1361 665x/aaefaa

Smart Materials and Structures
ACCEPTED MANUSCRIPT
Effects of debonding on lamb wave propagation in a bonded composite

structure under variable temperature conditions
To cite this article before publication: Shirsendu Sikdar et al 2018 Smart Mater. Struct. in press https://doi.org/10.1088/1361-665X/aaefaa
Manuscript version: Accepted Manuscript

Accepted Manuscript is “the version of the article accepted for publication including all changes made as a result of the peer review process,
and which may also include the addition to the article by IOP Publishing of a header, an article ID, a cover sheet and/or an ‘Accepted
Manuscript’ watermark, but excluding any other editing, typesetting or other changes made by IOP Publishing and/or its licensors”
This Accepted Manuscript is © 2018 IOP Publishing Ltd.
During the embargo period (the 12 month period from the publication of the Version of Record of this article), the Accepted Manuscript is fully
protected by copyright and cannot be reused or reposted elsewhere.
As the Version of Record of this article is going to be / has been published on a subscription basis, this Accepted Manuscript is available for reuse
under a CC BY-NC-ND 3.0 licence after the 12 month embargo period.
After the embargo period, everyone is permitted to use copy and redistribute this article for non-commercial purposes only, provided that they
adhere to all the terms of the licence https://creativecommons.org/licences/by-nc-nd/3.0
Although reasonable endeavours have been taken to obtain all necessary permissions from third parties to include their copyrighted content
within this article, their full citation and copyright line may not be present in this Accepted Manuscript version. Before using any content from this
article, please refer to the Version of Record on IOPscience once published for full citation and copyright details, as permissions will likely be
required. All third party content is fully copyright protected, unless specifically stated otherwise in the figure caption in the Version of Record.
View the article online for updates and enhancements.
This content was downloaded from IP address 142.132.1.147 on 16/11/2018 at 02:02

Page 1 of 35 AUTHOR SUBMITTED MANUSCRIPT - SMS-107419.R1
1
2
3
4 Effects of Debonding on Lamb Wave Propagation in a Bonded
pt
5 Composite Structure under Variable Temperature Conditions
6
7
8 Shirsendu Sikdar1*, Piotr Fiborek1, Paweł Kudela1, Sauvik Banerjee2, Wiesław
9
Ostachowicz1
cri
10
11
12 1
13 Institute of Fluid Flow Machinery, Polish Academy of Sciences, 14, Fiszera Street, Gdansk 80-
14 231, Poland.
2
15 Indian Institute of Technology Bombay, Civil Engineering Department, Powai, Mumbai-
16 400076, India.
us
17
18
19
*Corresponding author:
20 Email: [email protected]
21
22 Phone: +48 58-6995-251, Fax: +48 058 3416144
23
24
25
26
27
28
Abstract an
Bonded composite structures are the special type of sandwich-like structures in which multiple
dM
29 carbon-fibre laminates are bonded with adhesives, and debonding can appear at the bond-layer due
30
31 to variable loading and uncertain operating conditions. This study aims to investigate debonding
32
33 effects on Lamb wave propagation in a bonded composite structure under variable ambient
34 temperature conditions. In the process, a combined theoretical analysis, time-domain spectral
35
36 element simulation, and experimental analysis of elastic wave propagation in a carbon-fibre
37
38 reinforced adhesively-bonded composite structure have been carried out. It is shown that
pte
39
theoretical analysis and spectral formulation are effectively able to capture the behaviour of Lamb
40
41 modes in the healthy structure due to variations in ambient temperature. It is found that the primary
42
43 anti-symmetric Lamb wave mode amplitude and velocity decrease with an increase in ambient
44
45 temperature. Further, the spectral formulation accurately captures the effects of debonding on the
46 Lamb wave signals under variable temperature conditions that are consistent with the experimental
ce
47
48 results. Finally, temperature correction factors are proposed for the primary anti-symmetric mode
49
50 velocity and amplitude difference calculations and the effectiveness of the factors are verified
51
52
successfully for selected study cases.
Ac
53
54 Keywords: bonded composite structure, Lamb waves; debonding, piezoelectric disc transducers,
55
56 epoxy adhesive, dispersion.
57
58
59
60
AUTHOR SUBMITTED MANUSCRIPT - SMS-107419.R1 Page 2 of 35
1
2
3 1. Introduction
4
pt
5 Carbon-fibre composite structures are widely used in Marine, Automobile, and Aeronautic
6
7 industries due to their high energy-absorption capacity, fire-resistance, high stiffness-to-weight
8
9
ratio, effective acoustic-insulation, construction flexibilities, and lightweight advantages [1-4]. In
cri
10 some specialised applications, sandwich-like Bonded Composite Structure (BCS) made of
11
12 multiple pre-fabricated carbon-fibre laminates bonded with epoxy adhesive is used. However, due
13
14 to cyclic-loading, impact, ageing, and improper handling, debonding can occur at the bonded
15
junctions which may lead to a sudden structural failure [4-6]. Hence, it is important to identify
16
us
17 these debonding regions in such bonded structures effectively. Many researchers [4-9] focused on
18
19 the detection and mitigation of hidden defects at an early-stage before they propagate further and
20
21 lead to catastrophic structural failures. The ultrasonic guided wave (such as- Lamb wave)
22
23
24
25
26
27
28
an
propagation based structural health monitoring (SHM) of multi-layered composite structures has
shown the potential in localisation and characterization of such hidden structural
defects/discontinuities due to their long-distance monitoring potential and sensitivity to small
defects in [5,7-15]. The current SHM strategies involve the deployment of cost-effective, thin and
dM
29 lightweight bonded piezoelectric transducers or broadband transducers [5,7,9]. The knowledge on
30
31 dispersion phenomenon of the propagating Lamb waves in the structure is referred to be significant
32
33 for the success of SHM of composites using piezoelectric transducers [5,9,10,11,16]. Moreover,
34 knowledge of temperature effects on the Lamb wave propagation through such debonding regions
35
36 is also important, as the change in ambient temperature may significantly affect the SHM strategies
37
38 [17-22].
pte
39
40
Theoretical study on elastic wave propagation in a layered medium is presented by several
41 authors [23-27]. An analytical formulation of elastic wave propagation in sandwich plates under
42
43 specified conditions was presented in [23]. Higher-order theoretical wave propagation
44
45 formulations based on some prior kinematic-assumptions are also proposed for sandwich beams
46
[25-27]. A Thomson–Haskell method based on 2D theoretical model was formulated to calculate
ce
47
48 the dispersion, stress, displacement, and energy distributions of Lamb wave modes [28]. These
49
50 semi-analytical analysis results were only restricted to the frequency-wavenumber domain. A
51
52 global-matrix method based theoretical formulation was presented in [29]. It was shown that the
Ac
53 results obtained from this formulation are free from numerical instabilities in the high-frequency
54
55
56
57
58
59
60
1
2
3 range. The formulation was further extended for the solution of 2D response problems in a three-
4
pt
5 layer composite sandwich panel [30].
6
7 Many authors [9,31,32] presented the finite element method (FEM) based simulations of
8
9
elastic wave propagation in composite structures. In [33], the time-domain spectral element (SE)
cri
10 method based simulation technique was introduced that unites the flexibility of FEM and fast
11
12 convergence of spectral methods. A study on the advantages of higher-order FEM based
13
14 simulation techniques for solving elastodynamic problems was described in [34]. A relatively finer
15
discretisation (at least 15-nodes per wavelength) is required in case of FEM based simulations,
16
us
17 whereas, only 8-nodes per wavelength are required for the SE simulations. An appropriate node-
18
19 distribution in spectral-elements leads to the formation of diagonal elemental inertia-matrices and
20
21 controls the oscillations on the end of the element [35]. The SE simulation has the potential to
22
23
24
25
26
27
28
an
effectively solve the Lamb wave propagation problems in multi-layered structures [36-38]. In [39],
numerical integration operators which compose the stiffness and mass matrices of a modified SE
method were proposed for the numerical simulation of elastic wave propagation.
From the review of past literature, it is found that a very limited literature is available that
dM
29 deals with the temperature effects on Lamb wave propagation. Further, theoretical/numerical
30
31 studies on elastodynamic analysis of Lamb wave propagation in debonded BCSs subjected to
32
33 variable ambient temperature conditions are rather sparse. A theoretical dispersion analysis of
34 propagating Lamb waves, a series of time-domain three-dimensional (3D) SE simulation and
35
36 experimental analysis of Lamb wave propagation in a BCS under different temperature conditions
37
38 are presented in this paper. The SE simulation results in MATLAB are successfully verified with
pte
39
40
experiments using piezoelectric disc transducers (PDTs). In the numerical and experimental
41 signals, different Lamb wave modes are effectively identified using the theoretical dispersion
42
43 curve, which is obtained using a 2D semi-analytical model in FORTRAN. A 3D time-domain SE
44
45 simulation of Lamb wave propagation in a debonded BCS is carried out for different temperature
46
conditions, and the obtained simulation results are then successfully verified by conducting
ce
47
48 laboratory experiments using PDTs. It is envisaged that this research will help to understand the
49
50 debonding influence on the Lamb wave signals during SHM of BCS under variable temperature
51
52 conditions and would significantly guide the researchers to develop enhanced SHM strategies by
Ac
53 including appropriate temperature correction factors.

