Vous êtes sur la page 1sur 36

FINAL REPORT TO PROBLEM 1 IN MCE 613-FINITE ELEMENT METHOD:

EFFECTIVENESS OF FINITE ELEMENT


METHOD IN THE DEFORMATION ANALYSIS
OF 1-D ELASTICITY MEMBER
“I swear on my honor, I have not given nor received any inappropriate aid in the completion
of this report.”

Prepared by:

JEJOMAR U. DUQUE
MSCE Student
University of San Carlos

SEPTEMBER 2018
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
ABSTRACT
This report presents the solution of a sample 1-D Elasticity problem using Analytical Method and
Finite Element Method. The effectiveness of the FEM in solving the deformations is evaluated in
accordance with Analytical Method. The first part of this report shows the Analytical Method
including the derivations of differential equations. The deformation or displacement function using
Analytical Method is established that will be used in the gauging the efficiency of FEM. In the
FEM, two types of elements are used, namely the 2-node element and 3-node element. Each
element type has three cases based from the number of elements in the FE analysis, where author
use 2, 6 and 14 elements per element type. The author presents the possible minimum number of
elements in order to achieve an optimum accuracy of 99.99% or an error of 0.01%. The results
shows that the element type impacts the accuracy of the FEM. The results also reveals that the
level of accuracy proportionally increases with the number of elements. It is recommended to
review related literatures in the field finite element modeling inorder to have good idea in
identifying the suitable type of element to be used. Overall, finite element method is an efficient
way of solving elasticity problems compared with analytical method which is definitely
painstakingly time consuming.

1
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
I. Introduction
The basic philosophy in the design of the elements in a structure is the Limit State. When a structure
or element of the structure fails to perform its structural role, this is a circumstance of limit state.
There are two categories of limit states: strength and serviceability [1]. In the Strength Limit state,
the dimension of the member is determine by rationalizing the demand and the strength of the
member. Serviceability limit states are concerned to the performance of structures or elements of
these structures under normal (actual) service loads. It is good to design a structure with adequate
strength but it is not a guarantee that it will perform in the actual as expected even when the load
factor or factor of safety is 1.0. Serviceability of a structure is determined by considering its
deformations caused by the loads. It should be clearly emphasized that deformations do not usually
involve collapse but if too excessive may disrupt the use of structures. Therefore, the deformations
should be within tolerable values as specified the Regulating Organization so that in return
structures will have habitable conditions.
In the other point of view, recalling the principles under Mechanics of Materials, the analysis of
deformations, will help us determine the stresses within the member [2]. Stress distribution plots
and figures are very effective in identifying the locations within the member which are highly
stressed producing well modified design of the member. Mostly the computation of the stresses
are basically dependent to the internal forces in the element that is based on the concept of simple
stresses, although in reality stresses are indeterminate and variable along the transverse and
longitudinal direction. A drawback for those statically indeterminate structures where forces
cannot be obtained by just using the equations of static equilibrium. Thus, an additional relation
should be made between the deformation and stress known as the compatibility relation to obtain
the actual stresses. In other words, the stress field cannot be obtained without determining the
deformations within the member.
This paper aims to give a concrete example of the analysis of the deformation or displacement of
a structural member. A suspended axially loaded steel member is used and shown in Fig. 1.1. So
that the analysis will be thoroughly performed, the parameters are assumed given as follows:
Modulus of Elasticity, E = 200 GPa 2.4 m
Density of the member ρ = 7800 kg/m3
x
Dimensions: width, w = 1.2 m A
Height, L = 15 m P
Thickness, t = 0.1 m 7.5 m

The member is supported rigidly on its top and to carry


a point load of P = 10,000 kN at midheight. Since the B
member is assumed to be axially loaded only so the
problem qualifies that of a 1D elasticity problem. 7.5 m
Neglecting the Poisson’s effect, in 1D Elasticity
problem, it is assumed that the deformation is only C
dependent on one direction, same also with stress
distribution. In this particular sample problem, the 1.2 m
deformation or displacement of the suspended body is
y
only along the y-axis. This means that within the cross-
section the deformation is uniform. Fig. 1.1 1D Elasticity Problem

2
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
The report is devoted to present two methods in solving the displacement specifically at point C
on this particular problem. The first method is using the Analytical Solution and secondly using
Finite Element Solution. The analytical method is obtained inorder to check the robustness and
effectiveness of the Finite Element Analysis (FEA). The analytical solution of the problem with
the aid of spreadsheet software. In Finite Element Analysis, there two major type of elements used,
first is the Two-Node Element and second is the Three-Node Element. The goal of having two
different element in the analysis, so that later on the author can evaluate which element is effective
in approximating the true values in terms of displacements. Furthermore, there are three cases that
are being studied in the FEA these are using 2 elements, 6 elements and 14 elements for each type
of element.
In the FEA, the theoritical stress distribution in equilibrium equations is replaced with an
approximate finite element stress field, causing an error function which is a function of spatial
coordinates and unknown parameters in FEA formulation [3]. The error in the approximation
should be within the tolerable boundaries to obtain reliable results in the FEA. Inorder to evaluate
the precision and accuracy of FEA, a study is undertaken by determining the number of two-node
element that will yield a maximum error of 0.01% compared to the Analytical Solution.
As a summary, there are atleast seven (7) results expected for the FEA. Majority of the results
were done using manual FEA using spreadsheet and MathCad software, but the author make used
LISA 8.0 software as a tool for verification of results. However, the author solely used LISA 8.0
software in the optimization part in the FEA. The discussion of the analytical method is presented
in the next part and the Finite Element Method is presented in Chapter II in this paper. The
numerical values are tabulated, the displacement functions are presented with the corresponding
graphs and plots in Chapter 4 for the Analytical and FE Method.
Analytical Formulation/Solution of the Sample Problem
The analytical solution of the problem involves the equation/s of equilibrium and compatibility
equation. On this part of the paper the author only provides direct and sparse formulation of the
solution while the complete or detailed solution is attached in the appendices of this paper.
The analytical formulation begins with the calculation of the Ry
support reaction (Ry) with the aid of an equation of
Equilibrium. As shown in Fig. 1.2, applying summation of
forces vertical will give the value of Ry. In deed Ry is x
A
depended on the weight of member and of course the load WAB
applied. 7.5 m
Applying Equilibrium of forces: P
 Fv  0    B
Ry  P  WAB  WBC (1)
Weight Calculation: WBC
7.5 m
WAB   gAAB t (2)
WBC   gABC t (3) C
The weight is determined by solving the volume of the
1.2 m
member multiplied by its unit weight. For convenience, the
member is assumed to be divided in two segments, the y
tapered part (AB) and the uniform part (BC), with the Fig. 1.2 Support Reaction
weights WAB and WBC respectively.

3
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
Noting that the primary goal of this paper is to calculate for the displacement at C which is
dependent on the displacement function of the entire member. In the same fashion with the
weights, the displacement function is determined on each segment.
To obtain the displacement function of the segment AB, a section is analyzed within the segment
and shown in Fig. 1.3. Utilizing the upper portion of the section 1-1, using summation of the forces
for equilibrium, it can be shown that the internal
Ry
resistance at segment AB is dependent on the
reaction Ry and arbitrary weight W1, which is the
x
weight of segment AB within distance “y” from 1 A 1
the top. By applying simple ratio and proportion
the width of segment AB at distance y is 7.5 m P
obtained which is useful in determining the
weight W1. Using the same concept in
B
determining the weight WAB, the weight W1 as a 2 2

function of y is derived. The internal resistance


produces the normal/axial stress distribution 7.5 m
σ(y) in the section and represents the stress
function of y at any point within segment AB. C
W1   gA1t (4)
1.2 m
Applying Equilibrium of forces:
 Fv  0   y
The internal resistance is simply equal to the 2.4 m
Ry
cross-sectional in terms of x, A(x) multiplied the
stress σ(y) basing from the definition of “simple y
normal stress”. dy
  y  A x   Ry  W1  y  (5) A1 W1(y)
7.5-y x x y
Assuming that the behavior of the member under
the loading condition is within the elastic range,
thus Hooke’s law can be applied in this problem. 0.6 m 1.2 m 0.6 m
The law states that, “the stress is directly σ(y)
proportional with strain”, and expressed as, SECTION 1-1
 E (6) Fig. 1.3 Segment AB
with σ and ε are the stress and strain respectively
and E is the Young’s Modulus or commonly known as the Modulus of Elasticity. The expression
above is known as the stress-strain constitutive relation. Strain (ε) is defined as the deformation
(δ) in the member over its original length or,

 (7)
L
and instantaneously, for a length of dy the deformation is dδ and using Equation 7, the strain is
expressed as,
d
 (8)
dy
If equation (8) is substituted to equation (6), the resulting equation now becomes,
 d 
  y    E (9)
 dy 

4
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
The equation above is known as the compatibility equation. Substituting equation (9) to equation
(5), will yield to what is called the 1D Elasticity differential equation given as,
 d 
  E A  x   Ry  W1  y  (10)
 dy 
where, A  x   2 xt and solving for the value of dδ, the equation becomes
 Ry  W1  y 
d   dy (11)
E A x
Equation (11) is simply a first order ordinary differential equation (ODE), the general solution to
the displacement function can obtained by direct integration. What is exactly the objective is to
know the particular solution for the problem which can determined by imposing the boundary
conditions. Since the member is rigidly connected at point A (y = 0), therefore, δ = 0 at y = 0.
Imposing Boundary Condition: δ = 0 at y = 0 and
  y     0   d
y
 Ry  W1  y  
  y    dy (12)
0 E A x 
The author reiterates that the function (12) is the
displacement function within segment AB. The 2.4 m
formulation of the displacement function for
segment BC is similar to that of segment AB.
Utilizing a section within segment BC as shown
in Fig. 1.4, the member is then analyze using the
y
part below the cutting plane 2-2. The following
equations are used in the determination of the σ(y)
displacement function for segment BC, y-7.5
W2  y    gA2  y  t (13) dy W2(y)
  y  A x   W2  y  (14) 15-y 15-y

where A  x   wt
1.2 m
 d 
  E A  x   W2  y  (15)
SECTION 2-2
 dy 
W2  y  Fig. 1.4 Segment BC
d  dy (16)
E A  x
Equation (16) is the ODE for segment BC and imposing boundary condition, that is at y = 7.5, the
displacement, δ = δB. Equation (12) is used to obtain the displacement at point B, δB. By integrating
Equation (16), becomes
  y     7.5   d  (17)
y
W2  y 
  y   E A  x  dy    7.5 (18)
7.5

The simplified expression in (18) gives the displacement function at segment BC. As mention, the
robustness and effectiveness of the FEA is determined using the displacement at C, which can be
obtain using the equation (18). The graph of the analytical displacement function is plotted and
presented in Chapter IV.

