CALCULATION OF SHOCK

RESPONSE SPECTRUM

VSB – TU of Ostrava
Faculty of Mechanical Engineering
Jiří Tůma & Petr Kočí
Motivation
„ Launch Vehicle Stage Separation Test

http://www.vibrationdata.com/SRS.htm
Outline
„ Motivation
… Research work for the Visteon Company
„ Calculation Shock Response Spectrum
… Single-Degree-of-Freedom Model
… Transfer Functions for Relative Velocity and Displacement Response
… Approximation of Continuous Transfer Function by Discrete Transfer
Function
… Approximation of the transfer function
G(s) = 1/(s + a) by G(z) = (β0 + β1z-1)/(1 + α1z-1)
… Assumptions about the function of x1(t) in the time interval
[ nTS , (n+1)TS ]
… Ramp invariant method
… Calculation SRS
… SRS of A Half-Sine Impulse
… Examples
… Conclusion
Shock Response Spectrum

A set of SDOF systems - substructures

Natural frequencies for SDOF system

fn1 < fn2 < fn3 < fn4 fnK
Amplitude of vibration
Response

Resonance gain Q = 10 Input impulse

A base structure exposed by vibration transient

 Single-Degree-of-Freedom Model
x2 Equation of motion m &x&2 = −c( x&2 − x&1 ) − k (x2 − x1 )

m Laplace transform
L{x1 (t )} = X 1 (s ), L{v1 (t )} = V1 (s ), L{a1 (t )} = A1 (s )
c k x1 L{x2 (t )} = X 2 (s ), L{v2 (t )} = V2 (s ), L{a2 (t )} = A2 (s )

x&1 = v1 , v&1 = a1 Transfer functions with parameters: mass, damping and stiffness

x&2 = v2 , v&2 = a2 X 2 (s ) cs + k X 2 ( jω ) jω c + k
G (s ) = = 2 G ( jω ) = =
X 1 (s ) ms + cs + k X 1 ( jω ) jω c + k − mω 2
2
Frequency response
V2 (s ) A2 (s )
10
cs + k cs + k
G (s ) = = 2 G (s ) = = 2
V1 (s ) ms + cs + k A1 (s ) ms + cs + k
0
Q
abs(G(jw))

10

Transfer functions with parameters: natural frequency and resonance gain

ωn s
+ ωn2
A 2 (s )
-2
10
1 k km 1 c Q
fn fn = , Q= , ξ= = =
1 2 3 4
2π m c 2Q 2 km A 1(s ) s 2 + ωn s + ω 2
10 10 10 10 n
Frequency [Hz] Q
Transfer Functions for Relative Velocity
and Displacement Response

Relative displacement response

Relative velocity response
spectrum

V2 − V1 − ms X 2 − X1 −m
= 2 = 2
A1 ms + cs + k A1 ms + cs + k

Relative displacement response

Pseudo velocity response spectrum spectrum expressed as equivalent
static acceleration

X 2 − X1 − mωn X 2 − X1 2 − mωn2
ωn = 2 ωn = 2
A1 ms + cs + k A1 ms + cs + k
Approximation of the Continuous Transfer
Function by the Discrete Transfer Function
Continuous time signals Sampled time signals

x2 (t ), v2 (t ), a2 (t ), x1 (t ), v1 (t ), a1 (t ) x2 [k ], v2 [k ], a2 [k ], x1 [k ], v1 [k ], a1 [k ] x1 [k ] = x1 (k TS )
t ... Continuous time k ... Index of a sample x2 [k ] = x2 (k TS )

Differential equation of motion Difference equation

m &x&2 = −c( x&2 − x&1 ) − k (x2 − x1 ) x2 [k ] = β 0 x1 [k ] + β1 x1 [k − 1] + β 2 x1 [k − 2] − α1 x2 [k − 1] − α 2 x2 [k − 2]

