuss the determ ination of stresses in a later section. We simply point out here that M r involves only the a d d i t i o n a l shear stresses due to warping restraint since the St. Venant shearing stresses correspond to
ffn = O .t
13-3.
FORCE-DISPLACEMENT RELATIONS DISPLACEMENT MODEL
To establish the relation between force param eters and the displacement parameters, we consider (13-4) to define the actual (as well as virtual) strain distribution and apply the stress-strain relations. W e also consider the m aterial to be isotropic and suppose there is no initial strain. The stress expansions are
f f n =
E eftS i
X 3(0 2 ,
1 -
* 2 3 .I
/ l 0 ]
<7l2 = Gyi 2 = G[us2 , 1 (O 3 colyl(x 3 * 3 ) + J<p, 2 ] ^13 Gy 13 G[us3 >! + C O 2 + C O i, i ( * 2 ~ * 2) + f. 3 ]

f M
r
(1 3 -8 )
0 f o r S t. V e n a n t (p u re ) to r s io n .
W e n e g le c t
and
fo r u n r e s tr a in e d v a r ia b le
to r s io n .
376
RESTRAINED TORSION-FLEXURE OF PRISMATIC M EM BER
CHAP. 13
where cff denotes the effective modulus. Although our displacement expansions correspond to plane strain (e2 = e3 0 ), the in-plane stresses vanish on the boundary. Therefore, it seems more reasonable to use the extensional stressstrain relations for plane stress. In what follows, we will take Eeff = Youngs modulus, E. Consider the expression for i . The term involving < j>is due to warping of the cross section. This additional stress must satisfy (13-1), which, in turn, requires < f> to satisfy the following orthogonality conditions:! [J0 dA Jfx 2 < f>dA J\ x 3 (f) dA 0 (13-9)
Assuming (13-9) is satisfied, and noting that X 2, X 3 are principal centroidal axes, the expressions for F u M 2, M 3, and the iVI^ reduce to :
F t M ~
=
E A u u
F I 20)2, I E l
M3
M #
3< X > 3_ 1
J
(13-10)
E rI J
where (Jf/)2 dA We have included the subscript r on E to keep track of the normal stress due to warping restraint. Inverting (13-10) and then substituting in the expression for < 7 U lead to F,- t. Mj Ml * * ~0 , 'M, (p I (13-11) a ix H r ~ * 2 + ~ r <l>
A l-> 1 2
'3
I#
The expressions for F2, F 3, M T, and M R expand to

j_
G
1
F 2 A(us2 , 1 (Vi + X3C 01 , 1 ) + f S 2 F 3 /4(u,3 1 + co2
G
1
1) + f S i
(13-12)
M f
11cu1 > t + flj,

M
3F 2
X zF
G where
Mr F2 + J Fi + IfC O1 , j + Iff
GA Z ' GA
1
S j t ff a jd A
= polar moment of inertia = I 2 + h

1$ jj( x 2 0 . 3 ~ x 3<ib , 2 ) d A
I<f> =
f f ( < A ,2 2
^ ,2 3 ) d A
(S
S f)
Fi M2 iW3 =
0 for crn due to warping restraint.
SEC. 1 3 -3 .
DISPLACEMENT MODEL
377
Also, the expressions for the shearing stresses can be written as
(13-13)
The essential step is the selection of (p which, to this point, must satisfy only the orthogonality conditions (13 9). To gain some insight as to a suitable form for 4 > , let us reexamine the St. Venant theory of unrestrained torsion. We suppose the section twists about an arbitrary point (x 2, x'3), instead of about the centroid as in Sec. 11-2. The displacement expansions are 2 = <W i(X3 - x'3)
fti
ih = co,0c2 - x'2)
(i. ifit
(a)
where tox^ t = M t/GJ = const. Operating on (a) leads to

a u =
0
/V iI o 12 = J [0;,2 - (Xj - x'3)]

= y [ $ '.3 + (X 2 * i)J
(b)
The equation and boundary condition for < f)'t follow from the axial equilibrium equation and boundary condition, V 24 > 't = 0 in A
(c)
on
We can express as < t> l = C - x'3x 2 + x'2x 3 + (f> t
(d)
where C is also an arbitrary constant. The boundary condition and expressions for the stresses become
cd>t ~ ~
cn
K n2*3
~~ a n3 Y 2
(e)
Since < f> t depends only on the cross section, it follows that the stress distribution
378
RESTRAINED TORSIO N-FLEXURE OF PR ISM ATIC M EM BER
CHAP. 13
and torsional constant are independent of the center of twist. Also, one can show f that
2) d A
~ \ \ { x z ( t > r, 3
x 3 (j)t ,
2) 2 +
( 0 ,, 3) 2] = < f)[. The constants (C, x'2, x 3) are evaluated by requiring (f> 't to satisfy (13-9), and we obtain
C = I f a d A
x'z =
i i
fjx 3< /> , dA
(g)
x '3 = ^ $\x 2 (f> , dA

1 i
Now, one can show j that the equations for x 2 , x 3 are identical to the equations for the coordinates of the shear center when the cross section is considered to be rigid with respect to in-plane deformation. T hat is, the warping function for unrestrained torsion about the shear center is orthogonal with respect to 1, x 2, x 3. Summarizing, we have shown that
(j> C
x 3x 2
x 2x 3
<j), =
4 f,Q
(13-14)
is a permissible warping function. The cross-sectional properties and forcedisplacement relations corresponding to this choice for (f> are listed below:
Cross-Sectional Properties
S
= =
x 3A - r ; h + r+ =
S 3
x 2A
I* j I#
(13-15)
h + r i i f t t , j ) 2] d A
jj[ ( ^ r . l ) 1
Shear Stresses
F 2
0 12 = ~ 7 + G ( X 3W 1, i + f(j), 2)
(13-16)
+ G { x 2 c o i, ! 1 + f < p t, 3)
F =
7
A
t See Sec. 11 -2 and Prob. 11 -2. j See Prob. 13-1.
SEC. 1 3 -4 .
R ESTRAINED TO R SIO N D ISP LA C E M E N T M O D EL
379
Force-Displacement Relations M T - G J, !., - G l ' i f + x }F 2 - x 2F 3 M r G l ^ i f coit i) x 3 F 2 4- x 2F 3

= m *2, i 3
CO2 +
* 3( / - i, 1 )
(13-17)
F3
- Q ^ UsZ, 1 +
_
X2 ( / 0 i_ i )
We introduce the assum ption of negligible restraint against w arping by setting Er = 0. Then, 0, and the seventh equilibrium equation reduces to M r = 0. Specializing (13-17) for this case, we obtain
w
M f - G J < D \ i
(13 -
18 )
and
Fi f I
u,2. . 3 +
o
-F3 (
+ ? ; ) +
g
- - n
F2 f
x 2 x 3\ r
F3 f 1
x f\
(13-19)
+ ? ( T
G \i + n)
The shearing stress distributions due to F 2, F 3 do not satisfy the stress boundary condition & n2 & 1 2 + K3 7i 3 0 On S (a) However, one can show that they satisfy { ^ 2 ^ 1 2 + at3 <rii)dS = 0 (b)
for arbitrary F 2, F 3. Equations (13-19) are similar in form to the results obtained in C hapter 12, which were based on shear stress expansions satisfying (a) identically on the boundary. Finally, we point out that torsion and flexure are uncoupled only when warping restraint is neglected (Er 0). Equations (13-17) show that restrained torsion results in translation of the shear center. We will return to this point in the next section. 13-4. SOLUTION FOR RESTRAINED TORSION DISPLACEMENT MODEL
To obtain an indication of the effect of w arping restraint, we apply the theory developed in the previous section to a cantilever member having a rectangular cross section. (See Fig. 13-2). The left end (x t = 0) is fixed with
380
C H AP. 13
*3
2b
M
- x 2
V ~ 2a H
Se c t. A - A
Fig. 13-2.
R e s t r a i n e d t o r s i o n - c a n t i l e v e r w it h r e c t a n g u l a r c r o s s s e c t i o n .
respect to both rotatio n and w arping while th e right end ( x i = L ) is free to warp. The boundary conditions are
x x =
0
coi f
Xi L
M *
M
=
0
(a)
F o r convenience, we list the governing equations for restrained torsion:

Equilibrium Equations (See ( 1 3 - 7 ) )
M lt j 4- t t i i
(b) (c)
N ote that
MR M< t) <j + m$
Force-Displacem ent Relations (See ( 13- 10) and ( 13- 12) .
f
= M i =
E J t f ,
i + G l 't f
G lia H ,!
(d)
M R = G/â>1(1 + G r^ f Boundary Conditions (for this example)
At x \ = 0,
f = 0 3 1 = 0
(e)
At x r = L,
Mi = M
/ , !
= 0
W e start with (b). Integrating (b) and enforcing the b o u ndary condition at Xi = L leads to (13-20) Mi = M
SE C . 1 3 -4 .
R E S T R A IN E D T O R S I O N D I S P L A C E M E N T M O D E L
381
Next, we combine (c) and (d):

G/itt)!.! + G l 't f = M (f)
(g)
i,i + Solving (f) for co1 :

Q > i,i
ErI< j> f i11
/i
(h)
and then substituting in (g) lead to

/ .
ii
- 3
2/
= E r h L
M
(n r
(i)
where
is defined as
J2 m J L E r r h y
i_
'
17
(13-21)
N ote that P has units of ( 1 /length)2. The solution of (i) and (h) which satisfies the boundary conditions (e) is (we drop the subscript on x for convenience) / = M
ir j M ( /" I'A
1
, T {1 cosh Xx + tanh XL sinli Xx} (13-22)
wi = 7 < x + [sinh Xx + (1 cosh Xx)tanh XL] > GJ ( /J, Ail J

I'i " i +
The rate of decay of the exponential terms depends on X For 1L > as 2.5, we can take tanh XL x 1 , and the solution reduces to
e~*x)
1 fh " *
(13-2 3 )
GJ As a point of interest, the St. Venant solution is dcox M ~ 'dx ~ GJ
( j )
We see that 1/2 is a m easure of the length, L, of the interval in which warping restraint is significant. We refer to L b as the characteristic length or boundary layer. By definition, 0 (13-2 4 )
In w hat follows, w e shall take
L X
(13-25)
The results obtained show that X is the key param eter. Now, X depends on the ratio G/Er and on terms derived from 4> , the assumed w arping function. If
382
RESTRAINED TORSION-FLEXURE OF PRISM ATIC M EM BER
CHAP. 13
we take (/> 0 ,c, the warping function! for unrestrained torsion defined by (13-14), the various coefficients are related by = -n
j
= h-i;
(13-26)
j' j At this point, we restrict the discussion to a rectangular section (see Fig. 13-2) and (p = (f> 5 tc. We evaluate the various integrals defined by (13-15) and write the results as / = K ja 3b
11 I, * = K j = + K ; a 3b K t a 3b K K t y v
k
(1 3 -2 7 )
where the K 's are dimensionless functions of b/a. W ith these definitions, the expression for X takes the form X(13-28)
K j
The coefficients are tabulated in Table 13-1. We see that K x is essentially constant. Assuming E 2.6 G and K x 3.2, we find X 2jb and L b & 2b. The influence of warping restraint is confined to a region of the order of the depth. Although this result was derived for a rectangular cross section, we will show later that it is typical of solid and also thin-walled closed cross sections.
Table 13-1
b a Kj K* K'4 K,
1 2
3
10
2.25 3.66 4.21 4.99
.0311 .165 .283 .425
.156 .450 .083 .964
3.36 3.16 3.23 3.32
We consider next the problem of locating the center of twist. We utilize the solution corresponding to (j> and large AL:
M
M f
(13-29)
| C - T 2 ^3 = 0 for a rectangular section and < /> ,jiliear ;C T U rf reduces to <p,\,
SEC. 1 3 -5 .
M IXED F O R M U LA T IO N
383
The translations of the shear center follow from (13-17): " > = * < /
W S3, 1 =
~ X
( , 3 .30)
COl,x)
2( / -
By definition, the translations are zero at the center of twist. Setting u 2 u 3 0 in (13-3) and letting x ' 2 , x '3 denote the coordinates of the center of twist lead to x l2 = g x 2 x 3 = g x 3 1 i; /I (13-31)
\ \
We see that the center o f twist approaches the shear center as x increases. The m axim um difference occurs at x = 0 and the m inim um at x L.
I y |, = 0 = ------------77, =
I _ J
1
1
d \x - L ~
i -
(13-32)
77,
= y ~ Xu ,
For unrestrained warping, E r = 0 , 1 = oo, and g = I . 13-5. FORCE-DISPLACEMENT RELATIONS MIXED FORMULATION
We first review briefly the basic variational principles for the three-dim en sional form ulation. The principle of virtual displacem ents requires jJJ<rT dz d (v ol.) = jJJbT Au rf(vol.) + [JpT Au ^(surface area) (a)
to be satisfied for arbitrary Au and leads to the stress-equilibrium equations and stress-boundary force relations. N ote that ds is a function of Au and is obtained using the strain-displacem ent relations. The stress-strain relations can be represented as e T So =* 5V* (b) since, by definition of the com plem entary energy density,
dV* dV*
y,A) =
(0
By com bining (a) and (b), we obtain a variational principle which leads to both sets of equations. The stationary requirem ent, <5[JfJ(orT 5 Tu F*)d(vol.) Jjp Tu (/(surface area)] = 0 (13-33)
considering a, u as independent quantities, = c(u), and p, b prescribed, is called R eissners principle, t

f See Ref. 11 a n d P ro b . 1 0 - 2 8 . R e issn er s p r in c ip le a p p lie s for a rb itrary g e o m e tr y a n d e la s tic m aterial. T h is d isc u s sio n is r estricted to lin ear g e o m etr y . T h e n o n lin e a r c a se is trea ted in S ec. 1 3 - 9 .
384
C H AP. 13
The essential point to recognize is th at Reissners principle allows one to w ork with ct and u as independent quantities. In a displacem ent form ulation (Sec. 13-3), we take a as a function of u, using the stress-displacem ent relations < r = D t = <y(u), and <rr e - V * reduces to V, the strain-energy density. In a mixed form ulation we start by introducing expansions for the displacem ents. The Euler equations for the displacem ent param eters are obtained by ex panding (a). This step leads to the definition of force param eters and forceequilibrium equations. We then generate expansions for the stresses in term s of the force-param eters from an equilibrium consideration. The relations between the force and displacem ent param eters are obtained from the second stationary requirem ent: [jT( T ^C T ~ 3V*)dA]dxi = 0 (13-34)
The first step was carried out in Sec. 1 3 -2 and the expanded form of JJ 5&ra dA is given by (b) of Sec. 13-2. Letting V* represent the com plem entary energy per unit length along X u and using (13-4), the stationary requirem ent on the stresses (E quation 13-34) expands to 5 F iu i, i + dF 2 (usl' i 0U 3) + 8F 3 (uS3' 1 + 0J2 ) + 8 M 2 co2, i + <5M3ft>3, 1 -f- dMj-coi i + <5M^/t j + SM jiJ 5V* 0 . ^ ,,,
In order to proceed further, we m ust express V* in term s of the force param eters (F 1 } F 2, , M r ). E quating the coefficients of each force variation to zero results in the force-displacement relations. Instead of applying (13-34), one can also o btain (13-35) by applying the principle of virtual forces to a differential element. We followed this approach in C hapter 12 and, since it is of interest, we outline the additional steps required for restrained torsion. O ne starts with (see Fig. 13-3)
SV* d x t = d i S p i = U K d A l ^ + [ jI uT d A ~\xl +<ixi
(a)
The boundary forces are the stress com ponents acting on the end faces. Taking u according to (13-3) and considering only M T) M#, M R > we have fj <5pTu dA + 1[ <5(yTu dA
+ ( 5 M j- o ji
+ S M ^ f)
where the plus sign applies for a positive face. The virtual-force system must be statically permissible, i.e., it m ust satisfy the one-dim ensional equilibrium equations. This requires S M T const
SEC. 1 3 -5 .
M IXED FO R M U LATIO N
385
The first procedure (based on (13-34)) is m ore convenient since it avoids introducing the equilibrium equations. However, one has to have the straindisplacement relations. In certain cases, e.g., a curved member, it is relatively -8MT
-5 m T + 5M $ tld x y
Xx
C Ji /
F ig . 1 3 - 3 .
V ir tu a l fo r c e s y s te m .
easy to establish the force-equilibrium equations by applying the equilibrium conditions to a differential element. We obtain the force-displacement relations by applying the second procedure (principle of virtual forces) w ithout having to introduce strain expansions.! In what follows, we consider the m aterial to be homogeneous, linearly elastic and isotropic. To simplify the treatm ent, wc also suppose there is no initial strain. The complementary energy density is (<Xj2 + o'? 3) ^ (13 36)
l G
It remains to introduce expansions for the stress com ponents in terms of the force param eters such that the definition equations for the force param eters are identically satisfied. Considering first the normal stress, we can write %
_ F,
a" ~ a
Mi
Af,
'i '. * (a)
I T Xl - 7 T
where dA = jj.v2(j) dA = . The complementary energy due to o! 1 expands to
/ P * l
1
( F '
j + 7r + 7r j + 2 T
u.
i.
( M * \
r j
(c)
t T h e a p p r o a c h b a s e d o n t h e p r i n c i p l e o f v i r t u a l f o r c e s is n o t a p p l i c a b l e f o r t h e g e o m e t r i c a l l y n o n lin e a r ra se . J Sec (13 -11), 5 F, = M 2 = P r o b le m M 3 = 1 3 - 8 tr e a ts th e c a s e o f a n o n h o m o g e n e o u s m a te r ia l.
0 f o r <7 , 1 d u e
fo
w a r p in g r e s tr a in t.
386
RESTRAINED TORSION-FLEXURE OF PRISMATIC MEMBER
CHAP. 13
F i n a l l y , s u b s t i t u t i n g f o r ( F * ) 0 ll i n ( 1 3 - 3 5 ) , w e o b t a i n F i M 2
Ulrl ~ AE
-M l
< X > 2' 1
~ El 2
(d )
-M i
3' 1
T h e s e e x p a n s io n s c o in c id e w ith
El 3
f '1
ErU
w ith th e
th e c o r r e s p o n d in g r e la tio n s o b ta in e d
d is p la c e m e n t m o d e l (se e ( 1 3 - 1 0 ) ) . T h e s h e a r in g
t
s tre s s
d is tr ib u tio n
m u st
s a tis fy o b ta in r e s u lt s
th e
d e fin itio n
e q u a tio n s
fo r
F 2, F j, M a te rm
and to
M R id e n tic a lly . w a r p in g
W e can to th e
s u ita b le e x p a n s io n s b y a d d in g fo r u n r e s tr a in e d to r s io n and
due
r e s tr a in t
fle x u r e .
W e w r ite
a i j o { j + a i j + o \ j
w h ere to r s io n c
(13- 37)
th e u n r e s tr a in e d to r s io n . fo r
r{j
is
th e
fle x u r a l
d is tr ib u tio n
d ue
to
F 2 , F 3 ; c r " ( is due
d is tr ib u tio n ; a n d
< rj j i s t h e d i s t r i b u t i o n
to r e s tr a in e d
S in c e w e a r e a s s u m in g n o in - p la n e d e fo r m a tio n , th e fle x u r a l d is tr ib u tio n a th in - w a lle d s e c tio n Sec. 1 1 - 7 . . fo r u n r e s tr a in e d to r s io n is t r e a t e d can be o b ta in e d by a p p ly in g th e e n g in e e r in g
th e o ry
d e v e lo p e d in ta k in g v =
0
F o r a s o lid s e c t io n , w e u tiliz e th e r e s u lt s o f S e c . 1 1 - 5 ,
T h e s h e a r s tre s s d is tr ib u tio n th ro u g h to a 1 1-4 .
in S e c s . 1 1 - 2
S in c e th e r e s tr a in e d - to r s io n h ave to
d i s t r i b u t i o n is s t a t i c a l l y e q u i v a l e n t b e tw e e n th e
to r s io n a l m o m e n t, w e
d is tin g u is h
unrestrained
and
restrained
to r s io n a l m o m e n ts : M r = = c\j = M 't f { m + ) M t (13 -3 8 )
y (M 'r ) fo llo w th e sam e ap p roach th e as in th e e n g i
It r e m a in s n e e r in g
to
d e te r m in e o f fle x u r a l
o '[j.
W e
th e o ry
sh ear
stre ss,
i.e .,
w e
u tiliz e
a x ia l
e q u ilib r iu m
e q u a tio n s a n d s tre s s b o u n d a r y c o n d itio n :
Ol 2 , 2 2 < ? i2 D iffe r e n tia tin g w e o b ta in th e +
<?1 3, 3 a n3 f f U fo r cru =
~ O 1 1 , 1 0 and n o tin g
in A (a ) o n S th e e q u ilib r iu m e q u a tio n s ,
e x p r e s s io n
5 i u
F 2 ~ f~ x
^ +
F i
L M r + ~y~<P
i 3
S in c e o f s a tis fie s (a ) fo r a r b itr a r y
J2
F 2, F 3 and a"
*(j)
(b )
co rresp o n d s
to
<rn
0 , it
f o l l o w s t h a t o r is d u e t o M R :
J6
SEC. 13-5.
MIXED FORMULATION
387
The orthogonality conditions on < f>and boundary condition on < jr ensure thalf F\ dA - 0 Fr 3 - JjV13 dA - 0 (13-40) We solve (13-39) and then evaluate M'r from
M>r = J'J[ (-^3 ~ X3)(T12 + (* 2 3c2) ( 7 i 3 ] ^ (c)
Noting (13-40), we see that Mr T = M i. Finally, we write (c) as M'r - + C+Mr (13-43)
where Q, is a cross-sectional property which depends on < j> . With this definition, MT - Mr + Q M K (13-42) When the cross section is thin-walled, we neglect aln and (a) reduces to
ffu' s
crls 0 . at a free edge
"
f f i u
( d)
We take ft
q' s ~ qr 0
03-43) at a free edge
The orthogonality conditions on < f>and boundary condition on qr ensure that
^3 JaS 3qr dS.~ 0 Finally, we determine Q> by evaluating M'T and equating to (1.3-41). We consider next the complementary energy density. We write the expanded form of the shear contribution as
^ *sh ear J
q
JJ
K 0^ 2
^ 3
&*13 +
<^\^)2] d A
V f
F?
V?
VJr +
K -
We have evaluated Vf, Vf( and V% in Sec. 11-5. For convenience, these results are summarized below (See Equation 11-98)
= -L ( l + E i + l
2G \ A 2
V
A2$
(a)
*
M
(M r )2
2 GJ 0
See Prob. 13-2.
388
RESTRAINED TORSION-FLEXURE OF PRISMATIC MEMBER CHAP. 13
The coupling term, l /A 23, vanishes when the section has an axis of symmetry. Also F*f 0 is a consequence of our assuming the cross section is rigid with respect to in-plane deformation. We evaluate P*, using (13-39) ((13-43) for the thin-walled case), and write the results as V* ~ ~ r 2G [(tf'12)2 + (c> r t 3)2]dA =
2G
~ j^
(13-46)
where Cr is a dimensioniess factor which depends on < j> . The coupling between unrestrained and restrained torsion is expressed as
V r* nr = _
2 + ^ 3^13 )dA M'tM r
(13-47)
It is obvious that Cur = 0 for a thin-walled open section sincc a" is an odd function of n whereas ar is constant over the thickness. We will show later that it is possible to make C,ir vanish for a dosed section by specializing the homogeneous solution of (13-43). Therefore, in what follows, we will take
Cur = 0.
Finally, we write the coupling between flexural and restrained torsion as

VJr = j
j (o'{ 2 c i 2
+ <^{30^ 3 ) d A
j JJ = 7 7 7 (^ 31-^ 2 ^ ^ 4-
(13-48)
x 2rF 3M R)
where x jr have units of length. If X 2 is an axis of symmetry, x 3r = 0 since a J is symmetrical and a r is antisymmetrical with respect to the X 2 axis. We substitute for f * in (13-35), replace M r with M + C^,MR, and equate the coefficients of < 5 F2, 5F3, 5M u r, and 6MR. The resulting force-displacement. relations are 1 us2, 1 ~ W3 ^ ~ ~ A
G
u sl. I + =
(F 2
F,
x 3r
1 7
+
A 22 F3
-J -
"f
r~ Mr
7-
1 ( F2
I -J -
x 2r
H---------- M K
M't
Wl- 1 ~ G J
Al
'
(13-49)
C < !,0)irl + f 7TT Mr + (Xir^Z + -<2rF 3) y jJ (jJ

The corresponding relations for the displacement model are given by ( 1 3 - 1 2 ) . Up to this point, we have required (j> to satisfy the orthogonality relations and also determined ar such that there is no energy coupling between 0 and ( C = 0). If, in addition, we take
(f) = ~(C x 3x 2 + x 2x 3 + < /> ,) =
- rf>f
SEC. 13-6.
RESTRAINED TORSION MIXED FORMULATION
389
thent
C
M r =
-l-1
(13-50)
K '
+ M r
Note that < j)s tc is the warping function for unrestrained torsion about the shear center. We discuss the determination of 4 > in Secs. 13-7 and 13-8. One neglects shear deformations due to flexure by setting - L = 1 = - 7 = 0
A 2 A 3 ^23
(13-51)
Similarly, we neglect shear deformation due to restrained torsion by setting Cr = x2r - x3r = 0 (13-52)
This assumption leads to the center of twist coinciding with the shear center and /= One now has to determine
M r T
(13-53)
from the equilibrium relation,

i +
M T r =
If M f is known, it is more convenient to work with

M t = M
t
M 'f
(b)
In what follows, we outline the solution procedure for restrained torsion and list results for various loadings. We then discuss the application to open and closed cross section.
13-6. SOLUTION FOR RESTRAINED TORSION MIXED FORMULATION
We suppose only t o r s i o n a l loading is applied. The force-displacement rela tions are obtained by setting F 2, F 3, co2, o)3 equal to zero and C# = + 1 in (13-49). For convenience, we summarize the governing equations below.
E q u ilib r iu m E q u a t io n s
M t, i T "
~ 0 M j = Mf'X
(a) (b)
Foi ce-Displacement Relations (.([> < fitst;)

x
M t (jJo)i, ! GJ
M r U yL
, + / )
,C)
f See Prob. 13-3. We include the minus sign so that Q , will be positive.
390
RESTRAINED TORS ION-FLEXURE OF PRISMATIC MEMBER CHAP. 13

B o u n d a r y C o n d it io n s M r
M.
or or
a>i|
/ i
prescribed at each end
(d)
T r a n s l a t io n s o f th e S h e a r C e n t e r
us2, 1 yrj M 't

\JJ
L 2r U s3 , ! ~~~ A fj(jj
(e)
We start by integrating (a): M r - C\ - \m T d x l = C, + M Tp (13-54)
Substituting (c) in (b) and (13-54) leads to the governing equations for co1 and / :
(1 + C r)o j i . i + j
LrJ
0
(C , +
Tp)
(0
C r E r l+ fn - G J iW !,!
f)
After some manipulation, (f) becomes

M Tp d x y
GJ GJ G J
(g)
/ 11 AY
GJ
(Cx + M Tp)
where A2 is defined ast

c, =
1 + c. P = c -
(13-55)
Equation (g) corresponds to (h), (i) of Sec. 13-4. The general solution for / and has the following form:
Q
f C3 cosh '
C' C
a x
+ C4 sinh Xx - + f p GJ
j.
f M Tp d x
(13-56)
a x
= G J X + Cl +
(il J
)
- (C3 sinh Xx + C4 cosh XL w here/p is the particular solution due to on xx for convenience.
M Tp.
9 ld A l l dx
We have dropped the subscript
f T h e correspon ding param aler for the d isp iacem cn t-m od el form u lation is /, (see (1 3 -2 1 )).
SEC. 13 6.
391
The significance of k has been discussed in Sec. 13-4. We should expect, on the basis of the results obtained there, that XL will be large with respect to unity for a closed section. We will return to the evaluation of X in the next section. In the examples below, we list for future reference the solution for various loading and boundary conditions.
Example 13-1 ----------------------------------------------------------------------------------------C antilever C o n cen tra ted M o m e n t
Fig. E 1 3 - 1
,Y, ------ M
The boundary conditions (Fig. E13-1) are x = 0 -x

L
Mt = / = 0
M T = M
(a)
/., = o Starting with (13-54), we set M Tp 0 and C\ M . The remaining constants are deter mined from != = / = 0 /., = 0 and the final solution ist cosh H L x) cosh XL ------ {sinh {sinh XL sinh /(L x)} / cosn /.L = M Mr = M
Mr = M X cosh XL
at
a:
= 0
at x L
(b)
sinh X{L x)
(13-57)
1 - C,
M l
cosh X(L x) cosh XL
t The corresponding solution based on the displacement model is given by (13-22), (13-26). The expressions /or / differ by a minus sign. This is due to our choice of < /> . We took < f> = 0 in the displacement model and < f> 0 in the mixed model.
392
CHAP. 13
N ote that Cs = 1 when the complementary energy term due to the restrained torsion shear stress (ar) is neglected. The translations of the shear center are obtained by integrating
u , 2, , -
~ M r T (J j
u s3 . , =
~ M t K jJ
(c)
and requiring us2, u s3 to vanish at x = 0. We write the result as

US2 = X 3rU lls 3 = X2 .rll
11 =
...
, M M j d x = - 7- x (Ox
GJ
(13-58)
Let x j , i/j denote the coordinates and translations o f the center o f twist. By definition,
u 2 = s2 ~ d 3 = usi
0 )i(X3 .^3) = 0 + co 1 (.Vi - x 2) = 0
(d)
Substituting for usj and <yt, we obtain t

x'2 g = x2 =
g x 2r
x '3 - x 3 = g x 3r (13-59)
1 + ------------ < = ---------------------------------------------x ---------- - - [sinh X L sinh /(L x)] a cosh / . L
The limiting values}: of g occur at x = 0, L.
1
3|.v = 0 =
_ l+
I (13-60)
d \ x - - i . -------------------
-1 +
XL
C, tanh XL
N ote that x Jr = 0 if X k (j 7^ k) is an axis o f symmetry for the cross section. Also, x 2r = x 3r = 0 if we neglect shear deformation due to the restrained shear stress and, in this case, the center of twist coincides with the shear center throughout the length.
Example 13-2 - - ------------- ----- -------- ----- - -------

We consider next the case where warping is restrained at b o t h ends; the left end (x = 0) is fixed and the right end rotates a specified amount a> under the action of a torsional moment. The boundary conditions are
x
(ax = / co 1 = co
= 0
f
(a) = 0 (b)
x = L
f S e e ( 1 3 - 3 1 ) , ( 1 3 - 3 2 ) for th e d isp la c e m e n t m o d e l s o lu tio n . j T h ere is no tw ist o r tr a n sla tio n at x = 0. W e d e te rm in e q( 0) b v a p p lv in ? L H o s p ita ls rule to (1 3 -5 9 ).
SEC. 13-6.
393
To simplify the analysis, we suppose there is no distributed load. Starting with the general solution,
M T = C\
/ C 3 cosh Ax + C 4 sinh Ax
C !X
Cx
GJ
(c)
Cs u>i --------h C z ------- { C t , sinh Ax 4- C4 cosh Ax}

GJ /.
and enforcing the boundary conditions leads to the following relations:
C3 =
C,
GJ
G, - c ,
1 - c
c = cosh AL
s = sinh AL
C,C 4
2(1 GJ c)
AT
,
f
~ 7 ^cosh J x + i GJ (
c, r
(\ - c \
1 c
1 sinh Ax 1
(1 cosh Ax) sinh
ax
(13-61)
C, f
C.
0h = G J X x + T
M t = C, s M
M *t = C i { 1 C
cosh Ax +
I.
M 't Mj- M t
M * = Erl+X (
(~)
L
sinh Ax
| sinh }-x + M
) cosh A
We write the relation between the end rotation, co, and the end moment M , as
M co =
(77
* '
where Lcff denotes the effective length: , - iW ir J AL V s = L(1 - C,CA ) The following table shows the variation of C? with AL. For AL > 4, Cj, * 2/AL. N ote that C3 = 1 if transverse shear deformation due to restrained torsion is neglected. AL 0.5
Cx
(13-62)
1 2
3 4
0.98 .924 .76 .60 .48
394
CHAP. 13
Example 13-3 Uniform Distributed M om ent-Sym m etrical Supports

The general solution for m T = constant (we let m T = m for convenience) is: M r = Ci m x
, Ci , C+ . C, 77Z.X / = cosh /oc -I------sinh / jc I -------L L GJ GJ
Ci m fxz CA Cs -----7 I ------- (C 3 sinh ax + C 4 cosh Ax) co, = x + C 2 --------i ------ 1 1 GJ GJ \ 2 )}) }. L
(a)
We consider the boundary conditions to be identical at both ends and measure x from the midpoint (Fig. E l 3-3). Symmetry requires
M T = 0)
/ = oj and (a) reduces to
M r = mx C,l L m \x2
n
a tx = 0
(b)
. m / = sinh Ax + x
GJ
(13-63) CSC 4 .
Cv)
2 ~ g j | f
I fj _ T tT
We treat first the case where the end section is fixed with respect to both rotation and warping. Requiring (13-63) to satisfy
= 0
at x - L/2
(a)
results in /' = - m~ i x
GJ ' m L 2 {1
- - - sinh Ax t 2s J
r
x C,
/ x \ 2'
= G7 ( 8 . * - 4
M t mx
u ).
C5 + 2 u L (cosh
} c)i
f
mC, f
M = A
(13-64)
Ml j m L < --------F v ' sinh Ax J ( L zs
AL s 1 ------- cosh Ax AL c cosh 2

s sinh -
AL 2
The solution represents an upper bound. A lower bound is obtained by allowing the section to wrap, i.e., by taking
to! = / , * = 0
at x = -
(b)
SEC. 13-7.
APPLICATION TO THIN-WALLED OPEN CROSS SECTIONS
395
Fig. E13-3
-Xi
mL
-LI2-
u 2-
and the result is

m { 1 )
mL2 U QJi = ~GJ
_C ,
1 - 4 [ |
(8
2 (cosh Ax c)
c(XL)
M r m x
Mu r m i x H - sinh /_:c Ac
C,
(13-65)
m\ --j ! 1 ----------
fC /
cosh AX
c
U V
IL
c = cosh
13-7.
In what, follows, we apply the mixed form ulation theory to a wide flange section and also to a channel section. We first determ ine the cross-sectional properties corresponding to < j> = - W and then obtain general expressions for the stresses in terms of dim ensionlcss geom etric param eters. Before discussing the individual sections, we briefly outline the procedure for an arb itrary section. C onsider the arbitrary segm ent shown in Fig. 13-4. We select a positive sense for S and an arbitrary origin (point P). The unrestrained torsion w arping function is obtained by applying (11-29) to the centerline curve and requiring the section to rotate about the shear center, t
Îcemer!,ne = J
ou
M (
y
( P,c + ~ 4>V j
( 13 - 66 )
where psc is positive when translation in the + S direction rotates the position vector about the -f X 1 direction. The unrestrained torsional shear flow is zero
t B y d e fin itio n , /<, M r/G J . W e w o r k w ith ct rather th an a n d m ix ed s e c tio n s w h ere o n e g e n e r a te s q" in te rm s o f M ' j / J . y"s
to fa c ilita te tr e a tm e n t o f c lo s e d
396
R ESTR AIN ED TO R S IO N -FLE X U R E OF P R IS M A T IC M E M B E R
C H AP. 13
for an open section. Then, taking (f? = 0 fc and integrating leads to

r
+ js S r Ac
c lS
(13-67)
N ote th at one can select the sense of S arbitrarily. Also, < j>varies linearly with S when the segment is straight. The constant (f)P is evaluated by enforcing the orthogonality condition (a[ i - = 0), \(j)t dS = 0 (a)
If the section has an axis of symmetry, (pP 0, if we take P on the symm etry axis. The rem aining orthogonality conditions ((j'n -> M 2 M i = 0), j ^ x 2f dS 1 dS = 0 (b)
are identically satisfied by definition of the shear center, t

*3
------------------------------ *2
F ig. 1 3 - 4 . N o ta tio n fo r d e te rm in a tio n o f th e w a rp in g fu n c tio n .
W hen the section has branches, we apply (13-67) to each branch. O ne has only to require continuity of (f>at the junction point. As an illustration, consider the section shown in Fig. 13-5. The distribution of (j) for the three branches is given by A B B C B D (f> = < j)f> + jo psc dS (j) < j> B + jo psc dS cj) = (/> + Jo Psc dS (c)
We are taking the origin at B for branches B C and B D.

t See Prob. 1 3 -1 .
SEC. 13-7,
397
The shear flow due to M T is obtained by integrating (13-43) and noting (13-50). F o r convenience, we let qr = - V W ith this notation, the resulting expression simplifies to qr q 'P + T 0 / dS = < fP + Q
jsp
(13-68)
(13-69)
W e start at a free edge and work inward. A + q points in the + 5 direction (see Fig. 13-5). Then, a -I- qr corresponds to q \ i.e. qr acting in the S direction. If the section has an axis of symmetry, (j) is an odd function with respect to the axis and ~ q r is an even function.
*3
xl
Fig. 13-5. E x a m p le o f a s e c tio n w ith b ra n c h e s .
Once 

J
and Cr with (13-10), (13-46):
dA \(j)2t dS
Cl
C' = m
\ \ C(r,12)2 + K j , 2 ] i'4
(13-70)
In order to evaluate x lt, .x3r, we need the flexural shear stress distributions. W e let qU ) be the distribution due to Fj and write
398
CHAP. 13
The coupling terms are defined by (13-48), which reduces to , dS F 2M t ? v T = Ft ,M r T

+
x*
(a)
for a thin-walled section with 0 = (f> s tc. Substituting for qr and qf results in J
*2r
dS
q rq 3 ' t
T Jl
j
X3 r~ j j
dS
i
(13-72)
If X 2 is an axis of symmetry, Zf is an even function of x 3, qm is an odd function, and x 3r = 0. By analogy, x Zr = 0 if X 3 is an axis of symmetry. The definition equations for Cr, 1$, x 2r, and x 3r apply for an arbitrary thinwalled section. W hen the section is closed , we have only to modify the equa tions for < p, qr, and 7f>\ We will discuss this further in the next section. Example 13-4 --------------------------------------------------------------------------------- - Symmetrical / Section
The I section shown (Fig. E 13-4A ) has two axes of symmetry; it follows that the shear center coincides with the centroid and the warping function is o d d with respect to X 2, X 3.
Fig. E13-4A
*3
'X 2
Applying (13-67), we obtain

<p =
for web
^ - S
(a)
for flange
N ote that the sense of S is reversed for the bottom flange.
SEC. 13-7.
399
The shear flow vanishes at S' = b j 2. Applying (13-69) and starting from pt. A , we find s
-m S t dS =
------ - 1 16
b 2ht
-(f)
2 S \2"|
(b)
The distributions of 0 and < f are shown in Fig. E 13-4B , where the arrows indicate the sense of c f for + M' r .
Fig. E13-4B
b 2h ! ( M t \ 16 \ l j +
Plot o f < p
We express the cross-sectional properties in terms o f /?, f, and a shape factor : < = = b.lh ./ = _ _ (l + 2c)
ht 3
24
( h 2
(c)
1 +
{i 10
2{'
The dimensionless parameters occurring in the solution o f the differential equations for the mixed formulations are C $ and X L (see (13-55)). Using (c) and assuming a value of 1/3 for P oisson s ratio, we write
3(1 + 2 c ) '1/2
400
R E S T R A I N E D T O R S IO N -F L E X U R E O F P R I S M A T IC M E M B E R
C H A P . 13
The coefficients i, c 2 are tabulated below:

b
--------
i = h
^2
2.4 3 4.22 6.93
1
0.75 0.50
2.66
3.2
Since (f//i)2 l a n d ^ 0(1), we see that Cs sr 1. The warping parameter, I L , depends on t/h as well as L j h . This is the essential difference between open and closed cross sections. For the solid section, we found that X L = 0 ( L / h ) and, since L j h is generally large in com parison to unity, the influence of restrained warping is localizcd.t The value of XL for an open section is 0 ( L j h ) 0( t / / i ) and the effect o f warping restraint is no longer confined to a region on the order o f the depth at the end but extends further into the interior. We consider next the determination of the stresses due to restrained warping. The general expressions are
. crn =
,
J< * >
q' (C)
Using the distribution for r/; and q r shown above, the maximum values of normal and shear stress are j^nlmax 3 -Mr (0
2 h 2tc,
The shearing stress due to unrestrained torsion is obtained from

M j - f t -
3
Mr
<fu -
(g)
To gain some insight as to the relative magnitude of the various stresses, we consider a member fully restrained at one end and subjected to a torsional moment M at the other end. This problem is solved in Example 13-1. The maximum values o f the moments are , .
m a x=
MLC.S
tanh X L ) r~-- I
XL
at x = 0
(h)
M r U = C *M We substitute for the moments in (f), (g) and write the results in terms o f a , the maximum
t We defined the b oundary layer length, L b. (sec (1 3 -2 4 ). ( 1 3 - 2 5 ) ) as
SEC. 13-7.
401
shear stress for unrestrained torsion: k 'iiL = f o C i ' 2 tanh l L ) a u m

0) Mt
O n, =
y
U
= K c2
= KCs)'
The variation o f these coefficients with b / h is shown below:
< * 4 1 0.75 0.50 2 2.11 2.31
1.5
1.67 2
Since Cs, c 3. and c4 are o f 0(1), it follows that k u U = o(ff")
t
K | , = -O K ,)
(j)
The additional shearing stress (crr ls) is small in comparison to the unrestrained value. Therefore, it is reasonable to neglect the terms in the complementary energy density due to </i. i.e., to take C r = 0 and C, * 1 for an open section. We will show in the next section that this assumption is not valid for a closed section.
Example 1 3 - 5 --------------------------------------------------------------------------------------Channel Section

