Original Article

Structural Health Monitoring

1–9

� The Author(s) 2017

Reprints and permissions:

sagepub.co.uk/journalsPermissions.nav

DOI: 10.1177/1475921717707356

journals.sagepub.com/home/shm

Toward identifying crack-length-relatedresonances in acoustic emissionwaveforms for structural healthmonitoring applications

Md Yeasin Bhuiyan1, Jingjing Bao1, Banibrata Poddar2 and VictorGiurgiutiu1

AbstractIn this study, we focus on analyzing the acoustic emission waveforms of the fatigue crack growth despite the conventionalstatistics-based analysis of acoustic emission. The acoustic emission monitoring technique is a well-known approach inthe non-destructive evaluation/structural health monitoring research field. The growth of the fatigue crack causes theacoustic emission in the material that propagates in the structure. The acoustic emission happens not only from thecrack growth but also from the interaction of the crack tips during the fatigue loading in the structure. The acousticemission waveforms are generated from the acoustic emission events; they propagate and create local vibration modesalong the crack faces (crack resonance). In-situ fatigue and acoustic emission experiments were conducted to monitorthe acoustic emission waveforms from the fatigue cracks. Several test specimens were used in the fatigue experiments,and corresponding acoustic emission waveforms were captured. The acoustic emission waveforms were analyzed anddistinguished into three types based on the similar nature in both time and frequency domains. Three-dimensional har-monic finite element analyses were performed to identify the local vibration modes. The local crack resonance phenom-enon has been observed from the finite element simulation that could potentially give the geometric information of thecrack. The laser Doppler vibrometry experiment was performed to identify the crack resonance phenomenon, and theexperimental results were used to verify the simulated results.

KeywordsStructural health monitoring, non-destructive evaluation, acoustic emission, fatigue crack, finite element analysis

Introduction

The researchers in structural health monitoring (SHM)and non-destructive evaluation (NDE) always thrivefor the techniques to assess the health of the struc-ture.1–4 The acoustic emission (AE) technique of NDE/SHM has been used for structural defects detection formany years.5–12 It can be categorized in the passivedetection of the structural defects since it uses the defectas the passive source of AE. Recently, the fatigue-crack-related AE detection has attracted much atten-tion to the researchers. The detection of fatigue crackpropagation is important to prevent any potential cata-strophic failure of a structure. The detection of fatiguecracks was investigated in both active detection13–15

and passive detection8 in the past.The AE technique has a wide range of applicability

such as damage source localization, characterization,

crack-growth monitoring in the isotropic materials,16–19

and anisotropic composite and concrete structures.20–25

Many methods were observed to focus on the first fewpeaks of the waveform, and the examination of thesefew wave peaks had given very limited informationsuch as AE wave amplitude, average frequency, time ofarrival, and duration. To understand the AE sourcemechanism, finite element and some analytical works

1Department of Mechanical Engineering, University of South Carolina,

Columbia, SC, USA2Intelligent Automation Inc., Rockville, MD, USA

Corresponding author:

Md Yeasin Bhuiyan, Department of Mechanical Engineering, University of

South Carolina, Room#A236, 300 Main Street, Columbia, SC 29208,

USA.

Email: mbhuiyan@email.sc.edu

had been reported.22,26 In fact, the complex nature ofthe AE wave generation, propagation, and interactionwith the structural features makes it difficult to analyzethe AE waveforms.

Piezoelectric wafer active sensors (PWAS) were usedto detect the AE waveforms in the past, and the perfor-mance of PWAS was well characterized for detectingultrasonic guided waves.27–29 Efforts were made toretrieve the structural-feature-related information fromthe AE waveforms.8 Bao et al.30 discussed the identifi-cation of geometric feature of the fatigue crack growthfrom AE signal. Gagar et al.12 developed some correla-tions between AE signals generated during fatigue crackgrowth and corresponding cyclic loads to determine thecrack length. Sagar and Prasad31 reported a parametricrelationship of AE and fracture mechanics to estimatecrack length. However, these methods predominantlyrely on experimental data-driven statistical models tounderstand the AE signatures. To successfully extractgeometric feature information from the AE events dueto fatigue crack growth, it is important to analyze theAE waveforms and to find any clue on estimating thecrack-length information.