5
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
II. Numerical Model Formulation
As declared in Chapter I, the numerical analysis are Finite Element Analysis (FEA) in solving the
problem. In the FEA, the member is being discretized in two several elements and shapes. There
are two types of elements used by the author in the FEA, the two-node element and three-node
element good for 1D elasticity problems. In the same manner to what is discussed in the analytical
method in Chapter I, the goal in FEA is the displacement function or specifically the displacement
at point C shown in Fig. 1.1. The displacement functions in the FEA is governed by the “shape
functions” or “basis functions”. The shape functions
ω(x)
for the two-node and three-node elements are derived
ω(x)=a0 + a1x
in this chapter. The number of nodes in an element d2
also represents the number of shape functions that to
be defined inorder to approximately capture the exact d1
displacement function. In FEA, in general the shape
functions are polynomial expressions, thus
displacement functions are polynomials with variable
L
degree. The degree of the polynomial displacement x
function is once again related to the number of nodes x1 Element # 1 x2
in a discretized element.
Fig. 2.1 Linear Shape Function
Two-Node Element
In a two-node element, the shape function used is linear function. Moreover, in a two-node
element, there two shape functions that should be obtained as described above. Fig. 2.1 depicts an
example of two-node element with the displacement function, ω(x). The author used the symbol ω
to emphasize that the obtained displacement function is simply an approximate function with
respect to the true displacement function u(x) presented in Chapter I. Based on the figure, the
displacements at node 1 and node 2 are d1 and d2 respectively, and at node 1 and node 2 the x
ordinates are x1 and x2 respectively.
The displacement function for two-node element is,
 ( x)  a0  a1 x (19)
where a0 and a1 are undetermined coefficients in the polynomial expression. The displacement
function can also be expressed in terms of the shape functions,
 ( x)  N1 ( x)d x1  N 2 ( x)d x 2 (20)
where N1 and N2 are shape functions at node 1 and node 2 respectively. Imposing the conditions:
1. At x = x1, ω(x) = d1, and using Equation (19)
d1  a0  a1 x1 (21)
2. At x = x2, ω(x) = d2, and using Equation (19)
d 2  a0  a1 x2 (22)
Equations (21) and (22) are systems of linear equations, using elimination method or substitution
method, the undetermined coefficients is obtained. Subtracting equation (22) by equation (21) will
result to,
d 2  d1  a1 x2  a1 x1 (23)
therefore coefficient a1 is equal to
1
a1   d 2  d1  (24)
x2  x1

6
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
Substituting equation (24) to equation (21), the value of a0 is equal to
 1 
d1  a0    d2  d1  x1 (25)
 x2  x1 
 1 
a0  d1    d 2  d1  x1 (26)
 x2  x1 
Substituting equations (24) and (26) to equation (19), the displacement function ω(x) is given as,
 1  1
 ( x)  d1    d 2  d1  x1   d 2  d1  x (27)
 x2  x1  x2  x1
Re-arranging equation (27), the displacement function now becomes
 x2  x   x  x1 
 ( x)    d1    d2 (28)
 x2  x1   x2  x1 
It can be shown from equation (28) that the shape functions N1 and N2 are as follows:
x2  x
N1  x   (29)
x2  x1
x  x1
N2  x   (30)
x2  x1
In matrix form, the displacement function can be expressed as,
d 
 ( x)   N1 N 2   1  (31)
d 2 
The derived shape functions should have the properties N(x)
of (1) Kronecker Delta property, (2) Compatibility and N1(x) N2(x)
(3) Completeness as mention in reference [4]. The 1
graphical representation of the linear shape function is
shown in Fig 2.2. Kronecker delta property means, that
the shape function at any node has a value equal to 1 at
the node and zero at all other nodes. For example, N1 is
1 at node 1 but zero at node 2 and same. Compatibility
and completeness of the shape function is measured by L
x
the convergence of the FE displacement function with x1 x2
respect to the exact displacement function. Element # 1
Fig. 2.2 Kronecker Delta Property
Stiffness Matrix
The displacement function given above clearly shows that having obtained the shape functions,
the next task is to determine the nodal displacements per element. From the principles obtain in
Matrix Analysis of Structures, the nodal displacements can be determined using the equation,
F   K d (32)
where {F} is the nodal force vector, [K]is the structure’s global stiffness matrix and {d} is the
displacement matrix. The nodal force vector is consist of all external nodal force and the equivalent
nodal force due to its own weight. The structural stiffness matrix is sum of all global elemental
stiffness matrix. To determine the elemental stiffness the expression below is used,
x2

k   BT EBAdx (33)
x1

with B matrix (BT means transpose of B) as the displacement-strain relation matrix, E is the
Young’s Modulus, A is the cross-sectional area of the element, and x2 and x1 are the x-boundaries

7
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
of an element. In Mechanics of Materials we’ve been relating displacement or deformation to strain
that is,
d
  x  (34)
dx
by substituting (31) to (34), the strain becomes,
d d 
  x   N1 N2   1  (35)
dx d 2 
it can also be,
 dN1 dN 2   d1 
  x     (36)
 dx dx  d 2 
differentiating equations (29) and (30) with respect to x and substituting to equation (36)
 1 1   d1 
  x     (37)
 x2  x1 x2  x1  d 2 
Since, x2 – x1 is also the length of the element (L), and therefore the B matrix,
1
B  1 1 (38)
L
1  1 (39)
BT 
L  1 
and substituting equations (38) and (39) to equation (33) and where the area, A and E are constants
within an element,
AE 2  1 1
x

L2 x1  1 1 
k  dx
(40)

simplifying equation (40) further,


AE  1 1
k (41)
L  1 1 
Nodal Force Vector
Commonly body forces such as inertial and weight are uniformly distributed load, since the target
is to determine the nodal displacement, thus the body forces should be converted to nodal forces.
Inorder to distribute the body forces to the nodes of an element, the shape function is also utilized
and the nodal forces are expressed as,
x3

 fb     N T  bdx (42)
x1

where b = the uniformly distributed body force and NT is the transpose of the N matrix.
Three-Node Element
In a three-node element, the shape function used is quadratic function. In addition, in a three-node
element there are three shape functions that should be obtained. Fig. 2.3 shows a three-node
element with the displacement function, ω(x). Based on the figure, the displacements at node 1,
node 2 and node 3 are d1, d2 and d3 respectively, and at node 1, node 2 and node 3 the x ordinates
are x1, x2 and x3 respectively. The displacement function for two-node element is,
 ( x)  a0  a1 x  a2 x 2 (43)
where a0, a1 and a2 are undetermined coefficients in the polynomial expression. The displacement
function can also be expressed in terms of the shape functions, where N1, N2 and N3 are shape
functions at node 1, node 2 and node 3 respectively.

8
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
ω(x)
ω(x)=a0 + a1x + a2x2

d1
d3
d2

x
x1 x2 x3
Element # 1
N1 (x) N 2 (x) N 3 (x)

x
x1 x2 x3
Element # 1
Fig. 2.3 Quadratic Shape function
 ( x)  N1 ( x)d x1  N 2 ( x)d x 2  N3 ( x)d x 3 (44)
Imposing the conditions:
1. At x = x1, ω(x) = dx1, and using Equation (32)
d x1  a0  a1 x1  a2 x12 (45)
2. At x = x2, ω(x) = dx2, and using Equation (32)
d x 2  a0  a1 x2  a2 x2 2 (46)
3. At x = x2, ω(x) = dx2, and using Equation (32)
d x 3  a0  a1 x3  a2 x3 2 (47)
Equations (34), (35) and (36) are systems of linear equations given that x1, x2 and x3 are known,
using matrix operation, the undetermined coefficients is obtained.
Equations (34), (35) and (36) can be written in matrix form as,
 d x1  1 x1 x12   a0 
    
d x 2   1 x2 x2 2   a1  (48)
 d  1 x x32  a2 
 x3   3

Applying inversion of the matrix collecting the numeric values of x1, x2 and x3 of the equations
mentioned above, the coefficients a0, a1 and a2 are shown in (38). Substituting the values obtained
in (38) to equation (32), the polynomial displacement function is given in (39). Therefore, with the
result in (39) the shape functions for three-node element is given in (40), (41) and (42). Indeed,
the shape functions are quadratic polynomial, which means that the highest degree of the
polynomial is 2.

