SHM442441 1. - University of South Carolina

http://shm.sagepub.com/Structural Health Monitoring

http://shm.sagepub.com/content/early/2012/06/27/1475921712442441The online version of this article can be found at:

DOI: 10.1177/1475921712442441

published online 28 June 2012Structural Health MonitoringGiola Santoni-Bottai and Victor Giurgiutiu

Damage detection at cryogenic temperatures in composites using piezoelectric wafer active sensors

Published by:

http://www.sagepublications.com

can be found at:Structural Health MonitoringAdditional services and information for

http://shm.sagepub.com/cgi/alertsEmail Alerts:

http://shm.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

What is This?

- Jun 28, 2012OnlineFirst Version of Record >>

at UNIV OF SOUTH CAROLINA on July 19, 2012shm.sagepub.comDownloaded from

http://shm.sagepub.com/

http://shm.sagepub.com/content/early/2012/06/27/1475921712442441

http://www.sagepublications.com

http://shm.sagepub.com/cgi/alerts

http://shm.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://shm.sagepub.com/content/early/2012/06/27/1475921712442441.full.pdf

http://online.sagepub.com/site/sphelp/vorhelp.xhtml


Article

Structural Health Monitoring

0(0) 1–16

� The Author(s) 2012

Reprints and permissions:

sagepub.co.uk/journalsPermissions.nav

DOI: 10.1177/1475921712442441

shm.sagepub.com

Damage detection at cryogenictemperatures in composites usingpiezoelectric wafer active sensors

Giola Santoni-Bottai and Victor Giurgiutiu

AbstractAn experimental evaluation of the structural health monitoring capability of piezoelectric wafer active sensors on com-posite structures at cryogenic temperatures is presented. The piezoelectric wafer active sensor–based electromechani-cal impedance and the pitch–catch methods were first qualified for cryogenic temperatures using piezoelectric waferactive sensor–instrumented composite specimens dipped in liquid N2. Subsequently, damage detection experimentswere performed on laboratory-scale composite specimens with (a) impact damage and (b) built-in Teflon patches simu-lating in service delaminations. Finally, a comprehensive damage detection test was performed on a full-scale specimensubjected to pressure and cryogenic temperature cycles. Based on these tests, we conclude that piezoelectric waferactive sensor–based structural health monitoring methods show promise for damage detection in composite materialseven in extreme cryogenic conditions. Recommendations for further work are also included.Keywords

KeywordsPiezoelectric wafer active sensor, damage detection, composite structures, cryogenic temperature, nondestructiveevaluation, structural health monitoring

Introduction

Structural health monitoring (SHM) is an emergingtechnology with multiple applications in the evaluationof critical structures. The goal of current SHM researchis to develop a monitoring system that is capable ofdetecting and identifying various damage modes duringthe service life of the structure with minimal humanintervention. Numerous SHM approaches have beenproposed1–4; they can be broadly classified into twocategories: passive SHM and active SHM. PassiveSHM methods, such as monitoring strain, acceleration,and acoustic emission, are relatively mature methodsbut have limited utility because they can only infer thepresence of damage from the passive measurements butcannot directly interrogate the structure to detect thedamage. Active SHM methods are currently of moreinterest because of their ability to interrogate a struc-ture and detect damage presence, location, and inten-sity. One of the promising active SHM methods utilizesarrays of piezoelectric wafer active sensors (PWASs)bonded to a structure that can both transmit andreceive ultrasonic elastic waves to achieve damage

detection.5–7 When used to interrogate thin plate struc-tures, PWAS transducers are effective Lamb wavetransducers. The PWAS transducers couple their in-plane motion with the Lamb wave particle motion onthe material surface. The in-plane PWAS motion isexcited by the applied oscillatory voltage through thed31 piezoelectric coupling. Optimum excitation anddetection happen when the PWAS length is an oddmultiple of the half wavelength of particular Lambwave modes. The principle of wave generation throughPWAS is fundamentally different from that of conven-tional ultrasonic transducers. Conventional ultrasonictransducers act through surface tapping, applyingvibrational pressure to the object’s surface. PWASs, onthe other hand, act through surface pinching and are

University of South Carolina, Columbia, USA

Corresponding author:

Victor Giurgiutiu, Department of Mechanical Engineering, University of

South Carolina, 300 Main St., Columbia, SC 29208, USA

Email: [email protected]



strain coupled with the object’s surface. This allowsPWAS to have a greater efficiency in transmitting andreceiving ultrasonic guided wave compared to conven-tional ultrasonic transducers.

