Structural Health Monitoring By Vibration Measurement with Non-contact Laser Excitation

Feblil HUDA Candidate for the degree of Doctoral

Supervisor: Prof. Itsuro Kajiwara Division of Human Mechanical System and Design

Introduction

Damage identification as a part of structural health

monitoring (SHM) plays an important role to make sure

the current condition of structures. In general, there are

two main steps of diagnostic procedure in damage

identification: the experimental measurement and the

data processing.

The diagnostic methods largely studied and applied

are those based on vibration measurements, since they

allow non-destructive evaluation on investigated

structures. The usual approach in damage detection

based on vibration measurement consists in the

determination of dynamic characteristics changes due to

the presence of the damage. The dynamic characteristics

of a structure can be assessed from the frequency

response functions (FRFs) which are obtained by

applying an excitation force to the structure and

measuring the excitation force and vibration response.

Some problems and inconvenient are found in these

methods, that usually involving part directly mounted

and contacted to the surface of tested parts for applying

excitation force, can possibly contaminate tested parts

and add safety constraints such as the need for

additional inspection, endanger measurement operators

if the measurements last on narrow operational area and

even do not give the possibility of conducting the

measurement in narrow area. These conditions make

non-contact method become important.

In order to solve the problems in vibration excitation

for damage detection, an integrated non-contact damage

detection system based on vibration is important to be

developed. The measurements must be employing of

non-contact structural excitation testing with high

reproducibility, enable to extract measurement in high

frequency and enable working in small area. Non-

contact vibration excitation is known well using laser

system which may fulfill the requirements.

Laser Excitation

In this dissertation, the vibration testing and health

monitoring system based on an impulse response

excited by laser is proposed to detect damage on

structures. This idea is built based on the applicability of

laser in giving excitation to structures and the needs to

have high reproducibility non-contact excitation system

on structure. The source of excitations is proposed in

two methods: laser ablation [1] and laser-induced

breakdown [2].

Laser ablation

The process of the laser ablation is presented in Fig. 1.

When a laser beam is irradiated on a metal surface, it

will be absorbed by the metal, and the atoms absorbing

the laser light release ions. The absorption of the energy

in the laser beam by metal will also generate high

temperature plasma, and large quantities of particles are

then released (in the form of a plume) from the metal

(Fig. 1(a)). Momentum is then generated when a mass

Δm released at a velocity v from the metal, represented

by Δmv, and this expresses the laser-induced impulse.

Fig. 1 Process of the laser ablation [1]

To generate a larger excitation force on the

structure, a water droplet is placed on the metal surface

during laser ablation as shown in Fig. 1(b). The total

resulting momentum is Δmv+ΔMV, resulting in a larger

impulse with than without the water droplet. This

impulse constitutes the excitation force on the structure.

The input characteristics different between laser

excitation and impulse hammer can be seen in Fig. 2.

Fig. 2 shows that the excitation by laser occurred in

really short time, while the excitation by impulse

hammer shows curvy excitation force. This is a fact that

ideal impulse force to metal structure could be sourced

from non-contact excitation by laser ablation.

Fig. 2 Different characteristics between laser excitation and

impulse hammer [1]

Laser-induced breakdown

The laser-induced breakdown (LIB) refers to the

formation of plasma through the cascade process caused

by electrons emitted from atoms and molecules that

have absorbed multiple photons through a multi-photon

Lo

ad

[N

]

:Laser :Impulse hammer

0.024 0.025 0.026 0.027 0.028-50

0

50

100

Time [s]

Metal

Plume(m)

Atoms etc.

Metal

Laser

v

Plume(m)

Vapor

(M)

Metal

Water

droplet Atoms etc.

Metal

Laser

v

V

(a) Laser ablation without water droplet

(b) Laser ablation with water droplet

process when a laser beam is focused in a gas. A portion

of this plasma energy is transformed to a shock wave,

which is the source of the sound generated by LIB,

which has the potency to be acoustic excitation force.

This is one of the high power laser effects.

The LIB threshold in air is I ≥ 1015 W/m

2. When the

laser's local intensity is smaller than this threshold, a

convex lens can be used for focusing the laser beam,

thereby reducing the spot radius. By passing the laser

beam through the convex lens which focuses the beam,

the local intensity of the laser beam reaches to or above

the minimum LIB threshold of 1015 W/m

2 [3-5].

The process to achieve acoustic excitation by LIB is

presented in Fig. 3. From the figure, it can be seen that

Nd: YAG pulse laser is used to produce laser beam, and

laser beam is then passed to convex lens. The LIB

occurs in the distance of focal length of the convex lens

from the convex lens position.