9
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
 d x1  x2 x32  x3 x2 2   d x 2  x1 x32  x3 x12   d x 3  x1 x2 2  x2 x12  
 
  x2  x1  x3  x1  x3  x2  
 a0   
   d x1  x3  x2   d x 2  x3  x1   d x 3  x2  x1 
2 2 2 2 2 2

(49)
 a1    
a    x2  x1  x3  x1  x3  x2  
 2 
d x1  x3  x2   d x 2  x3  x1   d x 3  x2  x1  
 
  x2  x1  x3  x1  x3  x2  
 

d x1  x2 x32  x3 x2 2    x32  x2 2  x   x3  x2  x 2   d x 2  x32  x12  x   x1 x32  x3 x12    x3  x1  x 2   d x3  x1 x2 2  x2 x12    x2 2  x12  x   x2  x1  x 2  (50)
 ( x) 
 x2  x1  x3  x1  x3  x2 
  x  x3  x 
N1 ( x)  2
x
(51)
 x2  x1  x3  x1 
N 2 ( x) 
 x3  x  x  x1  (52)
 x2  x1  x3  x2 
N 3 ( x) 
 x2  x  x1  x  (53)
 x3  x1  x3  x2 
It should be clear that the derived quadratic shape functions should have the good qualities similar
to what is described in linear shape functions. The dense derivation of the quadratic shape function
is presented in the Appendices of this paper. The derivation of the stiffness matrix is similar to that
of the two-node element, so the author only provides the simplified derived expression below. The
detailed derivation of the stiffness matrix for the three node element is also shown in Appendices.
 7 8 1 
AE 
k 8 16 8 (54)
3L 
 1 8 7 
The expression in (54) can only be used for three-node element with constant cross-sectional area,
A and modulus of elasticity E.
Lisa 8.0 Software
LISA is an easy and free FEA
package for Windows where
modeler is being integrated, can
execute multiple processes or
threads at the same time and
having a good post-process
graphics. However, the software
is limited only to 1300 nodes of
free use making it
disadvantageous for structures
with thousands of nodes.
In LISA 8.0, only the geometry
(such as the nodes and elements)
and material properties (such as cross-sectional for 1D element, dimensions for 2D and 3D
element, modulus of elasticity, poison’s ratio, material density and etc.). The shape functions, B
matrix, stiffness matrix, nodal force vector, displacement, reaction and stresses are being solved
by the software. However, another drawback for Lisa 8.0 for free use is that it cannot solved 3-
node element for 1D elasticity. The results can be extracted in csv file format or liml file format.

10
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
III. Methodology for Finite Element Analysis (FEA)
In this chapter, the process of solving the problem presented in Chapter I using FEA is given
attention. The problem is solved using three cases, 2, 6 and 14 discretized elements using linear
and quadratic displacement function, the general procedure is provided for both element type as
follows:
1. The member is discretized into 2, 6 and 14 elements, shown in Fig. 3.1 below. Thus, the member
is idealized as having a piecewise constant cross-section for element with two-nodes and
variable area per element with three nodes.
2. Two types are used in each cases, two-node (linear displacement) element and three-node
(quadratic displacement) as also shown in Fig. 3.1. In the three-node element, the nodes are
located at the ends and at the midpoint on each element.
3. The geometric properties of each element are determined these includes, the nodal coordinates
in this case along the y-axis, the cross-sectional areas and also the material properties (i.e.
Young’s Modulus E and thickness t) are assigned. The nodal coordinates are obtained by using
the dimension of each discrete elements and the cross-sectional areas are based from the average
of the end areas per element with two-nodes.
4. The shape function matrix N is evaluated using the expressions (29-30) and (51-53).

1 E#1
E#1 E#2
E#3
E#1 E#2 E#4
E#5
E#6
2 E#3 E#7
E#8
E#4 E#9
E # 10
E#2 E#5 E # 11
E # 12
E # 13
3 E#6 E # 14

2 Elements 6 Elements 14 Elements


a) Two- Node Element

1 E#1
E#1 E#2
E#3
E#1 2 E#2 E#4
E#5
E#6
3 E#3 E#7
E#8
E#4 E#9
4 E # 10
E#2 E#5 E # 11
E # 12
E # 13
5 E#6
E # 14

2 Elements 6 Elements 14 Elements


b) Three- Node Element
Fig. 3.1 FEA Cases

11
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
5. The stiffness of each element for two-node is determined using the derived expressions in (41),
since it assumed that per element, A and E are constant values. The stiffness of each element
with three nodes is determined using the expression (33). Afterwards, the global stiffness is
acquired by summing-up all elemental stiffness matrices but taking into considerations the
common nodes for adjacent elements.
6. The nodal forces are determined using the expression in (42) and set-up the nodal force vector.
7. Apply the boundary conditions and reduced the stiffness matrix and force vector. The equation
in (32) is used inorder to obtain the nodal displacements for the entire member.
The given procedure above is purposely presented for manual FEA that means without using FEM
software. However, the author as clearly stated in Chapter I, made use LISA 8.0 software as a tool
in the FEA. In addition, only steps 1-3 are to be followed in using the software since the application
can already provide the quantities stated in steps 4-7. For clarity, the author provides an example
in FEA using quadratic function (three-node) for 2 elements.
Numerical Example of FEA (Three-Node) 2 Elements
1. Discretization of the structure into 2 (three-node) elements
2. For 2 elements with 3-nodes each element, so there are 5 1.8 m
nodes in the member. The nodal coordinates are as follows:
Node y (m)
1
1 0
2 3.75 3.75 m
3 7.50 2 7.5 m
4 11.25
5 15.00 3.75 m
3
3. The nodes for each element are as follows: 15 m
Element Node 1 Node 2 Node 2
1 1 2 3 3.75 m
2 3 4 5
4. Given that: E = 200 GPa and ρ = 7800 kg/m3 4

5. The shape functions are evaluated using the derive 3.75 m


5
expressions in (51-53).
y
Element 1: N1 (y)  
3.75  y  7.5  y  y  7.5  y 
N 2 (y) 
28.125 14.0625 1.2 m
 y  3.75  y  Fig. 3.2 FEA for 2 elements
N 3 (y) 
28.125
  3.75  y  7.5  y  y  7.5  y   y  3.75  y  
N1   
 28.125 14.0625 28.125 

Element 2: N1 (y)  
11.25  y 15  y  15  y  y  7.5
N 2 (y) 
28.125 14.0625

N3 (y) 
11.25  y  7.5  y 
28.125
 11.25  y 15  y  15  y  y  7.5 11.25  y  7.5  y  
N2   
 28.125 14.0625 28.125 

12
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
6. The elemental and the structural stiffness are solved using the expression in (33) for element
1 since it is having variable area and expression (54) for element 2 for having uniform area.
x2  7 8 1 
A2 E 
Element 1: k1  B
T
EBAdx Element 2: k2  8 16 8
3L 
x1
 1 8 7 
Overall: K  K1  K 2
The overall structural stiffness matrix is 5x5 square matrix, and take into
consideration the common node 3.
7. The nodal body forces are determined using the expression in (42).
 f b1   f b 3  15
  7.5  
 f b2     N1  b1 dy  f b4     N 2  b2 dy
T T
Element 1: Element 2:
  0   7.5
 f b3   fb5 
The member is subjected to an external nodal force of 10 kN at node 3, it should be considered
in the nodal force vector.

8. Apply the boundary condition: At y = 0, d = 0


Reduced the 5x5 stiffness matrix to 4x4 matrix by omitting the first row and first column of
the 5x5 structural stiffness matrix. Set-up the equation similar to (32) and omit the first row
in the nodal force vector. Solve for the displacements using matrix inversion of the reduced
stiffness matrix and multiply it to the reduced nodal force vector.

The numerical values of the displacements are tabulated and graphed in the next chapter.
Optimization Study
As introduce in Chapter I, the author also presents the minimum number of 2-node elements
(linear shape function) inorder to obtain an optimum accuracy or margin of error of 0.01%. It
can be shown in the results that having discretized the member in 14 elements is enough to
attain an accuracy of 0.01%. Inorder to achieve the task, a trial and error process is undertaken
using the algorithm and flow chart below.
Algorithm
1. Discretize the member into specified number of elements (i.e., 28 elements)
2. Using LISA 8.0 software execute FEA analysis
3. Determine the displacement at C.
4. Calculate the relative error
5. If error > 0.01% then go back to step 1 (i.e., 35 elements)
If error < 0.01% then the number of elements is obtained (i.e., 28 elements)
6. End

13
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
Flow Chart
START

INPUT No Yes
Is error
Number of elements & Now the loop continues with new < 0.01 %
Material properties number of elements
OUTPUT
Minimum Number of
elements is obtained
Execute FEA using
Determine the
LISA 8.0 Software &
relative error
determine the
displacement at C
END
The report presents three more additional FEA with arbitrarily choosing 28 elements, 40 elements
and 42 elements for the optimization study.

