DEPARTMENT OF CIVIL ENGINEERING,

IIT DELHI

D
IIT
LOW-FREQUENCY VIBRATION (GLOBAL
DYNAMIC) TECHNIQUES FOR

4
86
STRUCTURAL HEALTH MONITORING
VL
Dr. Suresh Bhalla
Professor
C

Department of Civil Engineering

Indian Institute of Technology Delhi, Hauz Khas, New Delhi 110 016

Website: [Link]
Email: sbhalla@[Link] Tel: (011) 2659-1040 Fax: (011) 2658-1117
 SHM TECHNIQUES

D
IIT
LOCAL GLOBAL
Involving application of

4
86
• Global Static Response
NDE techniques • Global Dynamic Response
VL
• Ultrasonic Wave Propagation

• Acoustic Emission
C

• Eddy Currents

• Impact Echo

23/01/2019 2
 GLOBAL SHM TECHNIQUES
Global Dynamic Techniques
(Dynamic Response Based Techniques)
F(t)

D
IIT
Measure structural Modal or
dynamic response Structural

4
Parameters

86
VL
C

F(t)
Finite Model
Element Updating
Model Dynamic Test

23/01/2019 3
 CONTENTS
• EXPERIMENTAL MODAL ANALYSIS

D
(NON MATHEMATICAL TREATMENT)

IIT
• GLOBAL VIBRATION TEHNIQUES FOR

4
86
STRUCTURAL HEALTH MONITORING
VL
C

• USE OF MODERN SMART SENSORS IN

GLOBAL VIBRATION TECHNIQUES
 EXPERIMENTAL MODAL ANALYSIS
Modal analysis is a process whereby the structure is described in terms
of its natural dynamic characteristics, namely the natural frequencies,
mode shapes and damping, by applying an external excitation and
measuring structure’s response at representative locations.

D
Voltage measuring

IIT
device
Accelerometer

4
ISMB 150

86
Armature
coil
Stinger
Function
VL
Generator
Shaker
C

Let us apply a sinusoidal signal of constant amplitude, but vary the

frequency gradually.
 TIME AND FREQUENCY DOMAINS

Time domain
response

D
IIT
4
Natural frequencies Frequency domain

86
Natural frequencies
response
VL
Fast Fourier
Transform (FFT)
C

The frequency domain response is much

easier to evaluate as compared to the time
domain.
 MODE SHAPES
MODE 2
MODE 1 MODE 3

D
IIT
4
86
VL
C

Mode shape is the deformation pattern of the structure when

excited at a particular natural frequency.
 DAMPING
Damping is the energy dissipation propertiy of a material or system
under vibration/ dynamic loading.

Half power band width method Logarithmic decrement method

D
IIT
4
86
VL
C

 2  1  xn  
   ln  
2
2 n  xn 1 
 FREQUENCY RESPONSE FUNCTION (FRF)
It is the ratio of the output response (displacement/ velocity/ acceleration) of a
structure (undergoing vibration) at a point to the force applied at the same or
other point, at a particular frequency. Both the force and the response are
simultaneously measured.

D
F (t )  Fo cos t  jFo sin t  Fo e jt

IIT
Y (Imaginary Axis)
We may just consider the real component and
Force phasor
ignore the imaginary, or vice versa, use of

4
Fo exponential function makes mathematical

86
t Fo sin t representation simpler and compact…..
X (Real Axis)

VL
Ro
R  Ro cos(t   )  jRo sin(t   )  uo e j (t  )
C

Response phasor

Fo cos t Ø = Phase difference between force

and displacement Why???
R Ro e j (t  ) R o  j Unique function of
FRF (ω) : h( )   j t
 e structural stiffness,
F Fo e Fo damping and mass
 FREQUENCY RESPONSE FUNCTION (FRF)
Respnose could be displacement/ velocity/ acceleration/ strain etc…
Being sinusoidal in nature, displacement, velocity and acceleration are
mathematically related and can easily substitute one another
Displacement u  uo e j (t  )

D
du
Velocity u   uo je j (t  )

IIT
dt

 uo ej / 2e j (t  )

4
86
Leads displacement by (π/2)
 uoe j (t   / 2)
VL
du
Acceleration  
u  uo 2e j (t  )  uo 2e j (t   )
C

dt
Acceleration is generally measured in majority Leads displacement by π
of the modal tests.

