2. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 2
Transfer path analysis
Application background
Structureborne noise - structural path between source and receiver
from engine
from suspension
from exhaust, gearbox, drive-line
Airborne noise - no structural path between source and receiver
engine noise
aerodynamic noise
radiated noise from panels
What are the main contributors, the important transfer paths ?
Treated in ASQ
3. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 3
Transfer path analysis
Principles
Simplify noise/vibration path into source-receiver-transfer system
Transfer
- structure
- sound field
Receiver
- human ear
- steering wheel
Source
- connection between
excitation source and
target structure
4. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 4
Quantification of source-receiver-transfer system
Transfer path analysis
Principles
Transfer
- vibro-(acoustic) FRF
- Pa/N or (m2/s)/N
Receiver
- pressure or acceleration
- Pa or m2/s
Source
- force
- N
5. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 5
Transfer path analysis
Principles
Structure-borne noise/vibration breakdown and ranking
Preceiver = Pi
Partial contribution Pi
Pi = Fi . Hi
PF
Source : N Transfer : Pa/N
Xreceiver = Xi
Xi = Fi . Hi
XF
Source : N Transfer : (m/s2)/N
:
:
:
:
6. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 6
Transfer path analysis
Two applications
Transfer path ranking, contribution analysis
Preceiver = FiHi
PF
source or transfer problem ?
post processing for FBS, ASQ, FE calculations
Multiple force estimation methods
Fi ?
7. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 7
Transfer path analysis
Source identification
Complex dynamic stiffness method
Acceleration measurements
Mount rates
Full matrix inversion
Acceleration measurements
FRF measurements (full matrix)
Driving point inversion
Acceleration measurements
Driving point FRF
measurements
8. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 8
Transfer path analysis
Complex dynamic stiffness method
t
x
s
x
t
s X
X
K
F
Acceleration measurements on both sides of
mount
Stiffness characteristics of mount
9. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 9
Transfer path analysis
Complex dynamic stiffness determination
determination of stiffness in three direction
rotational degrees of freedom neglected
cross-coupling effects neglected
Automatic integration/differentiation of mount stiffness data
Behavior of mounts is not very linear
apply correct pre-load
temperature effects may be important
adapt excitation according to actual operational conditions
X
F
K
)
(
10. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 10
Transfer path analysis
Complex dynamic stiffness determination
Mount
Ground
Mass (pre-load)
F
vert
x
hor
x
11. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 11
Transfer path analysis
Inverse method
F1
x4
x3
T14
T13
F2
T24
T23
2
24
1
14
2
1
4
2
23
1
13
2
1
3
.
.
,
.
.
,
F
T
F
T
F
F
f
x
F
T
F
T
F
F
f
x
F : excitation
X : response
T : transfer function
2
1
24
14
23
13
4
3
F
F
T
T
T
T
x
x
4
3
1
24
14
23
13
2
1
x
x
T
T
T
T
F
F
12. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 12
Transfer path analysis
Inverse method : example
Operational measurements
acceleration at three points
Laboratory measurements
full FRF matrix between
excitation at point 1 and
response at the other points
1
x
1001
x
1002
x
1
1001
1002
Finput
13. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 13
1001
1002
F1
oper ?
