3 General forced response

• So far, all of the driving forces have been sine or cosine

excitations
• In this chapter we examine the response to any form of
excitation such as
– Impulse
– Sums of sines and cosines
– Any integrable function
Linear Superposition allows us to break up complicated
forces into sums of simpler forces, compute the response
and add to get the total solution

If x1 , x2 are solutions of a linear homogeneous

equation, then
x  a1 x1  a2 x2 is also a solution.
If x1 is the particular sol of x  n2 x  f1
and x2 the particular sol of x  n2 x  f 2
 ax1  bx2 solves x  n2 x  af1  bf 2
 3.1 Impulse Response Function

F(t) Impulse excitation

0 t  
Fˆ
 ˆ
2 F
F (t )      t  
 2
0 t  


 
 is a small positive number
 


Figure 3.1
From sophomore dynamics
The impulse
imparted to an object is equal to the change in the
objects momentum
i.e. area impulse force =  F (t )dt  F t
F(t) under
  
pulse
Fˆ I ( )  
 

F (t )dt   F (t )dt N  s

2
Fˆ
 2  Fˆ
2

   


We use the properties of impulse to define the impulse
function: Dirac Delta
Equal
function
impulses
F(t)

F (t   )  0, t  

 F (t   )dt  Fˆ

t
If Fˆ  1, this is the Dirac Delta  (t)
The effect of an impulse on a spring-mass-damper is
related to its change in momentum.
Just after Just before
impulse impulse
impulse=momentum change
 
F t  mv  m[v(t )  v(t )]
0 0

ˆ F t
F
Fˆ  mv0  v0  
m m

Thus the response to impulse with zero IC is equal to the free

response with IC: x0=0 and v0 =FΔt/m
Recall that the free response to just non zero initial
conditions is:
The solution of:
mx  cx  kx  0 x(0)  x0 x(0)  v0
in underdamped case:

 v0  n x0    x0d 
2 2
x0d
x(t )  e-nt sin(d t  tan 1 )
d v0  n x0

For x0  0 this becomes:

v0e-nt
x(t )  sin d t
d
Next compute the response to x(0)=0 and v(0) =FΔt/m

The solution of:

Fˆ
mx  cx  kx  0 x(0)  x0 x(0)  F t / m 
m
in underdamped case from the previous slide is:
ˆ -nt
Fe
x(t )  sin d t
md

Response to an impulse at t = 0, and zero initial conditions

 So for an underdamped system the impulse response is (x0 = 0)
ˆ nt
Fe
x(t )  sin d t (response to Fˆ ) (3.6)
md

ˆ (t ), where h(t )  e n t

Fˆ
x(t )  Fh sin d t (3.8)
md
unit impulse response function x(t) M

0.5

h(t 0 k c
)
-0.5

-1
0 10 20 30 40
Time
The response to an impulse is thus defined in terms of the
impulse response function, h(t).

So, the response to  (t ) is given by h(t ).

e nt
h(t )  sin d t (3.8)
md
What is the response to a unit impulse applied
at a time different from zero?
The response to  (t - ) is h(t - ).
This is given on the following slide
  0 t 
 n (t  )
h(t   )   e
 m sin d (t   ) t  
 d

for the case that the impulse occurs at  note that the effects of
non-zero initial conditions and other forcing terms must be super
imposed on this solution (see Equation (3.9))
1
=0
For example: If two h
1
0

pulses occur at two -

1 0 1 2 3 4
1
0 0 0 0
different times then =10
h2 0
their impulse
-1
responses will 1 0 10 20 30 40

superimpose h1
+h 0
2
-1
0 10 20 30 40
Time
Consider the undamped impulse response

Setting   0 in the equation (3.8)

Response to unit impulse applied at t   ,
i.e.  (t - ) is:
1
h (t   )  sin n (t   )
mn
Example 3.1.2 Design a camera mount with a
vibration constraint

