2.
4 Base Excitation
• Important class of vibration analysis
– Preventing excitations from passing
from a vibrating base through its mount
into a structure
• Vibration isolation
– Vibrations in your car
– Satellite operation
– Disk drives, etc.
© D. J. Inman
1/51
University of Michigan
FBD of SDOF Base Excitation
System Sketch
System FBD
x(t) m
m
k c
y(t) base
© D. J. Inman
2/51
University of Michigan
SDOF Base Excitation (cont)
For a car,
The steady-state solution is just the superposition of the
two individual particular solutions (system is linear).
© D. J. Inman
3/51
University of Michigan
Particular Solution (sine term)
With a sine for the forcing function,
Use rectangular form to
make it easier to add
the cos term
© D. J. Inman
4/51
University of Michigan
Particular Solution (cos term)
With a cosine for the forcing function, we showed
© D. J. Inman
5/51
University of Michigan
Magnitude X/Y
Now add the sin and cos terms to get the
magnitude of the full particular solution
© D. J. Inman
6/51
University of Michigan
The relative magnitude plot
of X/Y versus frequency ratio: Called the
Displacement Transmissibility
40
ζ =0.01
30 ζ =0.1
ζ =0.3
ζ =0.7
20
X/Y (dB)
10
-10
-20
0 0.5 1 1.5 2 2.5 3
Frequency ratio r
Figure 2.13
© D. J. Inman
7/51
University of Michigan
From the plot of relative Displacement
Transmissibility observe that:
• X/Y is called Displacement Transmissibility
Ratio
• Potentially severe amplification at resonance
• Attenuation for r > sqrt(2) Isolation Zone
• If r < sqrt(2) transmissibility decreases with
increase in damping ratio Amplification Zone
• If r >> 1 then transmissibility increases with
damping ratio Xp~2Yζr
© D. J. Inman
8/51
University of Michigan
Next examine the Force Transmitted to the
mass as a function of the frequency ratio
From FBD
Substituting for ωn = k/m and r = ωn/ω x(t) m
FT
k
c
y(t) base
© D. J. Inman
9/51
University of Michigan
Plot of Force Transmissibility (in dB)
versus frequency ratio
40
ζ =0.01
ζ =0.1
30
ζ =0.3
ζ =0.7
20
F/kY (dB)
10
-10
-20
0 0.5 1 1.5 2 2.5 3
Frequency ratio r
© D. J. Inman
Figure 2.14
10/51
University of Michigan
Figure 2.16 Comparison between force
and displacement transmissibility
Force
Transmissibility
Displacement
Transmissibility
© D. J. Inman
11/51
University of Michigan
Example 2.4.2: Effect of speed
on the amplitude of car vibration
2π 2π V
ωb
= =
τ λ
λ = 6 m = vτ
Figure 2.17
© D. J. Inman
12/51
University of Michigan
Model the road as a sinusoidal input to
base motion of the car model
Approximation of road surface:
2π 2π V
, and =
ωb =
τ λ
From the data give, determine the frequency and
damping ratio of the car suspension:
© D. J. Inman
13/51
University of Michigan
From the input frequency, input amplitude,
natural frequency and damping ratio use
equation (2.70) to compute the amplitude of
the response:
ωb 5.818
r
= =
ω 6.303
n
What happens as the car goes faster? See Table 2.1.
