You are on page 1of 64

Journal Pre-proofs

Environmental and Operational Conditions Effects on Lamb Wave Based


Structural Health Monitoring Systems: A Review

Rahim Gorgin, Ying Luo, Zhanjun Wu

PII: S0041-624X(20)30053-6
DOI: https://doi.org/10.1016/j.ultras.2020.106114
Reference: ULTRAS 106114

To appear in: Ultrasonics

Received Date: 21 November 2019


Revised Date: 20 February 2020
Accepted Date: 26 February 2020

Please cite this article as: R. Gorgin, Y. Luo, Z. Wu, Environmental and Operational Conditions Effects on Lamb
Wave Based Structural Health Monitoring Systems: A Review, Ultrasonics (2020), doi: https://doi.org/10.1016/
j.ultras.2020.106114

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. Please note that, during the production process, errors
may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

© 2020 Published by Elsevier B.V.


Environmental and Operational Conditions Effects

on Lamb Wave Based Structural Health Monitoring

Systems: A Review

Rahim Gorgin1, Ying Luo1, Zhanjun Wu*2

Faculty of Civil Engineering and Mechanics, Jiangsu University, Zhenjiang 212013, Jiangsu, China

2 State Key Laboratory of Structural Analysis for Industry Equipments, School of Aeronautics and Astronautics,
Dalian University of Technology, Dalian 116024, Liaoning, China

Abstract:

Lamb wave is widely recognized as one of the most encouraging tools for structural health

monitoring (SHM) systems. In spite of many favourable characteristics of Lamb wave for

SHM, real-world application of these systems is still quite limited. Beside the complexities

derived from multi-modal, dispersive and multi-path characteristics of Lamb waves, one of

the main challenges in Lamb wave based SHM is sensitivity of these systems to

environmental and operational conditions (EOCs) parameters. This paper provides a state of

the art review of the effects of EOCs parameters including: temperature, moisture, load,

vibration and bonding (adhesive layer shear modulus and thickness, bond defects), on Lamb

wave propagation. Moreover, this paper provides a summary of compensation strategies to

account for EOCs effects as well as baseline free techniques. An objective is also to

understand the future directions and areas requiring attention of the researchers.

Keywords

* Corresponding author, Professor


E-mail: wuzhj@dlut.edu.cn
Structural Health Monitoring (SHM); Lamb waves; Environmental and Operational

conditions; Compensation strategies; Baseline free techniques.

1. Introduction

Structural health monitoring (SHM) is an emerging technology with signal processing

algorithms to monitor the ‘health’ status of structures in real time or whenever necessary.

Toward this topic, Lamb waves, ultrasonic guided waves that propagate inside thin-wall

plates and shallow shells, have been increasingly employed to develop various SHM

techniques [1–27]. Thanks to the ability of Lamb waves to propagate over large distances

especially in metallic structures (in composite structures the Lamb wave attenuation is very

fast) and to transmit through various structural features, sparse sensor arrays consisting of a

small number of sensors per area that can be integrated into (or applied to the surface of)

complex parts and used to gather information about changes within structures, they are

suitable for SHM [28–33].

When Lamb wave is incident on damage, scattering happens in all directions. Such scattering

waves can be utilized to investigate the damage. An inherent issue with the use of Lamb

waves for SHM is the complexity of the measured signals. In most cases, more than one

mode as well as many reflections are present. It is thus difficult to identify the cause of each

feature in a given time trace and to detect and interpret small scatters due to changes in the

complex Lamb wave signals. A way to simplify the detection of damage scatters is to

compare a signal captured during the operational life of the structure (current signal) with a

signal captured when the structure is known to be undamaged (baseline signal) [34] (see Fig.

1(a)). The simplest way to do this is to subtract the two signals. Because the waves (including

different modes and reflections) from benign structural features remain unchanged, they are

removed (or reduced significantly) on subtraction. This eliminates the need for interpretation
of the complex raw time signal and any damage scatter will be seen clearly since the

amplitude of the residual signal obtained after subtraction is sufficiently low (see Fig. 1(b)).

Different damage scatter properties (e.g. time of flight (TOF), amplitude, energy, …) can

then be used in Lamb wave based SHM techniques for damage identification and

characterization.

Fig. 1. (a) Baseline and current signals, collected at the same EOCs, -- baseline, current; (b) Scatter signal

determined from subtraction of baseline from current signal, collected at the same EOCs; (c) Baseline and

current signals collected at different EOCs, baseline, current; (d) Scatter signal determined from subtraction

of baseline from current signal, collected at different EOCs.

It has been shown by different authors [35–48] that environmental and operational conditions

(EOCs) parameters including, temperature, vibration, load, moisture, bond defects, adhesive

layer shear modulus and thickness effects Lamb wave signals. EOCs parameters may change

signal properties such as phase and amplitude which leads to the changes in the shape of the

signal (see Fig. 1(c)). Therefore, simple subtraction of one baseline from the current is not

sufficient to discriminate between changes due to damage and those false scatters due to

EOCs parameters (see Fig. 1(d)). If this damage scatter signal be used in Lamb wave based

SHM techniques will results in false damage identification and characterization. Therefore,
the research of EOCs effects on Lamb wave propagation and compensation strategies to

account for these factors are required to enable the use of strategies developed under

laboratory conditions in real world applications.

Several comprehensive reviews [49–52] of Lamb wave based SHM systems already exist in

literature, however, those are mostly have been focused on development of suitable

modelling schemes, transducer technologies, signal processing, feature extraction, statistical

and machine learning schemes. Although, Lamb wave is a suitable tool for damage

identification and characterization in various structures, its real-world application is still quite

limited. This is not only due to the complexities derived from multi-modal, dispersive and

multi-path characteristics of Lamb waves, but also because of its sensitivity to EOCs

parameters. Hence, this paper, provides a summary of the effects of environmental and

operational conditions parameters on Lamb wave propagation as well as the characterization

of the system response under these parameters variations. Also, various developed

compensation strategies to account for these effects are introduced. Moreover, a brief

summary of baseline free techniques is presented in this paper. Furthermore, the future

directions and areas that need to be investigated by researchers are discussed.

The paper is organized as follows. Firstly, the effects of environmental conditions, including

temperature and moisture, on Lamb wave propagation and their compensation strategies is

summarized. Secondly, a literature review on the effects of operational conditions including,

vibration, applied loads and bonding (adhesive layer shear modulus and thickness, bond

defects) on Lamb wave propagation is provided. Third section briefly reviews the recent

developments on baseline free techniques for Lamb wave based SHM systems. The paper

ends with important conclusions and an attempt to enumerate the scopes of further research.

2. Environmental Conditions
It is clear that structures used in the field of aerospace, mechanical and civil engineering

experience different environmental conditions during their service. These environmental

variations will affect Lamb wave based SHM systems using for such structures. Hence, this

section, provides a summary of the effects of environmental condition parameters including

temperature, and moisture on Lamb wave propagation and their compensation strategies.

2.1. Temperature

Among various environmental condition parameters, it is well known that studying

temperature changes, is of particular interest because temperature is a global environmental

condition and its variations is substantially limiting Lamb waves based SHM techniques.

2.1.1. Temperature effects on Lamb wave propagation;

Many researchers have experimentally evaluated the effects of temperature variations on

Lamb wave propagation. Generally, temperature variations affect amplitude and arriving time

of Lamb waves [35]. Ahmed and Kopsaftopoulos have shown that temperature variations as

small as 0.5°C may cause changes in the amplitude of lamb waves [53]. Boon et al. [54] have

experimentally shown that by increasing temperature, the wave group velocity decreases. It

should be noted that the experimental measurement of TOF using a toneburst signal leads to

an estimate of group velocity (as correctly stated in reference [54]), not phase velocity which

is stated in some references listed in this review. Sikdar et al. [41] have shown that the

primary anti-symmetric Lamb wave mode amplitude and velocity decrease with an increase

in ambient temperature. Blaise and Chang [55] have shown the decrease in time of flight

(TOF) for temperatures going down to −90°C. They have also observed the increase of

response amplitude by decreasing the temperature. In another study, the decrease of the

amplitude with increasing temperatures from 30°C to 70°C has been observed by Lee et al.

[56]. Although these two studies have shown the decrease of response amplitude as a result of
temperature increasing, Marzani and Salamone [57] have observed that the maximum system

response amplitude is at the ambient temperature of +20°C. They have shown that by

increasing or decreasing the temperature the system response decreases. They have also

compared Lamb waves collected at two extreme temperatures, i.e. -40°C and +60°C, at the

frequencies of 200 KHz (Fig. 2(a)) and 350 KHz (Fig. 2(b)).

Fig. 2. Experimental received Lamb waves at -40°C and +60°C at the frequency of: (a) 200 kHz and (b) 350

kHz [57].

It can be observed from Fig. 2 that at the temperature of +60°C Lamb wave present a larger

time of flight (i.e., slower speed of propagation) if compared to those at the temperature of -

40°C which is similar to what is concluded in references [55,56]. However, as can be seen in

Fig. 2, while at 200 kHz the signal at +60°C has in average a bigger amplitude with respect to

that at -40°C, at 350 kHz the opposite behaviour appears. The same result was obtained by Di

Scalea et al. [58]. Hence, the system response cannot be linearly related to temperature and

many mechanisms induce changes in Lamb wave signals as a result of temperature variations,

including:

a) Changes in the stiffness of the plate’s material affect the waves phase/group velocity

at a given frequency, which results in variation in their measured time of flight (TOF)
and waves wavelengths. For many materials, the elastic and shear moduli vary with

temperature. The moduli typically decrease with increasing temperature and, as a

result, the longitudinal and transverse wave speeds decrease which subsequently

result in decreasing phase velocities of the propagating wave components. Thus,

signals propagated at higher temperatures are expected to arrive later than signals

propagated at lower temperatures. It will also cause inducing a shift in the frequency

response of transducers (wavelength tuning effect) [59].

b) The plate in-plane expansion/contraction, changes the distance between actuator and

sensor causing a deviation in the TOF of incoming and reflected modes. Due to the

plate out-of-plane expansion/contraction, wave propagates at a slightly different phase

speed or wavelength which in accordance to the constant frequency-thickness phase

velocity curves, cause a shift in the frequency response for both actuator and sensor

(wavelength tuning effect).

c) Temperature effects on actuator and sensor properties including the dielectric

permittivity and the piezoelectric coefficient [60,61] (especially 𝑑33) [62] of both

actuator and sensor cause changes in signal amplitude due to temperature variations.

The ratio of output signal amplitude to input signal amplitude of a piezoelectric sensor

is proportional to the piezoelectric coefficient and dielectric permittivity of that sensor.

d) Temperature effects on transducer/plate bond layer properties including; the shear

modulus and thickness [63,64]. The shear modulus changes of adhesive layer due to

temperature variation is the most influential parameter for the change in sensor signals

as compared to other mechanical properties [65]. Variation of the adhesive shear

modulus changes the transducer-plate bonding shear stress transmission and effective

length which leads to changes in Lamb wave signal [66]. For large temperature

variations, changes in transducer and bonding properties may become more


significant. However, this variability is one that can potentially be minimized via the

suitable choice of transducer types, materials and manufacturer [67].

In addition to experimental studies, the effects of these mechanisms on Lamb wave

propagation due to temperature variations have been studied by some researchers using

numerical simulation models. Roy et al. [68] have done parametric studies using numerical

wave propagation simulations in order to study the effects of changes in plate material

properties on Lamb wave propagation due to temperature variations. They have suggested the

existence of a linear relationship between change in sensor signals and specific combination

of material properties within a certain temperature range. Kijanka and his co-workers, have

modelled the effects of changes in plate material properties on Lamb wave propagation due to

temperature variations using local interaction simulation approach (LISA) [69,70]. In another

study [71], they have taken into account temperature-dependent physical properties of low-

profile PZT transducers and transducer bonding layers in their Finite Element Model. Their

study has shown relevant changes in Lamb wave amplitude response caused by temperature

fluctuations. Kijanka and his co-workers [72] have also presented a three-dimensional

temperature-dependent model. They demonstrated that temperature significantly affects

Lamb wave propagation through physical properties of all elements involved, i.e. PZT

transducers, host structure and bond layers. Hence, any simplifications or omissions—related

to these elements—applied to analytical or numerical models will inevitably lead to

discrepancies in simulated Lamb wave responses. Therefore, Marzani and Salamone [57]

proposed a numerical approach based on a Semi-Analytical Finite Element (SAFE) model

including the combined role of transducer elements (actuator and sensor), substrate structure,

and transducer/structure interaction, in the prediction of the Lamb wave response under

changing temperature.
Apart from numerical simulation models, analytical models were also developed to

physically understand the effects of changes in above-mentioned mechanisms on Lamb wave

propagation due to temperature variations. Croxford and his co-workers [73] have considered

the effects of changes in material properties of the structure on Lamb wave propagation due

to temperature variations and they have concluded that (1) the change in velocity with

temperature is the dominant effect rather than the thermal expansion of the structure, (2) the

time shift associated with a change in temperature is proportional to propagation distance,

hence the performance of reference signal subtraction decreases as propagation distance

increases, (3) faster modes are less affected by temperature than slower modes. Andrews et al.