Consider example 2.1.3 of the security camera again only this

time with an impulsive load
Using the stiffness and mass parameters of Example 2.1.3,
does the system stay with in vibration limits if hit by a 1 kg
bird traveling at 72 kmh?
The natural frequency of the camera system is
k 3Ebh3
n  
mc 12mc 3
(7.1 x 1010 N/m)(0.02 m)(0.02 m)3
  75.43 rad/s
4(3 kg)(0.55)3

From equations (3.7) and (3.8) with ζ = 0, the impulsive response is:
F t mv
x(t )  sin nt  b sin nt
mcn mcn

mb v
X
The magnitude of the response due to the impulse is thus mcn
Next compute the momentum of the bird to complete the
magnitude calculation:

km 1000 m hour
mb v  1 kg 72  20 kg m/s
hour km 3600 s

Next use this value in the expression for the maximum value:
mb v 20 kg m/s
X   0.088 m
mcn 3 kg 75.45 rad/s

This max value exceeds the camera tolerance

 Example 3.1.3: two impacts, zero initial conditions
(double hit).

m  1 kg, c  0.5 kg/s, k  4 N/m

Fˆ  2 N  s and F (t )  2 (t )   (t   )
n  2,   0.125
2e nt
x1 (t )  sin d t  1.008e 0.25t sin(1.984t ), t  0
md
x2 (t )  0.504e 0.25(t  ) sin(1.984(t   )), t  

x(t )  x1  x2
 1.008e0.25(t ) sin(1.984t ) 0  t 
 0.25( t  )
1.008e 0.25t
sin(1.984t )  0.504 e sin(1.984(t   )) t 
Example 3.1.3 two impacts and initial conditions
x  2 x  4 x   (t )   (t  4), x0  1 mm, x0  1 mm/s
Solve three simple problems and add the results.
Homogeneous solution (n  2rad/s,  =0.5, d = 3 rad/s)
v0  x0n
xh (t )  e nt [ sin d t  x0 cos d t ]
d
1  1
 et [ sin 3t  cos 3t ]  e t cos 3t
3

Note, no need to redo constants of integration

for impulse excitation (others, yes)
Computation of the response to first impulse:

Treat  (t ) as x0  0 and v0  1, 0  t  4
n t  v0  1 t
xI (t )  e  sin d t   e sin 3t
 d  3
0t  4
Total Response for 0< t < 4

x1 (t )  xh (t )  xI (t )
t 1
 e (cos 3t  sin 3t ),
3
for 0  t  4
Next compute the response to the second impulse:

1 t  4
x2  e sin 3(t  4), t  4
3
et  4
 sin 3(t  4) H (t  4)
3 Heaviside Step function

Here the Heaviside step function is used to “turn on” the

response to the impulse at t = 4 seconds.
To get the total response add the partial solutions:
1
t et  4
x(t )  e ( sin 3t  cos 3t )  sin 3(t  4)H (t  4)
3 initial condition 3
frist impulse second impulse
1

0.5

x t
0 2 4 6 8 10

0.5

t
 3.2 Response to an Arbitrary Input
The response to general force, F(t), can be viewed as a series
of impulses of magnitude F(ti)Δt
th
Response at time t due to the i impulse zero IC
.
xi(t) = [F(ti)Dt ] h(t-ti) for t>ti x
F(t) ti t
Impulses If t  t I (the i th time interval)
I
F(ti) x(t I )   [ F (ti )t ]h(t  ti )
i 1

t  0, ti   
t
x(t )   F ( )h(t   )d (3.12)
0

t1,t2 ,t3 ti t convolution integral

Properties of convolution integrals: It is symmetric
meaning:

Let   t   , t fixed so that  =t  

and d  d . Also  : 0  t   : t  0
t 0
x(t )   F ( )h(t   )d =  F (t   )h( )( d )
0 t
t
=  F (t   )h( ) d
0
The convolution integral, or Duhamel integral, for
underdamped systems is:

1 nt t

n
x(t )  e 
 F ( ) e sin d (t   ) d
md 0

1 t

md  0
F (t   )e n sin d d (3.13)

• The response to any integrable force can be computed with

either of these forms
• Which form to use depends on which is easiest to compute
 Example 3.2.1: Step function input