In summary, u  ju u   u
 2
FREQUENCY RESPONSE FUNCTION (FRF)
In general, FRF is a complex number, consisting of real and imaginary
parts. It is generally plotted over a frequency range
(magnitude/ real/ imaginary):

D
IIT
4
Magnitude

86
VL
C

Frequency
FRF can be used to predict the response of the structure against a
known force excitation.
FREQUENCY RESPONSE FUNCTION (FRF)

D
IIT
4
86
VL
C
 MODE SHAPES FROM FRF
hij = Response at ‘i’ due to force applied at ‘j’.
i, j = 1, 2, 3.

h11 = Response at “1’

D
due to force at ‘1’

IIT
4
86 h12 = Response at “1’
VL
due to force at ‘2’
C

h13
 MODE SHAPES FROM FRF
Similarly, we can experimentally derive:
Response at 2, point of
h21 h22 h23 impact varied.

D
and also
Response at 3, point of

IIT
h31 h32 h33 impact varied.

4
86
IF THE STRUCTURE IS LINEARLY ELASTIC,
VL
WHAT IS THE RELATION BETWEEN: hij AND hji
C

MUST BE EQUAL SINCE MAXWELL-BETTI’S THEOREM STATES:

The displacement at a point due to unit load at another point is equal to
that at the other point due to unit load applied at the first point.
 MODE SHAPES FROM FRF

D
h11 h12 h13 h11 h12 h13

IIT
4
86
h21 h22 h23 h21 h22 h23
VL
C

h31 h32 h33 h31 h32 h33

Real part of FRF Imaginary part of FRF

 MODE SHAPES FROM FRF
In general, any one row can provide us all the mode shapes. Let us use
the imaginary part of the third row i.e h3j i.e the measurement point is
same, excitation point is varied. In all, only one sensor required.

D
h31 =h13 h32 =h23
h33 =h33

IIT
f1
MODE 1 f1

4
f1

86
f2
f2 f2
VL
f2
MODE 2 f2
C

MODE 3 ???
Try yourself….
 MODE SHAPES FROM FRF
Since hij = hji
Any one column can also be used in place of a row.

If say the third column is used

i.e hj3 i.e three measurement

D
points, one excitation point.

IIT
We can use three sensors with
simultaneous measurements

4
86
and thus acquire all data in a
single go.
VL
Or can also repeat three times with
C

one sensor, each time changing

position of the sensor. This gives third
column…which is same as third row.
 A WORD OF CAUTION!
Let us try row 2 to obtain a mode shape, say the second mode

h21 h22 h23

D
IIT
We cannot!

4
We do not see any peak at the frequency corresponding to mode 2. Why?

86
VL
C

Measurement point 2 is a nodal point for mode 2 i.e. ZERO amplitude of vibration.

Hence, nodal points should be preferably avoided as measurement or

excitation points.
 BUT ROW 1 AND 2 ARE FINE!

D
h11 h12 h13 h11 h12 h13

IIT
4
86
h21 h22 h23 h21 h22 h23
VL
C

h31 h32 h33 h31 h32 h33

Real part of FRF Imaginary part of FRF

 WORD OF CAUTION!
Measurement at the mid point of a simply supported beam.

D
IIT
4
MODE 1
86 MODE 2 MODE 3
VL
C

Modes 1, 3, 5, 7….. will be identified. In fact very good point

Modes 2, 4, 6, 8….. will be missed out.
 SHAKER vs IMPACT TEST
Voltage
measuring
Accelerometer device

Function

D
Stinger Generator
Shaker

IIT
Generally one point for
Single point excitation, multiple response, multiple points for

4
response….hence a column of excitation……hence a row of FRF

86
FRF Theoretically no difference….but practically…..
Need to make elaborate Very quick
VL
arrangements
Frequency of excitation can Input excitation frequency
C

be very well controlled. dependent on the hardness of

the hammer tip, harder the tip,
Potential stiffening effect of wider the frequency range
shaker and stinger system in
shaker test

Effect of location of accelerometers, especially when same set is moved

around the structure, applicable in both impact test and shaker test. Can be
minimized by “dummy accelerometers”
 Suggested

D
reading
IIT
4
86
VL
Avitable (2001) pp. 1 to 6, Farrar and Jauregui (1998)
C

23/01/2019 22
 CONTENTS
• EXPERIMENTAL MODAL ANALYSIS

D
(NON MATHEMATICAL TREATMENT)

IIT
• GLOBAL VIBRATION TEHNIQUES FOR

4
86
STRUCTURAL HEALTH MONITORING
VL
C

• USE OF MODERN SMART SENSORS IN

GLOBAL VIBRATION TECHNIQUES
 CHANGE IN STIFFNESS METHOD
(Zimmerman and Kaouk, 1994)
Submitted in June 1992

Based on the principle that damage alters the stiffness matrix

D
of the structure

IIT
The stiffness matrix is derived from the measured natural

4
frequencies and the mode shapes of the structure.