We consider next the channel section shown in Fig. E 13-5A . Since X 2 is an axis of symmetry, x 3 = x }r 0. The expressions for the location of the centroid, shear center,
Xi
Fig. E13-5A
S h ear
c e n te r
Xj
T
)
402 and 1 2 are
CHAP. 13
A = b ^
1 + 2 1 + 6
e = b ----- - = b e th 3 = (1 + 6 0 b C - ~ h
(a)
The dimensionless coefficient? is essentially constant, as the following table shows:

h
1.00 0.75 0.50
0.429 0.409 0.375
We determine 4> by applying (13-67) to the three segments. Taking S as indicated above, and noting that 0 is odd with respect to X 2, we ob tain :
Segment 1 - 2
h
P sc ~ Z
j L
bh ( Segment 2 - 3
Psc = + e
.S \
(b)
bh (
2 S \_
The distribution is plotted in Fig. E 13-5B . Since? < 1/2, the maximum value of occurs at point 1 (and 4). We generate next the distribution o f Z f', starting at point 1 (since q 0 at that point) and using (b):
Segment I - 2
s = .o
<f>tdS = bht f I -
1 S'2) --------- >

2 b )
(c)
Segment 2 - 3
(
\
\ J2
bhte ( 2
V
S h
= ( f | + ~ Y ~ [ ~ S + T The distribution o f qr is plotted in Fig. E 13-5C .
SEC. 13-7.
APPUCATION TO THIN-WALLED OPEN CROSS SECTIONS
403
Fig. E 1 3 -5 B
- ( ! - e)
e < i
+ (1 Distribution of
- e )
Fig. E13-5C
- \ 2 M'r
Distribution of q rj The expressions for J , I#, C r, Cs and exam ple: /I +

( 2 + h
aL
are written in the same form as for the previous
= ht
= h 5r
3c
C,
({ 1 AY j w
,12 VI + 6 f + 2c)(3 + 16c
+ 42c2 + 363)1 52(2 + 3c)2

c
/ (d)
XL = C 1/2
GO 1/2
;)( ) =
Cjw U /
f? V L N )
404
CHAP. 13
The following table shows the variation of and c , with b j h for GJEr = 3/8. i.e., Poisson's ratio equal to 1/3. Note that the comments made for the wide-flange section also apply to the channel section.
____
Ci
?2
1 0 .7 5 0 .5 0
2 .3 3 2 .6 5 3 .4
2 .5 5 3 .3 9 5 .2 4
In order to evaluate x 2n we need the flexural shear stress distribution due to F 3. Applying (11-106) leads to
Segment 1 -2
Segment 2 - 3
Tf* = bht
(e)
- - [h - S)
St ^
The distribution is plotted in Fig. E I3 -5 D ; the arrows indicate the sense of q for a + F 3.
SEC. 13-8.
405
Substituting for if, c / 3\ and the cross-sectional constants in (13-72) leads to
*2r =
(1 + 2^)( 0.2 + 52 + 63) =2(1 + 6J(2 + 3 a The coefficient is of order unity, as the following table shows: c3 1 0.5 0.926 1.03
(f)
In Example 13-1, we determined expressions for the coordinates of the center of twist in terms of x jr and Q . It is of interest to evaluate these expressions for this cross section. The coordinates at x = 0 (see (33-59), (13-60)) are *f 3- 0 x 2 = x 2 - X2r|cjjx=0
(g)
bUo = 1
I + Substituting for C s, x 2r, and evaluating x 2.
1
c,
----- - v, (A + e) b \ ------- 1 " \ l + 2c 1 + 6c we obtain

A
(h)
= * 2(1 - i s )
7 7
(i)
<=5 -
- _
< 5 1 <51^,4
(j)
Typical values arc listed below: c 1 0.5 C d 0.476 0.625 0.836 0.485
13-8.
APPLICATION TO THIN-WALLED CLOSED CROSS SECTIONS
We treat first a single closed cell and then generalize the procedure for m ulti cell sections. Consider the section shown in Fig. 13-6. The + S direction is from X 2 toward A^3 (corresponding to a rotation about the + X x direction). Using the results developed in Sec. 11-4, the shear flow for unrestrained to r sion is . M'j2A c c = ~T T d s (a)
o r f
406
CHAP. 13
where A is the area enclosed by the centerline curve. The shearing stress varies linearly over the thickness,
ffls
M \{
j i
C\
J O [open "IQ (closed
(b)
but the open-section term has a zero resultant.
*3
- X- >
F ig. 1 3 - 6 . N o ta tio n fo r s in g le c lo s e d c e ll.
Substituting for qu in (13-66), taking < j> = - </;f sc, and integrating from point P lead to
rs
rs
jc
(j)
(j> p
I |
J Sp
P s dS c^-C p sc
J */> 1
(13-73)
We determ ine (j)P by enforcing

(j> (f)t d S 0
(c)
The two additional orthogonality conditions

j x 2 <pt d S
j x 3 d)l d S
= 0
(d)
are identically satisfied by definition o f the shear center, f The shear flow due to M -r is defined by (13-69), M'T q[ = 7fP +
t N o t in g th a t x 2f = dQ-i/ch, w e can w rite
rs P
(e)
<pt dS = qr P + Qt
x24 > t Is = Q3< b .s clS

W e m e re ly h a v e to id e n tify th is term a s the m o m e n t o f th e flex u ra l sh ea r str ess a b o u t th e sh ea r cen ter. S ee P r o b . 1 1 - 1 2 .
s e c . 13-8.
407
where cf is indeterm inate. O ur form ulation is based on no energy coupling between q and <f, i.e., we require (see (13- 47)) " dS O tfY = 0 J ^ N oting th at q" is constant for a single cell, and using (e), we obtain " qP = dS </S (13-75)
O Q * y
(13-74)
<D
The flexural shear flow distributions for F 2, are generated with (11-110). We merely point out here th at there is no energy coupling between qu and qf : j>q"qf ~ = 0 (f)
O ne can interpret (13-74) and (f) as requiring qf , qr to lead to no twist deform a tion, i.e., coi , ! = 0. We have expressed the flexural shear flows as (see (13-71)): n in Z 7/ c 'l F j S ^ < = - - g O > j = 2 j = 3 k 3 k ~ 2 fe)
Finally, the definition equations for the cross-sectional properties have the same form as for the open-section: Eq. 13-70 =>/*, Cr Eq. 13-72 => x 2r, x 3r Suppose X 2 is an axis of symmetry. Then, 0 is an odd function of x 3. If we take the origin for S (point p ) on the X 2 axis, < j)p = 0. Also, qr is an even func tion of x 3 and x 3(. = 0. In w hat follows, we illustrate the application of the procedure to a rectangular cross section. Example 13-6 ------------------ -------------------------------------------------------------
Rectangular Section Constant Thickness

Applying (13-73) and taking < f> = 0 at point 0
C =
shown in (Fig. E 13-6A ) leads to
2abr a+b (a)
* = a { h Ib ) s
The distribution is plotted in Fig. E 13-6B . N ote that < fi = 0 when a = b, i.e., a square section of constant thickness does not warp.
408
CHAP. 13
Z3
Fig. E13-6A
-X 2
2 a
Fig. E13-6B
We determine Q $ by integrating (a),

fa b\ S2 Qp = at a + bj 2
for segment 1 -2 (b) for segment 2 -3
and evaluate q p r with (13-75):
SEC. 13-8.
409
The distribution of q r follows from (b), (c).
?r =
- -Ji + J 3 \ 0 (d)
a b 2t f a b D = ----- ----------2 \ a 4- b
and is plotted in Fig. E13-6C . Note that + <f corresponds to t f acting in the clockwise ( S) direction for + M'T. Also, D is negative for b > a.
Fig. E13-6C
^3
We introduce a shape factor , depth

b
410
CHAP. 13
and express the various coefficients in terms of a, t. and I. The resulting relations are
f "2 J = 16a 3f [
M 4 c
1 + c
(neglecting the contribution of J)
4a r ~c2(l - C)2
_ 4 1 + 5 + 5 (2 + c3
r c
1 - C - c2 + CJ
5(1 - C ) 2
c ,=
f 2C 9 ll + y + ^
A
/
(f)
12CZ \ ) 1,1 L
^ 2 1 - = *3r 0 The variation of C r, C s, and

b a
with b / a is shown in the table below:
Cs
for - E
cc
10.43 4.41 We found
0 0.0877 0.185
0.98 1.27 1.39
Cr = 0 1 - ) w 0 = 1 + 0
't iJL = 0
(g)
for an open section. Our results for the single cell indicate that
iL 0 f ~
C l C, 1
(h)
for a closed section. We obtained a similar result for XL, using the displacement-model formulation for a solid section. Since C,. is due to the restrained shearing stress (i f), we see that shear deformation due to cf c a n n o t b e n e g l e c t e d for a closed cross section. We discuss next the determination of the normal and shearing stresses due to warping. The general expressions are
s e c . 13-8.
411
The maximum normal stress occurs at point 2 while the maximum shear stress can occur at either points 1 or 3. We consider the same problem as was treated in Example 13-4, i.e., a member fully restrained at one end and subjected to a torsional moment M at the other end. We ex press the stresses in terms of a u m, the maximum shear stress for unrestrained torsion,
M (
< r I+ 1 J \ J
(j)
which reduces to
O = M
c
t
M 2At
Ik)
since we are considering the section to be thin-walled. The maximum stresses are , = ^ -1 1^Wf u'L tanh a L v\l |mio| ai point,Z
O \ s |ina\ til I or 3 =
(i) 3 C,
G / E r_
1/2
Cj = s e e , ( - - ., \ l J W
The variation of ratio equal to 1/3.
and C 2 with height/width is shown below. We are taking Poissons
C = bja
C, (point 2)
( 2 (point 1)
C2 (point 3)
0 -1 .0 4
-1.51
0 -0 .3 5 -0 .4 6
0 + 0.44 + 0.65
For large XL, tanh XL ^ 1 and we sec that both the normal and shear stress are of the order of the unrestraincd-torsion stress. In the open section case, we found the restrainedtorsion shear stress to be of the order of (thickness/depth) times the unrestrained shear stress.
To illustrate the procedure for a multicell section, we consider the section shown in Fig. 13-7. The unrestrained-torsion analysis for this section is treated in Sec. 11-4 (see Fig. 11-11). F o r convenience, we summ arize the essential results here. We num ber the cells consecutively and take the + S sense from X 2 to X 3 for the closed segments and inward for the open segments. The total shear flow is obtained by superim posing the individual ccll flows q\, qu z. q" = 0 q = We let
Mr
for an exterior (open) segment constant lor an interior segment
, ,
<fj = - j
Cj
(b)
412
R E S T R A IN E D T O R S IO N -F L E X U R E OF PRISM AT IC M E M B E R
C H A P . 13
Fig. 1 3 -7 . N o ta tio n fo r m ix e d c ro s s s e c tio n .
The constants C l; C 2 are determined by requiring each cell to have the same twist deformation, 0)3, 1 - Enforcing (11-67), f
M u t % J S ,. G i , l s = 2a" - ' A > = 2 a r A
(c)
for each cell leads to aC = 2A where a, A are defined as aij ~~

Sj
(d)
dS ~7
1
12
(>2 \
dS t
(e)
A = { /lj, A 2} The warping function is generated by applying (13-6):

< t >=
d
- < t> ?
(13-76)
dS
(j)
psc
'jL y i
M r ) t
We start at point in cell 1 and integrate around the centerline, enforcing continuity of e and (j?Pl:

rb 4>b
=
rb
4>e
Psc dS = < j} Pl + I
- ) dS
(g)
Repeating for points C and d results in the distribution of < fi expressed in terms of 4> Pi. One can easily verify that Pi by enforcing f Jj(/> dA = j(pt dS = 0 (h)
where the integral extends over the total centerline. N ote that < f> Pi = 0 if P t is taken on an axis of symmetry. The shear flow for restrained torsion is obtained with (13-69): ~ cf = (j)t (i)
The steps are the same as for the flexural shear determ ination discussed in Sec. 11-7. We take the shear flow at points P x, P 2 as the redundants, qr\Pj = Cj and express the shear flow as qr = q0 + qc (13-78) where q0 is the open section distribution and Z /c is due to C\, C\. The dis tribution, qc, has the same form as q / ( M y j ) . We just have to replace C with C . We generate q0 by integrating (i) around the centerline, and enforcing equilibrium at the junction points. F o r example, at point b (see Fig. 13-7),
Qb
j ~ 1,2
(13-77)
qi?l>,
clb e
(j)
N ote that q0 = 0 at points P 2, e and f The redundant shear flows are evaluated by requiring no energy coupling between qu and qr which is equivalent to requiring qr to lead to no twist de formation, ci>[ j . N oting (c), we can write is qr = 0 t t Sj Finally, substituting for qr, we obtain
aCr = B
B j=
r -
7 = 1 ,2
(13-79)
is
(13- 80)
f See footnote on page 385.
414
REST R AIN E D TOR SIO N -F L EX U R E OF PRISMATIC M E M B E R
C H A P. 13
Once 4 > and 7f are known, the cross-sectional properties (/ , Cr, x 2r, x 3r) can be evaluated. Also we can readily generalize the above approach for an n-cell section. 13-9. GOVERNING EQUATIONS GEOMETRICALLY NONLINEAR RESTRAINED TORSION
In this section, we establish the governing equations for geometrically non linear restrained torsion by applying Reissners principle. This approach is a mixed formulation, i.e., one introduces expansions for both stresses and dis placements. The linear case was treated in Sec. 13-5. To extend the formulation into the geometrically nonlinear realm is straightforward. One has only to introduce the appropriate nonlinear strain-displaccment relations. O ur starting point is the stationary requirement t < 5[{j'J'(<rr E Br u V*)d(voL) JJpr u (-/(surface area)] 0 (a)
where <j, u, are independent variables, = g(u), V* = V*(ff), and p, b are prescribed. We take the displacement expansions according to (13-3) and use the straindisplacement relations for small strain and small finite rotations :% i + co2x 3 0)3X2 + j(j)
& 2 = s2 I?3 = Us3 +
i(^ 3 - S 3)
C U i(.V 2 -X2 )
(13-81)
Cl = l. 1 + i( 2 .1 + 1 0 712 & 1 . 2 + Lh , t + l h , l f ' 3 , 2

?13 = fl1 .3 + 3, 1 + 2. 1^2, 3
The in-plane strain measures (s2, e3, T23) are of 0(a;2), which is negligible according to the assumption of s m a l l f i n i t e rotations. Actually we assume C22 = <733 = <723 = 0, i.e., plane stress. Substituting for the displacements and noting the definition equations for the force parameters, the first term in (a) expands to jXfo-r </(vol.) =
+
1 + i (s2, 1 + * 3^ 1. i)2 + i( S3, 1 - x 2cai, 1 )2]

F 2 [ lls2.
1 +
U>i(u s3.
X 2 CO\ i
()]
F i [ u si, 1 +
0>l -
0 )l(U s2 , 1 +
*3 1 , i) ]
+ M 2[ o 2 1
Wl, l{l> s 2, 1 + x 3J.l)]
( 13- 82 )
+ M 3[a>3) 1 ft>i. i(Uj3, 1 ~ -^2^1.1 )] + 1 + M<pf, 1 ^ M u / + iMpwf, i + MqCOiOJi. i } d x 1

f See Eqs. 1 3 -3 3 and co rresp o n d in g fo o tn o te . W e are w ork in g w ith K irch h o ff stress and L agrangian strain here. J S ee Sec. 1 0 -3 , Eq. 1 0 -2 8 . The d isp lacem en t ex p a n sio n s assu m e sm all-fin ite ro ta tio n , i.e., sin a> a> and c o s co 1. T o be con sisten t, w e m ust use (10 -28).
SEC. 1 3 - 9 .
G O V E R N IN G E Q U A T IO N S
415
where the two additional force parameters are Mp ~ JjciiC xi 4- x\)dA M 0 = ) $ ( x 2 <ri2 + XiOysjdA The terms involving the external forces have the same form as for the linear case, but we list them again here for convenience (see (13-6)): jjjb Tu d(vol.) + j f PTu J(surface area) i 2 us2 f b l-3 us3 + m T< ^ m 2< rw \rxJ b i ui + b x > i + x >2 + m 3c/j3 4- m ^ j )dxt + + F 2us2 + p3lls3 + M Ta>i + h42to2 + M 3a>i + = (13-83)
where the end forces (the barred quantities) are defined as previously, for example. = = (Jj/h dA)x^ L etc. It remains to introduce expansions for the stresses in terms of the independent force parameters and to expand V*. In the linear case, there are 8 force measures, . . . , M 3, and M R. Two additional force measures (M P, M Q ) are present for the nonlinear case but they can be related to the previous force measures. We proceed as follows. We use the stress expansions employed for the linear case with < j> U ~ ~ T M
12
r ~X
M, M i - x 2 + --- < / >

*3 U
a ij - < *{j + i j + n\j G"ij - f W r o\j = I jMr

{j hj2p2 + hj i l " 3
(a)
MT = Mj- + M T r
A/ ji
IVI -f
where < /> , / , g, h2 and equations for M P and
M q
are functions of x 2, x 3. Introducing (a) in the definition leads to
Mp fiiFi 4- fi2 M 2 + P3 M 3 4/5i = tA I f

1 h h
+ x & lA ' x 3{xi + xj)dA
(13-84)
-1
(h
77
h
J x 2(xj 4- xf)dA
< /> (.x \ 4- x\)dA
416
CHAP. 13
andt
M q > 'i2 F 2 + 3 + x 3 h 2k) d A
+ t\\ M t
(k =
>lk = J'J'(-x:2^2 k +
2,3)
n _s
rj
ffl =
r,3 =+i / ?2
jj fe ./V
+ X 3f ( /jd A
[ ~ >
C ertain coefficients vanish if the cross section has an axis o f sym m etry.J O ne can readily verify th at
M P => ( h F , M q => 0
(1 3 -8 6 )
when the section is doubly sym m etric. F o r generality, we will retain all the term s here. T he com plem entary energy density function has the sam e form as for the linear c a se :
P * = JL
2 E \ A + M l\
(M J \
I2
2
h )
2E r \ l J
1 (F \ + 2G U + +
F \\ + t )
, +
, (13" 87)
+ 2G 7( W )
(x3fF 2 + X2rF 3)
W e have show n th a t it is quite reasonable to neglect transverse shear deform a tion due to w arping (Cr = x 2r = x 2r = 0) for a thin-w alled open section. S ubstituting E quations (13 82) (13 87) in R eissners functional and re quiring it to be stationary with respect to the seven displacem ent and eight force m easures leads to the following governing equations: Equilibrium Equations
1 + bt = 0
dxi
{-F l(s2 , 1 + * 3 ^ 1 , l) + F 2 0 ) y F 3 CO j 1M 2} + b 2 = 0
{ F i ( us3' j x zco1, 1 ) + h
(1
+ <x>iF2 - W i,iM 3} -f b 3
= 0
+ r]\[o>,)Mr,
F
(1
+ rji^ iW r.
GJ j (
+ mT
$
2 ^-si,
1 +
F z U s l <x +
^ 3 -f2, 1 +
2^
3 . 1)
{Fl(xdU s2il ~ X2U s3tl + ? ! !,!) + M2( U s2, i + 2fi2Q )u l )
+ M 3( ui3-1 + 2ft3co1;1) + M 0/^co1?1} = 0 M 2> 1 F 3 TYl2 - 0

M y
1 + F 2 + W3 = 0
j M ' j -f- i n # = 0
t See Prob. 1 3 - 1 1 . J See Prob. 1 3 - 1 2 .
SEC. 1 3-9.
GO VER NING EQ U A T IO N S
417
where
Pi Pi +
2
X2
+ X3 ( / i ) .(hear ceuier
l h t;P i
7P2
*3
+ *2
Fovce-Displacetnent Relations Fi = 1, 1 + K l , 1 + zwf 3 l 1 + a>!, t(.T3U,2, ! - X2Us3, 1 + i/îCOi, 1 ) 4E 1 G T - + T - + ~J T M r Ai A23

u x2,
i ~ Cf>3 + a > i [ l / s3j 1
l_ fr 2 3 X*?!^ -J + - ~ + - J - M t s3, 1 + G ^23 ^3 J Mr = l(l + ? ?> l)
w 2
W l [ S2 , 1
C 0 \'ip 2 ]
GJ M2 Mi
(13-88) 0)2,1 + Cl)i( t ( t/s2, 1 + f)>0>{, 1 )

(03,1 + <*>1 . i ( MS3,
E h
.r
El 3
/^ 3 l,l)
M,
[C rM'r 4- x 3rF 2 + x 2rF.?] j + Wi. i(l 4 fJiCOi)
Boundary Conditions ( + for ,xx = L, for x , 0) ! prescribed or F j = F i zy s2 prescribed or F i(u s2, i + x 3a>i, t) 4- F 2 <iF3 - aJi>iM 2 = F 2 u.,3 prescribed or f i ( u s3yl - x 2oju i ) 4- F 3 4- a>tF2 - cou l M 3 = + F 3 coi prescribed or F I(x 3ws2, 1 x 2us3tl 4- /5i(w1> )L )
C0i(/j2E2 4- J]3E3) 4- (1 4- if{oji)Mj 4- (1 4 t]$0$Mr + M2{ iiS 2, 1 + 2^ 2 ojj,i) 4- M 3( u 53j1 4- 2/?3 coi; i) + ci,iP<t>J ^ < t> = x M r
4 o)2 prescribed or M 2 = A ?2 oj3 prescribed or M 3 M 3 / prescribed or M ^ These equations simplify considerably when the cross section is symm etrical and transverse shear deform ation is neglected.! We discuss the general solution of (13-88) in C hapter 18. T he following exam ple treats one of the simplest cases, a m em ber subjected to an axial force and torsional m om ent.
t See Prob. 1 3 -1 3 .
418
CHAP. 13
Example 13-7
We consider a prismatic member (see Fig. E 13-7A ) having a doubly symmetric cross section, fully restrained at one end and loaded by an axial force P and torsional moment M . We are interested here in evaluating the influence o f axial force on the torsional behavior. The linear solution (with no axial force) was derived in Example 13 -1 . Fig. E 13-7A
-l -
Equilibrium Equations (symmetrical cross section and no distributed load)

M \ = M,. k Fu i = 0
(a)
(,W, + F , / ? ^ , . , ) = 0
dxi
Force-Displacement Relations
M'{
=
GJoii. i
GJ M \ = ( / + !.,)
^r
M* ~ E J + f , J-, = A E u t j 4-
(b)
Boundary Conditions
.X j =0
Uy 0) 1 f M# 0 M
Fx -
Mi + PiFX 0)t, 1
(c)
Integrating the last two equations in (a) and noting the boundary conditions, lead to
F i = const = P
Mi + ftF.w,., = const = M
The first equilibrium equation takes the form
(d)
/, ii - H2 f =
fi 2M
GJ{ 1 + P)
(e)
SEC. 13-9. where
GOVERNING EQUATIONS
419
P =
P0 1 GJ GJ
Ph GJA
1+ P
/!>-
E rL I + C,.( 1 + P)
This expression reduces to Equation (g) of Sec. 1 3 -6 when P - 0. Once / is known, we can determine the rotation by integrating (d), which expands to '
Oh, 1 GJ (
_ 1 1 + P + V r
(f)
when we substitute for M , using (b). The general solution is.

f C x cosh f i x + C2 sinh f i x M G J { 1 + P)
ai
GJ
1 + P +
)'
(g)
= C3 + Mx - > ----- --- {C t sinh u x + C 2 cosh f i x} 4 . I f -Cr(i -----= + P) j j.iCr t (We drop the subscript on Aj. for convenience.) Finally, specializing (g) for these particular boundary conditions result in
M
GJ
/ =
oh =
G J ( 1 + P) M GJ() + P)
{ 1 + cosh fix - tanh f i L sinh u x }

- L (
(h) ) (sinh / i x + (I cosh ^.x)tanli f i L ]
A t + G(i + P)
These equations reduce to (13- 57) when P 0. A tensile force ( P > 0) increases the torsional stiffness whereas a compressive force (P < 0) decreases the stiffness. Equation (h) shows that the limiting value of P is 1. We let P r represent the critical axial force and < T t, the corresponding axial stress
GJA
"77"
GJ h
(')
In order for a cr to be less than the yield stress, ( / / M must be small with respect to unity. As an illustration, consider the section shown in Fig. E 13-7B . The various coefficients (see Example 13-4) are
l = b!h
4 20
RESTRAINED T O R SIO N -FL EX U RE O F PRISMATIC M EMBER
CHA P. 13
Fig. E 13-7B
X-K
Xt
and
&a G
REFERENCES
V on K arm an , T., and N. B. C hristf.n s e n : " Methods of Analysis of Torsion with Variable Twist, J. Ae r o . Set. , pp. 110-124. April 1944. 2. T im o sh en k o , S. J.: Theory of Bending, Torsion a n d Buckling of Thin-Walled Members o f Open Cross Scclion, J. Franklin Inst., pp. 559-609, 1945. 3. V o n Karman, T.. a n d W. C. Chien: Torsion w ith V a r ia b le T w is t," J. Aer o . Sci., Vol. 13, No. 10, pp. 503-510, October 1946. 4. B enscoter , S. U . : Secondary Stresses in Thin-Walled Beam s with Closed Cross Sections, NAC A-TN 2529, Washington, D. C 1951. 5. B enscoter , S. U.: A Theory of Torsion Bending for Multicell Beams, 7. Appl. M e c h . , Vol 21, No. 1, 1954. 6. V lasov . V. Z .: Thin W a l l e d El a st i c B e a ms , Israel Program for Scientific Translations, Office of Technical Services, U.S. Dept, o f Commerce, Washington. D.C. 1961. 7. H eilig , R .: Der Schuberverformungseinfluss auf die Wolbkrafttorsion von Stuben mit offenem Profil, D e r Stal l Iban, April 1961. 8. H eilig , R .: Beitrag zur Theorie der Kastcntniger beliebiger Querschnittsform,' D e r St ahl bau, December 1961. 9. O d e n , J. T .: M e c h a n i c s o f Elast ic S t r uc t ure s , McGraw-Hill. New York, J967. 10. K ollbrun ner , C. F ., and K. B asler : Torsion in S t r u c t u r e s , Springer-Verlag. Berlin, 1969. 11. W a sh izu , K .: Vari at i onal M e t h o d s in El as t i c i t y a n d P l a s t i c i t y , Pergamon Press, 1968. 12. M aisel , B. I.: Review of Literature Related to the Analysis and Design of ThinWalled Beams, Technical Report 440, Cement, and Concrete Association, London, July 1970.
1.
PROBLEMS 13. 14. 15. 16. 17. 18.
421
D abrowski, R .: Gekriimmte diinnwandige Tracer,' Springer-Verlag, Berlin, 1968. G ala MBOS. T. V .: St r u c t u r a l M e m b e r s a n d Frames , Prentice-Hal I, 1968. B leich , F .: Buckli ng St r e n g t h o f M e t a l Structures, McGraw-Hill, New York, 1952. B urgermeister , G., and H. S tfxjp: S t abi l i t at Theorie, Part 1. Akademie Verlag, Berlin, 1957. C h i l v e r , A. H .: T h i n - W a l l e d Structures, Cbatto and Windus, London, 1967. Reisser, E .: Note on Torsion with Variable Twist, ./. Appt. Me c h . , Vol. 23, No. 2, pp. 315-316, June 1956.
PROBLEMS 13-1. The shear stress distribution due to F2 is given by (see (11-95))
0 " l2 7 ~ <p2r. 2 Ii < 7 1 3 ------- q>2 r , 3 *3
F *-r
where < p2r are flexural warping functions which satisfy V202,. = - x 2 : = 0
On d(f>2r a
(in A) (on M
This result applies when the cross section is assumed to be rigid with respect to in-plane deformation. The coordinate of the shear center is defined by *31 sc = Xy = Show that (a) reduces to A ' 3 = 71 ff
J j
(x24> 2r. 3 - X3 (j}2r<2)dA
(a)
x 2 <j>, d A
where (j> t is the St. Venant torsional warping function. H int : See Prob. 11-11 and Equation (11-97). 13-2. Verify (13-40) and (13-44). 13-3. This problem reviews the subject of the chapter in two aspects, (a) No coupling between the unrestrained and restrained torsional dis tribution requires 2 + G'\30r i 3)dA = 0 (a) The unrestrained torsional shear stress distribution for twist about the shear center (see Sec. 13-3, Equation (b)) is given by
1712 Mr
Mu <rh= / [ ^ 3 f X 3 + ,V ?]
(b)
-x2 x 2]
The restrained torsional shear stress distribution is determined from (13-39). Verify that M r T M R when 'c and (a) is enforced.
4 22
RESTRAINED TOR SION -FLEX U RE OF PRISMATIC MEM BER
C H A P. 13
(b)
When the cross section is thin-walled, (a) and (b) take the form
(c) (d)
where \psc\ is the perpendicular distance from the shear center to the tangent at the centerline. Equation (d) follows from (11-29) and Prob. 11-4. We determine cf from (13-43). Finally, the force param eters for the thin-walled case are defined as M'r - $qrPs,-dS Mr = Verify that M'r ~ M R when < /> 1. 2. 3. Open section Closed section Mixed section Consider the following cases:
13-4. Specialize (13-57) for AL 1 and compare M r vs. M u. Also evaluate at x = L and compare with the unrestrained value. 13-5. Refer to Examples 12-2 and 13-2. Discuss how you would modify the member force-displacement relations developed in Example 12-2 to account for restrained torsion. Consider C , ~ 1. x 2r = x.v 0, and (a) warping restrained at both ends (b) warping restrained only at x = L 13-6. Refer to Example 13-2. Determine the translations of the shear center. Consider the cross section fixed at x = 0. Discuss how the solution has to be modified when the cross section at x = L is restrained against translation. 13-7. Starting with the force-deformation relations based on the mixed formulation (13-49), derive the member force-displacement relations (see Example 12-2). Consider no warping at the end sections and take Q, = 4-1. Specialize for (a) symmetrical cross section (b) no shear deformation due to restrained torsion and flexurearbitrary cross section. 13-8. Consider a thin-walled section comprising discrete elements of d i f f e r e n t material properties ( l i , G ) . Discuss how the displacement and mixed formulations have to be modified to account for variable material properties. N o t e : The unrestrained torsion and flexural stress distributions are treated in Prob. 11-14 and 12-1. 13-9. Determine the distribution of 4 > , cf. and expressions for 1$, C,., x 2r, x 3r for the cross sections shown in parts a and b and part c -d of the accom panying sketch (four different sets of data). 13 10. Determine < fi and qr for the section shown. 13-11. Using the flexural shear distributions listed in Prob. 13-1, show that rh =
PROBLEMS
423
Prob. 1 3 -9
_L
T h
-2 h ----(a)
1
(b)
T
h
S e e part c,
Id)
l------- h ------ -L
2h
(c)
Prob. 13-10
' O
t
0 * 2
t t a 2a
424
CHAP. 13
H in t : O ne can write ^3 Y~ Also show that m 1 P1
^_ < ^2r + A '5 V'iMA
13-12. Specialize Equations (13-84) and (13-85) for the case where the cross section is symmetrical with respect to the X 2 axis. Utilize
* 3) H 0( x 2,
x 3)dA
where H e is an even function and Ha an odd function of x 3. Evaluate the co efficients for the channel section of Example 13-5. Finally, specialize the equations for a doubly symmetric section. 13-13. Specialize (13-88) for a doubly symmetrical cross section. Then specialize further for negligible transverse shear deformation due to flexure and warping. The symmetry reductions are
*2 ~
= 0 ft 2 Pi Q
%3
*2.- = *3r = 0 -1/^2 3 0

>fl =
> 12
> /3
= '/" =
13-14. Consider the two following problems involving doubly symmetric cross section. (a) Establish linearized incremental equations by operating on (13-88) and retaining only linear terms in the displacement increments. Specialize for a doubly symmetric cross section (see Prob. 13-12). (b) Consider the case where the cross section is doubly symmetric and the initial state is pure compression (F{ P). Determine the critical load with respect to torsional buckling for the following boundary conditions: 1. 2. coi = f = 0 oh = ~ = 0 dx at x 0, L at a' = 0, L (restrained warping) (unrestrained warping)
N eutral equilibrium (buckling) is defined as the existence of a nontrivial solution of the linearized incremental equations for the same external load. One sets Fl = - P
it2
= 3
c o i co2
= co.-,
and determines the value of P for which a nontrivial solution which satisfies the boundary conditions is possible. Employ the notation introduced in Example 13-7. 13-15. Determine the form of V, the strain energy density function (strain energy per unit length along the centroidal axis), expressed in terms of displace ments. Assume no initial strain but allow for geometric nonlinearity. Note that V V* when there is no initial strain.
14

14-1. INTRODUCTION: GEOMETRICAL RELATIONS A member is said to be planar if 1. 2. 3. The ccntroidal axis is a plane curve. The plane containing the centroidal axis also contains one of the principal inertia axes for the cross section. The shear center axis coincides with or is parallel to the centroidal axis. However, the present discussion will be limited to the case where the shear center, axis lies in the plane containing the centroidal axis.
We consider the ccntroidal axis to be defined with respect to a global reference frame having directions X , and X 2- This is shown in Fig. 14-1. The orthogonal unit vectors defining the orientation of the local frame (Xj, Y 2) at a point are denoted by t j , t2, where points in the positive tangent direction and x t2 t3. Item 2 requires Y z to be a principal inertia axis for the cross section.
425
424
RESTRAINED TORSION FLEXURE OF PRISMATIC MEMBER
CHAP. 13
H in t : O ne can write (xf V 202r + x \ Vl-)dA Also show that n3

=
702
13-12. Specialize E quations (13-84) and (13-85) for the case w here the cross section is sym m etrical with respect to the X 2 axis. U tilize
x 3 ) H 0( x 2, X i ) d A =
where H e is an even function and H an o dd function of x 3. E valuate the co efficients for the channel section of Exam ple 1 3-5. Finally, specialize the equations for a doubly sym m etric section. 13-13. Specialize (13-88) for a doubly sym m etrical cross section. T hen specialize further for negligible transverse shear defo rm atio n due to flexure and w arping. T he sym m etry reductions are
X 2 X3 = 0 X 2r ~ Xjr = 0
@ 2 ~ Pi ~ P < /> 0
1/^23 0
Vl = >h = > f\ = ff 1 = 0 13-14. C onsider the two following problem s involving doubly sym m etric cross section. (a) Establish linearized increm ental equations by o p eratin g on (13-88) and retaining only linear term s in the displacem ent increm ents. Specialize for a doubly sym m etric cross section (see Prob. 13-12). (b) C onsider the case where the cross section is doubly sym m etric and the initial state is pure com pression P). D eterm ine the critical load with respect to to rsio n al buckling for the following b o u n d ary conditions: 1. 2. C 0\ / 0 df coi = -j- = 0 ax at x = 0, L at .V = 0, L (restrained w arping) (unrestrained w arping)
N eutral equilibrium (buckling) is defined as the existence o f a nontrivial solution of the linearized increm ental eq u atio n s for the sam e external load. O ne sets Fl = -P
n2 = w3 = coi = <y2 = <y3 = / = 0

and determ ines the value o f P for which a nontrivial so lu tio n which satisfies the boundary conditions is possible. E m ploy the n o ta tio n introduced in Exam ple 13-7. 13-15. D eterm ine the form o f V, the strain energy density function (strain energy per unit length along the centroidal axis), expressed in term s of displace ments. Assume no initial strain but allow for geom etric nonlinearity. N o te th at V = V* when there is no initial strain.
14

14-1. INTRODUCTION: GEOMETRICAL RELATIONS A m em ber is said to be p la n a r if 1. 2. 3. T he centroidal axis is a plane curve. T he plane containing the centroidal axis also contains one of the principal inertia axes for the cross section. T he shear center axis coincidcs w ith o r is parallel to the centroidal axis. H ow ever, the present discussion will be lim ited lo the case w here the shear center axis lies in the plane containing the centroidal axis.
W e consider the centroidal axis to be defined with respect to a global reference fram e having directions X x and X 2. This is show n in Fig. 14-1. The o rth o g o n al unit vectors defining the o rien tatio n of the local frame ( Y ly Y 2) at a p o in t are denoted by t u t 2, w here tt points in the positive tangent direction and t y x t2 = f3. Item 2 requires Y 2 to be a principal in ertia axis for the cross section.
Fig. 1 4 -1 . G eom etrical notation fo r plane curve.
425
426
PLANAR DEFORMATION OF A PLANAR MEMBER
CHAP. 14
By definition, t (1 4 -1 ) Since we are taking t2 according to t l x t2 = h , it follows th a t to dx2_ . dxi _ (1 4 -2 )
I s '* + I s 12
T he differentiation form ulas for the unit vectors are dti d i2 _ where 1 d tx d2x l d x 2 d2x 2 d x { R ~ dS ' (l ~ ~~dST 'dS + J s T ~dS A ccording to this definition, R is negative when dt^jdS points in the negative f2 direction, e.g., for segm ent A B in Fig. 1 4 -1 . O ne could take t 2 = n, the unit norm al vector defined by 1 dh (14-4) d t { dS dS rath er th an according to h x h = h but this choice is inconvenient w hen there is a reversal in curvature. Also, this definition degenerates a t an inflection point, i.e., when dt/dS = 0. If the sense of the cu rv atu re is constant, one can always orient the X r X 2 fram e so th a t t2 coincidcs with n, to avoid w orking with a negative R. T o com plete the geom etrical treatm ent, we consider the general p aram etric representation for the curve defining the centroidal axis, x i = *i{y) x 2 = x 2(y) where y is a param eter. T he differential arc length is related to dy by dS = + (1 4 -5 ) 1 _
(14-3)
Is ~ R h 1 I S ~ ~ R t{
fd x 2
1/2
dy = a dy
(1 4 -6 )
Jy
A ccording to this definition, the -f S sense coincides w ith the direction of

j We summarize here for convenience the essential geometric relations for a plane curve which are developed in Chapter 4.
SEC. 14-2.
FORCE-EQUILIBRIUM EQUATIONS
427
increasing y. U sing (14-6), the expressions for t u t2, and l/ R in terms of y are
(14-7)
1 a
3
fi2.X j dx2 d 2x 2 dx i ' d y2 dy + d y2 dy
A planar m em ber subjected to in-plane forces ( X l - X 2 plane for our notation) will experience only in-plane deform ation. In w hat follows, we develop the governing equations for planar deform ation of an arbitrary planar m ember. This form ulation is restricted to the linear geometric case. The two basic solution procedures, namely, the displacem ent and forcc m ethods, are described and applied to a circular member. Wc also present a simplified form ulation (M arguerres equations) which is valid for a shallow member. Finally, we include a discussion of num erical integration techniques, since one m ust resort to num erical integration when the cross section is not constant. 14-2. FORCE-EQUILIBRIUM EQUATIONS
The notation associated with a positive norm al cross section, i.e., a cross scction whose outw ard norm al points in the -F S direction, is shown in Fig. 14-2. We use the same notation as for the prism atic case, except that now the vector
Fig. 1 4 -2 . Force and moment components acting on a positive cross section.
428
CHAP. 14
com ponents are with respect to the local fram e Y2, Y3) rath er th an the basic frame ( X lf X 2, X 3). The cross-sectional properties are defined by A = j j dy2 dy2 = J{ dA h = jtty z ) 2 dA I z = J'j( v.^2 ilA (14-8)
Since Y2, Y3 pass through the centroid and are principal directions, it follows that JJy 2 dA = Jj> 3 dA = $ j y 2y 3 dA = 0 (14-9) W hen the m em ber is plan ar ( X r X 2 plane) and is subjected to a p lan ar loading, F 3 Ml i = M 2 ~ 0 (1410) In this case, we w ork with reduced expressions for F+ and M+ (see Fig. 14-3) and drop the subscript on M 3: F+ F J i + F 2tz M+ = M 3t 3 = M ?3 N ote that ?3 is constant for a plan ar mem ber.
l'l 11
(14-11)
Xy
Fig. 14-3. F o rc e a nd m o m e n t c o m p o n e n ts in p la n a r b e h a v io r.
T o establish the force-equilibrium equations, we consider the differential volum e elem ent shown in Fig. 14-4. W e define b and in as the statically equiva lent externa] force and m om ent vectors per unit arc length acting at the centroid. F o r equilibrium , the resultant force and m om ent vectors m ust vanish. These conditions lead to the following vector differential equilibrium equations: dF + b = 0 dS dM ,
~dS~
(14-12)
4- m + fj x F+ = 0
SEC. 1 4 -3 .
P R IN C IP LE O F V IR T U A L FO R C ES
429
We expand b and m in term s o f the unit vectors for the local fram e: b = b1t l + b2t 2 m mil, (14-13)
Introducing the com ponent expansions in (1 4 -1 2), and using the differentiation formulas for the unit vectors (14-3), lead to the following scalar differential equilibrium equations:
+ T
+ h> = 0
( , 4 - I4)
JM + r2 + m = 0 Note th at the force-equilibrium equations are coupled due to the curvature. The m om ent equilibrium equation has the form as for the prism atic case. J c/6
Fig. 14-4. D iffe re n tia l e le m e n t fo r e q u ilib r iu m a n a ly s is .
The positive sense of the end forces is shown in Fig. 14-5. We work with com ponents referred to the local frame at each end. The end forces are related to the stress resultants and stress couples by ^Bj ~ Fils* m b = M\sB - - n Ma = -M \
(14-15)
14-3.
FORCE-DISPLACEMENT RELATIONS; PRINCIPLE OF VIRTUAL FORCES
We establish the force-displacem ent relations by applying the principal of virtual forces to a differential element. The procedure is the sam e as for the
430
CHAP. 14
prismatic case described in Sec. 12-3, except that now we w ork with displace m ent com ponents referred to the local frame at each point. We define u and To as u / ; = equivalent! rigid-body translation vector at the centroid. a) = a)jt) = equivalent rigid-body rotation vector (14-16)
F or planar deformation, only u u u2 and c% are finite, and the terms involving 3, cou and co2 can be deleted: u = uxi l + u2i 2
CO
(14-17)
= CJ,)^1 2
CO t
The positive sense of the displacement com ponents is shown in Fig. 14-6.
U 2ti
Fig. 14-6. Definition of d isp lacem en t m easu res,

f Equivalence refers to work. See (12-8).
SEC. 14-3.
431
We define V* as the com plem entary energy per unit arc length. F o r planar deformation, V* V* (F ls Fz, M). O ne determines V* by taking expansions for the stresses in terms of F t, F 2, M, substituting in the com plem entary energy density, and integrating with respect to the cross-sectional coordinates y 2, >'3. We will discuss the determ ination of V* later. Specializing the three-dim ensional principle of virtual forces for the one dimensional elastic case, and writing cV* dV"* dV* d V * = A F, + ~ ~ A F 2 + AM
Tr\
of 2
cm
= et AFj + < ? 2 A F2 + k A M lead to the one-dim ensional form |s(e i A Fi + < ? 2 A F2 + k AM )dS - V d, AP,
(14-18)
(14-19)
where di is a displacement measure and P t is the force m easure corresponding to dt. The virtual-force system (AFj, AF2, AM , AP ;) m ust be statically permis sible, i.e., it must satisfy the one-dimensional equilibrium equations.
-A M + +
d(
dS\
) f
Fig. 1 4 -7 . V irtu a l fo rc e s y s te m
We apply (14-19) to the differential element shown in Fig. 14-7. The virtual force system m ust satisfy the force-equilibrium equations (14-17), dS =0 (a)
A M + I ty x AT + 0 dS f du da) / i \ _ A F + - + A M + + AM + *
E dl APi =
dS
432
CHAP. 14
and then substituting in (14-19) results in the following relations between the force and displacement parameters: cV* uz diii d f\ = I s _ R dV* (Xll2 u 1 e2 = = -7F + + - o U tJ R w i dV* do k = dM ll \r * r >
(14- 20)
We interpret e x as an average extension, e2 as an average transverse shear deformation, and k as a bending deformation. Actually, k is the relative rotation of adjacent cross sections. In what follows, we discuss the determination of V*. Consider the differential volume element shown in Fig. 14-8. The vector defining the arc QQt is dr2 , Noting that dr - - = a t! dv [dr d t2 d t3\
dy
= 6
for a planar member, (a) can be written as dS2 = \QQX \ = (1 dy = (1 - ^ dS (c)
By definition, V* is the complementary energy per unit length along the centroidal axis. Substituting for dS2 in the general definition, we obtain V* dS = j j V* dSz dy2 dy3 y2,y* V | j V * fl dA
(14-21)
V* -
In general, V* = V* ( g i u a 12, a l3). We select suitable expansions for the stress components in terms of f u Fz, M, expand V*, and integrate over the cross section. The only restriction on the stress expansions is that they satisfy the definition equations for the stress resultants and couples identically: jJfTj j dA = Fj j 2 dA = F2 Jj'(>i 3 dA 0 i dA = 0 ~ [ \ y 2o lx dA = M JJ(?2ffl3 - y3(Tii)dA = 0 (a)
SEC. 1 4 -3 .
43 3
The most convenient choice for a x{ is the linear expansion^

