Académique Documents
Professionnel Documents
Culture Documents
1968
Recommended Citation
Mohta, Brijgopal R., "Evaluation of methods for analysis of multi-degree-of-freedom systems with damping" (1968). Masters Theses.
5272.
http://scholarsmine.mst.edu/masters_theses/5272
This Thesis - Open Access is brought to you for free and open access by Scholars' Mine. It has been accepted for inclusion in Masters Theses by an
authorized administrator of Scholars' Mine. This work is protected by U. S. Copyright Law. Unauthorized use including reproduction for redistribution
requires the permission of the copyright holder. For more information, please contact scholarsmine@mst.edu.
EVALUATION OF METHODS FOR ANALYSIS OF
MULTI-DEGREE-OF-FREEDOM SYSTEMS
HITH DAMPING
BY
BRIJ. R. HOHTA 1
\ C,qL
THESIS
Degree of
Rolla, Missouri
1968
_ Approved by
~ (advisor)
. })yf 111.
ii
ABSTRACT
Five methods are discussed. These are: (1) Normal Mode Technique
(2) Holzer's Method (3) Impedance Method (4) Graphical Technique (5) A
by K. A. Foss has been employed. This method is quite complex, but does
for both undamped and damped systems are derived. A sample table is
dampers between masses as well as between the masses and ground have been
Method, the four-pole parameters of a mass, spring and damper are derived
and the formulas for solving tandem and parallel connections are pre-
illustrate the use of this method. The results obtained have been com-
and the results are compared. Suggestions for further work are made.
iv
ACKNOwLEDGEMENTS
TABLE OF CONTENTS
page
V. HOLZER' S METHOD 23
Undamped System .. 24
Selection of Trial Frequencies
Forced Vibrations
.. 27
29
Branched Systems 30
Damped Systems 35
VITA . . . . . . 129
vii
LIST.OF FIGURES
- ------
Figure
Foundation Motion 55
Foundation Motion 97
Figure Page
LIST OF SY11BOLS
= kinetic energy
u = potential energy
v = dissipation function
p = virtual work
w = circular frequency
Mi = mass of ith element
[N) =
[r] = identity matrix
= a constant
a
ne
Ji :::; mass moment of inertia of ith disk
w
0
= selected value of w within frequency range of interest
S = phase angle
1
I. INTRODUCTION
with only basic knowledge of the subject of vibrations often have diffi-
culty in understanding and applying these ideas, many of which are pre-
these ideas together and to describe, evaluate, and compare them in such
if the degrees of freedom are not more than three, is to obtain the
=-
F
n
damp . cnn xn
For a less accurate solution, the damping forces associated with the
Fl cl 0 0 0 .
nl
F2 0 c2 0 0 n2
0 0 c3 0
=
F
n
damp 0 cn
Note that the off-diagonal, or coupling terms have been assumed negligibly
multi-mass system.
presented.
3
(1 2 3 4)*
Generally textbooks ' ' ' very thoroughly cover the analysis
(2) where the author has derived the equations of motion for a damped
and demonstrate the use of Holzer's method, which is one of the most
mechanical systems.
5
order:
2. Holzer's Method
3. Impedance Method
4. A Graphical Technique
undamped systems will be considered first and then damping will be intro-
analysis.
scribe it.
more coordinates for their description than there are degrees of freedom.
(1)
7
is explained(l 4 ) as follows:
in each case. Nmv imagine that, at a given instant, the system is given
and is known as the virtual work. TI1e small displacements are called
virtual displacements because they are imaginary in the sense that they
are assumed to occur without the passage of time, the applied forces
remaining constant.
for all i when the system is in stable equilibrium. For such a system,
n
1 ~ . 2
T = 2 I.. mixi (2)
i=l
(3)
8
Mttz.
( (.-1, L
Ci+t
Figure 1
n
1
V=2 I
i=l
(4)
oP
oxi =
Let i =j =2
Since K
12
= K
21
1 2 1 2
U= 2 Kllxl + Kl2xlx2 + 2 K22x2
au
-=
au
--= Kllxl + K22x2 + Kl2xl + K12x2
2
= I Kijxj for i = 1,2
j=l
n
av
ax1 = ) c1 jxj
J=l
Substituting the above expressions in equation (1), the equations of
i = 1,2, . , n
(5)
where
0 0
Mass Matrix
0
0 0 M
n
Spring Matdx
K
nn
Damping Matrix
[c] =
cnl .. cnn
{x}
xll
= x2
..
a column vector of
order Nxl
X
n
and Fl
{F} = F2 a column vector of
order Nxl
F
n
matrices. This follows from the fact that the system obeys the reci-
Undamped Systems:
undamped system.
= {q} iwt
{x} e (9)
where {q} is a column vector of order Nxl, the elements of which are
or
(10)
(11)
Since both [M] and (K) are symmetric and positive definite, the roots
of this equation are all real and positive(lS). Neglecting for the
i
vector {q } which satisfies the following equation:
Since [M] and [K] are symmetric and positive definite, there exists
a transformation(lB)(Q]such that
where Q is known as the 11odal Hatrix, the columns of which are the
T
Premu1tip1ying equation (13) by fQ] ,
or
g (t)
n
14
(15)
pi
ni = Acoswit + Bsinwit + K _ M w2 coswt
ii ii
xl nl
x2 n2
= ~1 q
2
.q_j
X
n
1 2 n
where the column vectors q , q , , q are the eigenvectors of the
system.
system can be solved by the Normal Mode Technique provided the roots
of the frequency equation are distinct. This method has been presented
in great detail in reference (2) and the author has solved several
representative problems.
