Vous êtes sur la page 1sur 9

Undamped free and forced vibrations of a two-degrees-of-freedom line...

1 of 9

http://fsinet.fsid.cvut.cz/en/U2052/node123.html

Next: Comparison of analytical and Up: Systems with 2 DOF Previous: Systems with 2 DOF Contents

Undamped free and forced vibrations of a two-degrees-of-freedom


linear system

Figure 5.31: A two-degree-of-freedom system


The equations of motions are
(5.160)

The spring forces are proportional to their extensions


(5.161)

Substituting, rearranging
(5.162)

(5.163)

and rewriting into the matrix form


+

(5.164)

14.10.2016 09:50

Undamped free and forced vibrations of a two-degrees-of-freedom line...

2 of 9

http://fsinet.fsid.cvut.cz/en/U2052/node123.html

we finally get
M

where

+ K

(5.165)

M , K are mass and stiffness matrices respectively. These equations are linear and

homogeneous of the second order and their solutions can be assumed in the form

Substituting into the equations of motion and 'dividing out by the sine function' we get
(5.167)

(5.168)

This is a set of homogeneous linear algebraic equations in C1 and C2, which have a nontrivial solution
only if the determinant of the coefficient vanishes, that is
= 0.

(5.169)

This is called the characteristic - or the - frequency equation of the system from which the values of
natural frequencies can be determined. Solving the frequency equation we get

(5.170)

The natural frequencies, also called eigenfrequencies, can be found by Matlab both analytically and
numerically.
Generally, the eigenfrequencies of a mechanical system with n degrees of freedom described by
M

(5.171)

14.10.2016 09:50

Undamped free and forced vibrations of a two-degrees-of-freedom line...

3 of 9

http://fsinet.fsid.cvut.cz/en/U2052/node123.html

could be found assuming


x

ei

x0

=-

x0

ei

t.

(5.172)

Substituting the assumed vibrations into the equation of motion we get a so-called generalized eigenvalue
problem
K

x0

(5.173)

that could be transformed into a standard eigenvalue problem

In both cases the problem is to find n couples

x0

x0

where

or

, x0

(5.174)

satisfying the above

equations.
Matlab can solve both problems numerically; for analytical solution, however, only the standard
eigenvalue problem can be treated by the Matlab symbolic toolbox. See the program V2Etwodofa.m.
clear; format compact
m1=400; m2=300; k1=60000; k2=50000;
m=sym('[m1,0;0,m2]'); k=sym('[k1+k2,-k2;-k2,k2]'); minv = inv(m); c=sym(minv)*sym(k);
ei=eig(c); disp('the formula for the first eigenvalue is') la1s=sym(ei,1,1) disp('the source code for TEX processing') latex(la1s)
la2s=sym(ei,2,1); disp('evaluate numerically the formula derived by Matlab') disp('eigenfrequencies are square roots of
eigenvalues')
einumeric = sqrt(subs(ei))
disp('evaluate numerically the formula (5.173) derived in the text') a=(k1+k2)/m1; b=k2/m2; c=k1*k2/(m1*m2); om1=sqrt(0.5*
(a+b)+sqrt(0.25*(a+b)^2-c)) om2=sqrt(0.5*(a+b)-sqrt(0.25*(a+b)^2-c)) disp('The results are the same - what else could you
expect?')

The text output of this program is


V2Etwodofa the formula for the first eigenvalue is la1s = 1/2/m1/m2*(k1*m2+k2*m2+m1*k2+
(k1^2*m2^2+2*k1*m2^2*k2-2*k1*m2*m1*k2 +k2^2*m2^2+2*k2^2*m2*m1+m1^2*k2^2)^(1/2)) the source code for TEX
processing ans = 1/2k1m2+k2m2+m1 k2+k1^2m2^2+2 k1m2^2k2-2k1 m2m1k2+k2^2m2^2+2 k2^2m2 m1+m1^2k2^2m1m2
evaluate numerically the formula derived by Matlab notice that eigenfrequencies are square roots of eigenvalues einumeric =
19.3649 8.1650 evaluate numerically the formula (5.173) derived in the text om1 = 19.3649 om2 = 8.1650 The results are the
same - what else could you expect?

The complementary function satisfying the equations of motion could be written in an alternative form

14.10.2016 09:50

Undamped free and forced vibrations of a two-degrees-of-freedom line...