< r u = ' J - >2 y 04-22)
where I = I 3. A logical choice for a iy (when the cross section is thin-walled) is the distribution predicted by the engineering theory of flexural shear stress distribution described in Sec. 11-7:
Oxj = \q(F2) q = F^ (14-23)
where t denotes the local thickness, and q is the flexural shear flow due to F 2. Both expansions satisfy (a).
Xi
In what follows, we consider the material to be linearly elastic. The comple mentary energy density is given by V* = g V ii + + ^ 13) fa)
where is the initial extensional strain. Substituting (a) in (14-21) and taking the stresses according to (14-22), (14-23) results in the following expression
t T his app lies for a h o m o g e n e o u s beam . C o m p o site beam s are m ore con ven ien tly treated with the app roach described in the next section.
434
P L A N A R D E F O R M A T IO N OF A P L A N A R M E M B E R
CH AP. 14
for V*\ V = + + ^ 4 -2 4 )
where
e ?( 1
k g L4
. Vi R dA
k =
-I
~T
y 2ei
/*
/ I
/R
J 2 i _ i f
- =
If the section is symmetrical with respect to the Y2 axis, I* = / and ,4* = A 2. The deform ation-force relations corresponding to this choice for V* are o Fi M e' = e ' + AE + AER e-> = dih GAj
, n
duA dS 03 dS
d (0
u2 R (14-25)
dS + R "
F l M
k = k + + -----= AER El*
N ote that the axial force and m om ent are coupled, due to the curvature. Inverting (14-25) leads to expressions for the forces in term s of the d efo rm atio n s:
(*-*> M = E lA R( 1 - 6) (et - ef) + ~ - s (k - fc) - I 1 IR '

y \ dA
(14-26)
F 2 GA*e 2 1* 1 < 5= AR AR" We observe th at
= 0
(a)
where p is the radius of gyration and d is the depth of the cross section. F o r example, I ^2 A R 2 ~ 12R 2 ^
for a rectangular cross section. Then, 5 is of the ord er of (d /R 2) and can be neglected when (d/R)2 1. A curved m em ber is said to be thin when 0(d/R) 1 and thick when 0(d/R)2 1. We set < 5 = 0 for a thick member. The thinness assum ption is introduced
SEC. 14-4.
435
by neglecting y 2/'R with respect to unity in the expression for the differential arc length, i.e., by taking dSl X U dA J fj V*
'a
<14- 27> '
Assuming a curved m em ber to be thin is equivalent to using the expression for V* developed for a prism atic m ember. The approxim ate form of (14-25) for a thin m em ber is , 0 , FI *1 = e i + A E e7 F2 du2 dUl dS l h R co (14-28)
u + dS R
k = k + M = EI dS T o com plete the treatm ent of the linear elastic case, we list the expanded forms o f the principle of virtual forces for thick and thin m embers. N ote that these expressions are based on a linear variation in norm al stress over the cross section. Thick M em ber M \ + AE + A E R
r
Fn (14-29) = y d , AP,
AF[ + G A $ A F l
Thin Member F, A F 2 A2 2 e + J e ) A / 1 + G '* + ( k + ^ j } A M \ d S = d{ AP, (14-30)
14-4.
FORCE-DISPLACEMENT RELATIONS DISPLACEMENT EXPANSION APPROACH; PRINCIPLE OF VIRTUAL DISPLACEMENTS
In the variational procedure for establishing one-dim ensional force-displace m ent relations, it is not necessary to analyze the deform ation, i.e., to determ ine the strains at a point. O ne has only to introduce suitable expansions for the stress com ponents in term s of the one-dim ensional force param eters. N ow , we can also establish force-displacem ent relations by starting with expansions for the displacem ent com ponents in term s of one-dim ensional displacem ent p a ram eters a nd determ ining the corresponding strain distribution. We express the
436
CHAP. 14
stresses in terms of the displacement param eters using the stress-strain relations, and then substitute the stress expansions in the definition equations for F t , F 2, and M. The effect of transverse shear deform ation is usually neglected in this approach. To determine the strain distribution, we m ust first analyze the deformation at a point. This step is described in detail below. Figure 14-9 shows the initial position of two orthogonal line elements, QQr and QQ2, at a point (y, y2, y3). The vectors defining these elements are
(14-31)
We use a prime superscript to denote quantities associated with the deformed position of the member, which is shown in Fig. 14-10; for example: r ' ?'(>') = position vector to point P(y) in the deformed position (point P). t \ = tangent vector to the deformed centroidal axis. V2 ~ position vector to <2(y, y2, v3) in the deformed position (point Q,').
Fig. 1 4 -9 . Initial geometry for orthogonal curvilinear line elements.
SEC. 14 4.
437
From Fig. 14-10, and noting (14-31):
(14-32)
The analysis of strain consists of determ ining the extensions and change in angle between the line elements. We denote the extensional strains by j (j = 1, 2) and the shearing strain by y, 2- The general expressions are
(14-33) sin y12 = 7 ICTQiI iQ'Qzl Now, we restrict this discussion to small strain. Substituting for the deformed vectors and neglecting strains with respect to unity, (14 33) expands to
(14-34)
2
^ (
u 2
a 2 dy
dv2
The nonlinear terms are associated with the rotation of the tangent vector. Neglecting these terms corresponds to neglecting the difference between the deformed and undeformed geometry, i.e., to assum ing linear geometry. The next step involves introducing an expansion for u2 in terms of y 2. We express u2 as a linear function of y 2. (14-35) where co = (o(y) and u = ul t 1 + u2t 2 = u(y) (14-36)
is the displacement vector for a point on the centroidal axis. Equation (14-35) implies that a norm al cross section remains a plane after deformation. O ne can interpret co as the rotation o f the cross section in the direction from t l tow ard 12- This notation is illustrated in Fig. 14-11. In w hat follows, we consider only linear geometry. Substituting for u2, taking y = S, and evaluating the derivatives lead to the following strain expansions:
438
CHAP. 14
i =
1 - y 2/R
(e i - y 2
duv u2 = - - = et w -o dS R
e2
fi2 o 1 712 = i - yJR ei
= 1 7 1 21 > 2 = o
(1 4 -3 7 )
dS
The vanishing of e2 is due to our choice for u2. One could include an addi tional linear term, f$yzt 2. This would give s2 P an<3 additional terms in the
Fig. 14-10. D e fo rm e d g e o m e try fo r o rth o g o n a l c u rv ilin e a r lin e e le m e n ts .

u 2f2
SEC. 14 4.
439
expressions for and y 12. N ote that the assum ption th at a norm al cross section remains plane does not lead to a linear variation in extensional strain over the depth when the mem ber is curved. We introduce the assum ption of negligible transverse deform ation by setting e2 = 0. The resulting expressions for oj and k in terms of ux and u2 are e -> = 0 du2 dS dco dS R d 2u7 d (uC\ d S2 ' d S \ R j (14-38)
co
W hen transverse shear deform ation is neglected, one must determine F 2 using the m oment-equilibrium equation. The next step involves expressing F l, F 2, and M in terms of the one-dim en sional deformation param eters e xe2 and k. In what follows, we consider the material to be linearly elastic and take the stress-strain relations for eru , a l2 a s :t = E(ei ~ ei) n X2 = Gy12 (a) Substituting for t, y l2, using (14-37), =
1
E
- yJR
(et - y 2k ) - Eef (14-39)

2
G
O 1 7
12
-~yJR
and then evaluating F u F 2i and M, we obtain F , - Eex F 2 = Ge2 dA 1 - v2/R dA J - Ek ' J i dA y - y 2/ & (14-40) + Ek _ y \ dA + y 2t ? dA - E $dA
M = -E e1
2 dA
1 - y 2/R
y 2/R
The various integrals can be expressed in terms of only one integral by using the identity 1 yJR (a) 1 - yz/R 1
y /R = 1 +
and noting that Y3 is a centroidal axis: JJ>2 dA = 0

f The reiation for o n is exact only when o 2 2 * ^33 = 0- We generally neglect member.
(b)
< x 33 for a
440
CHAP. 14
O ne can easily show th at dA 1 JJ ' y 2 dA i V y \ dA I y 2j R 1 (14-41)
1 - y 2/R ~ R r =
F o r com pleteness, we list the inverted form of (14-40), e, = eV + e2 = F2 GA X /? where / Fj EA M EAR
A'7
A (t +
? M V
-
(14-42) y_2 dA R)
-
ei = T fc - r
W r, V 1
dA
a r
T he expressions for e l are identical with the result (see (14 25)) obtained with the variational approach. However, the result for k differs in the coefficient for M . This difference (/' or /") is due to the nonlinear expansion used for < 71 Example 14-1 ---------------------------------------------------------------------------------------We determine /' for the rectangular cross section shown in Fig. E14-1.
r =
yi dA 1 - y 2/R
C dl2 y\ dy2 }-<i2 1 ~ yzfR
= R2bd + R 3b In
(l + d \\ 2R
(a)
\l ~ 2r !
To obtain a more tractable form, we expand the log terms, using
SEC. 14-4.
441
This series converges for Ixj < 1. Then / d \ 'l + 2R In T 1 ~ 2R !

an d
12R )
5 \2 RJ
7 \2R J
(c)
(d)
Fig. E14-1
The relations listed above involve exact integrals. Now, when the m em ber is thick, we neglect {y2/R )2 with respect to unity. This assum ption is introduced by taking 1 (a)
i -
y 2/ R
( )
in the expansions for a u , a l2 and >n
v f d /1
(14-43)
A + y i ) dA
1 - y 2/R
i rr
442
CHAP. 14
To be consistent, vve must also neglect I '/A R 2 with respect to unity in the expression for A 2 and I". W hen the mem ber is thin, we neglect y 2/R with respect to unity. 1 1 1 - yJR
Z ii
fi0
(1 4 _ 44)
E < ?12 I f ~ 2 It is of interest to establish the one-dimensional form of the principle of virtual displacements corresponding to the linear displacement expansion used in this development. The general three-dimensional form for an orthogonal coordinate system is (see Sec. 10-6): J fJ fru + + a n Syi 2 + ' ' -Kvol.) = M (a)
where Pt represents an external force quantity and d; is the displacement quantity corresponding to P{. We consider only r.y and y12 to be finite, and express the differential volume in terms of the cross-sectional coordinates y 2, v3 and arc length along the ccntroidal axes (see Fig. 14-9): d(vol.) dS2 d \> 2 dv3 = f 1
\ R j
dS dv-, d y3
(b)
Then (a) reduces to
n J
1
~ dS = ?,- Ad, (14-45)
We take (14-45) as the form of the principle of virtual displacements for planar deformation. The strains corresponding to a linear expansion for displacements and linear geometry are defined by (14-37), which are listed below for convenience: Si = T-----" V2k) 1- yJR v 1 1 r> 111 1 - v2/R 2 duv u2 e, = dS " R du2 , dS ~ R dai dS
(C)
(d)
Substituting for et , y12 and using the definition equations for F\, F2, and M,
SEC. 14-4.
443
we obtain Js[F i Set + F 2 5e2 + M 8k~]dS - P f Ad, (14-46) This result depends only on the strain expansions, i.e., (c). O ne can apply it for the geometrically nonlinear case, provided th at (c) are taken as defining the strain distribution over the cross section. We use the principle of virtual displacements to establish consistent forceequilibrium equations. O ne starts with one-dim ensional deform ation-displacem ent relations, substitutes in (14-46), and integrates the left-hand side by parts. Equating coefficients of the displacement param eters leads to a set of force equilibrium equations and boundary conditions that are consistent with the geo metrical assum ptions introduced in establishing the deform ation-displacem ent relations. The following example illustrates this application. Example 14-2 -------------------------------------------------------------------------------------The assumption of negligible transverse shear deformation is introduced by setting e 2 equal to zero. This leads to an expression for the rotation, co, in terms of the translation components, du 2 ui
" - y s- + K
and the relations for negligible transverse shear deformation reduce to JsCF, S e t + M * k \ d S =
du, l = ^ ^ dm u2
< >
Ad t
(b)
d fdu2
Mi' + R
(c)
i = ' d S =Z d S \ d S
Substituting for Aa> and the strain variations,

Au i d
4" T
dS
+ 7 s i,,!
R
< 5 < ? i = J - Au, - - - An2

r, d2 ok ^ Aw? dS1
(d)
1 d -----:Awi RdS
and integrating by parts, the left- and right-hand sides of (b) expand to
444 and
CHAP. 14
'Ss f
sA I
V
' A | b i 4- ) + A u 2 | b 2 -----777 ) (' d S
m \
dmX)
Rj
d b ))
+ I F bi H r~ ) Awbi + ( F p 2 + n i g ) ^ llu2 + M B A ( )
R/
J
\
\ds
(f)
( -
M a\
( _
(du2
H --- Al(,4i + ^ ^ 2
m AJ A UA2 + M a A |
The consistent equilibrium equations and boundary conditions for negligible transverse shear deformation follow by equating corresponding coefficients of the displacement variations in (e) and (f);
SA < S < s T
+ - + ------- + />, + = 0
dS R dS R
dF,
1 dM
F,
R
-------+ b2 dS2
d 2M
dm dS
= 0
S * $A
ux u2
prescribed or prescribed or prescribed or
Fi = dM
F.4i (g)
= F A2 - m
= fflA
(IS
--dS
S = SB
prescribed or
u2 F l = Fn
-i ,
prescribed or prescribed or
dM
dS
= F R2 in
M Mn
dS
One can obtain (g) by solving the last equation in (14-14) for F 2 and substituting in the first two equations. Suppose we neglect u,//? in the expression for co:
du 2 co ^ ----d-u2 k T dS2
,f
(h)
This assum ption! is generally referred to as M u s h t a r f s a p p r o x i m a t i o n . The equilibrium equations for the tangential direction reduce to
'f See Ref. 5.
SEC. 14-5.
CARTESIAN FORMULATION
445
The other equilibrium equation and the boundary conditions are not changed. Using (h) instead of (a) eliminates the shear term, F2/R , in the tangential force-equilibrium equation.
14-5.
W e consider the case where the equation defining the centroidal axis has the form x 2 = /(* i). The geom etrical relations for this param etric representation are obtained by taking y = in (14-7). They are sum m arized below f for convenience an d the notation is show n in Fig. 14-12: dS = a. d x { tan
6
df (1 3 C % 1 cos
6
a
tx =
1 a
1 dr a dxy
1 a r w f i + \ = U x t 2 = 73 d2 f j _ dx \
3/2
(14-47)
R dt y dS
1 + I 't2 R 2
f )J \ dfx L dl 2 dS I
-
In the previous form ulation, we w orked with displacem ent com ponents and external force com ponents referred to the local frame. An alternate approach, originally suggested by M arguerre,t involves w orking with com ponents re ferred to the basic frame rather than the local frame. The resulting expressions differ, and it is therefore of interest to describe this approach in detail. We start with the determ ination of the force-equilibrium equations. C onsider the differential elem ent shown in Fig. 14-13. The vector equilibrium equations are -- + p = 0
dXi
dM dr + x F+ + h = 0
t See Prob. 14-1. t See Ref. 6.
(14-48)
446
CHAP.
ii
Fig. 14-12. N o ta tio n fo r C a rte s ia n fo rm u la tio n .
X 2
Fig. 1 4 - 1 3 . Differential element for equilibrium analysis.
SEC. 14-5.
447
where p, h are the external applied force and m om ent vcctors per unit projected length, i.e., per unit ,k, . They are related to b and m (see Fig. 14-4) by p J x , = b dS (ab)dx 1 h d x x m dS (<y.m)dx1 Substituting for the force and m om ent vectors, F + F J i + F 2^2 M + Ml 3 h ~ hi 3 P = P J i + P it 2 Ny F j cos 0 F 2 sin N 2 = F t sin
6 0
(14-49)
-)- N 2 72 (14-50)
+ F 2 cos 0
the equilibrium equations expand to (i7! cos 0 - F 2 sin 0) = p { ax ^ < / (Fj sin # + F 2 cos 0) = p2 dx j
6
ciXi dN2
(14-51)
- - (d - ^ - + h | - F 2 = - N , sin 0 + N z cos a j
We restrict this treatm ent to an elastic m aterial and establish the forcedisplacement relations, using the principle of virtual forces, j*, a rfP* d x y = JX l [e* A F t + e2 A F2 + /< AAf]a c/.Yv = AP, (a)
where P* F* (F l5 F 2 M) is the com plem entary energy per unit arc length. Consider the differential element shown in Fig. 14-14. The virtual-force sys tem is statically permissible, i.e., it satisfies the force-equilibrium equations identically: t -A F + = 0 dx j dx i Expanding AF;, . { du _\ A F r I - - - + at t x c o l + A M h da> dx, dxy
(c)
1
(b)
AM + a ti x A F+ - 6
4 AFi =
and then substituting for the displacement and ro tatio n vectors, u = v j y + v2 ~ i2

C O COl3 = cot 3
(14-52)
448
CHAP. 14
we obtain V di APi = ( A N , + A N 2 -j- a AF 2co + A M dx, \ dxv dx1 d x lJ

(d)
Finally, substituting for N u N 2 in term s o f F u F 2 and equating coefficients of the force increm ents result in dvy wt' dt) cos2 9 ~7~ + sin 9 cos 9 2 dXy w ; e-, =
dV*
dv* dv, sin 9 cos 9 ~ ~ + cos 1 9 dxt ax j dF7 , dV* dw k = 8M = T cos p dx.
CO
(14-53)
The m em ber is said to be shallow when 9 2 1. O ne introduces this assum p tion by setting df cos 9 x 1 sin 9 % tan 9 (14-54) dx i in (14-50), which relate the cartesian and local forces.
A A /+
dx.
Fig. 1 4 - 1 4 . V irtual force system.
M arguerres equations are obtained by assum ing the m em ber is shallow and, in addition, neglecting the co n trib u tio n of F 2 in the expression for N ^
SEC. 14-6.
DISPLACEMENT METHOD OF SOLUTION
449
M arguerre starts with N i Fj N 2 % F2 + and the resulting equations are

+ Pi -
(a)
ft
dXy + rfAf = dx j m
(14-55) dvl d f dv 2 dxx + d \\ dxj
dFl
dP* c/u, < ? 2 = - - = ~ - - to (?F2 dxi d P* ~dM cho dxx
O ne step rem ains, namely, to establish the b o u ndary conditions. The general conditions are or Ni Vl v2 M or or N2 a> prcscribcd at each end (14-56)
We obtain the appropriate boundary conditions for the various cases considered above by substituting for N u N 2 and ox F o r example, the boundary conditions for the M arguerre form ulation are l or or
CO
Fi f
2
prescribed at each end
(14-57)
or
14-6.
DISPLACEMENT METHOD OF SOLUTION CIRCULAR MEMBER
The displacem ent m ethod involves solving the system of governing dif ferential equations which, for the planar case, consist of three force-equilibrium equations and three force-displacem ent equations. If the applied loads are independent o f the displacem ents, we can first solve the force equilibrium equations and then integrate the force-displacem ent relations. This m ethod is quite straightforw ard for the prism atic case since stretching and flexure are uncoupled. However, it is usually quite difficult to apply when the m em ber is
450
CHAP. 14
curved (except when it is circular) or the cross section varies. In w hat follows, we illustrate the application of the displacem ent m ethod to a circular member having a constant cross section, starting with 1. 2. the exact equations (based on stress expansions) for a thick m em ber M arguerres equations for a thin m em ber
The results obtained for this sim ple geom etry provide us with some insight as to the relative im portance of transverse shear deform ation and stretching deform a tion versus bending deform ation. W hen the centroidal axis is a circular segment, R = const, and the equations simplify somewhat. It is convenient to take the polar angle 0 as the independent variable in this case. We list the governing equations below for convenience and sum m arize the notation in Fig. 14-15: ,d F x dM . / T Eq. (14-19) dQ2
R F ,
m + X
= R %
R ~
do
(14-58)
1 dM
_ 1 (dtii
A E R ~ R WW E ,. (14-25) k = fc + F, M M .
CO
- ih (14-59)
1 dm
A E R + El* ~ J JO
Solution o f the Fovce-Equilibrium Equations We consider the external forces to be independent of the displacements. Integrating the first equilibrium equation, we have RFl = - M - R2 b l + I dO + C, (a)
where C x is an integration constant. Substituting for F^ in the second equation results in a second-order differential equation for M: d 2M dd 2 The general solution of (b) is M = C v + C 2 cos 9 + C 3 sin 8 + M p (14-60) 1 dm h> - R M ~
b + j ) de
(b)
SEC. 14-6.
451
where M p denotes the particular solution due to the external distributed loading and C 2, C 3 are constants. O nce M is know n, we find F \ using (a) and F z from the m om ent equilibrium equation. The resulting expressions are F{ F7 - J - (C 2 cos R
6
+ C3 sin
+ M p) R
J (b t +
dQ (14-61)
dM \ - 1. = ^ I C 2 sin 9 + C 3 cos 9 -I----- I m R \ dd J
F2
dS ~ Rdd
Fig. 1 4 -1 5 . N o ta tio n fo r c ir c u la r m e m b e r.
Integration o f the Force-Displacement Relations We start with (14-59) w ritten in a slightly rearranged form:
le ~ 2 = Re + I E < m + RF,)
du 2
1 0
RF 2 = GXf + dm _, R dO - R k " + EI*
(a)
To determ ine u x and u2, w6 transform the first two equations to
452
CHAP. 14
and d 2 u2
,
" 2
* R ga* 2 R II I* ~ AR1 X $ = dF2 ~dj dF2 d.0 = 0 * > _ *? + e g . M El* <14- 63)
= 1- < ê W e have previously sh o w n f th a t < 5 C is o f the ord er of (d /R )2. It is reasonable to neglect de with respect to 1 bu t we will retain it in order to keep track of the influence of extensional deform ation. W e solve (14-63) for u2, determ ine u t from (14-62), an d co from the second eq u atio n in (a),
= " e l f + R ' S 1 +
<14_64)
This leads to three additional integration constants. T he six co n stan ts are determ ined by enforcing the three b o u n d ary conditions at each end. V arious loading conditions are treated in the following examples. Example 1 4 - 3 ---------------------------------------------------------------------------------------Consider a member (Fig. IZ14-3) fixed at the negative end (A ) and subjected only to FBl at the right end (B). The boundary conditions for this case are F2 = M = 0 u, a^ = co = 0 at 0 = 0* at 0 0 (a)
Specializing the force solution for no external distributed loading and enforcing the boundary conditions at B, we obtain Fi -- F bi cos(0B 0) Fz = Fm sin(0fl - 6 ) M - R F m ( l - cos (t)B (b)
-6 ))
To simplify the analysis, we suppose there is no initial deformation. Using (b). ift takes the form rpr l>i a2
cos[0 B - 0)]
(c)
where
5s
~GA'*R2
\Rj
1 3e + Ss
(d)
Cl2 C l j "i- 3$ =
t See Sec. 14-3, Eq. (14-26).
SEC. 14-6.
453
N ote that < 5, is associated with t r a n s v e r s e s h e a r d e f o r ma t i o n . Substituting for i/> in (14 63) and integrating, we obtain
u 2 = Q cos 8 + C3 sin 0 + r ~ ~
a <
a , ^mR* tl*
a } -) - 9 sin(0e 6)
(e)
The solution for
follows from (14-62): ! = C4 sin 0 C 5 cos 0 + C 6
FB 1 P3 f 0 + ~ [0 cos(0# - (9) + sin(0fl - 0)] /*
(f)
Next, we determine co using (14-64),
C l) -f- ------ { 6 + a, sin(0B - 0)}

R El*
C6 , F B l R 2
(g)
Finally, the constants are found by enforcing the displacement boundary conditions at 0 = 0:
Cd
E l*
T juR*
~Ej*~ sin
r Q = -a j a sin ()
(h)
T o determine the relative importance of stretching and shear deformation versus bending deformation, we evaluate the displacements at 9 0 H and write the resulting expressions
Fig. E 14-3
Constant cross section
454
CHAP. 14
in the following form:

F OJB
=
R. ^ Sin
0 S)U + /?! (5,,)
hr
uBi = ~ j T f ~
~ 2 sin Ob + 2 sin 9 B cos 0,,)i\ + b 2 d e + b 3 6 S)
U B2 ~
F>R3 Bl ' (L cos 8 b

E l*
- i s i n 2 0B)(1 (i)
=
b2 = b
sin 9a
0 R - sin On i O B + 2 sin 0 B \ sin 0 B cos 9 B
1 9 S 2 sin 9 B + j sin 0B cos &B

2 &b -
_____ H P b - sin 9 b cos 0 n)
2 sin
i sin 2 0 B
+ 5 sin 0 B cos . I + cos 9 1 cos 9 b
I cos 9 S 5 Usin 2 9 b
The coefficients (6 (, . . . , 6 4) arc of order unity or less when 9 n is not small with respect to unity, i.e., when the segment is not shallow. Also, < 5 and 6S are of order { d / R )2. It follows that the displacements due to stretching and shear deformation are of order (<//R)2 times the displacement due to bending deformation for a no n s h a l l o w member. To investigate the shallow case, we replace the trigometric terms in (i) by their Taylor series expansions, sin 9 9 ( 1
\
01
04
04
6 + 120
92
cos 9 = - 1 sin 9 cos 9 = 0
+ ------ 2 24 - B1 + - - fl4 J
U)
/ / - r +02 2
r -
and neglect 6 2 B with respect to unity. The resulting expressions are
lB ~ BJ U S2 - '
Now,
v SI1 s 2 r ,
El*
Bl'
E l'
F f l l S 3 \9 g
1 I* 6 + AS2
* 5
m
L? 1
/* j + m
20 + 3 G A f S 2
El*
(k)
E l*
[2
4 + G A t S 2j
I*
AS2 ~
dY
\S
(1)
GA$S2
E ,t Q 2
SEC. 14-6. For example.
455
I* AS2 * E ll*
1 fd \l
12 I s E 2 /d u \v f d \. 2 ,u \SJ
----- -7 ------- GA%S
10 G \ S J
* 0-26
(m)
for a rectangular section and v = 0.3. Since ( d / S ) 2 J for a member, we can neglect the transverse shear terms in uBl, uB2 and the stretching term in ojb. However, we must retain the stretching term in uBl since it is of the same order as the bending term. The appropriate expression for uBl is
Ur I = ------- I - + E l * \2 0
FBlS 3 ( 0\ , I* )
------r
A S 1}
/ t
(n)
In sum, we have shown that the percentage of error due to neglecting stretching and transverse shear deformation is of the order of ( d f R ) 2 for a nonshallow circular member. If the member is shallow (6B < ^ 1 5 we cannot neglect stretching deformation. Actually, the stretching term dominates when the member is quite shallow. The error due to ne glecting transverse shear deformation for the shallow case is still only of the order of
{ d / R ) 2.
Example 14-4 ------------------ - --------------------- ----- ; --------The internal force distributions due to in Fig. E 1 4-4 are given by
F i *= Fi M =
F B2
acting oil the cantilever member shown
F a 2 sin (Q -
0)
Ftil
c o s ( 0 B 0) 0)
(a)
Fb2 R sin(0 -
We suppose the member is not s h a l l o w and neglect stretching and shear deformation. The force-displacement relations reduce to (we set A = A 2 = co in (14-59))
du,
dO du 2 Te +
- ~
- u2 Re'i
2
1 (b)
RM
-
do)
To =
d 2 u*> dO2
+w
R2 Rer + 1 E l*
Eliminating u l from the first two equations, we obtain
- + u2 =
z
dux
R 2k v -
M
(c)
~dB
= u2 + R e ? 1 (du - ( + 2/, R [ dO
C 'J -
We determine u 2, then u u and finally co. N ote that (c) corresponds to (14-62), (14-63) and (14-64) with A = A 2 = ce. The final expressions (for no initial deformation or support
456 movement) are
CHAP. 14
P jj^ > 3
U2 = ~2f / *
cos(0b ~ 0) ~ sin 0 cos 0B} { - 2 cos 0 B + cos cos 0 6 sin(0B - 0 ) 1 - c o s ( d B 0)} (d)
M i =
F i?3
(0 =
{ c o s (0 b - 6) - COS 0*}
Example 14-5
We analyze the shallow parabolic member shown in Fig. E 1 4 -5 using Marguerres equations. We consider the member to be thin and neglect transverse shear deformation. Taking / = a x \ / 2 and p , = m 0, the governing equations (see (14-55) and (1 4 -5 7 )) reduce to
l d
dxv d 2M
0
(a)
2 - ciFi - p 2 = 0 dx j
dM Fl ~ ~ dx,
n Fi dv i du? <i - e? 4- = + ax, - /I t rfx, dxi dv2 co (IXy
(b) - 7u.Xj T
* - ** + 7 7 iSi
i, u2, w prescribed at x , = 0
Ni = F, = WB 1
dM N 2 ---- , ---- h flx ,/', = O f a t x , = L
(c)
x.
M = 0 Integrating (a) and using the boundary conditions at x , - L, we obtain F, = N

M {L
Pi
*i)2
fliVB 1 (L2 - x?)

ax, A'
(d)
F i = P i { L - x ,) -
We suppose e? = k c = 0 to simplify the discussion. Integrating the moment-curvature relation, /
d2i;2
*1>2 - -V f) (e)
SEC. 14-6.
457
Fig. E 1 4 -4
Fig. E14-5
F-,
wwwwww
-x,
458
CHAP. 14
and noting that v 2 = d u 2/ d x l = 0 at Xi = 0 lead to the solution for v 2,

Pi E J v2 =
2 \2
[ LVi . -
j * ' , , ] . * 12' 7
1,1
2^.2
(0
The axial displacement is determined by integrating the extensional strain displacement relation,
dv,
dxy
Fi
AE
dv2
1 d\'i (g)
Vi
ap2 / I (~ 2 E /\3
1 r L + _ i ^ 4 _ i x? - 7 Lxt + T~ 1y ) + Nfli l_/i 6EJ \ 5 is X ' 4
We express the last term in (g) as
~ AE ;
iii,+ f^ Y d W s y 1 \ 6 J \l) \ l ) 5\ l a = 2 h/L 2
(h)
Now, (i) Then
a2 L * ( A } _
fh V
(j)
6 VI J
3V /V
and we see that this term dominates when h is larger with respect to the cross-sectional depth.
14-7.
O ur starting point is the principle of virtual forces restricted to planar deformation, AFj 4- e 2 AF 2 + k AM)dS - 7lk AR k = c/( AP; (14-65)
where the virtual-force system is statically permissible, dk represents a support movement, and A R k is the corresponding reaction increment. The relations between the deformation measures (eh e2, k) and the internal forces (F ,, F z, M) depend on the m aterial properties and on whether one employs stress or dis placement expansions. This discussion is limited to a linearly elastic material but one should note that (14-65) is valid for arbitrary material. F o r con venience, we list the force-deformation relations below. The notation for internal force quantities is shown in Fig. 14-3. Arbitrary Linearly Elastic Member ei =
o
+ T F + EAR AE (14-66)
Fy e 2 = f GA 2
* -
AER
- + "
El
SEC. 1 4 -7 .
459
where H, A 2, and 1 are defined by (14-24) for the stress-expansion approach and (14-42) for the displacem ent-expansion approach. Thin Linearly Elastic Member = ei + Ft AE (14-67) M El
OA 2
k = k +
where A 2, e, k are the same as for a prism atic member. W hen the m em ber is not shallow, it is reasonable to neglect stretching and transverse shear deform ation. As shown in Example 14-3, this approxim ation introduces a percentage error of 0 (d/R)2. Form ally, one sets A A 2 co. If the m em ber is shallow, we can still neglect transverse shear deform ation but we m ust include stretching deform ation. The basic steps involved in applying the force m ethod to a curved m em ber are the same as for the prism atic case. However, the algebra is usually m ore complicated, due to the geometry. We will discuss first the determ ination of the displacem ent at a point. To determ ine the displacem ent at Q in the direction defined by ~ tQ, we apply an external virtual force, APqtQ, generate a statically determ inate system of internal forces and reactions corresponding to AP Q, A F j = F j ' QA P Q AM = M ^q APq A Rk R k q &P q and substitute in (14-65): dQ = Jjyfo F 1iQ + e 2 F ltQ + /uV/ Q)WS Q\ (14-69) u = 1,2) (14-68)
This expression is valid for an arbitrary m aterial. We set e 2 = 0 if transverse shear deform ation is negligible and e} = if stretching deform ation is negligible. Example 14-6
W e consider the thin linearly elastic circular segment shown in Fig. E 14-6A . We suppose the member is n o t shallow and n e g l e c t s t r e t c h i n g and t r a n s v e r s e shear deformation. The reactions are the end forces at A for this example, and (14-69) expands to
dQ =
j* ( e ^ F ^ ' Q + ^ k + - - ) M , a ) cIS + i ^ i F a i . q + Ua 2 F a 2 . q + & a M a , q
(a)
In what follows, we illustrate the application o f (a)
460
CHAP. 14 Fig. E 14-6A
Expressions for Displacements at B
To determine w B1, we take APQ AFni. The internal virtual-force system corresponds to Fbi = +1. It is convenient to work with rj 0B 0 as the independent variable rather than 6 . The force-influence coefficients (FLQ, F 2.q , M, Q) follow directly from Fig. E14-6B:
F i , q *= F i j
+ 1 f iUo, - + 1 = cc s i/
F z, q = sin r] M :q = i?(l cos tj) Substituting (b) in (a) results in the following general expression for uin w bi = jR J f r, cos t] + R
0B
(b)
/ n
(1 cos r})d> j
(c)
+ uA i cos
-f
T ia 2 sin 0B
+ (bAR{ 1 - cos 0B)
Once the loading is specified, we can evaluate the integral. Terms involving the support displacements define the rigid body displacement at B. Taking APQ = AFb2, AM B leads to expression for uB2 and coB. We list them below for future reference: ''O b U B2 ~ R 0 M e sin t] + R ( lc + j sin i i f d r j
uA t
sin 0B+
u AZ cos 0B+ coA R
sin 0B (d)
O )/} = (O y j, + R
jo
Solution for a Concentrated Loading at an Arbitrary Interior Point
We consider an arbitrary force vector, Pc , and moment, M c, applied at point C as shown in Fig. E14-6C.
SEC. 14-7.
461
Fig. E 14-6B
Fig. E14-8C
462
CHAP. 14
Pc Pc\h + Pci 12 Me = McF 3
(C)
The expressions for the displacements at B due to an external loading are obtained by specializing (c) and (d) for no initial deformation or support movement and noting that
M M
= 0
rj
= -R-PciCl cos[t] f/c)]
+ RPc2
< r\c sin(f/ rjc) +
Mq
> / ^ >]c
(f)
The solution for constant I is

p cir
3 (
U B1 El +
El
0 C sin 0C -
sin 0B + sin t]c + cos i]c + ~ sin Bc cos 0 B

9 C sin 0 D
ec
PR3
MIr dR 2 El
0C 1 1 cos 9 C 4- -- sin > jc - - sin
(0C + sin t]c sin 0B) 10 C sin fjc - -1 sm 0 C sm 8 H N cos 0 B + cos fjc -(g)
u B2
PgR3
' El
+ + coB
P ciR 3 A
E i
(2
1 cos '7c ~ 2 Si
00
O S
R 2M C El
(cos rjc ~ cos 0 B)
R 2Pc\
EI
. R 2P C2 n ( KM. (6> c - sin 0C ) + - (1 - cos 6 C) + Or EI EI
If we take point C to coincidc with B, i]c 0 and 0 C 0 B. The resulting equations relate the displacement at B due to forces applied at B in the directions of the local frame at B and can be interpreted as member forcc-deformation relations. It is convenient to express these relations in matrix form:
Ua f
11Bl
(h)
- 2 sin 0B
+ jsin 6 b c o s Qb~ \
^ UB2 (X > B
R Symmetrical EI
K2[l cos 9b ~ isin 2 0 B ] R1 {Qb ~ sin 0 B cos
R[0B - sin 08]
^01
PB2 ' Mp
R[ 1 O b
c o s 0 a]
We call fB the
member flexibility m atrix.
We describe next the application of the principle of virtual forces in the analysis of a statically indeterm inate p lan ar m em ber. Let the m em ber be in determ inate to the rth degree and let Z l 5. . . , Z,. represent the force redundants. U sing the equilibrium equations, we express the internal forces and reactions in
SEC. 14-7.
463
terms of the applied loads and the force redundants: Y Fi F 10 + Y _, F \ ,v Z k r = F i.o + I

r
F 2 , uZ k (14-70)
k= 1
Substituting the virtual force system corresponding to AZ; (which is statically permissible and self-equilibrating) in (14-65) and letting j range from 1 to r lead to the com patibility equations relating the actual deform ations: js(e i ^ l . j + e 2 ^ 2, j + k M j)ilS tljRi' j = 0 (a)
W hen the m aterial is linearly elastic, the com patibility equations take the form
^ f jkZ k = A , [ j 1 , . . . , r)
(14-71)
We set 7 /, A 2 A 2, and 1jA R = 0 for a thin member. N ote that f jk is the displacem ent of the prim ary structure in the direction of Zj due to a unit value of Z k. Also, A _, is the actual displacem ent of the point of application of Z j minus the displacem ent of the prim ary structure in the direc tion of Z j due to support movement, initial deform ation, and the prescribed external forces. Example 14-7 -------------------------------------------------------------------------Consider the symmetrical closed ring shown in Fig. E l4 -7 . From symmetry,
464
CHAP. 14
We take the moment at 6 0 as the force redundant. To simplify the algebra, we suppose the member is thin and neglect stretching and shear deformation. The compatibility equa tions reduces to
f u Z i = A!
f u =
(b) JS &
A, = - I
Note that / u is the relative rotation (X ) due to a unit value of Z t and Aj is the relative rotation (X ) due to the applied load. Equation (b) states that the n e t relative rotation must vanish. Fig. E 14-7
r M y , F1
Now,
M o = R{ 1 2
cos 0)
0 < 9
(c) 2
We consider / to be constant. Then, (b) reduces to
SEC. 14-7.
465
Because of symmetry, we need to integrate over only a quarter of the ring. Finally, the total moment is
(c)
The axial and shear force variations are given by