For the case of repeated roots, see Appendix (B) where the pos-
Damped Systems:
(a) Classically Damped Systems are those systems in which the matrix
{b) Non-classically Damped Systems are those where [CJ cannot be dia-
damped system.
is a diagonal matrix
motion.
Mi.n.
~ ~
+c~~
.. rii + K~~
.. n.~ = gi(t)
where gi = Pi coswt
c and c
1 2 are constants of integration to be determined by initial
conditions.
a solution of type
{X}= [Q]{n}
formation which uncouples the undamped system will also uncouple the
2
damped system. Dr. T. K. Caughey(l ) derived sufficient conditions for
necessary conditions for uncoupling the damping matrix have not yet been
developed.
17
{X} = [N]{L} and reducing the equations of motion to the following form
where 1
(A] = (N)- [c)(N)
and
(B)= (Nll (K)(N)
It is then proved that if diagonal matrices (a] and [b1, obtained from
as (a] = (b]t where t is an integer then [A]= (B] t Thus, if [A1 = [B]t,
a method for solving some of these systems. In this method the original
Method of K. A. Foss
{Z} = r~ {x}
{x} {
l
{0} l
. {F} = { {F(t) I
with the equations of motion. With the above definitions the equations
fTo1 [M)l
lfMJ [cJJ
{on} [-(M]
{x}
+
[o)
(o~
(Kn
f {x} {
1{x}! =
f {O} l
1{f(t)} J
or [M]{*f - lM] jxJ = {o}
equation
lR){Z} + (PJ{Zf ={Of (19)
Assume fz}= {{~~} = e o<t{~f= eo(t[1~J (20)
on substituting equation (20) in equation (19) and rearranging
J
It is easily shown that
-1
(P J -1 - [M]
(OJ
[ [O] [Kr 1
(23)
symmetric only if
- tl] and
(as shown in Appendix B) and Foss's method will give a solution. However,
will always exist only if there are 2N distinct roots of the frequency
equation in 2N space
1
For each distinct root --- of the above equation there exists an independent
vCi
20
. i
vector {~ }. As shown in Appendix B, a root of multiplicity K may or
values of the system are obtained by solving equation (24). Equation (23)
i . .
then gives the eigenvectors ~i} in 2N space. Having found {! }, {xi}
Forced Vibrations
obtained by letting
{z J = [o]{~}
(25)
Since
S{x}
{z}::[ {x}
J
, {x} can be obtained by expanding. {Z}.
21
it is assumed that
{Z} = f*l }=
{x}
e<><tq_} = e-<t{~~ f (26)
o(,t .
{x} = e {a<~}, and (27)
It is shown below that for critically damped systems, equation (27) does
not apply.
equation
\vhere , and
Here
and
22
(29)
The time derivative of the above equation cannot be expressed in the form
of equation (28) and therefore Foss's method does not give a solution
for critically damped systems. Summarizing, Foss's metho~ does not work
(1) where eigenvalues are repeated and the eigenvectors do not form a
(2) when the system is critically damped at one or more of its modes of
free vibration.
23
V. HOLZER'S METHOD
method also gives the amplitude ratios of the masses and the nodes in
Undamped System:
are
3 8
1 1
+ K12 <a 1 - a 2> = 0 (30)
..
J2a2 - K12 <e 1 - a 2 > + K23 <8 2 - 8 3 ) = 0 (31)
3 8
3 3 - K2 3 <a 2 - a 3 > = 0 (32)
(33)
Hence the sum of the inertia torques must equal zero at all times.
el =A coswt
82 = B coswt
83 =c coswt
or
- -
J", s~
Figure 2
or
J 2 Bw2 + J Aw2
1
C = B -
K23
is presented below:
Trial Frequency w
~-------------------------------------------------------------------------~
2
Station Hass Moment Jw Amplitude Inertia Spring
Number of Inertia a Torque constant
J J,.,43 K ..
i,~-1
Holzer Table
placed by M and e by X.
follows:
Column 5: gives the value of the shaft torque beyond the disk in
question;
The magnitude and sign of remainder torque are a measure of how far the
(1) in estimating the initial trial value for w; and (2) in selecting
a second trial value for w if the initial trial value does not satisfy
equation (33).
28
torque for the first natural frequency indicates that the system has
the mode of the vibration is usually taken to be the same as the number
of nodes * , which means that the first natural frequency has one node,
the second two, and so on. Each time a node is passed, the sign in the
column indicates the number of natural frequencies which lie below the
trial value. If there are many sign changes in the amplitude column for
the first trial, a much lower value of w should be selected for the next
trial.
the inertia torque on the last disk changes sign as the frequency is
then inertia torque is taken zero and e1 has some arbitrary value.
Once the eigenvalues have been found, the amplitude column of the
Holzer table directly gives the eigenvectors for each of these eigen-
values.