© D. J. Inman
14/51
University of Michigan
Example 2.4.3: Compute the force
transmitted to a machine through base
motion at resonance
From (2.77) at r =1:
c 900
k: ζ
From given m, c, and= = ≅ 0.04
2 km 2 40, 000 ⊗ 3000
From measured excitation Y = 0.001 m:
© D. J. Inman
15/51
University of Michigan
2.5 Rotating
Figure 2.19
Unbalance
• Gyros
• Cryo-coolers
• Tires m0
• Washing machines e
Machine of total mass m i.e. m0
included in m
e = eccentricity k
mo = mass unbalance c
=rotation frequency
© D. J. Inman
16/51
University of Michigan
Rotating Unbalance (cont)
Rx
m0 What force is imparted on the
structure? Note it rotates
e
with x component:
θ Ry
From sophomore dynamics,
© D. J. Inman
17/51
University of Michigan
Rotating Unbalance (cont)
The problem is now just like any other SDOF
system with a harmonic excitation
x(t)
k
c Note the influences on the
forcing function (we are assuming that
the mass m is held in place in the y direction as indicated
in Figure 2.19)
© D. J. Inman
18/51
University of Michigan
Rotating Unbalance (cont)
• Just another SDOF oscillator with a
harmonic forcing function
• Expressed in terms of frequency ratio r
© D. J. Inman
19/51
University of Michigan
Figure 2.21: Displacement magnitude vs
frequency caused by rotating unbalance
© D. J. Inman
20/51
University of Michigan
Example 2.5.1:Given the deflection at resonance (0.1m),
ζ = 0.05 and a 10% out of balance, compute e and the amount of added
mass needed to reduce the maximum amplitude to 0.01 m.
At resonance r = 1 and
Now to compute the added mass, again at resonance;
Use this to find Δm so that X is 0.01:
Here m0 is 10%m or 0.1m
© D. J. Inman
21/51
University of Michigan
Example 2.5.2 Helicopter rotor unbalance
Given
Fig 2.22
k = 1×105 N/m
mtail = 60 kg
mrot = 20 kg
m0 = 0.5 kg Fig 2.23
ζ = 0.01
e = 0.15 m
Compute the deflection at
1500 rpm and find the rotor
speed at which the deflection is maximum
© D. J. Inman
22/51
University of Michigan
Example 2.5.2 Solution
The rotating mass is 20 + 0.5 or 20.5. The stiffness is provided by the
Tail section and the corresponding mass (page 52) is that determined in
the example of a heavy beam. So the system natural frequency is
The frequency of rotation is
© D. J. Inman
23/51
University of Michigan
Now compute the deflection at r =
2.91 and ζ =0.01 using eq (2.84)
At around r = 1, the max deflection occurs:
At r = 1:
© D. J. Inman
24/51
University of Michigan
2.6 Measurement Devices
• A basic transducer
used in vibration
measurement is the
accelerometer. Figure 2.24
• This device can be
modeled using the
base equations
developed in the
previous section Here, y(t) is the measured
response of the structure
© D. J. Inman
25/51
University of Michigan
Base motion applied to
measurement devices
Accelerometer
These equations should be familiar
from base motion.
Here they describe measurement!
Strain Gauge
© D. J. Inman
26/51
University of Michigan
Magnitude and sensitivity
plots for accelerometers.
Effect of damping on
proportionality constant
Fig 2.27
Magnitude plot showing
Regions of measurement
In the accel region, output voltage is
nearly proportional to displacement
© D. J. Inman Fig 2.28
27/51
University of Michigan
2.7 Other forms of damping
These various other forms of damping are all nonlinear. They can
be compared to linear damping by the method of “equivalent viscous
damping” discussed next. A numerical treatment of the exact response
is given in section 2.9.
© D. J. Inman
28/51
University of Michigan
The method of equivalent viscous
damping: consists of comparing the energy
dissipated during one cycle of forced response
Assume a stead state resulting from a harmonic
input and compute the energy dissipated per one cycle
The energy per cycle for a viscously damped system is
(2.101)
© D. J. Inman
29/51
University of Michigan
Next compute the energy dissipated
per cycle for Coulomb damping:
Here we let u = ωt and du =ωdt and split up the
integral according to the sign changes in velocity.
Next compare this energy to that of a viscous system:
(2.105)
This yields a linear viscous system dissipating the same amount of
energy per cycle.
© D. J. Inman
30/51
University of Michigan
Using the equivalent viscous damping
calculations, each of the systems in Table 2.2
can be approximated by a linear viscous system
In particular, ceq can be used to derive
amplitude expressions. However, as
indicated in Section 2.8 and 2.9 the response
can be simulated numerically to provide
more accurate magnitude and
response information.