IV. Results and Discussion


Analytical Solution
The simplified equations based from what is discussed in Chapter I and its numerical results are
presented in Table 1.
Table 1. Equations of the analytical solution of the problem
Single Variable Equations
1 WAB  103, 299.3 N
2 WBC  68,866.2 N
3 Ry  10,172,165.5 N
4 W1  y   18364.32 y  612.144 y 2
N
11 1.0172 x107 N  18364.32 y  612.144 y 2   mm
d    dy
4.8 x1010  3.2 x1010 y
12  15  y  mm
  y   0.002869425 y  3.13576034375ln    0.0000956475 y
2

 15 
13 W2  y   137732.4  9182.16 y N
16 137732.4  9182.16 y mm
d  dy
2.4 x107
18   y   0.03228103125  0.00573885 y  0.000191295 y 2 mm
Equations (12) and (18) gives the generalized displacement field of the member at segment AB
and BC respectively. The equation (12) which depicts the displacement field for the tapered part
AB is quadratic-logarithmic function. The uniform cross-section part BC has a quadratic
displacement function as shown in equation (18). Using equation (18) and substituting y = 15, the
displacement at C is obtained which is 2.200444301 mm. In addition using those equations, the
displacement field can be graph and plotted as displacement vs. distance “y” from the origin as
shown in Fig. 4.1.

14
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
Finite Element Solution
The results for the displacement at C from the sample problem are presented in Table 2. The results
shows that 14 elements with 3-nodes per element (quadratic element) gives the most robust
approximation of the exact displacement at C with a margin of error of 0.00010596421% to be
exact. In contrast, 2 elements with only 2-nodes per element (linear element) gives the weakest
approximation of the displacement at C at an error of 3.692125%.
ANALYTICAL VS FEM SOLUTION
DISPLACEMENT AT C (MM)

2.21
2.19
2.17
2.15
2.13
2.11
2.09
2.07
2 ELEMENTS 6 ELEMENTS 14 ELEMENTS
ANALYTICAL 2.200444301 2.200444301 2.200444301
FEM 2 NODES 2.119201146 2.190012429 2.198491372
FEM 3 NODES 2.199053445 2.200397233 2.200441969

Table 2. Displacement at C
Influence of element type
Based from the results, the type of the element used definitely affects the finite element solution.
Comparing the results for the 2-node elements and 3-node elements, there are significant
difference of the values. The results for 3-node elements generally show certain and good
approximation of the displacement at C. The reason behind why 3-node elements gives betters
results it is because the shape function and displacement field as derive in chapter II in this paper
is clearly a quadratic function. As per observation, the displacement function obtained using
analytical method shown in equations (12) and (18) contains quadratic term. It can be shown in
Fig 4.1 where the displacement field for the 3-node elements almost overlapped or coincided with
the analytical solution of the deformation field.
In addition, the general distribution of stresses and deformation in a member is varying spatially.
In other words, there is significant difference in the stress and deformation at all locations within
the member, although slight difference may be expected in between locations with short distances.
Going through in our derivation for linear element, the displacement function is linear and applying
differentiation of displacement will obtain strain, meaning the strain is constant at all locations
within the element. Since strain and stress are proportional based on the assumption by Hooke’s
law, therefore stress is constant which is not true at all points within the element. It is known in
mechanics of materials that deformation is caused by the stress induced, meaning inaccurate stress
distribution will yield in accurate deformation field. With this logic linear element is definitely
weak in our finite element calculations.
Indeed, using quadratic element higher accuracy compared with linear element considering only
this type of problem, since a general truth can’t be established for the author only provides single
example problem. However, the results obtained agree to observations of some authors based from
their respective research and studies, where usually elements with higher order functions yields
better approximation. Hence, choosing an element type is very significant in the finite element
method, since the element chosen may produce near or close approximation of the exact solution
of a given problem without going through numerous elements.

15
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members

Displacement Field for 2-Node Elements Displacement Field for 3-Node Elements
2.50 2.50

Displacement, δ (mm)
Displacement, δ (mm)

2.00 2.00

1.50 1.50

1.00 1.00

ANALYTICAL METHOD ANALYTICAL METHOD


0.50 2 elements (2-nodes) 0.50 2 elements (3-nodes)
6 elements (2-nodes) 6 elements (3-nodes)
14 elements (2-nodes) 14 elements (3-nodes)
0.00 0.00
0.0 5.0 10.0 15.0 0.0 5.0 10.0 15.0
Distance, y (m) Distance, y (m)
Fig 4.1. Displacement Field of the member
Influence of number of elements or element size
The values acquired shows that 6 elements yields better approximation compared with 2 elements
for the both linear and quadratic elements. Furthermore, results for 14 elements gives more
accurate result compared to 6 elements. There is inaccuracy in the approximations especially for
large size elements with two-nodes, but by decreasing the element size and so increasing the
number of elements gives favorable results most especially for the 2-node elements. As discussed
above, true stress distribution is variable at all locations. In finite element analysis, when the
element size decreases will result to numerous and non-uniform computed stresses, the finite
element stress field then converge approximately with the exact stress field which is varying at all
locations. See also Fig. 4.4.
Optimization Study
The finite element modeling of trial 2-node elements inorder to achieve an optimum accuracy of
99.99% or maximum error of 0.01% is done using LISA 8.0 software as shown in Fig. 4.2 and the
results are presented in Fig. 4.3. The percent error is based on the comparison of the FEM to
Analytical Method in solving the displacement at C of the sample problem given in this paper.

14 Elements 28 Elements 40 Elements 44 Elements

Fig. 4.2 Trial Number of Elements for 0.01% error

16
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
In Fig. 4.3, the results of the trial and error process using 2-node shows that the required accuracy
is attained after analyzing 44 elements. The accuracy at 14 elements is 99.91125% (error of
0.08875%), for the 28 elements is 99.97599% (error of 0.02401%), 40 elements has 99.98908%
(error of 0.01092%) and 44 elements being the most accurate at 99.9924% (error of 0.0076%).
Fig. 4.3 Accuracy of FEM for the Displacement at C
2.20

Displacement at C 2.20

2.20

2.20

2.19

2.19

2.19
ANALYTICAL 14 ELEMENTS 28 ELEMENTS 40 ELEMENTS 44 ELEMENTS
DISPLACEMENT AT C (mm) 2.200444301 2.198491372 2.199915942 2.200204114 2.200276976
RELATIVE LOCAL ERROR (%) 0.08875 0.02401 0.01092 0.00760
ACCURACY (%) 99.91125 99.97599 99.98908 99.99240
NUMBER OF NODES 15 29 41 45

It is very fact that using small element


size or numerous elements is time- Fig. 4.4 Relation of Number of
consuming and needs rigorous Elements and Accuracy
computation, although with the aid of 100.000
software time and effort may be reduced. 99.990
In the contrary, it is already established
99.980
above how the number of elements
influence our FEA results. 99.970
y, Accuracy(%)

99.960
It is presented in Fig. 4.4 how the
accuracy progresses as the number of 99.950 R² = 0.8886
elements increases. Based from the 99.940
graph, a linear regression or relationship 99.930
can be assumed between the number of
99.920
elements and accuracy where number of
elements is the independent variable and 99.910

accuracy is the dependent variable. By 99.900


this bi-variate’s linear regression, we can 0 4 8 12 16 20 24 28 32 36 40 44 48
make use of the square Pearson’s x, Number of Elements
correlation coefficient, R2 to describe the
relationship which is in this case equal to 0.8886. The R2 value means that 88.86% of the variations
in the accuracy can be explained by the changes in the number of elements. The remaining 11.14%
of the variance in the accuracy is believed to be due to random variability since we only choose 4
data points. Truly accuracy has high dependence to the number of elements which supports our
initial argument above.

17
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
V. Conclusion and Recommendations
The first part of this paper gives the solution of the problem using analytical method. The second
part of this paper is solving the problem using finite element method. The results of the study give
the following key points:
1. The type of element used affects the overall finite element modeling. Even with less
number of elements used in FEM, the results may converge early with the exact solution if
the displacement or stress functions in the particular type of element approximately similar
to the real displacement or stress field.
2. The author would like to recommend that before choosing an element type in the modeling,
one should know the behaviour and implications of a particular element in the entire FEA.
Thus, review of related literatures is very useful in the field of finite element modeling.
3. The number of element or the element size impacts the overall accuracy of the finite
element modeling. Using numerous elements or small element size will yield lesser relative
error with respect to the exact solution.
4. It is recommended to increase the number or elements or lessen the size of elements inorder
to obtain a required accuracy. However, manually using numerous elements is very tedious
but with the aid of finite element software the difficulty diminishes.
5. Overall, finite element method is an efficient way of solving elasticity problems compared
with analytical method which is definitely painstakingly time consuming.
V. References:
[1] McCormac, J.C. et al, Design of Reinforced Concrete, John Wiley & Sons, New York, 9th Edition, 2009
[2] Beer, F.P. et al, Mechanics of Materials, McGraw-Hill, New York, 5th Edition, 2009
[3] Richmond, B. G. et al, 2005, Finite Element Analysis in Functional Morphology, George Washington
University, USA, https://doi.org/10.1002/ar.a.20169
[4] Logan, D., First Course in the Finite Element Method, Cengage Learning, USA, 5th Edition, 2012
VI. Appendices:
Tables
Analytical and Finite Element Method Results
Analytical FEM 2-NODES FEM 3-NODES
y
Method 2 elements 6 elements 14 elements 2 elements 6 elements 14 elements
0.000 0.0000000000 0.0000000000 0.0000000000 0.0000000000 0.0000000000 0.0000000000 0.0000000000
0.250 0.0534144734 0.0702813601 0.0576768355 0.0548910054 0.0497859367 0.0532050459 0.0533948211
0.500 0.1077179420 0.1405627201 0.1153536710 0.1097820108 0.1011689194 0.1074376145 0.1077148758
0.750 0.1629417455 0.2108440802 0.1730305065 0.1646730162 0.1541489480 0.1626977056 0.1629601641
1.000 0.2191188879 0.2811254403 0.2307073420 0.2195640217 0.2087260226 0.2189853194 0.2191306861
1.250 0.2762841575 0.3514068003 0.2883841776 0.2775155364 0.2649001432 0.2763004559 0.2762601220
1.500 0.3344742574 0.4216881604 0.3460610131 0.3366912548 0.3226713098 0.3346431149 0.3344620050
1.750 0.3937279495 0.4919695205 0.4037378486 0.3958669732 0.3820395224 0.3940132966 0.3937451532
2.000 0.4540862106 0.5622508806 0.4614146841 0.4550426916 0.4430047809 0.4544110009 0.4541095666
2.250 0.5155924050 0.6325322406 0.5190915196 0.5163778791 0.5055670854 0.5158362279 0.5155684121
2.500 0.5782924732 0.7028136007 0.5767683551 0.5805923589 0.5697264359 0.5782889775 0.5782678765
2.750 0.6422351406 0.7730949608 0.6469965777 0.6448068387 0.6354828324 0.6418551570 0.6422471230
3.000 0.7074721467 0.8433763208 0.7172248003 0.7090213184 0.7028362748 0.7069643031 0.7075061514
3.250 0.7740584994 0.9136576809 0.7874530228 0.7740941297 0.7717867632 0.7736164157 0.7740452020
3.500 0.8420527559 0.9839390410 0.8576812454 0.8443169300 0.8423342976 0.8418114949 0.8420128738
3.750 0.9115173352 1.0542204010 0.9279094680 0.9145397303 0.9144788780 0.9115495406 0.9115182845
4.000 0.9825188646 1.1245017611 0.9981376905 0.9847625305 0.9882205044 0.9828305528 0.9825614341
4.250 1.0551285674 1.1947831212 1.0683659131 1.0549853308 1.0635591767 1.0556545317 1.0551423225
4.500 1.1294226954 1.2650644813 1.1385941357 1.1314515777 1.1404948950 1.1300214770 1.1293661714
4.750 1.2054830135 1.3353458413 1.2088223582 1.2089583991 1.2190276593 1.2059313889 1.2054644892