[74] developed an analytical model which takes into account the phase shift of Lamb wave

signals due to temperature variations. However, their model does not account for amplitude

attenuation due to propagation and temperature variations. A simple model that describes the

dependence of elastic properties of the CFRP plates on temperature is presented by Putkis et

al. [75] Their model predicts guided wave velocity response to temperature variations.

Dodson and Inman [76] extended the theory of acoustoelasticity by allowing thermally

induced strain in unconstrained isotropic media, to develop theoretical formulations to

capture the change of Lamb wave velocity with temperature. They have shown the theoretical

thermally induced acoustoelastic Lamb wave thermal sensitivity is an upper bound

approximation of the experimental thermal changes, but the acoustoelastic Lamb wave theory

is not valid for predicting the antisymmetric (A0) phase velocity at low frequency-thickness

values, <1.55MHz mm for various temperatures. This discrepancy indicates that velocity

changes due to temperature predicted by the acoustoelastic theory in this frequency range

cannot be predicted by only a first order derivative. They have confirmed this observation by

comparing theoretical predictions with experimental data obtained from an aluminum plate

with simulated free-free boundaries over a large temperature range (20°C to 70°C ).
Raghavan and Cesnik [77] proposed a theoretical model which can capture the three-

dimensional Lamb waves excited and sensed by arbitrarily shaped, surface-bonded finite-

dimensional piezos in isotropic plates. The inputs of their model are the material properties

and dimensions of the structure and the transducer which are thermally sensitive parameters

as explained above. Their model was able to accurately determine phase-shift in the sensor

signals at different temperatures, but mismatch in signal amplitude with experimental data

which was attributed to the absence of temperature-dependent properties of the adhesive

material. Di Scalea and Salamone. [58] presented an analytical model which accounts for the

cumulative effects of changes in transducer properties including the dielectric permittivity

and the piezoelectric coefficient, the transducer-to-plate adhesive bond properties, and the

material properties of the plate, on Lamb wave propagation due to temperature variations.

Salamone et al. [78] Studied the wave propagation problem analytically by a model that

accounts for temperature effects on the piezo-mechanical properties of Macro-Fiber

Composite (MFC) piezocomposite transducer, the interaction between the transducer and the

fiber reinforced composite panel, and the panel wave dispersion properties. Theoretical and

experimental results obtained on a [0/ ± 45/0]𝑆 Carbon-Fiber Reinforced Plastic (CFRP)

laminate between ―40°C and 60°C show substantial changes in the predominant s0-mode

amplitude (as high as 40% change from the ambient-temperature value). Sikdar et al. [41]

have carried out a combined theoretical analysis, time-domain spectral element simulation

and experimental analysis of elastic wave propagation in a carbon-fibre reinforced

adhesively-bonded composite structure to capture the behaviour of Lamb modes in the

structure due to variations in ambient temperature. They have shown that their theoretical

analysis and spectral formulation are effectively able to define changes in Lamb modes due to

changes in temperature.
The Lamb wave velocity dispersion curves are dependent upon the elastic moduli and plate

geometry, both which vary with temperature. Hence, the change in dispersion curves with

temperature has been studied in some studies. Gandhi and Michaels [79], used changes in

wave speed and plate thickness presented in reference [67], to propose two methods of

approximating the solutions of the Rayleigh-Lamb equations under a small perturbation of

parameters such as the small thickness and wave speed changes caused by temperature

variations. Dan et al. [80] developed new Rayleigh-Lamb equations with temperature-

dependent variables. To do it, they introduced the Equivalent Wave Method (EWM). EWM

allows the prediction of the Lamb wave behaviour with temperature without prior knowledge

on the temperature dependencies of the material properties. This is because the only

parameter required to implement EWM is the frequency-sensitivity parameter, which can be

experimentally determined within a prior calibration using only one pitch-catch sensor-

actuator pair. In another study, the thermal sensitivity of dispersion curves is analytically

developed by Dodson and Inman [81] which shows how temperature affects Lamb wave

speeds in different frequency ranges.

2.1.2. Temperature effect compensation methods for Lamb wave based SHM;

As mentioned above, temperature changes the shape of the signal. Therefore, as demonstrated

in Fig.1, the subtraction of baseline from current signal will result in false scatters in the

scatter signal. If this false scatter is then used in Lamb wave based SHM techniques, it will

lead to false damage detection scenarios. Therefore, it is important to compensate the effect

of temperature on collected signals. Fendzi et al. [82] have compared the results of a Lamb

wave based damage identification method, with and without using temperature compensation

method (see Figure 3). As can be seen in Fig. 3(a), without using compensation method,

damage imaging results in false damage locations. However, as demonstrated in Fig. 3(b),

after using compensation method, the actual location of damage is identified precisely.
Fig. 3. Damage imaging of the composite laminate, baseline signal collected at 25°C, and damaged signal

collected at 40°C. black circle represents true damage location – numerical study. (a) Without temperature

compensation. (b) With temperature compensation [82].

Hence, utilizing temperature compensation methods can enhance the efficiency of Lamb

wave based damage identification and characterization methods.

Various strategies for compensating the effect of temperature on Lamb wave based SHM

techniques have been proposed in recent years. These compensation methods can be mainly

divided into two categories: data-driven methods and model-driven methods.

In data-driven methods, temperature compensation is achieved by comparing new

observations with collections of previously obtained data. A data-driven methodology, often

referred to optimal baseline subtraction (OBS), is introduced in references [83,84]. The

optimal baseline subtraction method attempts to minimise residual signal amplitude levels by

using M baseline data sets, collected at different temperatures, 𝑇𝑚 = 𝑇0 +𝛿𝑇𝑚 referred to as

the baseline dataset. Where, 𝑇0 is the initial temperature and 𝛿𝑇𝑚 represents the change in

temperature.

The mth signal from the baseline dataset can be expressed as:

𝑁
𝑗 𝑆𝑗 [𝑡 ― 𝑡𝑗 𝛽(𝛿𝑇𝑚)]
𝑢𝑚(𝑡;𝑇𝑚) = ∑𝑗 = 1𝐴𝑚 𝑚 𝑚
(1)
Where, 𝑡 is time, 𝑁 is the number of superimposed wavepackets, 𝐴𝑚 𝑚 𝑚
𝑗 , 𝑆𝑗 , 𝑡𝑗 are respectively

the amplitude, waveform and arriving time of the jth wavepacket at temperature 𝑇𝑚, and 𝛽(𝛿

𝑇𝑚) is the fractional shift in arrival times of wavepackets in each signal with respect to their

values at an arbitrary fixed temperature.

OBS is based on the identification of a baseline within the database which is most similar to a

current signal taken from the structure during inspection. To achieve this, data-driven

approaches have been established based on data mining techniques such as mean square

deviation [72], maximum residual amplitude [85] Principal Component Analysis (PCA) [86–

89], Singular Value Decomposition (SVD) [90,91] and Independent Component Analysis

(ICA) [92–94].

Principal component analysis (PCA) is a simple, nonparametric method for data compression

and information extraction that finds combinations of variables or factors that describe major

trends in a confusing data set. Torres-Arredondo et al. [89], have presented a pattern

recognition strategy using PCA which allows for automatic baseline selection from a set of

baseline measurements containing different temperature conditions under which the structure

is supposed to be in operation. In this pattern recognition, baseline signals recorded at

different temperature conditions, are clustered in different groups and then used as optimal

baseline models for a multi-variate statistical model based on PCA. This step guarantees not

only a reduction of the total size of the model to be employed, since not all the recorded data

are required for building the models, but also provides a reduction in the variability inside the

models so that a better description of the process which is desired to be described can be

achieved. They have shown the validity of their method by performing experiments on a

simplified aircraft composite skin panel made of carbon fibre reinforced polymer (CFRP).
Singular Value Decomposition (SVD) is a linear decomposition method that is widely used

for dimensionality reduction, and is closely related to PCA. Liu et al. [91] have demonstrated

that by applying SVD on ultrasonic records, it is possible to separate the change produced by

damage from the change caused by temperature, without a prior knowledge of the

temperature variations, and thereby robustly detect damage in a complex environment. In this

method, the orthogonality of singular vectors ensures that the effect of damage and that of

temperature variations are separated into different singular vectors. Damage-sensitive

features are then selected from the singular vector with damage effect. They have shown their

method’s efficacy on data collected in real world piping systems experiencing significant

variations in temperature.

Independent Component Analysis is a statistical technique for revealing the hidden trends and

groupings that underlie a set of data. The technique takes a set of multidimensional data and

transforms it into components that are as statistically independent as possible. ICA is similar

to SVD in that it seeks to extract meaningful trends in data. The difference between these

techniques is that SVD uses a decorrelation of the data to separate the information, while ICA

minimizes the mutual information between groupings. The intention of using ICA on guided

wave data is that after applying the transform, one of the independent components will

contain data relating to the defect, while temperature effect will be rejected to other

independent components; so giving a clearer representation of the defect. Moll et al. [94]

have shown that ICA can compensate the effect of temperature from data collected by PZT

transducers attached on a carbon fibre reinforced plastic (CFRP) plate.

Recently, a data-driven temperature baseline reconstruction approach that is applicable for

various structures made from the same material is presented by Yue and Aliabadi [95]. In this

method, firstly the influence of temperature on the amplitude and phase of collected signals

are experimentally quantified as dimensionless compensation factors. Then, the derived


compensation factors are used to reconstruct baselines at various temperatures. They have

shown that with a single baseline measurement at 20° C and the reconstructed baseline using

the predetermined temperature compensation factors, impact damage can be identified and

localized in a simple flat plate and a stiffened panel, when current signals were up to 25° C

and 20° C higher than the baseline temperature, respectively.

Another data-driven method for a non-uniform structural temperature field is proposed by

Sun et al. [96].

Model-driven techniques are based on an approximate model of the effects of temperature on

ultrasonic waves. A model-driven methodology, often referred to baseline signal stretch (BSS)

is developed. In this technique, only one baseline would be needed, since current signals

would be stretched or compressed until they matched this universal baseline. Baseline signal

stretch method can be presented based on time domain stretching [97] of either the reference

signal or the current signal, estimation of delay as a function of time through local coherence

and use of these values as factors for subsequent time-domain stretch [98,99], or simple

frequency domain stretch [100,101]. Salmanpour et al. [102] have presented an improved

BSS method which is effective for temperature difference of up to 13° C.

Stretching of the signal in time or frequency causes dilations of wave-packets as well as

arrival times; this means that these techniques do not compensate perfectly for temperature

effects since the frequency content of the signal is altered. Therefore, a correction was made

in the frequency domain by stretching the frequency axis. Harley et al. [103] developed three

model-driven optimal temperature compensation methods: the scale-invariant correlation

method, the iterative scale transform method, and a combination of the two. By using these

tools, they improved computational speed of baseline signal stretch method. Liu et al. [104]

have shown that the instantaneous phase difference of the two Lamb wave signals with
different temperature is proportional to the temperature difference, with the limit range of

temperature. Their proposed method could effectively work for temperature intervals of at

least 18°C with the baseline signal temperature as the centre. Recently, an improved BSS

method is proposed by Mariani et al. [105] which seeks to compensate for both wave

propagation speed change and transducer phase response change.

When the optimal baseline subtraction method is used alone as a temperature compensation

technique, a large number of baselines are needed since only small temperature steps

(typically 0.1°C) will ensure amplitude levels in the residual signal which are low enough for

good sensitivity (around -40 dB relative to the first arrival) [82]. On the other hand, the

success of the baseline signal stretch method is strongly dependent on mode purity and

structural complexity. Furthermore, there are two other aspects that make the applicability of

BSS method onto real complex structures a challenging endeavour. Firstly, the time traces

stretch with propagation distance and secondly, the change in wave velocity with temperature

depends on the frequency of the wave [106].