The nonlinear acoustic technique for crack detectionand the generation of harmonics for breathing-type crackshad been analyzed numerically and experimentally.32–34

The results showed the higher harmonic generation due tocrack formation. However, these analyses focused oninput voltage amplitude change with harmonics32 andlocalizing the breathing cracks in the structure.33

This article presents an experimental and computa-tional analysis of AE waveforms resulting from fatiguecrack growth. It focuses on the analysis toward identifyingcrack-length-related resonance phenomenon in the experi-mental AE waveforms. The measured AE waveformswere analyzed in both time and frequency domains, andthey were classified into three major types. To understandthe insight of the AE waveforms and their interaction withlocal crack boundaries, three-dimensional (3D) finite ele-ment harmonic analysis was applied. The experimentalresults provided the guidelines for choosing the frequencyrange of the finite element model (FEM) harmonicanalysis and excitation frequencies of the laser Dopplervibrometry (LDV) experiments. FEM results showed thecrack-length-related resonance phenomenon that was alsoconfirmed by the LDV experiments. Both the FEM andLDV experimental results suggest that the crack reso-nance phenomena could be used to estimate the fatiguecrack length from the AE waveforms.

Experimental setup for fatigue–AEexperiments

The fatigue experiments were designed to capture thereal-time AE signals. Aircraft grade aluminum Al-2024

T3 test coupons of 100-mm-wide, 300-mm-long and1-mm-thick dimension were used. A small (1-mm dia-meter) hole was created at the center of the specimen toinitiate the crack growth. It was important to minimizethe boundary reflections from the coupon edges toextract the AE waveform that was related to a fatiguecrack. Thus, an absorbing clay boundary around theedges of the coupon was used and the measured AEwaveforms were free of coupon edge reflection contam-ination. The clay absorbing boundary is outside thefatigue crack region and the sensor.

Thus, the recorded AE waveform may contain infor-mation from any possible interaction between thecrack-tip AE and the crack boundaries.

The fatigue loading level was selected based on theyield limit (345 MPa) of the material. The MTSmechanical testing machine was used to apply the fati-gue loading in the specimen. The highest and lowestload level was, respectively, kept at 65% and 6.5% ofthe yield strength maintaining an R ratio (smin=smax)of 0.1. At the beginning, an initial crack was created bythe cyclic fatigue loading without any AE instrumenta-tion. Then, the AE sensor was bonded to the specimenwith AE instrumentation. The fatigue loading with AEinstrumentation was conducted in two stages: fast fati-gue loading (0.25 Hz) and slow fatigue loading rate(0.05 Hz). At fast loading rate, the coupon undergoesone complete cycle of loading in 4 s. While the crackwas growing faster, we wanted more control on captur-ing the AE events and thus conducted the experimentat the slow fatigue loading rate which was five timesslower than the fast loading rate. In both cases, thesame cyclic load level was maintained.

The AE sensor (PWAS) was installed 20 mm awayfrom the initially drilled hole as shown in Figure 1(b).Since the AE signals were low in amplitude, the pream-plifiers were used in between the sensor and the AEinstruments. The preamplifier also filtered the very low-frequency vibration signals. Those vibration signalswere not related to the AE signals and were persistentlypresent there originated from the machine and othersources. The digital AE system from the physical acous-tics was used to detect the AE signals.

Experimental AE waveformmeasurements

The fatigue loading was monitored by the MTS controlmodule. At slow fatigue loading rate, the AE hits werecaptured and the corresponding fatigue loading cyclewas recorded. The most of the AE hits were observedat near the peak load of the fatigue load cycle as shownin Figure 2(a). The AE hits happened during both load-ing and unloading. Each of the AE hits corresponds to

2 Structural Health Monitoring

an AE waveform. The AE waveforms had very shortperiod of time (in the range of microsecond), while thecyclic fatigue loading rate was in the range of second.An AE waveform corresponding to an AE hit is shownin Figure 2(b).

The fatigue experiment was continued at fast loadingrate (0.25 Hz) and the corresponding crack growth wasvisually monitored by a video camera. Simultaneously,the PWAS captured the AE signals by the AE instru-mentation. We noticed that the number of recorded AEhits increased significantly as the crack grew faster(;6mm per cycle, on average). Since a large number ofmolecular bonding had to break as the crack grew atsuch a higher rate (;6mm per cycle, on average); thusreleasing the stored energy at a higher rate. The releaseof stored energy at higher rate preserved the marking asthe higher rate (;8–10 hits per cycle) of AE signalsrecorded in the experiments.