Transfer path analysis
Inverse method : example
oper
oper
oper
oper
x
x
x
F
x
F
x
F
x
F
1002
1001
1
1
1
1002
1
1001
1
1
1
14. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 14
Transfer path analysis
Singular value decomposition
Singular value decomposition of transfer function matrix
n
i
n
i
i
m
i
nxn
mxn
mxm
mxn
V
U
V
U
H
1
*
*
0
0
0
0
0
0
0
2
1
n
mxn
M : number of responses
N : number of sources
M N
15. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 15
Transfer path analysis
Singular value decomposition
Example : 28x15 accelerance FRF matrix
16. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 16
Transfer path analysis
Singular value decomposition
n
i
m
i
i
n
i
T
mxm
nxm
nxn
mxm
mxn
T
nxn
nxm
U
V
U
V
U
V
H
1
*
1
1
1
1
1
1
Matrix inversion by singular value decomposition
0
0
0
0
0
0
0
0
0
0
1
1
2
1
1
1
n
nxm
17. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 17
Matrix inversion by singular value decomposition
Over determination allowed (M N)
Least squares estimate
Calculation of condition number : upper bound for
multiplication coefficient of relative error
Transfer path analysis
Singular value decomposition
n
number
condition
1
_
18. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 18
Improving the matrix condition
Delete p singular values : condition number drops
Three possibilities : number, relative, absolute
Overdetermination : 2/1
Transfer path analysis
Singular value decomposition
p
n
number
condition
1
_ n-p > n
21. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 21
Transfer path analysis
Relation condition number / overdetermination
22. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 22
Transfer path analysis
Selection of overdetermination points
Overdetermination : 2n responses to estimate n forces
Location of OD points
good coherence with signal in transfer path
not to close to transfer path : no extra information
3 extra points in one direction better then 1point measured in 3
directions
Pressure response can be included in matrix inversion
23. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 23
Transfer path analysis
SVD example : fully trimmed car
Original FRFs + condition number Elements from inverted matrix
24. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 24
Transfer path analysis
SVD example : body in white
Original FRFs + condition number Elements from inverted matrix
25. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 25
Transfer path analysis
Transfer functions
Measured in uncoupled condition
engine, drive-line removed
suspension removed
Follows from substructuring formulation
P=HA+BFi=HB([HA+HB+K-1]-1HAFi)=HBFc
HA HB
K
source target
Fc
Fi P
TPA
26. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 26
Transfer path analysis
Transfer functions
Always check
Linearity
Coherence
Reciprocity
Impact or shaker testing
Coherence with hammer excitation
Misalignment of shaker
Small errors become important after matrix inversion
Excitation
direction
27. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 27
Transfer path analysis
Vibro-acoustic reciprocity
i
j
j
i
PF
F
P
Q
x
H
F1
F2
Pj
1
x
2
x
Qj
Reduction of measurement effort
j
i
Q
x
unit
define )
(
28. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 28
Transfer path analysis
Volume velocity source
Volume velocity sources
Sufficiently excite cavity and/or structure : high power
Not influenced by acoustic boundary
Ease of use : constant calibration factor
Omni-directional
29. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 29
Transfer path analysis
Features : easy model definition
Model definition based on PID and SID annotations
mic:1:S
body:2
body:1
Data type Primary ID Secondary ID
Force body:1:+Z -
body:2:+Z -
Pressure mic:1:S -
FRF mic:1:S body:1:+Z
mic:1:S body:2:+Z
30. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 30
x1 x2 ... xr
x11…x21…xr1
PCA
Force
id.
TPA
x12…x22…xr2 x1m…x2m…xrm
x11…x21…xr1 x12…x22…xr2 x1m…x2m…xrm
f11…f21…fn1 f12…f22…fn2 f1m…f2m…fnm
phenomenon 1 phenomenon 2 phenomenon m
Transfer path analysis
Features: support of multi-reference problems
Support of single
reference problems :
engine noise e.g.
Multi-reference
problems (road
noise e.g.) require
Principal
Component Analysis
31. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 31
Transfer path analysis
Features : structure and air-borne problems
Study contribution of structure and air-borne noise in one
analysis : Preceiver = Fi . Hi
PF+ Qi’ . Hi’
AA
Pre-processor required for Panel Contribution Analysis
32. LMS
INTERNATIONAL
CMC - TPA training - May, 2005 32
Transfer path analysis
Extensions to TPA
Easy data grouping mechanism in post-processing
grouping based on PID
simplifying visualization
Frequency spectra, ordercuts and autopowers (new) supported
Automatic squaring of FRFs in case of autopowers
Condition calculation : specify frequency range