54
55
56
57
58
59
60
1
2
3 2. Experimental analysis
4
pt
5 2.1. Experimental setup
6
7 Experimental analysis has been carried out on a 7 mm thick rectangular BCS (250 mm × 250 mm)
8
9
sample plate with a 10 mm diameter debonding region at the bond-layer, in order to validate the
cri
10 SE simulation results. In the process, 0.4 mm thin circular PDTs of 10 mm diameter are mounted
11
12 at predefined locations to study the Lamb wave propagation in the presence and absence of
13
14 debonding region. The 17-layer BCS is made of two 3.49 mm thick 8-layer (each) carbon-fibre
15
reinforced cross-ply (0o/90o) composite laminates (CFRCL) bonded with 0.02 mm (approximate)
16
us
17 thin epoxy adhesive layer. The elastic material properties of the BCS are described in Table 1.
18
19 Table 1. Material properties of the sample BCS
20
21
Material E11 E22 E33 G12 G23 G13 12 13 23 
3
22 (GPa) (GPa) (GPa) (GPa) (GPa) (GPa) (kg/m )
23
24
25
26
27
28
CFRCL
Adhesive
73.02
4.052
73.02
4.052
9.8
4.052 an
3.61
1.447
3.38
1.447
3.38
1.447
0.03
0.40
0.37
0.40
0.37
0.40
1568
1150
An advanced multi-channel data-acquisition-cum-signal-generator (DAQ) instrument is

dM
29
30 employed to generate and receive the user-defined Lamb wave signals through the PDTs mounted
31
32
on the plate surface. A detailed description of the configured laboratory experimental setup with
33 sample BCS bonded PDTs is shown in Figure 1(a). The PDT actuator-sensor path#1-2 is
34
35 considered as the debonding -influenced signal path as the signals directly pass through the
36
37 debonding region and the actuator-sensor path#3-4 is assumed to be the healthy signal path as this
38
path is away from the debonding location. In each PDT path, the sensor is located at a distance of
pte
39
40 150 mm from the actuator location. In order to perform the temperature dependent experiments, a
41
42 temperature controller has been used along with the configured experimental setup as shown in
43
44 Figure 1(b). The temperature controller enables to perform the experiments range of ambient
45 temperature conditions, i.e. 0o C, 20o C, 40o C, and 60o C for the sample BCS.
46
ce
47
48
49
50
51
52
Ac
53
54
55
56
57
58
59
60
1
2
3
4
pt
5
6
7
8
9
cri
10
11
12
13
14
15
16
us
17
18
19 (a)
20
21
22
23
24
25
26
27
28
an
dM
29
30
31
32
33
34
35
36 (b)
37 Figure 1. Experimental setup to (a) analyse effects of debonding on propagating Lamb waves
38
and (b) to study the temperature influence on debonding in the sample BCS.
pte
39
40
41
42 In order to determine the optimal frequency of excitation of the PDTs for the BCS, a series
43
44 of Hanning window modulated 5-cycle tone-burst sine pulse with different central frequency is
45
46
actuated through PDT#3 and recorded at PDT#4 as healthy signals. A frequency-response plot is
ce
47 shown in Figure 2(a), which showed higher response magnitudes around 100 kHz central
48
49 frequency. Hence, the 100 kHz input pulse (Figure 2(b)) is selected for the experiment and
50
51 subsequent theoretical and SE based numerical studies unless otherwise stated.
52
Ac
53
54
55
56
57
58
59
60
1
2
3 1.0
4
pt
5
6
7
8
0.8
Normalized response
9
cri
10
11
12
13
14
15 0.6
16
us
17
18
19
20
21 0.4
22
23
24
25
26
27
28
50 100
an 150
Freqency (kHz)
200 250
dM
(a)
29
30 1.0
31
32
33
34
35 0.5
Normalized amplitude
36
37
38
pte
39
40 0.0
41
42
43
44
45 -0.5
46
ce
47
48
49
50 -1.0
51 0 10 20 30 40 50 60 70 80 90 100
52
Ac
53 Time (s)
54
55 (b)
56 Figure 2. (a) Frequency modulation curve and (b) Selected 100 kHz actuation signal.
57
58
59
60
1
2
3 3. Theoretical model of Lamb wave propagation in the BCS
4
pt
5 A global-matrix formation method based elastic wave propagation formulation is used for the
6
7 study of time-history response and dispersion characteristics of the Lamb wave propagation in the
8
9
BCS sample. The theoretical model is prepared for the 17-layer BCS by using a semi-analytical
cri
10 formulation for the solution of 2D response problems in a 3-layer composite panel as described in
11
12 [30]. In the analysis, the out-of-the-plane far-ﬁeld time-domain elastodynamic responses are
13
14 obtained at the reception point located at a distance of 0.15 m from the actuation point as shown
15
in Figure 3.
16
us
17
18
19
20
21
22
23
24
25
26
27
28
an
Figure 3. Schema of the semi-analytical model.
dM
29
30
In the semi-analytical analysis, double integral transforms are performed for the calculation of
31
32 nodal displacements in the time-space domain. The formulations of the semi-analytical model are
33
34 briefly described here and the detailed formulation can be found in [30]. In the 2D formulation,
35
36 the equations of motion in a layer of finite thickness are defined as:
37
 11  13  2u1
38   2 (1)
 x1  x3  t
pte
39
40
41  13  33  2u
42   23 (2)
43
 x1  x3  t
44
where ‘u’ represents the displacement components, ‘σ’ represents the stress components, ‘t’ is
45
46 time, and ‘x’ represents the coordinates (ref. Figure 3). The governing equations of motion are then
ce
47
48 represented in terms of potential functions and the field variables in the equations in the spatial
49
50 and time domain are converted to the frequency and wavenumber domain by performing Fourier
51 transformations. The solution of the frequency and wavenumber domain equations is referred to
52
Ac
53 as eigenvalue problem and for the non-trivial solution of the potential functions, the eigenvalues
54
55 are obtained as the functions of wavenumbers (k). Thus, the stresses and displacements in the
56
57
58
59
60
1
2
3 frequency and wavenumber domains are expressed in a condensed matrix form (ref. [30] for
4
pt
5 details). The boundary conditions and interface conditions are then applied to the equations.
6
7
8
The stress boundary conditions at the topmost surface are described as Q1   A1  Fˆ  (3)
and boundary conditions at the lowermost free surface are given as Q3   A3   0
9
cri
10 (4)
11
12 where Q (21) combines the eigenvectors with the wavenumbers and stiffness components (ref.
13
14 [30] for details),  A(21) composed of positive and negative unknown constants (A+ and A_) that
15
16
us
17 control the terms associated with upward going and down going waves corresponding to the
18
 0 
19 vertical wavenumbers, and F̂  is defined as   for out-of-the-plane excitations.
20 (21)
 F̂  21
21
22
23
24
25
26
27
28
no debonding at the two interfaces) are defined as:
an
The traction and displacement components are assumed to be continuous across the interfaces
parallel to the X1-X2 plane. The interface conditions for the healthy 3-layer sandwich structure (i.e.
ui ( m )  ui ( m 1) ,  i(3m )  i(3m 1) , X 3  X 3( m 1) (5)