Forced Vibrations:
quired to obtain the angular displacements of all disks which result due
J 8 +
1 1
K (e -
12 1
e 2 ) = T0 coswt (35)
As before, let
e1 = A coswt
e2 = B coswt (38)
e3 = C coswt
30
- T0 (39)
B =A -
Branched S~stems:
tables have to be made for each branch of the system. The application
the primary branch with a unit angular deflection at the left-hand end.
c
Js
JT
~
J6
Ks-t> 8-- ~67 1178
~
1.-
Figure 3
the deflection at section C-C is already known, the two deflections are
are added. Calculations are then carried for branch 3 starting from
section C-C with inertia torque equal to the sum of inertia torques of
of the entire system, it will meet the boundary conditions at the end
trial frequency.
Use of this method requires that all branches of the system must
n is the speed ratio of the geared shaft to the reference shaft. This
1 . 2 1 2 . 2
T = 2 J 28 2 = 2 n J 26 1
2
and the equivalent inertia of disk 2 referred to shaft 1 is n J
2
(continued)
corresponding to twist of n is
1
is
1 . 2
v = 2 c2e 2
Since 82 :::: nel
1 2 2
v = (c n )e
2 2 1
Though the damping has not yet been introduced in the system, for
t
n,J7 2.
n, Ja
2
,, /{67
J,
Branch 2
Branch 1 2..
n2.J!>
Branch 3
nl = S:2eed of branch 2
Speed of branch 1
n2 S:2eed of branch 3
Speed of branch 1
Figure 4
Damped Systems:
phase by 180 degrees with the disturbing torque and no energy is dis-
disk will in general be out of phase by an angle other than 180 degrees.
The basic Holzer table equations for a general case may be derived
with the aid of figure 5. This includes viscous damping between the
Cij' from the viscous damping between the disks and the fixed member,
which is designated as hi
.. . .
=0 (40)
Jl 9 1 + h191 + K12 ( 9 1 - 92) + C12<91 - 92)
.
J2 9 2 + h292 - K12( 9 1 - 9 2 ) - c12 <9 1 - 9 2 )
. .
J3 9 3 + h393 - K23( 9 2 - 93) - C23< 9 2 - 93) =0 (42)
9i = Aiejwt
where j = v:T
36
h~
Figure 5
Substituting for et and its derivatives into equations (40), (41), and
2
-J 3w A3 + jwh A3 + (K + jwc 23 )(A - A ) = 0 (45)
3 23 3 2
or
(46)
(47)
(48)
38
disk i,
(49)
Equation (49) is a general form of the Holzer equation for damped systems
various trial frequencies has a minimum where the real part of the
out any external torque at its natural frequencies, a damped system re-
This explains why the minimum point corresponds to one of the natural
method(S) makes use of this fact and describes the steady state system
behavior.
Fl Elastic F2 + Direction
(1) v (2)
vl System 2
Figure 6
(1) and (2) which are called the input and output points respect-
system which preceeds it. At the output point (2) there exists a
force F
2
and a velocity v2 which result from the application of F
1
41
and v at point (1) and the reaction of the portion of the passive
1
mechanical system between point (1) and point (2).
(50)
(51)
body,
(52)
(53)
where FA' FB, etc., may be complex functions of mass, stiffness, resis-
tance, and frequency, but are time independent. Note that no phase
iwt
angles are necessary in thee term since the FA' etc., terms are
42
v = VA e iwt
iwt VA iwt =v-
X = !VA e dt =iw
- e
iw
..X =d-(V) = iw VA e
iwt
= iwv
dt
(55)
(56)
Comparing equations (55) and (56) with equations (50) and (51), the
input point of the spring is the same as the force which the spring
(57)
Also, (58)
Vl = iw F + V
K 2 2
Fl = F2 (59)
and
F2 = C(Vl - v2 ) (60)
F
1
= F2 + o.v2
1
vl = c F2 + v2
and the four-pole parameters are A = 1, A = O, A 1
= C'
11 12 21 and A22 = 1.
tandem connection when the output from the first is precisely the input
(61)
where the subscripts signify that the parameters belong to the four-
pole network numbered (1).
replacing
[~Din equation (61) by its value from (62),
four-pole network and that any changes made in that network affect only
so that
45
(1) all their input and output junctions move with the same velocity;
(2) the input force to the composite four-pole network is the sum of
(3) the output force from the composite four-pole network is the sum
c
Figure 7
Two Four-pole Elements Connected in Parallel
1 c
= B' and A22 = B
where
i=n A R,
A = r
i=l
<A)
A21
i=n
B = r
i=l
1
(~)
21
i=n A R,
c = r <A)
A21
i=l
With the use of above equations, the four-pole parameters of the com-
bination spring and damper system shown in figure 7 can be found as follows:
A A
A = (___!_!_) 1 + ( 11)2
A21 A21
.K
= -.-
J.W
+c
B = (_!_)1 + (_!_)2
A21 A21
K
= -
iw
+ c
A A
c = (__E._) 1 + (~)2
A21 A21
K
= -.-
J.W
+c
which gives,
AC
=--B=O
B
A =-=1
c
22 B
points (1) and (2), the composite system is reduced to a single element
are now used in tandem connections and the system can be simplified by
6
This technique( ) for determining the frequency response of damped
or computer methods.
springs and dampers are described by the spring constants Kj and the
damping constants Cj. The masses have linear motion only. The harmonic
MX
0 0
= Cl(Xl - x>+
0
Kl(Xl- Xo) (64)
.