β 0 , β1 , β 2 , α1 , α 2 ... parameters
ωn s
+ ωn2
A 2 (s ) Q A2 ( z ) β 0 + β1 z −1 + β 2 z −2
= H ( z) = =
A 1(s ) s 2 + ωn s + ω 2 A1 (z ) 1 + α1 z −1 + α 2 z − 2
n
Q
Decomposition into partial fractions
ωn s
+ ωn2
A 2 (s ) Q R1 R1*
= = +
A 1(s ) s 2 + ωn s + ω 2 s − s1 s − s!*
n
Q
Approximation of the transfer function
G(s) = 1/(s + a) by G(z) = (β0 + β1z-1)/(1 + α1z-1)
The input signal is an unknown function except for a sequence of samples x1 [0], x1 [1], x1 [2], ...
t x1 [n] = x1 (n TS )
Convolution integral x2 (t ) = ∫ g (t − τ )x1 (τ ) dτ x1 [1] x1 [2] x1 [7]
x1 [0] x1 [3] x1 [6] x1 (t )
x1 [4] x1 [5]
0

Impulse response g (t ) = L−1{G (s )} = exp(− a t )

0 TS 2TS 3TS 4TS 5TS 6TS 7TS t
The output signal for t = nTS
nTS +TS
⎛ nTS nTS +TS
⎞
x2 (nTS + TS ) = ∫ exp(− a(nTS + TS − τ )) x1 (τ ) dτ = exp(− aTS ) ∫ exp(− a(nTS − τ )) x1 (τ ) dτ + ∫ (...) dτ ⎟
⎜
⎜ 0 ⎟
0 ⎝ nTS ⎠
TS

= exp(− aTS ) x2 (nTS ) + exp(− aTS ) ∫ exp(a u ) x1 (u + nTS ) du

0

The result depends on the assumption dealing with x1(t) in between nTS and (n+1)TS

x2 (nTS + TS ) = exp(− aTS ) x2 (nTS ) + β 0 x1 (nTS + TS ) + β1 x1 (nTS )

X 2 (z ) β 0 + β1 z −1
G(z ) = =
X 1 ( z ) 1 − exp(− aTS ) z −1 http://www.bth.se/tek/amt/nonlinear.nsf/(WebFiles)/F75E7709B9C6
F47DC12571D50033078C/$FILE/ISMA_2006_444.pdf
Assumptions about the function of x1(t)
in the time interval [ nTS , (n+1)TS ]
x1(t) Impulse invariant TS
x1(n) x1(n+1) x1 (t ) = x1 ((n + 1)TS )δ (t − (n + 1)TS ) ∫ exp(a u ) x (u + n T ) du =
0
1 S

(n+1)TS
= x1 ((n + 1)TS ) exp(a TS )
nTS

x1(t) Step invariant

TS
x1(n) x1(n+1) x1 (t ) = x1 (n TS ), n TS ≤ t < (n + 1)TS ∫ exp(a u ) x (u + n T ) du =
1 S
nTS (n+1)TS 0

= x1 (n TS ) (exp(a TS ) − 1) a
Centered step invariant
x1(t)
⎧ x (nT ), n TS ≤ t < (n + 0.5)TS
x1(n) x1(n+1)
x1 (t ) = ⎨ 1 S
nTS (n+1)TS ⎩ x1 ((n + 1)TS ), (n + 0.5)TS ≤ t < (n + 1)TS

x1(t) Ramp invariant

x1(n) x1(n+1) x1 (t ) = x1 (nTS ) + ( x1 ((n + 1)TS ) − x1 (nTS ))(t − n TS ) TS , n TS ≤ t < (n + 1)TS
nTS (n+1)TS
Ramp Invariant Method
Digital filter Accuracy
A2 (z ) β 0 + β1 z −1 + β 2 z −2 Frequency response
H ( z) = = 2

A1 ( z ) 1 + α1 z −1 + α 2 z −2
10

Coefficients 1
10 Continuous system
β 0 = 1 − exp(− A) sin( B) / B G ( jω ) =
0

β1 = 2 exp(− A)[sin( B) / B − cos( B)]

10
jω ωn Q + ωn2
=
jω ωn Q + ωn2 − ω 2

abs(G(jw))
β 2 = exp(−2 A) − exp(− A) sin( B) / B -1
10

α1 = −2 exp(− A) cos( B) -2
10

α 2 = exp(−2 A) Bilinear transform

-3 Ramp with taking on
where 10
invariant account prewarping
ωnTS 1 method frequency
A= , B = ωnTS 1 − -4
2Q 4Q 2 10
1 2 3 4
10 10 10 10
Frequency [Hz]
fsfs2/ /2
Calculation SRS
10000 Shock Response Spectrum Peak values of responses
SRS as a function of the natural
G 1000 frequency.