Fig. 3 Process to achieve acoustic excitation by LIB

Fig. 4 shows times response sound pressure

generated by LIB. The ten measured waveforms of the

sound pressure are plotted. It is shown on the figure that

the point sound source generated by LIB on that

configuration has high reproducibility and can generate

ideal impulse excitation, occurs in short time.

Fig. 4 Time responses of sound pressure generated

by LIB [2]

The laser excitation can apply an ideal impulse

excitation, making it possible to use the Fourier

transform of the output response for the evaluation of

the vibration characteristics of the system as the

frequency response function. The Fourier transforms of

time histories of an input force f(t) and output response

x(t) into the frequency domain are described by

( ) ∫ ( )

(1)

( ) ∫ ( )

(2)

The output response x(t) can be expressed by the

convolution integral with an impulse response function

h(t) of the system:

( ) ∫ ( )

( ) (3)

The Fourier transform of Eq.(3) results in

( ) ( ) ( ) (4)

where, ( ) is the frequency response function of the

system. When f(t) is an impulse force, ideally the Dirac

delta function, the Fourier transform of f(t)

becomes ( ) , and Eq.(4) is expressed by

( ) ( ) (5)

The result is that the Fourier transform of the response

x(t) becomes the frequency response function.

Non-contact laser excitation and damage identification

In damage identification using vibration approach, the

ideal excitation method is required to ensure the

accuracy, reliability and easiness of measurement. In

applying non-contact laser excitation to damage

identification, the mechanism of laser excitation needs

to be carefully considered, so that the excitation method

not only gives the easiness in measurement, but also

enhances the quality of measurement.

Non-contact laser excitation by laser ablation is a

type of excitation that is usually applied to metal

structures. Besides giving ideal impulse excitation, this

excitation method has strength point in measuring high

frequencies vibration, but laser ablation causes a small

damage on measured structure, so this excitation

method will be appropriate to be applied to metal

structures that have no problem with the effect of small

damage caused by ablation. It will be suitable to be

combined with damage identification on bolted joint

structure.

Acoustic excitation by LIB is non-contact excitation

which gives possibility to conduct measurement in

really narrow area; this kind of excitation causes no

damage. So, this kind excitation is suitable to be

combined with damage detection on extremely light and

flexible structure. Acoustic excitation by LIB will be

applied and combined with damage identification

method to detect damage in membrane structures.

Bolt Loosening Analysis and Diagnosis by Non-contact Laser Excitation Vibration Test

Detecting a loosening of bolted joints is important for

ensuring the function of structure or subsystem.

Vibration measurements approach for bolt loosening

detection usually use contact excitation or hammering

method to excite a structure to get the frequency

response, requiring trained technicians over a long

period. The reproducibility of hammering method by

technicians is low; the measurement of dynamics

characteristics requires high reproducibility. This issue

led researchers to find other excitations method with

higher degree of reproducibility. Vibration measurement

systems using laser excitation can guarantee an

extremely high degree of reproducibility of

measurement [1]. This research concerns about the use

of laser excitation for detecting the bolt loosening.

Finite element model of bolted joint

The finite element analysis software ANSYS 14.0 has

been used to model bolted joint with pretension force

and mating part contact. The model is constructed in

form of cantilever, where one of the flanges is fixed, as

shown in Fig. 5. The SOLID186 element of ANSYS is

used to construct physical model of bolted joint. The

total number of nodes and elements are 29635 and 5412

respectively. The pretension force is given by using

pretension element PRETS179.

9.8 10.0 10.1 10.29.9-600

00

600

0

1200

Time [ms]

So

un

d p

ress

ure

[P

a]

Fig. 5 Simple bolted joint model

In this model, the tightening torque will be applied

to the bolt in form of pretension force. The relationship

between tightening torque and occurred pretension is

described by

d (6)

where F, T, K, and d are pretension force, tightening

toque, torque coefficient, and nominal diameter of bolt

respectively.

The contact modeling is presented by surface-surface

contact elements, which is a pair of contact element

CONTA174 and target element TARGE170. There are

three types contact used in this model: bonded,

frictional, and no separation.

In order to determine the contact type on contact

between flanges, the region, where the stresses due to

pretension that produced clamping force are

predominant, is set to be bonded each other. That region

in normal condition looks like a conical shape and

covers a range between 25o ≤ α ≤ 33

o, suggested by

Osgood [6]. The bolt head and the nut are also assumed

to be glued to the flanges due to the clamping force.