MjXj = cj+l(xj+l - Xj ) + Kj+l (xj+l - xj) - cj <Xj - X.J- 1)-K.J (X.-Xj 1)
J -
+{f~ @j =
@ j "I
u
u
(65)
MX = - C (X
. -X )- K (X - X ) (66)
nn n n n-1 n n n-1
* 2
Sinee t h e exc i tat i on i s assume d t o b e h armon i c , -w xJ an d i wCj can b e
Figure 8
(67)
2
-(iCjw+Kj) Xj-l +(-Mjw +iCjw+Kj + iCj+lw+Kj+l) Xj
2
(-M w +iC w+K ) X -(iC w+K ) X = 0 (69)
n n n n n n n-1
w
iwCj+Kj l+i2bj
Woj
Tdj = -M.w2+iwC.+K. = (70)
J J J l-(~) 2 +i2b ..!:'!_
woj j woj
2 -(~)2
-M w w.
j 0
-MjwZ+iC.w+K. =1-(3_.) 2+i2b
Tf. (71)
J w
J J w . j w
OJ oj
where
w. = (~)~
OJ
~
cj
bj = 2(MjKj)~
Tf j = l-Tdj (72)
50
(73)
(74)
(75)
(M.w2 ~ X j -1 + ( - M.w2
1 -
M
j+lw
2 Tdl.+l) X
- j (76)
J Tf Tfj Tfj+l
J
Td j=U
2 1-.+1)- Xj+l
- ( - Mj+1w
Tfj+l
2
2 Td X (-Mnw y O (77)
- ( -~w _!l) n-1 + ) ~"'Il =
f
T n Tf n
Dividing equations (75), (76), and (77) through by - w2, the following
equations are obtained:
j =u (79)
j 1: u
51
(80)
(81)
or
X I, M Tf 1] - l
. X~ =L + M~ Tdl
j < u
j = u
and j > u
(82)
Ml Td 2 Td 2
+ (Tf! + M2 Tf/ Xl -(M2 Tf ) Xz =O
2
(83)
Equation (83) gives
52
rearranging,
X
M Td1
!-
substituting~ for .... Tfl h b
t +
M1.'l'f2. 1
M Tf Td
2 1 2
(1 (84)
X
_j.=!_ =~
1 +
M Tf -:-=1=--- (1 - Td . -l X
_j.=!_ -=.:::.J_ ) Xj-2~- 1 (86)
X Mj Tdj Tfj-l J j-1
j
f
u
- w2
(87)
M X w
u f~
2
= _
[
TTdfuu
X
u-1 + _!_ +
Xu Tf u
M
u+l
Td
u+l Mu+l Tdu+1 Xu+l
Mu Tfu +l - ~ Tfu+1 x:- J - 1
(88)
For _j_~: ~-!rite equation (79) as
53
(90)
Xj+l = Td
Xj j+l
[1 + Mj+2
Mj+l
Tdj+2
Tfj+Z Tfj+l
(1 Xj+2)
- Xj+l
J
-1 (91)
X
n
- - = Td (92)
xn-1 n
Equations (81), (86), (88), (91), and (92) are used for determining the
the relation
~
M.
= _J_ (93)
M X
u u
~ = (94)
X.J- 2
If the external forces are applied at more than one location, then there
cations can be made. If the input is applied to an end mass, begin the
analysis at the opposite end and progress toward the input. If the
force is applied at each end, the analysis is carried out by taking into
account one force at a time and beginning the analysis at the opposite
position.
For the case where the first mass is attached to a rigid foundation,
xl
M is assumed to be infinite. Then X0 vanishes and the ratio - i s
0 x2
available as a function of the parameters describing the first and second
elements.
the nodal mass is simply a function of the sum of the branch transmis-
sibilities -the bracketed term in equations (81), (86), (88), (91), and
Let n = 3, which is the number of masses in the longest chain, and let
j = 0. Then from equation (92),
x3
-= Td (95)
x2 3
x4
- = Td (96)
xl 4
55
J:l
J,;o
7/ll/ll//lll/lll///ll/l//l//
Figure 9
(97)
noting Tf
4
= l-Td 4 , the above equation has the final form
Xl
--
X0
= Td
1
~1 M2 Td2
+- -
Ml Tf 2
X2 M4
Tf (1- - ) + - Td Tf
1 x1 Ml 4
J -1
(98)
. x2
Using equations (97) and (98) g~ves X:'
0
(99)
(100)
xo x2
and from equations (96) and (98),
x4 x4 xl
-=-. (101)
xo xl X
0
Equat:tons (98) through {101) give the ratio between the amplitude at
Note that the rotational motion has been neglected and the masses
(81), (86), (88), (91), and (92) involves the plotting of the individual
transmissibilities Tdj and Tfj. Magnitude and phase angle plots of these
and phase versus frequency ratio are shown in figures 10 and 11. To
motion and evaluate all the known parameters such as masses, damping
the magnitude curve of the function about the 0-decibel line and reverse
40
tl)
30
...-i
a,)
..0
ri 2
0
a,)
~ lO
.
a,)
"0
;:I 0
~
ri
Q -to
00
Cll
;:E.:
"0 -
E:-1
-30 too
-40 Frequency Ratio ~
woj
tl)
a,)
a,)
~
0
00
a,)
bj:::.lO
~ bj~O.S
... -60
a,)
...-i
00
~
<I)
-12.0
tl)
Cll
.a
~
-180
~
Figure 10.