The AE signals corresponding to the crack growthfrom 20 to 35 mm (tip to tip length) was recorded.Several test coupons of the same geometry and sameexperimental setup were used to generate the repeatabil-ity of the AE signals. The test coupons were manufac-tured from a single, large aluminum plate to maintainthe same material properties in all the test coupons. Itwas observed that the measured AE waveforms couldbe predominantly grouped into three types of wave-forms. The waveforms were distinguished based on thesimilar nature in the both time and frequency domains.

‘‘Type I’’ AE waveforms are the short length in thetime domain as shown in Figure 3. The frequency spec-trum of the Type I signal shows that they have higheramplitude in the 250–400 kHz frequency band. The‘‘Type II’’ AE signals have relatively longer time periodas shown in Figure 4. They have relatively lower fre-quency components (30–100 kHz) as depicted in their

Figure 1. (a) Fatigue experimental setup and (b) PWAS transducer (7 mm diameter) mounted on the specimen.

Figure 2. (a) AE hits near the peak load of the fatigue cyclic loading (slow fatigue cycle, Tc = 20 s) and (b) AE waveformcorresponding to an AE hit.

Bhuiyan et al. 3

frequency spectrum. ‘‘Type III’’ AE waveforms thatwere identified as mixed-type waveforms are shown inFigure 5. They have both components of ‘‘Type I’’ and‘‘Type II’’ signals and their frequency spectrum alsosupport the mixed nature of the waveforms. However,more than 90% of the AE signals can be classified into‘‘Type I’’ and ‘‘Type II’’ waveforms. Both types ofwaveforms appeared in almost equal proportions.

All three types of signals had the low-amplitudenoises in the raw experimental time domain signals.These noises are persistently present in the entire lengthof all three types of waveforms. We have identified thatthe source of noise is the surrounding environment ofthe experiment that includes MTS machine, electro-magnetic control module, and servo motor. However,the signal processing easily removed those noises fromthe raw AE signals.

To explain the types of AE waveforms, we couldobserve the physical phenomena. As the load levelreaches to the maximum of the fatigue load cycle, thestresses at the crack tip rise at a significant amount thatcauses the failure of the material at the crack tip. Thisreleases the stored energy from the crack tip and gener-ates AE event. This happens instantaneously while thecrack tips are open. This physical situation may repre-sent one type of the AE waveforms. The generated AEwaves travel along the crack tips as surface waves.

At lower load level, the crack tips tend to close andespecially at the crack-tip region, the rubbing and clap-ping happen between the faying surfaces. These rubbingand clapping motions of the faying surfaces generateanother type of AE waveform. The third type of AEwaveform may correspond to the mixed situation of thetwo previous physical phenomena and only 10% of thetotal AE waveforms was observed.

Since we focused on the AE waveforms capturedfrom the experiment, we may note the interaction ofthe AE waveform with the crack boundaries. While thefatigue crack grows, the generated AE waveform fromone of the crack tips may propagate along the crack tipas surface waves and reflected back from the other tipof the crack. This local interaction of the AE waveformfrom the crack faces may cause the crack resonancethat heavily depends on the geometry of the crack.

In order to observe the crack resonances, FEMsimulation of AE waveform is performed. It can benoted from the experimental result analyses of the AEwaveforms that all three types of AE waveform containthe frequency content up to 600 kHz. This informationguides us performing the harmonic and transient FEManalysis in the interested frequency domain. The wave-length, thereby, frequency dictates the element size ofFEM. While we know the maximum AE experimentalfrequency content (600 kHz), it guides us setting up themeshing parameters and others as detailed in section‘‘FEM simulation of AE waveform generation andcrack resonance.’’ To maintain proper accuracy inFEM simulation, the design model parameters shouldbe good for a frequency that exceeds the maximum fre-quency of any type of AE experimental waveform.

The frequency spectrum of the experimental AEwaveform also provides the guidance for the LDVexperiment of crack resonance observation as detailedin section ‘‘Experimental verification of crack reso-nances.’’ In LDV experiment, PWAS transducer wasused to excite the manufactured crack to simulate anAE waveform at the crack tip. The frequency spectrumof the experimental AE waveform dictated the fre-quency spectrum of the PWAS exciter in LDV experi-ment. Hence, the knowledge of experimental AEwaveform analysis facilitated performing FEM simula-tion and LDV experiment for simulated AE waveform.