© D. J. Inman
31/51
University of Michigan
Hysteresis: an important concept
characterizing damping
• A plot of displacement
versus spring/damping
force for viscous
damping yields a loop
• At the bottom is a stress
strain plot for a system
with material damping
of the hysteretic type
• The enclosed area is
equal to the energy lost
per cycle
© D. J. Inman
32/51
University of Michigan
The measured area yields the energy
dissipated. For some materials, called
hysteretic this is
Here the constant β, a measured quantity is called
the hysteretic damping constant, k is the stiffness
and X is the amplitude.
Comparing this to the viscous energy yields:
© D. J. Inman
33/51
University of Michigan
Hysteresis gives rise to the
concept of complex stiffness
Substitution of the equivalent damping coefficient
and using the complex exponential to describe a
harmonic input yields:
© D. J. Inman
34/51
University of Michigan
2.8 Numerical Simulation and
Design
• Four things we can do computationally to help
solve, understand and design vibration
problems subject to harmonic excitation
• Symbolic manipulation
• Plotting of the time response
• Solution and plotting of the time response
• Plotting magnitude and phase
© D. J. Inman
35/51
University of Michigan
Symbolic Manipulation
Let
and
What is
This can be solved using Matlab, Mathcad or Mathematica
© D. J. Inman
36/51
University of Michigan
Symbolic Manipulation
Solve equations (2.34) using Mathcad symbolics :
Enter this
Choose evaluate
under symbolics to
get this
© D. J. Inman
37/51
University of Michigan
In MATLAB Command Window
>> syms z wn w f0
>> A=[wn^2-w^2 2*z*wn*w;-2*z*wn*w wn^2-w^2];
>> x=[f0 ;0];
>> An=inv(A)*x
An =
[ (wn^2-w^2)/(wn^4-2*wn^2*w^2+w^4+4*z^2*wn^2*w^2)*f0]
[ 2*z*wn*w/(wn^4-2*wn^2*w^2+w^4+4*z^2*wn^2*w^2)*f0]
>> pretty(An)
[ 2 2 ]
[ (wn - w ) f0 ]
[ --------------------------------- ]
[ 4 2 2 4 2 2 2 ]
[ wn - 2 wn w + w + 4 z wn w ]
[ ]
[ z wn w f0 ]
[2 ---------------------------------]
[ 4 2 2 4 2 2 2]
[ wn - 2 wn w + w + 4 z wn w ]
© D. J. Inman
38/51
University of Michigan
Magnitude plots: Base Excitation
%m-file to plot base excitation to mass vibration
r=linspace(0,3,500);
ze=[0.01;0.05;0.1;0.20;0.50];
X=sqrt( ((2*ze*r).^2+1) ./ ( (ones(size(ze))*(1-r.*r).^2) + (2*ze*r).^2) );
figure(1)
plot(r,20*log10(X))
40
The values of ζ can
then be chosen ζ =0.01
30 Design
directly off of the plot. ζ =0.05
value
20 ζ =0.1
For Example: ζ =0.2
X/Y (dB)
If the T.R. needs to be 10 ζ =0.5
less than 2 (or 6dB)
and r is close to 1 0
then ζ must be more
than 0.2 (probably -10
about 0.3).