18
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
5.000 1.2833973443 1.4056272014 1.2790505808 1.2864652204 1.2991574696 1.2833842674 1.2834442860
5.250 1.3632601821 1.4759085615 1.3690707313 1.3639720418 1.3808843258 1.3624653136 1.3633055617
5.500 1.4451733858 1.5461899215 1.4590908817 1.4466273835 1.4642082280 1.4441117412 1.4451040014
5.750 1.5292469651 1.6164712816 1.5491110321 1.5331441154 1.5491291762 1.5283235501 1.5291967516
6.000 1.6155999736 1.6867526417 1.6391311826 1.6196608473 1.6356471704 1.6151007403 1.6156432463
6.250 1.7043615282 1.7570340017 1.7291513330 1.7061775792 1.7237622105 1.7044433118 1.7044434854
6.500 1.7956719766 1.8273153618 1.8191714835 1.7959589637 1.8134742967 1.7963512647 1.7956077756
6.750 1.8896842385 1.8975967219 1.9091916339 1.8939019799 1.9047834288 1.8908245989 1.8895848331
7.000 1.9865653535 1.9678780819 1.9992117844 1.9918449961 1.9976896069 1.9878633145 1.9865895438
7.250 2.0864982746 2.0381594420 2.0892319348 2.0897880124 2.0921928309 2.0874674114 2.0866219076
7.500 2.1896839568 2.1084408021 2.1792520853 2.1877310286 2.1882931010 2.1896368896 2.1896819247
7.750 2.1903893571 2.1087994802 2.1798498821 2.1883971451 2.1889985013 2.1903422899 2.1903873037
8.000 2.1910708455 2.1091581583 2.1804476790 2.1890632616 2.1896799897 2.1910237783 2.1910687749
8.250 2.1917284221 2.1095168365 2.1810454759 2.1897293781 2.1903375663 2.1916813549 2.1917263342
8.500 2.1923620868 2.1098755146 2.1816432728 2.1903954946 2.1909712310 2.1923150196 2.1923599817
8.750 2.1929718396 2.1102341927 2.1822410696 2.1909884115 2.1915809838 2.1929247724 2.1929697199
9.000 2.1935576805 2.1105928708 2.1828388665 2.1915520486 2.1921668247 2.1935106133 2.1935555451
9.250 2.1941196096 2.1109515490 2.1834366634 2.1921156856 2.1927287538 2.1940725424 2.1941174594
9.500 2.1946576268 2.1113102271 2.1840344603 2.1926793227 2.1932667710 2.1946105596 2.1946554627
9.750 2.1951717321 2.1116689052 2.1846322571 2.1931990400 2.1937808763 2.1951246649 2.1951695481
10.000 2.1956619255 2.1120275833 2.1852300540 2.1936601976 2.1942710697 2.1956148583 2.1956597275
10.250 2.1961282071 2.1123862615 2.1855887321 2.1941213552 2.1947373513 2.1960811399 2.1961259959
10.500 2.1965705768 2.1127449396 2.1859474103 2.1945825127 2.1951797210 2.1965235096 2.1965683534
10.750 2.1969890346 2.1131036177 2.1863060884 2.1950290304 2.1955981788 2.1969419674 2.1969868019
11.000 2.1973835805 2.1134622958 2.1866647665 2.1953877085 2.1959927247 2.1973365133 2.1973813374
11.250 2.1977542146 2.1138209740 2.1870234446 2.1957463867 2.1963633588 2.1977071474 2.1977519619
11.500 2.1981009368 2.1141796521 2.1873821228 2.1961050648 2.1967100810 2.1980538696 2.1980986755
11.750 2.1984237471 2.1145383302 2.1877408009 2.1964637429 2.1970328913 2.1983766799 2.1984214780
12.000 2.1987226455 2.1148970083 2.1880994790 2.1967345815 2.1973317897 2.1986755783 2.1987203593
12.250 2.1989976321 2.1152556865 2.1884581571 2.1969907802 2.1976067763 2.1989505649 2.1989953399
12.500 2.1992487068 2.1156143646 2.1888168353 2.1972469788 2.1978578510 2.1992016396 2.1992464095
12.750 2.1994758696 2.1159730427 2.1889363946 2.1975031775 2.1980850138 2.1994288024 2.1994735681
13.000 2.1996791205 2.1163317208 2.1890559540 2.1977008164 2.1982882647 2.1996320533 2.1996768100
13.250 2.1998584596 2.1166903990 2.1891755134 2.1978545356 2.1984676038 2.1998113924 2.1998561438
13.500 2.2000138868 2.1170490771 2.1892950728 2.1980082548 2.1986230310 2.1999668196 2.2000115666
13.750 2.2001454021 2.1174077552 2.1894146321 2.1981619740 2.1987545463 2.2000983349 2.2001430785
14.000 2.2002530055 2.1177664333 2.1895341915 2.1982864134 2.1988621497 2.2002059383 2.2002506791
14.250 2.2003366971 2.1181251115 2.1896537509 2.1983376531 2.1989458413 2.2002896299 2.2003343680
14.500 2.2003964768 2.1184837896 2.1897733103 2.1983888928 2.1990056210 2.2003494096 2.2003941459
14.750 2.2004323446 2.1188424677 2.1898928696 2.1984401326 2.1990414888 2.2003852774 2.2004300128
15.000 2.20044430050 2.11920114583 2.19001242902 2.19849137231 2.19905344471 2.20039723334 2.20044196882

Computations and Derivations


1. GENERAL ANALYTICAL SOLUTION TO ELASTICITY PROBLEMS
Governing Equations:
A. Equilibrium within the plate
Where fb = body force per unit volume
Applying summation of forces horizontal:
 Fh  0  +
 xx x  x  dx  dydz   yx y  y  dy  dxdz  f bx dxdydz   xx x  dydz   yx y  dxdz  0
  xx    yx 
  xx  x dx  dydz   xx  y dy  dxdz  f bx dxdydz   xx dydz   yx dxdz  0
   
 
 xx dydz  xx dxdydz   xx dxdz  dxdydz  fbx dxdydz   xx dydz   yx dxdz  0
yx

x y
 xx  yx
dxdydz  dxdydz  f bx dxdydz  0
x y

19
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
σyy|y=y+dy
y
Figure 3.0

τyx|y=y+dy

fby τxy|x=x+dx

σxx|x dy σxx|x=x+dx
fbx

τxy|x

x dx
τyx|y
y
σyy|y
x
 xx  yx
  fbx  0 → Equation 1a
x y
 xx  yx
 0 (Neglecting body force) → Equation 1b
x y
 Fv  0 +
 yy y  y  dy  dxdz   xy x  x  dx  dydz  f by dxdydz   yy  y  dxdz    dydz  0
xy x

  yy    xy 
  yy  dy  dxdz   xy  dx  dydz  fby dxdydz   yy dxdz   xy dydz  0
 y   x 
 yy  xy
 yy dxdz  dxdydz   xt dxdz  dxdydz  fby dxdydz   yy dxdz   xy dydz  0
y x
 yy  xy
dxdydz  dxdydz  fby dxdydz  0
y x
 yy  xy
  fby  0 → Equation 2a
y x
 yy  xy
 0 (Neglecting body force) → Equation 2b
y x
Applying summation of moment at center of element:
 M center  0  +

 xy
 x  x  dx  dydz   xy x  dydz   dx2   yx y  y  dy  dxdz    dxdz   dy2  0
yx y

  xy   dx   yx   dy
 xy  dx  dydz   xy  dydz     yx  dy  dxdz   yx  dxdz   0
 x   2  y   2
dx  dx dx dy  dy dy
 xy dydz   xy dxdydz    xy dydz    yx dxdz   yx dydxdz    yx dxdz   0
2 x 2 2 2 y 2 2