Figure 3. ‘‘Type I’’ AE signals from the experiment.

“Type II” AE waveform

Figure 4. ‘‘Type II’’ AE signals from the experiment.

4 Structural Health Monitoring

FEM simulation of AE waveformgeneration and crack resonance

Based on the physical phenomena discussed in section‘‘Experimental AE waveform measurements,’’ the AEwaveforms travel along the crack tips. Since the bound-ary condition changes as the waveforms approach tothe other end of the crack, reflection and transmissionmay happen at the other tip of the cracks. That alsomeans that the other crack tip would act as a secondarysource of scattered AE waveforms. This local interac-tion depends on the geometry of the crack. Hence, theremay be some resonance frequencies for different modesof local vibration. This phenomenon may be termed asthe crack resonance.

Crack on each side of the small hole is modeled usingthe discontinuity at the adjacent pair of nodes along thecracks. There are actually two sets of nodes along the

crack faces and each set represents the nodes on eachface and the nodes are discontinuous along the crackfaces. Two sets of nodes are adjacent to each other andthe solid elements are disbanded along the crack faces.The modeling of the cracks in finite element using theabove approach is fair enough to model the actualcracks in a small hole in the plate-like structure.35,36

In order to prove the concept of crack resonance, 3DFEM simulation was performed. The AE source wasmodeled using dipole concept suggested by Hamstad etal.37 To model the fatigue crack growth, the conceptwas extended and two dipoles were modeled along thethickness that represents a line dipole source. The illus-tration of the 3D FEM modeling is shown in Figure 6.The non-reflective boundary (NRB) was used in theFEM to avoid the reflections from the plate edges. TheNRB was modeled using the criteria mentioned in Shenand Giurgiutiu.38

The criteria that followed for better FEM resultswere lm=cs � 15, D=s � 7, and lm=s � 5 followingHamstad et al.,37 where lm is the minimum wavelength,cs is the cell size in the FEM, D is the distance betweenthe source and the sensing location, and s is the sourcesize. The minimum wavelength lm is defined aslm = cs=f , with cs being the shear wave speed and fbeing the maximum frequency. The cell sizes (cs) variedfrom 0.9 to 0.25 mm with D = 20mm and s = 0:2mm.The AE source dipoles were chosen as half cycle cosinestep function with 15-ms rise time. The amplitude ofthe function is chosen to be unity (1). The aluminum2024-T3 material properties of the plate were chosen tobe E = 73:1GPa, n = 0:33, and r = 2780 kg=m3 withcs = 3140m=s. With these parameters being set, theFEM simulation can accurately predict the results forthe frequency range of up to 840 kHz with D=s = 100

following Hamstad et al.37

Harmonic FEM analysis of the 3D model was per-formed for the plate with hole + 10 mm cracks. Thetwo models were designed to show the differences in thefrequency spectrum of the response signals. The FEMresults were illustrated in Figure 7. The crack openingdisplacement at the mouth of the crack was recordedover the wide frequency band as shown in Figure 7(a).This showed that the crack-opening resonancesoccurred at regular frequency intervals. Although theresults were shown for wide range of frequencies, wewere interested in the results up to 1000 kHz wheremost of AE signal frequency contents were available.

Transient analysis of the same model was performedfor the plate with hole and hole + 10 mm cracks. Thefrequency spectrum of the displacement responses ofthe two situations was illustrated in Figure 7(b). Thisshowed that the AE signals are modified by the pres-ence of the crack. This gave us an indication that thegeometry of the crack may also be estimated from the

Figure 5. ‘‘Type III’’ AE signals from the experiment.

Figure 6. 3D FEM for harmonic analysis: (a) top view, (b) frontview, (c) dipole loading at the crack tip, and (d) line load alongthe thickness.

Bhuiyan et al. 5

recorded AE signals. The length of the crack is directlyrelated to the resonance frequencies of the crack. It isalso interesting to note that the resonance phenomenacaptured at the mouth of the crack (from harmonicanalysis) were similar to that captured at 20 mm awayfrom the crack (transient analysis). The comparisonbetween the frequency peaks is illustrated by the dottedlines.

The animation snapshot of the transient FEM simu-lation is demonstrated in Figure 8. The acoustic waveemitted from one crack tip travels along the crackfaces. When the waves hit the other tip of the crack, itgenerates secondary waves. The reflected waves travelback along the crack faces. This may cause the crackresonances.