To tackle these problems, a combination of these two methods can be used as the most

effective strategy. The closest matching baseline is first selected from the set and is then

further improved by BSS method [107–111]. This reduces the number of baselines necessary

to ensure good sensitivity in comparison to that needed when the OBS technique is used

alone. The reduction in the number of baselines in the database is limited by the maximum

temperature gap between baselines which can be compensated for by the BSS method

without loss of sensitivity; this is a function of mode purity, signal complexity and the

maximum propagation distance to cover the whole structure expressed in wavelengths. Lu et

al. [112] presented the possible improvements to increase the efficiency of combined OBS

and BSS method by optimizing each step of the original method.


However, there are a few issues arising from the current approach of the combined OBS and

BSS method. First of all, most structures cannot be practically placed into an environmental

chamber for the collection of a baseline set. Therefore, long time might have to be dedicated

for collection of a complete set of baselines. Secondly, the assumption that the structure is in

a healthy condition during this time has to be made. Finally, the structure might undergo

changes during its life-cycle that would affect guided wave signals but are not associated with

damage, Hence, the assumption that a complete baseline set could be pre-recorded becomes

weak.

To overcome this obstacle, new approaches have been developed by researchers. Putkis and

Croxford [113] proposed an approach for the acquisition of a baseline set, which makes it an

evolutionary process and lifts the restrictions on set completeness. A method based on an

interpretation of multiple signals acquired in distinct points of the structure was introduced by

Dworakowski et al. [114]. They used data fusion to merge this method with a number of

methods into one with a substantially increased efficiency. Roy et al. [68] developed a

physics-based temperature compensation model that would need limited set of sensor

measurements for model training. A method for reconstruction of the baseline time-trace

corresponding to the current signal is presented by Aryan et al. [115]. In this method the

baseline signal can be reconstructed based on the measurements of the 3D

velocity/displacement points near the PZT and prescribing these fields to the 3D transient FE

model. A methodology including the use of multivariate analysis, sensor data fusion and

machine learning approaches is proposed by Vitola et al. [116]. In another study, Anaya et al.

[117] used artificial immune systems (AIS) and sensor data fusion to develop a temperature

compensation method. Other methods are based on dynamic time warping [118], Gaussian

mixture model [119,120], dictionary learning method [121], and combining an adaptive filter

and OBS method [122]. Recently, an iterative compensation method is suggested by


Herdovics and Cegla [123], which estimates the phase and propagation speed change solely

from the ultrasonic time signal.

Dworakowski et al. [124], have verified the efficiency of four methods designed for

temperature compensation. The first method is optimal baseline selection approach. The

second method is damage index based on a signal alignment with respect to instantaneous

phase. The third one is a group measurement approach capable of distinguishing local

damage-related changes from temperature-induced global ones. The basis of group

measurement approach is that a tiny damage at the beginning of its formation only affect

structure locally. Therefore, in a sparsely located net of sensors which cover the whole

structure, only those signals which are collected from sensor pairs at the location of damage,

are influenced due to the presence of damage. In contrast, the temperature variations usually

affect the whole structure at the same time, changing collected signals of all sensor pairs.

Hence, if damage index is calculated for each sensor pair, it can be seen that the damage

indexes of sensor pairs located at the location of damage, are not only increased due to

temperature variations but also increased due to the presence of damage. While, the damage

indexes of other sensor pairs are only increase due to temperature variations. This difference

in the increase of damage index value can be used to distinguish damage from temperature

variations. The last method relies on fusion all these solutions simultaneously. By comparing

these methods, it is found that although all the methods have their pros and cons, a

cooperation of all solutions allows for significant increase of the damage detection efficiency.

The technique of cointegration—from the field of econometrics [125]—has been proposed as

a method for compensating the temperature effects on SHM, in references [126–131].

Cointegration is in fact a property of multivariate nonstationary time series, which if present,

implies that some linear combination of the variables in question would be stationary.

However, some constraints fall on the nonstationary variables if they are to be cointegrated.
First of all, they must share common trends and they must also be integrated to the same

order. Therefore, the first step in any cointegration analysis is to determine the order of

integration of each variable in question. This can be achieved with the Augmented Dickey

Fuller (ADF) test, which determines whether a variable will exhibit stationary or

nonstationary behaviour by examining its characteristic roots when fitted to an error-

correction model. In ADF test, choosing an appropriate lag length is a matter of importance

because the effectiveness of this method depends on the lag length. Upon ascertaining the

order of integration of each of the variables under consideration, any that are of the same

order are eligible for cointegration analysis. One of the most common approaches for

cointegration analysis is the Johansen’s cointegration procedure, which is a maximum

likelihood based method for combining nonstationary variables whose first difference is

stationary.

The idea of using cointegration for dealing with the problem of temperature variability in

SHM data is that if a number of variables from a process under investigation are cointegrated,

the stationary linear combinations of them found during the cointegration process will be

purged of all common trends in the original datasets, leaving residuals equivalent to the long-

run dynamic equilibriums of the process. In this case the common trends removed by the

cointegration process could be the environmental and operational conditions such as

temperature that drive the response of the structure.

Dao and Staszewski [130] used cointegration approach for the compensation of temperature

effects, on Lamb wave based damage detection. In their study, instead of directly using Lamb

wave responses for damage detection, they have proposed two approaches: (i) analysis of

cointegrating residuals obtained from the cointegration process of Lamb wave responses, (ii)

analysis of stationary characteristics of Lamb wave responses before and after cointegration.

Moreover, they have used the ADF test not only for the purpose of testing the order of
integration of each variable, but also to create a damage detection indicator. Their

experimental results obtained from undamaged and damaged aluminium plate showed that

the cointegration process can successfully remove the undesired temperature effect from

Lamb wave data and thus the cointegrating residuals obtained are free from temperature

variations and they can be used to detect damages with different severities in plate.

In another study [131], the cointegration approach in Lamb wave based damage detection is

investigated with two temperature variation cases: (1) single-temperature trend – data

corrupted by a single value of temperature and (2) multiple-temperature trends – data

corrupted by different values of temperature. They have shown that by using the

cointegration process, undesired temperature effects can be successfully removed from Lamb

wave data in both temperature effect cases investigated. Hence, the cointegrating residuals

obtained were free from temperature variations. In addition, they have shown that despite of

the fact that the first cointegrating residual was sufficient to detect the largest severity of

damage investigated, the remaining cointegration residuals also provide valuable information

on reliable damage detection.

The advantages and disadvantages of some temperature compensation methods, is

summarized in Table 1.

Table 1. The advantages and disadvantages of some temperature compensation methods

Method Advantages Disadvantages


Doesn’t need to collect a large set of baseline signals.
Certain model limitations do exist for
physics-based model [68] Instead, they can be numerically generated at any specified
generating baseline signals.
temperature.
The efficiency of the process depends on
Baseline set acquisition
The monitoring phase starts without the prior period solely the value of time offset. However this
using an evolutionary
dedicated for baseline acquisition. value is unique for each SHM system
process [113]
and type of damage.
The proposed algorithm is robust not only to the
It needs to collect a large set of baseline
Data Fusion [114] temperature-related changes but also to outliers caused by
signals
external factors.
The use of multivariate
analysis, sensor data Gives the current state of the structure directly from the
The big quantity of data required
fusion and machine acquired data and the reduction of false positives
learning approaches [116]
Dynamic time warping
Its flexibility to better align data Computational cost
[118]
Adaptive filter joint with Requires a few baselines for a large temperature range The process needs to determine a
OBS [122] (temperature interval for baselines of low-frequency signals, compensation standard for different type
can be up to 20°C and for high-frequency signals, can be up of damages.
to 12°C).
Can successfully remove the undesired temperature effect The effectiveness of the method depends
from Lamb wave data and thus the cointegrating residuals on the selection of an appropriate lag
Cointegration [126–131] obtained are free from temperature variations that can be length in ADF test. The lag length value
used for damage detection in structure. It can detect damages would be different for different damage
with different severities. severity and also different damage types.

2.2. Moisture

This section reviews the effect of moisture on Lamb wave propagation in composite materials.

Composite materials are made of fibres and matrix. Sateesh et al [132] have shown that

moisture generally affects any property of composite materials which is dominated by the

matrix. For example, they have shown that the flexural modulus of glass fibre reinforced

plastic (GFRP) composites rapidly decreases by hydro aging. However, the flexural strength

of composite materials which is a fibre dominated property, reduces if the fibres themselves

are affected by moisture. For instance, they have shown that moisture can cause degradation

at fibre level in glass fibres. This is because water can extract ions from the fibre and in this

way alter its structure. In another study [133], Alamri and Low have investigated the effects

of moisture on cellulose fibre epoxy composites. They have shown that moisture absorption

makes cellulose fibres susceptible to swelling, resulting in formation of micro-cracks and

voids at the fibre-matrix interface region. This in turn reduces the dimensional stability and

mechanical properties of composites. However, cellulose fibres are not representative of the

fibres most commonly used in composites e.g. in aerospace applications. It should be also

noted that in polymer-matrix fibre composites, the effect of moisture on polymer matrix is

generally more pronounced than effects on the fibres.

Since, moisture absorption changes the dimensional and mechanical properties of composite

materials, it affects Lamb wave propagation in composite materials. Lee and Cho [36]

defined the A1 mode velocity of Lamb waves in carbon/epoxy laminates with different

moisture contents and demonstrated in Fig. 4.


Fig. 4. Wave propagation velocity in carbon/epoxy laminate with different moisture contents [36].

As can be seen in Fig. 4, the wave velocity is decreased at higher moisture contained

specimen. In another study, Schubert and Herrmann [134] investigated the influence of

moisture absorption on Lamb waves propagation in viscoelastic composite materials. They

have shown, moisture absorption leads to a massive drop of the amplitude of the sensor

response, which is related to both the changes of the material properties and that of the

adhesive layer, and a relatively small drop of Lamb wave velocity in composite materials.

Since, changes of moisture cause changes in wave velocity (TOF) and the amplitude of the

signal, subtraction of baseline from current signal collected at two different moisture content

will lead to false positives. If this scatter signal is used in Lamb wave based SHM techniques

will results in false damage detection process. Therefore, it is necessary to develop moisture

effect compensation methods for Lamb wave based SHM in composite structures. However,

the effects of moisture content change on Lamb wave propagation in metallic structures is

less important than composite structures because in metallic structures moisture only affects

the transducer and the adhesive layer. However, this variability is one that can potentially be

minimized via the choice of transducer types, materials and manufacturer.

3. Operational Condition
Even though the effects of operational conditions including vibration, applied loads and

bonding (adhesive layer shear modulus and thickness, bond defects) on Lamb wave

propagation may be unavoidable in real application, relatively little attention have been paid

to them. This section provides a summary of such factors on Lamb wave propagation.

3.1. Vibration

Structures are generally and inevitably subjected to vibrations from varied sources during

their service lives. Extra stress induced by vibrations could possibly lead to changes in Lamb

wave signals and hence the misinterpretation of the structural behavior. Lu et al. [37]

compared the Lamb wave signals measured in static condition and vibration at 10 Hz

frequency under free boundary condition (see Fig. 5).

Fig. 5. Comparison of sensor signals measured in static condition and vibration at 10 Hz frequency [37].

As can be seen in Fig. 5, they have concluded that the received time domain wave signals in

vibrating structures will be deviated. This is due to the external vibrations which generate

flexural waves in addition to the Lamb waves actuated by the PZT elements. Moreover, since

time of flight (TOF) is a crucial parameter in most Lamb wave based SHM studies, they have

investigated the influence of the external vibrations on the TOF. To achieve this, they have

used two specimens (Two aluminium plates with different dimensions and different boundary

conditions) in the experiment. The TOF extracted for both specimens are shown in Fig. 6.
Fig. 6. Extracted TOF for different vibration frequencies: specimen 1 (solid line); specimen 2 (dotted dash line)

[37].

It can be observed that for both specimens, the TOFs only fluctuate slightly as the vibration

frequency varies, which could be attributed to measuring errors. Therefore, TOFs were not

affected by the external vibrations at the frequencies studied. Since, the vibration frequencies

studied cover most of the dominant frequency ranges in practical operating conditions for

most structures, their study signify that SHM algorithms developed based on TOF of Lamb

waves are still effective even if the structures being monitored are subjected to vibrations.