4 of 9

http://fsinet.fsid.cvut.cz/en/U2052/node123.html

(5.175)

Since (5.171) is a set of homogeneous equations we can obtain the ratio of amplitudes and express them
for both natural frequencies obtained from (5.173)
=

(5.176)

(5.177)

(5.178)

(5.179)

(5.180)

(5.181)

From the four ratios defined by (5.180), (5.181), (5.183) and (5.184) only two are independent. Defining
=

(5.182)

we could set
C21 =

C11,

C22 =

C12.

(5.183)

Finally, the equation (5.178), describing the transient response of the system, can be rewritten into
(5.184)

14.10.2016 09:50

Undamped free and forced vibrations of a two-degrees-of-freedom line...

5 of 9

The unknown constants C11, C12,

http://fsinet.fsid.cvut.cz/en/U2052/node123.html

can be determined from initial conditions as it is shown in

5.10.2.
Forced vibrations of a two-degrees-of-freedom system
If a harmonic excitation of the same frequency and of the same phase is applied to both particles, then the
equations of motion are
(5.185)

The particular integrals, describing the steady-state response of the system, could be assumed in the form
(5.186)

Substituting into the equations of motion we get


(5.187)

(5.188)

The amplitudes are

(5.189)

Read carefully the program twodof2a.m where three different approaches to the determination of amplitudes
of the steady-state motion are compared
clear m1=400; m2=300; k1=60000; k2=50000; om=5; p1=10; p2=15;
ms=sym('[-m1*om^2+k1+k2,-k2;-k2,-m2*om^2+k2]'); ims=inv(ms); p=sym('[p1;p2]'); a=symmul(ims,p); disp('the first
amplitude') a1a=sym(a,1,1) disp('the second amplitude') a2a=sym(a,2,1) disp('their numeric values are') a1an=subs(a1a)
a2an=subs(a2a)
aalter=linsolve(ms,p); disp('numeric values found by an alternative way') aalter1n=eval(sym(aalter,1,1))
aalter2n=eval(sym(aalter,2,1))
disp('evaluate numerically the formula (5.192)') jm=(k1+k2-m1*om^2)*(k2-m2*om^2)-k2^2; a1=(p1*(k2-m2*om^2)+p2*k2)/jm
a2=(p2*(k1+k2-m1*om^2)+p1*k2)/jm

14.10.2016 09:50

Undamped free and forced vibrations of a two-degrees-of-freedom line...

6 of 9

http://fsinet.fsid.cvut.cz/en/U2052/node123.html

t=0:0.01:10; x1=a1*sin(om*t); x2=a2*sin(om*t); figure(1) subplot(2,1,1); plot(t,x1,'linewidth',2); title('the first particle


displacement','fontsize', 12); subplot(2,1,2); plot(t,x2,'linewidth',2); title('the second particle displacement','fontsize', 12);
xlabel('time','fontsize', 12) print V2Etwodof2a -deps; print V2Etwodof2a -dmeta

The text output of the program is


V2Etwodof2a the first amplitude a1a = -(m2*om^2-k2)/(m1*om^4*m2-m1*om^2*k2-k1*m2*om^2+k1*k2k2*m2*om^2)*p1+k2/(m1*om^4*m2-m1*om^2*k2-k1*m2*om^2+k1*k2- k2*m2*om^2)*p2 the second amplitude a2a =
k2/(m1*om^4*m2-m1*om^2*k2-k1*m2*om^2+k1*k2-k2*m2*om^2)*p1- (m1*om^2-k1-k2)/(m1*om^4*m2-m1*om^2*k2k1*m2*om^2+k1*k2- k2*m2*om^2)*p2 their numeric values are a1an = 6.7143e-004 a2an = 0.0011 numeric values found by an
alternative way aalter1n = 6.7143e-004 aalter2n = 0.0011 evaluate numerically the formula (5.192) a1 = 6.7143e-004 a2 =
0.0011

The graphical output is in Fig. 5.32 and is not too interesting

Figure 5.32: Forced vibration of a two-degrees-of-freedom system


Solution by means of the Newmark integration method.
Time history of the response of more complicated mechanical systems with many degrees of freedom is
usually found numerically. A number of integration methods are available. One of them, very popular in
technical practice, is the so called Newmark integration method. Let us apply the Newmark method for
the mechanical system described in the text above and in the program V2Etwodof3a.m and compare the time
history of displacements of both particles. Numerical integration is provided by the function V2Enewmd.m.
See paragraph 5.8.2.
clear
m1=400; m2=300; k1=60000; k2=50000; om=5; p10=10; p20=15; tmax=10; h=0.02; tt=0:h:tmax;

14.10.2016 09:50

Undamped free and forced vibrations of a two-degrees-of-freedom line...