20
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
 xy dx 2 dydz  yx dxdy 2 dz
 xy  dxdydz     yx dxdydz   0
x 2 y 2
Dividing Both Sides by dxdydz: Assuming that:
 xy dx  yx dy  xy dx  yx dy
 xy     yx   0  0 and  0
x 2 y 2 x 2 y 2
 xy   yx → Equation 3
B. Hooke’s Law

Basic Stress-Strain Relation:  → Equation 4
E
Lateral Strain
Poisson’s Ratio: 
Longitudinal Strain
It follows:  xx   yy and  zz   yy
If the shown object is also stress along x and z direction:
 yy   xx and  zz    xx (due to σxx)
 xx    zz and  yy   zz (due to σzz)

σyy

εyy σyy

τyx
τyz
τxy
τzy
σxx
τzx τxz
σzz

εxx
εzz
σyy
Figure 4.0 Stress Along Y - Direction
By superposition principle, the total strain:
 xx  xx
 xx    yy   zz or  xx     yy   zz 
E E
 
 yy  yy   xx   zz or  yy  yy    xx   zz 
E E
 zz  zz
 zz    xx   yy or  zz     xx   yy 
E E
Generalized Hooke’s Law:
 xx    yy   zz  
1
 xx  → Equation 5a
E 

21
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
1
 yy   yy    xx   zz   →
E
Equation 5b

 zz   zz    xx   yy 
1
→ Equation 5c
E
C. Shearing Strain-Shearing Stress Relation
y
b'
τ
b
γ/2 τ
b
τ τ τ
σxx
45° c 45°- γ/2 σxx σyy
a O x σ
45°
O c' c
τ τ

d σyy

(a) Stress Flow (b) Deformed State (c) Mohr’s Circle


Figure 5.0 Pure Shear
   Ob ' 1   yy
It follows that: tan  45     → Equation 6a
 2  Oc ' 1   xx

tan 45  tan
  2
tan  45   
 2 
1  tan 45  tan
2

   
1
For very small γ: tan  and tan  45    2 → Equation 6b
2 2  2  1 
2
1
From Hooke’s Law:  xx   xx   yy  since  yy   xx
E

 xx  1    xx → Equation 6c
E
1
 yy   yy   xx 
E

 yy   1    xx → Equation 6d
E
 xx
1   yy 1  1    E
Right Side of Equation 6a: 
1   xx 

1  1   xx 
E

22
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
 
It follows that: 1  1  1    xx
2 E
 
Or  1    xx since  yy   xx
2 E
2 1   
 → Equation 7a
E
Let G (Shear Modulus/Modulus of Rigidity):
E
G → Equation 7b
2 1   

It follows that:  → Equation 7c
G
D. Stress Boundary Conditions:
y
psy
psn psy n
α β
psx
σxx α
psx
dy ds

x τxy dx β

Surface forces, fs
σyy
Figure 6.0
Where psn = surface force per unit area
Applying summation of forces horizontal:
 Fh  0 +
pn  dsdz  cos    xx  dydz   xy  dxdz  0
dy dx dy dx
Divided by dsdz: pn cos   xx    xy   0 or pn cos   xx    xy 
ds ds ds ds
psx   xx  cos   xy  cos 
Let: cos  l and cos   m
psx  l xx  m xy → Equation 8a
 Fv  0 +
pn  dsdz  sin    yy  dxdz   xy  dydz  0
dx dy dx dy
Divided by dsdz: pn sin    yy    xy   0 or pn sin    yy    xy 
ds ds ds ds
psy   yy  cos    xy  cos
psy  l xy  m yy → Equation 8b
Where α, β are the angle of inclination of the Normal axis from the global x and y axes respectively.
In addition, l and m are the direction cosine of N-axis to the global x and y axes respectively.

23
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
E. Compatibility Equation:
𝜕𝑢
𝑑𝑦
𝜕𝑦
𝜕𝑣
𝑣+ 𝑑𝑦
𝜕𝑦 𝐶 y, v

𝜕𝑣

𝑑𝑦 𝜕𝑦 𝜕𝑣 x, u
𝐵
𝜕𝑥 𝜕𝑣
𝐴 𝑑𝑥
𝜕𝑥
𝑣

𝑢 𝜕𝑣
𝑢+ 𝑑𝑥
𝑑𝑥 𝜕𝑥

u
It can be shown that the longitudinal strain along the xx-axis:  xx  → Equation 9a
x
v →
It can be shown that the longitudinal strain along the yy-axis: y  Equation 9b
y
u v → Equation 9c
It can be shown that the shear strain along the xy-axes:  xy  
y x
It follows that, the Partial Derivatives:
2  2  u  →  2 xx  3u →
2  xx 
  2   Equation 10a
y y  x  y 2
xy 2
2  2  v   2 yy  3v
x 2
  yy    
x 2  y 

x 2

x 2y
→ Equation 10b

 2   xy   2  u v   2 xy  3u  3v
    →   → Equation 10c
xy xy  y x  xy xy 2 x 2y
 2 xx   yy  2 xy 2
Substitution of Eq. 10a and Eq. 10b to Eq. 10c:  →  Equation 10d
xy y 2 x 2
The differential equation 10d is known as the strain compatibility equation.
1
Reducing Equation 5a:  xx   xx   yy 
E
1
Reducing Equation 5b:  yy   yy   xx 
E
2 1    xy
From Equation 7a:  xy 
E
Substitution of the 3 equations above to Eq. 10d:
 2  2 1    xy   2  1  
2
1 
   2   xx   yy    2  E  yy   xx  
xy  E  y  E  x  
 2 1      2 xy 1 2 1 2
    2  xx   yy     2  yy   xx  
 E  xy E y E x
 2 xy 2 2
2 1    
xy
 
y 2 
  xx   
yy 
   yy   xx  
x 2  
→ Equation 11a

24
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
Using the Equations of Equilibrium, differentiating the first and second with respect to x and then
y respectively and adding them afterwards:
 2 xx   yx fbx
2
From Eq. 1a:   0 → Equation 12a
x 2 xy x
 2 yy  2 xy fby
From Eq. 2a:   0 → Equation 12b
y 2
xy y
Adding Eq. 12a and Eq. 12b:
 2 xx   yy  2 xy fbx fby
2

  2   0
x 2 y 2 xy x y
 2 xx   yy fbx fby  2 xy
2

    2 → Equation 12c
x 2 y 2 x y xy
Substitute Eq. 12c and Eq. 11a:
  2 xx  2 yy fbx f by   2 2
 1     

   

 
2       
  2  yy
    xx  

 x y x y  y x
2 2 xx yy

 2 xx   yy  2 xx   yy  fbx fby   2 xx  2 yy  2 yy  2 xx
2 2

     1          
x 2
y 2
x 2
y 2
 x y  y 2
y 2
x 2
x 2
 2 xx  2 xx   yy   yy  f fby 
2 2

     1      bx   → Equation 13a
x 2
y 2
x 2
y 2
 x y 
 2 2   fbx fby 
 2  2   xx   yy    1        → Equation 13b
 x y   x y 
 2 2 
2   xx
 2     yy   0 (Neglecting body force) → Equation 13c
 x y 
 
Let  , the Laplace operator:  
x y
 2  xx   yy   0 (Neglecting body force) → Equation 13d
F. Strain-Displacement Equations
In Compact form:
 
 xx   1  0   xx   xx  1  0   xx 
  1     E   

 yy      1 0   yy  or  yy    1 0   yy 
  1  
2
  E
 xy 
 0 0 2 1      xy   xy 
 1     xy 
0 0 
 2 

From:  xx 
E
1  2
 xx   yy  → Equation 14a

 yy 
E
1  2
 yy   xx  → Equation 14b

E xy
 xy  → Equation 14c
2 1   
E   xx  yy  E  xy
Using Eq. 1a:    0
1   2  x x  2 1    y

25
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
E  2u 2 v  E  2u 2 v 
Using Eq. 9a, 9b and 9c:        0
1   2  x 2 xy  2 1     y 2 xy 
2u 2 v 1      2 u  2 v 
    
x 2
xy 2  y 2 xy 
 2u

 2v

1     2u  1  2v    2v
x 2
xy 2 y 2 2 xy 2 xy
 2 u  2 u 1      2 u 2 v  → Equation 15a
    
x 2 y 2 2  y 2 xy 

It follows that:  2 v  2 v 1      2 v  2 u  → Equation 15b


    
x 2 y 2 2  x 2 xy 
G. Displacement Natural Boundary Conditions
From Eq. 8a and 8b: psx  l xx  m xy

E xy
2  xx
   yy   m
E
psx  l
1  2 1   

Using Eq. 9a, 9b and 9c: psx  l E 2  u   v   m E  u  v  → Equation 16a


1    x y  2 1     y x 
E  u v  E  v u  →
It follows: psy  l   m 2 
  Equation 16b
2 1     y x  1    y x 
The equations of equilibrium (Eq. 1a or 1b) including with the boundary conditions (Eq. 8a and 8b)
and a compatibility equation (Eq. 11a or 13b) will produce a system of equations which is generally
enough for the determination of the stress distribution in a two-dimensional problem.
To summarize, the following systems of equations used to solve the stress distribution are as
follows:
1. Equations of Equilibrium:
 xx  yx
  fbx  0
x y

Or  xx  yx (Neglecting body force)