Ramsey [135] evaluated the effects of wind-induced vibration on Lamb wave based SHM

systems. To achieve this, he placed the test specimen in a wind tunnel and at different wind

speeds and collected three signals. He has shown that the shape of the pulses from the

receiving transducer remained intact however the signals will be deviated as is also shown by

Lu et al [37]. He has also demonstrated that although the amplitude of the signals at high

frequencies remained intact, but at low frequencies the amplitude increased with wind speed

and could become large enough to saturate a typical data acquisition system.

The influence of random vibration on Lamb wave signals was evaluated by Salmanpour et al.

[136]. They have shown that the random low frequency noise can have higher amplitude than
the diagnostic signals which results in the signal to be clipped or even saturated if the

recording voltage range is close to recorded signal amplitude. To tackle this problem, they

have designed a hardware high-pass filter.

Due to the results of these studies, vibration of the structure doesn’t change the phase (TOF)

of the signals and also the amplitude of the signals at high frequencies remained intact. Hence,

the shape of the signals remains intact and only signals will be deviated. On the other hand,

the dominant frequency ranges in practical operating conditions are much lower than the

frequency of ultrasonic signals. Amplitude normalization of the baseline and current signal,

will eliminate the vibration induced deviation. Then, since vibration doesn’t change the phase

and amplitude of the current signal, subtraction of baseline from current signal will result in

no false positives. In this way, it is possible to compensate the effect of vibration on Lamb

wave based SHM systems.

3.2. Load

Structures are designed to carry loads and they will experience different loading conditions

during their service. Therefore, in order to enhance the efficiency of Lamb wave based SHM

techniques, it is important to evaluate the effects of applied loads on Lamb wave propagation.

The effects of axial load on the properties of guided wave are investigated by Chen and

Wilcox [137]. They have shown that at low frequency-thickness values the change in phase

velocity with applied strain is non-linear. However, above a frequency-thickness of around 1

Hz m, the relationship between phase velocity change and applied strain becomes

approximately linear. They have also shown that changes of group velocity are depending on

the frequency-thickness and the level of strain. At frequency-thickness values above around 5

Hz m, the group velocity sensitivity to small strains is equal and opposite to that of phase
velocity; in fact, an applied strain will result in an increase in phase velocity and an equal

decrease in group velocity.

Lee et al. [38] have demonstrated the first arrivals of S1 mode for two paths at different

directions (vertical and horizontal) under varying loads (see Fig. 7). It is clear in Fig. 7 that

the primary effect of a change in applied load on the collected Lamb wave signals is a time

shift, or change in phase. However, the magnitude of the time shift for two paths are different.

In fact, applied loads changes, affect propagation of Lamb waves in homogeneous and

isotropic materials by changing the dimensions and the wave speed of the target structure.

Dimensions are changed as per the strain tensor and Lamb waves speeds change because of

the acoustoelastic effect; both of these changes are generally anisotropic [138].

Therefore, applied loads, change an initially isotropic medium into a slightly anisotropic one,

and the so-called effective elastic constants do not have the same symmetry as the usual

second order elastic constants. Hence, the magnitude of the time shift due to varying applied

loads, depends on the propagation angle.

Fig. 7. Zoomed views of the first arrivals of the S1 mode at 600 kHz under varying loads. (a) Two transducers

oriented in the vertical direction (axis of uniaxial loading). (b) Two transducers oriented in the horizontal

direction [38].
The same results have been demonstrated by Michaels et al. [139] In their full scale fatigue

test of an aircraft wing, they have shown that applied loads cause a time shift in Lamb wave

signals and the magnitude of the time shift are a function of the angle of the line connecting

the source and the receiver and their separation distance. Gandhi et al. [140] have developed

the theory of acoustoelastic Lamb wave propagation for isotropic media subjected to a biaxial,

homogeneous stress field. It is shown that dispersion curves change anisotropically for most

stresses, modes, and frequencies. Moreover, for some mode-frequency combinations, changes

in phase velocity are isotropic even for a biaxial stress field.

In another study, Michaels et al. [141] have investigated changes in signals caused by

uniaxial loading. They have plotted the scatter signal as a function of load for an intact plate

(see Fig. 8).

It can be seen that the amplitudes of the scatter signals increase monotonically as the load

increases and that their shape does not change significantly. The effect of applied or

thermally-induced stresses on the propagation characteristics of the Lamb waves is

investigated by Mohabuth et al. [142]. They have shown that the effect of noise generation

due to variations in applied or thermally induced stresses is comparable with that of moderate

temperature fluctuations. Therefore, load effect need to be considered when developing SHM

systems for real-world applications.

Roy et al. [45], have developed a load compensation model for ultrasonic guided waves based

SHM accounting for the changes in both the phase-shifts and signal amplitude.
Fig. 8. Scatter signals as compared to the signal at zero load (intact structure) [141].

3.3. Bonding Effects

Typically, the piezoelectric transducers used for Lamb wave based SHM systems are

permanently bonded to the host structure with adhesive. The adhesive forms an interfacial

layer of finite thickness between the piezoelectric transducer and host structure. The adhesive

interface provides the necessary mechanical coupling needed to transfer the forces and strains

between the piezoelectric transducer and the host structure. Therefore, a fundamental

understanding of the effects of bonding condition including adhesive layer shear modulus and

thickness and also bond defects, on Lamb wave propagation is one of the important issues in

real-world application of Lamb wave based SHM systems.

3.3.1. Adhesive Layer Shear Modulus and Thickness

The adhesive layer significantly affects a sensor signal. Hence, several studies have been

conducted on the effects of adhesive layers between surface-mounted PZTs and host

structures for very low excitation frequencies up to several kHz, which correspond to much

larger wavelengths than the size of the PZT [143–146]. In this frequency range, a shear lag
effect has been reported. According to Crawley’s work [143], the shear lag effect becomes

more dominant with a lower shear modulus and thicker adhesive layer, and shear transfer

between the PZT and host structure becomes less effective. As a result, the signal amplitude

is reduced as the shear modulus decreases and the thickness of the adhesive layer increases.

However, the aforementioned shear lag effect cannot explain what was observed

experimentally by Qing et al [39]. They have collected signals experimentally from two

sensors with different adhesive thicknesses (10 and 40 µm) at 50 kHz and 500 kHz excitation

frequency (see Fig. 9).

As can be seen in Fig. 9, at lower frequency, around 50 kHz, the signal for actuator–sensor

PZTs with thin adhesive layer is stronger than for those with thicker adhesive layer but at

higher frequency, around 500 kHz, the amplitude of the signal for actuator–sensor PZTs with

thin adhesive layer is smaller than for those with thicker adhesive layer. It could be associated

with the fact that the resonant phenomenon of the PZT is less constrained with a thicker and

softer adhesive layer so more energy might be generated from the PZT excitation.

Fig. 9. Sensor signal with 10 and 40 µm adhesive layer; (a) at 50 kHz, (b) at 500 kHz (experiment from [39]).

This resonant phenomenon is very useful in applications because a better signal-to-noise ratio

in sensor signals can be achieved with the same energy input. Therefore, there have been

some studies on adhesive layer effects around the resonant frequency [39,147]. Ha and Chang
[148] implemented a hybrid spectral element method (SEM) and have demonstrated that the

resonant affects signal amplitudes in a manner opposite to shear lag, which is a dominant

effect in frequencies far from the resonance.

In addition to experimental studies, analytical and numerical simulation models have been

developed to study the effects of the adhesive layer on Lamb wave propagation. Crawley and

De Luis [143] presented a one-dimensional static elastic model of the adhesive layer that

couples the displacement of the actuator to the host structure through shear deformation.

Their model requires a double-actuator configuration to excite either symmetric or

antisymmetric mode. However, this limitation was lifted by Giurgiutiu [149]. He extended

the model to the case of a single actuator bonded to one side of the beam and considered the

total effect as a superposition of symmetric and antisymmetric contributions. Giurgiutiu and

Santoni-Bottai [150] and Yu et al. [151] generalized Crawley model to nonlinear guided

wave modes. Crawley’s work was also extended by Dugnani [152]. He discussed some of the

limitations of the Crawley model and offers a simple, yet more generic solution to the

problem by considering the inertia term in the piezoelectric transducer.

Santoni-Bottai and Giurgiutiu [153] developed an iterative solution approach for the multiple

generic wave modes. In this model, the shear-lag effect of the adhesive layer was represented

by an aggregate number that expresses the participation of the various wave modes. This

model has been later used by Kamal et al. [154] to evaluate the power transfer for multimode

Lamb waves. An analytical solution for the two-dimensional shear lag problem has been

proposed by Giurgiutiu [155]. Another two dimensional shear lag model is developed by

Kapuria and Agrahari [156]. They presented an iterative analytical solution of the governing

equations using the recently developed mixed-field multiterm extended Kantorovich method

(MMEKM).
Willberg et al. [157] optimized the Lamb wave generation by performing parametric studies

on the actuator length as well as the shear modulus and thickness of the adhesive layer using

finite element simulation. They demonstrated that the amplitude of the Lamb wave signal is

influenced by the excitation frequency as well as the adhesive layer thickness.

The analytical formulation of the longitudinal and flexural models defined by Islam and

Huang [158], indicated that the effect of the adhesive layer is governed by two parameters, i.e.

the shear transfer parameter and the thickness-shear modulus ratio. Their Parametric studies

based on the simulation model and experiments suggested that there exists an adhesive

thickness at which the longitudinal and/or flexural mode Lamb wave signals can be

maximized.

3.3.2. Bonding Defects

Bonding defects (between a PZT patch and a host structure) may occur over time during the

in-service life of the structure compromising the performance and reliability of the Lamb

wave based SHM system. Hence, some researchers investigated the effects of bonding

defects on Lamb wave propagations into the host structure. Park et al. [40] compared the

Lamb wave propagation in paths with equal distance through the same material but from

transducers with healthy and defective bonding as demonstrated in Fig. 10.


Fig. 10. Lamb wave propagation from transducers with healthy and defective bonding (transducers 2,3,4,5,7,8,

and 9 have a complete bonding but transducers 1 and 6 have a defective bonding) [40].

They have found that the effects of bonding defects on Lamb wave propagation are

remarkable, modifying the phase, amplitude, and shape of propagated Lamb waves. The same

results have defined by Lanzara et al [159]. According to their study, the electro-mechanical

coupling between PZT actuators and a host structure vary due to bond defects over a wide

frequency range causing a signal delay and an amplitude decrease for: increasing debonding

area, different debond shape, and location underneath the PZT actuators.

As mentioned above, transducer bond defects affect Lamb wave propagation and these

effects are more significant as the size of the actuator increases [160]. Hence, many

researches have studied the measured variables such as admittance measurement [161,162],

impedance measurement [163], reciprocity between time response of two PZT sensors and

time reversal of the signals using Lamb waves [164], electrical properties of PZT sensors

[165,166], frequency shift in the maximum amplitude spectra [167] and maximum amplitude

[159] to diagnose bond defects between transducer and the host structure.

Mulligan et al. [168] used Finite Element Method to evaluate the effects of varying the

bonding layer coverage area and Young’s modulus on (a) admittance and (b) amplitude and

phase of the ultrasonic signals to develop a compensation method for bonding layer

degradation effects on Lamb wave signals. Gao et al. [169] presented a bonding defects

compensation method based on frequency response characteristics of the surface strain. A

solution to the bonding issue of sensors (to have consistent and repeatable bonding with

possibility of removal and repair) is proposed by Bekas et al. [170].

4. Baseline Free Techniques


Generally, the Lamb wave based SHM techniques use damage scatter signals for damage

identification and quantification. Usually damage scatter signals are determined by

comparing the current signal with a baseline signal, which corresponds to the intact state of

the structure. However, there are certain drawbacks in using baseline signal. Firstly, baseline

signal might not be available. Secondly, as mentioned above, many factors can change the

current signal even in absence of any damage and it may be construed as damage resulting in

false alarm. To overcome these limitations, researchers have proposed baseline free damage

detection techniques. Since the main attention of this paper is to review the EOCs effects on

lamb wave propagation, baseline free techniques are summarized here very briefly.

Time reversibility of Lamb wave is widely used for baseline free damage detection

techniques [171–175]. Nevertheless, there are possible concerns with an incorrect

interpretation of the physics of time reversal and also incorrect approach to assessing time

reversibility in several published papers. Hence, it is recommended to refer to reference [176]

for a critical discussion of these issues.