 0 0  t  t0
mx  cx  kx  
 F0 t0  t
x0  0, v0  0, 0    1

Figure 3.6 Step function To solve apply (3.13):

t
1 nt 0 1 nt t
 
n n
x(t )  e (0) e sin  (t   ) d  e F e sin d (t   )d
md md
d t 0
0
0

F0 nt t n

md
e t0
e sin d (t   )d
Integrating (use a table, code or calculator) yields the
solution:

F0 F0
x(t )   en (t t0 ) cos d (t  t0 )    , t  t0 (3.15)
k k 1  2

  tan 1 (3.16)
1  2
 Example: undamped oscillator under IC and constant force

For an undamped system: F(t)

F0
1
h(t )  sin nt
mn
The homogeneous solution is t1 t2
v
xh  0 sin nt  x0 cos nt , t  t1
n F (t )

Good until the applied force acts at t1, then:

t
x12   F ( )h(t   )d , t1  t  t2 x(t) m
0
t1 t
  F ( )h(t   )d   F ( )h(t   )d k
0 t1
0
 Next compute the solution between t1 and t2

For t1  t  t2
t
1
x12   F0 sin n (t   )d
t1
mn

F0  (1)(1)
t

  cos n (t   ) 
mn  n t1 
 
F
 0 2 [1  cos n (t  t1 )]
mn
 Now compute the solution for time greater than t2

For t  t2
0 0
t1 t2 t
x2   F ( )h(t   )d   F ( )h(t   )d   F ( )h(t   )d
0 t1 t2


F0  1
t2

  cos n (t   ) 
mn  n t1 
 
F0
 [cos n (t  t2 )  cos n (t  t1 )]
mn 2
 Total solution is superposition:
 v0
 sin nt  x0 cos nt t  t1
  n
 v0 F
x(t )   sin nt  x0 cos nt  0 2 [1  cos n (t  t1 )] t1  t  t2
 n mn
 v0 F
 sin nt  x0 cos nt  0 2  cos n (t  t2 )  cos n (t  t1 )  t  t2
 n mn

m  F0  1, n  8, t1  2, t2  4, x0  0.1, v0  0
Check points: x increases after application of F . Undamped response around x  0
0.3

Displacement x(t)
0.2

0.1

-0.1
0 2 4 6 8 10
Time (s)
 Example 3.2.3: Static versus dynamic load

md g t  0
mx  cx  kx  
 0 t0

md g  1  nt
 x(t )  1  e cos d t   
k  1   
2

md g
  0  x(t )  (1  cos d t )
k
md g
This has max value of xmax 2 , twice the static load
k
 Numerical simulation and plotting

• At the end of this chapter, numerical simulation is used to

solve the problems of this section.
• Numerical simulation is often easier then computing these
integrals
• It is wise to check the two approaches against each other
by plotting the analytical solution and numerical solution
on the same graph
 3.3 Response to an Arbitrary Periodic Input
x  2n x  n2 x  F (t ) where F (t )  F (t  T )
2
• We have solutions to sine and
cosine inputs. 1.5 T
1
• What about periodic but non-
Displacement x(t)
harmonic inputs? 0.5

• We know that periodic 0

functions can be represented -0.5

by a series of sines and cosines -1
(Fourier)
-1.5
• Response is superposition of as -2
many RHS terms as you think 0 2 4 6
Time (s)
are necessary to represent the
forcing function accurately Figure 3.11
 Recall the Fourier Series Definition:

a0 
Assume F (t )     an cos  nt  bn sin  nt  (3.20)
2 n 1
2n
where  n   n
T
T
a0  2
T  0
F (t ) dt (3.21) : twice the average
T
an  2
T  0
F (t ) cos  nt dt (3.22) : Oscillations around average
T
bn  2
T 
0
F (t ) sin  nt dt (3.23)
The terms of the Fourier series satisfy orthogonality
conditions:

T  0 mn
0 sin nT t sin mT tdt  T mn
(3.24)
 2
T  0 mn
0 cos nT t cos mT tdt  T mn
(3.25)
 2
T
0
cos nT t sin mT tdt  0 (3.26)
 Fourier Series Example
F(t)
Step 1: find the F.S. and
F0
determine how many terms
you need
0 t1 t2=T

 0, t  t1

F (t )   F0
 t  t  t  t1  , t1  t  t2
2 1
 Fourier Series Example

1.2

1
F(t)
2 coefficients
0.8 10 coefficients
100 coefficients
0.6F(t)
Force

0.4

0.2

-0.2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)
 Having obtained the FS of input

• The next step is to find responses to each term of the

FS
• And then, just add them up!
• Danger!!: Resonance occurs whenever a multiple of
excitation frequency equals the natural frequency.
• You may excite at 100rad/s and observe resonance
while natural frequency is 500rad/s!! Backwards?
Solution as a series of sines and cosines to
x  2n x  n2 x  F (t )
The solution can be written as a summation

x p (t )  x0 (t )   xcn (t )  xsn (t )
n 1

where x0 (t ) is a solution to
a0 a0
x  2n x   x   x0 (t ) 
2

2n2
n
2
and xcn (t ) and xsn (t ) are a solutions to
x  2n x  n2 x  an cos(nT t ) Solutions calculated from
x  2n x  n2 x  bn sin(nT t ) equations of motion (see
section Example 3.3.2)
 3.4 Transform Methods

An alternative to solving the previous problems, similar

to section 2.3
 Laplace Transform
• Laplace transformation

F (s)   f (t )e st dt  L{ f (t )} (3.41)
0

Laplace transforms are very useful because they change

differential equations into simple algebraic equations
• Examples of Laplace transforms (see page 244) in book)

f(t) F(s)

Step function, u(t) 1/s

-at
e 1/(s+a)
2 2
sin( t )  / ( s +  )
 Laplace Transform
• Example: Laplace transform of a step function u(t)

  e  1  st
u(t)
L{u (t )}    st
e dt    
0
 s 0 s
-at t
• Example: Laplace transform of e
  -at
L{e }   e e dt   e
 at  at  st  ( s  a )t
dt e
0 0


 at e 
 ( s  a )t
1
L{e }     t
 ( s  a )  0 ( s  a)
 Laplace Transforms of Derivatives
• Laplace transform of the derivative of a function
 df (t )   df (t )
 0
 st
L  e dt
 dt  dt
Integration by parts gives,
 df (t )   st 

  
 st
L 
  f (t ) e   s f (t ) e dt
 dt  0 0

 df (t ) 
L    f (0)  sL  f (t )
 dt 
 Laplace Transform Procedures
• Laplace transform of the integral of a function

L 

t
 1
f (t )dt  L  f (t )   f (t )dt
s
0



Steps in using the Laplace transformation to solve DE’s

• Find differential equations
• Find Laplace transform of equations
• Rearrange equations in terms of variable of interest
• Convert back into time domain to find resulting response
(inverse transform using tables)
Laplace Transform Shift Property
Note these shift properties in t and s spaces...

eat f  t  
L
 F s  a
f  t  a  (t  a ) 
L
 e  as F  s 
thus

 (t ) 
1   (t  a) 
L
e L  as
Example 3.4.3: compute the forced response of a
spring mass system to a step input using LT
The equation of motion is
mx(t )  kx(t )  (t )
Taking the Laplace Transform (zero initial conditions)
1 1 1/ m
( ms 2  k ) X ( s)   X ( s)  
s s (ms 2  k ) s ( s 2  n2 )
Taking the inverse Laplace Transform yields:
1/ m 1
x(t )  2 1- cos nt   1- cos nt 
n k

Compare this to the solution given in (3.18)

 Fourier Transform w =2 and M=1
n
0.1
• From Fourier series of non-periodic
functions 0.05
• Allow period to go to infinity
• Similar to Laplace Transform h(t 0
• Useful for random inputs )

-0.05

X ( )   x(t ) e jt dt

-0.1
• Corresponding inverse transform 0 1 2 3 4
Time (s)
 Fourier
 X ( )e d
jt
x(t )  1
2 Transform
 20