F, --- cos
0
(f)
F, = -----sin 9 When the equation defining the centroidal axis is expressed in the form x 2 = /(a'j), it is more convenient to work with force and displacement quan tities referred to the basic frame rather than to the local frame, i.e., to use the cartesian formulation developed in Sec. (14 5). The cartesian notation is sum marized in Fig. 14-16. *2
Fig. 1 4-16 . N otation fo r C a rte sia n fo rm u la tio n .
The geometrical quantities and relations between the internal force com ponents are tan
6
= f dx i
cos 6 F j = jVj cos 6 + N 2 sin 6 Fz = Ny sin 0 + N 2 cos 9
</ s =
(14-72)
466
CHAP. 14
We first find N u N 2 and then determine F u F2. To obtain the equations for the cartesian case, we just have to replace dS by d x je o s 0 in the general ex pressions ((14-69) and (14-71)). In what follows, we suppose the member is thin and linearly elastic. When the member is not shallow, we can neglect the stretching and transverse shear deformation terms. The equations for this case reduce t o :
Displacement at Point Q
dq
( V l .c + (fc
Compatibility Equations
If,A
= Aj cos 9 elF \,j + ( k + - J - f j M .j dxt cos 0 (14-74)
We can express the terms involving F l-() in terms of /V, ,, and /V2 (, since = cos 0 N l (1 + sin 0 /V2,o Then, f
J.X, U
(a )
f
jxt
<
(14-75)
One must generally resort to numerical integration in order to evaluate the integrals, due to the presence of the term 1/cos 9. When the member is shallow, 0 2 1, and we can approximate (14-72) with cos 0 1 sin 9 as tan ds % dx j f (14-76)
2
F t JV4 + f ' N
Fj = - f'N , + N 2 We cannot neglect the stretching deformation term in this case. However, it is reasonable to take F 1 & N^. We also introduced this assumption in the devel opment of M arguerres equations. The equations for the shallow case with negligible transverse shear deformation and F, ^ A', have the forms listed below :
Displacement at Point Q
dQ ~
K e ?
+ %e
) N i q + ( k
dx, - R , . a t i,
+ & ) '*
(14-77)
SEC. 14-7.
467
Compatibility Equation
Zfz> = a ,
fik 4 ; = 2 X j 3, l' Example 14-8 -------Consider the two-hinged arch shown in Fig. E14-8A. We work with reaction com ponents referred to the basic frame and take the horizontal reaction at B as the force redundant.
(14-78)
*1
+(*+^ K
dxx
Xl,V2
Fig. E 1 4 -8 A
Primary Structure
We must carry out two force analyses on the primary structure (Fig. E14-8B), one for the external forces (condition Zj = 0) and the other for Z, = I. The results are displayed in Figs. E14-8C and D, respectively.
Fig. E 1 4 - 8 B
468
CHAP, 14
F ig. E 1 4 -8 C
Vi
h MN,
F ig. E 1 4 -8 D
Compatibility Equation
We suppose the member is not shallow. The compatibility equations for from ( 1 4 - 7 4 ) : follow
f n Z i = A, fix = AI 1
0 El COS 0
(a)
,, , M0\ M , *?(*!., + /'iV2. ,) + fc +
E / 7 -os 0
SEC. 14-7.
469
Using the results listed above, the various terms in (a) expand to
,/n -
] J f - i
Jo E l I
'
12 i'/x ,
L A,_ cos 0
v-.. T L Ai, 1 > ~ ~
hh^ U 42 + t'Bl + ~ < -'B 2
(b
1 0 E l cos 0
- M 0M j. c/a-i
rL
------ I f x. If P.Xi 1d x , El cos 0 V ^ 7 \ ^ 7

J, El cos 0 V
< ( + H + * ( / - r I , ) i . / ft \ / /> \
} \
Oncc the integrals are evaluated, we can determine Z x from
- A -i
(c)
P'inally. the total forces are obtained by superposition of the two loadings:
= t y .o + Z ^ , , , 7 = 1 .2
M = M . o + z iM .i
R -i ~ Ri.o + R4 = Zj
/ = I. 2, 3
(d)
To evaluate the vertical displacement at point Q, we apply a unit vertical load at Q on the p r i m a r y structure and determine the required internal forces and reactions plotted in Fig. E14-8E.
470
CHAP. 14
Applying (14-73), we obtain
V Q 2 & A2 +
* Q 1
2 ~ ^ 2) Y j
eJ' d x L -
C L
eif dx1
+
)
* -* Jo
X qt
,
(. o dXi X| ( + 1:1 / ---< 2 \ E l) cos n 0~ A

M \ dxi
E l j cos 0
(e)
h i } cos 0
A numerical procedure for evaluating these integrals is described in the next section.
Example 14-9 -------------------------------------------------------------------------------------The symmetrical nonshallow two-hinged parabolic arch shown in Fig. E 1 4 -9 A is sub jected to a uniform load per unit h o r i z o n t a l length, that is, per unit x t. The equation for the centroidal axis is V 2 4/l/
7 \x
(a)
where h is the elevation at mid-span (x, = L j 2). W e take the horizontal reaction at the right end as the force redundant and consider only bending deformation. Figures E 14-9B and C carry through an analysis parallel to that of the preceding example.
Determination of Z
and Total Forces
The equation for Z x follows from (14-74):
Z l
.. _
A,
1
rL
M. 0M . t
dx,
E I cos 0
C L
8h
E I cos 0
(b)
N ote that this result is valid for an arbitrary variation of E I . Finally, the total forces are
pL2
N2 = N
M M
2,0
0
+
+
Z t N 2_ ! =
Z j M . , =
p(
(c)
Since M = 0, the deformed shape of the arch coincides with the initial shape when a x i a l deformation is neglected. It follows that (c) also apply for the f i x e d n o n s h a l l o w case. When the arch is s h a l l o w, the effect of axial deformation cannot be ncglected. The expression for Z , follows from (14-78):
SEC. 14-7.
471
Fig. E 1 4 -9 A
Fig. E14-9B
^2
Force System Due to p
472
CHAP. 14
If E is constant, (d) reduces to
rL 1
7 =
pL2
_ _
/ 2 d x j
o 1
pL
8/j
L!
- d x ,
0
. J
+ - p d x ,
8h 1 +
Jo /
(e)
-1 w dx.
b=
.
dx ,
The parameter < 5 is a measure of the influence of axial deformation. As an illustration, we consider A and I to be constant and evaluate 5 for this geometry. The result is 15
l _ = 15 ( p V
ih2 8 A
8 \h )
(f)
where p is the radius o f gyration for the cross section.
*2
F ig . E 1 4 -9 C
Force System Due to Z ] + l O ne should note that (e) applies only for the shallow case, i.e., for ( f ) 2 1. Now,
Ah ( 2x, y ~ r =T [1 ~ ~ Tr )
(g)
For the assumption o f shallowness to be valid, 16 ( h / L ) 2 must be small with respect to unity. The total forces for the shallow case arc
SEC. 14-8.
NUMERICAL INTEGRATION PROCEDURES
473
It is o f interest to determine the rotation at B. The <2" loading consists of a unit moment applied at B to the primary structure (see Fig. E14- 9D). Applying (14-77) (note that Fig. E14-9D
*1
the stretching terms vanish since N l y = 0), we obtain
'L 1
(Of, =
J . H I
2 \1 + < 5
0)
When E I is constant, (i) reduces to
.
)B
pjJ ( s 24EJ \ l +
< 5
(j)
Since u)H # 0, the results for the fixed end shallow case will differ slightly from (h).
14-8.
O ne of the steps in the force m ethod involves evaluating certain integrals which depend on the mem ber geometry and the cross-sectional properties. Closed-form solutions can be obtained for only simple geometries, and one usually m ust resort to a numerical integration procedure. In what follows, we describe two procedures! which can be conveniently autom ated and illustrate their application in deflection com putations. We consider the problem of evaluating J = \XB f ( x ) dx
JX A.
(14-79)
f See Ref. 8 for a more detailed treatment of numerical integration schemes.
474
CHAP. 14
w here f ( x ) is a reasonably sm ooth function in the interval x A < x ^ x B. W e divide the total interval into n equal segm ents, of length h: h = Xb- ~ - A n (14-80)
I f /( x ) is discontinuous, we w ork with subintervals and use a different spacing for each subinterval. F o r convenience, we let x 0, x , , . . . , x represent the coordinates of the equally spaced points on the x axis, and ,./ the corresponding values of the function. T his n o ta tio n is show n in Figure 14-17. /(* )
F ig . 1 4 - 1 7 . C o o rd in a te d is c r e tiz a tio n fo r n u m e r ic a l in te g ra tio n .
T he sim plest approach consists in approxim ating the actual curve by a set of straight lines connecting ( / 0,/i), ( / 1;/ 2),etc., as show n in Fig. 14 -1 8 . W ith this approxim ation, (14-81) *h- i
.'.TO
If only the total integral is desired, we use,

J =
f{ x )d x =
i=l
A J_ 1>;- h \ \ ( f 0 + /) +
I
X
i= 1 J
(14-82)
which is called the trapezoidal rule. A m ore accurate form ula is obtain ed by ap p ro x im atin g the curve connecting three consecutive points with a second-degree polynom ial, as show n in Fig. 14-19. This leads to AJ k.k + Z j dx - [ j k + 4 /k+1 + f k+2] * !< Jk+ 2 Jk + A J k. k + 2 (14-83)
T o apply (14-83), we m ust take an even num ber o f segm ents, that is, n m ust be an even integer. If the values of J at odd points are also desired, they can
SEC. 14-8.
475
be determ ined using rh AA , k + I Finally, one can express J n as h

Jn
[/o
(14-84)
+ fn +
M fl
./3
' ' ' +
f n - l)
+ 2 (/2 + .U + ' ' ' + / j i - 2)] Equation (14-85) is called Simpsons rule. /
(14-85)
F ig . 1 4 -1 8 . L in e a r a p p ro x im a tio n .
Fig. 14-19. Parabolic approxim ation.
476
CHAP. 14
Example 14-10 ------------------------------------ -------- ------------------------Consider the problem of determining the vertical displacement at Q for the straight member o f Fig. E 14-10. We suppose shear deformation is negligible. The deflection due
F ig. E 1 4 -1 0
M U J jV
EI
to bending deformation (we consider the material to be linear elastic) is given by
fL M
dQ '
*,0
r,M Q d x
(a)
where M is the actual moment and M Q is due to the Q " loading. Substituting for M Q, (a) expands to
SEC. 14-8.
477
To evaluate (b), we divide the total length into n equal segments of length h. number the points from 0 to n, and let
With this notation, (b) takes the form

(d)
If, in determining J, H, we also evaluate the integrals at the interior points, then we can readily determine the displacement distribution using (d).
Example 14-11 ----------------------------------------------------------------------------------Consider the simply supported nonshallow arch shown. We suppose there is some distribution of M and we want to determine the vertical deflection at Q. Considering
F ig. E 1 4-1 1
I
A
X
/7 7 ^ 7
ii El
M .q
only bending deform ation,
dQ
is given by
478
CHAP. 14
N ow , the distribution of M Q is the same as for the straight member. Then, the procedure followed in Example 14-10 is also applicable here. We just have to replace M j E I with M j E I cos 0 in Equation (c) of Example 14-10.
REFERENCES
1. 2. 3. 4. 5. 6. 7. 8.
T im o s h e n k o , S. J . : A d v a n c e d S t r e n g t h o f M a t e r i a l s , Van Nostrand, New York, 1941 . B o r g . S. F., and J. J. G
en n a ro : A dvanced Structural Analysis,
Van Nostrand,
New York, 1959. R e is s n e r , E .: Variational Considerations for Elastic Beams and Shells, ./. Eng . M e c h . D i o . , A.S.C.E, Vol. 88, N o. EM 1, February 1962. M a r t in , H. C .: I n t r o d u c t i o n t o M a t r i x M e t h o d s o f S t r u c t u r a l A n a l y s i s , McGraw-Hill, New York, 1966. M u s h t a r i , K. M ., and K. Z. G a u m o v : Nonlinear Theory o f Thin Elastic Shells," Israel Program for Scientific Translations, Jerusalem, 1962. M a r g u e r r e , K .: Zur Theorie der gekriimmten Platte grosser Formanderung," P r o c . 5 t h Int. C o n g r e s s A p p l . M e c h . pp. 9 3-101, 1938. O d e n , J. T .: M e c h a n i c s o f E l a s t i c S t r u c t u r e s , McGraw-Hill. New York, 1967. H il d e b r a n d , F. J.: I n t r o d u c t i on to N u m e r i c a l A n a l y s i s , McGraw-Hill, New York, 1956.
PROBLEMS 14 -1 . Specialize (14-7) for the case where Vj = x ,. Let x 2 f { x i) and let 9 be the angle from to Yx as shown below. Evaluate the various terms for a parabola / = Q x j + C zx \ Finally, specialize the relations for a shallow curve, i.e., where
02
1.
14-2. Evaluate I* and 5 (see E quation 14-24) for the section defined by the sketch. 14-3. Verify (14-34). 14-4. Verify (14 41) and (14-42). 14-5. Discuss the difference between the deform ation-force relations based on stress and displacem ent expansions (Equations (14-25) and (14-42)). Illustrate for the rectangular section treated in Exam ple 14-1. W hich set of relations would you employ?
PROBLEMS
479
21 I
Prob. 1 4 - 2
h ~ O. ISd I = d f 20
T t
Prob. 14-6
14-6.
Evaluate /' and I" for the symmetrica! section shown.
h= - O.ISd I = d/20 I
14-7. Consider a circular sandwich member comprised of three layers, as shown. The core layer is soft (E 0), and the face thickness is small in com parison to the depth (dc % d). Establish force-dcformation relations based on strain expansions (see (14-37)).
P r o b .1 4 -7
'///////////
d
Core ///////////z
_L T
dc
14-8. Starting with (14 34) and (14-35), derive a set of nonlinear strain displacement relations for a thin member. Assume small finite rotation, and
480
CHAP. 14
linearize the expressions with respect to y 2, * < ?., take
1 = <?i - v2/<
})2 ~ < ?2 D eterm ine the corresponding force-equilibrium equations with the principle of virtual displacements. 14-9. Refer to Fig. 14-10 and E quation (14-31). If we neglect transverse shear deform ation, t '2 is orthogonal to t \ and we can write (1 + y | ) a' i dr ~ ~ a ay
1
+ ( 2 ^2 (a)
2 = /?2h + P J 2
2
rfti _ 1 + e x ~dS ~ Rr a = a (a)
dV2 . IS ~ du = (1 +
1+ < ?! R' 1
+ ds
Verify, that ^ can be expressed as
'! ; i
(b) -1
'> = -----L _ (e - v,A

I - y2//< 11
Also determ ine e x and R' for small strain. Express i'l in term s of the initial tangent vectors, u u xt | +- u2t 2 and take y = S (i.e., a ~ 1). (b) D erive the force-equilibrium equations, starting with the vector eq u a tions (see (14-12) and Fig. 14-4), dF dS
d ---M -- + p+ = 0 n m + t, x F dS
and expanding the force vectors in term s o f com ponents referred to the deformed frame: F+ = Fjji + F 2 V 2 M+ = M t 3 (c) Assume small strain. Derive the force-equilibrium equations with the principle o f virtual displacem ents. T ake the strain distribution according to E quation (b). b = b j \ + b 2 t '2
PROBLEMS
481
(d)
D erive the n onlinear deform ation-displacem ent an d equilibrium eq u a tions for the cartesian form ulation. Refer the tran slatio n s and loading to the basic frame, i.e., take ft = i\X x + v2i 2
P = P
i'i +
P z 'h
Specialize the equations for the case of a shallow m em ber. 1 4 -1 0 . The accom panying sketch applies to b o th phases of this problem .
Prob. 1 4 - 1 0
b2 =
co n st
D eterm ine the com plete solution for the circular m em ber shown. U tilize sym m etry at point A (i(t = co = f 2 0) and w ork with (14-58), (14-59). Discuss the effect of neglecting extensional and sh ear d e form ation, i.e., setting (1/^4) = (1//12) = 0. (b) R epeat (a), using M u sh taris equations for a thin m em ber w ith no transverse shear deform ation, which arc developed in Exam ple 14-2. Show th a t M u sh ta ris approxim atio n (i^ du 2/dO) is valid w hen the segm ent is shallow. . 14-11. The sketch presents the inform ation relevant to the problem :
P r o b . 1 4 -1 1 p2
(a)
= const
I A
T h
1
/.
482 (a)
CHAP. 14
Apply the cartesian formulation to the symmetrical parabolic arch shown. Consider the member to be thin and neglect transverse shear deformation. (b) Specialize (a) for negligible e.xtensional deformation {set 1/4 0). (c) Specialize (a) for the shallow case and investigate the validity of Marquerres approximation. 14 - 12 . Refer to Example 14-6. Determ ine u B2 due to a uniform distributed loading, b 2 = constant. 14 - 13 . Determine the displacement measures at B (see sketch). Consider only bending deformation. N o t e : It may be more convenient to integrate the governing equations rather than apply (14-69). Prob. 14-13
14 - 14 .
Solve two problems with the information sketched: Prob. 14-14
T7777T
(a)
(b)
Determ ine the fixed end forces and radial displacement at point B with the force method. Consider only bending deformation and utilize symmetry at B. Generalize for an arbitrarily located radial force.
PROBLEMS 14 - 15 .
483
(a) (b)
Refer to Example 14-7. Determine the radial displacement at B defined in Fig. E l4-7. Determine the force solution for the loading shown.
Prob. 14-15 P
14 - 16 .
The sketch defines a thin parabolic two-hinged arch.

P rob.14-16
*2
I - I0 /cosd
Determine the horizontal reaction at B due to the concentrated load. Consider the arch to be notishallow. (b) Utilize the results of (a) to obtain the solution for a distributed loading p 2(x) per unit (c) Determine the reactions resulting from a uniform temperature increase, AT. . (d) Suppose the horizontal support at B is replaced by a prismatic member extending from A to B. Assume the connections are frictionless hinges. Repeat parts a and c.
(a)
484
CHAP. 14
14 -1 7 .
C onsider the arbitrary two-hinged arch shown. Discuss how you
P ro b . 1 4 -1 7
would generate the influence line for the horizontal reaction. U tilize the results contained in Examples 14-10 and 14-11.
15 Engineering Theory of an Arbitrary Member

15-1. INTRODUCTION; GEOMETRICAL RELATIONS In the first part of this chapter, we establish the governing equations for a m em ber whose centroidal axis is an arbitrary space curve. The form ulation is restricted to linear geom etry and negligible w arping and is referred to as the engineering theory. Exam ples illustrating the application o f the displacem ent and force m ethods are presented. Next, we outline a restrained w arping for m ulation and apply it to a p lan ar circular member. Lastly, we cast the force m ethod for the engineering theory in m atrix form and develop the m em ber force-displacem ent relations which are required for the analysis of a system of m em ber elements. The geom etrical relations for a m em ber are derived in C hapter 4. F o r convenience, wc sum m arize the differentiation form ulas here. Figure 15-1 shows the natural and local frames. They are related by
t t 12 t
= cos (f> n + sin (j)b 1 3 = sin (j> n + cos 4 > b
(a)
where (j) = 0(,s). D ifferentiating (a) and using the Frenet equations (4-20), wc obtain dt
JS = at
'd iC dS dt 2 dS dt 3 dS
0 K cos (j) K sin 4)
K cos (f> 0
K sin cf) d(j) < T + dS 0 h (15-1) ._____ ______
N ote th at a is skew-symmetric.
485
484
CHAP. 14
14 -17.
Consider the arbitrary two-hinged arch shown. Discuss how you
P r o b .1 4 -1 7
would generate the influence line for the horizontal reaction. Utilize the results contained in Examples 14-10 and 14-11.
15
Engineering Theory of an Arbitrary Member

15-1. INTRODUCTION; GEOMETRICAL RELATIONS In the first part of this chapter, we establish the governing equations for a member whose centroidal axis is an arbitrary space curve. The formulation is restricted to linear geometry and negligible warping and is referred to as the engineering theory. Examples illustrating the application of the displacement and force methods arc presented. Next, we outline a restrained warping for mulation and apply it to a planar circular member. Lastly, we cast the force method for the engineering theory in matrix form and develop the member force-displacement relations which are required for the analysis of a system of member elements. The geometrical relations for a member are derived in Chapter 4. For convenience, we summarize the differentiation formulas here, fig u re 15-1 shows the natural and local frames. They are related by = cos < ji + sin (f)b 1 3 = sin (j> n + cos 4 > b
12
(a)
where $ = (j)(s). Differentiating (a) and using the Frenet equations (4-20), we obtain dt = af dS pfi' K sin c f) K cos K cos - U + J s 0 . dS y (A Note that
a
is skew-symmetric.
485
486
ENGINEERING THEORY OF AN ARBITRARY MEMBER

Y2
CHAP. 15
F ig. 1 5 -1 . N atural and loca l re fe re n c e fra m e s fo r a m e m b e r e le m e n t.
The curvilinear coordinates'^ of a point, say Q, are taken as 5 and Letting R be the position vector to Q (see Fig. 15-2). R = r(S) + y 2 f 2(S) + y3f 3(S) and differentiating, we find dR dS (i ~ y ia u y i a \ i )11 y3a23{ 2 + y 2^ 23*3 dR V- 2 dyj dR dyi The differential volume at Q is </(vol.) = (1 - y 2o l2 - y3(tu )cfS dy 2 dy 3
y 2, y$.
(1 5 -2 )
(15-3)
(1 5 -4 )
= ( 1 - j H dS dy 2 dyz where y n is the coordinate of Q in the normal (55) direction and the radius of curvature. Also,
Rc
l/K
is
SEC. 15 1.
INTRODUCTION; GEOMETRICAL RELATIONS
487
and the local vectors at Q are orthogonal when a23 0, which requires
( l 23
0
\ Rt (15-6)
J L
d< p dS = ~ T =
It is reasonable to neglect y/R terms with respect to unity when the member is thin, i.e., when the cross-sectional dimensions are small in comparison to
Fig. 1 5 -2 . C u rv ilin e a r d ire c tio n s .
Rc and R t. We express d.<j)/dS as ich d(f> dS 0 (15-7)
~L
where L is the total arc length and A(j) is the total increment in (p. The non orthogonality due to < j> cafl he neglected when the member is only slightly twisted, i.e., when bA < f> - y; - 1 (15-8) vvhere b is a typical cross-sectional dimension. In what follows, we will assume the member is thin, (15-8) is satisfied, and < fi defines the orientation of the principal inertia directions.
488
CHAP. 15
Example 1 5 - 1 --------------------------------------------------------------------------The curvature and torsion for a circular helix are derived in Example 4-5:
R,
\2nR j R ,
where R is the radius of the base circle and H is the rise in one full revolution. The helix is thin when b/R 1, where h is a typical cross-sectional dimension. Example 1 5 - 2 ----------------------------------------------------------------------------------------By definition, a member is planar if t = 0 and the normal direction (n) is an axis of symmetry for the cross section. We take the ccntroidal axis to be in the X x-X 2 plane and define the sense of t 2 according to t2 x r3 = l3. The angle < />is constant and equal to either 0 (f2 = ii) or 180 (f2 = / Only a l2 is finite for a planar member:
Example 15 -3 -------------------------- _ _ _ - ........ ----- ---- -------------Consider the case where the ccntroidal axis is straight and r/> varies linearly with S. The member is said to be naturally twisted. Only azs is finite for this case:
n = = 0
a23
(l(i>
, = const = k
If bk 1, we can assume dR/dS is orthogonal to f2, fj.
15-2.
T o establish the force-equilibrium equations, we consider the differential elem ent show n in Fig. 15-3. Wc use the sam e n o tatio n as for the p lan ar case. T he vector equilibrium equations follow from the requirem ent th a t the resultant force and m om ent vectors m ust v a n ish :
(15-9)
W e express the force and m om ent vectors in term s o f com ponents referred to the local frame,
SEC. 15-2.
bdS dF+ dS dS 2
d l lLt dS
dS 2
F ig . 1 5 - 3 . D iffe re n tia l e le m e n t fo r e q u ilib r iu m a n a ly s is .
where F ( F 1; F 2, F 3} etc. The vector derivatives are <n\ Jt-T <lMr

,r
Is
iM + Also,
= I s ' + * at
(a)
~ j f T s + M a'
(b)
Substituting in (15-9), and noting th at a 7 = a, lead to the following equi librium equations: (IF -- - aF + b 0 (IS 0 ~dS aM + in >F $ > = 0 (+ F 2
dFx ds
fl12^2 ~ a iiF ) + 1,71 0 ~ ^2 3-^3 + 62 - 0 (15-11)
dF, + ^1.3^! + 6*23^2 + Ji 2M 2 dM j ds dM 3 "dS
= 0
+ /! = 0
W hen the m em ber is planar, a 13 = a 23 = 0 and the equations uncouple naturally into two systems, one associated with in-plane loading (bu b2, rn3,
490
CHAP. 15
Fi, F 2, M 3) and the other with out-of -plane loading (b3, m u m 2, F 3, M b M 2). The in-plane equations coincide with (14-14) when we set c712 = 1/R and the out-of-plane equations take the form
(15-12)
15-3.
FORCE-DISPLACEMENT RELATIONS-NEGLIGIBLE WARPING RESTRAINT; PRINCIPLE OF VIRTUAL FORCES
We consider the m aterial to be elastic and define V* as the com plem entary energy per unit arc length. Since we are neglecting w arping restraint, V * is a function only of F and M. We let
(15-13)
and write the one-dim ensional principle of virtual forces as

d V * dS = \ s (eT AF + k T AM)rfS - 7 dt A P ;
(15-14)
1/
(f)
- A F+
Fig. 1 5-4. Virtual force system.
SEC. 15-3.
NEGLIGIBLE WARPING RESTRAINT
491
Now, we apply the principle of virtual forces to the element shown in Fig. 15-4. We define u and co as j = uTt = equivalent rigid body translation vector at the centroid (15-15) co = y ]cojtj o Tt equivalent rigid body rotation vector The virtual system satisfies the equilibrium equations (15-5) identically and therefore is statically permissible. Evaluating d( APb I d, AP, * { dit &F - -1 - t, x co 1 + AM.. dS A F7 f du [ i s - au +
0 -ft) 3
02 v+ C /
ft =
dm dS dS I do \d S
(a)
')
and substituting in (15-14) lead to the following force-displacement relations: du e = d s- + dm k = dux d?* I f ; dS dF* du 2 Ts~ cV* dF 3 cV* du 3 Is ^1 2 ^ 2 ^ 13^3
Cl2 3CO3
l 12u 2 ~ a 13u 3
0 w (15-16)
+ C0 2
do.) j M l ~ dS k2
*3
QV* dM 2 dV* dM 3
dco2 dS
C/l 2 0 ) \
dco3 + ^13^1 + dS
Once V * is specified, the left-hand terms can be expanded. The form of V* depends on the material properties, the particular stress expansions selected, and the member geometry. In what follows, we consider the m aterial to be linearly elastic and approxim ation V* with the complementary energy function
492
CHAP. 15
for the prismatic case, which is developed in Sec. 12-3: = ^ + 1 Mr + 2 k F> + 2 < k F U 2GJ
(15-17)
+ k \ M l + - i M \ + k3 M 3 + - M i 2El, 2EU where M t = M i + F 2y3 F 3y 2 ~ torsional moment with respect to the shear center y2, y3 = coordinates of the shear center with respect to the centroid 1
k 2 = 2 I 1-0 k 3 ____~ .
dA y A dA
Note that (15 17) is based on taking a linear expansion for the normal stress, Fx M7 0,1 = T + 1 7 n M,
(a)
and using the shear stress distribution predicted by the engineering theory, = <*\i + ff{; (b)
where <j' is the unrestrained torsional distribution due to M r and a 1 is the flexural distribution due to F 2, F ,. In addition to these approximations, we are also neglecting the effcct of curvature, i.e., we are considering the member to be thin. The approximate fora;-displacement relations for a linearly elastic thin curve member are Fx _ duy = To~ ~ a U U2 ~ 1 3 3 A~E dS F7 iV/7_ dtij e " o t 2 + o J y i = I s + a ' lU' du-. Mr . + a lJul + c/2:ih2 + < x >2 dS e3 ga3 g jI y2 dcî k = mt = - Cli 2 ^ 2 ~ a i 3 (^ 3 ~d, 1 GJ dS dco2 i.O , M i k2 + + @ 1 2 ^ 1 ~ a23o :> 2 E l, dS
ei = el
+
(15-18)
^3 = K -3 + "FT- = b l.
tn
doj3
+ 013^! + a 23C U 2
SEC. 15-4.
CIRCULAR PLANAR MEMBER
493
W hen the member is planar, the shear center is on the Y2 axist and there is no coupling between in-plane (ul5 u2, c d 3 ) and out-of-plane (u3, co2) dis placements. T hat is, an out-of-plane loading will produce only out-of-plane displacements. The approxim ate force-displacement relations for out-of-plane deformation for a thin planar member are F3 Mt _ e3 = GAn, ~ GJ MT d(x>j k ' = GJ I S da 3 = IS + h 1 ' R
0 ,2
Q ( 5_19)
where M T M x y zF 3. N ote that flexure and twist are coupled, due to the curvature, even when the shear center coincidcs with the centroid.
1 5 -4 . D IS P LA C E M E N T M ETH O D C IR C U LA R P LA N A R M E M B E R
Since the displacement m ethod involves integrating the governing differential equations, its application is restricted to simple geometries. In what follows, we apply the displacement m ethod to a circular planar member subjected to out-of-plane loading. We suppose the cross section is constant and the shear center coincides with the centroid. It is convenient to take the polar angle 6 as the independent variable. The governing equations are summarized below and the notation is defined in Fig. 15-5. Equilibrium Equations (see (15-12))
+ Rh> - 0
dj ^ - M 2 + Rnh = 0 dM 2 + M j + Rm2 RF3 0 dO Force-Displacement Relations (see (15-19)) F3 1 du 3 e* ~ GAj ~ R d.9

l
(a)
0h iu\ \
_ M * _ 1 ( d(h \ GJ R \d 6 i2J o M2 1 ( 2 ( da) do) 2 k> = k > + E f 2 R 1
U + )
t T he sh ear cen ter a x is lies in th e p la n e c o n ta in in g th e cen tro id a l axis, w h ich , by d efin itio n , is a plane o f sym m etry for th e c ro ss se c tio n .
494
CHAP. 15
Boundary Conditions
F3 m2
or or or
w3 Oil prescribed at each end (pts. A, B)

0)2
(c)
The solution of the equilibrium equations is quite straightforward. We integrate the first equation directly: F 3 = C\ - R } b3 dO The remaining two equations can be transformed to
d2M x
(15-20)
( di71, rj I- M , = 1<( Ft, - m 2 ----

dO
(< *)
(e)
dMx w + Rm
We solve (d) for M i and determine M 2 from (e). The resulting expressions are Mj C 2 cos 0 + C} sin M2
=
6
+ Ml p
9 + M, p + Rm,
C2 sin 0
C3 cos
(15-21)
where AfliP is the particular solution of (d).
Fig. 1 5-5. Notation for circular member.
SEC. 15-4.
CIRCULAR P L A N A R M E M B E R
495
The solution of the force-displacement relations is also straightforward. First, we transform (b) to
d2
+ C > 2 du3 7(9
" m
do ) i
~ i f nh + e72 (1 + Ct)Mz
RM l
d() "dO RF 3 G/4a
GJ Rea-,
(0
where c, is a dimensionless parameter.

fj
C'
GJ
(15-22)
which is an indicator for torsional deformation. Solving the first equation for ! and then determining u>2 and 3 from the second and third equations lead to w, = C4 cos 9 + C5 sin 0 + ojl p RMj = C\ sin 9 + C5 cos 0 + ~ (oi p ~ o .r (15-23) u^ /?cuj -+
' ~ R F j R 2M . 1
v, / '
where a>liP is the particular solution for The complete solution involves six integration constants which are deter mined by enforcing the boundary conditions. The following examples illustrate the application of the above equations.
E x a m p le 1 5 - 4 -----------------------------------------------------------------------------------------------------
The member shown is fixed at A and subjected to a uniform distributed loading. Taking
Fig. E15-4
b j = const in (15-20), we obtain
f 3 = c, - Rb3 e
(a)
496
CHAP. 15
The equation for M , reduces to
d2M , T + Mi = RF* =
Then,
- R2M
W
(c)
M x p = R F 3 = R C X R ~ b 30 and the solution for M, and M, follows from (15-21), M l = C2 cos 0 + C3 sin 0 + RFj A Y 2 C x sin 0 + The boundary conditions at B require = Mj =
m
cos 0
(d>
2= 0
at
0 = 0,
(I
= R b 36 B C2 - C3 = sin 0B cos 0H (e)
Replacing 0 B - 0 by ij, the final solution is
F3 = i?k3f/
Mj =
~
- sin > ?] K2/j3[1 c o s /]
(f)
Example 15-5
The force system due to the end action, F n , can be determined by applying the equi librium conditions directly to the segment shown in Fig. EI5-5A. This leads to
F3
M x = F n i R(\ -
cos /f) - F03R [ I - cos(0 - 0)] - FB:iR sin(0B 0)
(a)
M 2 = F B3R sin rj -
We suppose there is no initial deformation. Using (a), the equation for a>t becomes
d 2OJx
il0'r + ' at)
+ co, =
EI2
- R ZP' b 3
(1 + cV )sin(0, - 0)
(b)
The particular solution of (b) is
< -'h.P = - ^jFf 3 (' +
C0SW fl -
0 )]
(c)
Using the above results and specializing (1.5-23) for this support condition lead to the following expressions for the displacements:
0 31 *= a,A i cos 0 + &A2 sin 0
r.h
1 - c, 1 4- c. - cos < )B + c, sm 0 - - 6 cos(0B - 0)
SEC. 15-4.
CIRCULAR PLANAR MEMBER
497
---- - <c,[ 1 4- cos 0] 7 = o)Ai sin 6 + ajA2 cos 0 H

0)
(d )
{ Eh I 1 r \ / 1 .L q\ sin 0 B sin 0 - ( - - ) 6 sin($i; 2 y T j s* n s * n ^ ("- - - - - \ 2 uj *= ^^3 + R loAi ( 1 cos 0) R6) j 2 s 11 0 +

El 2
1 a - c, ---- ccos 0
sin 0 - c, sin &B
1 + ct
2
Eh
" S(9 ^ + C ' + GAJP
-\- c, sin(0fl - 9)
Terms involving o)Al, <oA1 and itAi define the rigid body displacements due to support movement. Also, terms involving c, are due to twist deformation. The rotations and Fig. E15-5A
translation at B are listed below:
R 2f*>
+ El
- -- cos 0 + c, sin ()B
c,
1+
c,
a>Bi ~ - ai sin Of, + a>A2 cos 9B
-f
1 - c, . R 2F* - sin2 6 b /f iC '[CS 9* " *] - 2

R zT-b1 f Eh I
(e)
uB ?i = uA1 + RaiAl (l - cos Op) - R w A1 sin Ba
1 - cr 3
cos 9 b sin Bb 2c, sin &B
+O a
1
2
Eh
C ' + GA7R1
498
CHAP. 15
To investigate the relative importance of the various deformation terms, we consider the rectangular cross section shown in Fig. E15-5B. The cross-sectional properties aref
I _ 6 1 /Ij 5A
1 5 di dj,
(f)
J = ~ d } d j (for d 2
/ - d jJ l
2 12
Then,
e i2 = n
G 7
(dj y <g)
G j 4fc \ d 2J
Eh
G Aj,R2
E G
I fd2
10 \dm
The values of 4 k and c, for c j'3/c/2 = 1, 2, 3 and v 0.3 are tabulated below:
c, = E hJ G. J d 3/ d 2 4k
(for 1- - 0.3) 1.54 3.8 7.4

Fig. E 1 5 -5 B
1 2 3
1.69 2.75 3.16
y 2 -------------------------- d-i
^3
f The torsional constant for a rectangular cross section i s developed in See. 11-3.
SEC. 15-5.
FORCE METHOD EXAMPLES
499
Since ( d 2/ R ) 2 1, we see that it is reasonable to neglect transverse shear deformation. In general, we c a n n o t neglect twist deformation when the member is n o t shallow. For the shallow case, we can neglect c, in the expressions for con2, uB3-
Example 15-6 -----------------------------------------------------------------------------------Consider a closed circular ring (Fig. El 5-6) subjected to a uniformly distributed twisting moment. From symmetry, F z = 0 and M u M z are constant. Then, using ( 1 5 -J6), we find
(b)
F ig. E 1 5 -6
bj = m2 - 0
m i ~ const
Xx
15-5.
In this section, we illustrate the application of the principle of virtual forces to curved members. The steps involved are the same as for the prismatic or planar case and therefore we will not reiterate them here. We restrict this discussion to the case where the material is linearly elastic, the member is thin and slightly twisted, and warping is neglected. The general form of the expression for the displacement at an arbitrary point and the compatibility equations corresponding to these restrictions (see (15-14), (15-17)) follow.
500

D isp la c e m e n t a t P o in t Q
CHAP. 15
F, e? + l ) F '- fi + F3 + { '5 a ; ) F3' < , + ( ^ j ) M t -q + ( k + , 7 t )

Compatibility Equations
G/t-
F 2,Q
(15-24) 1 ( k + iFyM ;V/= e r/S
Z u Z 2, . . . , Z r force redundants
Fj = Fh o +
r
(a)
F j.
Mj = M jm o + X!
k= 1
kZ k
(b)
= î,0 + X &i,kZk
k~ 1
j= 1 where .fkj fjk ~

+
(/c = 1 , 2 , -----r) 1 + 0 r -A T2 ^ 2 .
+
(15-25)
"Tr ^ AE
M T j M Tt
.k +
TTT" ^ 3-' f GA3
i - k
- M 2j ; M 2 i ,
4- - g j -
jM }
dS
F
A . - 5 X * 3, +
V1 "sty *+IcirJ J'*
( p0
+ / M
* -
M r
M t. q Mr,k + I ^2 H 1M 2, k '2. I t + I ^3 + GJ /2 = M j + v3 F 2 - V 2F 3
M 3,*
The reduced form for out-of-plane deformation is obtained by setting Fj = F 2 = M 3 = e + /c? - 0. Example 15-7 ---------------------------------------------------------------------------------Consider the nonprismatic member shown below. The centroidal axis is s tra ig h t but the orientations of the principal inertia axes vary. We take X , to coincide with the centroidal axis and X 2, X 3 to coincide with the principal inertia directions at the ieft end (point A). The principal inertia directions are defined by the unit vectors t 2, t 3.
t 2 = cos 0z2 + sn 4>l3
t3 = sin < jf> 7 2 + cos 4 ^ 3 $ 0 at Xj = 0
(a)
SEC. 15-5.
501
Now, we consider the problem of determining the translations of the centroid at B due to the loading shown in Pig. E15-7A. It is convenient to work with translation components (vB2, vS3) referred to the basic frame, i.e., the X 2, X 3 directions. We suppose that the shear Fig. E15-7A
X2
X\
center coincides with the ccntroid and transverse shear deformation is negligible. Spe cializing (15-24), and noting that M, = 0 for a transverse load applied at the centroid, the displacement expression reduces to
(Iq =
I f L ( 1
\ M 2M 2 q + M \ M j Q j d x 1
V 2
i
(b)
Jo
Force Systems
The moment vectors acting on a positive cross section due to P 2, P 3 applied at B (Fig. E 15-7B) are _ (M ),2 = P 2{ L ^1,^3
( M ) Pi = - P 3(L x {) h
To find M 2, M 3, we must determine the components of M with respect to the local frame. These follow from Fig. E15-7C: For Pz,
M 2 = P 2{L -
.xt)sin (/>
M 3 = P 2(L - x j c o s < j>
id)
502
CHAP. 15 Fig. E15-7B
Fig. E 15-7C
For P } ,
M 2 X[ )CDS (j)
M 3 = +P.,(L x jsin <p
(e)
Determination o fv B 2 Due to P2
The virtual-force system for vB1 corresponds to l \ = +1. Introducing (d) in (b), we obtain sin2 <b cos2 < / ; P, ----L + -----L (L - xt )2 dxl (0 B 2
. ^2 h
Determination o f vf!3 Due to P 2

The virtual-force system for vB3 corresponds to P3 = + 1. Using (e) leads to
SEC. 15-5.
503
Example 15-8
We rework Example 15-6 with the force method. Using symmetry, we see that
M, = 0
M 2 = miR
(a)
Suppose the rotation coi in the direction of m L is desired. The virtual loading for this displacement is = + 1 . Starting with
cot Ait d S = rt>
J E l i
A M 2 dS
(b)
and substituting for M2, we obtain

m j i 2
e
(c)
/7
Example 15-9
Consider the closed ring shown. Only M, and M 2 arc finite for this loading. Also, the behavior is symmetrical with respect to X , and we have to analyze only one half the ring. Fig. E15-9
*2
2
Zi
\
-Xi
We take the torsional moment at 8 0 as the force redundant. The moment distributions arc
M y sin 9 4 Zj cos 8 = M l 0 + Z t M x j
K
M 2 --
cos
sin
** M 2, 0 +
M 2.
^ 9 ^ n /2 2 '
(a )
504
CHAP. 15
Specializing (15-25) for this problem,
A i^ i = ry2
A, = - 2 R
"n}!
J nj2
~ M \.x GJ
GJ w l.,1
e i
/?/,
dO 2 J
d6
(b)
A i = 2K
J -w/2
and then substituting for M u M 2,