100
10 100 1000 10000 Natural frequency in Hz
6000 6000 6000 6000
4000 4000 4000 4000
2000 2000 2000 2000 A set of responses
G 0
G 0 G 0 G 0 to the input signal
-2000 -2000 -2000 -2000
0,0 0,1 0,2 0,3 0,4 0,5 0,0 0,1 0,2 0,3 0,4 0,5 0,0 0,1 0,2 0,3 0,4 0,5 0,0 0,1 0,2 0,3 0,4 0,5

Time [s] Time [s] Time [s] Time [s]

fn1 = 10 Hz fn2 = 100 Hz fn3 = 1000 Hz fn4 = 10000 Hz

Substructures

Input signal
Input impulse
4000
3000
2000
G
1000
0
Resonance gain Q = 10 A base structure -1000
0,0 0,1 0,2 0,3 0,4 0,5
Time [s]
SRS of A Half-Sine Impulse
Half-sine impulse (11 ms) Peak values
12
10
Absolute value of SDOF-system responses
8
6
G

4
2
G
abs(G)
0 24
20
0,00 0,01 0,02 0,03 0,04 0,05 18
15 Index of
Time [s] 12 the SDOF natural
10
5 6 frequency Peak values
Input impulse 0 0 as a function of fn
0,00 0,02 0,04 0,06 0,08 0,10 0,12
Time [s] Shock Response Spectrum
100,0

10,0
Shock

G
Response
1,0
Spectrum
0,1
1 10 100 1000
Frequency [Hz]
Signal Analyzer Script
‘Shock Response Spectrum’;
‘CrLf’;
ymax = [ ]; ff = [ ]; fmin = 1; fmax = 1000; n = 90;
qv = (fmax/fmin)^(1/n); T = 1/get(input1,'freq'); Q = 10;
for(i=0;i<n;i=i+1)
{
fn = fmin*qv^i; ff = [ff,fn]; wn = 2*pi*fn;
A = wn*T/2/Q;
B = wn*T*sqr(1-1/4/Q/Q);
b0 = 1-exp(-A)*sin(B)/B;
b1 = 2*exp(-A)*(sin(B)/B-cos(B));
b2 = exp((-2)*A)-exp(-A)*sin(B)/B;
a1 = (-2)*exp((-1)*A)*cos(B);
a2 = exp((-2)*A);
BB = [b0,b1,b2];
AA = [-a1,-a2];
y = filter(input1,BB,AA);
yy = max(y);
ymax = [ymax,yy];
};
save(ff);
save(ymax);
Metal hammer and plastic shield
Examples: impact of metal to plastic
Acceleration during impact of a car light system and corresponding SRS
Time History : Acc Shock Response Spectrum
400 10000
300
200 1000
100
0 100
G

G
-100
-200 10
-300
-400 1
0,0 0,1 0,2 0,3 0,4 1 10 100 1000 10000
Time [s] Natural frequency [Hz]

Acceleration during impact of a steel hammer to the plastic shield and corresponding SRS
Time History : Acc - Hamer Shock Response Spectrum
1000 10000

0
1000
-1000

m/s2
m/s2

-2000 100
-3000

-4000 10
0,00 0,05 0,10 0,15 0,20 1 10 100 1000 10000 100000
Natural frequency [Hz]
Time [s]
Conclusion
„ The paper presents the method of calculation the shock response spectrum, which is
corresponding to an acceleration signal exciting the resonance vibration of
substructures. SRS determines the maximum or minimum of the substructure
acceleration response as a function of the natural frequencies of a set of the single-
degree-of-freedom systems modeling the mentioned substructures.

„ The vibration (or shock) is recorded in digital form, commonly as acceleration signal.
The single-degree-of-freedom systems are approximated by an IIR digital filter and
the filter response to the sampled acceleration signal may be easily calculated. This
shock response spectrum shows how the individual component of the impulse signal
excites the mechanical structure to resonate.