• Fourier transform of the unit impulse Normalized H() (dB)

10
response is the frequency response
function 0

 -10
H ( )   h(t ) e  jt
dt

-20
0 1 2 3 4 5
Frequency (Hz)
3.5 Random Vibrations
• So far our excitations have been harmonic, periodic, or at least known in advance
• These are examples of deterministic excitations, i.e., known in advance for all time
– That is given t we can predict the value of F(t) exactly

• Responses are deterministic as well

• Many physical excitations are nondeterministic, or random, i.e., can’t write explicit
time descriptions
– Rockets

– Earthquakes

– Aerodynamic forces

– Rough roads and seas

• The responses x(t) are also nondeterministic

Random Vibrations
• Stationary signals are those whose statistical properties
do not vary over time
• Functions are described in terms of probabilities
– Mean values*
– Standard deviations
• Random outputs related to random input via system
transfer function

*ie given t we do not know x(t) exactly, but rather

we only know statistical properties of the response
such as the average value
Autocorrelation function and power spectral density
The autocorrelation function describes how a signal is
changing in time or how correlated the signal is at two
different points in time.
1 T
Rxx (t )  lim  x(t ) x(t   )d
T  T 0

The Power Spectral Density describes the power in a signal as

a function of frequency and is the Fourier transform of the
autocorrelation function.
1 
S xx ( ) 
2 

Rxx ( )e j d
 Examples of signals
RANDOM
HARMONIC
T
A Arms
Signal
x(t) x(t)
time time

-A

2
2 T A rms
A /2
Autocorrelation

Rxx(t) Rxx(t)
t time shift
2
-A /2 t time shift
 Power Spectral Density
Sxx(w) Sxx(w)

Frequency Frequency
1/T (Hz) (Hz)
More Definitions

T
1
x  lim  x(t )dt (3.47)
Average : T  T
0

T
1 2
x  lim  x (t )dt
2
(3.48)
Mean-square: T  T
0

T
1 2
xrms  x  lim  x (t )dt
2
(3.49)
rms: T  T
0
Expected Value
(or ensemble average)

T
x(t )
E[ x(t )]  lim  dt  x (3.63)
The expected value = T 
0
T

The Probability Density Function, p(x), is the probability

that x lies in a given interval (e.g. Gaussian Distribution)
The expected value is also given by

E[ x]   xp( x)dx

(3.64)
Recall the Basic Relationships for Transforms:

Recall for SDOF

1
transfer function : G (s ) 
ms 2  cs  k
1
frequency response function : G ( j)  H () 
k  m 2  cj
1
unit impulse response function : h(t )  e  nt sindt
md

1
L[h(t )]   G (s )
ms  cs  k
2

And the Fourier Transform of h(t) is H(ω)

What can you predict?
The response of SDOF with f(t) as input:

Deterministic Input: Random Input:

S xx ( )  H ( ) S ff ( )
2
X ( s)  G( s) F ( s)
t 
E[ x ]   H ( ) S ff ( )d
2
x(t )   h(t   ) f ( )d
2


0

In a Lab, the PSD function of a random input and the output can be measured simply in one experiment. So the
FRF can be computed as their ratio by a single test, instead of performing several tests at various constant
frequencies.
Here we get an exact Here we get an expected
time record of the output value of the output given
given an exact record of the a statistical record of the
input. input.
Example 3.5.1 PSD Calculation
Consider mx  cx  kx  F (t ), where the PSD of F (t ) is constant S0
The corresponding frequency response function is:
1
H ( )  (2.59)
k  m 2  c j
2
1 1 1
 H ( )  
2

k  m 2  c j k  m 2  c j k  m 2  c j
1

(k  m 2 ) 2  c 2 2

From equation (3.62) the PSD of the response becomes:

S0
S xx  H ( ) S ff 
2

(k  m 2 )2  (c )2
Example 3.5.2 Mean Square Calculation
Consider the system of Example 3.5.1 and compute:

2
 1
E  x 2   S0  d
 k  m  c j
2
n

m  S0
 S0 
kcm kc

Here the evaluation of the integral is from a tabulated value

See equation (3.70).
Section 3.6 Shock Spectrum

Arbitrary forms of shock are probable (earthquakes, …)

The spectrum of a given shock is a plot of the maximum
response quantity (x) against the ratio of the forcing
characteristic (such as rise time) to the natural period.
Maximum response gives maximum stress.

t
x(t )   F ( )h(t   )d (3.71)
0
Using the convolution equation as a tool, compute the
maximum value of the response
Recall the impulse response function undamped system:

1
h(t   )  sin n (t   ) (3.73)
mn

1
F ( ) sin n (t   )  d
t
x(t ) max 
mn 
0 max
(3.74)

Such integrals usually have to be computed numerically

Example 3.6.1 Compute the response spectrum for
gradual application of a constant force F0. Assume zero
initial conditions
mx(t )  kx(t )  F (t )
t1=infinity, means static loading F (t )  F1 (t )  F2 (t )
F(t) t
F1 (t )  F0
t1
F0 time shift and negative, like half sine problem
 0 0  t  t1

F2 (t )   t  t1
t1 t ( t ) F0 t  t1
 1

The characteristic time of the input

Split the solution into two parts and use the convolution
integral

n t F0 F0  t sin nt 

x1 (t ) 
k 
0
t1
sin n (t   )d   
k  t1 nt1 
 0  t  t1 (3.75)

For x2 apply
F0  t  t1 sin n (t  t1 ) 
time shift t1 x2 (t )      , t  t1 (3.76)
k  t1 nt1 

F0  t sin nt  F0  t  t1 sin n (t  t1 ) 

x(t )  x1 (t )  x2 (t )        (t  t1 )
k  t1 nt1  k  t1 nt1 
(3.77)
Next find the maximum value of this response
To get max response, differentiate x(t).

In the case of a harmonic input (Chapter 2) we computed this

by looking at the coefficient of the steady state response,
giving rise to the Magnitude plots of figures 2.8, 2.9, 2.14.

Need to look at two cases 1) t < t1 and 2) t > t1

For case 2) solve: (what about case 1? Its max is Xstatic)
d  F0 
  n1
 t  sin  t  sin  (t  t1   0 
)
dt  knt1
n n

Solve for t at xmax, denoted tp

 cos nt  cos n (t  t1 ) t t  0

p

cos nt p  cos nt p cos nt1  sin nt p sin nt1
 1  cos nt1 
 nt p  tan 
1

 sin  t
n 1 
sin 2 nt1  (1  cos nt1 ) 2
 2 1  cos nt1 
1 cos nt1
nt p

sinnt 1
From the triangle: sinnt p  
1
(1  cos nt 1)
2
 sinnt 1
cos nt p 
2(1  cos nt 1)

Minus sq root taken as + gives a negative magnitude

Substitute into x(tp) to get nondimensional Xmax :

xmax k 1
 1 2 1  cos nt1 
F0 nt1
st nd
1 term is static, 2 is dynamic. Plot versus:
t1 nt1 Input characteristic time

T 2 System period
Response Spectrum

• Indicates how normalized max output

changes as the input pulse width
increases.
• Very much like a magnitude plot.
• Shows very small t1 can increase
the response significantly: impact, rather
than
smooth force application
• The larger the rise time, the smaller
the peaks
• The maximum displacement is
minimized if
rise time is a multiple of natural period
• Design by MiniMax idea
x maxk
X 
F0
Comparison between impulse and harmonic inputs
Impulse Input Harmonic Input
Transient Output
Harmonic Output
Max amplitude versus
Max amplitude versus
normalized pulse “frequency” normalized driving frequency
Review of The Procedure for Shock Spectrum

1. Find x(t) using convolution integral

2. Compute its time derivative
3. Set it equal to zero
4. Find the corresponding time
5. Evaluate the max possible value of x (be careful about
points where the function does not have derivative!!)
6. Plot it for different input shocks
3.7 Measurement via Transfer Functions