*,r/2
Al =
-R T
-it/2
GJ
sin 0 cos 9 dO
Eh
(C)
and it follows that Zj = 0. We could have arrived at this result by noting that the behavior is also symmetrical with respect to X 2. This require? M 2 to be an ev en function of 0. The virtual-force system for cuM is T + 1 . Using (15-24) and (a) leads to
r/2 I / j. sjn a \ sjn o

2coAl = 2 R
{ j. cos + I 21:.!,
j
cos q 2
dO
t t /2
2 G J J "2
(d)
RTn <*>A 1 =
_ L L GJ +
EU
Example 15-10 ------------------------------------------------------------------------------------We analyze the planar circular member shown in Fig. Cl 5 - 10A. The loading is out-of plane, and only M v, and M 2 are finite. To simplify the algebra, wc consider the shear center to coincide with the ccntroid and neglect transverse shear deformation. It is con venient to take the reaction at B as the force redundant.
SEC. 15-5.
FORCE METHOD-EXAMPLES
505
Primary Structure
The primary structure is defined in Fig. E l5 -JOB:
Ri =
- Fa * haz Z t = F B3
R z M a2 d-2 ^.42
- MAl
(a)
R4 -
Fig. E 1 5 -1 0 B
Force Analyses
The force solutions for the loadings shown in Fig. E I5-10C are: For P :
^ 3, 0 + P
Mi, p - PR[ 1 - cos(i; - 7 C )]

M 2, o =
s i n (>7 -
;/c )
(b)
* lC
For Z* = + 1 :
^ Fi.i ~ +1
M , ! i? (l COS Jj)
(c)
M 2, ! /? sin r/
506
CHAP. 15
Fig. E15-10C
Compatibility liquation (15-25) A t Jn
0 B
GJ
Ei
dr,
(d )
Ai = S Rt. t h - R
'Mi . qMi.i + -( k %+
GJ
) Mi t
c/ij
Substituting for the internal force and reactions, we obtain the following expressions for f n and A,:
f
= *-
E !1
- 2c, sin 0B _ r .
V 2
sin 0 n c o s
Ai = 83 - W /43 + R ( oa2 sin 8 b - RoiAl( l ~ cos ffB)

+ R2 C
Jtf/i ~9c PR3 r Eh

c,
*2 sin(0s 0)d6
1 + - C O S(0B 0C )
sin (?B sin 0C
(e)
+ sin flBcos 0C - cos 9 sin 6C

+ ~ | $ c cos(0B f)c) -
cos 0 8 sin 0,
c, =
Eh
GJ
Note that we could have determined A, and / n using the results of Example 15-5.
SEC. 15-6.
RESTRAINED WARPING FORMULATION
507
15-6.
In what follows, we consider the member to be thin and slightly twisted. Referring to Fig. 15-2, these restrictions lead to cfR
-
(15-26) </(vol.) sr dS dyz dy3 Therefore, in analyzing the strain at Q (S, y 2, _ y 3), we can treat the differential line elements as if they were orthogonal. The approach followed for the pris matic case is also applicable here. One has only to work with stress and strain measures referred to the local frame (f1; f 2, h ) rather than the global frame. Our formulation is based on Reissners principle (13-33): <5[fJJ(<rr b ; ti F*)<7(vol.) jfp Tu (-/(surface area)] = 0 <r, u independent quantities = e(u) p, b = prescribed forces V* = F*(<t) = complementary energy density
(a)
We introduce expansions for u, a in terms of one-dimensional displacement and force measures (functions of S) and integrate over the cross section. The force-equilibrium equations follow from the stationary requirement with respect to displacement measures. We start with the strain measures, r. (k^ y12, One can show th a tt
(15-27)
where u is the displacement vector for Q (S, yl7 y 2). We use the same displace ment expansion as for the prismatic case: u = u,f, -I- u2i, + u3t3 ~ "I" VjjVi 0)y}2 + f '4 > U2 = Usl ~ a>i(y3 - y 3) u 3 - us3 + Wi()>2 - TV ) (15-28)
< t >'= 4 > {y z , >'3)

Expanding jJ(rTc dyz dy3 = JKffnej +
t Sec Prob. 15-5.
<r12y l 2 - f
ffx3y 1 3 ) ^ 1 dy 3
(a)
508
CHAP. 15
leads to jfoÊ dy2 dy3 F l e-l + F 2e 2 + ^ 3 e3 + M Tk x + ' M 2k 2 + M 3/C 3 + M Rf + M J ' S eu e 2, . . . , k 3 (defined by (15-16)) (15-29) M* = ^2 ^3 = jJO l2 ($ ,2 + 120) + <7l3W>.3 + 01 3 ^)3 ^2 ^ 3 T he equilibrium equations consist of (15-11) and the equation due to w arping restraint, M R = M*,.f (.15-30) which can be interpreted as the stress equilibrium equation for the I j direction weighted with respect to (f> . Now, we use the stress expansion developed for the prism atic case. The derivation is discussed in Sec. 13-5, so we only list the essential results here. T he norm al stress is expressed as
fx ^ M 2 M3 M a u = - + y 3 -----, >2 + ~ r (l} a 12 3 
(1 5 -3 1 )
where (j) = the St. V enant w arping function referred to the shear center. We write the transverse shear stress distribution as a = (//2F 2 + \j/31' 3 + ij/,,M r -f i/z.-M r MT = Mu t + Mr r ( 1 5 - 32 )
(i//s are functions of >:2, y 3.) The corresponding com plem entary energy function is
*AS' * )* s(iw *
2 5
(jJ
y^r + F 3^2r ) ^ r
Also, (15-32) satisfies (see(13-50))
11 (^ 1 2 ^ 2 + o i 3< j> '3)dy2 d y 3 = M T

Finally, noting (b), we express M R as (1 + br)Mr T + b2F 2 + &3F 3 where the 6s involve the curvature (a12,
^2 3
(b)
(15-34)
3). If the cross section is symm etrical,

~ br ~ 0
= ,V 3 r J'2 r
(c)
SEC. 15-6.
509
and b2, b 3 are due to self-equilibrating stress distributions.! It is reasonable, in this case, to take b2 fr3 ~ 0 and com pute the shear coefficients (A 2, A 3) based on the prim ary flexural shear stress distributions. Expanding the stationary requirem ent with respect to force measures yields the force-displacement relations,
dV*
db\
k, = + b2f =
dV* m~2 dV* OF
dV* dM , f,s = dM , (15-35)
e3 + b3f -
dV*
BF 3 dV* dM'r
Cl _ dM'i-
l< i + (1 + br) f =
where et, e2, , /<3 are defined by (15-16). The corresponding unrestrained warping relations are (15-18). Example 15-11
To investigate the influence of warping restraint, we consider a planar circular member having a d o u b ly s y m m e t r ic a l cross section (Fig. 1:15 - 1 I), clamped at one end and subjected Fig. E15-11
to a torsional moment at the other end. We neglcct transverse shear deformation due to restrained torsion. The governing equations for this loading (see Sec. 15-4) follow.
Equilibrium Equations
dM j dO dM 2 d&
= M,
= -M , 1 dM,
(a)
d()
M, = M\ + M \ | See Prob. 15-6.
510
CHAP. 15
Force-D isplacem ent Relations

M,
E l 2 fcko E I 2k 2
R [To
1 fdoc>x
(b)
*
7
-h ! L A R dO
-k.
f m
-------
R \ dd
-r r -
M'l = GJkl Boundary Conditions

0 = 0 0 = 9b
CO 1 = C(?2 = / = 0
M, = M = 0
M, 0
(c)
One can write the equilibrium solution directly from the sketch: Mj = M cos(0B 0) M 2 M sin(0# - 0)
(d)
We substitute for the moments in the force-displacement relations. Mj GJ/cj = M cos{9B - 0) (e) sin(0B ~ (?)
R 2 JO 2
1 K 1 ( da>7 k2 " ~R \d0~ + (
and solve for k t, and then o)j. The resulting expressions are , GJ X2 = ----EJ, ky = M GJ +
R2
/. = RX
E l2 c, = ' GJ
{cos(0B 0) cos 0;i[cosh A0 sinh ^0 tanh A0C ]]
(0
^2 A
cos(0B - 0) - sin 0 cos 0j
C t ( 1\ \ V + *v X cos 0B
1 + X2
~ 0 cos(0B - 0) +
1 1+ F
r sin 0 cos 0n 2
[sinh X0 tanh X0B cosh X0 + cos 0 tanh XOg]
Warping restraint is neglected by setting Er = 0 and X = -- co.
SEC. 15-7. The rotation at B is
COMPLETE END RESTRAINT
511
C01
RM \
1
=
~ sin <J cos 0 B) + c , K
uB
K = -------- | - 9 B r sin Ojj cos 9 B
I 1 1 +~2 A
J+
If we set
R
1 2
(g)
I + A
co s2 0 D tanh /
, _ LI
O n
0 ~ n
(h)
and let 0 B~ >0, (g) reduces to (13-57), the prismatic solution. The influence of warping restraint depends on I and 0 B. Values or K vs. 1 for 0 S = tc/ 4 , n / 2 are tabulated below:
K
7 1 for 0 , = -
n
K
1 Ii n 1 +- ~i 1 + 1 / 12
A O r + sin
K.Z
T iI 1 1 + - \ 1 + --7 2 \ /r
t tanh 4 1 ------- ^ I
for 0B -
(i)
K =
cos 0)
r? s
/
1
4^
0H = n/2 0.500 0.96 0.99
5 10
We showed in Chapter 13 that
0.179 0.786 0.907
k = 0
il
(open scction) (j) (closed section)
;' = *
where t is the wall thickness and h is a depth measure. Since I = RX and R j h 1 for a thin curved member, the influence o f warping restraint is not as significant as for the prismatic case.
15 7.
MEMBER FORCE-DJSPLACEMENT RELATIONS COMPLETE END RESTRAINT
In the analysis of a m em ber system, one needs the relations betw een th e forces and displacem ents at the ends o f the m em ber. F o r a truss, these equations
512
CHAP. 15
reduce to a single relation between the bar force and the elongation. M atrix notation is particularly convenient for this derivation so we start by expressing the principle of virtual forces and the complementary energy density in terms of generalized force and deformation matrices. Referring back to Sec. 15-3, we define
rd V !
Ifl. " < dV* dML
(15-36)
and write the principle of virtual forces as Js dV* dS = \s S r ^ dS = dT AP (15-37)
N ote that we are working with M ,, not M T. We use the complementary energy function for a thin slightly twisted member with negligible warping restraint (i.e., (15-17)). With the above notation, Eq. (15-17) = s> P* where R= 8/ Tf _gf m
--------- [----------------------
I Sfm j Sm 0 '
A 2G
(15-38) 0 + l i
GJ
" 1 IE g/ = Sym
T 2J 3
GJ
1
A 3G
A +
GJ
0 gf m
?3
0 0
Vi
0 0 0 gm
1 GJ
0 1 Eh
0 0 1 Eh
GJ GJ
Sym
The force-deformation relation implied by (15-38) is (15-39) We will use these general expressions for planar and out-of-plane deformation as well as for the arbitrary case. One has only to delete the rows and columns
SEC. 15-7.
513
of g corresponding to the zero force measures. For example, -JF => {F1F2M 3}
y3 = o
1 AE 0 0 i o
1
a 2g
i i j 1
0 0
(15-40)
1 1 e F3_
for planar loading applied to a planar member. Finally, we substitute for S ' in (15-37) and distinguish between prescribed and unknown displacements. The principle of virtual forces expands to
js(<^ + g
A & dS - dT AR = d T AP
(15-41)
where 3 contains prescribed displacements and R are the corresponding re actions; d contains unknown displacements and AP are forces corresponding to d. The virtual-forcc system (AP, AR, A-^) must satisfy the force-equihbrium equations, (15 11). It is more convenient to generate and R with the equili brium equations for a finite segment rather than attempt to solve (15-11). Consider the arbitrary member shown in Fig. 15-6. Each end is completely restrained against displacement. The positive sense of S is from A toward B.
Fig. 15-6. Arbitrary curved member.
514
CHAP. 15
W e suppose the geom etry of the m em ber is defined with respect to a basic frame which we refer to as fram e n, and take the end forces at B as the force redundants. Then, the prim ary structure consists of the m em ber cantilevered from A. T hroughout the rem aining portion of the chapter, wc will em ploy the n o tatio n for force and displacem ent transform ations th at is developed in C h ap ter 5. A superscript n is used to denote a quantity referred to the basic frame. W hen no frame superscript is used, it is understood the quantity is referred to the local frame. F o r example, 2FQ represents the internal force m atrix at point Q referred to the local frame at Q. N o te th at acts on the positive face. The force m atrix for the negative face is Q. The end forces at A, B are denoted by (F'b and are related to the internal force m atrices by (15-42) Also, the displacem ent m atrix at point Q is w ritten as % Q. QIq = {ui% u2, u 3 | (Oi, (o2, co3}q | w | (15-43)
F o r this system, Un A and are prescribed. We determ ine ll"B for the prim ary structure, i.e. the m em ber cantilevered from A, due to displacem ent of A, tem perature, loads applied along the mem ber, and the end forces at B and then equate it to the actual e W%. The virtual-force system is AP = A # J AR = = - S T ba A;F?, (a) a & Q = .t 'Sq a & i = 0 f nq ?r;SQ&.W"B Also, a d = -n Introducing (a), (b) in (15-41), we obtain ( A # a r s = ( a ^ s ) t m A)T^
fS B
(b)
~S b
(-^zyxK + g
= 3cn i l w \ + JIS Next, we express 3F as
h V q + zq& qW
= :^ Q , o +
(15-44)
where . F q 0 is the internal force m atrix at Q due to the prescribed external loading applied to the m em ber cantilevered from A. Finally, substituting for
SEC. 15-7.
515
3? 0 leads to
K = a m fW k + P " 3Tn f c t( q + gQ^ Q .o )d S " ^ "* , + J s^ n Tg
(15-45)
The first term is due to rigid body m otion of the member about A whereas the second and third terms are due to deform ation of the member. We define ir " as the member deformation m atrix : i 1 ~ ^Jjjactual ''n U c 1-0! borly (15 ' 46)
m otion iihnui .4
By definition, i ~ n is equal to the sum of the second and third terms in (15-45). We also define r 0 = [ <^~BQr(^Q + Rq-^q, o)dS = initial deform ation matrix
f"
= j
*Sb ,
B{ ^ n f?QT g Q & ~ B , Q )d S
mem ber flexibility matrix
and (15-45) reduces to 1 = r% + f ' ^ n (15-48)
Equation (15-48) is the force-displacement relation for an arbitrary m em ber with complete end restraint. It is analogous to the force-elongation relation for (he ideal truss element that we developed in C hapter 6. The member flexibility matrix, f", is a natural property of the member since it depends only on the geometry and m aterial properties. F o r simple members such as a prism atic member or a planar circular member with constant cross section, one can obtain the explicit form of f. W hen the geometry is complex, one m ust generally resort to numerical integration such as described in Sec. 14-8 in order to determ ine f and //' 0. This problem is discussed in the next section. Finally, we point out th at the general definitions of f, V 0 are also valid for in-plane o r out-of-plane deform ation of a planar member. O ne simply has to use the appropriate forms for the various matrices. U p to this point, we have considered only a simple member. Now suppose the actual mem ber consists of a set of members rigidly connected to each other and the flexibility matrix for each m em ber is known. We can obtain the total flexibility m atrix by compounding the flexibility matrices for the individual elements. To illustrate the procedure, we consider two members, A A X and A 1B, shown in Fig. 15-7. The matrix, f", contains the displacements at B due to the end forces at B with A fixed: Un n = (a) Now, suppose point A x is fixed. Then, the displacem ent at B due to the deform a tion of member A rB is
member =
A,B
5 .5 ^ 5
(*>)
516
CHAP. 15
where f"AtB is the flexibility m atrix for m em ber A X B referred to fram e n. The additional displacem ent at B due to m ovem ent of A , is
m ) d isp lacem ent at A t ^ " bA i ^ A i (0
It rem ains to determ ine ^ l n Ax.
*3 "
x2 "
X"
F ig . 1 5 - 7 . S e g m e n te d m e m b e r.
The force system at A t due to the end forces at B is given by

A, ~
_ or"
H A ,'-'
(d)
and the resulting deform ation of m em ber ,4/1, is

01/" I _ * A j | member A A ,
* AA
f "
<2>
A\
f '1
0/-n
BA\* U
(e)
Finally, we have Oil" (fn

U B ~ V M iB ' '*> H A 11 A A X'A' B A x h *
tam.Trn
or"
1 '*
(15-49)
The end forces at B are found by inverting (15-48): k" (P )_ 1 = m em ber stiffness m atrix = k n{i/ n - i ' l ) = -k " Y ^ 0 + k**U% - k "STb a ^ a (15-50)
The first term is due to external load applied along the m em ber and represents the initial (or fixed-end) forces at B. F o r convenience, let y'n (15-51) n The second and third terms are the end forces at B due to end displacem ent at B, A. O nce is know n, we can evaluate the interior force m atrix at a point using (15-44), & Q = & Q .0 + (a) Thus, the analysis of a completely restrained m em ber reduces to a set of m atrix m ultiplications once the m em ber stiffness and initial deform ation m atrices are established.
SEC. 15-8.
GENERATION OF MEMBER MATRICES
517
W hen analyzing a system of members by the displacement method, expres sions for the end forces in terms of the end displacements are required. In addition to (15-50), we need an expression for $ A n . Now, = -& *A = - ^ l o Substituting for .jFg, leads to
A __ P * _ A, i '> fn \ r n O ) / n A BAK + VTn T V ? /n *-' / J.lK X R A U A
- % nA & \
(b)
A, i ~
A, 0 ~
A B A B , !
where &*AA represents the initial end forces. In order to express the equations in a more com pact form, we let K b = k" l,n _ __l,n<3">t,T
&BA K n r /M ^ K HA
(15-53)
- K nA h k "
( K a )T =
ar
BA*
I ' l ar- n, t
__
cam
J,''
uak ba
W ith this notation, the force-displacement relations simplify to

j
?/7 n _ -S'" I. I," < ?//' _L If'' Ol/" /j r p , t k /{/j f f R , k D/, a A

y y | .,- +
(15-54) and k",^
y .1
ff,n +
/Y 4
N ote that only k" and
are required in order to evaluate
15-8.
GENERATION OF MEMBER MATRICES
The member flexibility matrix is defined by f ,= \ly n u q TgQ- n $ d S N oting that (b) and letting
& =
Qm n*
(a)
( 1 5 - 55 ) (15-56)
we can write
r = [s (arSQrg a ^ Q ^
If numerical integration is used, the values of the integral at interm ediate points along the centroidal axis as well as the total integral can be determ ined in the same operation. This is desirable since, as we shall show later, the interm ediate values can be utilized to evaluate the initial deform ation matrix. We consider next the initial deform ation m atrix:
'n = f5s
JSa
T(Sl + gG .? v o)ds
(c)
We transform <f,g, and 3F from the local frame to the basic frame, using (15-55)
518
CHAP. 15
an d gQ n = m nq' The contributions of tem perature and external load are (ô).emp. (T^S)ioad = \$B m J ^ nq- T^ ) d S s
SB
(d)
(15-57) (15-58)
^ m q ^ Q . O )dS
Suppose there is an external force system applied at an interm ediate point, say C. Let P c, T c denote the force and m om ent m atrices and the total force m a trix : = |^ } (15-59)
N orm ally, the external force quantities are referred to the basic frame for the mem ber, i.e., fram e n. The initial force m atrix at Q due to this loading is given by ^ q, o = ^ cq^ c SA ^ S Q ^ S C (15-60)
^
q. o
Sc
^ S q ^ Sf{
W riting
** CQ
(ifn
__ orx oyn
'* B Q - t CB
(~\
and introducing the above relations in (15-59) result in en v , m j ^ " Q)ds] (src^ c ) (0
The bracketed term is an interm ediate value of the integral defining f". Finally, we let
jp =
J.ly1
fSp m j g & r BQ )ds
(i5-6i)
W ith this notation, (f) simplifies to = (J c ^ c n W 'c Also, f = J* (15-63) The determ ination of the m em ber flexibility m atrix reduces to evaluating J defined by (15-61). O ne can w ork with unpartitioned m atrices, i.e., SC, g, but it is m ore convenient to express the integrand in partitioned form. The partitioning is consistent with the partitioning of 3F into F, M. Since the form ulation is applicable for arb itrary deform ation, it is desirable to m aintain this generality when expanding SC, in partitioned form. Therefore, we define a as the row order of F and /? as the row order of M. & = fF )
(M J
(15-62)
(*!)
(flx l)
(15-64)
C ontinuing, we p artition SC, @ 1 and g symmetrically, consistent with (15-64),
SEC. 15-8.
G E N E R A T IO N OF M E M B E R M A T R IC E S
519
and simplify the notatio n som ew hat: arn >< iS

1 HQ ~
_ _ _ k _
1
0 "
(/?**)
! h(15-65)
!
(a xa) m nq =
K
0
(ft x ID
0
[a * a>
K,
(ax ft) J J li 2 ~ i g( 22
() !2 -
"sr() 61 1
'
(>, _g 12
(ft X ID'
The translation and rotation transform ation m atrices are developed in Secs. 5 -1 , 5 -2 and the form of g for a thin curvcd m em ber is given by (15-38). The local llexibility m atrix g, is defined by (15-55). U sing the above notation, the expressions for the subm atrices are g n = R jg n R * g u - R lg u R /. R22 (15-66)
N ote th a t g 12 = 0 and g U) g 22 are diagonal m atrices when the shear center coincides with the centroid. If, in addition, axial and shear deform ation are neglected, gj j = 0 . We let < **) (kx /))_
* - { - * * (1 5 -6 7 )
' The subm atrices follow from (15-65):
'I 'll
I ^2 2 (ft *ID'
' I ' l l = g l l + g" zX-BQ + ( g u X V > 7' + *1*1 2 g 12 + ^B Q ^22
%22^BQ
(15-68)
^22 a22 Next, we partition J consistent with vj/:

rsP
JP (
jS a
vj/ dS =
Jp. 11 jl J r , j 2
|7 I2 1 I
(a * x) 1
(a x m
Jp.
22
(15-69)
v.
J p , ij
U! * flV
Finally, we partition f":

(a x a) ~r n (axfl)
f" =
>n 1 T T f it. 1 12
|CS, .j u
rn
*12
f" I t2 , *2
= J,
(15-70)
< ,p x- py
5 = >.= J&A . >M S
520
ENGINEERING THEORY OF AN ARBFTRARY MEMBER
CHAP. 15
T he initial deform ation m atrix due to an arb itrary loading at point C can be determ ined with (15-62). Its p artitioned form is r a w =
( a x a)
(J
&
cb
X P
loJ
(v"o!>
(XI) 1 T ~Jc, 11 Jc, 12 ^BC J C. 12 f'P f ) f A 12 - j ~ c ~ M c 1 J C. 22_ I T c J (Px0) (fl* 1)
(15-71)
where Y q, 0 o denote the initial translation and ro tatio n matrices. T he m em ber stiffness m atrix, k", is obtained by inverting f". W e write
(ct x a) (a x ft)
'L " K11 k" - (P ) 1 = k k"> 12r
I !
-f-
If'1 k 12 k" k 22
f-j
(15-72) for convenience)
O ne can easily show th a t (we d ro p the frame superscript on k i i = (f n - f u f z t f i F 1

^"2 =
(15-73)
= f l i d , - f L K 2)
O nce k " is know n, the stiffness m atrices k",n, k n BA and k " ^ can be generated. Expanding (15-53) leads to the following partitioned forms:
k" kn k 0B
1 tn k ba
k
k aa
1 k" n T+ I K11\ aB /4 k". r V- T _k''1 2 'r ; k12 a /M + Mi 1 A _A r 1 1 _ xs: a v b _ k" KJ1
_LL Un' K 12T'
k"
k"
K22
k 2~
k "22
|1 A k""T ,% T~ _ !2 ! B
k" K11
(15-74)
A~ Ar ] C.
15-9.
MEMBER MATRICES PRISMATIC MEMBER
In C hapter 12, we developed the governing equations for a prism atic m em ber and presented a num ber of exam ples which illustrate the displacement and force m ethods o f solution. Actually, wc obtained the com plete set of forcedisplacem ent relations and also the initial end forces for concentrated and uniform loading. Now, in this section, we generate the m em ber flexibility m atrix using the m atrix form ulation. W e also list for future reference the various m em ber stiffness matrices. The n o tatio n is sum m arized in Fig. 15-8. F o r convenience, we d ro p the frame reference superscript n, since the basic fram e coincides with the local frame, i.e., R"9 = I. The positive sense of a displacem ent, external force, or end forces coincides with the positive sense o f the corresponding coordinate axis. S tarting w ith (15-66), we have g";- = gip since Ra, R /? are identity m atrices. O nce XBQ is assem bled we can determ ine the subm atrices of \j/ from (15-68).
SEC. 15-9.
521
X2
Mby^bi
-*----- X\
F ig . 1 5 - 8 . S u m m a ry o f n o ta tio n fo r a p r is m a tic m e m b e r.
Now, 0 Xfl<2 0
o (L
0 0
X Q1)
0
4I
I (L xQl) 0
(a)
Then, using g defined by (15-38), we obtain L /A E fu =
o
L GJ * z* 3
r ( 1 LM \ A 2G + G J J + 3 E Ij j Sym y
0 0 0 L2 2E l 2 ! ! o "
] Tf 1 x l\ L? i L -------------- I -I--------1 \ A 3G G J J 3 E l 2
Lx3 GJ Ex 2 GJ L /G J ^22 Sym
l 2 2E l i (15-75)
I ! 0 0
0 '
0
l /e i 2
| L /E I,_
522
CHAP. 15
The subm atrices o f k are generated w ith (1 5-73), (1 5 -7 4 ) and are listed below for reference. Transverse shear deform ation is neglected by setting a 2 = a-, = 0: _ a2 It = 12 E I 2
g a 3l
12 E l 3 G A 2L 2 h
12
1 I- Cl 2
GJ 12 E b i = ~Lr + T L T W 3 + x l 1*) AE 17 ku Sym 0 12 / f x 3 ki2 L? 12EI* x 2 L3 b1 ^22 ~ Sym " 0 - 12EI* x 3 A = L3 1 2 /|X 2 L3 0 0 -6 /| L2 0 6 /? L2 0 0 12El% L3 j o i i " i ! 0 ___ I 1 1 2 E lt \ Z> 0 0 6 /f ' L2 6E m 2 -'-2 L FA* (4 + a2) 0 6 / ; L2 0 6 /* x 3 (15-76)
0
(4 + 3) El*
(change sign of (2, 3) and (3, 2) in k j 2) - 6 /2x 2

b i L 2 L 2 n (2 __ \ m. fl2) ___ ^ 6 E m 2
6 /* x 3
L 2 0 _. .
B =
6E I$x3 L L2
(2
,) L _
SEC. 15-9.
523
-6 E I% x2 bi L2 E lf (4 + a 2) iSym
6 /*.x3 L2 0 (4 +
3)
/f T -_
(change sign of (1, 2), (1, 3) in k 22) Finally, the fixed end forces due to a concentrated transverse force and a uniform transverse loading are sum m arized below.
C oncentrated Force PC2
12E I 3 3 ~ g a J ?
x X-C i
X c"
M
b
L c I - flj/2^ 1 + a3 J (15-77)
M B .3
C(1 - x r) 1 = L P C2.X (*c)2^c2 y
F B2
M M
bi ai
= x 3(xr P C2 + //,2) ~ m bi =
~ P c 2
F a2
~ ~ ^ if2
+
a3 = L (^ c P C 2
FH2
Concentrated Force PCi
a-, =
12EIZ G A jL ~ X c) ( ~7~T V 1 + 2
P C l(x c)2 +
M [12 ~ ~
F
b
"r
M fll = X 2( x cP c) + Fb3)
M A 1 ~ M }n
^ A3 ~ ~ P c3 ~ F
b
524
ENGINEERING THEORY OF AN AR BITRAR Y M EM BER
CHAP. 15
Concentrated Torque T CI M Bl ~ T CIXC M j i = T C1{1 3cc) Uniformly Distributed Load, b2 b ?L
(15-79)
FB2 FA 2 ~ ~ ~
M iJ3 M a 3
M -2
12
'
(15-80)
Mb i = M ,u = 0
Uniformly Distributed Load, b3
F bi
F A*
2 fe,/?
12
M B2 M az M in Uniformly Distributed Torque, m y
(15-81)
= 0
m vL M Bi = M,,,
(1 5 - 8 2 )
15-10.
MEMBER MATRICESTHIN PLANAR CIRCULAR MEMBER
In this section, we generate the flexibility and initial deformation matrices for a thin planar circular member, of constant cross section, using matrix operations. We include extensional and transverse shear deformation for the sake of generality. Some of the relations have already been obtained as illustra tive examples of the force and displacement methods. In particular, the reader should review Example 14-6, which treats planar deformation, and Examples 15-4, 15-5, 15-10 for out-of-plane deformation. The notation is summarized in Fig. 15-9. By definition, Y2 and T3 are principal inertia axes and y 3 0, i.e., the shear center lies in the plane con taining the centroidal axis. It is convenient to take the basic frame (frame n) to be parallel to the local frame at B. The three-dimensional forms of Ra, R^, and X BQ are
=
Rb =
"cos tj sin rj 0
sin cos tj 0
fj
0' 0 1.
R?? o ' T
0 u
(15-83)
= R"*
SEC. 15-10.
THIN PLANAR CIRCULAR MEMBER

0 0 0
525
V"
V*
! R(l - cos rfi 1 R sin ij

ij
ft(l COS }]) We use
R sin
R, = Rj'' Rrt = I = [^(1 cos ij) R sin ;/] for planar deform ation and R,
Y A" Bq
(a)
Rp - R 29 R( 1 cos ?/) R sin t]
(b)
for out-of-plane deformation. Since the complete flexibility m atrix is desired, it is just as convenient to work with submatrices of order 3 as to consider separately the planar and out-of-plane cases.
-r.h h i A,BVUM T
1 *
., w *
F ig . 1 5 -9 . S u m m a ry of n o ta tio n (o r a p la n a r c irc u la r m e m b e r.
We consider the member to be thin and use the local flexibility matrix defined by (15-38). Expanding (15-66), (15-68) leads to the member flexibility matrix.
526
CHAP. 15
h AR2 e i2 C ,= GJ
ai ~ ae +
a2
(Ij
/3
g ^ 2r 2
GA3fi2
ae -
as
ci + cj) < = 2 = 1(1 - c,) y2 C3 = C t 1R c4 = c, + c, ( 1

V
(15-84)
h. R
RL &
E l 3 12
(3 + ,)
Symmetrical
+ sin dB
2+
cos 0B
+ a2)
s in 2 0B
(1 + a2) R' , r {^n(c 'i + c* ) Li,

2 c , sin 0 B c 2 sin 0 c o s 0]
0 0
ff3 =
0 0
+ c 3 s in
R2
- - (0 * 3 sin
0)
R2
- ( J cos0)
t-fj
R2
{- c A t/j
R2
c? sin
/*
+ Cj sin 0 cos 0B }
- c3(l - cos (?)}
Eh
ft
(c(0B ~ c2 sin 9b c o s 0 b)
Symmetrical
V
* 2 2_
i?
tT 2
(c2sin2 9g)
(c i^ b +
c2
sin 0 cos 0fl)
F >2
R0 Eh
SEC. 15-10.
THIN PLANAR CIRCULAR MEMBER
527
We consider next the determination of the initial deformation matrix due to an arbitrary concentrated load at an interior point, C. Now, the flexibility matrix for the segment AC referred (o the local frame at C, which we denote by fitc; is known. We just have to change 9S to 0C and superscript b to c in (15-84). When the external load is referred to the local frame at C, the displace ment at C is given by Wc (a) The displacement at B due to rigid body motion about C is
TW c
(b)
Finally, we can write e n k = m u

<
= w v '* - r vACw c u (15-85)
= b = ( R j f ic .n + XfecTR ^ c r 12)Pc
+ (RX c , i z + Xfer Kjfic. h YTc

0 = <
= ( R f e T 12)P i + (R/ftiC, 2 lW c
The uncoupled expressions follow. Planar Loading
528
CHAP. 15
O ut-of-Plane Loading
V0 , 3 = UB 3
C4 ) - C2 S ffi t l l I. + c3( sin 9C sin 0B + sin ^c)\Pcz

=
R3 f
TTT- < 0 c ( c ! COS ??c +
d c COS
0B
R 2 ( J ) + l - c . O c cos rjc + c2 sin 0C cos 9B + c3 sin 6C>TC C1 tl2 J
R2 f ) + CjOc sin rjc - c2 sin 0C sin 9n - c3(l - cos 9c ) y c2 r 2 ( 0h o, 1 = < 4 i - - <!-C!0c cos vie + c 3(sin dB - sin ;/c ) 2 (. -j 4- c2 sin 0C cos 0B>PC$ R f J 1 + cos nc - c2 sin 0C cos 0b >7ci ^ 2 (. J. + | Cl0C Sin + C2 Sn 0c Sn + C^ C S 0lt ~ C S lc ^
(15-87)
R2 [ 0&. 2 = B2 = Y f jClflc Sil1 ^ - c2 sin 0C sin 0B>PC3 +

R <
El
c t 0c sin r/c + c2 sin dc sin 0By c l 1 sin 0C cos 0By r cz
R f + -g j- jCiflc cos Vc +
c2
W hen the loading is sym m etrical, one can utilize sym m etry to determ ine the fixed end forces. T h e m ost convenient choice of unknow ns is the internal forces at the m idpoint, i.e., 9 9 Bj 2 \ F \ and M 3 are unknow n for the p lan ar case and only M 2 is unknow n for the out-of-plane case. Explicit expressions for the fixed end forces due to various loading conditions arc listed below.
Planar Loading
Fig. 1 5 -1 0 defines the n o tatio n for the p la n a r case. W e consider tw o loadings: a concentrated radial force P applied at C, and a uniform distributed radial load b2 applied per unit arc length over the entire segment. T he basic fram e is chosen to utilize sym m etry. W e determ ine the axial force and m om ent at C from the sym m etry conditions u x = co3 = 0.
C
ase
1 C
o n c e n t r a t e d r a d ia l f o r c e
P c2
~ ~2 ^
SEC. 15-10.
THIN P L A N A R C IR C U LA R M E M B E R
P
529
sin a . 1 + a2 . . ------- (1 - cos a) H -------------sin a a 2 sin a ( \ + a2\ . ---------- (_ j ---------- sm a cos a a V 2 7 cos a sin a \ , 1 J?P j l + 1 j --------" 2~ ( a \ a. F > > _ v _ r 1 til /li 1 ci P F" Fn [ B 2 1 AZ Mc
in
(15-88)
PR ( . M a = <sm a
cos a , /s m a -----------1 -------cos a a V a
C a s e 2 U n ifo r m d is t r ib u t e d r a d i a l l o a d b 2
Fd = 0 P ci = < P= - e4> ) sin a

( \ + a ,\
sin2 a
/ l + a2\
. ! sm a cos a
530
GHAP. 15
sin a \ 1 + ------ I a ' Fn Bi Fn Ai = ~ R b 2(cosa ae F"b2 ~ âz ~ ~Rbj sin a
_ __ gjjl # M b ~ M a = R 2b2ae<j) --------- cos a
(15-89)
Out-of-Plane Loading Figure 15-11 defines the notation for the out-of-plane case. We consider four loadings: a concentrated force P, and a couple T both applied at C; a uniform distributed force b3 \ and a uniform distributed couple The bending moment at C is obtained using the symmetry condition ft>2 = 0.
\ Mc\
f'Ci
Fig. 1 5 -1 1 . N otation for o u t-of-p lan e loading.