86
The stiffness matrix [K] was expressed by the researchers in terms of the
VL
mode shape matrix [φ], the mass matrix [M], and the modal stiffness
matrix [Ω] (diagonal ωi2 ) as
C

 n 2 T
[ K ]  [ M ][ ][][ ] [ M ]  [ M ]  i ii [ M ]
T

 i 1 
Mass normalized

where “n” is the total number of modes considered.

 1 2 3 4 5 6…………………………………………………………………..n

Consider a “n” degree of freedom structure

For the ith mode, the eigenvalue problem can be written as

D
i [ K ]i   i [ M ]i 2

IIT
i

i [ M ]i  mi
4
86
ith modal mass
VL
Let us define complete “modal matrix of the structure (nxn) by arranging all
C

“n” mode shapes vertically

[ ]  [123 ...i ...n ]

 The total eigenvalue problem can be written as

[ ] [ K ][ ]  [][ ] [ M ][ ]

T T

[]  [diag i ]
2

D
[ ] [ M ][ ]  ??  [diag mi ]  [ M p ]
T

IIT
[ ] [ K ][ ]  [][ M p ]
T

4
86
[] 
T 1
[]
1
VL
Pre-multiply both sides by Post-multiply by
C

[]  [ ] [ K ][ ] []  []  [][ M p ][]

T 1 T 1 T 1 1

[K ]  []  [][ M p ][]

T 1 1
 [K ]  []  T 1
[][ M p ][  ] 1

[ ] [ M ][ ]  [ M p ]
T

[ M p ]1

D
Pre-multiply both sides by

IIT
[ M p ] [ ] [ M ][ ]  I
1 T

4
86
Post-multiply both sides by [ ] 1
VL
C

1
[ ]  [M p ] 1
[ ] [ M ] T
 1
[ ]  [M p ] 1
[ ] [ M ]
T

[] 
T 1
 ?? [ M p ] [ ] [ M ]
1

D
IIT
If mode shapes are mass normalized

4
86
[ ] [ M ][ ]  [ M p ]  [ I ]
T
VL
1
[ ]  [ ] [ M ] T
C

[]  T 1
 [ ] [ M ]
 [K ]  []  T 1
[][ M p ][  ] 1

1
[ ]  [ ] [ M ] T
[]  T 1
 [ ] [ M ]

D
IIT
[K ]  [ ] [ M ] [] [] [ M ] T

4
86
VL

[K ]  [ M ][ ] [][] [ M ]
T
C

Even if we have incomplete modal information, in terms of “m” modes only,

we still obtain [K] matrix “nxn”
 CHANGE IN STIFFNESS METHOD
(Zimmerman and Kaouk, 1994)
Submitted in June 1992
Based on the principle that damage alters the stiffness matrix
of the structure

D
The stiffness matrix is derived from the measured natural

IIT
frequencies and the mode shapes of the structure.

4
The stiffness matrix [K] was expressed by the researchers in terms of the

86
mode shape matrix [φ], the mass matrix [M], and the modal stiffness
matrix [Ω] (diagonal ωi2 ) as
 
VL
n
[ K ]  [ M ][ ][][ ] [ M ]  [ M ]  i2iiT [ M ]
T

 i 1 
C

Mass normalized
where “n” is the total number of modes considered.

The main disadvantage of this method is that the higher modes are more important
in the estimation of the stiffness matrix, since the modal contribution to [K] increases
as the modal frequency increases. The dynamic vibration testing, on the other hand,
can yield only the first few mode shapes.
 CHANGE IN FLEXIBILITY METHOD
(Pandey and Biswas, 1994)
Paper was submitted on 15 Feb 1991
Based on the principle that damage alters the flexibility matrix of
the structure, that is the inverse of the stiffness matrix

D
IIT
n
1
[ F ]  [ ][] [ ] 1 T    T

i
2 i i

4
i 1

86
Mass normalized
VL

[K ]  []  1
[][ M p ]  1
C

T
[ ]

Just take inverse of both sides to obtain the flexibility matrix (HW)
 CHANGE IN FLEXIBILITY METHOD
(Pandey and Biswas, 1994)
Paper was submitted on 15 Feb 1991
Based on the principle that damage alters the flexibility matrix of
the structure, that is the inverse of the stiffness matrix

D
IIT
n
1
[ F ]  [ ][] [ ] 1 T    T

i
2 i i

4
i 1

86
Mass normalized
VL
C

As can be seen from the equation, [F] is proportional to the square of the inverse of
the modal frequencies.

