D. J.

Inman
1/24
Mechanical Engineering at Virginia Tech
Section 1.8 Stability
Stability is defined for the solution of free
response case:

Stable:

Asymptotically Stable:
Unstable:
if it is not stable or asymptotically stable
x(t ) < M, t > 0
lim ( ) 0
t
x t

=
D. J. Inman
2/24
Mechanical Engineering at Virginia Tech
Examples of the types of stability
x t sin
.
2 t y t
.
e
.
0. 1 t
x t z t e
.
0. 1 t
r t
.
z t x t
0 5 10
1
1
x t
t
0 5 10
1
1
y t
t
0 5 10
1
2
3
z t
t
0 5 10
4
2
2
4
r t
t
Stable Asymptotically Stable
Divergent instability Flutter instability
D. J. Inman
3/24
Mechanical Engineering at Virginia Tech
Example: 1.8.1: For what values of the
spring constant will the response be stable?
Figure 1.37
( )
2 2
2 2
sin cos sin 0 0
2 2
2 2 0 (for small )
k k
m mg m mg
m k mg
u u u u u u u
u u u
| |
+ = + =
|
\ .
+ =
kl > 2mg for a stable response
D. J. Inman
4/24
Mechanical Engineering at Virginia Tech
1.9 Numerical Simulation
Solving differential equations
by numerical integration
Euler, Runge-Kutta, etc.
Available in Mathcad,
Matlab, Mathematica and
Maple (or in FORTRAN)
Or use Engineering Vibration
Toolbox
Will use these to examine
nonlinear vibration problems
that do not have analytical
expressions for solutions
dx(t
i
)
dt
= lim
At 0
x(t
i +1
) x(t
i
)
At
At = t
i +1
t
i
0 0.5 1 1.5 2
0
0.1
0.2
0.3
0.4
0.5
Time (sec)
A
m
p
l
i
t
u
d
e

o
f

x

t A
0
t
4
t
D. J. Inman
5/24
Mechanical Engineering at Virginia Tech
0 0.5 1 1.5 2
0
0.1
0.2
0.3
0.4
0.5
Time (sec)
A
m
p
l
i
t
u
d
e

o
f

x
First order differential equation
| |
0
1
0
1
1
solve ( ) ( ), (0)
,
( ),
1
i i
i
i i i i
i i
x t ax t x x
x x
ax x
t
x x t t t t
x x a t
+
+
+
= =

=
A
= A =
= + A
The new value of x
is calculated from
the old value of x.
x
3
will be used to
calculate the
next term x
4

Time (sec)
a
m
p
l
i
t
u
d
e

D. J. Inman
6/24
Mechanical Engineering at Virginia Tech
Example 9.7.1 solve dx/dt=-3x, x(0)=1
0
1 0 0
2 1 1
3, take 0.5
1
(0.5)( 3)( ) 0.5
(0.5)( 3)( ) 0.25
=
a t
x
x x x
x x x
= A =
=
= + =
= + =
t
t t
t
e t x
A Ae x x
Ae Ae t x t x
Ae t x
3
0
0
) (
that so 1 1 ) 0 (
, 3
3 ) ( 3 ) (
) (

=
= = = =
=
= =
=

Numerical solution
Analytical solution
Note that the numerical
Solution is different for
Each choice of At
D. J. Inman
7/24
Mechanical Engineering at Virginia Tech
Time step
With time step at
0.5 sec the
numerical solution
oscillates about the
exact solution
Large errors can be
caused by choosing
the time step to be
too small
Small time steps
require more
computation
0 1 2 3 4
-0.5
0
0.5
1
Time (sec)
A
m
p
l
i
t
u
d
e

o
f

x

Numerical solution A t=0.5sec
Numerical solution t=0.05sec
Exact solution
A
D. J. Inman
8/24
Mechanical Engineering at Virginia Tech
Numerical solution of the 2nd
order equation of vibration:
1 2
2 2 1
x x
c k
x x x
m m
=
=
It is necessary to convert the second order equation into
two first order equations. To achieve this two new variables
x
1
and x
2
are defined as follows.
1 2
0
Let ,
mx cx kx
x x x x
+ + =
= =
From this two first order differential equations can be
written.
Called state space
D. J. Inman
9/24
Mechanical Engineering at Virginia Tech
Matrix form

(

(
(

=
(

2
1
2
1
1 0
x
x
m
c
m
k
x
x

x
A
x(t
i +1
) = x(t
i
) + AtAx(t
i
)
x
i +1
= I + AtA
| |
x
i
Combining these first order DEs in matrix form gives.
The Euler numerical method can then be applied to the
matrix form to give.
State
vector
State
matrix
D. J. Inman
10/24
Mechanical Engineering at Virginia Tech
Matlab Solutions ode23 and
ode45
Use Runge-Kutta. More sophisticated than the Euler
method but more accurate
Often picks At (i.e. if solution x(t) is rapidly changing At is
chosen to be small and visa-versa
Works for nonlinear equations too
function xdot=sdof(t,x)
k=2;c=1;m=3;
A=[0 1;-k/m -c/m];
xdot=A*x;
Create Matlab function
t0=0;tf=20;
x0=[0 ; 0.25];
[t,x]=ode45('sdof',[t0 tf],x0);
plot(t,x)
In the command window
Saved as sdof.m
D. J. Inman
11/24
Mechanical Engineering at Virginia Tech
Resulting solution
0 5 10 15 20
-0.2
-0.1
0
0.1
0.2
0.3
Time (sec)
A
m
p
l
i
t
u
d
e