dM
29
30 where ‘ X 3( m 1) ’ represents the location of the interface between the layers m (m = 2, 3) and m–1.
31
32 Thus at the first interface between the layer 1 and 2: Q1   A1   Q 2   A2  (6)
33
And at the second interface between layer 2 and 3: Q2   A2    Q 3   A3 
34
35 (7)
36
37 Eventually, the formulation for the 3-layer plate can be summarised as:
38
 Q21
1 1
E1     A1 1 
pte
39 Q22 0 0  Fˆ 
40    2  
 
  Q111 E1 Q121   Q112 Q122 E2   A    0 
41
 1 1  2   1 2 
1  2
0
42
 Q21 E Q22 
2
Q21 Q22 E    A   0 
43
  2     
3 
(8)
44
  Q112 E2 Q122   Q11 Q12 E  
3 3
 A    0  
45

0  2 2 2   3 3  1 3  0  
Q21 Q22 E  
3
46

Q21 E Q22 
  A    
 
ce
47  2   0 (121)
 Q213 E3 Q223  
48 0 0  (1212)   A  (121) 
49
50 where the {E}(2×1) represents the vertical propagation vectors (‘E_’ is corresponding to the upward
51
52 going and ‘E+’ for the downward going waves). The determinant of the [QE](12×12) represents the
Ac
53
54
dispersion condition for the BCS as:
55
56
G(k1 , f )  0 (9)
57
58
59
60
1
2
3
4 For different values of ‘f’’, the values of ‘ k1 ’ can be calculated, and the corresponding frequency
pt
5
6 versus phase-velocity (c) dispersion curves can be obtained as:
7
f
8 c (10)
9 k1
cri
10
11 The displacement condition in the wavenumber, ‘ k1 ’ and frequency, ‘f’’ domain can be described
12
13 as:
14
15 F ( k1 , f )
16 uˆ3 ( x1 , x3 , f )  (11)
us
17 G (k1 , f )
18
19 where ‘F’ and ‘G’ are the functions of frequency and wavenumber and obtained as described in
20
21 [30]. The Equation (11) can be solved for the real values of ‘ k1 ’ by applying the Cauchy’s residue
22
23
24
25
26
27
28
an
theorem described in [40]. In the analysis, the real roots of ‘ k1 ’ are only accepted for the solution
of propagating Lamb wave modes and the complex roots represent non-propagating wave modes
are rejected due to their exponential decay with the travel distance [41]. The solution for the
dM
29 spatial-domain dynamic response can be obtained as:
30  F (k1 , f )  k1x1
31 u3 ( x1 , x3 , f )   e dk1 (12)
 G ( k , f )
32 1
33
34 Equation (12) can be solved by applying Cauchy’s residue theorem [40-41] for the real roots of ‘
35
36 k R ’ where R = 1 to n and ‘N’ is the number of real roots:
37
38 N
F (k R , f )
u3 ( x1 , x3 , f )  2 i  eikR x1 (13)
pte
39
40 R 1 dG ( k1 , f )
41 dk1 k k
42 1 R
43
where ‘ u3 ’ is the out of the plane response in frequency domain. The time domain responses can
44
45
be obtained by conducting inverse Fourier transforms to Equation (13).
46
ce
47 The theoretical analysis is further extended for a range of temperature-dependent elastic

48
49 properties, in order to study the variable ambient temperature influence on the propagating Lamb
50
51 waves in the sample BCS and compared with SE based numerical simulations and possible
52 experimental results.
Ac
53
54
55
56
57
58
59
60
1
2
3 4. Spectral element based numerical simulation of Lamb wave propagation in the BCS
4
pt
5 A time-domain SE method [36] based numerical simulation of Lamb wave propagation in the
6
7 sample BCS under variable ambient temperature conditions have been carried out in MATLAB.
8
9
The SE analysis method is similar to the finite element analysis despite the difference in node
cri
10 distribution in spectral elements and the approximation functions that represents the changes in
11
12 calculated displacement components. In the spectral elements, the nodes are non–uniformly
13
14 distributed and their positions can be determined by solving the real roots of ‘ξi’ as described in
15
equation (14).
16
us
17
18 1    M2 '
a 1 
( )  0 

1   M  
19
b 1 ( )  0 
2 '
20 (14)
21 
22 1    M2 '
( 
)  0 
23
24
25
26
27
28
c 1

an
where ξ; η; ζ ∊ [-1; 1], M a 1 , M b 1 , M c 1 are the Legendre polynomials, a, b, c denotes the
number of nodes along ξ, η, ζ direction, and ‘′ ’ denotes the first derivative. A 3D shape function
is configured, which is a tensor product of 1D shape functions described in Figure 4 as: N j ( ) ,

dM
29
30
31 N k ( ) , Nl ( ) of Lagrange polynomials of (a – 1), (b – 1), (c – 1) degree can be represented as:
32
33
34
N p ( , ,  )  N j ( ) N k ( ) Nl ( ) (15)
35
36 where j  1  a, k  1  b, and l  1  c .
37
38
pte
39
40
41
42
43
44
45
46
ce
47
48
49
50
51
52
Ac
53
54
55
56
57
58
59
60
1
2
3
4
pt
5
6
7
8
9
cri
10
11
12
13
14
15
16
us
17
18
19
20
21
22
23
24
25
26
27
28
an
dM
29
30
31
Figure 4. Graphical representation of the 1D shape functions ( Ni ( ) ).
32
33
34
35 The element matrices are computed by using the Gauss–Lobatto–Legendre (GLL) integration
36 scheme where the GLL integration points coincide with the defined element-nodes. In the
37
38 formulation, the GLL quadrature is a tensor product of the 1D quadratures, which are associated
pte
39
40 with weights w j , wk , and wl :
41
42
2 
43 wj  
a (a  1)  M a 1 ( j )  
2
44
45 
46 2 
wk  2 
(16)
b(b  1)  M b 1 ( k )  
ce
47
48