Sohn and co-researchers have comprehensively studied the application of time reversed

Lamb wave based damage detection technique theoretically and experimentally including:

time reversed Lamb wave based damage index for delamination detection in composite

laminate [177], wavelet analysis for enhancement of time reversed Lamb wave [178], effect

of edge reflections [179], detection of debonding of PZT transducer [180].

Selection of parameters such as excitation frequency, modulating window for the tone burst

signal, and size of PZT transducers for better reconstruction of time reversed Lamb wave for

damage detection is presented for isotropic plates [181], woven fabric laminate [182],

stiffened aluminium plate [183]. Watkins and Jha [184] modified time reversal method where

the transducers act both as actuator and sensor which reduces the difficulty of switching the
transducers. Another modified time reversal method in which the frequency dependence of

the time reversal operator is compensated, is proposed by Zeng et al. [185]. An enhanced

Lamb wave virtual time reversal (VTR) algorithm with transducer transfer function

compensation to eliminate the transducer influence for dispersive, multimodal Lamb waves is

presented by Wang and Shen [186]. Some probabilistic diagnostic imaging technique using

time reversal approach were also presented [187,188]. Time reversibility in conjunction with

wavelet analysis of wave propagation in concrete rebar has been presented in reference [189].

Qu et al. [190] presents a transducer array based baseline free imaging technique using

decomposition of time reversal operator.

In addition to time reversal approach, some other baseline free damage detection techniques

have been proposed by researchers. Park et al. [191] presented the concept of transfer

impedance of PZT transducers for baseline free damage detection. This method uses crack

induced energy increase over a frequency range. In the method proposed by Anton et al. [192]

Lamb wave propagation along several paths within the structure is interrogated in pitch-catch

configuration and the common features in the undamaged paths act as instantaneous baseline

information for damage prediction. A method based on identifying the first arrival of A0

mode from a delamination in the path of wave propagation is proposed by Yeum et al [193].

A baseline free imaging method based on new PZT sensor arrangement is developed by

Qiang and Shenfang [194]. Bagheri et al. [195] presented a baseline free damage diagnostic

technique by using a round-robin sensor network. A Lamb wave based baseline free

technique which can be applied to specimens with complex structural geometries such as

holes, stiffeners, and varying thickness have been developed by Kim et al [196]. In another

study, they have presented a baseline free technique by investigating the electro-mechanical

(EM) signals from piezoelectric (PZT) wafers [197].


A baseline free method using a time-space analysis is presented by Li and Chattopadhyay

[198]. Alem et al. [199] developed a baseline free method based on cross-correlation (CC)

analysis. Hameed MS et al. [200] have presented a multistage damage detection method

which does not need any baseline signal, and it only uses the same arrangement of

transducers and the same data processing technique in all stages. An instantaneously recorded

baseline method is proposed by Salmanpour et al. [201]. This method eliminates need for

baselines required when operating at different temperatures by mapping a baseline area onto

the interrogation area. Alguri et al. [202] have proposed a dictionary learning framework

which combines wave propagation characteristics of a test structure with geometric

information from surrogate structures to create a baseline free damage detection method. In

another study, a baseline free method using a dual PZT network is developed by Lizé et al.

[203]. A baseline free damage detection method using a distance compensation algorithm for

pitch-catch pairs of different length, is proposed by Qiu et al. [204]. Another method based

on reciprocity principle is presented by Huang et al. [205].

Baseline free techniques only use current signal for damage identification and quantification.

However, EOCs factors affect features of current signals. Hence, some baseline free damage

identification techniques under varying temperature condition have also been developed

[206]. An and Sohn [207] proposed an instantaneous baseline free damage detection scheme

using Lamb waves in two independent environmental factors: temperature ranging from -

30°C to 50°C, and different static stresses on the structure.

5. Conclusions

The paper presents a review of the existing literature on environmental and operational

conditions effects on Lamb wave based structural health monitoring (SHM) systems.

Generally, in Lamb wave based SHM techniques, damage scatter signal is used for damage
identification and characterization. Damage scatter signal is usually determined by

subtracting a baseline from current signal. However, since Lamb wave signals can be easily

affected by EOCs parameters, simple subtraction of one baseline from the current is not

sufficient to discriminate between changes due to damage and those due to EOCs parameters

which is importance for a real-world SHM system. Therefore, this paper, provides a summary

of the works done on the effects of EOCs parameters including: temperature, moisture,

vibration, applied loads, and bonding (adhesive layer shear modulus and thickness, bond

defects), on Lamb wave propagation and the compensation strategies to account for these

effects. Baseline free techniques are also reviewed in this paper. A brief summary of the

effects of different EOCs factors on Lamb wave propagation is stated below:

(1) Temperature; According to many experimental, numerical and analytical studies on

the effects of temperature variations on Lamb wave propagation, it can be concluded

that the system response cannot be linearly related to temperature and generally, four

main mechanisms induce changes in Lamb wave signals as a result of temperature

variations, including: (a) Changes in the plate material stiffness due to temperature

variations; Since the moduli typically decreases with increasing temperature, the

longitudinal and transverse wave speeds decrease with increasing temperature. (b)

The plate in-plane expansion/contraction changes due to temperature variations

changes the distance between actuator and sensor causing a deviation in the TOF of

incoming and reflected modes. (c) Temperature effects on actuator and sensor

properties including the dielectric permittivity and the piezoelectric coefficient of both

actuator and sensor cause changes in signal amplitude due to temperature variations.

(d) Temperature effects on transducer/plate bond layer properties including the shear

modulus and thickness. Variation of the adhesive shear modulus changes the

transducer-plate bonding shear stress transmission and effective length which leads to
changes in Lamb wave signal. In order to compensate the effects of temperature

variations on Lamb wave based SHM techniques, various compensation methods are

developed.

(2) Moisture; In composite materials, moisture generally affects any property of

composite materials which is dominated by the matrix such as flexural modulus.

However, the flexural strength of composite materials which is a fibre dominated

property, reduces if the fibres themselves are affected by moisture. Generally,

moisture absorption leads to a massive drop of the amplitude, which is related to

changes in material properties of the composite material and the adhesive layer.

Moreover, it causes a relatively small drop of Lamb wave velocity in composite

materials.

(3) Vibration; Vibration of the structure doesn’t change the phase (or TOF) of the signals.

Also the amplitude of the signals at high frequencies remained intact. Hence, the

shape of the signals remains intact and only signals will be deviated. On the other

hand, the dominant frequency ranges in practical operating conditions are much lower

than the frequency of ultrasonic signals. The effect of vibration on Lamb wave based

SHM systems can be compensated by first eliminating the vibration induced deviation

using amplitude normalization of the baseline and current signal. Then, since

vibration doesn’t change the phase and amplitude of the current signal, subtraction of

baseline from current signal will result in no false positives.

(4) Load; The primary effect of a change in applied load on the collected Lamb wave

signals is a time shift, or change in phase. However, applied loads, change an initially

isotropic medium into a slightly anisotropic one, and the so-called effective elastic

constants do not have the same symmetry as the usual second order elastic constants.
Hence, the magnitude and direction of the time shift due to varying applied loads,

depends on the propagation angle.

(5) Adhesive layer shear modulus and thickness; The shear lag effect becomes more

dominant with a lower shear modulus and thicker adhesive layer, and shear transfer

between the PZT and host structure becomes less effective. As a result, the signal

amplitude is reduced as the shear modulus decreases and the thickness of the adhesive

layer increases. However, the resonant phenomenon of the PZT must be considered

since it is less constrained with a thicker and softer adhesive layer so more energy

might be generated from the PZT excitation.

(6) Bond defects; The electro-mechanical coupling between PZT actuators and a host

structure changes due to bond defects over a wide frequency range. This changes the

phase, amplitude, and shape of propagated Lamb waves.

The review presented here will help researchers to develop efficient Lamb wave based SHM

techniques for in-service monitoring of aerospace, automotive, civil and mechanical systems,

which are subjected to various EOCs parameters. The open areas of research which might

need attention are outlined as follows,

i. In the existing studies, the effects of only one of the EOCs factors on Lamb wave

propagation is investigated. However, in real practice, usually different EOCs

factors influence on Lamb wave propagation at the same time. Therefore,

evaluating the effects of simultaneous EOCs factors on Lamb wave propagation

needs research.

ii. Most of researchers have been focused on developing compensation strategies for

the effects of temperature on Lamb wave propagation. However, the effects of

other EOCs factors on Lamb wave propagation is unavoidable in real practice.

Therefore, compensation methods for these factors should also be developed.


iii. Even though the effects of moisture and applied loads on Lamb wave propagation

may be unavoidable in real application, relatively little attention have been paid to

them. Therefore, these factors need to be evaluated.

iv. Baseline free techniques only use current signal for damage identification and

quantification. In these methods, usually the signal amplitude or signal energy is

utilized for damage quantification. However, EOCs factors affect features (e.g.

amplitude) of current signals. Therefore, in order to increase the efficiency of

baseline free techniques, they can be joint with EOCs compensation strategies.

References

1. Dworakowski Z, Ambrozinski L, Packo P, et al. Application of artificial neural networks

for compounding multiple damage indices in Lamb-wave based damage detection. Struct.

Control Health Monit. 2015; 22: 50–61.

2. Ostachowicz W, Kudela P, Krawczuk M, et al. Guided Waves in Structures for SHM: The

Time—Domain Spectral Element Method. West Sussex, UK: John Wiley & Sons, Ltd, 2012.

3. Shen Y and Giurgiutiu V. WaveFormRevealer: An analytical framework and predictive

tool for the simulation of multi-modal guided wave propagation and interaction with damage.

Struct. Control Health Monit. 2014;13: 491–511.

4. Zhou W, Li H and Yuan FG. Guided wave generation, sensing and damage detection using

in-plane shear piezoelectric wafers. Smart Mater. Struct. 2014; 23.

5. Dai D and He Q. Structure damage localization with ultrasonic guided waves based on a

time-frequency method. Signal Process. 2014; 96: 21–28.


6. Pai PF, Deng H and Sudaresan MJ. Time-frequency characterization of Lamb waves for

material evaluation and damage inspection of plates. Mech. Syst. Signal Process. 2015; 62–63:

183–206.

7. Yu L, Leckey CAC and Tian Z. Study on crack scattering in aluminum plates with Lamb

wave frequency wavenumber analysis. Smart Mater. Struct. 2013; 22.

8. Quaegebeur N, Ostiguy PC and Masson P. Correlation-based imaging technique for fatigue

monitoring of riveted lap-joint structure. Smart Mater. Struct. 2014; 23.

9. Yu L, Cheng L and Su Z. Correlative sensor array and its applications to identification of

damage in plate-like structure. Struct. Control Health Monit. 2012; 19: 650–671.

10. Gorgin R, Wu Z, Gao D, et al. Damage size characterization algorithm for active

structural health monitoring using the A0 mode of Lamb wave. Smart Mater. Struct. 2014; 23.

11. Hall JS, Fromme P and Michaels JE. Guided wave damage characterization via minimum

variance imaging with a distributed array of ultrasonic sensors. J Nondest. Eval. 2014; 33:

299–308.

12. Keulen CJ, Yildiz M and Suleman A. Damage detection of composite plates by Lamb

wave ultrasonic tomography with a sparse hexagonal network using damage progression

trends. J Shock Vib. 2014; 2014.

13. Haynes C and Todd M. Enhanced damage localization for complex structures through

statistical modeling and sensor fusion. Mech. Syst. Signal Process. 2015; 54–55: 195–209.

14. Yelve NP, Mitra M and Mujumdar PM. Detection of stiffener disbonding in a stiffened

aluminium panel using nonlinear Lamb wave. J Appl. Acoust. 2015; 89: 267–272.
15. Shen Y and Giurgiutiu V. Predictive modeling of nonlinear wave propagation for

structural health monitoring with piezoelectric wafer active sensors. J. Intell. Mater. Syst.

Struct. 2014; 25: 506–520.

16. Hong M, Su Z, Wang Q, et al. Modeling nonlinearities of ultrasonic waves for fatigue

damage characterization: Theory, simulation, and experimental validation. Ultrasonics 2014;

54: 770–778.

17. Kudela P, Radzienski M, Ostachowicz W, et al. Structural health monitoring system

based on a concept of Lamb wave focusing by the piezoelectric array. Mech. Syst. Signal

Process. 2018; 108: 21–32.