7 of 9

http://fsinet.fsid.cvut.cz/en/U2052/node123.html

xm=[m1 0; 0 m2]; xk=[k1+k2 -k2; -k2 k2];


gama=0.9; beta=0.25*(0.5+gama)^2; a1=1/(beta*h*h); a1d=gama/(beta*h); ck=0; cm=0; xd=ck*xk+cm*xm; p=[0 0]';
xk=xk+a1*xm+a1d*xd; acc=xm; dis=zeros(2,1); vel=zeros(2,1); t=0; kk=0; kmax=round(tmax/h);
dis1=zeros(kmax,1); dis2=zeros(kmax,1);
while t<=tmax, [disn,veln,accn] = ... VTRnewmd(beta,gama,dis,vel,acc,xm,xd,xk,p,h); kk=kk+1; t=t+h; dis1(kk)=dis(1);
dis2(kk)=dis(2); dis=disn; vel=veln; acc=accn; p1=p10*sin(om*t); p2=p20*sin(om*t); p=[p1 p2]'; end;
jm=(k1+k2-m1*om^2)*(k2-m2*om^2)-k2^2; am1=(p10*(k2-m2*om^2)+p20*k2)/jm; am2=(p20*(k1+k2-m1*om^2)+p10*k2)/jm;
x1=am1*sin(om*tt); x2=am2*sin(om*tt); dis1t=dis1'; dis2t=dis2';
figure(1)
subplot(211); plot(tt(1:kmax),dis1t(1:kmax),'k:', ... tt,x1,'k', 'linewidth', 2); lab='Steady-state (solid) vs. transient + steady-state
(dotted)'; title(lab) ylabel('particle 1') subplot(212); plot(tt(1:kmax),dis2t(1:kmax),'k:', ... tt,x2,'k', 'linewidth', 2); lab=['gamma for
Newmark is ' num2str(gama)]; title(lab); xlabel('time'); ylabel('particle 2') sgama = gama*100; str_sgama = int2str(sgama);
file_name = ['V2Etwodof3a' str_sgama]; print('-deps', file_name); print('-dmeta', file_name);
function [disn,veln,accn] = ... VTRnewmd(beta,gama,dis,vel,acc,xm,xd,xk,p,h) a1=1/(beta*h*h); a2=1/(beta*h);
a3=1/(2*beta)-1; a4=(1-gama)*h; a5=gama*h; a1d=gama/(beta*h); a2d=gama/beta-1; a3d=0.5*h*(gama/beta-2); r = p + ...
xm*(a1*dis+a2*vel+a3*acc)+xd*(a1d*dis+a2d*vel+a3d*acc); disn=xk; accn=a1*(disn-dis)-a2*vel-a3*acc;
veln=vel+a4*acc+a5*accn;

Two graphical outputs are presented here.


In Fig. 5.33 you can notice that the analytical and numerical solutions differ significantly. This is quite
clear because the former describes the steady-state response only, while the latter corresponds to the
general solution comprising both transient and steady state solutions. It is easy to add the analytical part
of the transient solution and compare the results. See paragraph 5.10.2. It is, however, the steady-state
part of the response that is usually of interest in solving the engineering vibration problems.

14.10.2016 09:50

Undamped free and forced vibrations of a two-degrees-of-freedom line...

8 of 9

http://fsinet.fsid.cvut.cz/en/U2052/node123.html

Figure 5.33: Steady-state and transient results are compared - no numerical damping
The second graphical output, in Fig. 5.34, shows how the Newmark integration method could be used for
the removal of the transient part of solution. No numerical damping was introduced (gamma = 0.5) for the
calculation presented in Fig. 5.33. Employing the Newmark numerical damping by setting gamma = 0.9, we
can notice that the transient part of the solution quickly dies out.

14.10.2016 09:50

Undamped free and forced vibrations of a two-degrees-of-freedom line...

9 of 9

http://fsinet.fsid.cvut.cz/en/U2052/node123.html

Figure 5.34: Steady-state and transient results are compared - Newmark numerical damping
More details about methods for numerical integration could be found in 5.8.2.

Next: Comparison of analytical and Up: Systems with 2 DOF Previous: Systems with 2 DOF Contents
marcel 2001-08-14

14.10.2016 09:50

Vous aimerez peut-être aussi