Some pretension forces for normal and loosening

condition will be applied to the bolt to get the stress data

on contact between flanges. By applying conical shape

suggested by Osgood, the values of stress on outer

radius of conical shape from normal pretension force are

taken as the benchmark and limit for bonded area, and

the lower stress areas will be frictional and no

separation. A series of static structural analysis and

frequency response analysis by pre-stress modal

analysis are then conducted.

Verification of laser excitation vibration measurement result by finite element analisis

Vibration test by using laser excitation is conducted for

a simple model of bolted joint, and the results are then

validated by finite element analysis. The tightening

torques are given by digital torque wrench for normal

condition (24.5 Nm) and loose conditions (less than

24.5 Nm). A vibration testing arrangement using a high

power pulse laser is shown in Fig. 6.

Fig. 6 Vibration testing arrangement using the high power

pulse laser

Frequency responses with different tightening

torques got from experiments in vertical and horizontal

direction are shown in Fig. 7. The resonance peaks in

frequency responses are shifted to lower frequency and

the shifting is significant in high frequency, while in low

frequency the response seems not to change so much.

The Frequency response of simple one-bolt joint in

normal condition is presented in Fig. 8.

(a) (b)

Fig. 7 Frequency response of simple one-bolt joint by

experiment (a) in vertical (b) in horizontal

(a) (b)

Figure 8 Frequency response of simple one-bolt joint in

normal condition (a) in vertical (b) in horizontal

The frequency responses in normal condition by

simulation and experiment show a good agreement.

They are not only having the same peaks, but also

having the same tendency. This good agreement can be

seen both in vertical and horizontal directions of test.

Fig. 9 shows the frequency response of simple one-bolt

joint in vertical direction with loose conditions, and all

of simulation-experiment comparison results show the

same tendency.

Fig. 9 Frequency response of simple one-bolt joint in vertical

with loose condition and tightening torques of

(a) 20 Nm, (b) 18 Nm, (c) 15 Nm, (d) 10 Nm

Bolt loosening detection approach

The Recognition Taguchi (RT) method is a statistical

evaluation method used to detect the loose bolt in this

research. RT method is a type of MT (Mahalanobis–

Taguchi) method used in the field of quality engineering

as a pattern recognition method. In this method, data is

measured a multiple number of times under identical

conditions, and a unit space for normal condition is

defined. For unknown data (loose condition), distances

in relation to unit spaces are calculated and compared to

determine damage index (DI), which defined as

Damage Index 𝐷

�̅� (7)

Where D is the Mahalanobis distance for unknown

data, and �̅� is the covariance of the Mahalanobis

distances for normal condition. If DI > 1, the system is

under damage (in this case undergoes loosening). If DI

≤ 1, the system is normal.

In order to demonstrate loose bolt detection on

bolted joint, a six-bolt joint model is constructed;

several damages are applied to the joint by giving

different tightening torques on certain bolt position. The

list of damage cases can be seen in Table 1.

Table 1 the list of damage cases

Laser excitation vibration measurement system is

applied to six-bolt joint. The model, bolt numbering,

excitation point, and measurement points can be seen in

Fig. 10. The excitations by laser are given in a point,

and measurements of the responses are conducted in six

points which represent the measurement of response on

each bolt.

Fig. 10 The excitation and measurement points position for

bolt loosening detection

Result

The measurements were conducted first for normal

condition to get the unit space that was constructed from

ten sets of power spectrum data. The measurements for

several damage conditions were then conducted to get

the only one set power spectrum data for each case of

damage. By comparing damage data with the unit space

in the same range of interest, the damage index would

be gotten. The damage index on the bolt in every case of

damage by simulation and experiment data can be seen

in Table 2 and Table 3 respectively.

Table 2 Damage index based on simulation result

Post. Bolt

1

Bolt

2

Bolt

3

Bolt

4

Bolt

5

Bolt

6 Case Damage 1 36.00 14.00 14.60 11.90 18.00 6.32

Damage 2 16.86 16.82 15.08 55.68 14.08 14.27

Damage 3 15.75 15.65 15.90 16.15 17.30 16.30

Damage 4 56.70 20.74 20.05 31.20 27.80 10.30

Damage 5 25.22 19.30 25.88 40.03 16.51 18.97

Damage 6 25.73 26.05 24.54 28.90 33.10 27.70

Table 3 Damage index based on experiment result

Post. Bolt

1

Bolt

2

Bolt

3

Bolt

4

Bolt

5

Bolt

6 Case Damage 1 10.40 7.00 1.60 0.67 0.45 0.41

Damage 2 0.45 2.20 0.44 5.47 1.62 0.62

Damage 3 1.80 2.20 3.11 2.38 7.76 2.56

Damage 4 29.30 9.10 5.30 3.20 2.53 2.86

Damage 5 0.58 1.88 0.32 5.93 4.70 2.39

Damage 6 3.73 3.25 7.54 0.82 11.43 3.77

Table 2 and Table 3 show that most of DI is greater

than 1, which means that the damage condition

(loosening) of the joint almost can be detected in every

position of measurement, and the highest value of

damage index in the same damage case represents the

position of the loose bolt. These results correspond to

the scenario of some cases of loosening designed in this

research.