.3o
oo Zo
r-1
<1.1
..0
...-~ Jo
u
<1.1
~
.. 0
<1.1
e -lo
"0
.-I
~
~ -20
::.=
~ -3
100
w
1-0
Frequency Ratio,
Cll woj
<1.1
<1.1
1-1
00
<1.1
~
.
Cll
180
<1.1
r-1 I20
00
~
<1.1
60
Cll
~
..c: 0
P1
4-l
~
Figure 11
1 (85)
xl
It is desired to obtain the magnitude and the phase angle of X: . For
2
M1 Tf 2 1 M1 Tf2 Td1 X0
this, first evaluate the products M Tfl Td and M Tfl Td x The
2 2 2 2 1
two vectors thus obtained are then added in a vector diagram. Add
1
-x- = -A-L-:-t:-
2
xl now become
x2
range far less broad than the full range of resonant frequencies. This
ntethod makes use of this fact and reduces a large system of equations
the degrees of freedom of a system and then solving this reduced system.
sented below makes use of Staffeld's basic ideas but the solution is
obtained from it. For a given frequency range, the steady state system
following steps:
tions.
interest.
63
(3) Solve the reduced system for its steady state response by any of
ation, (2) the eigenvalues of the system which are greater than the
omission from. the system does not produce a large error in the system
response.
the inertia force produced by the missing masses. The original spring
2
constant K is replaced by K-mw0 , where w0 is the usual (or typical)
This permits coupling of this mass with the one which precedes it.
springs without mass and damper. The resulting reduced damped system
(2) Fewer forced vibration equations are required and the uncoupling
in figure 12 with
2
M
1
= M2 = M
3
= 1 lb-sec /in.
K
1
= 100 lbs/in., K
2
= 4 lbs/in., K3 =1 lb/in.
Figure 12
A Spring-Mass System
66
the conventional methods can be used; the tabular method of Holzer has
2 2 IMw~Xi
Mass M Mw xi Mw xi IHw2Xi Ki,i+l ---
No Ki,i+l
1 1 .3 1.0 .3 .3 100 .003
2 1 .3 .997 .297 .597 4 .149
3 1 .3 .848 .254 .851 1 .851
4 00 -.003
natural frequency of the system and since only one sign change in the
page 27.
67
2 2 .rMw2xi
Mass M Mw xi Mw xi LMW2Xi Ki,i+l
No Ki,i+l
1 1 6.64 1.0 6.64 6.64 100 .0664
2 1 6.64 .9336 6.195 12.835 4 3.209
4 00 -.0004
w1 = Jfi = ~ = 10 rad/sec
w2 ~~~- Ji
= = 2 rad/sec
w3 = Pi:=
M ~~ = 1 rad/sec
3
2
Table I: w - n .~
2 LHw~Xi
Mass 2 2 LMw Xi
M Mw xi Hw xi Ki,i+l
No Ki,i+l
1 2 .6 1 .6 .6 4 .15
2 1 .3 .85 .255 .855 1 .855
3 00 -.005
Table II: 2
w = 6.7
2 2 2 XMw2Xi
Mass M Mw xi Mw xi }:Mw Xi
No Ki,i+l
Ki,i+l
wl w2 xl x2
It can be seen from the above table that eigenvalues in the frequency
As a second example, consider the system shown below, where the mass
xt.T
Figure 14. A Spring-Mass System
Given data:
------
K = 1 lb/in., K = 5 lb/in., K3
2
= 10 lb/in.
1
2
Table I: w = 0.71
2 2 . .XHw2xi
Mass M Mw xi Mw xi LMw2Xi
No/ Ki,i+l
Ki,i+1
4 00 .008
2
Table II: w = 4.05
IMw2Xi
2
Mass M Mw xi Mw 2X. IMw2Xi Ki,i+l
~
No Ki,i+1
4 00 +.004
are
wl = 1.0 rad/sec
w2 = 2.27 rad/sec
w3 = 3.17 rad/sec
72
modified stiffness K = K3 - Mw 2
3 0
K = 10- (1)(4)
= 6 lb/in.
2
Table I: w = .675
2
2 2 2 }:Mw Xi
Mass M Mw xi Mw xi }:Mw Xi Ki,i+l
No Ki,i+l
3 00 -.002
2
Table II: w = 4.05
2
2 2 2 }:Mw xi
Mass M Mw xi Mw xi }:Mw Xi Ki,i+l
No Ki,i+l
3 00 -.01
K=-6
Figure 15
which is equivalent to
Figure 16
K K2K 5X 6 - 2 73 lb/in.
eq = K +K = 5+6 -
2
74
is presented below:
wl w2 xl x2
can reduce the degrees of freedom and at the same time give results
which compare well with the actual solution. With a little experience,
this method can save a considerable amount of labor when a large num-
points:
results.
(3) Application to systems such as (a) those with masses and dampers
connected to a reference frame; and (b) those having both linear and
angular motions.
(4) Ease of application when used (a) manually; or (b) with a digital
or analog computer.
(1) Even for systems with many degrees of freedom the derivation of
this method lies in the fact that to ascertain the nature of the system
employ.