Due to the increased use of composite materials innumerous types of structures, particularly in aerospacestructures, it is important to prove that an active SHMsystem is capable of reliably detecting composite mate-rial damage. For example, one of the most troublingforms of damage in laminated composites is the low-velocity impact damage. This type of damage may leaveno visual traces but can generate subsurface delamina-tions, which may significantly reduce the structuralstrength. Internal delaminations in composite struc-tures may also result from mechanical and thermalcycling; if not detected at an early stage, these delami-nations may lead to a catastrophic failure of the com-posite structure.

The present article investigates the use of PWAS togenerate and sense ultrasonic guided waves to detectdamage presence in composites. Section ‘‘State of theart’’ reviews the state of the art in PWAS-based damagedetection and the different methods being used. Section‘‘Cryogenic operability and survivability experiments’’presents operability and survivability experiments thatprove that PWAS-based SHM methods can operate onspecimens subjected to cryogenic temperature (CT).Section ‘‘Damage detection experiments on laboratoryspecimens’’ demonstrates how a PWAS-based SHMsystem can detect low-velocity impact damage andbuilt-in delaminations at both room temperature (RT)and CT. Section ‘‘Damage detection on full-scale com-posite specimen subjected to cryogenic and pressurecycling’’ presents results of a comprehensive test inwhich a full-scale composite specimen instrumentedwith PWAS transducers was subjected to pressure andCT cycling. This test showed that a PWAS-based SHMsystem is able to withstand and survive cryogenic ther-mal cycles under cycling pressure loading while per-forming meaningful damage detection tasks.

State of the art

Significant research effort has been invested in studyingthe possibility of using in situ ultrasonic nondestructiveevaluation (NDE) methods for the development ofSHM systems.1–12 Wave propagation methods wereused for detection of cracks, corrosion, and disbonds instiffened metallic panels. Also, the ability to detectcracks under bolts and rivets was investigated. It wasfound that successful damage detection can be achievedusing wave propagation methods (pitch–catch andpulse–echo methods) as well as the electromechanical(E/M) impedance method.

As most of the space applications are movingtoward composite material, a new set of experimentswas determined to validate the SHM system for spaceapplication on composite structures. The structuralcomponents for space applications are subjected tohigh loads and extreme low temperatures, that is,T = �1858C (�3008F).

Pitch–catch method

Pitch–catch method (Figure 1(a)) can be used to detectstructural changes that take place between a transmit-ter transducer and a receiver transducer. The detectionis performed through the examination of the Lambwave amplitude, phase, dispersion, and time of flight incomparison with a baseline signal corresponding to a‘‘pristine’’ condition. Typical applications include (a)corrosion detection in metallic structures, (b) diffuseddamage in composites, (c) disbonds detection in adhe-sive joints, (d) delamination detection in layered com-posites, and so on. The pitch–catch method can also beused to detect the presence of cracks from the wave sig-nal diffracted by the crack.

Pulse–echo method

The use of Lamb wave pulse–echo methods withembedded PWAS follows the general principles of con-ventional Lamb wave NDE. A PWAS transducerattached to the structure acts as both transmitter anddetector of Lamb waves traveling in the structure. Thewave sent by the PWAS is partially reflected at thecrack. The echo is captured at the same PWAS acting

V2

Receiver Pitch-catch

Damaged region

V1

Transmi�er

(a)

Pulse-echo

Damage

Transmi�er-receiver

V1V2

(b)

(c)

i(t) = I sin(ωt +φ )

m(ω)PWAS transducer

v(t)=V sin(ωt)

F(t)

( )u t� c(ω)

k(ω)

Figure 1. PWAS-based in situ damage detection methods:(a) pitch–catch method, (b) pulse–echo method, and (c) E/Mmethod.PWAS: piezoelectric wafer active sensor; E/M: electromechanical.