C a s e I C o n c e n t r a t e d f o r c e
PR c2 sin2 a + c3(l cos a)

cyOt + c 2 sin a cos a
M Ci -
SEC. 15-11.
FLEXIBILITY MATRIX CIRCULAR HELIX
531
M bi =
= ( 1 - co sa )
(1 5 -9 0 ),
_ PR Mg2 ~ ~M j 2 M C2 sin a
C ase 2 C o n c e n t r a t e d t o r q u e 7
^C3 0 T M c, = T
c2 sin2 a
(15-91)
2 a c t + c2 sin a cos a
Mn m = Si'Al - - I
M%2 = - M ^ = M cl F* = 0
C a s e 3 U n i f o r m d i s t r i b u t e d l o a d /?3
^C3 = Mcl = 0
R 2b Cl^ n a a) + c 2 sin a0 cos a) + Cj(sin a a cos a) act -f c2 sin a cos a Mg, = = R 23(sin a a cos a)
2
(15-92)
M"b2 ~ ~ ^ A
- M C2 K263(a sin a 1 + cos a)
Fh = Fa = -PRa
C a s e 4 U n i f o r m d is tr ib u te d c o u p le m ,
Fc3 ~ MC1 =
M C2
c,(a sin a) + c2 sin a(cos a 1) = m ,/? ----------------- ------- r-----------------------aCj + c2 sin a cos a
(15-93)
M"CI M"Al m t R sin a M p2 = M A n2 M C2 m-L-Rd cos a) 15-11. FLEXIBILITY MATRIX-CIRCULAR HELIX
In this section, we develop the flexibility matrix for a member whose centroidal axis is a circular helix. The notation is shown in Fig. 15-12. The principal inertia direction, T2, is considered to coincide with the normal direction, i.e., the inward radial direction, at each point. We also suppose the cross-sectional properties are constant. For convenience, we summarize the geometrical
532
CHAP. 15
relatio n s:!
Xj R cos 9 x z = R sin 0 x 3 = C0 dS a dd a = [ R 2 + C2)1/2 = constant t = t . = - ( K sin 07 i + J? cos 0z2 + Cz3) a n t 2 = cos 07 j sin 0 ;2 5 = f 3 (C sin ^7! C cos 07, + K73) a R . ----- sin 9 a. cos 6 C sin 0 n _a 0 ! ! R cos 0 i a . j sin 0 ! i C ----- cos 0 a 0) 1 c~ i a
7
(a)
Ra = R = R"* =
pi R i a .
-c( 0B-
R {$in 0n R(cos 0
X"
_ 0Q _
C(0B -
0 ] R (sin 0B - sin 9) j R (cos 9n - cos 0)
9)
The steps involve only algcbraic operations and integration. We first deter mine g "j using (15-66), then \|#tJ from (15 -68), and finally f" with (15-70). In what follows, we assum e the shear center coincides with the centroid and neglect extensional and transverse shear deform ation. W ith these restrictions, 0
'_ l_
g2 2 = K"r/' r g 2R"'7
(b)
GJ 1 El'z EL and the expressions for \J/i; reduce to ^22 22 vJ/12 = XS/ ^ 2 2 ^11 = tyll&BQ The flexibility m atrix for a constant cross section is given below.
f See Examples 4-6 and 5-3.
(c)
SEC. 15-11.
FLEXIBILITY MATRIX CIRCULAR HELIX
533
N otation Dimensionless Param eters
R 2 Ely
C2 h
C 1 + C ty
7~2 7 =
a6 =
El2 GJ 1 - ft!
2
a4 a6
C Z 4 +
ah +
do
3 cia (i 10
2
fle 3 04 2
534
CHAP. 15
Elem ents o f f"j /n hi hi I I Sym hi fz2 i sin 0B cos 0B
'--SI
+ i R2 ~ RF j
-C Z 7#j5
y ^2
+ 2î4 sin 0B
j
}
BJ ~
3 1 2 sin 0B cos Qbj
h i = - g j- j - y
0l ~
sin2 0B + ci4(0B sin 0B - cos 0B + cos2 0B)
A -.1 3
+ cos sin2 * {cos 0b(0b sin 0B eB - 1 1 Bl}) 0 ' ^ 1 2 0B n 1

2
Q2 b c o s 0 b + (fli -
/ 3 1 = -gTj^ j - y
a 4) ( l
cos 07J) a 4 sin 2 0B -I- a y0B sin0|
A2 =
C ^ 2a : f 1
a ) + y 0B - a 10 sin tfB cos 0B - 2a49B cos tfBj>
N
+ ^ r* r V i 5\ w
Sin ^ +
cos 0 sin 0,, >
/
Sin ^ + a lO0 B COS
hi = 1 732 = e i 2 {! - Y 2
0"
a8 + a 8 sin 0B -f a4 sin 0B cos 0B| [fll + a 5)0if 2ai sin 0}i a 6 sin 0B cos 0Bj
(1 5 -9 4 )
^ 3
e i2
Elements o f i 1 2 /l4 f" *12 In h$ f 14pi / h s hs /io /26 /36
t l 2
, a 4 sin 0B + a-jOB + a 8 sin
_ Ca 5 fl2 , Ca jt, (fls J -----< 0B 4- a 8 s n r 0B> ;i5 " ^ { Y ^
-2
/116
/M = ^
C c3a Ca f n
1 cos 0B - ~ ~ ~ a5 ,
sin 0B - 1 + cos
0 j
in 2 0B - a4(l cos 8' + , s m
0 B )j
/-S
j a 8(0B - sin 0B cos 0 )l
SEC. 15-12.
PARTIAL END RESTRAINT
535
fi6 ~
/ 34 =
| 0B j^ A
s in
s in
b| + R y - i(l
-
|^
COS
0g
sin 0BJ
do
COS
0fl)j
h s = - F r i^ C s in dB - 0B cos 0B )}E I2 I J Caâ {sin 6b - dB) fz6 ~ W 7 ~ ^ ln UB B<>

Eh
Ra (
Elem ents o f f j 2
a
e T2
f" l22 ~
{a5dB + a6 sin 0B cos 0 B} sin2 0B (1 - cos 0B)
Sym a . T7 7 2 aa3 . ~ El~2 . " 6 sm

0 COS B g }
aa6
e Y2
ocaj E l]
m i e, FA3
15-12.
MEMBER FORCE-DISPLACEMENT RELATIONS PARTIAL END RESTRAINT
In Sec. 15-7, we considered an arb itrary m em ber which is com pletely re strained at both ends. This led to the definition of the m em ber flexibility m atrix and a set of equations relating the end forces and the end displacem ents. Now, when the m em ber is only partially restrained, there is a reduction in the num ber of m em ber force unknow ns. F o r example, if there is no restraint against ro tatio n at B , M f, = 0, and there are only a unknow ns (where a is the order o f F 8), the rotation Wj, at B has no effect on the end forces. T o handle the case o f partial restraint, we first determ ine the com patibility equations corresponding to the reduced set o f force unknow ns. Inverting these equations and using the equi librium relations for the end forces results in force-displacem ent relations which are consistent with the displacem ent releases. Let Z denote the force redundants. N orm ally, one w ould w ork with the prim ary structure corresponding to Z 0. However, suppose we first express the force at a point, say Q, in term s of the end forces at B, using, as a primary structure, the member cantilevered from A :
<3T
Q
_
35r
Q, 0 T 0
tj
o j-n q
j i qi *
o frn
-S A
_ < 5 5A.
_ O jT S5?B BA
fa)
Next, using the primary system corresponding to Z 0, we express & B r in term s of the applied external load and the force re d u n d a n ts:
= E Z 4- G
(15-95)
536
CHAP. 15
The elements of G are the end forces at B (for Z 0) due to the applied external loads. Note that G = 0 if Z contains only end forces at B. Now, the principle of virtual forces requires
JSa
[Sb a . ^ V 4 g .r y is =
Tu\ +
J/w \
(b)
for any self-equilibrating virtual-force system. Talcing (he system due to AZ results in the compatibility equations for Z. It is convenient to work first with the virtual force system due to A # # . Equation (b) reduces to A#"B T(:n , 4- v& "B) = Tm - :riJ'V "A) = A # V 'r " where V"", f" are the initial deformation and flexibility matrices for the fu ll end restraint case. Substituting for using (15-95), and requiring the resulting expression to be satisfied for arbitrary AZ, we obtain (ETf"E)Z 4- E r( '/ ^ + f"G) = E T- r n = ET(< ?rB - ^ A r W'A) (15-96) It should be noted that %A n arc the displacements of the supports at B, A. We suppose Z is of order q x I, i.e., there are q force redundants. Also, we let i be the row order of ^ (and U). (a * i) (ixl) [ F ) * - -M f (l 5 -97) < /Jx 1) With this notation, E is i x q G is / x 1 and (15-96) represents q equations. For convenience, we let fr = ETf"E (q x q) r n Q ',= r n 0 + f"G (/ X (15-98)
1)
(d)
and the member force-deformation relations take the form frZ = ET(-Tn - i'ly z)
=
(15-99)
- r t z)
E Tm
ycn d u \
We refer to fr as the reduced flexibility matrix since, in general, q < i. Actually, fr is the flexibility matrix for Z and it is positive definite since E must be of rank q, i.e., the force systems corresponding to the redundants must be linearly independent. Note that one can determine f,. directly by working with the primary system corresponding to Z 0. This is the normal approach. The approach that we have followed is convenient when the member flexibility matrix is known.
SEC. 15-12.
PARTIAL END RESTRAINT
537
At this point, we summarize the force-displacement relations for partial end restraint: Z = member force matrix = EZ + G dfcn _ 3n jy-n /gru J A A,0 < * B A -y B = reduced flexibility matrix (q x q ) Erf"E ir l , z + f"G frZ - E T( r " - f \ z) (15-100) - Erm - s r t f w - t \ ,)
fr
Note that, for complete end restraint, Z = &n B

G = 0
E I,
= T 'l
(15-101)
We will use (15-100) in Chapter 17 when we develop the formulation for a member system. Continuing, we let k r - fr_ 1 (15-102) The force redundants are obtained by inverting (15--99): Z = K E rm - <r A rW \ - - n , z) (15-103)
Substituting for Z, the end forces at B are given by = \EKKrm + G (e)
We defined k! as the effective member stiffness matrix: k" = Ek,ET = E(Er f"E)_ lET
(15-104)
In general, k" is singular when q < i, since E is only of rank q. Equation (e) takes the form f e = + K n - K ^b/W a & a ,i = - K m . z + G = - k"'/' o + (I) - k"f")G The end forces at A are determined from (a): & a = & a. i - : n . x n + ?rBA &A.i = ~ & \ 0 - S n A & l.i Finally, we write the relations in the generalized form
(15-106)
(15-105)
538
CHAP. 15
where
kg
(15-107)
k AA
K a = ( K b)t - - w / IfH v or* l M _ otu \t"or">T t B A ^ B A ^ ^ A K eu BA
Comparing (15-107) with (15-53), the corresponding expressions for the com plete restraint ca^se, we see that one has only to replace k" by k^ in the partitioned forms for k"BB, k n BA, and k ^ . The equation for , is different, however, due to the presence of the G term.
Example 15-12 -----------------------------------------------------------------------------------Suppose there is no restraint against rotation at B. Then, Ml, - 0. We take Z - FJ, and generate E, G with (15-95).
{Fb3 = EZ +' G
(a)
For this case, 0 0. The reduced flexibility and stillness matrices follow from (15-98), (15-102), f f"
k jL 1-. K,'l 11j

and the effective stiffness matrix follows from (15-104): k" -
< >
u 0
f n.
0 I 0
(C)
Finally, the force-displacement relations are (see (15-99)):
MiTfl U /i
M l
(d)
Note that premultiplication of 1r " by E T eliminates 0", the relative rotation at B. There is
no compatibility requirement for the end rotations in this case; i.e., the support rotation
at B, which we have defined as oflh docs not introduce any member deformation.
REFERENCES
1. 2. 3. 4. 5. 6.
R etssnhr,
E.: Variational Considerations for Plastic Beams and Shells, J. Eng.
M ech . Div. A.S.C.E., Vol. 88, No. EMI, February 1962.
A. S. and R . W. W o o d h e a d : F ram e Analysis, Wiley, New York, 1967. J. M. and W . W e a v e r : Analy sis o f F ra m ed S tr u c tu r e s , Van Nostrand, New York, 1965. R u b i n s t e i n , M. F .: M a t r i x C o m p u te r A n alysis o f Str uctures, Prentice-Hall, 1966. L i v e s l e y , R. K .: M a t r i x M e t h o d s o f S tr u c tu ra l A n alysis, Pcrgamon Press, London, 1964. D a b r o w s k i , R.: G e k r u m m te dunmvandige Trtiger (Curved thin-walled beams), Springer-Verlag, Berlin, 1968.
H all. G ere,
PROBLEMS
539
7. 8.
V lasov, V. Z .: Thin W a l le d E la s tic B ea m s, Israel Program for Scientific Transla tions, Office of Technical Services, U.S. Dept, of Commerce, Washington, D.C., 1961. Bazant , Z. P.: Nonuniform Torsion of Thin-walled Bars of Variable Section, International Association for Bridge and Structural Engineering Publications, Zurich, Vol. 25, 1965, pp. 17-39.
PROBLEMS 15-1. Refer to Example 15-5. Determine c.t for a typical wide-flange section and a square single cell. Comment on the relative importance of torsional deformation vs. bending deformation (i.e., terms involving c, in Equation (e)). Distinguish between deep and shallow members. 15-2. Refer to Example 15-7. Consider a rectangular cross section and (f>
I fp
j 3\
varying linearly with x ,, as shown in the sketch. Evaluate vB2 /( . 2 ) and IVEhJ vB3 for a range of 4 > and a/b.
Prob. 15-2
C en tro id
XI7
15-3. Determine the reaction at B and translation (in the direction of P) at C for the member sketched. Neglect transverse shear deformation.
P
Prob. 15-3
/
Shear center
C e n tro id
|----- Vertical restraint at B
540
CHAP. 15
15-4. R epeat Prob. 15-3, considering com plete fixity at B. Utilize sym m etry with respect to point C. 1 5-5 . Derive (15-27). Start with the definitions for the strain m easures (see Fig. 15-2), p = R + u dR dS dp
.
dp OS
es
dp
dy2
dp { ? dS \dy2 neglect second-order terms, and note (15-26). 15 -6 . Sum m arize the governing equations for restrained torsion. Evaluate b2 and b3 (see (15-34)) for a sym m etrical wide-flange section and a sym m etrical rectangular closed cell. C om m ent on w hether one can neglect these terms. 1 5-7 . Refer to Exam ple 15-11. Specialize the solution (Equations f) for Id s = XL 1. Verify th at (g) reduces to the prism atic solution, (13-57), when 0B - 0. 1 5-8 . C onsider a m em ber com prising of three segments. A ssuming the flexibility m atrices for the segments are know n, determ ine an expression for the m em ber flexibility matrix in term s of the segm ental flexibility m atrices. G eneralize for n segments. 15 -9. Discuss how you would apply the num erical integration schemes described in Sec. 14-8 to evaluate J P, defined by (15-69). 15-10. Verify (15-73) and (15-74). 15-11. D eterm ine the fixed end forces for the m em ber shown, using (15-77) and (15-79). X2
Prob. 15-11
15-12. 15-13.
Solve Prob. 15-3 using (15-84) and (15-87). Verify (15-90) and (15-91). Apply them to Prob. 15-4.
PROBLEMS
541
15-14. Starting with (15-87). develop expressions for the initial deform a tions due to an aribitrary distributed loading, b3 = b3(0). Specialize for b3 constant and verify (15-92). 15-15. Using the geom etric relations and flexibility matrix for a circular helix (constant cross section; Y2 coincides with the norm al direction) developed in Sec. 15-11: (a) Develop a matrix equation for the displacements at B due to a loading referred to the global frame and applied at 6C . Hint: See (15-85). (b) Evaluate uj'3 for the loading and geometry shown.
Prob. 1 5 -1 5
r f/ 2
15-16. Determ ine the reduced member flexibility matrix for no restraint against rotation at an interior point P. 15-17. F o r the planar mem ber shown, determine E and G corresponding to Z =
m b m
a}
Then specialize for rotation releases at A , B and determ ine ke.
542
CHAP. 15
Prob. 15-17
*2
15-18. Determine E and G for (a) no restraint against translation in a particular direction at B (b) no restraint against rotation about a particular axis at B Hint: Review Example 15-12.
Part IV ANALYSIS OF A MEMBER SYSTEM
16
Direct Stiffness Method Linear System

16-1. INTRODUCTION We consider a system comprised of tn members which are connected at j joints. We suppose the geometry of the assembled system is defined with respect to a global framef and use a superscript o to indicate quantities referred to the global frame. The external force and displacement matrices for joint !< are denoted by ijP k, tf/'l : <*l> (1Pfc]1 (a x J)
-
ii n j |
1 ) f L< |
(0*1) (* * I ) <0*1)
( i* ll
where a is the number of translation (force) components, /? is the number of rotation (moment) components, and i ~ a + /?. Note that a = 2 , [i 1 for a planar system subjected to in-planc loading and a = J. /> = 2 for a planar system subjected to out-of-plane loading. F or an arbitrary system, a = /? = 3. In what follows, we assume the material is linearly elastic and the geometry is linear, i.e., we neglect the change in geometry due to deformation. The governing equations consist of joint force-equilibrium equations and member force-displacement relations. We have already developed the member forcedisplacement relations in Chapter 15, so that it remains only to establish the joint force-equilibrium equations. In this chapter, we apply the direct stiffness method, which consists in assembling the system stiffness and initial force matrices by superimposing the. contribution of each member. In the next chapter, we present the general formulation for a linear member system and obtain the equations corresponding to the force and displacement solution by
t By global frame, we mean a fixed cartesian frame.
II ca *
fvdfi
546
DIRECT STIFFNESS METHOD LINEAR SYSTEM
CHAP. 16
m atrix operations. Finally, in C hapter 18, we extend the direct stiffness m ethod to include geometrical nonlinearity. 16-2. MEMBER FORCE-DISPLACEMENT RELATIONS
In developing the relations between the end forces and end displacements for a member, we considered the mem ber geom etry and loading to be referred to a basic m em ber frame (frame rt) and used A, B to denote the negative and positive ends of the member. The general relations were w ritten (see (15-107)) as
(a)
N ote th at (a) also applies when there is only partial end restraint or internal releases. Now, we define +, n_ as the joints at the positive and negative ends of member n. Replacing B by n +, A by n_, the l'orce-displacement relations for mem ber n referred to the member frame take the form
where
(c)
We transform the force and displacement quantities from the m em ber frame to the global frame for the system by applying q t = m on6U opo _ on. T apn (16-2)
to (b). This step is necessary since we are w orking with joint, forces and displace ments referred to the global frame. The final expressions a re : (16 -3) where the global mem ber stiffness and initial force m atrices are generated with (16-4)
J (). i ~
GL-a
cman , T &?;.
y (). i
O nce the displacem ents are known, we evaluate using (16-3) and then transform to the m em ber frame. Since the initial end force and stiffness m atrices are generated in partitioned form, it is natural to express (16-4) in partitioned form. Using the notation
SEC. 16-3.
SYSTEM EQUILIBRIUM EQUATIONS
547
introduced in Section 15-8, we write
(a x 1)
' o _
(a * a)
(M )
(a x P )
(16-5)
( K). 12 ()(), 22_ (P */?> Expanding (16-4) leads to
( 16-6 )
N ote that k")() is a natural property of the m ember whereas k<)() depends on the orientation of the member frame with respcct to the global frame. The operations defined by (16-6) can be considered as the element m atrix generation phase. The member force-displacement relations satisfy the equilibrium conditions for the member and com patibility between the restrained end displacements and the corresponding joint displacements. Actually, the equilibrium condi tions were used to determ ine Com patibility is satisfied by setting Un = llnir and a ?/.A % n^. W hen there is only partial restraint at an end, there will be displacement discontinuities. For example, if there is a rotation release at the positive end, o>"+ will not be .equal to the end rotation matrix. We have treatedf partial end restraint by defining an effective m em ber stiffness matrix ke. In the derivation of ke, we consider QfA to be the displacements of the supports (i.e., the joints) and enforce continuity of only the restrained end displacements. 16-3. SYSTEM EQUILIBRIUM EQUATIONS
The equilibrium equations for joint k arc obtained by summing the end forces for the members incident on k: (a)
t See Sec. 16-12.
548
CHAP. 16
In general, 0*1 depends on tfl'l and the displacem ents of those jo in ts which are connected to jo in t k. We define 0*, a U as the to tal (or system) external jo in t force and jo in t displacem ent m atrices: ^ qi = ..., ..., 0; x l) (ij x 1)
and w rite the com plete set of ij jo in t force-equilibrium equations as

0
* = g>a +
(16-8)
where contains the jo in t forces required to equilibrate the initial end forces We have dropped the reference frame superscript for convenience. T he m ost efficient way to assemble and 3P0 is to w ork with subm atrices of order i, the natural partition size, and superim pose the contributions of each m em ber which follow directly from (16-3). This operation requires no m atrix m ultiplications. T he term s due to m em ber n are listed below. In 0*Q (Partitioned Form Is j x 1): i n r o w? / + n-.i In y f ( Partitioned Form Is j x j ) : k+n+ in row n +, colum n n + in row " + >colum n n_ in row colum n n +
k__
in row >i_
(16-9)
(16-10)
in row n_, colum n _
Since 3C is sym m etrical, only the upper o r low er half has to be stored. 16-4. INTRODUCTION OF JOINT DISPLACEMENT RESTRAINTS
In this section, we extend the procedure described in Sec. 8 -3 for introducing jo in t translation restraints in the form ulation for an ideal truss to an arb itrary m em ber system. Actually, only the n o tatio n for the jo in t force and jo in t displacem ent m atrices has to be changed. The governing equations are: - & >0 = & > N
J f i ! -* X 2 ^ 1 2 J C 22 ' ' J T u CM 2 j = <
_
-
(3f> , o. 1 + 07> , " a. 2 + > = <
.i N, 2
,/^ r 2j
m s
OT> v J
&
iV. j
The stiffness an d initial force m atrices are assem bled using (16-9) and (16-10). It rem ains to introduce the prescribed external forces and displace m ent restraints. If jo in t q is unrestrained, is prescribed, and we ju st add
SEC. 16-4.
INTRODUCTION OF JOINT DISPLACEMENT RESTRAINTS
549
& to -ô. q- If j i nt 4 is com pletely restrained, is unknow n. W e replace the m atrix equation for & >n q with the m atrix identity,
(a)
Finally, if jo in t q is partially restrained, som e of the elem ents in are unknow n. In this case, we replace the scalar equations for the unknow n reactions by scalar identities. W e suppose jo in t q is partially restrained and, for generality, consider the translation and ro ta tio n restraint directions to he arbitrarily orientated with respect to the basic frame. W e define X \ , . . . , X ' as the o rthogonal directions for the translational restraint frame and X '{,. . . , X " as the o rthogonal directions for the rotational restraint frame. Q uantities referred to the restraint frames are indicated with prim es and a single superscript is used for the total m atrix:
(16-12)
Now, (16-13) W e define . %oq as the total ro tatio n transform ation m atrix: (16-14) W ith this notation, the transform ation laws take the form (16-15) The m odification requires two operations. First, we transform in (16-11) to U\. This is accom plished by prem ultiplying row q of J f , with M oq and postm ultiplying colum n q of JdT with r. In the second step, we replace the equations corresponding to the unknow n elem ents in with identities. This operation can also be represented in m atrix form. Suppose the rth elem ent in r ) lf\ is prescribed. We assem ble four m atrices, E G r '# * , and as follows: 1. E g and Gq E - I, and set Err = 0 Gn. + 1 G =0,
We start with
550
DIRECT STIFFNESS METHODLINEAR SYSTEM
CHAP. 16
2. < W * We start with an jth-order column matrix having zero elements and set the element in row r equal to the prescribed displacement. 3. We start with an ith-order column matrix having zero elements and enter the values of the prescribed forces and moments referred to the restraint frames. Note that element r is zero. Premultiplying transformed row q of J with E? reduces the rth equation to 0 = 0. Then, adding to anc* to ~ q introduces the identity for the rth element of 4/q q and includes the prescribed external forces in We also operate on the q th column of :/ to preserve symmetry and include the terms due to prescribed displacements in The complete set of operations for joint q are listed below:
1. I = 1, 2, . . . , q ~
j r lq = ( j r f^ - r )E, 2. 0>Ni = + w* + & > * = E q ( m x \ q m r ) E (l + g (( 3. t = q + 1, q + 2 , . . . , j jr q t - e , o r - (-K<M0"'Ty K The operations defined by (16 16) are carried out for cach joint, working with successive joint members. We represent the modified equations as = iJ ?% (16-17) (i6 -i6 )
The superscript J is placed on V . to indicate that the joint displacement matrices are referred to the local joint restraint frames, which may not coincide with the global frame. Again we point out that the primary advantage of this modifica tion procedure is that no row or column rearrangement is required. Solving (16-17) yields the joint displacements (local restraint frame) listed in their natural order, i.e., according to increasing joint number. The modified stiffness matrix, JtT*, will be positive definite when the system is stable. Once is known, we transform the displacements from the restraint frames to the global frame, using (16-15), and evaluate the member end forces from (16-3). Next, we assemble the total external force matrix, (P. The contribution
SEC. 16-4. INTRODUCTION OF JOINT DSSPLACEMENT RESTRAINTS
551
of member n is < s? nl in row n+ in row /?_

(16-18)
Finally, we transform the external joint forces from the global frame to the local restraint frames. This step determines the reactions and also provides a statics check on the solution.
E xam ple 1 6 -1 Suppose joint q is completely restrained. Then. ;,U* W* and !P* -- 0. The forms for E, G are . E, = 0; G, = I, (a) and (16-16) reduces to
I. ( = 1,2........< 7 - 1 yr,
2.
O f
I,
3. I = q + j , q + 2......... j
(b)
XT,' = 0; &n.t = = &., ~

E xam ple 1 6 - 2 Suppose joint q is completely restrained against translation. Then, the translation matrix u" and external moment matrix are prescribed. The appropriate matrices for this ease are
E 0
! o' !
b-
X (* 9~ J
J*
= 1,
! o' r />. (a)
w
(0,1
E xam p le 1 6 - 3 -----------------------------------------------------------------------------------------We consider the case where joint q is restrained with respcct to translation in one direc tion and there is no restraint against rotation. This corresponds to a roller support. We take X \ to coincide with the restraint direction and X'r, X \ as mutually orthogonal directions comprising a right-handed system. The translation, u'q l, is prescribed. The prescribed forces are ^ 2 . f"t3 > and T.
552
CHAP. 16
We first assemble 0 t oq. From (16-14).
0 taq =
where
R f = [7 ]
tr
I *3
(a)
r, .9 - 1, 2, 3 (b)
i rs = cosfz;, x")
The forms of E, G, tfi*, and are
'0 E ,= ! h. uql 0 0 o7. < #* Ga =

0 ! 0 J1 ___ I
i
'0
K ,
0
(C)
It ; J
We specialize the results for a planar system subjected to planar loading. In order for only planar deformation to occur, the translation restraint direction must lie in the plane of the system, which we take as the X 1-X2 plane. It is convenient to select the orientation of X 2 such that X 3 coincides with A'j. The specialized forms are R ,"
T
R f' = M
r.s = 1 , 2
t - cos(X n X)
0 ! i ! P
j *Ci
! o !
(d)
Finally, we consider the case o f a planar system subjected to an out-of-plane loading. The translational restraint direction must be parallel to the X% direction in order for only out-of-plane deformation to occur. For this case, w"j, Tq l , and T 2 are prescribed. The specialized forms are
E =
Note that (e) is obtained by setting
i -i______ i i 1 i II
i ' .j ______ 1 1 ,
ii
>
W \
= < tfd
m 0* = I3 0 I I, ( to 'I II * K
a
-LL_
[ 1
(c)
[r 1 1-42)
= 1, /? = 2 in (a) of Example 16-2.
REFERENCES
553
REFERENCES
1. Lives le y , R. K .: M a t r i x M e t h o d s o f S tr u c t u r a l A n a ly s i s , P ergam on Press, L o n d o n ,
1964. 2. 3. 4. H. C .: In tro d u c tio n to M a t r i x M e t h o d s o f S t r u c t u r a l A n a ly s i s , McGraw-Hill, New York, 1965. R u b i n s t e i n , M. F .: M a t r i x C o m p u t e r A n a l y s i s o f S t r u c t u r e s , Prentice-Hall, New York, 1966. Gere, J. M. and W. Weaver: A n a ly s i s o f F r a m e d S tr u c tu r e s , Van Nostrand, New York, 1965.
M a r t in ,
17
General Formulation Linear System

17-1. INTRODUCTION We consider a system com prising m linear elastic m em bers interconnected at j joints. We suppose there are i degrees of freedom per jo in t (i.e., the jo in t displacem ent and force m atrices are of order i x 1) and the geom etry and jo in t quantities are referred to a global frame. Also, we neglect geom etry change due to deform ation. In the previous chapter, we applied the direct stiffness m ethod, which is actually a displacem ent m ethod, to this system. Now, in this chapter, we first develop the governing m atrix equations and then deduce the equations corresponding to the force and displacem ent solution procedures. We also establish variational principles for the force and displacem ent methods. Finally, we discuss how one can introduce m em ber deform ation constraints in the displacem ent m ethod. Since the basic steps involved in the m em ber system form ulation are the same as for the ideal truss form ulation, we recom m end that the reader review C hapters 6 through 9 before starting this chapter. Let r be the num ber of prescribed jo in t displacements. Then, the total num ber of jo int displacem ent unknow ns, nd, is = U- r (17-1)
The total num ber of force unknowns, nf) is equal to r (the reactions corre sponding to the prescribed displacem ents) plus q T, the total num ber of member force unknowns: nf = r 4- qT = r + {qx + + ' ' + 9m) (17-2)
where qn represents the num ber of force unknow ns for m em ber n. By definition, qn is equal to the num ber of force quantities that have to be specified in order to be able to determ ine the total internal force m atrix (J^) at an arb itrary point. If the m em ber is fully restrained at each end, qn = i. F o r partial restraint, qn is equal to i minus the num ber of independent force releases. N ote th at when the m em ber is pinned a t both ends, qn = 1 since there are only five independent m om ent releases.
554
SEC. 17-2.
MEMBER EQUATIONS
555
There are qT equations relating the member forces and the jo in t displacements. Also, there are ij equilibrium equations relating the external joint forces and the member forces. The form ulation is consistent, i.e., the num ber of equations is equal to the num ber of unknowns. If nf - ij, the system is said to be statically determinate since the force unknowns can be determined using only the equi librium equations. The difference, n f - ij, is generally called the degree of static indeterminacy, and represents the order of the final system of equations for the force method. F or the displacement m ethod, the final system of equations are of order rzrf. In what follows, we first establish the member force-joint displacement relations by generalizing the results of Sec. 15-12. Then, vve assemble the joint force-equiiibrium equations. Finally, we introduce the joint displacement restraints.
17-2.
MEMBER EQUATIONS
The reduced member equations were developed in Sec. 15-12. F o r conve nience, we summarize the notation and equations below (see (15-100)):
mem ber force matrix (q x 1)
=
A
EZ
^
+ G
A ,o BA'-* fi
(i x 1)
-
~ -^ A ,o
3T"b a G
ST'b a E Z
f"
= member flexibility matrix (/ x i )
(a)
f r reduced member flexibility m atrix (qn x q ) = i rn = member deform ation matrix (i x i) = U "R
ErfE
i 'n 0 , = initial mem ber deformation matrix (i x i) - 1'"0 + f G
f,.z - ET( r n - r n 0 . z)
These equations include the effect of partial end restraint, internal force releases, and reductions due to symmetry or antisymmetry. We can also use (a) for complete end restraint by setting E = I,- and G = 0. Now, we introduce new notation which is more convenient. First, we note that G contains the end forces at B due to the external mem ber loads acting on the primary structure defined by Z 0. Also, WbaG are the end forces at A. Then we write (17-3) Next, we note that the equation relating Z and V f '" ,, is a compatibility requirement. The term frZ + 2 is the relative deform ation in the positive sense of Z due to the mem ber loads and the member redundants, Z, whereas E r y n is the relative deform ation in the negative sense of Z due to support (joint) movement. T he net relative deform ation m ust be zero for continuity.
556
GENERAL FORMULATION LINEAR SYSTEM
CHAP. 17
Then, we define i r r reduced member deformation matrix (q x 1) = Err n = ET cJ /rB - zen B A rW'A) 'Vro = reduced initial member deformation matrix (q x 1)
(17-4)
= ET (Y /i: + f"G)
4 EZ
=
With this notation, the member equations take the form
& A .O - &
baE
(17- 5 )
rr = ^ ro + frz = et(w%-ar",Jwn A )
We generalize the relations for member n by setting B n+ E = E
^
A = n_ Z = Z
f r, n r
(17-6)
__ w / -
or'ff
BA
_
^ ' /
ro
to,n
V. n
Since the joint quantities are referred to the global frame, we must transform the end forces and displacements from the member frame (frame n) to the global frame (frame o), using < j)fn _ (%on0f{o L% n, T:jfr" The final equations follow.
Member Forces End Forces
(b)
- m " TE)zn + - - w m' t ^;;eiz +

Member ForcesJoint Displacements {q Equations)
r ,n ^ ' r o . n + f r, n^n
(1 7 -7 )
= (EZ0tmYK. ~
tM"YK
(1 7 -8 )
The force translation transformation matrix, 3, is a second-order tensor, i.e., it transforms according tof seq = Mqp' Tarpm qp where p and q are arbitrary orthogonal frames. Then, 3 C = Sr+- = S T ' T3 C "nm a and it follows that 0 ton' T'3Cn ^W "'T SCn K Tffi = T
t See Sec. 10-2.
(a)
(17-9)
(17-10)
SEC, 17-3.
SYSTEM FORCE-DISPLACEMENT RELATIONS
557
Using (17-10), we can express
and
i" r
as (17-11)
= &i..o ~ .r o n (-#n ,-TEn )zH

^ r . n = (Ela r w n ,. - 3 C V K .)
We prefer to work with since it is a natural property of the member whereas depends on the selection of the global frame. 17-3. SYSTEM FORCE-DISPLACEMENT RELATIONS
Equation (17-8) represents the q force-deformation relations for member n. By defining general flexibility and deformation matrices, we can express the complete set of qT member force-deformation relations as a single matrix equation. We let Z = total member force matrix {qT x 1) {Z Z 2, , Zm} 1 ' total reduced member deformation matrix {qT x 1) ir g total reduccd initial member deformation matrix (qT x 1)
-
i 'i/~
I 9 > < ? , 1? r ro, 2 * ' > 7 rof tn)
-//'
'i/ '
(17-12)
f = total reduced member flexibility matrix (qT x qT)
f,
Note that f is quasi-diagonal, symmetrical, and positive definite. W ith this notation, the qT forcc-deformation relations are given by i" = + fZ (17-13)
It remains to generalize the deformation-displacement relations. We define ll as the total joint displacement matrix referred to the global frame. (ijx 1) 6 u = q/2, .... i/ff (17-14) and express V as
/ = s / ft
(17-15)
The partitioned form is
558
GENERAL FORMULATIONLINEAR SYSTEM
CHAP. 17
Row n of < 5 / corresponds to member n. The submatrices in row n are of order qn x i. Now, we see from (17-8) that there are only two non-zero elements in row n and they are at columns +, h_. The assembly of s / is defined by
' n t t4
, M E rM" = -E = - E l m onSC n- T K
= 0 s ^ n+, n._ s ~ 1,2,... ,j

n
1 , 2 , . . . , m
It is of interest to express M in factored form. First, we define the following matrices: (im X1 ) _____ * _ } (im x .1) .yp01 (ini x im) M"
E -
E7 E
(im x < fT)
(1 7 -1 8 )
sr\
9C\ (im x im)
.T 0 ! 3C =
'JY 'O
(t>7?
im )
Using this notation, the expression fox ^ takes the form TT - E 7 ' + - E r a rr _ - e t .^("//+ - ar-T<_)
(a)
Next, we relate W +, to using member-joint connectivity matrices for the positive (C+) and negative (CL) ends: <%+ = C +U W- =
(1 7 -1 9 )
( 1
SEC. 17-4.
SYSTEM EQUILIBRIUM EQUATIONS
559
N ote that rows n of C +, C _ correspond to member n. There is only one nonzero element in a row. For row n, we enter + I ; in column n+ of C + and column of C_. Finally, combining (a) and (J7 19), we have
1
' = = (ETa r c + - E l : r l M C-Y% = (Et k c + - s r 'Tc - Y V
(1 7 -2 0 )
and it follows that = E t-# C . - E t * -tj C _ = (Er S?)(C + - % 'TC_) For an ideal truss, (17-21) reduces to (see Equation 6-28) s i = a(C+ - C_) where a contains the direction cosines for the bars.
1 7 -4 . SYSTEM EQUILIBRIUM EQUATIONS
(17-21)
(b)
We have used the member force-equi librium equations in developing the member force-displaccment relations, so it remains only to satisfy equilibrium of the joints. There are / equations for each joint, and a total of ij equations. The expressions lor the end forces in terms of the member forces are given by (17 7). Assembling the joint force-equilibrium equations involves only summing at each joint the end forces incident on the joint. We define as the total external joint force matrix referred to the global frame: & = # > \ , #>1} (17-22) <7* 1) and yP, as the initial (Z = 0) joint force matrix: (17-23) < 0 * 1) The elements of are the joint forces due to external forces acting on the members with Z = 0. We express the complete set of equations as & = & >, + M Z
12
~7~ i/> " '
.= <
>+ m 3.
_I I
I ' _|__' * : i
22
/l Z2
(1 7 -2 4 )
J2 |
i ; _)___ ;__ ' I 1 zw 1
We assemble and working with successive members. The contribution of member n follows from (17 7): In y ; , in row n. m row n
(17-25)
560
CHAP. 17
Column n o f 38: m n, n = = 8
?ion' Tj rE (17-26)
- m on' T?x; e
~ % nm on' T e
- o s = n +, rc_ 5 = 1, 2 , . . . , j C om paring (17-26) w ith (17-17), we see th a t 09 W e let & + .o = (17-28) s'JT (17-27)
Then, we can express (9*, as
0>t = C l T& +f 0 + C l . T# ' _ i0

17-5. INTRODUCTION OF JOINT DISPLACEMENT RESTRAINTS; GOVERNING EQUATIONS
(17-29)
The governing equations for the unrestrained system are = 3P, + + .^ rZ (17-30)
= i~0 + fZ = r/^/
N ow, we suppose r jo in t displacem ents are prescribed. W e rearrange W so th a t the prescribed displacem ents are last. We also rearrange and f^y. (H rf * i ) (r*I! ('Mx 1) (17-31)
p, -
( Pf f ,1 ( i X l ) ; K .> (r x I)
where U 2, P l5 and P ; are prescribed. W e use B, A to represent the rearranged forms of s it: jrf ^
- A - [A,
-> B =
( ? T x I d ) (<Jr x r )
| A2] (1 < 1 ('Id X<rr)

(r x yT )
fifi 1
b
(17-32)
; at _
SEC. 17-5.
JOINT DISPLACEMENT RESTRAINTS
561
Finally, we w rite the eq u atio n s for the restrained system as P i = P , x + B 1Z = P l j l + A [Z P 2 = P / , 2 + B2Z - P / . 2 + A lZ "y fZ + ~ i/ = A tU j + A 2 U 2 - B fU j + B2 rU 2
K ecls-)
(r eqs.) (qT eqs.)
(17-33) (17-34) (17-35)
T h e unknow ns are the q T m em ber forces (Z), the nd displacem ents ( U ; ), an d the r reactions (P 2). If the restrain ts are parallel to the directions o f the global frame, the tran s fo rm atio n o f .si to A (or 0b to B) involve only a p erm u tatio n o f the colum ns of s i (rows o f # ) . T h e sam e p erm u tatio n is applied to the row s o f # , . S uppose jo in t q is p artially restrain ed an d the restrain t d irections do n o t coincide w ith the global frame directions. We first tran sfo rm th e force and dis placem ent m atrices for jo in t q from the global fram e to the restrain t fram e, using ,J U q = T?/q q (17-36)
T his step involves postm ultiplying colum n q o f s i by M oq' T an d prem ultiplying row q o f y/*,, iiS by M oq. W e w rite the transform ed eq u atio n s as < 9 > J 0
>\ +
= t \ + fZ = si*'*/*
(.17-37)
w here the superscript .7 indicates th a t jo in t forces an d displacem ents are referred to local restrain t frames. T h e final eq u atio n s are o b tain ed by p erm u tin g the colum ns o f s i 3 (rows of ffi3). the row s o f # / , an d then partitioning. T he tran sfo rm atio n of l l to U can be expressed as a m atrix product, U - DV = I M T 'W w here M 0j contains the ro tatio n m atrices for the jo in t restrain t frames,
OJ
(17-38)
o2 (17-39) M o} and 11 is the row p erm u tatio n m atrix. O ne can generate I I by startin g w ith I and perm u tin g the row s according to the new listing o f the jo in t displace m ents, i.e., w ith the prescribed displacem ents last. N ow , D is a n o rth o g o n al m atrix, D -1 = D 1 (17-40) Then, % = D 7U , , P = D #
(a )
562
CHAP. 17
and it follows th a t P, = D ^ r A = c/D7 (17-41)
B = A 1' = T)SS
T he partitioned forms are obtained by partitioning D : D = Finally, we can write A, = x /D [ = B [ A2 = s S D 2 t = Bf Pm = t> .^ P/. 2 = D 20*r T o determ ine the requirem ent for initial stability, wc consider (17-33), B jZ = P j P j
1
D, D2
(",i x i j ) (r x i j )
(17-42)
(17-43)
(a)
which represents nd equations in qT unknow ns. F o r the equations to be con sistent for an arbitrary loading, the ran k of B t m ust equal nd. Therefore, the stability requirem ent for the system is r(B .) = r(A J = n(l (17-44)
Since B! is o f o rder nd x qT, a neccssary but not sufficient condition for stability is qT > nd = ij - r (17-45) E quation (17-44) is the stability requirem ent for a geom etrically linear system. It is also the initial stability requirem ent for a geom etrically nonlinear system. In the next chapter, we develop the stability criteria for a geom etrically n o n linear system subjected to a finite loading.
17-6.
NETWORK FORMULATION
In the form ulation presented in the previous articles, we w orked w ith the actual jo in t displacem ents and external jo in t forces referred to the global frame. T he governing equations are given by (17-30), which we list below for convenience: & = 0>1 + (a) t T = + fZ = where
s i = {Er)(c+ - <r-rc_)
0
> l = c \ m T. P + i 0 +
O ne assembles using (17-17), (17-25), which are actually the expansions of (b). By introducing new jo in t variables, we can express in term s of only one
SEC. 17-6.
NETWORK FORMULATION
563
connectivity m atrix, C+ - C_ = C. The rule (17-17) for assembling still applies except th at now s / _ = nn y. Let Y denote some arbitrary point. Suppose we express the actual force and displacem ent m atrices for jo in t k in term s of their equivalents at point Y. We define gPy, k = statically equivalent force at Y due to actual force m atrix at jo in t k. $ / y. k displacem ent at joint k.
Y Pk,
the (17-46)
due to rigid body m otion about
T he actual and equivalent quantities are related by or. Tffl/o

kY
(17-47)
Y. k
where
.A !(y
o 1 5 ^ C .* A .. XY 3) ~(xk2 ~ V Y 2 )
t planar
(X /c 3~
0
{X k2 xy?.)+ -(4i - -4i)