If there are combined linear and angular motions, the degrees of free-
dom will be more than the number of masses and consequently there will
provides th~ system transient response. Though not the topic of this
thesis, this is an advantage of the Normal Mode Technique over the other
methods.
(4) For classically damped systems having more than two or three
Holzer's Method:
(1) This method does not make use of the frequency equation. The
eigenvalues and system response are obtained by trial and error without
Also, Holzer's Method is ideal for obtaining steady state system response
task if the system has more than a few masses or if damping is present.
78
connected either between two masses or between a mass and the reference
and translational motions have not been treated by this method. In this
case the mass has a combination of inertia force and inertia torque
which makes the problem quite complex and beyond the scope of this work.
from one station to the next. This method has been developed by N. 0.
(1 23)
Myklestad and M. A. Prohl ' .
and labor.
Impedance Method:
natural frequencies nor the mode shapes can be obtained directly from
the Normal Mode Technique, this method is much simpler to use for steady-
The only practical exception is a system with coupled linear and angular
(4) Nost of the computations for this method are relatively simple
Graphical Technique:
form which is suitable for plotting comprises the major part of the
work for this method. The eigenvalues cannot be found directly by this
frequencies. The peaks in the plot of the output magnitude versus fre-
80
to problems having coupled linear and angular motions, because the com-
Reduced System:
the procedure for reducing the degrees of freedom is a simple one, con-
interest.
81
the responses of such systems may tax the capacity of a digital computer.
with the actual solutions. The accuracy of this method when applied to
damped systems will depend upon the accuracy of the conventional method
tl1e steady state response are sinusoidal in nature. To obtain the res-
2
ponse for other types of inputs, several methods( ) are available such
(1) A systematic procedure for solving systems with coupled angular and
linear motions by Holzer's Method, the Impedance Method, and the Graphi-
eliminating only those masses which have little effect on system behavior
BIBLIOGRAPHY
7. Staffeld, S. E., (1962) "A Hethod for Reducing the Number of Degrees
of Freedom in Hathematical Models of Damped Linear Dynamic
Systems". Journal of Engineering for Industry, Trans. ASHE,
p. 418-422.
10. Den Hartog, J. P., Li, L. P., (1946) "Forced Torsional Vibrations
\-lith Damping: An Extension of Holzer's Hethod". Trans. ASHE,
p. A276-280.
18. Gere, J. M., Weaver, W. Jr., (1965) "Matrix Algebra for Engineers".
N. J., Van Nostrand, p. 109-126.
APPENDICES
86
APPENDIX A
Reciprocity Theorem
2
Maxwell's reciprocity theorem will be proved( l) here for influence
2 is F o
2 12
The last two terms of this equation represent the additional potential
Fz
- -t_ - - -- _..
Figure 17
The last two terms of this equation are due to the application of
Fl.
Since the final states of the system are identical for the two
energies expressed in the two cases are the same. By equating the two
oij = oji
cij = cji
APPENDIX B
and consequently the eigenvectors of such a matrix are also real. This
(1)
(2)
T
respectively. Premultiplying equation (1) by Xj (transpose of the
(3)
(4)
Taking the transposes of equations (3) and (4) yields two more equations
(5)
(6)
The left-hand sides of equations (3) and (6) are the same, and similarly
for equations (4) and (5), the right-hand sides of these equations may
be equated. Thus,
(8)
Premultiplying equations (8) and (9) by the transposes of the jth and
taking the transpose of both sides of equation (10) and noting that
(12)
(13)
(14)
(15)
92
Equations (14) and (15) are the orthogonality conditions for the normal
E1 =3
Because two of the eiganvalues are the same and the matrix is unsym-
metric, it is not clear at this point whether or not [B] can be dia-
solutions:
That the above solutions are linearly independent can be verified from
x3 = B3m
[Q] -[ 4
5
0
-1
0
5 :J
-~
-23
1 [ 30
(Ql- =~ 40 -32
-25 20
and
=[
-~J
3 0
(<ifl (B] [Q) 0 3 = [n]
0 0
If all eigenvalues are positive and at least one is zero, the matrix
is Positive Semidefinite.
APPENDIX C
Parallel Four~pole Networks
To derive the formulas for a parallel four-pole network, we have
(1)
(2)
(3)
=1 (5)
necessarily true for all systems which follow the reciprocity principle.
Equations (1) through (5) are written for the spring and damper
individually. The total force carried by the spring and the damper is
then equated to the force at the input or at the output point. The
equations.
96
APPENDIX D
Sample.Problem:
illustrate this point and also to demonstrate the use of the math-
the rotational motion of any of the masses does not occur; and (2) the
vehicle fonvard speed is 20 mph, and the road surface varies sinusoid-
be compared.
Ml = M = 386 lbs
Kl = K = 150 lb/in.
cl = c = 17.13 lb/in./sec
M2 = 10M
K2 = SK 1-<z c2.
c2 = 7C
M5
J:l
M3 = l.SM
K3 = 25K
c3 = 9C ~) (.3
Figure 18
Solution:
. 28.8
w = 10.5 = 2.74 cps = 17.2 rad/sec
(1)
(2)
(3)
F
0
= (4)
M 0 0 0
. 0 0
cl -cl xl Kl -Kl xl
0
1
M2 0
.. + -C 1 Cl+C2 -c 2
.
x2 + -Kl Kl+K2 -K2 x2 0
0 0 M 0 -c x3 0 -K K +K x3 F
2 2 2 0
(5)
.. c .