• Apply a sinusoidal input and measure the response

• Do this at small frequency steps
• The ratio of the Laplace transform of these to signals then
gives and experiment transfer function of the system
Several different signals can be measured and these are
named

X ( s) 1
receptance:  2 (3.86)
F ( s ) ms  cs  k
sX ( s ) s
mobility:  2 (3.87)
F ( s ) ms  cs  k
s 2 X (s) s2
inertance:  2 (3.87)
F ( s) ms  cs  k
The magnitude of the compliance transfer function yields
information about the systems parameters

1
H ( j )  (3.89)
(k  m )  (c )
2 2

k 1
H( j ) (3.90)
m cn
1
H (0)  (3.91)
k
3.8 Stability

Stability is defined for the solution of free response case:

Stable: x(t )  M ,  t  0

Asymptotically Stable: lim x(t )  0

t 
Unstable:
if it is not stable or asymptotically stable
Recall these stability definitions for the free response

Stable Asymptotically Stable

Divergent instability Flutter instability

Stability for the forced response:

mx(t )  cx(t )  kx(t )  F (t )

• Bounded Input-Bounded Output Stable

 x(t) bounded for ANY bounded F(t)
• Lagrange Stable with respect to F(t)
 If x(t) is bounded for THE given F(t)
Relationship between stability of the homogeneous
system and the force response

• If xhomo is Asymptotically stable then the forced response is

BIBO stable (Bounded input, bounded output)
• If xhomo is Stable then the forced response MAY be
Lagrange Stable or Unstable
Stability for Harmonic Excitations

The solution to:

mx(t )  kx(t )  F0 cos t
is:
v0  f0  f0
x(t )  sin nt   x0  2 2 
cos nt  2 cos t
n  n    n   2

As long as ωn is not equal to ω this is Lagrange Stable, if the

frequencies are equal it is Unstable
For underdamped systems:
mx(t )  cx(t )  kx(t )  F0 cos t
f0  2n 
x p (t )  cos(t  tan  2 ) 1

(n2   2 )2  (2n ) 2  n   
2

X 

Add homogeneous and particular to get total solution:

x(t )  Ae nt sin(d t   )  X cos(t   )
homogeneous or transient solution particular or
steady state solution

Bounded Input-Bounded Output Stable

Example 3.8.1
M 0  m 2    k sin    cos    mg  sin  
Force from Spring moment arm force moment arm

The equation of motion after a small

angle approximation is given becomes:
m 2 (t )  k 2 (t )  mg  (t )
 m 2 (t )  (k 2  mg ) (t )  0
This will be stable if and only if
the coefficent of  is positive • The system is thus
Lagrange stable.
or if k  mg • Physically this tells
us the spring must
be large
enough to overcome
gravity
Find a force of the form F  t   a  b

to make the system asymptotically stable (BIBO)

m 2  (k 2
 mg )  a  b
 m 2  b  (k 2
 mg  a ))  0

Choose b  0 and a  mg
 m 2  b  k 2  0

Then the system is asymptotically stable and BIBO

3.9 Numerical Simulation of the response

• As before in Section 2.8 write equations of motion as state

space equations
• The Euler integration is just

x(t i 1)  x(t i )  Ax(t i )t  F(t i )t

Example 3.9.1 with delay

Let the input force be a step function a t=t0

F (t )
F0  30 N
F(t) k  1000 N/m
  0.1 x(t)
F0 M
n  3.6
t0  2 s
t0 k c
Example 3.9.1 Analytical versus numerical
 
x(t )  0.03  0.03e0.316(t t0 ) cos[3.144(t  t0 )  0.101] (t  t0 )