49 2

wl 
50 c(c  1)  M c 1 ( l )  
2
51 
52
The constitutive equation can be obtained for linear piezoelectric materials as per [42]:
Ac
53
54    B F eT   X 
55     (17)
56 d   e  X  F 
57
58
59
60
1
2
3
where the tensors: ‘ B F ’ represents the elastic constants, ‘  X ’ represents the dielectric constants,
4
pt
5 ‘e’ represents the piezoelectric constants, ‘d’ represents the electrical-displacement components
6
7 and ‘F’ represents the electrical-field components, ‘X’ represents the strain components, and ‘σ’
8
9 represents the stress components. The superscripts: ‘T’, ‘F’ and ‘X’ represents the transpose
cri
10 matrix, initial electric-fields and initial strains, respectively. The governing equation of motion at
11
12 the elementary condition can be described as:
13
14 muu 0 u e  cuu 0 u e   kuu kue  u e   f e 
e e e
15        e T e  e
 e (18)
16  0 0  e   0 0  e  ku k     q 
us
17
e e
18 where ‘ muu ’ represents the structural mass-matrix, ‘ cuu ’ represents the damping-matrix, and
19
20 e e e
21 ‘ kuu ’ represents the stiffness-matrix, ‘ ku ’ represents the piezoelectric coupling-matrix, ‘ k ’
22
23
24
25
26
27
28
e an
represents the dielectric permittivity-matrix; ‘ u e ’ represents the nodal displacement-vector, ‘  ’
e
e
represents the electric potential-vector, ‘ f ’ represents the external force-vector; ‘ q ’ represents
the nodal externally applied charge vector. In the formulation, cuu can be defined as per the
e
dM
29
30 Rayleigh damping model (ref. [36] for the detail formulation) as:
31
32 cuue  m muue  k kuue (19)
33
34 where ‘ m ’ represents the mass proportionality coefficients and ‘ k ’ represents the stiffness
35
36 proportionality coefficient which is assumed to be zero in the present study. In the process, an
37
38 efficient central-difference scheme can be applied to solve the Equation (18) with reduced
pte
39
40 computation time.
41
42
43
44
45
46
ce
47
48
49
50
51
52
Ac
53
54
55
56
57
58
59
60
1
2
3
4
pt
5
6
7
8
9
cri
10
11
12
13
14
15
16
us
17
18
19
20
21
22
23
24
25
26
27
28
an
Figure 5. Numerical model of the sample BCS with the predefined network of PDTs.
dM
29
30
31
The numerical model for the PDT induced Lamb wave propagation in the sample BCS is shown
32 in Figure 5. In the simulation, the selected elastic material properties for the BCS part is shown in
33
34 Table 1, and the material properties for the Noliac NCE51 PDTs are selected as per the
35
36 manufacturer’s supplied data:
37
38  0 0 0 0 13.5 0 
 0 13.5 0 0  [C/ m 2 ],
charge constant:  e   0
pte
39 0
40
41  6.1 6.1 15.8 0 0 0 
42
43 1940 0 0 
44  0  *8.85*10-12 [F/ m],
permittivity:     0
X
45
1940
46  0 0 1900 
ce
47
48
49
50
51
52
Ac
53
54
55
56
57
58
59
60
1
2
3
4
134 8.89 90.9 0 0 0 
 0 134 90.9 0 
pt
5 0 0

6  0 0 121 0 0 0 
7 elastic stiffness matrix:  B F     [GPa]
8  0 0 0 20.5 0 0 
9  0 0 0 0 20.5 0 
cri
10  
11  0 0 0 0 0 22.4 
12
13
14 and mass density, ρ = 7290 [kg/m3].
15
16
us
17 Both, BCS and PDTs are modelled using 3D spectral elements consists of three through-
18 thickness nodes and thirty-six in-plane nodes with a total of 108 nodes as shown in Figure 6. Each
19
20 node has three displacement degrees of freedom (DOF) and one additional DOF (i.e. electric-
21
22 voltage) is considered for the PDTs.
23
24
25
26
27
28
an
dM
29
30
31
32
33
34
35
36
37
38
pte
39
40
41
42
43
44
45 Figure 6. Schematic representation of node distribution in a typical spectral element in the BCS.
46
ce
47
48 The number of in-plane spectral elements per layer is 3499, and the PDTs are connected to 24
49
50
elements in the top layer of the BCS. In the simulation, the contact effect is not considered and the
51 debonding regions are modelled by untying the appropriate nodes at the interface between the bond
52
Ac
53 layer and the top-CFRCL layer of the BCS (ref. Figure 6). The debonding region nodes are selected
54
55 from four different sizes of circular areas with diameter, d = 5 mm, 10 mm, 15 mm and 20 mm.
56
57
58
59
60
1
2
3 The simulation with debonding size 10 mm is a reference for comparison with experimental
4
pt
5 results.
6
7
8
9
5. Temperature-dependent elastic properties of BCS
cri
10 In order to carry out the temperature dependent SE simulation and semi-analytical analysis of
11
12 Lamb wave propagation in the BCS, the elastic modulus of BCS layers are calculated as per the
13
14 methodology described in [43-44]. The calculation is done for a range of temperature (-40 oC, -20
15 o
C, 0 oC, 20 oC, 40 oC, 60 oC, and 80 oC) generally occurred in practical operating scenarios. The
16
us
17 obtained temperature-dependent elastic properties (E11 = E22, E33, G12, G13 = G23) for composite
18
19 laminate and adhesive (E, G) are presented in Figure 7 and Figure 8, respectively.
20
21
74.0
Composite
22
23
24
25
26
27
28
73.6
73.2
an
dM
E11 (GPa)
29
30
31
32
72.8
33
34
35 72.4
36
37
38
72.0
pte
39
40
-40 -20 0 20 40 60 80
41 o
42
Temperature ( C)
43 (a)
44
45
46
ce
47
48
49
50
51
52
Ac
53
54
55
56
57
58
59
60
1
2
3 11.2
4 Composite
pt
5
6
10.8
7
8 10.4
9
cri
10 10.0
11
E33 (GPa)
12 9.6
13
14
15 9.2
16
us
17 8.8
18
19 8.4
20
21
22
8.0
-40 -20 0 20 40 60 80
23
24
25
26
27
28 4.2
an
Temperature ( C)
(b)
Composite
o
dM
29
30
31 4.0
32
33
34 3.8
35
G12 (GPa)
36
37 3.6
38
pte
39
40 3.4
41
42
43 3.2
44
45
46 3.0
ce
47 -40 -20 0 20 40 60 80
48 o
49 Temperature ( C)
50 (c)
51
52
Ac
53
54
55
56
57
58
59
60
1
2
3 1.95
4 Composite
pt
5
6
7
8 1.80
9
cri
10
11
G23(GPa)
12
13
1.65
14
15
16
us
17 1.50
18
19
20
21
22 1.35
23
24
25
26
27
28
-40 -20 0
an
Temperature ( C)
(d)
20
o
40 60 80
Figure 7. Temperature-dependent Young's modulus: (a) E11, (b) E33, and shear modulus: (c) G12,
(d) G23 of the composite layer.
dM
29
30 4.8
31 Adhesive
32
33
34 4.5
35
36
37
4.2
38
E (GPa)
pte
39
40
41 3.9
42
43
44
45 3.6
46
ce
47
48 3.3
49 -40 -20 0 20 40 60 80
50 o
51 Temperature ( C)
52 (a)
Ac
53
54
55
56
57
58
59
60
1
2
3
1.7
4
pt
5
6
7 1.6
8
9
cri
10
11 1.5
G (GPa)
12
13
14 1.4
15
16
us
17
18 1.3
19
20
21
22
1.2
23
24
25
26
27
28
-40 -20
(b)
0
an 20
Temperature ( C)
o
40 60 80
Figure 8. Temperature-dependent (a) Young's modulus (E) and (b) shear modulus (G) of the
dM
adhesive layer.
29
30
31
32 In this model [44], reduction of Young’s modulus of resin ‘𝐸𝑚 ’ with variation in temperature is
33
34 assumed as:
35
36 Em T   Fm * Erm (20)
37
38 where ‘𝐸𝑟𝑚 ’ is the Young’s Modulus of resin at the reference temperature and ‘𝐹𝑚 ’ is the
pte
39
40
temperature degradation factor as proposed in [43]:
41
42  T T 
Fm   g 0  (21)
43  T T
44  g0 r 
45
where 𝑇𝑔0 is the glass transition temperature and 𝑇𝑟 is the reference temperature. In the study, the
46
ce
47 value of 𝑇𝑔0 = 215oC and 𝑇𝑟 = 20oC is selected as per [43].