M 0 0 xl -c 0 xl
.. .
0 10M 0 x2 + -c 8C -7C x2
0 0 l.SM
.. 0 -7C 16C
.
x3 x3
K -K 0 xl 0
+ -K 6K -SK x2 = 0 (6)
0 -SK 30K x3 Fo
M 0 0 K -K 0
0 lOH 0
..X + -K 6K = 0 (7)
2
0 0 -SK 0
Sinwt (8)
2
K-Hw 0 0
-K -SK = 0 (9}
2
0 -SK 30K-1.5Mw 0
6 K 4 . K2 2 K3
w - 21.6 M T,v + 31.0 2 w
M
8.34 M3" = 0 (10)
2 .K
wl = 0.363 M
2 K
w2 = 1.17 M (11)
2 K
w3 = 20.1 M
w = 7.38 rad/sec
1
w
2
= 13.25 rad/sec
w = 54.9 rad/sec
3
(9)' .
.637K -K 0
-K 2.37K -5K
0 -5K 29.45K
.637 A1 - A2 = 0
-A
1
+ 2.37 A2-5 A3 = 0
-5 A + 29.45 A = 0
2 3
Setting arbitrarily the amplitude of the first mass to unity, the mode
x1 = 1.0}
.637
{ .108
Repeating the above procedure for the remaining eigenvalues, the cor-
x2 = 1.0}
-.17 x3 = 1.0}
-19.1
{ -.032 {
760
The modal matrix whose columns are the vectors x1 , x2 , and x3 , becomes
1.0 1.0 1
'[Q] = 637 -.17 -19.1
[ 108 -.032 760
5.067M 0
0 !.29M =[M)
0 0
0
1.45K = (K J
0
~:~~:
-.72
-2130d
[QJT [c) [Ql = c 1.263 555
~2140 575 9.45xl0
Now
1.~
60
60 1.5
10 1.5
and
c
25K
2N space,
Jl [ u + ~ ]ll = 0
1 0 0 -1 0 0
o(
1
0 o( 0 0 -1 0
1
0 0 oe: 0 0 -1
6M 60M 1.5M
0
35C + !_ -26C
25K 25K 25K 25K o<. 2SK
.. M 10M l.SM 9C 1
25K 2SK 25K
0 0 25K +~
Solving the above determinant of order 6x6 is a long and tedious task.
Consider the following equation which has been derived in section IV:
2.193C 0
fc] = 0 1.263C
0 0
and
( Q) T (F) =
.108F
-. 032F:
1
{ 760F0
c
2.193C 0 0
r.067M 0
L~ 1.~9M
0
0
1.263C
0
0
"6
9.45xl0 C
1.82K 0
1.45K (12)
+ 0
0 0
104
(13)
(14)
Using equations (13) and (14), the follmving results are obtained
nl = 0.721 ~1 = -28
n2 = -.702 ~2 = -66.5
n3 = .00193, ~3 = 49.5
Using {Xi}= [Q]{ni}' the system response to the harmonic excitation is,
xl = .021 L.:L.
x2 = s764 L-31. 5
x3 = 1.564 /84.5
0
The phase angle of Xi is with respect to X0 (at 0 ). The vector F0
I)
given by equation (4) leads X0 by an angle of 35 Consequently this
wd 2 = 10.0 rad/sec
For wd 3 ' b is greater than one and therefore, the system becomes over-
3
damped. Hence, vibratory motion will not exist in this case.
Holzer's Method:
support has zero displacement. This is the boundary condition for con-
2 .K
w = 0.363 M
Mass 2 2 2
M Mw Hw X LMw2Xi .Z:Mw Xi
No xi i Ki,i+l
1----
Ki,i+l
1 M .363K 1.0 .363K .363K K .363
2 10M 3.63K 0.637 2.31K 2.673K 5K .534
3 l.SM .5445K 0.103 0.056K 2. 73K 25K .108
-.005
2 K
w1 0.363 M
or w = 7.38 rad/sec
1
2
Starting with a trial value of w = 1.17 ~ for the second natural
w2 Remainder
Amplitude
K
1.17 M 0.027
1.16 K 0.0071
M
1.15 K -0.01
M
2
A graph of the remainder displacement versus w is shown on the next
2
page. The points lie on a straight line which crosses w axis at
107
003
00~
001
.w
~
<li
m
0
cO
.-i
p..
en
rl
A
1-1
to 2
lL ~
1"1
12. k..
M
<li w
'1j
~
rl
~
~
-o.ot
Figure 19
2 'K
w = 1.155 M
Mass 2
X
. Mw2Xi
M Mw2 xi Mw X. LMw2Xi
No 1 Ki,i+l Ki,i+l
1 M 1.155K 1.0 1.155K 1.155K K 1.155
-.0006
therefore,
2 K
w2 = 1.155 M
or w
2
= 13.16 rad/sec
w
3
= 54.9 rad/sec
The mode shapes are obtained directly from the amplitude column
is shown on the next page. For this table, the displacement of the
xl.= a+ ib
Equating the real and the imaginary parts of the above equation provides
-1.393a - b = 2 (17)
a - 1.393b =0 (18)
which gives
a = -.95
b = -.681
Therefore,
Mass Column 6
Kj,j+1
Number M Mw2 xj Mw2xj LMw2Xj Column 7
+iwC j, j+1
(1) (2) (3) (4) (5) (6) (7) (8)
-1. 393X1
+(Xl)i
,_.,_.