2 Structural Health Monitoring 0(0)



as receiver (Figure 1(b)). For the method to be success-ful, it is important that a low-dispersion Lamb wave isused. The selection of such a wave is achieved throughthe Lamb wave tuning methods.5

E/M impedance method

The E/M impedance method is an active damage detec-tion technique complementary to the wave propagationmethods. E/M impedance method gives structuraldynamics identification at hundreds of kilohertz andlow megahertz.13 Because high frequency implies smallwavelength, PWAS-based E/M impedance spectro-scopy (EMIS) is able to detect subtle changes associ-ated with the presence of incipient damage, whichwould not be detected by conventional modal analysissensors that operate at much lower frequencies.

The E/M impedance principles are illustrated inFigure 1(c), where the left side represents the electricalpart (e.g. an impedance analyzer equipment) and theright side represents the structure offering to the interro-gating PWAS, a frequency-dependent mechanical impe-dance, Zstr(v) = ½�v2m(v) + ivc(v) + k(v)�=iv. Hence,the impedance measured by the impedance analyzer isa combination of the intrinsic PWAS impedanceZPWAS(v) = 1=ivC and structural impedance Zstr(v),that is

Z vð Þ= ZPWAS vð Þ 1� k231

Zstr vð ÞZPWAS vð Þ+ Zstr vð Þ

� �ð1Þ

where k231 is the E/M coupling coefficient of the PWAS

material. The E/M impedance method is applied byscanning a predetermined frequency range in the hun-dreds of kilohertz band and recording the compleximpedance spectrum. By comparing the impedancespectra taken at various times during the service life ofa structure, meaningful information can be extractedpertinent to structural degradation and the appearanceof incipient damage. Previous work (e.g. Ref. 14) hasshown that the E/M impedance method can be veryeffective in detecting incipient damage in aerospace-likestructures.

PWAS–Lamb wave tuning

Lamb wave theory is extensively documented in a num-ber of textbooks.15–17 The Lamb wave equations for anisotropic media can be expressed through two potentialfunctions and the longitudinal and shear wave velocitycharacteristic of the material. The shear horizontal(SH) wave can be studied separately because it isdecoupled from the pressure (P) and shear vertical (SV)waves. For the case of a plate with free boundaries, theP and SV waves are coupled and their interaction

generates the Lamb wave. Lamb waves modes can besymmetric and antisymmetric with respect to the mid-surface. Closed-form analytical solutions exist and theLamb wave speeds can be found as transcendentalroots of the Rayleigh–Lamb characteristic equation.Lamb waves are dispersive because the wave speedchanges with frequency. For a given plate, a thresholdexists below which only the fundamental guided-wavemodes (A0, S0, and SH0) exist. For laminated compo-site plates, closed-form solutions do not exist and thesolution to the problem has to be obtained numericallyusing the transfer matrix, the global matrix, the stiff-ness matrix, or other approaches.18–24

Commonly used Lamb wave transducers for ultraso-nic NDE are piezoelectric wedge transducers; the wedgeangle and transducer frequency can be designed toexcite particular Lamb wave modes as needed for struc-tural interrogation. These wedge transducers are bulkyand are not appropriate for aerospace SHM applica-tions. PWAS transducers are much smaller than wedgetransducers and can be used for SHM applications, butthey are broadband and hence excite all the Lamb wavemodes present at a given frequency-thickness product.The simultaneous presence of two or more Lamb wavemodes increases the difficulty of the damage detectionprocess. The ability to excite a single wave mode wouldbe useful for effective damage detection. ThroughPWAS–Lamb wave tuning,25 one can selectively rejectcertain Lamb wave modes and amplify others by judi-ciously combining PWAS size and excitation frequency.The PWAS–Lamb wave tuning concept is based on thestrain response formula