48
49
50
51 6. Results and discussions
52
Ac
53 6.1. Dispersion and mode identification of Lamb waves

54 In the semi-analytical analysis, the frequency versus phase-velocity dispersion curve and the time-
55
56 history responses are obtained for the Lamb wave propagation in a healthy BCS. The dispersion
57
58
59
60
1
2
3 plot in Figure 9 shows the presence of multiple Lamb wave modes corresponds to the real roots of
4
pt
5
‘ k1 ’ in Equation 10. The steepest-descent method is applied to draw these roots of ‘ k1 ’ as
6
7
described in [29]. In the plot, the anti-symmetric Lamb wave modes are designated as A0, A1, A2
8
9 and the symmetric-modes are designated as S0, S1, S2 [7]. The theoretical time-domain response
cri
10
11 generated by the predefined actuation signal (100 kHz) is obtained at a distance of 0.15 m from
12
13 the source location as shown in Figure 10. The theoretical signal at 100 kHz shows the existence
14 of two primary Lamb wave modes that are identified as S0 and A0 mode as per the dispersion plot
15
16 in Figure 9.
us
17
18 10
19 A1 S1 S2
A2
20
21
22 8
23
24
25
S0
an
Velocity (mm/s)
26 6
27
28
dM
29
30 4
31
32
33
34
35
2 A0
36
37
38
0
pte
39
40 0 100 200 300 400 500
41
42
Frequency (kHz)
43 Figure 9. Phase-velocity dispersion curves for the propagating Lamb waves in healthy BCS.
44
45
46
ce
47
48
49
50
51
52
Ac
53
54
55
56
57
58
59
60
1
2
3
4
1.0
S0 A0
pt
5
6
7
Normalized signal amplitude
8 0.5
9
cri
10
11
12
13 0.0
14
15
16
us
17
18 -0.5
19
20
21
22
23
24
25
26
27
28
-1.0
0.00000 0.00005
anTime (s)
0.00010
Figure 10. Semi-analytical signal at 100 kHz.