Displacement
Velocity
D. J. Inman
12/24
Mechanical Engineering at Virginia Tech
Why use numerical simulation when we can
compute the analytical solution and plot it?
To have a tool that we are confident
with that will allow us to solve for the
response when an analytical solution
cannot be found
Nonlinear systems to not have
analytical solutions, but can be
simulated numerically
D. J. Inman
13/24
Mechanical Engineering at Virginia Tech
Section 1.10 Coulomb Friction and
the Pendulum
Nonlinear phenomenon in
vibration analysis
D. J. Inman
14/24
Mechanical Engineering at Virginia Tech
Vibration of Nonlinear Systems
Sliding or Coulomb Friction
( ) 0
( ) 0 ( ) 0
( ) 0
c
N x t
f x x t
N x t

>

= =

<

m
k
x
0

N=mg
x(t)
The force due to Coulomb friction opposes motion, hence the
sgn function is used. The force is proportional to the normal
force and independent of the velocity of the mass.
D. J. Inman
15/24
Mechanical Engineering at Virginia Tech
The free body diagram split depending
on the direction of motion:
Figure 1.41
mass moving left
( ) 0 x t <
mass moving right
( ) 0 x t >
sgn( ) 0 (1.92) mx mg x kx + + =
D. J. Inman
16/24
Mechanical Engineering at Virginia Tech
The sgn function is nonlinear
Causes equation of
motion to be nonlinear
Can solve as piecewise
linear (see text)
Can solve numerically
Has more than one
equilibrium position
Decay is linear rather
then exponential
Comes to rest when
spring cannot overcome
friction at the instant the
velocity is zero
0 5 10 15 20 25
-0.2
-0.1
0
0.1
0.2
0.3
Time (sec)
D
i
s
p
l
a
c
e
m
e
n
t

x

Does not
settle at x=0
Linear
decay
D. J. Inman
17/24
Mechanical Engineering at Virginia Tech
Figure 1.42 shows the details of the free
response of a system with Coulomb damping
D. J. Inman
18/24
Mechanical Engineering at Virginia Tech
A General second order system can be
written as a single first order equation
1 2
2 1 2
( )
( , )
x x
x f x x
( (
= = =
( (

x F x
2 1
0 and
s s
mg mg
x x
k k

= < <
The equilibrium position is defined:

F(x
e
) =0
For Coulomb friction this is defined as:
i.e. the positions where the force due to the spring can no
longer overcome the sliding friction force
D. J. Inman
19/24
Mechanical Engineering at Virginia Tech
x + x |
2
x
3
= 0
Example 1.10.2: Calculating the
equilibrium position for nonlinear DEs
Equation of motion:
2
2 2
1 1
0
( 1) 0
1 1
0
, ,
0
0 0
e
x
x x
x
|
| |
=
=

( (
(
( (
=
(
( (

( (

Multiple equilibrium
positions possible
Equilibrium positions:
State space form:
1 2
2 2
2 1 1
( 1)
x x
x x x |
=
=
D. J. Inman
20/24
Mechanical Engineering at Virginia Tech
The pendulum
Unstable
Equilibrium
u = t, 3t, 5t...
Stable
Equilibrium
u= 0, 2t, 4t ...
u

m
mg
D. J. Inman
21/24
Mechanical Engineering at Virginia Tech
Example 1.10.2 Equilibrium of a Pendulum
Figure 1.44
1 2
1 2
2 1
( ) sin ( ) 0
,
sin
g
t t
x x
x x
g
x x
u u
u u
+ =
= =

=
=
2
1
2
1
( ) 0
sin
0
sin 0
x
g
x
x
x
(
(
= =
(

=
F x
x
2
= 0 and x
1
= nt, n = 0,1, 2L
D. J. Inman
22/24
Mechanical Engineering at Virginia Tech
Solution to the pendulum
Can use numerical simulation to examine both linear and
nonlinear response
Let (g/L)=(0.1)
2
so that e
n
= 0.1
a) use u(0)=0.3 rad & initial vel: 0.3 rad/s
b) change the initial position to: u(0)= t rad which is near the
unstable equilibrium
0 10 20 30 40 50
-1.5
-1
-0.5
0
0.5
1
1.5
Time (sec)
A
n
g
l
e

(
q
)

Non-linear
Linear
0 10 20 30 40 50
-10
0
10
20
30
Time (sec)
A
n
g
l
e

(
q
)

Non-linear
Linear
(a)
(b)
D. J. Inman
23/24
Mechanical Engineering at Virginia Tech
Pendulum with friction added
0 20 40 60 80
0
5
10
15
Time (sec)
A
n
g
l
e

(
u
)

After making a
single loop the
pendulum
cannot make a
second rotation
and settles to the
stable
equilibrium
position of u=4t
Friction
loss
causes
slow decay
u=4t
sin 0
g
c u u u + + =
D. J. Inman
24/24
Mechanical Engineering at Virginia Tech
Summary of Nonlinear Vibrations
Additional phenomena over linear case
Multiple equilibrium
Instabilities possible with positive coefficients
Form of response dependent on initial conditions
Closed form solutions usually not available
Can simulate numerically
Linear model has tremendous advantages
Linear combination of inputs yields linear combination of
outputs
Linear ode techniques very powerful
But dont make a design error by ignoring important nonlinear
situations
All systems have nonlinear ranges of operation
Need to sort out when nonlinearity is important to consider
and when to ignore it