Experimental verification of crackresonances

Finite element simulation results proposed that fatiguecrack resonates during AE events. An LDV experimentwas done to verify the results. The experimental setupis shown in Figure 9. A 1-mm-thick 2024-T3 aluminumplate with 1.2 m 3 1.2 m dimension was used. Thelarge size of the specimen was used to avoid the plateedge reflections. A 16-mm-long through-thickness slitwas made to simulate a fatigue crack. A pair of 7-mm-diameter circular piezoelectric active wafer sensors(PWAS) was mounted on the opposite surfaces of theplate at the tip of the slit as the excitation source.

A single-count tone-burst signal of center frequencyof 350 kHz was used to excite the sensors placed at thecrack tip. The reason for choosing the single-counttone-burst signal was to achieve a wider band in the fre-quency (up to 800 kHz). The results of the actual AEwaveform presented in section ‘‘Experimental AE wave-form measurements’’ provided the guidance of selectingthe frequency band. To correctly simulate an AE signalat the crack tip, the frequency content of the PWASexcitation at the crack tip should cover the entire fre-quency range of the actual AE waveform. LDV mea-sured the out-of-plane vibration velocity of the platesurface at 20 mm away from the crack.

The out-of-plane velocity of a point at 20 mm awayfrom the center of the crack was measured by LDV.Fourier transform of the time domain signal was usedto obtain the frequency spectrum of the signal, and it isshown in Figure 10. Several peaks are clearly identifiedand labeled in the frequency spectrum. Evidently, thesepeaks represent the frequencies for which the out-of-plane velocity reaches a maximum, and the crack reso-nances happen at anti-nodes.

Comparing the experimental data with the FEMsimulation results, we can see that both the results wereable to capture several resonance modes of the crackand higher harmonics. However, the amplitude trendincreases with frequency in the experiment while it isopposite in the simulation. The reason is that the exci-tation signal in the experiment was tone burst whichhas higher amplitude near the center frequency, whilein the FEM simulation, the excitation is a cosine bellstep function which has higher amplitudes in the lower

Figure 7. FEM simulation results for (a) the crack resonancecaptured at mouth of the crack in harmonic analysis and(b) comparison between FFTof AE waveform of the hole andhole + 10 mm crack (sensing at 20 mm away from the hole) intransient analysis.

Figure 8. Animation snapshots of transient FEM simulation asthe time progress: (a) t = 1:5 ms, (b) t = 3 ms, (c) t = 3:9 ms, and(d) t = 4:8 ms.

6 Structural Health Monitoring

frequencies. That is why the trend in the amplitude ofthe frequency response curve from the two results isnot directly comparable. Nonetheless, in both experi-ment and FEM simulation, some frequency peaks havebeen observed within the AE frequency band caused bythe crack resonance which is the objective of the article.

Summary and conclusion

In this research, in-situ fatigue and AE experiment wasconducted to capture the AEs due to fatigue crackgrowth using PWAS transducers. Two stages of fatigueexperiments were conducted: slow fatigue and fast fati-gue. The slow fatigue experiments provided more con-trol on capturing AE events with the fatigue load cycle.It had been shown that most of the AE hits happenednear the peak load of the cyclic fatigue loading. The fastfatigue experiment was conducted to allow about 15

mm crack growth from the initial crack length withsimultaneous measurement of AE. It was observed thatthe number of AE hits increases significantly as thecrack grows faster. By analyzing the pool of AE wave-forms, three types of AE waveforms were identifiedbased on the similarity in both time and frequency spec-tra. The AE waveform generated from the crack tipmay interact with the crack itself and cause crack reso-nance. Transient and harmonic analyses were carriedout to the 3D FEM to simulate the AE wave genera-tion, propagation, and its interaction with fatiguecrack. Both analyses confirm the local crack resonancephenomena due to the interaction between AE wave-form and fatigue crack and showed that they are relatedto the crack length. The LDV experiments were con-ducted to verify the computational results.

Future work

We will analyze the AE waveforms from the fatigueexperiments to extract the crack-length information.The observation of crack-resonance-related frequencypeaks in the real fatigue experiment AE waveform is yetto be determined. In the future research in the sequel,the crack resonance in the real AE waveform would beaddressed using the simulated results in the presentanalysis. A combination of physics of material-basedunderstanding and experimental validation with simula-tion may be used to synthesize the AE waves to harnessmore information. We will conduct more experimentsto generate the individual physical phenomenon in acontrolled way to collect the AE signals from eachphenomenon.