 0
x y
 yy  xy
  fby  0
y x
 yy  xy
Or  0 (Neglecting body force)
y x
2. Boundary Conditions:
psx  l xx  m xy and psy  l xy  m yy
3. Compatibility Equations:
 2 xy 2 2
2 1    
xy
 
y 2 
  xx   
yy 
   yy   xx  
x 2  
 2 2   fbx fby 
Or  2  2   xx   yy    1       
 x y   x y 
 2 2 
Or  2  2   xx   yy   0 (Neglecting body force)
 x y 

26
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
Or  2  xx   yy   0 (Neglecting body force)

2. SPECIFIC ANALYTICAL SOLUTION WITH DERIVATIONS


a) Support Reaction
Solving for the weights: Ry
W   V   gV and V  Aplanet
x
For Segment AB: A
WAB
WAB   gAABt
7.5 m
 w  2 w  L  P
WAB   g   t
 2  2 
3 B
WAB   gLwt
4
3 kg  m WBC
WAB   7800 3  9.81 2  15 m 1.2 m  0.1 m  7.5 m
4 m  s 
WAB  103, 299.3 N
C
For Segment BC:
WBC   gABC t 1.2 m
 L 1
WBC   g  w  t   gLwt y
 2  2
Fig. 1.2 Support Reaction
1 kg   m
WBC   7800 3   9.81 2  15 m 1.2 m  0.1 m 
2 m  s 
WBC  68, 866.2 N
Applying Equilibrium of forces:
 Fv  0  
Ry  P  WAB  WBC
Ry  10,172,165.5 N or 1.0172 x107 N
b) Displacement
Segment AB (0 ≤ y ≤ 7.5 m) Relative to A:
Ry
2.4 m
Ry

x
1 A 1 y
dy
7.5 m P W1(y)
7.5-y x x y

B
2 2 0.6 m 1.2 m 0.6 m
σ(y)
7.5 m

SECTION 1-1
C
1.2 m
y

27
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
Weight W1(y): Solving for x:
x  0.6 0.6
W1  y    gA  y  t 
7.5  y 7.5
 2 x  2 w   2
W1  y    g   y t x  7.5  y   0.6
 2   25
W1  y    g  x  w  t  y x  1.2  0.08 y
 kg  m
W1  y    7800 3  9.81 2  1.2  0.08 y  1.2  0.1 m  y
 m  s 
W1  y   y 18364.32  612.144 y 
W1  y   18364.32 y  612.144 y 2
Applying Equilibrium of forces: Applying Hooke’s Law:
 Fv  0     E

  y  A x   W1  y   Ry Strain-Deformation Relation:
 d
  y  A x   Ry  W1  y   and 
L dy
 d 
  y    E
 dy 
By substitution/Differential Equation: Cross-sectional Area:
 d  A  x   2 xt  2 1.2  0.08 y  0.1 m  106 
  E A  x   Ry  W1  y 
 dy  A  x   240,000  16,000 y
 Ry  W1  y  
d   dy
E A x
Displacement Function:

1.0172 x107 N  18364.32 y  612.144 y 2  
d    dy
200,000  240,000  16,000 y 
Imposing Boundary Condition: δ = 0 at y = 0

y 1.0172 x107 N  18364.32 y  612.144 y 2 

  0   d     dy
0 200,000  240,000  16,000 y 
 15  y 
  y   0.002869425 y  3.13576034375ln    0.0000956475 y
2

 15 
Deformation at B (At y = 7.5 m):
 15  7.5 
  7.5   0.002869425  7.5  3.13576034375ln    0.0000956475  7.5 
2

 15 
δ = 2.19 mm
Segment BC (7.5 ≤ y ≤ 15 m) Relative to B:
Weight W2(y):
W2  y    gA y  t
W2  y    g 15  y  w t

28
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members

Ry
2.4 m
2.4 m

x
1 A 1

y
7.5 m P σ(y)
y-7.5
B dy
15-y W2(y)
15-y
7.5 m
1.2 m
2 C 2

1.2 m

y SECTION 2-2
 kg  m
W2  y    7800 3  9.81 2  1.2 m  0.1 m 15  y 
 m  s 
W2  y   9182.16 15  y 
W2  y   137732.4  9182.16 y
Applying summation of forces: Applying Hooke’s Law:
 Fv  0     E
  y  A x   W2  y  Strain-Deformation Relation:
 d
 and 
L dy
 d 
  y    E
 dy 
By substitution/Differential Equation: Cross-sectional Area:
 d  A  x   wt  1.2 m  0.1 m  106 
  E A  x   W2  y 
 dy  A  x   120,000 mm2
W2  y 
d  dy
E A x
Relative Displacement Function:
137732.4  9182.16 y
d   1000  dy
200,000 120,000
Imposing Boundary Condition: δ = 0 at y = 7.5
y

   d     0.00573885  0.00038259 y  dy
7.5

29
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
  y   0.00573885 y  0.000191295 y 2 7.5
y

  y   0.00573885 y  0.000191295 y 2  0.00573885  7.5   0.000191295  7.5  


2
 
  y   0.03228103125  0.00573885 y  0.000191295 y 2
Deformation at B (At y = 7.5 m):
  7.5  0.03228103125  0.00573885  7.5   0.000191295  7.5
2

δ=0
Deformation at C (At y = 15 m):
 15  0.03228103125  0.00573885 15   0.000191295 15 
2

δ = 0.01076 mm
Summary:
Displacement Function:
 15  y 
 AB / A  y   0.002869425 y  3.13576034375ln    0.0000956475 y
2

 15 
 B / A  2.18968395681 mm
 BC / B  y   0.03228103125  0.00573885 y  0.000191295 y 2
  AB  y  0  y  7.5 m
  y  
 B / A   BC  y  7.5 m  y  15 m
  15  y 
0.002869425 y  3.13576034375ln    0.0000956475 y 2 0  y  7.5 m
  y    15 
 2.1574029255+0.00573885 y  0.000191295 y 2 7.5 m  y  15 m

Deformation at B (At y = 7.5 m):
  7.5  2.1574029255+0.00573885 7.5  0.000191295 7.5
2

δ = 2.1897 mm
Deformation at C (At y = 15 m):
 15  2.1574029255+0.00573885 15   0.000191295 15 
2

δ = 2.2004 mm
3. DERIVATIONS FOR THE THREE-NODE ELEMENT
Quadratic Shape Function
ω(x)
ω(x)=a0 + a1x + a2x2

d1
d3
d2

x
x1 x2 x3
Element # 1

30
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
DERIVATION:
1. Quadratic Displacement function is:  ( x)  a0  a1 x  a2 x 2
Can be also express as:  ( x)  N1 ( x)d x1  N 2 ( x)d x 2  N3 ( x)d x 3
 d x1 
In Compact form:  ( x)   N1 ( x) N 2 ( x) N 3 ( x)   d x 2 
 d x 3 
N1 (x) N 2 (x) N 3 (x)

x1 x2 x3 x
Element # 1
At Node 1: x = x1 and ω(x) = dx1, thus → d x1  a0  a1 x1  a2 x12 → Equation 1
At Node 2: x = x2 and ω(x) = dx2, thus → d x 2  a0  a1 x2  a2 x2 2
→ Equation 2
At Node 3: x = x3 and ω(x) = dx3, thus → d x 3  a0  a1 x3  a2 x3 2
→ Equation 3
Note: The unknowns are a0, a1 and a2.
In Matrix Form:
1
 d x1  1 x1 x12   a0   a0  1 x1 x12   d x1 
     and solving for the unknowns →  a1   1 x2   
d x 2   1 x2 x2 2   a1   x2 2  d x 2 
 d  1 x x32    a  1 x x32  d 
 x3   3 a2   2  3  x3 
1 x1 x12 
  Adjoint of  A
Let: A  1 x2 x2 2  and the inverse of Matrix A → A1 
det  A
1 x3 x32 

Adjoint of  A  CT of  A
 x2 x32  x3 x2 2   x32  x2 2  x3  x2 
 
Cofactor of Matrix A, C     x x 2  x x 2  x 2  x 2   x3  x1  
 
A 1 3 3 1 3 1

 x1 x2 2  x2 x12
   x 2
2
 x 1
2
 x 2  x1 

 x2 x32  x3 x2 2   x1 x32  x3 x12  x1 x2 2  x2 x12 


 
Adjoint of  A     x32  x2 2  x32  x12   x2 2  x12  
 
 x3  x2   x3  x1  x2  x1 
 
1 x1 x12 1 x1 
 
Determinant of  A  1 x2 x2 2 1 x2   x2 x32  x1 x2 2  x12 x3   x2 x12  x3 x2 2  x1 x32 
1 x3 x32 1 x3 