18. Wang Q, Ma S, and Yue D. Identification of damage in composite structures using

Gaussian mixture model-processed Lamb waves. Smart Mater. Struct. 2018; 27: 045007.

19. Qing X, Li W, Wang Y, et al. Piezoelectric transducer-based structural health monitoring

for aircraft applications. Sensors 2019; 19: 545.

20. Abbas, M.; Shafiee, M. Structural health monitoring (SHM) and determination of surface

defects in large metallic structures using ultrasonic guided waves. Sensors, 2018; 18: 3958.

21. Cantero-Chinchilla S, Chiachío J, Chiachío M, et al. A robust Bayesian methodology for

damage localization in plate-like structures using ultrasonic guided-waves. Mech. Syst. Sig.

Process, 2019; 122: 192–205.

22. He J, Leckey CAC, Leser PE, et al. Multi-mode reverse time migration damage imaging

using ultrasonic guided waves. Ultrasonics, 2019; 94: 319–331.

23. Zheng Y, Liu K, Wu Z, et al. Lamb waves and electro-mechanical impedance based

damage detection using a mobile PZT transducer set. Ultrasonics, 2019; 92: 13–20.
24. Mori N, Biwa S and Kusaka T. Damage localization method for plates based on the time

reversal of the mode-converted Lamb waves. Ultrasonics, 2019; 91: 19–29.

25. He J, Rocha DC, Leser PE, et al. Least-squares reverse time migration (LSRTM) for

damage imaging using Lamb waves. Smart Mater. Struct. 2019; 28: 065010.

26. Zhou K, Zheng Y, Zhang J, et al. A reconstruction-based mode separation method of

Lamb wave for damage detection in plate structures. Smart Mater. Struct. 2019; 28: 035033.

27. Gao D, Wu Z, Yang L, et al. Integrated impedance and Lamb wave–based structural

health monitoring strategy for long-term cycle-loaded composite structure. Struct. Health

Monit. 2018; 17: 763–776.

28. Kamal AM, Lin B and Giurgiutiu V. Exact analytical modeling of power and energy for

multimode Lamb waves excited by piezoelectric wafer active sensors. J. Intell. Mater. Syst.

Struct. 2014; 107: 87–94.

29. Moll J, Golub MV, Glushkov E, et al. Non axisymmetric Lamb wave excitation by

piezoelectric wafer active sensors. Sensors Actuators A. 2012; 174: 173–180.

30. Manka M, Rosiek M, Martowicz A, et al. Lamb wave transducers made of piezoelectric

macro-fiber composite. Struct. Control Health Monit. 2013; 20: 1138–1158.

31. Wang W, Bao Y, Zhou W, et al. Sparse representation for Lamb-wave-based damage

detection using a dictionary algorithm. Ultrasonics. 2018; 87: 48–58.

32. Cawley P. Structural health monitoring: closing the gap between research and industrial

deployment. Struct. Health Monit. 2018; 17: 1225–44.

33. Muller A, Soutis C and Gresil M. Image reconstruction and characterisation of defects in

a carbon fibre/epoxy composite monitored with guided waves. Smart Mater. Struct. 2019; 28:

065001.
34. Worden K, Farrar CR, Manson G, et al. The fundamental axioms of structural health

monitoring. Proc. R. Soc. A 2007; 463: 1639–1664.

35. Radecki R, Staszewski WJ and Uhl T. Impact of changing temperature on Lamb wave

propagation for damage detection. Key Eng. Mater. 2014; 588: 140–148.

36. Lee J and Cho Y. Using Lamb waves to monitor moisture absorption in thermally

fatigued composite laminates. J Korean Soc. Nondest. Test. 2016; 36: 175–180.

37. Lu X, Soh CK and Avvari PV. Lamb wave propagation in vibrating structures for

effective health monitoring. In: Proceedings of SPIE 9438, Health Monitoring of Structural

and Biological Systems, 2015, 94381Z.

38. Lee SJ, Gandhi N, Michaels JE, et al. Comparison of the effects of applied loads and

temperature variations on guided wave propagation. AIP Conf. Proc. 2011, vol.30, pp.175–

182.

39. Qing X, Chan H, Beard S, et al. Effect of adhesive on the performance of piezoelectric

elements used to monitor structural health. Int. J. Adhes. Adhes. 2006; 26: 622–628.

40. Park G, Farrar CR, di Scalea FL, et al. Performance assessment and validation of

piezoelectric active- sensors in structural health monitoring. Smart Mater. Struct. 2006; 15:

1673–1683.

41. Sikdar S, Fiborek P, Kudela P, et al. Effects of debonding on Lamb wave propagation in a

bonded composite structure under variable temperature conditions. Smart Mater. Struct. 2019;

28: 015021

42. Wang P, Zhou W, Bao Y, et al. Ice monitoring of a full- scale wind turbine blade using

ultrasonic guided waves under varying temperature conditions. Struct Control Health Monit.

2018; 25: e2138


43. Mohabuth M, Kotousov A and Ng CT. Effect of uniaxial stress on the propagation of

higher-order Lamb wave modes. International Journal of Non-Linear Mechanics. 2016; 86:

104–111

44. Yang B, Xuan FZ, Xiang Y, et al. Lamb wave-based structural health monitoring on

composite bolted joints under tensile load. Materials. 2017; 10: 652.

45. Roy S, Ladpli P and Chang FK. Load monitoring and compensation strategies for guided-

waves based structural health monitoring using piezoelectric transducers. Journal of Sound

and Vibration, 2015; 351: 206–220.

46. Heinlein S, Cawley P and Vogt T. Validation of a procedure for the evaluation of the

performance of an installed structural health monitoring system. Structural Health

Monitoring, 2019; 18: 1557–1568

47. Salmanpour MS, Khodaei ZS and Aliabadi MHF. Impact damage localisation with

piezoelectric sensors under operational and environmental conditions. Sensors, 2017; 17:

1178.

48. Salmanpour MS, Khodaei ZS and Aliabadi MH. Airborne transducer integrity under

operational environment for structural health monitoring. Sensors 2016; 16: 2110.

49. Raghavan A and Cesnik CES. Review of guided wave structural health monitoring. Shock

Vib. Digest 2007; 39: 91–114.

50. Su Z, Ye L and Lu Y. Guided Lamb waves for identification of damage in composite

structures: A review. J. Sound Vib. 2006; 295: 753–780.

51. Mitra M and Gopalakrishnan S. Guided wave based structural health monitoring: A

review. Smart Mater. Struct. 2016; 25.


52. Qing X, Li W, Wang Y, et al. Piezoelectric transducer-based structural health monitoring

for aircraft applications. Sensors, 2019; 19: 545.

53. Ahmed S and Kopsaftopoulos F. Uncertainty quantification of guided waves propagation

for active sensing structural health monitoring. Presented at the Vertical Flight Society 75th

Annual Forum & Technology Display, Philadelphia, Pennsylvania, May 13–16, 2019.

54. Boon MJ, Zarouchas D, Martinez M, et al. Temperature and load effects on acoustic

emission signals for structural health monitoring applications. In: Proceedings of the 7th

European Workshop on Structural Health Monitoring, Nantes, France, 8–11 July 2014.

55. Blaise E and Chang FK. Built-in diagnostics for debonding in sandwich structures under

extreme temperatures. In: Proceedings of the 3rd international workshop on structural health

monitoring, Stanford, CA, 12–14 September 2001, pp. 154–163.

56. Lee BC, Manson B and Staszewski WJ. Environmental effects on Lamb wave responses

from piezoceramic sensors. Mod. Pract. Stress Vibr. Anal. 2003; 440: 195–202.

57. Marzania A and Salamone S. Numerical prediction and experimental verification of

temperature effect on plate waves generated and received by piezoceramic sensors. Mech.

Syst. Signal Process. 2012; 30: 204–217.

58. Di Scalea FL and Salamone S. Temperature effects in ultrasonic Lamb wave structural

health monitoring systems. J. Acoust. Soc. Am. 2008; 124: 161–174.

59. Di Scalea FL, Matt H and Bartoli I. The response of rectangular piezoelectric sensors to

Rayleigh and Lamb ultrasonic waves. J. Acoust. Soc. Am. 2007; 121: 175–187.

60. Lee HJ and Saravanos DA. The Effect of Temperature Dependent Material Nonlinearities

on the Response of Piezoelectric Composite Plates. Report, NASA/TM-97, 1997.


61. Hooker MW. Properties of PZT-based piezoelectric ceramics between _150 and 250° C.

Report, NASA/CR, 1998.

62. Li YH, Kim SJ, Salowitz N, Chang FK. Development of high-performance BS-PT based

piezoelectric transducers for high-temperature applications. In: Proceedings of the 7th

European Workshop on Structural Health Monitoring, Nantes, France, July 2014.

63. Chambers JT, Wardle BL and Kessler SS. Durability assessment of Lamb wave-based

structural health monitoring nodes. In: Proceedings of the 47th AIAA/ASME/ ASCE/AHS/ASC

structures, structural dynamics, and materials conference, Newport, RI, 1–4 May 2006, pp.

1–12.

64. Schulz MJ, Sundaresan MJ, Mcmichael J, et al. Piezoelectric materials at elevated

temperature. J. Intell. Mater. Syst. Struct. 2003; 14: 693–705.

65. Ha S, Lonkar K, Mittal A, et al. Adhesive Layer Effects on PZT-induced Lamb Waves at

Elevated Temperatures. Struct. Health Monit. 2010; 9: 247–256.

66. Sirohi J and Chopra I. Fundamental understanding of piezoelectric strain sensors. J. Intell.

Mater. Syst. Struct. 2000; 11: 246–257.

67. Lu Y and Michaels JE. A methodology for structural health monitoring with diffuse

ultrasonic waves in the presence of temperature variations. Ultrasonics 2005; 43: 717–731.

68. Roy S, Lonkar K, Janapati V, et al. A novel physics-based temperature compensation

model for structural health monitoring using ultrasonic guided waves. Struct. Health Monit.

2014; 13: 321–342.

69. Kijanka P, Radecki R, Packo P, et al. Local interaction simulation approach for

temperature effect modelling in lamb wave propagation. In: Proceedings of the 6th European

Workshop, Structural Health Monitoring, 2012, vol.1, pp. 856–862.


70. Kijanka P, Radecki R, Packo P, et al. GPU-based local interaction simulation approach

for simplified temperature effect modelling in Lamb wave propagation used for damage

detection. Smart Mater. Struct. 2013; 22.

71. Kijanka P, Packo P, W. Staszewski J, et al. Temperature effect modelling of piezoceramic

transducers used for Lamb wave propagation in damage detection applications. In:

Proceedings of SPIE 8695- Health Monitoring of Structural and Biological Systems, 2013,

86952C.

72. Kijanka P, Packo P, Zhu X, et al. Three-dimensional temperature effect modelling of

piezoceramic transducers used for Lamb wave based damage detection. Smart Mater. Struct.

2015; 24.

73. Croxford AJ, Wilcox PD, Drinkwater BW, et al. Strategies for guided-wave structural

health monitoring. In. Proceeding of the Royal Society A, 2007, vol.463, pp. 2961–2981.

74. Andrews JP, Palazotto AN, DeSimio MP, et al. Lamb waves propagation in varying

isothermal environment. Struct. Health Monit. 2008; 7: 265–270.

75. Putkis O, Dalton RP and Croxford AJ. The influence of temperature variations on

ultrasonic guided waves in anisotropic CFRP plates. Ultrasonics 2015; 60: 109–116.

76. Dodson JC and Inman DJ. Investigating the thermally induced acoustoelastic effect in

isotropic media with Lamb waves. J. Acoust. Soc. Am. 2014; 136: 2532–2543.

77. Raghavan A and Cesnik C. Effect of elevated temperature on guided-wave structural

health monitoring. J Intel. Mat. Syst. Str. 2008; 19: 1383–1398.

78. Salamone S, Bartoli I, di Scalea FL, et al. Guided-wave health monitoring of aircraft

composite panels under changing temperature. J Intel. Mat. Syst. Str. 2009; 20: 1079–1090.
79. Gandhi N and Michaels JE. Efficient perturbation analysis of Lamb wave dispersion

curves. Rev. of Quant. Nond. Eval. 2010; 29.

80. Dan CA, Kudela P, Radzienski M, et al. Temperature effects compensation strategy for

guided wave based structural health monitoring. In. Proceeding of 6th International

Symposium on NDT in Aerospace, 12-14th November 2014, Madrid, Spain.