Response to step input  0 1   0 

 x1    x 
c      F0 
clear all
x    k
1

(t  t0 ) 
%% Analytical solution (example 3.2.1)
Fo=30; k=1000; wn=3.16; zeta=0.1; to=0;  2    x2 
theta=atan(zeta/(1-zeta^2));  m m m 
wd=wn*sqrt(1-zeta^2);
t=0:0.01:12;
Heaviside=stepfun(t,to);% define Heaviside Step function for 0<t<12
xt = (Fo/k - Fo/(k*sqrt(1-zeta^2)) * exp(-zeta*wn*(t-to)) *
cos (wd*(t-to)- theta))*Heaviside(t-to);
plot(t,xt); hold on
%% Numerical Solution
xo=[0; 0];
ts=[0 12];
[t,x]=ode45('f',ts,xo);
plot(t,x(:,1),'r'); hold off
%---------------------------------------------
function v=f(t,x)
Fo=30; k=1000; wn=3.16; zeta=0.1; to=0; m=k/wn^2;
v=[x(2); x(2).*-2*zeta*wn + x(1).*-wn^2 + Fo/m*stepfun(t,to)];
Matlab Code

x0=[0;0]; 0.06

ts=[0 12]; 0.05

Displacement (x)
[t,x]=ode45('funct',ts,x0); 0.04

plot(t,x(:,1)) 0.03

0.02

0.01

0
0 2 4 6 8 1 1
Time (s)

function v=funct(t,x)
F0=30;
k=1000;
wn=3.16;
z=0.1;
t0=2;
m=k/(wn^2);
v=[x(2); x(2).*-2*z*wn+x(1).*-wn^2+F0/m*stepfun(t,t0)];
Problem 3.22
A wave consisting of the wake from a passing boat impacts a
seawall. It is desired to calculate the resulting vibration.
Figure P3.22 illustrates the situation and suggests a model.
Calculate the resulting response.
Numerical solution of Problem 3.22
%problem 3.19
m=1000; F (t )  
E=3.8e9;
F(t)  F (1  t ) 0  t  t0
A=0.03;  0 t0
L=2; 
 F0 t  t0
k=E*A/L;
t0=0.2;
F0=100;
global F0 k m t0
%numerical solution
x0=[0;0];
ts=[0 0.5];
[t,x]=ode45('f_3_19',ts,x0);
plot(t,x(:,1))

function v=f_3_19(t,x)
global F0 k m t0
A=x(2);
F=(((1-t./t0).*stepfun(t,0))-((1-t./t0).*stepfun(t,t0)))*F0/m;
B=(-k/m)*x(1)+F;
v=[A; B];
P3.22
3.9 Nonlinear Response Properties

Euler integration formula:

x(t i 1)  x(t i )  F(x(t i ))t  f(t i )t

Nonlinear term

Analytical solutions not available so we must interrogate the

numerical solution
Example 3.10.1 cubic spring subject to pulse input

mx(t )  cx(t )  kx(t )  k1 x3 (t )  1500 (t  t1 )  (t  t2 )

The state space form is:
x1 (t )  x2 (t )
x2 (t )  2n x2 (t )  n2 x1 (t )   x13 (t )  15 (t  t1 )  (t  t2 ) 
Nature of Response

Red (solid) is nonlinear response. Blue (dashed) is linear response

Is there any justification? Yes, hardening nonlinear spring.
The first part is due to IC.
Matlab Code
clear all
xo=[0.01; 1];
ts=[0 8];
[t,x]=ode45('f',ts,xo);
plot(t,x(:,1)); hold on % The response of nonlinear system
[t,x]=ode45('f1',ts,xo);
plot(t,x(:,1),'--'); hold off % The response of linear system
%---------------------------------------------
function v=f(t,x)
m=100; k=2000; c=20; wn=sqrt(k/m); zeta=c/2/sqrt(m*k); Fo=1500; alpha=3;
t1=1.5; t2=5;
v=[x(2); x(2).*-2*zeta*wn + x(1).*-wn^2 - x(1)^3.*alpha+ Fo/m*(stepfun(t,t1)-
stepfun(t,t2))];
%---------------------------------------------
function v=f1(t,x)
m=100; k=2000; c=20; wn=sqrt(k/m); zeta=c/2/sqrt(m*k); Fo=1500; alpha=0; t1=1;
t2=5;
v=[x(2); x(2).*-2*zeta*wn + x(1).*-wn^2 - x(1)^3.*alpha+ Fo/m*(stepfun(t,t1)-
stepfun(t,t2))];