=  x2 x3  x2 x1  x12  x1 x3   x3  x2 

Determinant of  A   x2  x1  x3  x1  x3  x2 

31
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
 x2 x32  x3 x2 2   x1 x32  x3 x12  x1 x2 2  x2 x12 
 a0     d x1 
  1  2  
 1
a 
 A  
  x3
2
 x 2
2
 x3
2
 x1
2
  x 2
2
 x1  
d x 2 
a   
 2  x3  x2   x3  x1  x2  x1   d x 3 
 
 d x1  x2 x32  x3 x2 2   d x 2  x1 x32  x3 x12   d x 3  x1 x2 2  x2 x12  
 
  x2  x1  x3  x1  x3  x2  
 a0   
   d x1  x3  x2   d x 2  x3  x1   d x 3  x2  x1 
2 2 2 2 2 2

 a1    
a    x2  x1  x 3  x1  x 3  x2  
 2 
d x1  x3  x2   d x 2  x3  x1   d x 3  x2  x1  
 
  x2  x1  x3  x1  x3  x2  
 
d x1  x2 x32  x3 x2 2   d x 2  x1 x3 2  x3 x12   d x 3  x1 x2 2  x2 x12 
a0 
 x2  x1  x3  x1  x3  x2 
 d x1  x32  x2 2   d x 2  x32  x12   d x 3  x2 2  x12 
a1 
 x2  x1  x3  x1  x3  x2 
d  x  x   d x 2  x3  x1   d x 3  x2  x1 
a2  x1 3 2
 x2  x1  x3  x1  x3  x2 
d x1  x x  x x    x  x  x   x  x  x   d  x  x  x   x x
2 2 2 2 2 2 2 2
 x3 x12    x3  x1  x 2   d x3  x1 x2 2  x2 x12    x2 2  x12  x   x2  x1  x 2 
 ( x)   2 3 3 2 3 2   3 2 x2 3 1 1 3

 x2  x1  x3  x1  x3  x2 
 x2 x3  x3 x2    x3  x2  x   x3  x2  x   x3  x2   x2 x3   x3  x2  x  x 2   x2 x3   x3  x2  x  x 2 
2 2 2 2 2

N1 ( x)      
 x2  x1  x3  x1  x3  x2   x2  x1  x3  x1  x3  x2   x2  x1  x3  x1 
x2 x3  x3 x  x2 x  x 2 x3  x2  x   x  x2  x   x2  x  x3  x 
N1 ( x)   → N1 ( x) 
 x2  x1  x3  x1   x2  x1  x3  x1   x2  x1  x3  x1 
 x32  x12  x   x1 x32  x3 x12    x3  x1  x 2   x3  x1   x3  x1  x  x1 x3  x 2   x3  x1  x  x1 x3  x 2 
N 2 ( x)      
 x2  x1  x3  x1  x3  x2   x2  x1  x3  x1  x3  x2   x2  x1  x3  x2 
x3 x  x1 x  x1 x3  x 2 x3  x  x1   x  x  x1   x3  x  x  x1 
N 2 ( x)   → N 2 ( x) 
 x2  x1  x3  x2   x2  x1  x3  x2   x2  x1  x3  x2 
 x1 x2 2  x2 x12    x2 2  x12  x   x2  x1  x 2   x2  x1   x1 x2   x2  x1  x  x 2   x1 x2  x2 x  x1 x  x 2 
N 3 ( x)      
 x2  x1  x3  x1  x3  x2   x2  x1  x3  x1  x3  x2   x3  x1  x3  x2 
 x1 x2  x2 x  x1 x  x 2  x2  x1  x   x  x1  x   x2  x  x1  x 
N 3 ( x)    → N 3 ( x) 
 x3  x1  x3  x2   x3  x1  x3  x2   x3  x1  x3  x2 

CHANGE OF VARIABLES:
By Ratio & Proportion:
 1  L
 and let x2 = L/2, 
1 → x 1   
x  x2 x3  x2 L L 2
x
2 2
Thus the shape functions can be express as:
Set every value of x equal to its equivalent ξ value,
 x  x  x3  x   0   1    2 
N1 ( x)  2 → N1 ( )  → N1 ( ) 
 x2  x1  x3  x1   0  11  1 2

32
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members

ω(x) ω(x)=a0 + a1x + a2x2

d1
d3
d2

x
x1 x2 x3
Actual Element
ω(x)
2
ω(x)=a0 + a1x + a2x

d1
d3
d2

-1 0 +1
Using Change of Variable, ξ
ξ(x)
x +1

0 ξ
x
x1 x2 x3

-1
L
Applying Linear Interpolation

N 2 ( x) 
 x3  x  x  x1  → N 2 ( ) 
1     1 → N 2 ( )  1   2
 x2  x1  x3  x2   0  11  0 
 x2  x  x1  x  → N 2 ( ) 
 0    1    → N3 ( ) 
2 
N 3 ( x) 
 x3  x1  x3  x2  1  11  0  2

1. Strain and B matrix


 d x1 
d
 d x 2 
→ d
 ( x) 
dx
 ( x)   1
N ( x ) N 2 ( x ) N 3 ( x )
dx
 d x 3 
 d x1   d x1 
 ( x)   N1 ( x) N 2 ( x) N 3 ( x)   d x 2  N 3 ( x)   d x 2 
d d
→  ( x)   N1 ( x) N 2 ( x)
dx dx
 d x 3   d x 3 

33
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
d 
  2 x  ( x2  x3 )   2 x  ( x1  x3 )   2 x  ( x1  x2 )    x1 
 ( x)        d x 2 
   x2  x1  x3  x1     x1  x2  x3  x2     x3  x2  x3  x1     
d x3 

  2 x  ( x2  x3 )   2 x  ( x1  x3 )   2 x  ( x1  x2 )   
B       
   x2  x1  x3  x1     x1  x2  x3  x2     x3  x2  x3  x1   
 
 d 1 
d d d     d 
Or  ( x)   →  ( x)    N1 ( ) N 2 ( ) N3 ( )   d 2   
d  dx  d     dx 
 d 3 
 d 1 
 dN ( ) dN 2 ( ) dN 2 ( )     d 
 ( x)   1 d 2  
 d d d     dx 
 d 3 
 
d L d 2
1   
L L 
Since, x 1    → dx   → dx  d →
2 d  2  2 dx L
 d 1 
 2   2  1 1  2     2   2  1 1  2 
 ( x)     2 B    2
2 
d 2 and Let
 L  2 2     L  2
 d 3 
 
Elemental Stiffness Matrix
x3

k   BT EB  Adx  where:
x1

  2 x  ( x2  x3 ) 
2
 2 x  ( x2  x3 )   2 x  ( x1  x3 )   2 x  ( x2  x3 )   2 x  ( x1  x2 )  
        
   x2  x1  x3  x1     x2  x1  x3  x1     x1  x2  x3  x2     x2  x1  x3  x1     x3  x2  x3  x1   
 
  2 x  ( x2  x3 )   2 x  ( x1  x3 )   2 x  ( x1  x3 )   2 x  ( x1  x2 )  
2
 2 x  ( x1  x3 ) 
B B  
T
      
   x2  x1  x3  x1     x1  x2  x3  x2     x1  x2  x3  x2     x1  x2  x3  x2     x3  x2  x3  x1   
 2 
  2 x  ( x2  x3 )   2 x  ( x1  x2 )   2 x  ( x1  x3 )   2 x  ( x1  x2 )   2 x  ( x1  x2 )  
 x  x x  x   x  x x  x   x  x x  x   x  x x  x    
   2 1  3 1     3 2  3 1     1 2  3 2     3 2  3 1     x3  x2  x3  x1   
Or:
x 1 1 1
 L  AEL AEL

3

k   BT EB  Adx   E  BT B  A d    k BT Bd 
x1 1  2  2 
1
BT Bd  →
2 1

 2  1 
 2 
 2   2  1 1  2   2  
B    2 and → B     2 
2 
T

 L  2  L 
1  2 
 
 2 
  2  1     2  12 4 2  1 
      2  1 
 2   2    2   2  1 1  2   4  4 4 
B B     2    
T
2  → BT B  2    2  1 4 2  1  2  
 L  1  2    L   2 2   L  
 2  1 
2
    4 2  1
  2     1  2  
 4 4 

34
Effectiveness of Finite Element Method in the Deformation Analysis of 1-D Elasticity Members
 k11 k12 k13 
Elemental Stiffness: k   k21 k22 k23 
 k31 k32 k33 
Assuming a uniform cross-section throughout the element:
1
1   2  12  AE  1  2  1 
3
7 AE
AEL 4 AE 
 2  1   2  1  k11 
2 1 L2
k11   d   
3 3
  →
 4  2 L  2 3  1 12 L  3L

1
1
AE   2 2 3  AE  1 2   1 2   8 AE
k12 
AEL 4

2 1 L2    2 2
 d    
2L  2
      
3  1 2 L  2 3   2 3  
→ k12 
3L
1 1
AEL 4  4 2  1  AE  4 3  AE  4   4   k13 
AE
k13   2 
2 1 L  4 
 d   
2L  3
     1      1 
 1 2 L  3   3  

3L
1
1
AE   2 2 3  AE  1 2   1 2   8 AE
k21 
AEL 4
2 1 L2
   2 2
 d    
2L  2
      
3  1 2 L  2 3   2 3  
→ k21 
3L
1 1
8 AE   3   8 AE  1 1  16 AE
2  
AEL 4 k22 
k22           →
2
4 d
2 1 L L  3  1 L  3 3  3L
1
1
AE   2 2 3  AE  1 2   1 2   8 AE
k23 
AEL 4

2 1 L2    2 2
 d    
2L  2
      
3  1 2 L  2 3   2 3  
→ k23 
3L
1 1
AEL 4  4 2  1  AE  4 3  AE  4   4   k31 
AE
k31   2 
2 1 L  4 
 d   
2L  3
     1      1 
 1 2 L  3   3  

3L
1
1
AE   2 2 3  AE  1 2   1 2   8 AE
k32 
AEL 4

2 1 L2    2 2
 d    
2L  2
      
3  1 2 L  2 3   2 3  
→ k32 
3L
1
1   2  12  AE  1  2  1 
3
7 AE
AEL 4 AE 
 2  1   2  1  → k33 
2 1 L2
k33   d   
3 3
 
 4  2 L  2 3  1 12 L  3L

 7 8 1 
AE 
Elemental Stiffness: k  8 16 8
3L
 1 8 7 

35

Vous aimerez peut-être aussi