81. Dodson JC and Inman DJ. Thermal sensitivity of Lamb waves for structural health

monitoring applications. Ultrasonics 2013; 53:677–685.

82. Fendzi C, Rébillat M, Mechbal N, et al. A data-driven temperature compensation

approach for Structural Health Monitoring using Lamb waves. Struct. Health. Monit. 2016;

15: 525–540.

83. Konstantinidis G, Drinkwater B and Wilcox PD. The temperature stability of guided

wave structural health monitoring systems. Smart Mater. Struct. 2006; 15: 967–976.

84. Konstantinidis G, Wilcox PD and Drinkwater BW. An investigation into the temperature

stability of a guided wave structural health monitoring system using permanently attached

sensors. IEEE Sensors Journal 2007; 7: 905–912.

85. Clarke T, Simonetti F and Cawley P. Guided wave health monitoring of complex

structures by sparse array systems: influence of temperature changes on performance. J.

Sound Vib. 2010; 329: 2306–2322.

86. Tibaduiza DA, Mujica LE, and Rodellar J, Damage classification in structural health

monitoring using principal component analysis and self-organizing maps. Structural Control

Health Monitoring, 2013; 20: 1303–1316.


87. Yan AM, Kerschen G, Boe PD, et al. Structural damage diagnosis under varying

environmental conditions—Part I: A linear analysis. Mechanical Systems and Signal

Processing, 2005; 19: 847–864.

88. Anaya M, Tibaduiza DA, Torres-Arredondo MA, et al. Data-driven methodology to

detect and classify structural changes under temperature variations. Smart Mater. Struct. 2014;

23: 045006

89. Torres-Arredondo MA, Sierra-Pérez J, and Cabanes G. An optimal baseline selection

methodology for data-driven damage detection and temperature compensation in acousto-

ultrasonics. Smart Mater. Struct. 2016; 25: 055034

90. Liu C, Harley JB, Ying Y, et al. Singular value decomposition for novelty detection in

ultrasonic pipe monitoring. In Proc. Sensors Smart Structures Technol. Civil, Mech., Aerosp.

Syst. San Diego, CA, USA, 2013; 8692.

91. Liu C, Harley JB, Berges M, et al. Robust ultrasonic damage detection under complex

environmental conditions using singular value decomposition. Ultrasonics 2015; 58: 75–86.

92. Zang C, Friswell MI, and Imregun M. Structural damage detection using independent

component analysis. Struct. Health. Monit. 2004, 3: 69–83.

93. Dobson J and Cawley P. Independent component analysis for improved defect detection

in guided wave monitoring. Proceedings of the IEEE. 2016; 104: 8.

94. Moll J, Kexel C, Pötzsch S, et al. Temperature affected guided wave propagation in a

composite plate complementing the Open Guided Waves Platform. Sci Data 2019; 6.

95. Yue N and Aliabadi MH. A scalable data-driven approach to temperature baseline

reconstruction for guided wave structural health monitoring of anisotropic carbon-fibre

reinforced polymer structures. Struct. Health. Monit. 2019; 1475921719887109.


96. Sun H, Yi J, Xu Y, et al. Identification and compensation technique of non-uniform

temperature field for Lamb wave-and multiple sensors-based damage detection. Sensors 2019;

19: 2930.

97. Lobkis OI and Weaver RL. Coda-wave interferometry in finite solids: Recovery of p-to-s

conversions rates in an elastodynamic billiard. Phys. Rev. Lett. 2003; 90.

98. Michaels JE and Michaels TE. Detection of structural damage from the local temporal

coherence of diffuse ultrasonic signals. IEEE Trans. Ultrason. Ferroelectr. Freq. Control

2005; 52: 1769–1782.

99. Weaver RL and Lobkis OL. Diffuse fields in ultrasonics and seismology. Geophysics

2006; 71: S15–S19.

100. Clarke T, Simonetti F, Rokhlin S, et al. Evaluation of the temperature stability of a low

frequency A0 mode transducer developed for SHM applications. AIP Conf. Proc. 2008,

vol.975, pp. 910–917.

101. Croxford AJ, Wilcox PD, Konstantinidis G, et al. Strategies for overcoming the effect of

temperature on guided wave structural health monitoring. In: Proceedings of SPIE 6532,

Health Monitoring of Structural and Biological Systems, 2007, 65321T.

102. Salmanpour MS, Khodaei ZS and Aliabadi MH. Guided wave temperature correction

methods in structural health monitoring. Journal of Intelligent Material Systems and

Structures, 2017; 28: 604–618.

103. Harley J and Moura J. Scale transform signal processing for optimal ultrasonic

temperature compensation. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 2012; 59:

2226–2236.
104. Liu G, Xiao Y, Zhang H, et al. Baseline signal reconstruction for temperature

compensation in Lamb wave-based damage detection. Sensors 2016; 16: 1273–1287.

105. Mariani S, Heinlein S and Cawley P. Compensation for temperature-dependent phase

and velocity of guided wave signals in baseline subtraction for structural health monitoring.

Struct. Health Monit. 2019, DOI: 10.1177/1475921719835155

106. Dan CA, Kudela P and Ostachowicz W. Compensation of temperature effects on guided

wave based structural health monitoring systems. 7th European Workshop on Structural

Health Monitoring, La Cité, Nantes, France July 8-11, 2014.

107. Croxford AJ, Moll J, Wilcox PD, et al. Efficient temperature compensation strategies for

guided wave structural health monitoring. Ultrasonics 2010; 50: 517-528.

108. Clarke T, Cawley P, Wilcox PD, et al. Evaluation of the damage detection capability of

a sparse-array guided-wave SHM system applied to a complex structure under varying

thermal conditions. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 2009; 56: 2666–2678.

109. Moll J and Fritzen CP. Guided waves for autonomous online identification of structural

defects under ambient temperature variations. J. Sound Vib. 2012; 331: 4587–4597.

110. Michaels JE. Sparse array imaging with guided waves under variable environmental

conditions. Duxford: Woodhead Publishing, 2016; 255–284.

111. Douglass ACS and Harley JB. Flexible, multi-measurement guided wave damage

detection under varying temperatures. AIP Conference Proceedings. July 2017

112. Lu Z, Lee SJ, Michaels JE et al. On the optimization of temperature compensation for

guided wave structural health monitoring. AIP Conf. Proc. 2010, vol.1211, pp.1860–1867.

113. Putkis O and Croxford AJ. Continuous baseline growth and monitoring for guided wave

SHM. Smart Mater. Struct. 2013; 22.


114. Dworakowski Z, Ambrozinski L and Stepinski T. Data fusion for compensation of

temperature variations in Lamb-wave based SHM systems. In: Proceedings of SPIE 9438,

Health Monitoring of Structural and Biological Systems, 2015, 94381S.

115. Aryan P, Kotousov A, Ng CT, et al. Reconstruction of baseline time-trace under

changing environmental and operational conditions. Smart Mater. Struct. 2016; 25: 035018.

116. Vitola J, Pozo F, Tibaduiza DA et al. Distributed piezoelectric sensor system for damage

identification in structures subjected to temperature changes. Sensors, 2017; 17: 1252

117. Anaya M, Tibaduiza DA and Pozo F. Artificial immune system (AIS) for damage

detection under variable temperature conditions. 8th European Workshop On Structural

Health Monitoring (EWSHM 2016), Bilbao, Spain, 5-8 July 2016.

118. Douglass ACS, Harley JB. Dynamic time warping temperature compensation for guided

wave structural health monitoring. IEEE Transactions on Ultrasonics, Ferroelectrics, and

Control. 2018; 65: 5.

119. Ren Y, Qiu L, Yuan S, et al. Gaussian mixture model–based path-synthesis

accumulation imaging of guided wave for damage monitoring of aircraft composite structures

under temperature variation. Struct. Health Monit. 2019; 18: 284–302.

120. Ren Y, Qiu L, Yuan S, et al. Multi-damage imaging of composite structures under

environmental and operational conditions using guided wave and Gaussian mixture model.

Smart Mater. Struct. 2019; 28: 115017.

121. Supreet K, Michaels JE and Harley JB. Robust baseline subtraction for ultrasonic full

wavefield analysis. AIP Conf Proc, 2017; 1806: 020005.

122. Wang YS, Gao LM, Yuan SF, et al. An adaptive filter-based temperature compensation

technique for structural health monitoring. J. Intell. Mater. Syst. Struct. 2014; 25: 2187–2198.
123. Herdovics B and Cegla F. Compensation of phase response changes in ultrasonic

transducers caused by temperature variations. Struct. Health Monit. 2019; 18: 508–523.

124. Dworakowski Z, Ambrozinski L and Stepinski T. Multi-stage temperature compensation

method for Lamb wave measurements. Journal of Sound and Vibration, 2016; 382: 328–339.

125. Engle RF and Granger CWJ. Co-integration and error-correction: representation,

estimation, and testing. Econometrica 1987; 55: 251–276.

126. Chen Q, Kruger U and Leung AYT. Cointegration testing method for monitoring non-

stationary processes. Ind. Eng. Chem. Res. 2009; 48: 3533–3543.

127. Cross EJ and Worden K. Approaches to nonlinear cointegration with a view towards

applications in SHM. J. Phys.: Conf. Ser. 2011; 305.

128. Cross EJ, Worden K and Chen Q. Cointegration: a novel approach for the removal of

environmental trends in structural health monitoring data. Proc. R. Soc. A 2011; 467: 2712–

2732.

129. Worden K, Cross EJ and Kyprianou A. Cointegration and nonstationarity in the context

of multiresolution analysis. J. Phys.: Conf. Ser. 2011; 305.

130. Dao PB and Staszewski WJ. Cointegration approach for temperature effect

compensation in Lamb-wave-based damage detection. Smart Mater. Struct. 2013; 22.

131. Dao PB and Staszewski WJ. Data normalisation for Lamb wave–based damage

detection using cointegration: A case study with single- and multiple-temperature trends. J.

Intell. Mater. Syst. Struct. 2013; 25: 845–857.

132. Sateesh N, Rao PS, Ravishanker DV, et al. Effect of moisture on GFRP composite

materials. Materials Today: Proceedings. 2015; 2: 2902–2908.


133. Alamri H and Low IM. Mechanical properties and water absorption behaviour of

recycled cellulose fibre reinforced epoxy composites. Polymer Testing 2012; 31: 620–628.

134. Schubert KJ and Herrmann AS. On the influence of moisture absorption on Lamb wave

propagation and measurements in viscoelastic CFRP using surface applied piezoelectric

sensors, Compos. Struct. 2012; 94: 3635–3643.

135. Ramsey JJ. Effects of wind on piezoelectric Lamb wave-based health monitoring. MS.

Thesis, University of Akron, USA, 2006

136. Salmanpour M, Sharif Khodaei Z and Aliabadi MH, Damage detection with ultrasonic

guided wave under operational conditions. 9th European Workshop on Structural Health

Monitoring Manchester, United Kingdom, July 10-13, 2018.

137. Chen F and Wilcox PD. The effect of load on guided wave propagation. Ultrasonics.

2007; 47: 111–122

138. Pao YH and Gamer U. Acoustoelastic waves in orthotropic media. J. Acoust. Soc. Am.

1985; 77: 806–812.

139. Michaels JE, Michaels TE and Martin RS. Analysis of global ultrasonic sensor data

from a full scale wing test, Rev. of Quant. Nond. Eval. 2009; 28.

140. Gandhi N, Michaels JE, Lee SJ. Acoustoelastic Lamb wave propagation in biaxially

stressed plates. J. Acoust. Soc. Am. 2012; 3: 1284–1293.

141. Michaels JE, Lee SJ, and Michaels TE. Impact of applied loads on guided wave

structural health monitoring. Review of Progress in Quantitative Nondestructive Evaluation.

2011; 30: 1515–1522.

142. Mohabuth M, Kotousov A, Ng CT, et al. Implication of changing loading conditions on

structural health monitoring utilising guided waves. Smart Mater. Struct. 2018; 27: 025003.
143. Crawley EF and De Luis J. Use of piezoelectric actuators as elements of intelligent

structures. AIAA J. 1987; 25: 1373–1385.

144. Seshu P and Naganathan N. Finite-element analysis of strain transfer in an induced

strain actuator. Smart Mater. Struct. 1997; 6: 76–88.

145. Rabinovitch O and Vinson J. Adhesive layer effects in surface-mounted piezoelectric

actuator. J. Intell. Mater. Syst. Struct. 2002; 13: 689–704.