-20
0 0.5 1 1.5 2 2.5 3
Frequency ratio r
© D. J. Inman
39/51
University of Michigan
Force Magnitude plots: Base Excitation
%m-file to plot base excitation to mass vibration
r=linspace(0,3,500);
ze=[0.01;0.05;0.1;0.20;0.50];
X=sqrt( ((2*ze*r).^2+1) ./ ( (ones(size(ze))*(1-r.*r).^2) + (2*ze*r).^2) );
F=X.*(ones(length(ze),1)*r).^2;
figure(1)
plot(r,20*log10(F))
40
ζ=0.01
30 ζ=0.05
ζ=0.1
20 ζ=0.2
FT /kY (dB)
ζ=0.5
10
-10
-20
0 0.5 1 1.5 2 2.5 3
© D. J. Inman Frequency ratio r
40/51
University of Michigan
Numerical Simulation
We can put the forced case:
Into a state space form
© D. J. Inman
41/51
University of Michigan
Numerical Integration
Using the ODE45 function Zero initial conditions
>>TSPAN=[0 10]; 5
>>Y0=[0;0]; 4
>>[t,y] =ode45('num_for',TSPAN,Y0);
3
>>plot(t,y(:,1))
2
Displacement (m)
Including forcing 1
function Xdot=num_for(t,X) 0
m=100;k=1000;c=25; -1
ze=c/(2*sqrt(k*m));
-2
wn=sqrt(k/m);
w=2.5;F=1000;f=F/m; -3
f=[0 ;f*cos(w*t)]; -4
A=[0 1;-wn*wn -2*ze*wn];
Xdot=A*X+f; -5
0 2 4 6 8 10
Time (sec)
© D. J. Inman
42/51
University of Michigan
Example 2.8.2: Design
damping for an electronics
model
• 100 kg mass, subject to 150cos(5t) N
• Stiffness k=500 N/m, c = 10kg/s
• Usually x0=0.01 m, v0 = 0.5 m/s
• Find a new c such that the max
transient value is 0.2 m.
© D. J. Inman
43/51
University of Michigan
Response of the board is;
transient exceeds design specification value
0.4
0.2
Displacement (m)
-0.2
-0.4
0 10 20 30 40
Time (sec)
© D. J. Inman Figure 2.33
44/51
University of Michigan
To run this use the following file:
function Xdot=num_for(t,X)
m=100;k=500;c=10;
Create function ze=c/(2*sqrt(k*m));
wn=sqrt(k/m);
to model forcing w=5;F=150;f=F/m;
f=[0 ;f*cos(w*t)];
A=[0 1;-wn*wn -2*ze*wn];
Xdot=A*X+f;
>>TSPAN=[0 40];
>> Y0=[0.01;0.5];
Matlab >>[t,y] = ode45('num_for',TSPAN,Y0);
command >> plot(t,y(:,1))
window >> xlabel('Time (sec)')
>> ylabel('Displacement (m)')
>> grid
Rerun this code, increasing c each time until a
response that satisfies the design limits results.
© D. J. Inman
45/51
University of Michigan
Solution: code it, plot it and change c
until the desired response bound is
obtained.
0.3
Meets amplitude limit when c=195kg/s
0.2
Displacement (m)
0.1
-0.1
0 10 20 30 40
Time (sec) Figure 2.34
© D. J. Inman
46/51
University of Michigan
2.9 Nonlinear Response
Properties
• More than one equilibrium
• Steady state depends on initial conditions
• Period depends on I.C. and amplitude
• Sub and super harmonic resonance
• No superposition
• Harmonic input resulting in nonperiodic motion
• Jumps appear in response amplitude
© D. J. Inman
47/51
University of Michigan
Computing the forced response
of a non-linear system
A non-linear system has a equation of motion given
by:
Put this expression into state-space form:
In vector form:
© D. J. Inman
48/51
University of Michigan
Numerical form
Vector of nonlinear dynamics Input force vector
Euler equation is
© D. J. Inman
49/51
University of Michigan
Cubic nonlinear spring (2.9.1)
2
1
Displacement (m)
-1
Non-linearity included
Linear system
-2
0 2 4 6 8 10
Time (sec)
Subharmonic resonance
© D. J. Inman
50/51
University of Michigan
Cubic nonlinear spring near
resonance
3
2
Displacement (m)
-1
-2 Non-linearity included
Linear system
-3
0 2 4 6 8 10
Time (sec)
Response near linear resonance
© D. J. Inman
51/51
University of Michigan