Therefore, it converges rapidly with increasing frequencies. Hence, only a few lower
modes are sufficient for an accurate estimation of [F]. The technique is thus an
improvement over the change in stiffness method.
 CHANGE IN FLEXIBILITY METHOD
(Pandey and Biswas, 1994)

D
Damage location
Change in flexibility

IIT
4
86
VL
1 6 11 16 21 26 31
Node
C

Identification of damage location from a plot of the change in flexibility for

a beam structure
 CHANGE IN MODE SHAPE CURVATURE
METHOD (Pandey et al. , 1991)
The curvature at any location of the beam subjected to a bending
moment M (x), could be expressed as

M ( x) 

D
 "( x)   Curvature mode shape is obtained as the second

IIT
derivative of the displacement mode shape
EI y

4
where E is the Young’s modulus of elasticity and I the moment of inertia of the section.

86
A reduction in stiffness associated with the damage will lead to an increase in
curvature.
VL
In this method, the pre and post damage displacement mode shapes are first extracted
C

by experimental modal analysis.

Differences in pre and post damage curvature mode shapes are maximum in the vicinity
of the damaged region. This method is computationally simpler as compared to the
change in stiffness method and the change in flexibility method.
 CHANGE IN UNIFORM LOAD SURFACE
CURVATURE METHOD
(Zhang and Aktan, 1995)
The coefficients of the ith column of the flexibility matrix represent the deflected
shape assumed by a structure when a unit load is applied at the ith degree of

D
freedom.

IIT
The sum of all the columns of the flexibility matrix represent the deformed shape if
a unit load is applied at each degree of freedom.

4
86
Zhang and Aktan used the curvature of this uniform load surface to determine the
location of damage. The curvature change at the ith location was expressed as
VL
Fi  Fi  Fi
'' *'' ''
C

Post-damage curvature Pre-damage curvature

The damage location is identified as the element with maximum value of ΔF

 DAMAGE INDEX METHOD
(Stubbs and Kim, 1994)

1 1
U   dV   EI ( y" ) 2 dx

D
Strain energy
2V 2L

IIT
4
86
VL
1
Modal strain energy U   EI ( " ) 2 dx
2L
C
 DAMAGE INDEX METHOD
(Stubbs and Kim, 1994)
The damage index β is calculated based on the strain energy stored
in the structure when it deforms in a particular mode shape. For the

D
jth location and the ith mode, the damage index βij can be defined
as

IIT
 b L
 L
  [ i*'' ( x)]2 dx   [ i*'' ( x)]2 dx  [ i'' ( x)]2 dx
 

4


 ij  b

a
86 0


 0
L
VL
L
  [ i ( x)] dx   [ i ( x)] dx   [ i ( x)] dx
'' 2 '' 2 *'' 2
 
C

a 0  0

 i'' ( x) Undamaged i*'' ( x) Damaged

structure structure
 DAMAGE IDEX METHOD
(Stubbs and Kim, 1994)
 b *'' L
 L ''
  [ i ( x)] dx   [ i ( x)] dx  [ i ( x)]2 dx
  Strain U  EI ( y) 2 dx

D
2 *'' 2

2 L


 ij  ba 0  0 energy

IIT
 L
 L
  [ i'' ( x)]2 dx   [ i'' ( x)]2 dx   [ i*'' ( x)]2 dx
  Modal EI
a  0     2

4
0
Strain U  ( ) dx
2 L

86
Here, ‘a’ and ‘b’ are the limits of a segment of the
beam where damage is being evaluated. energy
VL
C

This technique has an advantage over the previous ones in that it

has a specific criterion for determining whether damage has
occurred at a particular location. Other techniques look for the
largest change in a particular parameter. It is therefore suitable for
multiple damage locations.
COMPARATIVE STUDY OF DIFFERENT ALGORITHMS
Farrar and Jauregui (1998)
Farrar and Jauregui (1998) did a comparative study of the techniques on a
single structure, the I-40 bridge over Rio Grande in Albuquerque, New
Mexico, USA.