Damage Detection in Membrane Structures Using Non-contact Laser Excitation and Wavelet Transformation

The vibration of membranes has been investigated with

different types of non-contact excitation and vibration

measurement methods, such as horn-acoustical

excitation and capacitance displacement sensor,

speakers-acoustical excitation and laser vibrometer,

laser excitation by laser ablation and laser Doppler

vibrometer (LDV), but these measurement methods do

not fulfill the requirements of ideal non-contact

measurements method for membrane. The ideal

measurement method for membrane should be non-

destructive, allow conducting experiment in small area,

attach no sensors/exciters to objective structure which

causes the changes of mass and stiffness, and give ideal

point-excitation to the structure. Vibration test using

laser-induced breakdown (LIB) is non-contact and non-

destructive excitation method which involves no any

attachments to objective structures, and offers the

possibilities of applying acoustical excitation on a point

with some distance between exciter and membrane

structure. This research concerns about the use of laser

excitation by LIB for detecting damage in membrane

structure.

Vibration testing system using LIB

In this research, an excitation to a membrane structure is

applied in the form of acoustic excitation achieved by

generating an ideal point sound source at specific

location via LIB which offers the possibility of

conducting experiments with acoustic excitation in

limited space. To measure the output response, two

LDVs are set to obtain the responses of the membrane

structure. These measurements enable to extract the

mode shapes of membrane. A spectrum analyzer (A/D;

NI-4472B, Software; Catec CAT-System) is used for

Condition Tightening Torque (Nm)

Bolt 1 Bolt 2 Bolt 3 Bolt 4 Bolt 5 Bolt 6

Normal 24.50 24.50 24.50 24.50 24.50 24.50

Damage 1 20.00 24.50 24.50 24.50 24.50 24.50

Damage 2 24.50 24.50 24.50 20.00 24.50 24.50

Damage 3 24.50 24.50 24.50 24.50 20.00 24.50

Damage 4 16.50 24.50 24.50 24.50 24.50 24.50

Damage 5 24.50 24.50 24.50 16.50 24.50 24.50

Damage 6 24.50 24.50 24.50 24.50 16.50 24.50

analysing the Fourier spectrum of the structure. The

vibration testing set-up is shown in Fig. 11.

Fig.11 Vibration testing set-up for the membrane structure

The membrane material is Kapton, manufactured by

Du Pont-Toray Co., Ltd. Owing to its good resistance to

temperature; Kapton is often used in actual space

applications. The membrane here is 200 mm 200 mm

in size and 50 µm thick which is clamped on four sides

by metal clamps and kept stretched by three 700 gram

masses and one side is fixed. The masses and the fixed

side are connected to the metal clamps by steel wires.

FEA investigation of damage detection on the membrane structure

The finite element analysis software ANSYS 14.0 is

used to conduct the pre-stress modal analysis to

generate the mode shapes of normal and damaged

membrane structures. The SHELL181 element of

ANSYS is used to construct the physical model of the

membrane, and the SOLID186 element is employed to

model the clamps at the corners of the membrane. The

membrane is uniformly divided into approximately

10,000 2 mm 2 mm four node membrane elements.

The finite element model of the membrane structure is

shown in Fig. 12.

Fig. 12 Finite element model of the membrane structure

Two-dimensional continuous wavelet transformation for damage detection in membrane structure

The two-dimensional continuous wavelet transformation

(2-D CWT) considered in this study is based on the

formulation by Antoine [7], and wavelet computations

are performed using MATLAB® and the YAWTb

toolbox. The procedure to detect the position of damage

in the membrane structure is adopted from the

procedure suggested by Fan [8], using the 2-D CWT

derivative Gauss (Dergauss2d) and iso-surface concept,

which was used successfully for detecting damage on

plates using mode shape data. The mode shape data is

transformed by 2-D CWT with several scales and by

using iso-surface concept, the position of damage can be

recognized.

Application of damage detection

To verify and to examine the applicability the damage

detection approach using 2-D CWT by simulation, three

different single-damage conditions were induced in the

membrane finite element models. The details of the

three damage cases are listed in Table 4. These damages

are induced to membrane by simulation and experiment.