146. Bhalla S and Soh C. Electromechanical impedance modeling for adhesively boned

piezo-transducers. J. Intell. Mater. Syst. Struct. 2004; 15: 955–972.

147. Dugnani R. A modified global-local analysis model of a PZT disk transducer bonded to

a host structure. In: Proceedings of 6th international workshop on structural health

monitoring, Stanford University, Stanford, CA, 11–13 September 2007, vol.26, pp.859–868.

148. Ha S and Chang FK. Adhesive interface layer effects in PZT-induced Lamb wave

propagation, Smart Mater. Struct. 2010; 19.

149. Giurgiutiu V. Tuned Lamb wave excitation and detection with piezoelectric wafer active

sensors for structural health monitoring. J. Intell. Mater. Syst. Struct. 2005; 16. 291–305.

150. Giurgiutiu V and Santoni-Bottai G. Extension of shear lag solution for structurally

attached ultrasonic active sensors. AIAA J. 2009; 47. 1980–1983

151. Yu L, Bottai-Santoni G, Giurgiutiu V. Shear lag solution for tuning ultrasonic

piezoelectric wafer active sensors with applications to Lamb wave array imaging.

International J Eng. Science. 2010; 48. 848–861.

152. Dugnani R. Extension of the Crawley’s adhesive model to dynamically actuated

piezoelectric transducers. J. Intell. Mater. Syst. Struct. 2016; 27. 2112–2124.


153. Santoni-Bottai G and Giurgiutiu V. Exact shear-lag solution for guided waves tuning

with piezoelectric-wafer active sensors. AIAA J. 2012; 50: 2285–2294.

154. Kamal A.M. Lin B. and Giurgiutiu V. Exact analytical modelling of power and energy

for multimode Lamb waves excited by piezoelectric wafer active sensors. J. Intell. Mater.

Syst. Struct. 2014; 25: 452–471.

155. Giurgiutiu V. Structural health monitoring with piezoelectric wafer active sensors.

Elsevier Academic Press. Oxford. 2014.

156. Kapuria S. and Agrahari J.K. Two dimensional shear lag solution for stress transfer

between rectangular piezoelectric wafer transducer and orthotropic host plate. European J.

Mech. A/Solids. 2016; 55. 181–191.

157. Willberg C, Duczek S and Gabbert U. Increasing the scanning range of Lamb wave

based SHM systems by optimizing the actuator–sensor design. CEAS Aeronaut. J. 2013; 4:

1–12.

158. Islam MM and Huang H. Effects of adhesive thickness on the Lamb wave pitch-catch

signal using bonded piezoelectric wafer transducers. Smart Mater. Struct. 2016; 25.

159. Lanzara G, Yoon Y, Kim Y, et al. Influence of interface degradation on the performance

of piezoelectric actuators. J. Intell. Mater. Syst. Struct. 2009; 20: 1699–1711.

160. Sirohi J and Chopra I. Fundamental behavior of piezoceramic sheet actuators. J. Intell.

Mater. Syst. Struct. 2000; 11: 47–61.

161. Park G, Farrar CR, Rutherford AC, et al. Piezoelectric active sensor self-diagnostics

using electrical admittance measurements. J. Vib. Acoust. 2006; 128: 469–476.

162. Overly TG, Park G, Farinholt KM, et al. Piezoelectric active-sensor diagnostics and

validation using instantaneous baseline data. IEEE Sensors J. 2009; 9: 1414–1421.


163. Park S, Park G, Yun CB, et al. Sensor self-diagnosis using a modified impedance model

for active sensing-based structural health monitoring. Struct. Health Monit. 2009; 8: 71–82.

164. Lee SJ, Sohn H, Michaels JE, et al. In situ detection of surface-mounted PZT transducer

defects using linear reciprocity. Rev. of Quant. Nond. Eval. 2010; 29: 1844–1851.

165. Tinoco HA and Serpa AL. Voltage relations for debonding detection of piezoelectric

sensors with segmented electrode. Mech. Syst. Signal Process. 2012; 31: 258–267.

166. Tinoco HA, Serpa AL and Ramos AM. Numerical study of the effects of bonding layer

properties on electrical signatures of piezoelectric sensors. Mecanica Computacional 2010;

29: 8391–8409.

167. Sathyanarayana CN, Ashwin U and Raja S. Effect of sensor debonding on Lamb wave

propagation in plate structures. ARPN J Eng. Appl. Sci. 2014; 9.

168. Mulligan KR, Quaegebeur N, Masson P, et al. Compensation of piezoceramic bonding

layer degradation for structural health monitoring. Struct. Health Monit. 2014; 13: 68–81.

169. Gao D, Wu Z, Wang Y, et al. Influence of bonding quality for piezoelectric sensors with

applications to Lamb wave damage detection. Appl. Mech. Mater. 2013; 330: 494–499.

170. Bekas DG, Khodaei ZS, Aliabadi MH. An innovative diagnostic film for structural

health monitoring of metallic and composite structures. Sensors 2018; 18: 2084.

171. Ing RK and Fink M. Self-focussing and time recompression of Lamb waves using a time

reversal mirror. Ultrasonics 1998; 36: 179–186.

172. Ing RK and Fink M. Time-reversed Lamb waves. IEEE Trans. Ultrason. Ferroelectr.

Freq. Control 1998; 45: 1032–1043.


173. Wang CH, Rose JT and Chang FK. Computerized time-reversal method for structural

health monitoring. In: Proceedings of SPIE 5046, Nondestructive Evaluation and Health

Monitoring of Aerospace Materials and Composites II, San Diego, CA, 2003, 48.

174. Wang CH, Rose JT and Chang FK. A synthetic time reversal imaging method for

structural health monitoring. Smart Mater. Struct. 2004; 13.

175. Blanloeuil P, Rose LRF, Guinto JA, et al. Closed crack imaging using time reversal

method based on fundamental and second harmonic scattering. Wave Motion 2016; 66: 156–

176.

176. Blanloeuil P, Rose LRF, Veidt M, et al. Time reversal invariance for a nonlinear

scatterer exhibiting contact acoustic nonlinearity. J.Sound Vib. 2018; 417: 413–431.

177. Sohn H, Park HW and Law KH. Combination of a time reversal process and a

consecutive outlier analysis for baseline-free damage diagnosis. J. Intell. Mater. Syst. Struct.

2007; 18: 335–346.

178. Park HW, Lee C and Sohn H. Time reversal active sensing for health monitoring of a

composite plate. J. Sound Vib. 2007; 302: 50–66.

179. Park HW, Kim B and Sohn H. Understanding a time reversal process in Lamb wave

propagation. Wave Motion 2009; 46: 451–467.

180. Lee SJ, Sohn H and Homg JW. Time reversal based piezoelectric transducer self-

diagnosis under varying temperature. J Nondest. Eval. 2010; 29: 75–79.

181. Poddar B, Kumar A, Mitra M, et al. Time reversibility of a Lamb wave for damage

detection in a metallic plate. Smart Mater. Struct. 2011; 20.

182. Poddar B, Bijudas CR, Mitra M, et al. Damage detection in a woven-fabric composite

laminate using time-reversed Lamb wave. Struct. Health Monit. 2012; 11: 602–612.
183. Bijudas CR, Mitra M and Mujumdar PM. Time reversed Lamb wave for damage

detection in a stiffened aluminum plate. Smart Mater. Struct. 2013; 22.

184. Watkins R and Jha R. A modified time reversal method for Lamb wave based

diagnostics of composite structures. Mech. Syst. Signal Process. 2012; 31: 345–354.

185. Zeng L, Lin J and Huang L. A modified Lamb wave time-reversal method for health

monitoring of composite structures. Sensors, 2017; 17: 955–969.

186. Wang J and Shen Y. An enhanced Lamb wave virtual time reversal technique for

damage detection with transducer transfer function compensation. Smart Mater. Struct. 2019;

28: 085017.

187. Miao X, Wang D, Ye L, et al. Identification of dual notches based on time-reversal

Lamb waves and a damage diagnostic imaging algorithm. J. Intell. Mater. Syst. Struct. 2011;

22: 1983–1992.

188. Zhu R, Huang GL and Yuan FG. Fast damage imaging using the time-reversal technique

in the frequency wavenumber domain. Smart Mater. Struct. 2013; 22.

189. Mustapha S, Lu Y, Li J, et al. Damage detection in rebar-reinforced concrete beams

based on time reversal of guided waves. Struct. Health Monit. 2014; 13: 347–358.

190. Qu W, Xiao L, Zhou Y, et al. Lamb wave damage detection using time reversal DORT

method. Smart Mater. Struct. 2013; 22.

191. Park S, Lee C and Sohn H. Reference-free crack detection using transfer impedances. J.

Sound Vib. 2010; 329: 2337–2348.

192. Anton SR, Inman DJ and Park G. Reference-free damage detection using instantaneous

baseline measurements. AIAA J. 2009; 47: 1952–1964.


193. Yeum CM, Sohn H, Lim HJ, et al. Reference free delamination detection using Lamb

waves. Struct. Contol Health Monit. 2014; 21: 675–684.

194. Qiang W, Shenfang Y. Baseline-free imaging method based on new PZT sensor

arrangements. J. Intell. Mater. Syst. Struct. 2009; 20: 1663–1673.

195. Bagheri A, Li K and Rizzo P. Reference-free damage detection by means of wavelet

transform and empirical mode decomposition applied to Lamb waves. J. Intell. Mater. Syst.

Struct. 2013; 24: 194–208.

196. Kim SB, Lee CG, Hong JW, et al. Applications of an instantaneous damage detection

technique to plates with additional complexities. J Nondest. Eval. 2010; 29: 189–205.

197. Kim EJ, Kim MK, Sohn H, et al. Investigating electro-mechanical signals from

collocated piezoelectric wafers for the reference-free damage diagnosis of a plate. Smart

Mater. Struct. 2011; 20.

198. Li G and Chattopadhyay A. Reference-free damage localization in time-space domain

for structural health monitoring of X-COR sandwich composites. Journal of Intelligent

Material Systems and Structures, 2019; 30: 371–385.

199. Alem B, Abedian A and Nasrollahi-Nasab K. Reference-free damage identification in

plate-like structures using Lamb-wave propagation with embedded piezoelectric sensors. J.

Aerosp. Eng. 2016; 29: 04016062.

200. Hameed MS, Li Z, Chen J, et al. Lamb-wave-based multistage damage detection method

using an active PZT sensor network for large structures. Sensors, 2019; 19: 2010.

201. Salmanpour MS, Khodaei ZS and Aliabadi MH. Instantaneous baseline damage

localization using sensor mapping. IEEE Sensors Journal 2017; 17.


202. Alguri KS, Melville J and Harley JB. Baseline-free guided wave damage detection with

surrogate data and dictionary learning. The Journal of the Acoustical Society of America,

2018; 143: 3807.

203. Lizé E, Rébillat M, Mechbal N, et al. Optimal dual-PZT sizing and network design for

baseline-free SHM of complex anisotropic composite structures. Smart Mater. Struct. 2018;

27: 115018.

204. Qiu J, Li F, Abbas S, et al. A baseline-free damage detection approach based on distance

compensation of guided waves. Journal of Low Frequency Noise, Vibration and Active

Control, 2019; 38: 1132–1148.

205. Huang L, Zeng L and Lin J. Baseline-free damage detection in composite plates based

on the reciprocity principle. Smart Mater. Struct, 2018; 27: 015026.

206. Lim HJ, Sohn H, Yeum CM, et al. Reference free damage detection, localization, and

quantification in composites. J. Acoust. Soc. Am. 2013; 133.

207. An YK and Sohn H. Instantaneous crack detection under varying temperature and static

loading conditions. Struct. Control Health Monit. 2010; 17: 730–741.


The author(s) declared no potential conflicts of interest with respect to the research,
authorship, and/or publication of this article.
Highlights

 Lamb wave based structural health monitoring (SHM) is an emerging technology to


monitor the ‘health’ status of structures in real time or whenever necessary.
 In Lamb wave based SHM techniques, discriminate between changes due to damage and
those due to environmental and operational conditions (EOCs) effects is complicated.
 This paper mainly provides a literature review of EOCs effects on Lamb wave propagation
in structures.
 EOCs effects’ compensation strategies for Lamb wave based SHM systems are also
reviewed in this paper.
 Moreover, baseline free damage identification techniques are briefly summarized in this
paper.

You might also like