0.00015
dM
29
30
31 6.2. Analysis of debonding effects
32
33 The spectral element simulation results are obtained and a comparison of healthy and debonding-
34 influenced Lamb wave signals is shown in Figure 11(a). In the simulation signals, the A0 and S0
35
36 modes are effectively identified based on the semi-analytical response in Figure 10. Further, for a
37
38 better understanding of the debonding effects, the Hilbert-transform is performed on the signals in
pte
39
40
Figure 11(a) and presented in Figure 11(b). It is observed that the presence of debonding region in
41 the BCS slightly reduces the A0 wave mode amplitude. The amplitude reduction might have
42
43 happened due to a certain degree of energy-absorption at the debonding region, which may lead to
44
45 a higher dispersion and attenuation in A0 mode with respect to the S0 mode in the Lamb wave
46
signal at 100 kHz frequency. The debonding influence on Lamb wave propagation is clearly visible
ce
47
48 in the numerical waveform plot shown in Figure 12
49
50
51
52
Ac
53
54
55
56
57
58
59
60
1
2
3
4
1.0
A0 Healthy
pt
5
6 S0-reflection Debonding
7
8
S0
Normalized displacement
9
0.5
cri
10
11
12
13
14 0.0
15
16
us
17
18
19 -0.5
20
21
22
23
24
25
26
27
28
-1.0
0.00000 0.00005
an Time (s)
(a)
0.00010 0.00015
dM
29 1.00
30 A0 Healthy
31 Debonding
32
33 S0-reflection
34 0.75
35
36 S0
HT coefficient
37
38
0.50
pte
39
40
41
42
43
44 0.25
45
46
ce
47
48
49 0.00
50 0.00000 0.00005 0.00010 0.00015
51
52 Time (s)
Ac
53 (b)
54 Figure 11. Comparison of numerical (a) healthy and debonding signals, and (b) Hilbert
55 transform (HT) of (a) at 100 kHz.
56
57
58
59
60
1
2
3
4
pt
5
6
7
8
9
cri
10
11
12
13
14
15
16
us
17
18
19
20
21
22
23
24
25
26
27
28
an
dM
29
30
31
32
33
34
35
36 Figure 12. Waveform plot at 60 µs in the simulation shows the debonding effect on the
37
38 propagating Lamb waves.
pte
39
40
41
The comparison of experimentally obtained healthy and debonding-influenced signals and their
42
43 Hilbert-transform are presented in Figure 13(a) and Figure 13(b), respectively which also shows a
44
45 decrease in A0 mode amplitude due to the existence of debonding as noticed in the numerical
46
simulation. In all the cases, a good agreement among the semi-analytical, numerical simulation,
ce
47
48 and experimental results is observed. However, some S0 reflections appeared in the experimental
49
50 signals due to the high-velocity of the S0 mode and the finite size of the sample BCS.
51
52
Ac
53
54
55
56
57
58
59
60
1
2
3
4
1.0
A0 Healthy
pt
5
6
Debonding
S0-reflection
7
8
S0
Normalized displacement
9
0.5
cri
10
11
12
13
14
0.0
15
16
us
17
18
19 -0.5
20
21
22
23
24
25
26
27
28
-1.0
0.00000 0.00005 an Time (s)
(a)
0.00010 0.00015
dM
29
30
31
32
33
34
35
36
37
38
pte
39
40
41
42
43
44
45
46
ce
47
48
49
50
51
52
Ac
53 (b)
54 Figure 13. Comparison of experimental (a) healthy and debonding signals, and (b) Hilbert
55
56
transform (HT) of (a) at 100 kHz.
57
58
59
60
1
2
3 In order to study the debonding size effects on the A0 mode propagating Lamb wave signals in
4
pt
5 the BCS, the numerical simulation is further extended for a range of debonding size (diameter: 0
6
7 mm (healthy), 5 mm, 10 mm, 15 mm, and 20 mm). The simulation results are also compared with
8
9
available experimental results which shows a good agreement as presented in Figure 14.
cri
10
11 0.0130
12 Simulation
13
14 0.0125 Experiment
15
16
us
Amplitude (magnitudes)
17 0.0120
18
19
20
21
0.0115
22
23
24
25
26
27
28
0.0110
0.0105
an
dM
29
30 0.0100
31
32
33 0.0095
34
35
0 5 10 15 20
36 Debonding region diameter (mm)
37
38 Figure 14. Debonding size effect on the A0 mode amplitudes at 100 kHz.
pte
39
40
41 6.3. Analysis of temperature effects on Lamb waves
42
43
In order to study the temperature effect on propagating Lamb waves in the BCS, the semi-
44 analytical analysis for the healthy condition, and the SE simulation and experimental analysis for
45
46 healthy and debonding conditions are carried out. In the semi-analytical analysis and SE
ce
47
48 simulations, the temperature dependent elastic material properties (ref. section 5) are used for the
49
BCS model to simulate different ambient temperature effects. The temperature effect on A0 mode
50
51 velocity and amplitude is obtained for a range of general operating temperature conditions (-40
52
Ac
o
53 C, -20 oC, 0 oC, 20 oC, 40 oC, and 60 oC) and presented in Figure 15 and Figure 16, respectively.
54
55 It is observed that the variation in ambient temperature condition can substantially influence the
56
57
58
59
60
1
2
3 amplitude difference and change the propagation velocity of the wave mode. The increase in
4
pt
5 temperature is leading to the reduction in propagation velocity and amplitude of the A0 mode. In
6
7 all the study cases, only the A0 wave mode effects are specifically considered due to its high-
8
9
debonding sensibility and lower propagation velocities in comparison to the S0 mode.
cri
10
11
1435
12
13
14
15 1400
16
us
17
18
19
Velocity (m/s)
20 1365
21
22
23
24
25
26
27
28
1330 Healthy_Th.
Healthy_Num.
Healthy_Exp.
an
dM
29
30 1295 Debonding_Num.
31 Debonding_Exp
32
33
34 -40 -20 0 20 40 60
35 o
36 Temperature ( C)
37
Figure 15. Temperature effect on A0 mode velocity in the healthy and debonding signals.
38
pte
39
40
41
42
43
44
45
46
ce
47
48
49
50
51
52
Ac
53
54
55
56
57
58
59
60
1
2
3
4 0.014
pt
5
6
7
8
9
cri
Amplitude (magnitude)
10
11 0.012
12
13
14
15
16
us
17
18
0.010 Healthy_Th.
19
20 Healthy_Num.
21
22
Healthy_Exp.
23
24
25
26
27
28
0.008
-40
Debonding_Num.
Debonding_Exp.
-20
an
0 20 40 60
dM
o
29 Temperature ( C)
30
31
Figure 16. Temperature effect on A0 mode amplitude in the healthy and debonding signals.
32
33 The temperature-dependent debonding effects distinctly emerged from the waveform plots in
34
35 Figure 17, obtained from the 3D SE simulations corresponding to the PDT path: Actuator#1-
36
37 Sensor#2 (ref. Figure 5). The plots clearly indicate that the debonding influence is higher for both
38
high (positive) and low (negative) temperature ranges with respect to the reference temperature
pte
39
40 (i.e. 20 oC).
41
42
43
44
45
46
ce
47
48
49
50
51
52
Ac
53
54
55
56
57
58
59
60
1
2
3
4
pt
5
6
7
8
9
cri
10
11
12
13
14
15
16
us
17
18
19
20
21
22
23
24
25
26
27
28
an
dM
29 (a)
30
31
32
33
34
35
36
37
38
pte
39
40
41
42
43
44
45
46
ce
47
48
49
50
51
52
Ac
53
54
55
56
(b)
57
58
59
60
1
2
3
4
pt
5
6
7
8
9
cri
10
11
12
13
14
15
16
us
17
18
19
20
21
22
23
24
25
26
27
28
an
dM
29 (c)
30 Figure 17. Waveform plots at (a) -40 oC, (b) 20 oC, and (c) 60 oC showing the temperature
31 influence on debonding effects at 100 μs.
32
33
34 From the analysis of results, it is evident that the change in ambient temperature can significantly
35
36 influence the amplitude and/or velocity change based SHM strategies and may lead to anomalous
37
measurements if the material wise temperature-corrections are not included in these strategies.
38
pte
39 Therefore, the study is further extended for temperature influence on SHM strategies. In order to
40
41 effectively identify the optimum debonding size and location in BCS by using the A0 mode
42
43 amplitude difference (healthy – debonding) based SHM strategies, a temperature correction factor
44 (CFTA) for different ambient temperature conditions is proposed as-
45
46 ( Arh  Ard ) Ar
CFTA  
ce
47 (22)
48 ( Aih  Aid ) Ai
49
50 where ‘Arh’ and ‘Ard’ are the A0 mode amplitude at the reference temperature for the healthy and
51 debonding-influenced Lamb wave signals, respectively, and ‘Aih’ and ‘Aid’ are the A0 mode
52
Ac
53 amplitudes at the actual operating temperature for the healthy and debonding-influenced signals,
54
55 respectively. Based on the proposed formula in Equation 22, the temperature correction factors are
56
57
58
59
60
1
2
3 obtained for the BCS for the given conditions and presented in Figure 18. Temperature-wise value
4
pt
5 of the given correction factor is needed to be multiplied by the amplitude difference calculated
6
7 beyond the reference temperature conditions for the BCS.
8
9
1.8
cri
10
11
12
13
1.6
14
15
16 1.4
CFTA (magnitude)
us
17
18
19
20 1.2
21
22
23
24
25
26
27
28
1.0
0.8
an
dM
29
30
31
32
0.6
33
-40 -20 0 20 40 60
o
34
35
Ambient temperature ( C)
36 Figure 18. Temperature correction factor (CFTA) for A0 mode amplitude difference (healthy-
37
38 debonding) in BCS at 100 kHz.
pte
39
40
41 Similarly, a temperature correction factor (CFTv) for the A0 mode velocity difference (healthy –
42
43 debonding) based SHM strategies is proposed as-
44
45 (vrh  vrd ) vr
46 CFTv   (23)
(vih  vid ) vi
ce
47
48
where ‘vrh’ and ‘vrd’ are the A0 mode velocity at the reference temperature for the healthy and
49
50 debonding-influenced Lamb wave signals, respectively, and ‘vih’ and ‘vid’ are the A0 mode velocity
51
52 at the actual operating temperature for the healthy and debonding-influenced signals, respectively.
Ac
53
54 Based on the proposed formula in Equation 23, the temperature correction factors are obtained for
55 the BCS for the given conditions and presented in Figure 19. An appropriate value of the given
56
57
58
59
60
1
2
3 correction factor is required to be multiplied by the velocity difference determined beyond the
4
pt
5 reference temperature for the BCS.
6
7 1.8
8
9
1.6
cri
10
11
12
13 1.4
14
15
16
1.2
us
17
18 1.0
CFTV
19
20
21 0.8
22
23
24
25
26
27
28
0.6
0.4
an
dM
29
30
0.2
31
-40 -20 0 20 40 60
32 o
33
Ambient temperature ( C)
34
35
Figure 19. Temperature correction factor (CFTV) for A0 mode velocity difference (healthy-
36 debonding) in BCS at 100 kHz.
37
38
pte
39
40 Further, the debonding size effects on the A0 mode amplitudes (Ref. Figure 14) in the propagating
41 Lamb wave signals in the BCS under variable temperature conditions are obtained and presented
42
43 in Figure 20. The results show that the variation in ambient temperature significantly influences
44
45 the debonding size effects on the A0 Lamb mode. The proposed temperature correction factor for
46
amplitude difference (CFTA) is then applied to the results described in Figure 20. The analysis
ce
47
48 results in Figure 21 clearly shows that the proposed CFTA has the potential to effectively minimize
49
50 the temperature variation influence on the A0 mode amplitude difference under the given
51
52 conditions (frequency, material, and damage type).
Ac
53
54
55
56
57
58
59
60
1
2
3
4
0.0168 o
-40 C
pt
5 o
-20 C
6
7
0.0156
o
8 0C
9 0.0144
Amplitude (magnitude)
cri
10
11 o
12 0.0132 20 C (Reference)
13
14
15 0.0120
o
16 40 C
us
17
18
0.0108
19
20 0.0096 o
21 60 C
22
23
24
25
26
27
28
0.0084
0.0072
0 5
an 10 15
Debonding diameter (mm)
20
dM
29
30
31
Figure 20. Debonding size effects on the A0 mode amplitudes under variable temperature
32 conditions.
33
34
35
36
37
38
pte
39
40
41
42
43
44
45
46
ce
47
48
49
50
51
52
Ac
53
54
55
56
57
58
59
60
1
2
3
4 0.0130
pt
5
6
7
8
0.0125
9
Amplitude (magnitudes)
cri
10
11 0.0120
12
13
14 o
15 0.0115 -40 C
16 o
-20 C
us
17
o
18
19
0.0110 0 C
o
20 20 C
21
o
22 0.0105 40 C
23
24
25
26
27
28
0.0100
0
60 C
o
5
an 10 15 20
dM
29
30 Debonding size (mm)
31
32 Figure 21. Debonding size effects on the A0 mode amplitudes under variable temperature
33
conditions after applying the proposed CFTA values described in Figure 18.
34
35
36
37 Analysis of temperature dependent debonding effects on the primary antisymmetric mode in the
38
propagating Lamb waves is the prime focus of this paper as the differential features of the A0
pte
39
40 mode is widely used for damage detection in composites. Perhaps, a more detail investigation is
41
42 required to carry out for a wide range of composite structures and for a range of operating
43
44 frequencies to support robust online monitoring strategies.
45
46
ce
47 7. Conclusion
48
49  Lamb wave propagation in the BCS at 100 kHz generates only the fundamental wave modes
50
51 (A0 and S0) for the structure considered in this work.
52