At lower load level, the crack tends to close andcrack surfaces lead to a nonlinear contact problem forthe AE source. The nonlinearity of the breathing crackmay generate harmonics in multiples of primary fre-quency. It would require further investigation to

Figure 9. Laser Doppler vibrometry (LDV) experimental setup.

Figure 10. LDV experimental results (out-of-plane velocity)measured at 20 mm away from the slit. The resonancefrequencies are labeled in the frequency spectrum.

Bhuiyan et al. 7

compare the nonlinear effect in the breathing cracksand the crack resonance phenomena.

Acknowledgements

Dr Ignacio Perez, ProgramManager, is thankfully acknowledged.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest withrespect to the research, authorship, and/or publication of thisarticle.

Funding

The author(s) disclosed receipt of the following financial sup-port for the research, authorship, and/or publication of thisarticle: This study was supported by the Office of NavalResearch (no. N00014-14-1-0655).

References

1. Wu Q, Yu F, Okabe Y, et al. Acoustic emission detection

and position identification of transverse cracks in carbon

fiber-reinforced plastic laminates by using a novel optical

fiber ultrasonic sensing system. Struct Health Monit 2015;

14: 205–213.2. Janapati V, Kopsaftopoulos F, Li F, et al. Damage detec-

tion sensitivity characterization of acousto-ultrasound-

based structural health monitoring techniques. Struct

Health Monit 2016; 15: 143–161.3. Cuadra J, Vanniamparambil PA, Servansky D, et al.

Acoustic emission source modeling using a data-driven

approach. J Sound Vib 2015; 341: 222–236.4. Chang F-K, Janapati V, Kopsaftopoulos F, et al. Quanti-

fication of structural health monitoring for damage detec-

tion. In: Proceedings of the 30th aircraft structural integrity

program (ASIP), San Antonio, TX, 2–4 December 2014.5. Niri ED, Farhidzadeh A and Salamone S. Adaptive mul-

tisensor data fusion for acoustic emission source localiza-

tion in noisy environment. Struct Health Monit 2013; 12:

59–77.6. Ohtsu M and Ono K. Ageneralized theory and source

representations of acoustic emission. In: Proceedings of

the 2nd international conference on acoustic emission, Lake

Tahoe, NV, 28 October–1 November 1985, pp.S50–S53.

Bellingham, WA: SPIE.7. Perez IM, Cui H and Udd E. Acoustic emission detection

using fiber Bragg gratings. In: Proceedings of the SPIE’s

8th annual international symposium on smart structures

and materials, Newport Beach, CA, 4 March 2001, vol.

4328, pp. 209–215. Bellingham, WA: SPIE.8. Roberts TM and Talebzadeh M. Acoustic emission mon-

itoring of fatigue crack propagation. J Constr Steel Res

2003; 59: 695–712.9. Pollock AA. Acoustic emission—2. Acoustic emission

amplitudes. Non-Destruct Test 1973; 6: 264–269.10. Chang H, Han EH, Wang JQ, et al. Acoustic emission

study of fatigue crack closure of physical short and long

cracks for aluminum alloy LY12CZ. Int J Fatigue 2009;

31: 403–407.

11. Farhidzadeh A, Salamone S, Luna B, et al. Acoustic

emission monitoring of a reinforced concrete shear wallby b-value-based outlier analysis. Struct Health Monit

2013; 12: 3–13.12. Gagar D, Foote P and Irving P. A novel closure based

approach for fatigue crack length estimation using theacoustic emission technique in structural health monitor-

ing applications. Smart Mater Struct 2014; 23: 105033.13. Bhuiyan MY, Shen Y and Giurgiutiu V. Guided wave

based crack detection in the rivet hole using global analy-tical with local FEM approach. Materials 2016; 9: 602.

14. Connor ZM, Li W, Fine ME, et al. Fatigue crack initia-

tion and growth in riveted specimens: an optical and

acoustic microscopic study. Int J Fatigue 1997; 19:331–338.

15. Bhuiyan MY, Shen Y and Giurgiutiu V. Ultrasonic

inspection of multiple-rivet-hole lap joint cracks using

global analysis with local finite element approach. In:Kundu T (ed.) Health monitoring of structural and biolo-

gical systems 2016 SPIE proceedings, Las Vegas, NV, vol.