0
out-of plane
(17-48)
(*Z i - *n)
Wc could operate on (b), but it is m ore convenient to start with (17-11): i r r,n = (E lM n){K + ~ 3 T TK ~ ) N ow, by definition,
A"
(c)
(d)
Jjro
jf"
Substituting for we obtain
using (1 7 -4 7 ), and noting that

jy o , T .jfo . T _ ^ n+ n_ * * * 'n _ V < jfo , T A n + Y
(e) j (17-48)
= ( e ^ m M3cn\ T , ) m iH + -
The rem aining steps are the same as followed previously. We let w Y = { r r i l , ^ , 2 and write = SjfyWy The generation of ,stfy follows from (17-48). F o r now n, .< X ?v Y, nn+ s
m%_j)
(17-49) (17-50)
(17-51)
(qn x 1)
^ Y , ns = 0
n_ _
5 ] j 2, . . , j
564
CHAP. 17
T o express
i*1 factored form , we let < 7fO "M t y GfO 9C % 2 +y oy'o

in + r
f im x im)
(17-5 2 )
Then, - E T^ f r ( C ( - C _ ) = E r .^ r + r C W e transform the jo in t forces, using (17-47), and w rite the resulting equations as gpY = + .s /J z + fZ = T o relate corresp o n d in g term s in (a) an d (17-54), w e generalize (1 7-47): (17-5 5 ) m
JC\ y
(17-53)
(17-54)
3CC JY
m /'o ^ 2 Y.
(v x y) (nra
J
j Y
It follows th a t
jrfy = SfHTjY & >Yl =
(17-56)
T he expression fo r ^ reduces to (1 7 -5 3 ) when (d) an d (e) a re introduced. T he fo rm u latio n developed above can be in terp reted as a network form ula tion since the connectivity term ap p ears seperately in the factored form o f c/. A simplified version w hich does n o t allow for m em ber force releases has been presented by Fenves an d B ranin (see Ref. 1). T he only o p eratio n al ad v an tag e o f n o t w orking with the actu al jo in t quan tities is in th e g en eratio n o f ,c/,m+ an d This ad v an tag e is trivial com pared to the ad d itio n al o p eratio n s required to generate iP Yy 0 > Yl, to intro d u ce the displacem ent restraints, and finally, to transform llY to Uonce the solution is o b tained. A n o th er serious disad v an tag e is th a t the eq u atio n s tend to becom e ill-conditioned. Fenves a n d B ran in s p rim ary objective was to show th a t the governing equ atio n s for a m em ber system can be cast in a form such th a t geom etrical an d topological effects are separated, i.e., a netw ork form ulation. D iM aggio and Spillars (Ref. 2) have also presented a n etw ork fo rm u latio n for a rigid jo in te d m em ber system. A ctually their form u latio n is a special case of o u r first fo rm u la tion. It is not, strictly speaking, a true n etw o rk fo rm u latio n since connectivity is n o t com pletely sep arated from geom etry (see (17- 21)). T he only way th a t one can sep arate connectivity from geom etry is to redefine th e jo in t variables. N o te
SEC. 17-7.
DISPLACEMENT METHOD
565
th a t the ideal truss is an exception. C onnectivity an d g eom etry are naturally uncoupled for this system . W hether o n e in terp rets the governing eq u atio n s for a m em ber system from a netw ork view point is of academ ic in terest only. In the displacem ent m ethod, the eq u atio n s reduce to the eq u atio n s for the direct stiffness m ethod. T he only possible ad v an tag e o f th e n etw o rk in te rp re ta tio n is in the force m ethod. T h ere one can use certain concepts o f the m esh m e th o d ! to select a prim ary stru ctu re, provided th a t th ere are no m em ber force releases o r p artial jo in t restraints. H ow ever, th e selection o f a p rim ary stru c tu re for a rigid-jointed fram e having fixed su p p o rts is q u ite sim ple, and even this ad v an tag e is debatable.
17-7.
DISPLACEMENT METHOD
T he governing eq u atio n s are given by (17-33), (17-34), an d (17-35). O nce the m em ber forces are know n, we can find th e reactio n s from (17-34). N ow , we sta rt by solving (1 7 -3 5 ) for Z in term s o f the displacem ents, Z - Z, + k A ^ w here Z r = initial m em ber force m atrix (qr x I) = Ik T 1
_
+ kA 2U 2
(1 7 -5 7 )
= {kr, 1 1 _______
1
(17-5 8 )
k = f -1 rcduced m em b er stiffness m atrix (qT x qr )
----------
f-l l r, 2 f t r- _
k, 2
N o te th a t k is q u asi-d iag o n al, sym m etrical, an d positive definite. T he m atrix, Z j, co n tain s th e initial m em ber forces due to external loads actin g on the m em bers an d initial d efo rm atio n resulting from fab ricatio n erro rs o r te m p era ture changes. W e su b stitu te for Z in (17-3 3 ) an d w rite the result as Pi = K i + w here K u = A fk A j K 12 = A [k A 2 K i ~ i + A fZ / (nd x nd) (n(, x r) (tid x i) (17 -6 0 ) K U i + K 12u
2
(1 7 -5 9 )
T he elem ents o f P 0 , are the jo in t forces d u e to th e initial end forces. Since A x is o f ran k nd (w hen the system is stable) and k is positive definite, it follows J
t See Sec. 9 - 5 . j See P rob. 2 - 1 8 .
566
CHAP. 17
that Kj j is positive definite. Conversely, if K j, is n o t positive definite, the system is unstable. The jo in t displacements are determ ined by solving (17-59) and the mem ber forces are obtained by back substitution in (17-57). O perating on the restrained equations, as we have done above, is n o t efficient since the various coefficient matrices m ust be generated by m atrix m ultiplica tion. By first m anipulating the unrestrained equations and then introducing the displacement restraints, one can avoid any m atrix multiplication. This procedure corresponds to the direct stiffness method. O perating on (17-30), we obtain Z = Z, + VM m and
= &>, + .c/TZ , + .c /'k * /* # (17-62)
(17-61)
&>0
E quation (17-62) is identical to (16-8). The generation of reduces to (16-9), (16-10) when we introduce the factored forms of Z f. First, we review the definitions of the m em ber stiffness matrices, k"++, k*+n_, k"_n_. The effective m em ber stiffness m atrix (see (16-104)) has been defined as k" - E X . X (a) Transform ing k" to the global frame and applying (16- 107) leads to
1 ,0 _ cm o n . ru n anon
= (E :^ Y K ,n (& l and k U" n "11 ++ = --f,
1.9
*'+l:.o
_ _
"p. n '-*' n
ojro\.a
oro, T mro. T
/\
'
Now, substituting for 0 2 / using (17-21), the expression for jtr = ( c r + - c l s r ) K(c+ - s r o Tc . ) where K = (Er ?)Tk(Er .#) Finally, we expand (d): X = C l.k "C + + C T f ( - k ^ : T)C -
takes the form (d) (17-63)
+ c i ( - K W o r )Tc + + c l ( . r (,k : ^ 'T)C O ne can easily show th a t (17-64) reduces to (16-10) when the properties of C + , C - are taken into account. The initial end actions for member n are f
SEC. 17-8.
FORCE METHOD
567
Using the factored forms for form

& 0
.$4, and Z ;, the expression for & a takes the
= CT +& T& +^ +
+ (CT + - CZ% ')0?tE{ kyo )

The general form o f d e f i n e d according to (16-9) is
& 0
(17-65)
(17-66)
= C ^ +. f + C 7 : # !,,-
Substituting (e) in (17-65) results in (17-66). In Sec. 16-4, we presented a procedure for introducing jo in t displacement restraints and represented the modified equations as (ij eqs.) (f)
Now, (f) consists of (17 -59) plus r relations for the prescribed displacements. We obtain (f) by starting with Kn
u ,) _ fP, - ro. t
U:
12u 2
(g)
and perm uting the rows and columns. This operation can be represented in terms of the perm utation matrix, fl, defined by
(h)
Then,
jf * = n '
'K n j 0
O' I,
n
(i)
=n
r [P i - K . K_uU2 o2
It follows that ,#"* is positive definite when K 1 V is positive definite, i.e., when the system is stable.
17-8.
FORCE METHOD
We start with the governing equations for the restrained system: B jZ - P j - P A , B [U t + B l u 2 - r - r 0 + fZ Pz = P/, 2 + B2Z tar eqs.) (qT eqs.) (r eqs.) (a) (b) (c)
Equation (a) represents nd equations in qT unknow ns where qT > nd. Also. is of rank nd. The system is statically determ inate when qT nd. We let qR be the degree of static indeterminacy, i.e., the num ber of m em ber force red u n d an ts: qR = qT - nd (17-67)
568
CHAP. 17
Since B t is of ran k d, we can solve (a) for nd m em ber forces in term s o f the net prescribed jo in t forces (Pt P 7 J and r m em ber forces. T he com patibility equations for the m em ber force red u n d an ts are obtained by elim inating from (b). This is possible since (b) represents qr equations whereas U t is only o f o rd er nd x 1. In the next section, we specialize the principle o f virtual forces for a m em ber system and utilize it to establish the com patibility equations. We suppose the first nd colum ns o f Bj are linearly independent. If the system is initially stable, the m em ber force m atrix Z can always be rearranged so th at this condition is satisfied. We p artition Z after row nd: (17-68) The elements o f Z R are the force redundants for the system. We refer to the system obtained by setting Z R 0 as the prim ary system. C ontinuing, we partition an d B2 consistent with (17-68):
(<I x q r )
("cl * d)
('i,i x t i n )
B,
B 2
- [ B 1P
[ B 2p
j
!
B ,* ]
B 2 /< ]
(17- 6 9)
(r X<jr)
('x "a)
('*
The equilibrium equations take the form B 1PZ , = Pj. - P / . ! - B lRZ R P 2 = P /,2 + BzpZp + B2Z f( We write the solution of (17 -7 0 ) as Z P = Z P>0 + Z P R Z R The force influence m atrices can be expressed as (n,i x 1)
~
0
(17 -7 0 ) (17 71)
(17-72)
W ir)
l(P 1 ~
P / , i)
u * qn) %p. r "(Bi/) B ,*
(17-73)
but it is n o t necessary to determ ine (B1P) *. Actually, the solution procedure can be com pletely a u to m ate d .t T he com plete solution for Z is
(qT
xqR)_
Z =
[Z p J r 0 1
Zp. R r i 7 j
(17-74)
N ote th a t the m em ber forces due to Z R are self-equilibrating, i.e., they satisfy BX Z = 0. Finally, we substitute for Z P in the expression for P 2 an d write the result as p 2 = P2,o + P i . r Z r (17-75)
t See Sec. 9-2 .
S E C . 17 8.
FORCE METHOD
(r x 1)
1*2. o ~
^ 2 . 7? =
569
where
P j, 2 +
B 2 fl +
B 2p Z P 0
B 2pZ p R
(17 76)
0 x9r) It rem ains to determ ine Z K. E quation (b) represents q r equations m nt, unknow ns, U t. Since qT nd + qR, there are qR excess equations. W e partition (b) consistent with the partioning o f Z,
B r ip B ZP r B 2 fR
T~ B7 IK
1 ^ 4 t * ' T
fpp
PR !
fpR 'KK
IM lz R J
(C)
and obtain the following two sets of equations relating to U j and Z R : BfpU! + B 2p U 2 'fr p f 0, p 4- f//>Zp + fpZK (nd eqs.) B IrU j + B ^ U 2 Y R \'f R + fp>jZp 4- fRRZ R (qR eqs.) (17 77) (d)
T he jo in t displacem ents can be determ ined from (17-77) once Z R is known. E lim inating U t from (d) leads to . . p2 t k02 r R + z l Rr P
fpR^p + fRRZ R + Z Pt R( ' f g_ p + fppZp + fprZ r )
(17-78)
"1r a. R +
E quation (17-78) represents the com patibility equations for the force redundants. Finally, we substitute for Zp using (17-72) and write the resulting equations as fZj,Z - A (17-79) where
(<i R x < ?K )
f7n
7 JP. f / ' p Z p K +
Z p , RfpR + ~
( Zp R f PR) T
Z p , jW p +
f p p Z p 0)
(K * 1)
A P 2 , / ? U 2 ( ^"o . R +
fp R ^P .o)
(17-80)
These equations are sim ilar in form to the corresponding equations for the ideal truss developed in See. 9 -2. T he flexibility m atrix, fZr, can be expressed as ..~r Zip, U \n ' ^P.R ipp i fpR
~P. R
RR Wn 1 lfT L P, R
(17-81)
I, Now, f is positive definite for a deform able system. Then, it follows th a t fZjl is also positive definite. In a later article, we consider the case where certain m em ber deform ations may be prescribed. O nce the prelim inary force analyses have been carried out, the rem aining steps arc straightforw ard. We generate f Zn, A, solve for Z K, and then determ ine
570
CHAP. 17
Z P, P 2 by b ack su b stitu tio n . If th e disp lacem en ts arc also desired, they can be d eterm ined by solving (17-77). T h e final n u m b e r o f e q u atio n s for the force m e th o d is usually sm aller th an for the disp lacem en t m e th o d (qR vs. n(l). H ow ever, th e forcc m e th o d requires considerab ly m o re o p eratio n s to g en erate the eq u atio n s. T h e force m e th o d can be co m pletely a u to m ate d , b u t n o t as co n v eniently as th e d irect stiffness m ethod. Also, a u to m a tin g th e p relim in ary force analyses req u ires solving an additional set o f nd eq u atio n s. A n o th er d isad v an tag e o f th e force m eth o d is th a t th e co m p atib ility eq u atio n s ten d to be ill-co n d itio n ed unless one is careful in selecting force red u n d an ts.
17-9.
VARIATIONAL PRINCIPLES
In C h a p te r 7, we developed v aria tio n a l principles for th e d isp lacem en t an d force fo rm u latio n s for an ideal truss. N ow . in this section, we develop th e co rresp o n d in g v aria tio n a l principles for a m em b er system . T h e ex ten sio n is q u ite straig h tfo rw a rd since the g o v ern in g eq u a tio n s are alm o st identical in form . W e s ta rt w ith th e force-eq u ilib riu m e q u a tio n s,f P = P, + Ar Z T h e p artitio n e d form is P ^ P ^ + A f Z = P ,. 2 + A 'fZ (b)
(C)
(a)
T o in terp ret (a) as a sta tio n a ry req u ire m en t, we co n sid er the d cfo rm atio n disp lacem en t relatio n , i r = A U = /4l L/1 + A 2 U 2 * (d) T h e first differential o f d u e to an in crem en t in U is (1 7 -8 2 )
drT = A AU = A j A U , + A 2 A U 2 T hen, the req u irem en t th a t P r AU - P J AU 4- Z 7' d r
(17-8 3 )
be satisfied for a rb itra ry A U is eq u iv alen t to (a). If we co n sid er U 2 to be p re scribed, (1 7 -8 3 ) results in on ly (b). W e refer to (1 7 -8 3 ) as the p rin cip le o f virtu al displacem ents for a m em b er system . In the d isp lacem en t m eth o d , we s u b stitu te for Z in th e jo in t force-equilibrium eq u atio n s, using Z - k { i r - - r o) = (A U - 1r 0) (e) T h e form o f (1 7 -8 3 ) suggests th a t we define a scalar q u an tity , V = F(U ), having the p ro p e rty d V = Z Td r r - dV(\J) (1 7 -8 4 )
t We work with the governing equations for the restrained system. See (17-33), (17-34), (17-35).
SEC. 17-9.
VARIATIONAL PRINCIPLES
571
O ne can in terp ret V as th e strain energy function for th e m em bers. F o r th e lin ear case, V can be expressed as
v = {{tr - v'SMl ' - -r0 )

= i( A U r o)Tk ( A V - i r o) y ') C o n tin u in g , we define th e p o te n tia l energy function, JJp, as R p *= V + P j U - P r U (17-86)
T he E u ler e q u atio n s for U p are the u n p a rtitio n e d jo in t force-equilibrium e q u atio n s expressed in term s o f U. F inally, we in tro d u ce the jo in t d isplacem ent co n stra in t co n d itio n , U 2 = U 2> by w riting (1 7 -8 6 ) as n p = V + P ,. + P,r 2U 2 - P [ U X - P 2 r (U 2 - 0 2) (1 7 -8 7 )
w here U x, U 2, an d P 2 are variables. T h e E u ler e q u atio n s for (1 7 -8 7 ) are the p a rtitio n e d eq u ilib riu m e q u atio n s (E q u atio n s (b), (c)) expressed in term s o f the displacem en ts w ith U 2 set equal to U 2, i.e., they are th e g o v erning eq u atio n s for the d isp lacem en t fo rm u latio n p resen ted in Sec. 1 7-7. If only the e q u atio n s for P t a re desired, we set U 2 = 0 2 in (17-87), U p = V + P I ,U , - P f U a w here V = H A xU ! + A 2D 2 - i r oy k (A 1U I + A2U 2 - tT ,) (1 7 -8 9 ) (1 7 -8 8 )
T h e E u ler e q u atio n for (1 7 -8 8 ) is (1 7 -5 9 ), an d the seco n d differential h as the form d 1 FL = A U [(A 'fk A j)A U j - A l l [ K ,, A u ' ( , 7 - 90,
Since K u is positive definite, we can state th a t th e d isplacem ents defining th e equilib riu m p o sitio n co rresp o n d to a m in im u m value o f ITp defined by (1 7 -8 8 ) o r (17-87). W e co n sid er next the fo rce-m eth o d fo rm u latio n . W e let A P, A Z be a statically perm issible v irtual-force system . By definition, A P - A r AZ = B A Z (1 7 -9 1 )
P rem u ltip ly in g b o th sides o f (d) w ith A Z r a n d in tro d u cin g (1 7 -9 1 ) leads to the prin cip le o f v irtual forces, AjPt U = & Z T1r (1 7 -9 2 )
N o te th a t (1 7 -9 2 ) is valid only for a statically perm issible virtual-force system , i.e., o ne w hich satisfies (17-91). T he co m p atib ility e q u atio n s follow directly from the principle o f v irtu al forces by req u irin g th e v irtual-force system to be self-equilibrating. If AZ satisfies A P! = B i A Z = 0 (1 7 -9 3 ) then (17-92) reduces to A P jU 2 - A Z ttT (17-94)
572
CHAP. 17
This result is valid for an arbitrary self-equilibrating virtual-force system. The form ulation presented in the previous section corresponds to taking AZ =
ap2
AZ,
(17-95)
= p 2. r a z r
We define the m em ber com plem entary energy function, V* = F*(Z), such that (17-96) dV* = A ZT'//'(Z) F o r the linear case,
ir = r 0 + fZ
and
(0
V* = iz rfz
7 /f
(17-97)
We also define the total com plem entary energy function, n c, as n c f * p i o 2 = r y z , p 2) (17-98)
The deform ation com patibility equations, (17-94), can be interpreted as the stationary requirem ent for n c subject to the following constraints on Z, P 2: B tZ -- P t P/ i P , = P , , + B 7Z The constraint conditions are the jo in t force-equilibrium equations. O perating on (g), and noting th at P l5 P 7. P, 2 are prescribed, lead to the constraint conditions on the force variations B, AZ = 0 A P 2 = B 2 AZ
(h) .
N ote that (h) require the virtual-force sy&fem to be statically permissible and self-equilibrating. In the previous section, we expressed Z, P 2 as
__
o| 0
7, R
Z r
_ < 1R
(i)
P 2 = P2,y + ^ 2 , h Z k This representation satisfies (g) and (h) identically for arbitrary A ZR. Sub stituting for Z, P 2 in (17 - 98) and expanding V* using (17-97), we obtain
nc = z l [ z l R
j I ,J
'\ t
ZR J/
(17-99)
- Z K 8C 2 + const The Euler equations for (17-99) are (1 7-79), and the second differential has the form d 2U c = A Z l f ZR AZ* (17-100)
SEC. 17-10.
MEMBER DEFORMATION CONSTRAINTS
573
Since f7n is positive definite, it follows that the true forces, i.e., the forces th a t satisfy com patibility as well as equilibrium , correspond to a m inim um value of n c. Instead of developing separate principles for the displacem ents and force redundants, we could have started with a general variational principle whose Euler equations are the com plete set of governing equations. O ne can easily show th a t the statio n ary requirem ent for n R = z r( B [ u , + b J u 2) -
v* - P f u ^
+ p;;1u 1 + p ; \ u 2 - p 2 r(u2 - u 2)
(17-101)
considering Z , U ,, U 2, an d P 2 as variables, lead to the p artitio n ed jo in t force-equilibrium equations and the m em ber force-joint displacem ent relations. This principle is a specialized form of Rcissners principle. We o b tain (17-87) from (17-101) by introducing the force-displacem ent relations as a constraint condition on Z , z = M Blu.! + B2 ru 2 - k(:r and noting that, by definition, Z ^ B f U j -I- B2 r U 2) V* = V (k) r 0) (j)
r 0)
Introducing the jo in t force-equilibrium equations as co n strain t conditions reduces I I K to ITC as delined by (17-98).
17-10.
INTRODUCTION OF MEMBER DEFORMATION CONSTRAINTS
Suppose a m em ber is assum ed to be either com pletely o r partially restrained with respect to deform ation due to force. T he rigidity assum ption is introduced by setting the corresponding deform ation param eters equal to zero in the local flexibility m atrix, g. F o r exam ple, if axial extension is to be neglected, we set \ / A E = 0. F o r com plete rigidity, we set g = 0,. N ow , in w hat follows, we discuss the case w here neglecting m em ber deform ation param eters causes the m em ber flexibility m atrix f t o be singular. This happens, for example, when axial extension is neglected for a straight m em ber. T he ran k is decreased! by 1 and the axial force-deform ation relation degenerates to
(a)
where y" x is the initial axial deform ation due to tem p eratu re and fabrication error. N o te that now the axial force has to be determ ined from the equilibrium equations. F o r com plete rigidity, fr 0. and the force-displacem ent relations (see (1 7 -5 )) degenerate to
(b)
f See (1 6 -7 5 ).
574
CHAP. 17
O n e can in te rp re t (a), (b) as either m em b er d efo rm a tio n co n strain ts o r as co n stra in t co n d itio n s o n th e jo in t displacem ents. In general, the decrease in ra n k o f the system flexibility m atrix f is equal to the n u m b e r of c o n strain t conditions. W e con sid er first th e force m ethod. T h e gov ern in g e q u atio n s a re given by
f/ .Z r = A
i q R e q s .)
(c)
w here
Zp, R
Zp, R
_
f*r
bT j
(d)
Suppose these are c d efo rm atio n con strain ts. T h en , f is o f ran k tjT c. In o rder to solve (c), fZ) m u st be non sin g u lar, i.e., it m u st be o f ra n k qR. T his requires qT c qR (17 102) T h a t is, th ere m u st be a t least qR u n co n strain ed m em b er defo rm atio n s. T his co n d itio n is necessary b u t n o t sufficient as we will illu strate below. A side from insuring th a t the flexibility m atrix is o f ra n k qR, th ere is n o difficulty involved in in tro d u cin g m em b er d efo rm a tio n co n strain ts in th e force form u latio n .
Example 17-1 - ---------------------------------Consider the ideal truss shown. For this system, q-r = 4
<Ir = 2
We take the forces in bars 3, 4 as the redundants:

Ft
*'2
ZR =
F-
Then,
Zp. R 1 O' 0 1
and
1 o' T 0 1 1 0 _0 1_
7i + h
7;
ji f3 u .
! o
"1 o ' 0 l 1 0 _ S> 1
T f2 + /4_
We can specify that, at most, two bars are rigid. No difficulty is encountered if only one bar is rigid. However, we cannot specify that bars 1, 3 or 2, 4 are rigid.
W e consid er next th e d isp lacem en t fo rm u latio n . Since f is singular, k does n o t exist and, therefore, we can n o t invert the com plete set o f force-displacem ent
SEC. 17-10.
575
Fig. E17-1
relations, i.e., (17-57) are n o t applicable. In w hat follows, we first develop the ap p ro p ria te eq u atio n s by m an ip u latin g th e original set o f governing equations. W e then show how th e e q u atio n s can be deduced from th e v ariational principle for displacem ents. The governing eq u atio n s are p i = P /, i + A fZ (ntl eqs.) V = / + f Z = A ,U ( + A2U 2
(a)
(qT cqs.) (b)
N ow , we suppose there are c d eform ation co n strain ts and the elem ents of are listed such th a t th e last c elem ents are the prescribed deform ations. W e p artitio n i r an d Z as follows: (17-103) w here '/ contains the co n strain ed m em ber deform ations an d Z c the co rre sponding m em ber forces. W e use subscripts c, u to indicate quantities asso ciated with the con strain ed an d u n co n strain ed deform ations. C o n tinuing, we p artitio n A t, A 2, '/ , an d f consistent w ith (17-103): A,
(i t x
A, 1 rvll/
A Ic
( 9 T - c ) x "d
(c x I'd)
A2
(i n
A , ii ^*2 A?c
(<?r - 0 y r
x r)
,c*r)
(17-104)
(q-r xqT)
576
CHAP. 17
T h e defo rm atio n co n strain ts are in tro d u ced by settin g fr = 0. N o te that, in order for f to be singular, there m u st be no coupling betw een ir u an d Z c, i.e., f m ust have th e form show n above. U sing this n o ta tio n , the governing eq u atio n s take the form P i = P /. 1 + ArZ 4- A [C ZC T ' = c,o = A lcU i + A 2cU 2 _ (17-105) (17-106) (17-107)
+ fuZ = A 1U U1 _ + A 2((U 2
E q u atio n (17-107) represents c c o n strain t co n d itio n s on the u n k n o w n joint displacem ents, T h e ran k o f A lc is equal to the n u m b er o f in d ep en d en t c o n strain t equations. O n e can easily d em o n strate th a t c in d ep en d en t co n strain t conditio n s are req u ired in o rd er to be able to analyze th e system for an arb itra ry loading.
Example 17-2 -
....- ...... ' - ----- ------------ -------
Suppose we specify that bars 1, 3 arc rigid for the system considered in Example 17-1. The constraint equations arc (wc take i " c = { e u <?,}}
f l = e l,o * nU " 21
~ p3.n =; ~ U\X " I"

For (a) to be consistent, we must have
*\.o + 3. =
- 21 +
4i
(b)
Even if (b) is satisfied, we cannot find the forces in bars 1, 3 due to p n .
In w hat follows, we assum e A ic is of ra n k c. W e solve (17-106) for Z and su b stitu te in (17-105). T his is perm issible since fMis nonsingular. T he resulting equations are = k "Vu,n) = M j ./U i + k(A2u0 2 and (A L kuA lu)U i + A [C Z C = P , - P ; 1 - A r j a A 2llU 2 = - T c , 0 - A 2cU 2 r uJ (17-109) (17-110) (17-108) : r ii ( ,)
N ow , the coefficient m atrix, A[JkAlH, is nonsingular only w hen the stru ctu re o b tain ed by deleting the restraint forces (Z c) is stable. By suitably redefining Z c, we can tran sfo rm (17-109) such th at the coefficient m atrix is always n o n singular. Suppose we w rite
z c = z ; + K (;r c - i r c g)
= Z ' + k;(A lcu t + a 2cu 2 r cj (17- 111)
where Z ' represents the new force variable and k ' is an arbitrary symmetrical
SEC. 17-10.
577
positive definite m atrix of o rd er c. Substituting for Z c in (17-109), we o b tain P i = Pz, i + A [C Z'

+ [A [ i A j j
v.< i
r
A i^
k:
A ir
fui\ I A2c U M
j _ A 2ll
(a)
J ;
By defining
k'
_
f i" k ]
1 1
-1
(1 7 - 1 1 2 )
and noting (17-104), we can w rite (a) as P i = P /. t + A [k'(A 20 2 - i \ ) + (A fk 'A jJU , + A feZ ; U sing the n o ta tio n in tro d u ctio n in Sec. 17 -7 (see (1 7 -6 0 )), we let K - A,r k'A , Z/ = . (17-113) (b)
Finally, the governing equations take the form
K n U i + A J X = P i - K i - K I2U 2 = H , A uU , = - A 2c U2 = H 2
Since A , m ust be o f ran k nd for stability and we have required k' to be positive definite, it follows th a t K n is positive definite. Also, the solution for U 3 m ust satisfy (17-116) and we see from (17-111) th a t Z ' is equal to Z f, the actual constraint forcc m atrix, for arbitrary k'. , The expressions for Z an d P1 ( w ith Z c deleted, have the sam e form as the unconstrained expressions (17-57) and (17-59). N ow, it is n o t necessary to rearrange Z such th a t th e co n strain t forces are last. O ne can w ork w ith the natural m em ber force listing, z = { Z . . . , Z IM } (a)
and take arbitrary values for the m em ber deform ation param eters that are to be negelected. W e o b tain K u and H , by first generating <#"*, using the direct stiffness m eth o d and then deleting th e row s and colum ns corresponding to the prescribed displacem ents. T he constrained deform ations, 'f '~ c, can be listed arbitrarily. It is only necessary to specify the locations o f the co nstraint forces (elements of Z'.) in. th e n atu ral m em ber force listing. O nce the displace m ents an d co n strain t forces are know n, we can determ ine the force m atrix for m em ber n by first evaluating (see (1 7 -8 ) and (1 7 -1 1 ))
7 k HiHr V - n = *V. ' r, n ' '" ro, nf)

y \ . n - (E
1 lc
K i = P / ,. + A [z;
+k, ( r ' '
( i 7" u4>
(17-115) (17-116)
(1 7 -1 1 7 )
578
CHAP. 17
w here k' is th e m odified stiffness m atrix, an d then adding the co n strain t forces in the ap p ro p ria te locations. In w hat follows, we describe tw o procedures for solving (17-115) an d (17-116). In the first m ethod, we solve (17-115) for U l5 u t = K T /H i - K f M J X and then substitute in (17-116), ( A ic K f /A f jZ ' Ak Ku Hi H2 (17-119) (17-118)
The coefficient m atrix for Z ' is positive definite since K n is positive definite and A lc is o f ra n k c.. N o te that, with this procedure, wc m ust invert an ndth order m atrix an d also solve a set of c equations. F o r the u n constrained case, we have to solve only nd equations.
Example 1 7 - 3 ------------------------------------------------------------We suppose bar 2 (Fig. E17- 3) is rigid. The constraint equation is
e-, u-y e i.
(a)
To simplify the example, we consider only the eflcct of joint forces. Using the notation introduced above, the various matrices for this example arc
U, {W |, u2}
P i = {P. - Pi } r u = < ? i r c = qT = 2
Z = F l
K =
k,
(b)
nd = 2
K - k'2 c - I
li
(P /,i, o 2, ~tr 0 are null matrices.)
Si
U2 ,PZ
Fig. E17-3
Bar
We start by assembling A (,
i
- A .U ,
SEC. 17-10.
579
and then partition according to (17-104);
' 1
1
1]
(d)
A ,f = [0
Note that we cannot invert (17-109), since A[kAlH is singular. Now, we assume an arbitrary value for the stiffness of bar 2,
k'2 = a k j
(e)
where a is an arbitrary positive constant, and assemble K u : k' =

k
"
j j __ ! k' 1 Kf-
fcj
'I 0 0 a
(f)
Kn = Afk'A, = * * AJ__
.iti + 2a
The governing equations (17-114), (17-115), and (17 -116) reduce to
(g)
|F,1 f0 v/2 7 2 K T ' I ? ? + * 0 " ~a K u U ; + A [cF'2 = P,

A IC U, = 0 The solution follows from (17-118), (17-119). We just have to take
_L
i_
(h )
(i) (j)
h , - P,
The inverse of K, j is
h2= 0
1
ak,
z; =
(k)
K l =
Then,
1 + 2a
-
I
+1
(1 )
I
AuK ri ~ a k , [ 1
A uK u'A u ** f l i t j
and (17-119) reduces to
+ l]
(m )
F i = P i - P i
Substituting for f ' 2 in (17-118), we obtain
(n)
2pj w ?= 0
F in a lly , w e s u b s titu te fo r F 2, Kj, u 2 in (h ):
(o)
f i -=
f 2 = F2 = p2 ~ Pi
(P)
580
CHAP. 17
Instead of first solving (17-115) for U j in term s of Z'n one can start with (17-116), A lcU j = ^ c. 0 - A 2t ) 2 = H 2 (a) which represents c relations betw een the nd displacem ents. Since A lc is o f ran k c, we can express c displacem ents in term s o f nd c displacem ents, i.e., there are only nd c independent displacem ents. W e suppose the first c colum ns of A lc are linearly independent. Since A lc is o f ran k c, we can always perm ute the colum ns such th a t this requirem ent is satisfied. W e let n = nA ~ c (17-120) an d p artitio n A lc, U j :
}TJJ (cxi)
<xl)
(cxk)
Ui
(cxrtd)
[U
(cxc)
(17-121)
Alc - [A lc> |
j A lc 2 ]
The elem ents of U are the independent displacem ents. By definition, A iC i t is nonsingular. Then, solving (a) for U c, the constrained displacem ents, we have u c = Al 11H 2 - A reS A lf. 2U Finally, we express U x as IJ, - BU + H 3 where
(C X )
(17-122) (17-123)
(ii xn) (17-124)

H
0
A lcB =
J Ml)
(c ' 1 1)
(fXXl
N o te th a t B is of rank n and 0 (17-125)
T he generation o f B, H 3 from A lr, H 2 can be com pletely au to m ated using the sam e procedure as em ployed in the force m ethod to select the prim ary structure. W e consider next the jo in t force-equilibrium equations, (17-115), K u U i + A [C Z; = H i S ubstituting for U j leads to (K U B)U + A fcZ ' = H , - K u H 3 = H , (b) K , eqs.) (a)
W e elim inate Z ' from (b) by prem ultiplying by B T an d no tin g (17-125). The resulting system o f n equations for U is (B7K n B)U = B r H 4 (n eqs.) (17-126)
Since B is of rank n, the coefficient matrix is positive definite. One can interpret
SEC. 17-10.
581
B r K n B as the reduced system stiffness m atrix. W e solve for U and then evaluate from (17-123). It rem ains to determ ine the restrain t forces, 7JC . W e consider again Eq. (a). A ssum ing U , is know n, we can write A?'cz ; - Hy - K u U j = H 5 K eqs.) (17-127)
The m atrix, H 5, is the difference between the external applied force, P j, and the jo in t force due to m em ber force with the co n strain t forces deleted, i.e., Hs where (see (17-114)) Z - k'(TT = k '{A 1 V
1
P , - P /tl - A [Z i r 0) + A 2U 2 - n )
(c)
(d)
We determ ine Z using th e m em ber force-displacem ent relations and assem ble P/, i + A [Z by the direct stiffness m ethod. N ow A fr has c independent rows. In determ ining B, we p erm uted the colum ns o f A le such th at the first c colum ns are linearly independent. W e apply the sam e perm u tatio n to (17-127) and p artitio n after row c:
\ T 'Me
H
A[ 1 -AiC )2'H s .,
H 5> 2
(17-128)
C onsidering the first c equations, we have

A [c . I H 5, j
(.17-129)
Since A !c t is nonsingular, we can solve (17-129) for Z'.. W e o b tain the final m em ber forces by ad d in g the elem ents o f Z^ in the a p p ro p ria te locations of Z defined by (17-114) and (d). In this appro ach , we have to invert a m atrix of o rd er c an d solve a system of nt! c equations. A lthough the final nu m b er of equations is less th a n in the first approach, there is m ore prelim inary co m p u tatio n (generation of B) and the procedure can n o t be au to m ated as easily.
Example 17-4 -----------------------------------------------------------------------------------For this example (Fig. El 7-4), rtr f= 5 The constraint conditions are
U] 2 U45 U u 32 Ii4 i t-d. 0 * 1.0 2.
c 4
n = l
(a)
~i~r = 'N
< ? 3 e*.
>
<
l(2 , ~
W 22 ~
= r'
(b)
U31 -
582
CHAP. 17
Note that (b) corresponds to (17-107). The form corresponding to (17-116) is
+1 -1 +1 +1 +1
A lc
U n
*1.0
e 2 ,o u 2l > = < U 2 2 A .o U l l
Ul 2
(c)
T'
u ,
H2
Columns 2,4, 5, and either 1 or 3 comprise a linearly independent sel. Then, we can take either un or k21 as U. It. is convenient to take U = u . We permute the columns according to 1 -* 5
2 - t 3 -2 4 -> 3 5 -> 4
The rearranged form of U x is
= {u 12' u21> 1Y 22U31 !
(d)
Ul 1 }
= {u c | U}
(e)
We determine U cby applying (17-122). This step is simple for this example since A rf, j I. Finally, we assemble U 4 defined by (e) and then permute the rows to obtain the initial
Fig. E17-4
Bars 1, 2, 3, 4 are rigid listing of U t. The final result is

U\ 1 ull V 21 > = u 22
. 31
41
0
+ 1 {M n} +
Ll, 0
e 2 ,a
42 (0
0 0
t B
e 3 ,o + l l S 2
4, a + u4l
T
h
3
SEC. 17-10.
583
The constraint forces are determined from (17-127), which for this example has the form
-1 +1 +1 +1 + 1J
r
F' i r 2 F\ 1 S
h 5i2 ^ 5 .4 .n .
(g)
T
Z'
t
H5
A[(
We permute the ro w s of (g) according to (d) and consider only the first four equations. The resulting equations correspond to (17-129).
It is of interest to derive the equations for the constrained case by suitably specializing the variational principle for displacem ents. We start w ith the unconstrained form of TIp developed in Sec. 17-9, n , = V + P ^ U , - PJU, where v =
t t A tUj
(a)
- r ' 0)Tk(:r
+ a 20 2
- ~ r 0) (b)
N ow, the displacem ents are constrained by r c = A lcu l + a 2 cv Then, V reduces to v = - v ' , , 0)r ku( n ^ u A iuU t + A 2uU2 (d)
2
= 7 - ~ ,0
(c)
We obtain the ap p ro p riate form of Ylp by substituting for V using (d) and introducing the constraint condition, i / \ 1 "c ~ 0: I I , ~ V + P l ,U , - P [ U X + Z ( 17- 130)
The elements o f Z c are Lagrange multipliers. O ne can easily show th at the stationary requirem ent for (17 130) considering U t and Z c as independent variables leads to (17-109) and (17-110). Since i r c Y xC j , we can add the term \(-rc to (d). T aking v = H 'f 1
/- , )
(e) (17-131)
/'0)Tk'(i-" - ~ r 0)
in (17-130) leads to (17-115) and (17- 116). In the second approach, we substitute
U j BU + H 3
(f)
584
GENERAL FORMULATION-LINEAR SYSTEM
CHAP. 17
in (a) and (17-131): n p = V + ( P 1 1 - P[)(B U + H 3) v = i c r - * '0)Tk'(* ' - n ) t T - A ,B U + A ,H 3 + A2tJ 2 The variation of n p considering U as the independent variable is d l l p - AUr [B r (P fi j - P J + ( B W k 'A ^ U + B ^ f k ^ A i H s + A 2U 2 - r j ] = A U r [(B r K u B)U - B r H 4] (g) (17-132)
Requiring n p to be stationary for arbitrary AU results in (17-126). N ote that we could have used the reduced form for V, i.e., equation (d). Also, we still have to determ ine the constraint forces.
REFERENCES
1. F enves, S. J., and F. H. B ra n in , Jr., Network-Topological Formulation o f Struc tural Analysis, ./. S tr u c tu ra l D iv ., A.S.C.E., Vol. 89, No. ST4. August, 1963, pp.
483-514. 2. 3.
D i M aggjo , F. L., and W. R. S pillars , Network Analysis of Structures. J. Eng.
M e c h . D iv .h A .S .C .E ., Vol 91. No. EM3, June, 1965, pp. 169- 188.
A rgyris , J. H., The Matrix Analysis o f Structures with Cut-Outs and Modifica
tions, P ro c. N in th In tern ation aI C o n g re ss A p p l. M e c h ., Vol. 6, 1957, pp. 131-142.
18
18-1. INTRODUCTION
In this chapter, we extend the displacem ent form ulation to include geom etric nonlinearity. The derivation is restricted to small rotation, i.e., where squares of rotations are negligible with respect to unity. We also consider the m aterial to be linearly elastic and the m em ber to be prismatic. The first phase involves developing ap p ro p riate m em ber force-displacem ent relations by integrating the governing equations derived in Sec. 13-9. We treat first p lan ar deform ation, since the equations for this case are easily integrated and it reveals the essential nonlinear effects. T he three-dim ensional problem is m ore form idable and one has to introduce num erous approxim ations in order to generate an explicit solution. We will briefly sketch out the solution strategy and then present a linearized solution applicable for doubly sym m etric cross sections. . The direct stiffness m ethod is em ployed to assemble the system equations. This phase is essentially the same as for the linear case. However, the governing equations are now nonlinear. Next, we described two iterative procedures for solving a set o f nonlinear algebraic equations, successive substitution and N ew ton-R aphson iteration. These m ethods are applied to the system equations and the ap p ro p riate re currence relations are developed. Finally, we utilize the classical stability criterion to investigate the stability of an equilibrium position.
18-2.
MEMBER EQUATIONS PLANAR DEFORMATION
F igure 18-1 shows the initial and deform ed positions o f the member. The centroidal axis initially coincides with the direction and X 2 is an axis of sym m etry for the cross section. We w ork with displacem ents (uu u2, (0 3 ),
585
584
CHAP. 17
in (a) an d (17-131): n , = V + (P,T x - P [)(B U + H :1 )
v = \{;r - ^ 0)rk'(;r - r 0 )
t T
(17-132)
= A iB U + A t H 3 + A 2U 2
T he variatio n o f n p considering U as the in d ep en d en t variable is d U p = A U r [B T( P j . ! - P j) + (BTA [k 'A 1B)U + B ^ f k 't A i H a + A 2U 2 = A U r [(B 7K u B)U - B r H 4] R equiring TIp to be statio n ary for a rb itra ry A U results in (17-126). N o te th at we could have used the reduced form for V. i.e.. eq u atio n (d). Also, we still have to determ ine the co n strain t forces. r -j] (g)
REFERENCES
1.
F enves , S. J., and F. H. Br a n in , J r ., Nctwork-Topological Formulation o f Struc tural Analysis, J. S tr u c tu r a l D iv ., A.S.C.E., Vol. 89, No. ST4, August, 1963, pp. 483-514. D i M a g g i O, F, L., and W. R. S pilla r s . Network Analysis o f Structures, J . E ng. M e c h . D iv ., A . S . C . E . , Vol 91, No. EM3, June, 1965, pp. 169-188. A r g y r is , J. H., The Matrix Analysis o f Structures with Cut-Outs and Modifica tions, P r o c . N in th I n te r n a tio n a l C o n g re ss A p p l. M e c h ., Vol. b, 1957, pp. 131- 142.
2. 3.
18

18-1. INTRODUCTION
In this chapter, we extend the displacem ent form ulation to include geom etric nonlinearity. T he d eriv atio n is restricted to small rotation, i.e., w here squares of ro tatio n s are negligible with respect to unity. We also consider the m aterial to be linearly elastic and the m em ber to be prismatic. T he first phase involves developing a p p ro p ria te m em ber force-displacem ent relations by integrating the governing equations derived in Sec. 1 3 -9 . W e trea t first p la n ar deform ation, since the eq u atio n s for this case are easily integ rated and it reveals the essential nonlinear effects. T he three-dim ensional p ro b lem is m ore form idable and one has to in tro d u ce num erous ap p ro x im atio n s in o rd er to generate an explicit solution. W e will briefly sketch o u t th e solution strategy and then present a linearized so lu tio n applicable for doubly sym m etric crosssections. T he direct stiffness m ethod is em ployed to assem ble th e system equations. T his phase is essentially the sam e as for th e linear case. H ow ever, the governing equatio n s are now nonlinear. N ext, we described tw o iterative procedures for solving a set o f n o n lin ear algebraic equations, successive su b stitu tio n an d N ew ton-R aphson iteration. These m ethods are applied to the system eq u atio n s an d the ap p ro p ria te re::-.rrence relations are developed. Finally, we utilize the classical stability criterion to investigate the stability of an equilibrium position.
18-2.
F igure 18 -1 show s the initial an d deform ed positions o f the m em ber. T he centro id al axis initially coincides w ith the X x direction and X 2 is an axis of sym m etry for the cross section. W e w ork with displacem ents (uls u 2, a>3),
585
586
ANALYSIS OF GEOMETRICALLY NONLINEAR SYSTEMS
CHAP. 18
distributed external force (b2), and end forces (Fu f 2- M 3) referred to the initial (X l ~X 2 - X 3) member frame. The rotation of the chord is denoted by p 3 and is related to the end displacements by ------L -----UB2 ~ UA2
(18
The governing equations follow from (13-88). F or convenience, we drop the subscript on x u and M 3, co3, J3. Also, we consider bx = m3 = 0.
Xi,u 2
Deformed position
-p
Initial position
X\,U\
Centroidal axis
F ig . 1 8 - 1 .
N o ta tio n fo r p la n a r b e n d in g .
Equilibrium Equations F i.x - 0 ax (FlU2 ,x + ^ 2) + ^2 = 0 Fr = - M . , (a)
Force-Displacement Relations
SEC. 18-2.
587
Boundary Conditions F o r x 0: l u2 For x