Ac
53 Presence of debonding region at the bond-interphase significantly reduces the A0 mode

54 amplitude and the amplitude difference further increases due to the increase in ambient
55
56
57
58
59
60
1
2
3 temperature. The increase in temperature also decreases the propagation velocity of the A0
4
pt
5 wave modes in the BCS.
6
7  It is found that the increase in debonding size leads to the reduction in A0 mode amplitudes
8
9 and the increase in ambient temperature further decreases the amplitudes.
cri
10  These temperature influences can potentially lead to an incorrect measurement for the popular
11
12 online SHM frameworks that work on the basis of amplitude and/or velocity difference of A0
13
14 modes in guided wave signals. Therefore, it is important to effectively adopt specific
15
16
temperature influence factors in such SHM strategies to improve accuracy and avoid
us
17 measurement anomalies during the inspection.
18
19  The proposed temperature correction factors have shown the effectiveness to minimize the
20
21 amplitude change difference for the given conditions and would significantly help to develop
22
23
24
25
26
27
28
an
enhanced SHM strategies for more precise identification of debonding in composite structures
under variable temperature conditions. A more detail investigation on higher order wave modes
and for a wide range of operating frequencies for a different type of composites would be the
future research scope to support the development of robust SHM strategies, which is an
dM
29
ongoing research by the authors.
30
31
32
33 Acknowledgement
34
35 The research work is funded by the Polish National Science Centre (NCN) under grant agreement
36 number: UMO-2016/23/N/ST8/01326 in the frame of the PRELUDIUM 12 project.
37
38
pte
39
40 References
41
42
[1] Peters ST. Handbook of composites. Boca Raton, FL: Chapman and Hall; 1998.
43 [2] Gay D, Hoa SV, Tsai, SN. Composite materials: design and application. New York, CRC;
44 2003.
45 [3] Safri SN, Sultan MT, Jawaid M, Jayakrishna K. Impact behaviour of hybrid composites for
46 structural applications: A review. Composites Part B: Engineering 2017;133:112-121.
ce
47
[4] Kersemans M, De Baere I, Degrieck J, Van Den Abeele K, Pyl L, Zastavnik F, Sol H, Van
48
49 Paepegem W. Nondestructive damage assessment in fiber reinforced composites with the
50 pulsed ultrasonic polar scan. Polymer Testing 2014;34:85-96.
51 [5] Giurgiutiu V, Zagrai AN, Bao JJ. Piezoelectric wafer embedded active sensors for aging
52 aircraft structural health monitoring. Structural Health Monitoring 2002;1: 41–61.
Ac
53
54 [6] Bogenfeld R, Kreikemeier J, Wille T. An analytical scaling approach for low-velocity impact
55 on composite structures. Composite Structures. 2018;187:71-84.
56
57
58
59
60
1
2
3 [7] Giurgiutiu V. Tuned Lamb wave excitation and detection with piezoelectric wafer active
4
pt
5 sensors for structural health monitoring. Journal of Intelligent Material Systems and
6 Structures 2005;16:291-305.
7 [8] Hay TR, Wei L, Rose JL. Rapid inspection of composite skin-honeycomb core structures
8 with ultrasonic guided waves. Journal of Composite Materials 2003;37(10):929–939.
9
[9] Sikdar, S., Kudela, P., Radzieński, M., Kundu, A. and Ostachowicz, W., “Online detection of
cri
10
11 barely visible low-speed impact damage in 3D-core sandwich composite structure”,
12 Composite Structures 2018;185,646-55.
13 [10] Maslov KI, Kundu T. Selection of lamb modes for detecting internal defects in
14
laminated composites. Ultrasonics 1997;35:141–150.
15
16
[11] Mitra M, Gopalakrishnan S. Guided wave based structural health monitoring: A review.
us
17 Smart Materials and Structures. 2016;25(5):053001.
18 [12] Wandowski T, Malinowski PH, Ostachowicz WM. Circular sensing networks for guided
19 waves based structural health monitoring. Mechanical Systems and Signal Processing
20 2016;66:248-67.
21 [13] Yang B, Xuan FZ, Xiang Y, Li D, Zhu W, Tang X, Xu J, Yang K, Luo C. Lamb wave-
22
23
24
25
26
27
28
2017;10(6):652.
an
based structural health monitoring on composite bolted joints under tensile load. Materials
[14] Balasubramaniam K. Lamb‐wave‐based structural health monitoring technique for