9805, pp. 1–15. SPIE Digital Library.16. Jingpin J, Bin W and Cunfu H. Acoustic emission source

location methods using mode and frequency analysis.

Struct Control Hlth 2008; 15: 642–651.17. Kim KY and Sache W. Characteristics of an acoustic

emission source from a thermal crack in glass. Int J Fract

1986; 31: 211–231.18. Gorman MR and Prosser WH. Application of normal

mode expansion to acoustic emission waves in finite

plates. J Appl Mech 1996; 63: 555–557.19. Buttle DJ and Scruby CB. Acoustic emission source loca-

tion in fiber reinforced plastic composites. J Acoust Emiss

1988; 7: 211–223.

20. Zelenyak A, Hamstad M and Sause M. Modeling of

acoustic emission signal propagation in waveguides. Sen-sors 2015; 15: 11805–11822.

21. Prosser WH, Hamstad MA, Gary J, et al. Finite element

and plate theory modeling of acoustic emission wave-

forms. J Nondestruct Eval 1999; 18: 83–90.22. Sause MGR and Richler S. Finite element modelling of

cracks as acoustic emission sources. J Nondestruct Eval

2015; 34: 4.23. McLaskey GC and Glaser SD. Acoustic emission sensor

calibration for absolute source measurements. J Nondes-

truct Eval 2012; 31: 157–168.24. Abdelrahman M, ElBatanouny MK and Ziehl PH.

Acoustic emission based damage assessment method for

prestressed concrete structures: modified index of dam-age. Eng Struct 2014; 60: 258–264.

25. Abouhussien AA and Hassan AAA. Detection of bond

failure in the anchorage zone of reinforced concrete

beams via acoustic emission monitoring. Smart Mater

Struct 2016; 25: 075034.26. Andreykiv O, Skalsky V, Serhiyenko O, et al. Acoustic

emission estimation of crack formation in aluminium

alloys. Eng Fract Mech 2010; 77: 759–767.27. Giurgiutiu V. Embedded NDE with piezoelectric wafer-

active sensors in aerospace applications. J Mater 2003;

55: 1–5.

8 Structural Health Monitoring

28. Giurgiutiu V, Zagrai A and Bao J. Piezoelectric wafer

embedded active sensors for aging aircraft structural

health monitoring. Struct Health Monit 2002; 1: 41–61.29. Yu L, Santoni-Bottai G, Xu B, et al. Piezoelectric wafer

active sensors for in situ ultrasonic-guided wave SHM.

Fatigue Fract Eng M 2008; 31: 611–628.30. Bao J, Poddar B and Giurgiutiu V. Identifying fatigue

crack geometric features from acoustic emission signals.

In: Kundu T (ed.) Health monitoring of structural and bio-

logical systems 2016 SPIE proceedings. Las Vegas, NV,

vol. 9805, pp. 9805J1–9805J8. SPIE Digital Library.31. Sagar RV and Prasad BKR. A review of recent develop-

ments in parametric based acoustic emission techniques

applied to concrete structures. Nondestruct Test Eva 2012;

27: 47–68.32. Dutta D, Sohn H, Harries P, et al. A nonlinear acoustic

technique for crack detection in metallic structures. Struct

Health Monit 2009; 8: 251–262.

33. Kim G, Johnson DR, Semperlotti F, et al. Localization

of breathing cracks using combination tone nonlinear

response. Smart Mater Struct 2011; 20: 055014.34. Sholeh K, Vafai A and Kaveh A. Online detection of the

breathing crack using an adaptive tracking technique.

Acta Mech 2007; 188: 139–154.35. Trendafoilova I, Gorman DG and Manoach E. An inves-

tigation on vibration-based damage detection in circular

plates. Struct Health Monit 2009; 8: 291–302.36. Friswell MI and Penny JET. Crack modeling for structural

health monitoring. Struct Health Monit 2002; 1: 139–148.37. Hamstad MA, O’Gallagher A and Gary J. Modeling of

buried monopole and dipole sources of acoustic emission

with a finite element technique. J Acoust Emiss 1999; 17:

97–110.38. Shen Y and Giurgiutiu V. Effective non-reflective bound-

ary for Lamb waves: theory, finite element implementa-

tion, and applications. Wave Motion 2015; 58: 22–41.

Bhuiyan et al. 9