D
IIT
4
86
VL
C
COMPARATIVE STUDY OF DIFFERENT ALGORITHMS
Farrar and Jauregui (1998)
The bridge had a maximum span length of about 50m and a girder depth
of about 3m.

D
Damage was induced by means of a torch cut, at the middle of the 50m
span and was started from the mid depth of web as a 0.6m long by

IIT
10mm wide crack .

4
It was then extended in three stages to the entire bottom half of the web

86
and the bottom flange. The bridge was excited by a hydraulic shaker (both
in undamaged and damaged conditions) and the modal data was
VL
extracted.
C
COMPARATIVE STUDY OF DIFFERENT ALGORITHMS
Farrar and Jauregui (1998)
The major conclusions drawn based on the analysis of the response data by
various methods on a real bridge being decommissioned, are as follows :

D
1. Standard modal properties such as mode shapes and resonant frequencies are
poor indicators of damage. No noticeable change in the measured resonant

IIT
frequencies and mode shapes was observed until the final level of damage.

4
2. All the methods identified the damage location correctly for the most severe

86
damage case, i.e a cut through the entire bottom half of the beam.

3. In some of the methods, human judgment was required to correctly identify

VL
the location of damage. If they were applied blindly, it would have been
difficult to locate the damage.
C

4. When the entire set of tests are considered, the damage index method was
found to be the one with the best performance.
 CONTENTS
• EXPERIMENTAL MODAL ANALYSIS

D
(NON MATHEMATICAL TREATMENT)

IIT
• GLOBAL VIBRATION TEHNIQUES FOR

4
86
STRUCTURAL HEALTH MONITORING
VL
C

• USE OF MODERN SMART SENSORS IN

GLOBAL VIBRATION TECHNIQUES
 PZT PATCH AS STRAIN SENSOR

Using Q  D3 A  CV

D
IIT
 d 31hY E 
V   S1  kS1

4
 T 
 86
33 
VL
C

Voltage α Strain
Sensitivity = 200mV/m/m
23/01/2019
 PZT PATCH AS
CURVATURE SENSOR Vo
Compression + Stop
- PZT patch
D

D
+
Tension - PZT patch Sbott

IIT
+ - + -

4
Stop  Sbott
Vo
86 
VL
D
Vo  k ( Stop  Sbott )
C

Stop  Sbott Vo
Curvature  
23/01/2019
D kD 44
 EXPERIMENTAL SETUP

LCR meter

Multi meter

D
IIT
4
PZT Sensor
86
VL
Weight Damage Location
C

Steel beam

Artificial damaged steel beam

 DETECTION OF INCIPIENT DAMAGE

Damage location

D
PZT Sensor

IIT
4
80 0mm 1204 mm

86
4016 mm
VL
C

Experimental test setup

 1.6 cm 1.5 cm 1.6 cm 1.5 cm

1.7 cm
DAMAGE INDUCED
IN STAGES
Stage 3

D
Stage 1 Stage 2

IIT
3.9 cm

4
Stage 4 Stage 5 86 Stage 6
VL
C

7.8 cm

Stage 7
 PROCEDURE
• Beam was divided into 12 nodes.

• Sensor response was measured corresponding to

D
weight (200g) dropped at each point from 1.5m.

IIT
• Time domain data was transformed to frequency

4
86
domain.
VL
• Three measurements were made at each point
and average values obtained.
C

• Ordinates (real part) corresponding to first three

modes were utilized to obtain the strain mode
shapes.
 NATURAL FREQUENCIES
First natural Second natural frequency
Third natural frequency /
[Link]. Damage state frequency / /
(% change)
(% change) (% change)

1 Undamaged 45 (-) 190 (-) 410 (-)

D
2 State-1 45 (0.0%) 189 (0.52%) 410 (0.00%)

IIT
3 State-2 45 (0.0%) 189 (0.52%) 408 (0.49%)

4
4 State-3 41 (8.9%) 185 (0.97%) 402 (1.95%)

86
5 State-4 39 (13.3%) 182 (4.21%) 399 (2.68%)
VL
6 State-5 37 (17.8%) 179 (5.79%) 392 (4.39%)

36 (20%)
C

7 State-6 174 (8.42%) 387 (5.61%)

8 State-7 34 (24.4%) 170 (10.52%) 380 (7.32%)

Poor damage indicators in beginning!!