inaccessible regions in complex composite structures. Structural Control and Health
Monitoring 2014;21(5):817-32.
dM
[15] Lowe MJS, Challis REC, Chan W. The Transmission of Lamb waves across adhesively
29 bonded lap joints. Journal Acoustical Society of America 2000;107(3):1333–1345.
30
31
[16] Baid H, Schaal C, Samajder H et al (2015). Dispersion of Lamb waves in a honeycomb
32 composite sandwich panel. Wave motion 56:409-416.
33 [17] Clarke T, Simonetti F, Cawley P. Guided wave health monitoring of complex structures by
34 sparse array systems: Influence of temperature changes on performance. Journal of Sound and
35 Vibration. 2010 Jun 7;329(12):2306-22.
36 [18] Anaya M, Tibaduiza DA, Torres-Arredondo MA, Pozo F, Ruiz M, Mujica LE, Rodellar J,
37
Fritzen CP. Data-driven methodology to detect and classify structural changes under
38
temperature variations. Smart Materials and Structures. 2014 Feb 28;23(4):045006.
pte
39
40 [19] Dan CA, Kudela P, Radzieński M, Ostachowicz W. Temperature effects compensation
41 strategy for guided wave based structural health monitoring. InProceedings of the 6th
42 international symposium on NDT in aerospace, Madrid 2014 Nov (pp. 12-14).
43
[20] Putkis O, Dalton RP, Croxford AJ. The influence of temperature variations on ultrasonic
44
45 guided waves in anisotropic CFRP plates. Ultrasonics. 2015 Jul 1;60:109-16.
46 [21] Salmanpour MS, Sharif Khodaei Z, Aliabadi MH. Guided wave temperature correction
ce
47 methods in structural health monitoring. Journal of Intelligent Material Systems and Structures.
48 2017;28(5):604-18.
49 [22] Sohn H, Park G, Wait JR, Limback NP, Farrar CR. Wavelet-based active sensing for
50 delamination detection in composite structures. Smart Materials and structures
51
52
2003;13(1):153.
Ac
53 [23] Nilsson AC. Wave propagation in and sound transmission through sandwich plates. Journal
54 of Sound and Vibration 1990;138(1):73-94.
55 [24] Qiao P, Yang M. Impact analysis of fiber reinforced polymer honeycomb composite
56 sandwich beams. Composites Part B: Engineering. 2007;38(5):739-50.
57
58
59
60
1
2
3 [25] Sorokin SV. Analysis of propagation of waves of purely shear deformaton in a sandwich
4
pt
plate. Journal of sound and vibration 2006;291(3):1208-20.
5
6 [26] Backstöm D, Nilsson A. Modeling flexural vibration of a sandwich beam using modified
7 fourth-order theory. Journal of Sandwich Structures & Materials 2006;8(6):465-76.
8 [27] Nosier A, Kapania RK, Reddy JN. Free vibration analysis of laminated plates using a layer
9 wise theory. AIAA Journal. 1993;31:2335–2346.
cri
10 [28] Liu L, Bhattacharya K. Wave propagation in a sandwich structure. International Journal of
11 Solids and Structures 2009;46(17):3290-300.
12
13
[29] Mal A.K. Wave propagation in layered composite laminates under periodic surface loads.
14 Wave motion 1988;10(3):257-66.
15 [30] Banerjee S, Pol CB. Theoretical modeling of guided wave propagation in a sandwich plate
16 subjected to transient surface excitations. International Journal of Solids and Structures
us
17 2012;49(23):3233-41.
18 [31] Zhang B, Sun XC, Eaton MJ, Marks R, Clarke A, Featherston CA, Kawashita LF, Hallett
19
SR. An integrated numerical model for investigating guided waves in impact–damaged
20
21 composite laminates. Composite Structures 2017;176:945-960.
22 [32] Hosseini SM, Gabbert U. Numerical simulation of the Lamb wave propagation in
23
24
25
26
27
28
an
honeycomb sandwich panels: a parametric study. Composite Structures 2013;97:189-201.
[33] Patera AT. A spectral element method for fluid dynamics: laminar flow in a channel
expansion. Journal of computational Physics 1984;54(3):468-88.
[34] Willberg C, Duczek S, Perez JV, Schmicker D, Gabbert U. Comparison of different higher
order finite element schemes for the simulation of Lamb waves. Computer methods in applied
dM
29 mechanics and engineering 2012;241:246-61.
30 [35] Ostachowicz W, Kudela P, Krawczuk M, Zak A. Guided waves in structures for SHM: the
31 time-domain spectral element method. John Wiley & Sons; 2011.
32 [36] Kudela P, Żak A, Krawczuk M, Ostachowicz W. Modelling of wave propagation in
33 composite plates using the time domain spectral element method. Journal of Sound and
34 Vibration 2007;302(4-5):728-45.
35
36
[37] Kudela P. Parallel implementation of spectral element method for Lamb wave propagation
37 modeling. International Journal for Numerical Methods in Engineering 2016;106(6):413-29.
38 [38] Ha, S., & Chang, F. K. Optimizing a spectral element for modeling PZT-induced Lamb
pte
39 wave propagation in thin plates. Smart materials and structures 2009;19(1):015015.

40 [39] Hasegawa K, Fuji N, Konishi K. Improvement of accuracy of the spectral element method
41 for elastic wave computation using modified numerical integration operators. arXiv preprint
42
arXiv:1801.02829. 2018 Jan 9.
43
44 [40] Ahlfors LV. Complex Analysis. McGraw Hill. 1979;ISBN: 0-07-085008-9.
45 [41] Vasudevan N, Mal AK. Response of an elastic plate to localized transient sources. ASME
46 J. Appl. Mech. 1985;52(2):356-62.
ce
47 [42] Giurgiutiu V, Lyshevski SE. Micromechatronics: Modeling, analysis, and design with
48 MATLAB. CRC Press; 2016.
49
50
[43] Chamis, C.C. Simplified composite micromechanics equations for hygral, thermal and
51 mechanical properties. Thirty-eighth Annual Conference of the Sodety of the Plastics Industry
52 (SPI) Reinforced Plastics/Composite Institute, Houston, Texas, February 7-11, 1983.
Ac
53 [44] Salamone S, Bartoli I, Lanza Di Scalea F, Coccia S. Guided-wave health monitoring of

54 aircraft composite panels under changing temperature. Journal of Intelligent Material Systems
55 and Structures. 2009;20(9):1079-90.
56
57
58
59
60

1361 665x/aaefaa

Uploaded by

Copyright:

Available Formats

1361 665x/aaefaa

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1361 665x/aaefaa

Uploaded by

Copyright:

Available Formats

Smart Materials and Structures

Effects of debonding on lamb wave propagation in a bonded composite

Manuscript version: Accepted Manuscript

This Accepted Manuscript is © 2018 IOP Publishing Ltd.

View the article online for updates and enhancements.

This content was downloaded from IP address 142.132.1.147 on 16/11/2018 at 02:02

53 including appropriate temperature correction factors.

An advanced multi-channel data-acquisition-cum-signal-generator (DAQ) instrument is

ui ( m )  ui ( m 1) ,  i(3m )  i(3m 1) , X 3  X 3( m 1) (5)

47 The theoretical analysis is further extended for a range of temperature-dependent elastic

is configured, which is a tensor product of 1D shape functions described in Figure 4 as: N j ( ) ,

47 value of 𝑇𝑔0 = 215oC and 𝑇𝑟 = 20oC is selected as per [43].

53 6.1. Dispersion and mode identification of Lamb waves

Figure 10. Semi-analytical signal at 100 kHz.

53 Presence of debonding region at the bond-interphase significantly reduces the A0 mode

[14] Balasubramaniam K. Lamb‐wave‐based structural health monitoring technique for

39 wave propagation in thin plates. Smart materials and structures 2009;19(1):015015.

53 [44] Salamone S, Bartoli I, Lanza Di Scalea F, Coccia S. Guided-wave health monitoring of

You might also like