Table 4 Details of the damage induced on the membrane

Damage

type

Damage size

(mm)

Closest measurement

point (x,y) in mm

L-cut

(120,120)

L-tear

(140,100)

I-cut

(80,100)

The process of analysis for detecting the damages can

be explained by using Fig. 13. The mode shape

extracted from simulation is interpolated to get more

data points of mode shape which then transformed by 2-

D CWT to get wavelet coefficient. The boundary

treatment is applied to move very high value on the

corners and the edge. The position of damage can be

recognized by applying iso-surface concept to wavelet

coefficient. The same procedure is applied to vibration

mode shape extracted from vibration measurement by

laser excitation. The comparison between simulation

and experimental results is presented in Table 5.

Fig. 13 Process of the analysis for detecting the L-cut

damage on the membrane by simulation data (a) first

principal mode shape (41.78 Hz), (b) interpolated first

principal mode shape, (c) wavelet coefficient, (d) wavelet

coefficient with boundary treatment, (e) 3-D view of iso-

surface, (f) top view of iso-surface.

Table 5 shows that the damages on membrane could

be detected well by using proposed method. The

position of L-cut and L-tear damages could be detected

using first principal mode shape, but I-cut damage could

not be detected using first principal mode shape, thus

the higher order mode shape (in the frequency of 843.8

Hz by simulation and 851 Hz by simulation) was

employed to detect the position of damage. The

simulation and experimental results are in good

agreement.

Table 5 Comparison of damage detection between simulation

and experimental results

Damage

type

Simulation

result

Experimental

result

L-cut

L-tear

I-cut

In order to make sure applicability the proposed

approach to detect more damage, I-cut and square-hole

damages are introduced to the membrane. The positions

of damage on the membrane can be seen in Fig. 14.

Because these damages cannot be detected well using

first principal mode shape, the higher order mode is

considered to detect these two damages on the

membrane.

Fig. 14 Membrane with I-cut and square-hole damages

Vibration testing using laser excitation is conducted

on the membrane. Some vibration mode shapes at high

frequency got from experiment are checked to find the

mode shape which contains dominant peak of both types

of damages. It is found at natural frequency of 799.1

Hz, the mode shape contains dominant peaks which

show the local mode of damage. The analysis process

for detecting these two damages is presented in Fig. 15.

The position of the two damages can be detected well.

Fig. 15 Process of the analysis for detecting I-cut and

square-hole damage on the membrane by experimental

data (a) mode shape (799.1 Hz), (b) wavelet coefficient

with boundary treatment, (c) 3-D view of iso-surface, (d)

top view of iso-surface

Conclusion

In this dissertation, two methods of non-contact

excitation by laser were elaborated. These excitation

methods have high reproducibility, can apply ideal

impulse excitation to structure, have the big potency to

be implemented to vibration testing without measuring

the input, and really powerful in extracting high

frequency measurement, cause the easiness of

implementing them to structural health monitoring

(SHM). These have been proofed by the application of

non-contact laser excitation by laser ablation in

detecting bolt loosening and by the application of non-

contact laser excitation by laser induced breakdown in

detecting damage in membrane structure.

References

[1] I. Kajiwara, and N. Hosoya, Vibration testing based on

impulse response excited by laser ablation, Journal of Sound

and Vibration, 330, 5045 – 5057, 2011.

[2] N. Hosoya, M. Nagata, I. Kajiwara, Acoustic testing in a

very small space based on a point sound source generated by

laser-induced breakdown: stabilization of plasma formation,

Journal of Sound and Vibration, 332, 4572 – 4583, 2013.

[3] Q. Qin , K. Attenborough, Characteristics and application of

laser-generated acoustic shock waves in air, Applied

Acoustics, 65, 325-340, 2004.

[4] M. Oksanen, J Hietanen, Photo acoustic breakdown sound

source in air, Journal of Ultrasonics, 32, 327-333, 1994.

[5] V.B. Georgiev, V.V. Krylov, Q. Qin, K. Attenborough,

Generation of flexural waves in plates by laser-initiated air

borne shock waves, Journal of Sound and Vibration, 330,

217-228, 2011.

[6] Shigley, J.E., [Mechanical Engineering Design Eight

Edition], McGraw-Hill, USA, 413-414, (2006).

[7] J.P. Antoine, R. Murenzi, P. Vandergheynst, Two-

dimensional wavelets and their relatives, Cambridge

University Press, Cambridge, 2004

[8] W. Fan, P. Qiao, A 2-D continuous wavelet transform of

mode shape data for damage detection of plate

structures, International Journal of Solids and Structures

46 (2009) 4379-4395.