=
Wll
UA2 &A 3
or or or or or or
\P 2 + +
\ f 2
N o
r-4 O
=
- F - M 42 ^ 3
(c)
II
OJ =
|M j 0
L: ut &B1 u2 = % 2
CO = &>B1
\F i \l =
F 1U 2 , x \ l W VB2
(d)
= + M b3
Integrating (a) leads to

= =
P b 2 dx C 2P x (e)
+ J X( J X b 2
2 +
P w2 , a =
PC2
3 P 2 =
- c 3P
dx)d
where C2, C 3 are integration constants. W e include the factor P so th at the dim ensions are consistent. The axial displacem ent u x is determ ined from the first equation in (a), PL 1 CL (18-2) (i<2..,)2 dx > Com bining the rem aining two equations in (a), we obtain ( P \ M = E I [ 1 + - r j lh . x, + GA El ' (0
Finally, the governing equation for u2 follows from the third equation in (e), 2 ,
XX
^2 j' p j *= M2(C2x + C 3) + b2
\ ^2 dx ) Jx \v -x / (18-3)
- (
where ( P El ( 1 + T \ ga2 The solutions for u 2 and M are u 2 = C4 cos t-ix + C 5 sin jux + C 2x + C 3 + u2b ay + Ci + ( C4 sin f i x + C 5 cos f i x )
(18-4)
*2b, x
where u2fr denotes the particular solution due to b2. If b 2 is constant, b ( El 1/ , 2 U2 b = P5 ( G / l 2 ----- t2 \s x ~ "2 /i2
) !
(18-5)
538
CHAP. 18
E nforcing th e b o u n d a ry co n d itio n s on u2, co a t x = 0, L leads to four linear equatio n s relatin g (C 2 C 5). W hen th e coefficient m atrix is singular, the m em ber is said to have buckled. In w hat follows, we exclude m em ber b u ck ling. W e also neglect transverse sh ear d efo rm atio n since its effect is sm all for a hom ogeneous cross section. W e consider th e case w here th e end displacem ents are prescribed. T h e net displacem ents are
l i nct ~ U' ~ U2 b ) x = 0 , L
(1 8 -6 )
net
^26, x).t* 0. i.
E valuatin g (1 8 -4 ) w ith A 2 = oo, we o b ta in C 2 = (o'A - n C 5

C i UA ~
^4
c4
1 cos jiL sin f.iL
co',} o/A fi sin fiL 1 cos fiL

{( o 'b CO'a ) }
(1 8 -7 )
C s = = ~D
(u'g u'A a ^ L )sin jiL
J 27r. T his defines the u pper lim it on P, i.e., th e m em ber 4 n 2FJ = ~ L
D = 2(1 cos jiL) fxL sin fiL N o te th a t D -* 0 as fxL bucklin g lo a d :
(1 8 -8 )
T he end forces can be o b tain ed w ith (c-e). W e o m it the algeb raic details since they are obvious an d list th e final form below. = M
a ' 3
4-
Eh
17
0 1 ^ 3 + 02<W3 - ~
02f ^/t3 +
(UB2 - U^2)
~ ^2)
E ls M b 3 M 53 4 ~ L F a 2 = PA 2 + F B 2 = Fb 2 ~
0 1 W B3 ~ - (
03 E l 3 L2 03^3 L2
P
W iU + W /13 ( UX?2 W /12)
~ '
+
[Ub2
UA 2 )
(1 8 -9 )
2
W B3 + ~ T~ ( UB2 ~ UA 2 )
( U B2 ~
11A i )
Bl
P ax = ~ P CL 1 - { u 2. x)2 d x Uai ~ A E ~ PL
P = F Bl
PL t - - e,.L AE
where
D = 2(1 - cos /xL) - jj.L sin jjL D <f>i l-iL{sin (xL f.iL cos /tL) D(p2 = liL {u L sin fj.L) D(j>z = m , + 0 2) = M 2(l - cos fxL)
SEC. 18-2.
589
T he < pi functions w ere in tro d u ced by Livesley (Ref. 7), an d are p lo tted in Fig. 18-2. They degenerate rapidly as f i L -> 2 k . T he initial end forces depend on the transverse loading, b 2. If b 2 is constant,
f i
1 A2
r B n2 t
pi
_~
BS
b l 2 v2 ^ 4(01 02))
(18-1 0 )
= - M *3 In o rd er to evaluate th e stiffness coefficients, P has to be know n. I f one end, say B, is unrestrained w ith respect to axial displacem ent, there is no difficulty since F B 1 is now prescribed. T he relative displacem ent is determ ined from P F Bl
UB \ = 41 + T F
P L
AE
Le r
'L er 2 L iu 2 ,x ) 2 dx er{fiL, uA2, uB2 , ida, coB) 0

'\2
L + 0 7(<^B3 ~ W43)2 + D(j) 4 11 L sin jxL C s 0 3

U B2
0 6 ( < W -
OJA 3 )
C'5
UA2
(18-11)
(^ L )2
_____ 1
C0A 3
\f.iL(fiL sin /.iL cos j.iL) + 2(1 cos [iL ) 2 (juL sin /.iL)2 I f 2 ) 0 6 = (1 - cos ^L) j 0 7 0 7= " 1 C O S jlV 0 5=
I ^
1 ff i
sin liL cos uL) 9 ----------------------ttL 1
+ (7 IF j+ r
W e call er th e relative end sh ortening due to rotation. H ow ever, w hen both axial displacem ents are prescribed, we have to resort to iteration in o rd er to evaluate P since er is a n o n lin ear function of P. T he sim plest iterative schem e is
P{<+1) = - u^ + AEer)
(18-12)
an d convergence is rapid w hen fxL is n o t close to In, Expressions for the increm ental end forces due to increm ents in the end displacem ents are needed in the N ew to n -R ap h so n procedure an d also for stability analysis. If //L is n o t close to 2 k . we can assum e the stability functions
590
CHAP. 18
Mi
F ig . 1 8 - 2 .
P lo t o f t h e cj> fu n c tio n s .
are constant and equal to their values at the initial position, when operating on (18-9). The resulting expressions are
(f> 1 + (j>2 A l t > B 3 ------ ( A i / f l 2 A t / ^ 2 )
_ /, < t> : dM R 3 = dM'n-i H L < f)3 E I 3 A C D B3 + A CO .43 -(AuB2 ~ &UA2) dF A 2 = dF A 2 + - ^ b r I
(18-13)
~ dF B2 =
^ 1\
~ A ua2) ~
UB2 ~
UA2 j D
- (dFA2 - dF A2) dFBl = dP
dF A 1 = - d P
/4 dP ~ (Awgj A1 / 4J + .4J5 dey
SEC. 18-3.
MEMBER EQUATIONS ARBITRARY DEFORMATION
591
where the increm ental initial end forces are due to loading, Ab2. We can obtain an estim ate for der by assum ing Au2. x is constant. de,. = i - J u2' XA u 2 x d x -^ 4 ^ (18-14)
The coefficients in (18 13) are tangent stiffnesses. They are not exact since we have assum ed (j> ; an d Au2 x constant. T o obtain the exact coefficients, we have to add EJ 3 T
<f>'i = ^
" j
(1 8 ~ 15)
d(fiL) = ~ ~ ~ ~ ~ dP 2E I3 h to d M /t and sim ilar term s to dM, . . . , dVn. The derivatives o f the stability functions are listed below for reference: <*>, = ^ 2{(iL ) 2 sin fj.L fa fa ---------------- +
= a 2" 4> \ = {1 - < !> i + 2 0.i} +
<18_l6)
- /^ )
02 03. 01 We also have to use the exact expression for < f*> der Ai/^2 +
o u B2
eu a 2
Aug2 + --- Aco^ + ~

cw A d co B
AcoB +
-r A(pL)
<( u L )
(18-17)
in the equation for dP. An im provem ent on (18-14) is obtained by operating on (18-11), and assum ing jiiL is constant.
18-3.
MEMBER EQUATIONSARBITRARY DEFORMATION
The positive sense of the end forces for the three-dim ensional case is shown in Fig. 18-3. N ote th at the forcc and displacem ent m easures are referred to the fixed m em ber frame. The governing equations for small rotations were derived in Sec. 13-9. They are nonlinear, and one m ust resort to an approxim ate m ethod such as the G alerkin schem e,t in which the displacem ent m easures are expressed in term s of prescribed functions (of x) and param eters. The problem is transform ed into a set of nonlinear algebraic equations relating the param eters. Some applications o f this technique are presented in Ref. 5.
t This m ethod is outlined in Sec. 10 G .
592
CHAP. 18
N o te :
T h e c e n tro id a l axis coin cides w ith X \ . X 2 and X 3 are p rin cipal in e rtia d ire c tio n s .
Fig. 18-3.
N o t a t io n f o r t h r e e - d i m e n s i o n a l b e h a v i o r .
If we consider b x = 0, the axial force F y is co n stan t along the m em ber an d the nonlinear term s involve and coupling term s such as F 2 uX i, co1M 2; etc. N eglecting these term s results in linearized equations, called the Kappus equations. T heir form is: Equilibrium Equations Fi = P 1 + *31, 1) + F 2] + b 2 = 0
* 2^ 1, 1) + ^ 3] + b3 = 0
J - [P(Us3, 1 -
M jt 1 +
j + mT +
X
[/> (x3s2. 1 x 2 usi ! f PiOJi, i)] ~ 0 AJ 2 ^ I'w " M *t l - M r +

75
__- _t
^ (18-18) - 0
A^s.i + F 2 + m3 = 0
Pi
*2
X3
SEC. 18-3.
MEMBER EQUATIONS ARBITRARY DEFORMATION
593
Fovce-Displacement Relations
f + col%j -prACrM r T + x 3 rF 2 -I- x 2 rh 3) Boundary Conditions (4- for x L, for x = 0) P = +F

1 > S 2 .
X +
1)
F2 = F X
P{Us3. 1 X 2 (l)\. l) + ^3 = i ^ 3 Mr + Mr T + P{x 3 us2>l - x 2w ,,3>J !- ^ w u ) = M r M

2
= +M 2
M 3 + A?3
= + Mq If one neglects the
To in terp ret th e linearization , wc consider (1 3 -8 1 ). n onlinear term s in the shearing strains,

> '1 2
u i,
2 +
U2 .
(a)
y 13 ~ 1,3 + ^ 3. 1 takes the extensional strain as

*1 M l. I + 2 L |fi2. 1 +
3.lU= = 0
(* 2
J C s W .l
(b)
an d assum es J jx 2( x | + x])dA = 0 + x j)d A - 0 Jf (f>{xj + x j)d A = 0 one obtain s (13-81). E q u atio n s (c) are exact w hen the section is d o u b ly sym m etric. A ssum ptions (a) an d (b) are reasonable if cou x is sm all w.r. to 2 j and u3 > H ow ever, they in tro d u ce considerable e rro r w hen c o t is the d o m in an t term . This has been dem o n strated by Black (Ref. 5). W hen the cross section is d o u b ly sym m etric, (c)
594
CHAP. 18
(r is the radius o f gyration with respect to the centroid) and the problem uncouples to 1. 2. 3. Flexure in X l - X 2 plane Flexure in X r X 3 plane R estrained torsion
2
W e have already determ ined the solution for flexure in the X r X If we introduce a subscript for fi and < f?j, 7
W
$21
plane.
~ P - E li = frliUlL)
2 _
-P EI2 (18-20)
031
rA l ( ^ 3 ^ )
and then replace U2 * u3 U3 - ~ u 2 F3 -F ,

CO2
OJ3
C l)^
U )2
M r -> Af, M i -A f2
(18-21)
in (18-9), (18-13), we obtain the m em ber relations for flexure in the frame. F o r example, El 3 M a 3 = A ft, + ~ L < l>
(p 2 l < -'} A 3 + (t) 2 2 a ) B 3 7 ~ ( % llA z )
(18-22}
m
42 = A fAt, 2 +
E h
L
$ 31w /12 + 4}32^>B2 ^
$^ 3
j' (u3 ~ yo)
and
C O B 2 0)A2 --- (U 3 M .o)
The expressions for the axial end forces expands to r Bl = p

fm
= - p
AE P = ~ j r K t ~ w/ii) + > l (^ i + er2 + ^ 3) (18-23) *rl = 2L Jo *r3 = 2L * '2 2L (2. A -)2
( 3. x ) 1 d x i
where er 3 is obtained from e r2 by applying (18-21). We generate the restrained torsion solution following the procedure described in Exam ple 13-7. If the joints are m om ent resisting (i.e., rigid), it is reasonable to assume no warping, which requires / 0 at x 0, L. The corresponding solution is sum m arized below :
S E C . 1 8 -3 .
M EMBER
E Q U A T IO N S A R B I T R A R Y D E F O R M A T I O N
595
_ GJ
Mffi
r2P
r ~ GJ
~j
u
1
GJ
I + P
ErL, 1 + C ,(l + F)
(0A l)
M Al ~ $ bi 1 + P
f a = -
1 +
1 cosh juL sinh fxL

M m ___ GJ{ 1 + P)
(18-24)
piUl + Q 1 + P)) -x 1 - cosh iuL sinh ixx h-------;------------(1 cosh jxx) sinh /.iL
)I = (Oai + +
1 fi(l + Cr(l + P))
W e ncglect shear deform ation due to restrained torsion by setting C,. 0. If w arping restraint is neglected, / r = co < j > t = 1 + P A t this point, we sum m arize the m em ber force-dispiacem ent relations for a doubly sym m etric cross section. F o r convenience, we introduce m atrix n otatio n : 3F B = {i* yF 2 1' 3 M t M 2M 3}B :ifl B {ul U2M3ft)\W 2 (Oi)p etc. (18-26) (18-25)
+ k BnUB + kB/ U A + & r & A - 3F\ + (kBA)TWB + k A/ U A - 3Pr

where &'r contains nonlinear term s due to chord ro tatio n and end shortening = ÂE{erl + er2 + er3); y (uB 2 - uA2 ) ; ^ ( u B3 - uA3); 0; 0; o | ; c o n t a i n s the initial end forces due to m em ber loads; and
596
CHAP. 18
O p eratin g on & B , 8F A leads to the increm ental equations, i.e., the th ree dim ensional form of (18-13). Assum ing th e stability functions are constant and taking r2 d(j)t a; dP dP GJ
1 r
d e r ~ ~ 2 \ ( UB2
we obtain
d & B = d & B + (kB B + k r)A<W B + (kfl/1 - k r) M / A d& A = + (k ,i ~ JU T + (k^,, + k , ) A ^
^ 1 1 .
2023 jJ , p /2 - 2^ 033 -023
E I3 LT
r
- 0 ( -r
/ v2
i GJ 7 2
^33 V33 ^ jr 2
03 2 - j
,
023 L 2
922 T -
/3
AE T ,
2 0 2 3^ 3
e j3
^23 ~ jT
e i2
2 0 3 3* j3 '
e i2
0 3 3
GJ
0 , y ;
Sym
0 3 1T
0 2 ,
EL l>
(1 8 27)
UA 2 ) { & UB2 ~ & UA 2 ) + ( UB3 ~ l l A i ) ( ^ UB?, ~ A u^)
+ r 2(o>zn
- A oja i )\
(18-28)
S E C . 1 8 -4 .
S O L U T I O N T E C H N I Q U E S ; S T A B I L IT Y A N A L Y S IS
597
w here k r is the increm ental stiffness m atrix due to ro tatio n , 0 P3 2 , Pa k, = AE ~L Sym m etrical AE Pi r2P i
0 0 0 0 0
0 0 0 0 0 0
~~ P 2 P 3 1 LI' B1 p 2 + 4E
2
r 2 P\P*
~ r 2p i P 2
(> '2P i )2
Pi
(UB2
UA 2 )
Pi ~
(W J?3 " ^43)
Pi =
K i l
<Al)
If P is close to the member buckling load, one m ust include ad d itio n al term s due to the v ariatio n in the stability functions and use the exact expression for der. K a p p a ss eq u atio n s have also been solved explicitly for a m onosym m etric section w ith w arping an d sh ear deform ation neglected. Since the eq u atio n s are linear, one can w rite dow n the general so lu tio n for an arb itra ry cross section. It will involve twelve in teg ratio n co n stan ts which are evaluated by enforcing the displacem ent b o u n d ary conditions. T h e algebra is u n tractab le unless one introduces sym m etry restrictions.
18-4.
SOLUTION TECHNIQUES; STABILITY ANALYSIS
In this section, we present th e m athem atical b ack g ro u n d for tw o solution techniques, successive su b stitu tio n and N ew to n -R ap h so n iteratio n , an d then apply them to the governing eq u atio n s for a n o n lin ear m em ber system. C onsider the problem of solving the n o n lin ear eq u atio n i//(x) = 0 Let x represent one o f the roots. By definition, i//(x) = {j} = 0 (18-30) (18-29)
In the m eth o d o f successive su b stitu tio n ,! (18--29) is rew ritten in an equivalent form, x - g(x) (18-31) and successive estim ates of the so lu tio n are determ ined, using x +1 = g(x(k> ) = g(k) (18-32)
where x(i) represents the /cth estimate. The exact solution satisfies .Y = g
j See Ref. 9.
(a)
598
CHAP. 18
T hen, x x {k+1) = g g{k) E x p an d in g g in a T ay lo r series a b o u t x {k\ v =

9
(b)
{k) + g[k &
- * w ) + k i % ( * - x {k)) 2 +
(c)
an d reta in in g only th e first tw o term s lead to th e convergence m easu re X - x < t + I > = ( X - x ,* > )|0 .,|,.fc (1 8 -3 3 )
w here is betw een x {k) an d x. In th e N ew to n -R a p h so n m e th o d ,t ijj(x) is ex p an d ed in a T a y lo r series a b o u t x (k) ijj(Z) - \j/k) + i//^ A x + i-^ * .v(Ax)2 + = 0 (a) w here Ax is th e exact c o rrectio n to x {k\ A x x x (k)
(b)
A n estimate for Ax is o b ta in e d by neglecting second- an d h ig h e r-o rd e r term s: jAk) = _ r_ ,//<';> X{k+D = x (k) + A x (k)
a x (*>
(1 8 -3 4 )
T h e convergence m easu re for this m e th o d can be o b ta in e d by co m b in in g (a) an d (1 8 -3 4 ), an d h as th e form (X = -JP - x ) 2|v > . |,=6, (1 8 -3 5 )
N o te th a t the N ew to n -R a p h so n m e th o d has second -o rd e r convergence w hereas successive su b stitu tio n h as only first-order convergence. W e co n sid er next a set o f n n o n lin e a r e q u atio n s: v|/ = {,/zj \jj 2 ' ' ' fA,,} - 0 'A; = * 2, , -*) (1 8 -3 6 )
A n exact so lu tio n is d en o ted by x. Also, vJ/(X) = vp. In successive su b stitu tio n , (1 8 -3 6 ) is re a rra n g e d to ax = c - g (1 8 -3 7 )
w here a, c are c o n stan t, g = g(x), an d th e recu rren ce rela tio n is ta k en as ax(k+I) = c g(f< ) T h e exact so lu tio n satisfies ax = c - g T hen, (a) (1 8 -3 8 )
a(x - x(ft+1> ) = - ( g - g(k> )

f See Ref. 9.
(b)
SEC. 18-4.
599
E x p an d in g g in a T a y lo r series a b o u t x (k), g = g(k) + g ^ (x x (k>) +

9
(c)
9 1,n 92, n (d )
1, i
9 1,2 92, 2
'''
' ' '
8, x
39j
CXr
92, 1
_9n, I
9n,
9n, n_
a n d reta in in g only the first tw o term s resu lts in the convergence m easure (x - x (k+1)) a |g,xj (x x(fc)) (1 8 -3 9 )
w here t;k lies betw een x k a n d x. F o r convergence, th e n o rm o f sT :g x m u st be less th an unity. T h e generalized N ew to n -R a p h so n m e th o d consists in first ex p an d in g \|/(X) a b o u t x (,< ),
x |,(x ) = xj/(k> +
d \ | / (k)
w here d\j/w = Ax = dxr (1 8 -4 0 )

' ? J j
f?X,. f x s
d2ijj {dîj'j} d 2 ij/j = (Ax)r (Ax)
N eglecting the second differential leads to th e recu rren ce rela tio n c/v|/(k) = v|/^> A x(k) X<k+D =, x (k) + A x(k> T h e c o rresp o n d in g convergence m easu re is xj/^X - x (k+1)) = (1 8 -4 2 ) - \} /(k) (1 8 -4 1 )
Let us now apply these so lu tio n te ch n iq u es to th e s tru c tu ra l p roblem . T h e govern in g e q u atio n s a re the n o d al force-eq u ilib riu m e q u a tio n s referred to th e global system fram e, \|/ = 0>t - sj = 0 (1 8 -4 3 ) w here 0Pe co n tain s th e extern al n o d al forces an d - Pm , is the su m m a tio n o f th e m em b er end forces in cid en t on n o d e L O n e first h as to ro ta te the m em b er end forces, (1 8 -2 6 ), from th e m em b er fram e to the global fram e using = { m on)r& n k = ( & on)Tk ni%on ^
In o u r fo rm u latio n , th e m em ber fram e is fixed, i.e. 3%on is co n stan t. W e in tro d u ce th e d isp lacem en t restrain ts an d w rite th e final e q u atio n s as
< A= Pe - Pm= o Pm~ P( + Pr + K U
(18-44)
600
CHAP. 18
N o te th a t K an d P ; d ep en d o n th e axial forces w hile P,. depends o n b o th th e axial force an d th e m em b er rigid b o d y ch o rd ro ta tio n . If the axial forces are sm all in co m p ariso n to th e m em b er b u ck lin g loads, we can replace K w ith K{ , th e linear stiffness m atrix. A pplying successive su b stitu tio n , we w rite K U = P e - P, - P r an d iterate on U , h o ld in g K co n sta n t d u rin g th e ite r a tio n : k u (,,) = p e - p ; - r
; 1 - 11 '
(a)
(1 8 -4 5 )
W e em ploy (1 8 -4 5 ) to g eth er w ith a n in crem en tal lo ad in g schem e since K is actu ally a variable. T he steps are o u tlin ed here: 1. 2. A pply th e first lo ad increm ent, P c(i), an d solve for U (1), using K = K,. U p d ate K using th e axial forces c o rresp o n d in g to P e{1). T h en apply P e(2) a n d iterate on
(n ) _ P I P _ p(n-l) K( S )IT ^ r e(l) + 1 e (2 ) r r
3.
C o n tin u e for successive load increm ents.
A convenient convergence criterio n is the relative change in th e E uclid ean norm , N , o f th e n o d al displacem ents. N = (U r U )1/2 1
I abs value
^ e ,
(a specified value)
(1 8 -4 6 )
This schem e is p articu la rly efficient w hen th e m em b er axial forces are sm all w ith respect to th e E uler loads since, in this case, we can tak e K = K ( d u rin g th e entire so lu tio n phase. In the N ew to n -R ap h so n p ro ced u re, we o p erate on // acco rd in g to (1 8 -4 1 ): d ^ n) = \|/(n) N ow , P e is p rescrib ed so th a t d\\f(n) = d P $ = - |K t A U V w here Kr d en o tes the tangent stiffness m atrix. T h e iteratio n cycle is K(") A u (n) = P e - P ^ U (,,+ 1) = U (n) + A U (n) (1 8 -4 8 ) d u e to A U (1 8 -4 7 )
(a)
= - | d P , + d P r + K AU + (</K)U|u(.,
W e iterate o n (1 8 -4 8 ) for successive load increm ents. T his schem e is m o re expensive since K, h as to be u p d ated for each cycle. H ow ever, its convergence ra te is m ore rap id th a n d irect su b stitu tio n . If we assum e the stab ility functions
SEC. 18-4.
601
are c o n stan t in form ing dP, d u e to AU, the tan g en t stiffness m atrix reduces to dK ^ 0 dPi 0 (1 8 -4 9 )
K, x K + Kr
w here Kr is gen erated w ith (18-28). We include the increm ental m em b er loads in P e at the s ta rt o f th e iteratio n cycle. R ath er th an up d ate K, a t each cycle, one can hold K ( fixed for a lim ited n u m b er o f cycles. This is called modified N ew to n -R ap h so n . T h e convergence rate is low er th an for reg u lar N ew tonR ap h so n b u t higher th a n successive sub stitu tio n . W e consider next th e q u estio n o f stability. A ccording to the classical stab ility criterio n ,! an equilibrium p o sitio n is classified as: stab le n eu tral unstab le d 2 Wm - d 2 WE > 0 d 2 Wm - d
2
We = 0
(1 8 -5 0 )
d z Wm d 2 We < 0
w here d 2 We is the seco n d -o rd er w o rk d o n e by the external forces d u rin g a displacem ent increm ent AU, and d 2 Wm is the seco n d -o rd er w ork d o n e by the m em ber end forces actin g on the members. W ith o u r n o tatio n , d 2 We = d 2Wm = = A U % AU and th e criteria tran sfo rm to < 0 (AU)r K, AU AU = 0 > 0 stable n eutral u nstable (1 8 -5 2 ) AU (1 8 -5 1 )
T he m o st frequent case is P e prescribed, an d for a c o n stan t loading, th e tan g en t stiffness m atrix m ust be positive definite. T o d etect instability, we keep trac k o f the sign of th e d eterm in an t o f the tan g en t stiffness m atrix d u rin g the iteratio n . T he sign is o b ta in ed a t no cost (i.e., no ad d itio n al co m p u tatio n ) if G au ss elim ination o r the factor m eth o d are used to solve the correction eq u atio n , (18-48). W hen the d eterm in an t changes sign, we have passed th ro u g h a stability transition. A n o th er in d icatio n o f the existence o f a b ifurcation p o in t (K, singular) is the deg en eratio n o f the convergence rate for N ew to n -R ap h so n . T h e co rrectio n tends to diverge an d oscillate in sign and one has to em ploy a higher iterative scheme. F inally, we co n sid er the special case w here th e lo ad in g does n o t p ro d u ce significant ch o rd ro tatio n . A typical exam ple is show n in Fig. 1 8 -4 . B oth the
t See Secs. 7 - 6 and 1 0 -6 .
602
CHAP. 18
fram e and lo ad in g are sym m etrical a n d the displacem ent is due only to s h o rt ening o f th e colum ns. T o investigate the stability o f this stru ctu re, we d e le te ! the ro ta tio n term s in K(. an d w rite K t - K + ZK;. '= K,(/.) (18-53)
w here K' is due to a u nit value o f th e load p aram eter / . T he m em b er axial forces are d eterm ined from a linear analysis. T hen, the b ifu rcatio n problem reduces to determ in in g th e value of X for w hich a non triv ial so lu tio n of (K + /JC )A U = 0 exists. This is a n o n lin ear eigenvalue problem , since
|2 X
K
(18 -5 4 ) =
2\
K (/L ).
/////
/7 7 7 7
Fig. 18-4. Exam ple of structure and loading for which linearized stability an alysis is applicable.
In linearized stability analysis, K is assum ed to be K, an d one solves K, AU = - /K;. AU (1 8 -5 5 )
B oth K ; an d K' are sym m etrical. Also, K, is positive definite. U sually, only the low est critical load is o f interest, and this can be o b ta in ed by ap p lying inverse iteratio n J to ( - K ;) A U = IK ,A U T 1 I REFERENCES
1. 2. 3.
T im o s h e n k o , S. P., and J. M. G ere : T h e o r y o f E la s tic S t a b i l i t y , 2d ed., McGraw-Hill,
(18 -5 6 )
New York, 1961.

K o l l b r u n n e r , C. F., and M. M e is t e r : K n ic k e n , B ie g e d r illk n ic k e n , K ip p e n , 2d ed.,
Springer-Verlag, Berlin, 1961. B l e i c h , F .; B u c k lin g S tr e n g th o f M e t a l S tr u c tu r e s , McGraw-Hill, New York, 1952.
t Set p i = p 2 p z 0 in (18-28). J See Refs. 11 and 12 of Chapter 2.
REFERENCES 4. 5. 6.
7.
603
B u r g e r m e is t e r , G.. and H. S t e u p : S ta b ilitiits th e o r ie , Part 1, Akademie-Veriag.
Berlin, 1957.
C h il v e r , A. H., e d .: T h in -W a lle d S tr u c tu r e s , Chatto & Windus, London, 1967. V l a s o v , V . Z .: T hin W a lle d E la s tic B e a m s, Israel Program for Scientific Transla
8. 9. 10. 11.
tions, Office o f Technical Services, U.S. Dept, o f Commerce, Washington, D.C., 1961. Li v b sle y , R. K . : M a t r i x M e th o d s o f S tr u c tu r a l A n a ly s is , Pergamon Press, London, 1964. A r g y r is , J. H .: R e c e n t A d v a n c e s in M a t r ix M e th o d s o f S tr u c tu r a l A n a ly s is , Pergamon Press, London, 1964. H i l d e b r a n d , F. B .: In tr o d u c tio n to N u m e r ic a l A n a ly s is , McGraw-Hill, New York, 1956. G a l a m b o s , T. V .: S tr u c tu r a l M e m b e r s a n d F r a m e s , Prcntice Hall, 1968. B r u s h . D. and B. A l m r o t h : B u c k lin g o f B a rs, P la t e s , a n d S h e lls , McGraw-Hill, New York, 1975.
Index
A ssociative m ultiplication, 8 A ugm ented branch-node incidence m a trix, 124, 222 A ugm ented m atrix, 33 A xial deform ation, influence on bending o f p la n a r m em ber, 472 B ar stiffness m atrix, 180 B ifurcation; see N eu tral equilibrium Bim om ent, M<f), 373 B ranch-node incidence table, 121, 145 C < j> } C rt C,. coefficients appearing in com plem entary energy expression fo r restrained torsion, 387, 388, 416 C anonical form , 58 C artesian form ulation, principle o f v ir tual forces fo r a p lanar m em ber, 465 C astig lian o s principles, 176 C a y le y -H a m ilto n T heorem , 63 C en ter o f twist, 383, 389 C haracteristic values o f a m atrix, 46 C hord rotatio n , p , 586 C ircu la r helix, definition equation, 84, 86 C ircu la r segm ent out-of-plane loading, 504 restrained w arping solution, 509 Classical stability criterion continuum , 256 m em ber system , 601 truss, 170 C losed ring, out-of-plane loading, 503 C ofactor, 19 C olum n m atrix, 4 C olum n vcctor, 4 C om plem entary energy continuum , 261 m em ber system , 572 p lan a r curved m em ber, 434 restrained torsion, 385; 387, 388 u n restrained torsion-flexure, 301 C onform able m atrices, 8, 35 C onnectivity m atrix, m em ber system , 563 C onnectivity table fo r a truss, 121, 143 C onsistency, o f a set o f linear algebraic equations, 31, 44
"onstraint conditions treated with grange multipliers, 76, 80 "urved member definition of th in and th ic k , 434 thin, 487 slightly twisted, 487
La
Dcfect, of a system of linear algebraic equations, 31 Deformation for out-of-plane loading of a circular member, transverse shear, twist, and bending, 498 for planar member, stretching and transverse shear vs. bending, 454 Deformation constraints force method, 573 displacement method, 576 variational approach, 583 Deformed geometry, vector orientation, 239 Degree of statical indeterminacy member, 555, 567 truss, 210 Determinant, 16, 37, 39 Diagonal matrix. 10 Differential notation for a function, 70, 72, 79 Direction cosine matrix for a bar, 119 Discriminant, 40, 59 Distributive multiplication, 8 Echelon matrix, 29 Effective shear area, cross-sectional prop erties, 302 Elastic behavior, 125, 248 End shortening due to geometrically non linear behavior, 589 Engineering theory of a member, basic assumptions, 330, 485 Equivalence, of matrices, 27 Equivalent rigid body displacements, 334, 414, 430 Euler equations for a function, 73 Eulerian strain, 234
606 First law of thermodynamics, 248 Fixed end forces prismatic member, 523 thin planar circular member, 528 Flexibility matrix arbitrary curved member, 515 circular helix, 534 planar member, 462 prismatic member, 345, 521 thin planar circular member, 526 Flexural warping functions, 296, 300 fn Frenet equations, 91
INDEX Maxwells law of reciprocal deflections. 356 Member, definition, 271 Member buckling, 588 Member force displacement relations, 537, 546, 556 Member on an elastic foundation, 384, 369 Mesh, network, 220 Minor, of a square array, 19 Modal matrix, 52 Modified Neuton-Raphson iteration, 601 Moment, M n , 375 Mushtaris equations, 444 Natural member reference frame, 92 Negative definite, 58 Negligible transverse shear deformation, planar member, 443, 454, 498 Network, topological, 220 Neutral equilibrium, 170, 256, 601 Newton-Raphson iteration, member sys tem, 598 Normalization of a vector, 49 Null matrix, 4 One-dimensional deformation measures, 335, 338, 432 arbitrary member, 491 Orthogonal matrices and trnasformations, 50, 53 Orthotropic material, 250, 251 Permutation matrix, 42, 135 Permutation of a set of integers, 16, 37 Piecewise linear material, 126, 146 Plane curve, 98, 425 Poisson's ratio, 252 Positive definite matrix, 58, 63 Positive semi-definite matrix, 58 Postmultiplication, matrix, 8 Potential energy function, member system, 571 Premultiplication, matrix, 8 Primary structure member system, 568 planar member, 463 prismatic member, 354 truss, 211 Principle minors, 55 Principle of virtual displacements member system, 570 planar member, 442 Principle of virtual forces arbitrary member, 490, 492, 512 member system, 571 planar member, 435, 458 prismatic member, 338, 351 Quadratic forms, 57
Gausss integration by parts formula, 254 Geometric compatibility equation arbitrary member, 499 continuum, 259, 264 member system, 569 planar member, 463, 466 prismatic member, 355 truss, 160, 212, 216, 223 unrestrained torsion, 279, 315 Geometric stiffness matrix for a bar, 200 Geometrically nonlinear restrained torsion solution, 595 Greens strain tensor, 234 Hookean material, 126, 249 Hyperelastic material, 248 Incremental system stiffness matrix member system, 601 truss, 193 Inelastic behavior, 125 Initial stability member system, 562 truss, 137 Invariants of a matrix, 59, 62 Isotropic material, 252 Kappus equations, 592 Kronecker delta notation, 11 Lagrange multipliers, 76, 80, 583 Lagrangian strain, 234 Lame constants, 253 Laplace expansion for a determinant, 20, 38 Linear connected graph, 218 Linear geometry, 120, 143, 237 Linearized stability analysis, 602 Local member reference frame, 92 Marguerre equations, 449, 456 Material compliance matrix, 249 Material rigidity matrix, 249 Matrix iteration, computational method,
201
INDEX Quasi-diagonal matrix, 15, 38 Quasi-triangular matrix, 39 Radius of gyration. 434 Rank of a matrix, 27, 42, 43 Rayleighs quotient, 75, 79 Reissners principle continuum, 270 member, 383, 414 member system, 573 Relative minimum or maximum value o f a function, relative extrema, 66 Restrained torsion solution, prismatic member linear geometry, 391 nonlinear geometry, 595 Restrained torsion stress distribution and cross-sectional parameters channel section, 401 multicell section, 411 symmetrical I section, 398 thin rectangular cell, 407 Rigid body displacement transformation, 109 Rotation transformation matrix, 10 1, 232 Row matrix, 4 Self-equilibrating force systems, 160, 211, 258 . member systems, 568 Shallow member, assumptions, 448 Shear center, 297, 300, 309, 378, 389 Shear flow, 287 Shear flow distribution for unrestrained torsion, 308 Similarity transformation, 53, 62 Simpson's rule, 475 Singular matrix, 22 Skew symmetrical matrix, 11 Small strain, 120, 235 Small-finite rotation approximation, 238 Square matrix, 4 Stability of an equilibrium position, 171, 195 Stability functions (</>), prismatic member, 589 Statically equivalent force system, 103, 106 Statically permissible force system, 159, 216, 257 Stationary values of a function, 67, 79 Stiffness matrix arbitrary curved member, 516, 520
607
modification for partial end restraint, 535 prismatic member, geometrically non linear behavior, 588, 595 prismatic member, linear geometry, 522 Strain and complementary energy for pure torsion, 280 Strain energy density, 248 Stress and strain component trnasformations. 249 Stress components Eulerian, 242 Kirchhoff, 246 Stress function, torsion, 276 Stress resultants and stress couples, 272 Stress vector, 240 Stress vector transformation, 242 Submalrices (matrix partitioning), 12, 36 Successive substitution, iterative method member system, 597 truss, 193 Summary of system equations, force equi librium and force displacement, 561 Symmetrical matrix, 11, 35 System stiffness matrix member system, 548, 550, 565 truss, 179, 180, 188, 206 Tangent stiffness matrix for a bar, 193 prismatic member, 590, 596 Tensor invariants, 232 Torsion solution, rectangle, 281 Torsional constant, J, 276, 278, 323 Torsional warping function, 274, 377 Transverse orthotropic material, 252 Transverse shear deformation planar member, 454, 498 prismatic member, 355 Trapezoidal rule, 474 Tree, network, 220 Triangular matrix, 12 Two-hinged arch solutions, 467, 470 Unit matrix, 10 Variable warping parameter, /, for re strained torsion. 372 Vector, definition (mechanics), 4 fn Work done by a force, definition, 153, 156

Connor Structural Analisys

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Connor Structural Analisys

Transféré par

Droits d'auteur :

Formats disponibles

JER O M E J. C O NNO R, Sc.D ., M assachusetts In stitu te of Technology, is P rofessor of Civil E ngineering a t M assa chusetts In stitu te of Technology.

ANALYSIS OF STRUCTURAL MEMBER SYSTEMS

M assach u setts Institute of T echn ology

THE RONALD PRESS C O M P A N Y NEW YORK

Library of Congress Catalog Card Number: 74-22535

Cam bridge, M ass. January, 1976 .

I MATHEMATICAL PRELIMINARIES 1 Introduction to Matrix Algebra

Characteristic-Value Problems and Quadratic Forms

Relative Extrema for a Function

Differential Geometry of a Member Element

vi 4-3 4-4 4-5 4-6 4-7 4-8

Matrix Transformations for a Member Element

II ANALYSIS OF AN IDEAL TRUSS Governing Equations for an Ideal Truss

115 116 120 125 130 130 132 134 137

Variational Principles for an Ideal Truss

Displacement Method Ideal Truss

Incremental Formulation; Classical Stability Criterion Linearized Stability Analysis

Force Method Ideal Truss

III ANALYSIS OF A MEMBER ELEMENT 10 Governing Equations for a Deformable Solid

St. Venant Theory of Torsion-Flexure of Prismatic Members

Engineering Theory of Prismatic Members

viii 12-3 12-4 12-5 12-6

333 339 340 349

Restrained Torsion-Flexure of a Prismatic Member

Planar Deformation of a Planar Member

14-5 14-6 14-7 14-8

435 445 449 458 473

Engineering Theory of an Arbitrary Member

15-9 15-10 15-11 15-12

520 524 531 535

IV ANALYSIS OF A MEMBER SYSTEM 16 Direct Stiffness Method Linear System

General Formulation Linear System

Analysis of Geometrically Nonlinear Systems

Part I MATHEMATICAL PRELIMINARIES

Introduction to Matrix Algebra

o r a tw o-dim ensional array, such as

a 21, a 2 2 > > a 2 n

EQUALITY, ADDITION, AND SUBTRACTION OF MATRICES

If a is of o rd er m x n, th e m atrix eq u atio n a = b co rresp o n d s to mn eq u atio n s: i = 1, 2, . . . , m aij = bfj j = 1 ,2 ,..., n

Scalar m ultiplication is com m utative. T h at is,

+ am2x 2 + + amnx = cm This set can be w ritten as n Z aikx k = C t= i i = 1, 2, . . . , m (a)

ax = < (8)(2) I- ( 4)(3) - =

M ATR IX M ULTIPLIC ATIO N

Substituting for x k in (1-10), I Z au f E = (ci) i = 1, 2, . . . , m (a)

Interchanging the o rd er of sum m ation, and letting Pij = L

the transform ed equations take the form ( l Pijy, j = i = 1, 2 , . . m (1 -1 5 )

N oting (1-12), we can write (1-15) as py C

- 1)(1) + (1)(0) ( - l ) ( i ) + (l)(l) (0)(1) + (2)(0) (0)(1) + (2)(1)

- 1X0) + ( i ) ( ~ n (0)(0) + (2)( - 1 j

a22_ J>21 bi2 a li

aml ' ' &m2 & m n

T h e tran sp o se of p will be o f o rd e r s x m an d th e typical elem ent is Pij = Pji (c)

SPECIAL SQUARE MATRICES

Diagonal M atrix, Order 3

Unit Matrix, Order 2

SPECIAL SQUARE MATRICES

W e in tro d u ce th e K ro n eck er d elta n o ta tio n : i * J i = j (1-2 8 )

W ith this n o tatio n , th e unit m atrix can be w ritten as = [<5y] Uj = 1, 2 , . . . , (1 -2 9 )

Also, the diagonal m atrix, d, takes th e form

O ' "3 1' 2 7 -1 1'

7 (< 2(j )" fljj)

OPERATIONS ON PARTITIONED MATRICES

#12 I ' ----- 1 <?22 1 | f l 32

OPERATIONS ON PARTITIONED M ATRICES

Using (1-44), we obtain ii 12 13 < ^21 a22 #23