0
111
x2 = 1.16/-91.2
and
x3 = 1.74 ~25.5
above is exact.
Impedance Hethod:
In order to apply this method, the various elements of the system are
connected between the points which have been numbered as shown in figure
20. Thus across points (1) and (2) are a spring and a damper in parallel
and across points (2) and (3) is a mass, and so on. The velocity v1
at point (1) is wx and the force F at this point is unknown. The force
1
F at point (7) is zero as it is a free end. With these boundary con-
7
ditions and knowing the four-pole parameters which relate any two points,
Mass
Column 6
Number M Mw2 Xj M\v2xj IMw2xj Kj,j+l+iwcm,j+l Column 7
-.003+.707i
.....
.....
N
113
6'K 7C.
Figure 20
the steady state system response can be determined. Using the theory
developed in section VI~ the following relationships are obtained:
[::J 25~8~ J
= [: L:: J (20)
[::] =
~.00:36+i(.00268) :] [:j (21)
[::] =
[: 172i
1
Fs
vs J (22)
[::] [: =
:J17.2i
1
(24)
and (7),
[-5+8.04i J
[::] = -.0327+.076i
-170+100J [ 0
-1.4+1.0:1. v
7
(25)
L5+8.04i
[:~ =
.00246+.074i
-170+1001] [
-. 73+1.31
0]
v
7
(26)
v
1
= (-1.4 + l.Oi)V
7
(27)
and
vl = 1.15 v 2 /25.5:
6
or v2 = .87 vl /-25.5
Now x1 = X0 = 2
Therefore,
x2 = 1.74 /-25.5
(30)
F
1
= (113 + 67i)V2 (32)
F
3
= (113 + 41.2i)V 2 (33)
F
3
= F4 (35)
V
3
= (.00736+.00268i)F4+v4 (36)
v
4
= (.277 - .606i)v2 (37)
-.73 + 1.3li
or Xz .
x6 = -.73 + 1.3li
coordinates,
x1 = 1.163 /215.5
x2 = 1.16 f-9!':._
x3 = 1.74 ;'-25.5
Method.
yraphical Technique:
For obtaining the steady state system response by this method, define
the coordinates of the system as shown in figure 21. This permits the
and
w
w
- - = .345, - - = 1.985, -w- = 1.4
WQ2 W03
The ratio of the damping coefficient and the critical damping for
individual M, K, and C is
Kt
Figure 21
w
Corresponding to--- and b., the value of Tdj is read from the dis-
woj J
2
placement transmissibility curve(l, ). Tfj is then obtained by the
relation
The values of Tdj and Tfj are presented below in a polar notation:
0
Td
2 = o. n L-6o ' T2 = 0.885 /42.4_
Td
3
= 1. 0 /-54 ' T3 = 0.918M
following equations:
x3
- = Td (39)
x2 3
x2
-=
xl
Td 2 [1 +M3M2-
Td
3 x
Tf Tf2 (1 - _2_)
3 x2
rl
(40)
(41)
X
0
section VII. The vector diagram for solving equation (40) is shown
120
72 (_,_Q.
= ----~~~~~==~~
1 + .0885 ;-11.6
= . 72 /-60.
!.086 L-2b_
(43)
(44)
coordinates,
1.125 t-_21 .6
. 663 L-58
1.0 1-st!:_
which gives
x3 = 2.25 ;'-27.6
x2 = 1.49 L-85.6~
121
122
x1 ~ 1.49 /-139.6
or x1 ~ 1.49 /220.4
-x3 =
xo
t
0.87-25.5~
~---------
- = 1.125 /--27
X
.6
0
x3
TI1e calculations for Xo were checked and rechecked. However, extensive
Red~ced System:
150 . 17.13
10
. 12.0
Figure 23
150
10
610
XoSin 17.2t
Figure 24
obtained-:
Hethod.
equations. TI1us,
where x1 = a + ib
As before, the real and the imaginary parts of the above equation are
a = -.583, b = -1.13
Therefore,
x1 == 1.27/242.8~
Substituting for x in the expression for X2 (obtained from the dis-
1
placement column of the Holzer's table),
w2 = 294.0
-.715Xl
+(1. 394X1) i
.....
N
VI
126
x2 = 1.266 /-64
to the exact system response. However, the phase angles of the displace-
error. This is the case for this problem as can be observed by comparing
**
7.38 L!_.o .637 .1~ .87 f2s. S0
Holzer 13.16 lJ.. 0 -.lSS -.02~ . 666 /-65.7
S4.9 L!.:o -19.1 741_J 1.06 /-s3. 2
87 L-2s. s:_
Impedance
* * .666 L-6S.5_
1.02 L-53. s__o
1.125 (_-21. Go_
Graphical
* * . 663 0_
1.0 /-54
*~fhesc ratios are for the approximate system obtained by ignoring the
off-diagonal terms of the damping matrix.
*Values cannot be determined.
128
For the above table, the natural frequencies and the mode shapes are
order:
129
VITA
University of Bombay and the School of Mines and Metallurgy (now the