 FIRST
FIRSTCURVATURE
MODE MODE SHAPE
SHPAE
20
undamage
damage 1
damage 2
16 damage 3
damage 4

D
damage 5
Relative ordinate

12 damage 6

IIT
damage 7

4
8

86
VL
4
C

0
0 100 200 300 400
Distance (cm)
 SECOND
SECOND CURVATURE
MODE MODE SHAPE
SHPAE
25
undamage
damage 1
20
damage 2
damage 3
15 damage 4

D
damage 5
10 damage 6

IIT
Relative ordinate

damage 7

4
86
0
VL
-5
C

-10

-15
0 100 200 300 400
Distance (cm)
 THIRD MODE SHPAE
10
undamage
damage 1
damage 2

D
damage 3
5 damage 4

IIT
damage 5
damage 6
Relative ordinate

4
damage 7

86
0
VL
C

-5

-10
0 100 200 300 400
Distance (cm)
 10

damage 1
damage 2
8
damage 3
damage 4
damage 5

D
Damage Index

6
damage 6

IIT
damage 7
4
D.I .   Cdamage  Cundamage 
N

4
n 1

86
2
VL
0
C

1 2 3 4 5 6 7 8 9 10 11
Element No
EXPERIMENTAL SETUP FOR STEEL PLATE
(Kumar, 2008)

D
IIT
Artificial Damage

4
86 Stage-2, 3
VL
Stage-1
C

Steel plate: 1.3x0.65m MS plate

 FIRST CURVATURE MODE SHAPE
30

25

D
20

IIT
15

10

4
86
5
VL
0
0.6
C

0.4 1
0.2 0.5
0 0
SECOND CURVATURE MODE SHAPE

10

D
5

IIT
0

4
-5

-10
86
VL
0.6

0.4 1
C

0.2 0.5
0 0
 THIRD CURVATURE MODE SHAPE
3

D
0

IIT
-1

4
86
-2

-3
VL
0.6
C

0.4 1
0.2 0.5
0 0
DAMAGE INDEX OF MS PLATE AFTER
FIRST DAMAGE

2.2

D
IIT
1.8

4
D. I.

1.4
86
VL
C

1 S4
1 2 3 4 5 S1
6 7 8 9 10
 DAMAGE INDEX OF MS PLATE
AFTER SECOND DAMAGE

D
IIT
2.5

4
D.I.

2
86
VL
C

1.5 S4
1 2 3 4 5 6 S1
7 8 9 10
DAMAGE INDEX OF MS PLATE AFTER THIRD DAMAGE

3.8

D
3.3

IIT
D.I.

4
2.8

2.3 86
VL
C

1.8 S4
1 2 3 4 5 6 S1
7 8 9 10
DAMAGE LOCATION ALGORITHM
We define (Talwar, 2011)
𝑖 = Curvature of 1st mode shape at node i before damage

D
𝑖 = Curvature of 1st mode shape at node i after damage

IIT
𝑖
𝐷𝑖 = 𝑎𝑏𝑠(1 − ) (6.3)

4
𝑖

86
𝐷𝑖 is the Damage index for i th node
VL
The damage index for the element 𝐷𝑖𝐸 was obtained by summing the damage indices of the four nodes
C

making up the element. Damage index of an element

=
Sum of Di of four surrounding nodes
 EFFECT OF DAMAGE

D
IIT
4
86
VL
Pre-damage
C

23/01/2019
Post-damage 62
DAMAGE INDEX FOR ELEMENTS
DAMAGE LOCATION

D
IIT
4
86
VL
C

DAMAGE LOCATION CLEARLY

IDENTIFIED BY MAXIMUM
DAMAGE INDEX FOR
ELEMENT
 IN CONCLUSION…..
• Surface bonded PZT patches can be employed to obtain
curvature mode shapes directly. Expensive
accelerometers substituted by low-cost PZT patches.

D
• The experimental undamaged and damaged mode

IIT
shapes have been used, circumventing any numerical
modelling of structure.

4
• Damage detection, location and severity covered.
86
VL
C

UNANSWERED…..
• How to identify incipient damage????
 THANK YOU

D
IIT
Suggested reading: Farrar and Jauregui (1998)
Supplementary reading: Pandey and Biswas (1994),

4
Pandey et al. (1991)

86
VL
C

Compulsory reading: Shanker et al. (2011)

23/01/2019 65