ex x, tð Þ= �iat0

m

XjS

sin jSa� �NS jS

� �D0S jS� � ei jSx�vtð Þ

�iat0

m

XjA

sin jAa� �NA jA

� �D0A jA� � ei jAx�vtð Þ ð2Þ

where a is the PWAS half-size, j is the wave number,whereas superscripts A and S signify antisymmetricand symmetric modes. (The other Lamb wave analysisnotations, which can be found in Ref. 25 or in Ref. 26,p.327, equation (99), were not included here for brev-ity.) The PWAS–Lamb wave tuning process can bearticulated as follows: It is apparent from equation (2)that if the product ja equals an odd multiple of p=2,then the function sin ja takes extreme values and theassociated Lamb wave mode is maximized. Conversely,if the product ja equals an even multiple of p=2, thensin ja vanishes and that particular Lamb wave mode isrejected. This tuning concept was developed in Ref. 25for straight-crested Lamb waves; subsequently, it wasextended in Ref. 27 to circular-crested Lamb waves.Experimental confirmation of this tuning concept was

Santoni-Bottai and Giurgiutiu 3



also provided.25,27 Lamb wave tuning has also beenobserved with PWAS mounted on anisotropic compo-sites28; an extension of equation (2) to compositelaminates has recently been attained.29 Pitch–catchexperiments performed on composite materials usingone PWAS transmitter and several PWAS receiversplaced at various angular directions with respect to thecomposite material fiber orientation showed that incomposite materials, the tuning depends on wave pro-pagation direction.28 In another experiment, roundPWAS transducers (7 mm in diameter, 0.2 mm in thick-ness, American Piezo Ceramics APC-850 material)were bonded to a quasi-isotropic composite plate.Three waves were detected: S0, A0, and SH0. Rejectionof the A0 mode was observed at certain frequencieswhen the A0 packet almost disappears and the S0packet becomes maximized. However, the actuation ofa pure S0 mode in quasi-isotropic was not entirely pos-sible because the SH0 mode also appeared because ofthe intrinsic coupling in the quasi-isotropic compositematerial.

Cryogenic operability and survivabilityexperiments

The basic element used for damage detection in theseexperiments was a round PWAS transducer (7 mm indiameter, 0.2 mm in thickness, American PiezoCeramics APC-850 material). In a previous study,30 weproved that the piezoelectric material APC-850 and thePWAS transducers are able to retain their operationalabilities after exposure to cryogenic conditions. Toreproduce cryogenic conditions, we used containersfilled with liquid N2, which ensure a temperaturearound �2008C. Free PWAS resonators as well asPWAS transducers attached to metallic plates weresubmerged in the liquid N2 container, kept there for

10 min, and had their E/M impedance signature taken.Then, they were returned to RT and had the E/Mimpedance signature taken again. The process wasrepeated for 10 times with good results.30

In this study, we used a similar approach but withthe focus on composite material specimens. In addition,we also augmented the experiments with pitch–catchmeasurements besides the E/M impedance measure-ments. The experiments performed in this study arelisted in Table 1. The set of experiments were dividedinto two groups. The aim of first set of experimentswas to determine the CT operability of a PWAS-basedSHM approach for composite materials. The aim ofsecond set of experiments was to perform damagedetection with a PWAS-based SHM system operatingon composite specimens subjected to cryogenicconditions.

The adhesive layer between the PWAS and the struc-ture and solder material used to connect the PWASelectrodes to the electric wiring was carefully selected.The selected adhesive was the two-component VishayM-Bond AE-15 with an operating range down to�2708C (�4508F). The selected solder was indium-based 97In3Ag with good performance at CT. (Sn/Pbsolder was not used because it was previously found30

that it becomes brittle and fragile at CT.)A strip of unidirectional carbon/epoxy composite

material was used to test the PWAS pitch–catch oper-ability at CTs. Figure 2(a) shows the specimen with sev-eral PWASs attached to it; Figure 2(b) shows theexperimental setup.

Figure 3 shows the impedance signatures.30 Fromthe data recorded, it can be seen that the SHM systemimpedance curves did not change significantly aftersubmersion in liquid N2; the peak amplitude and theirrelative frequency location seem to remain the samethroughout the experiment.30

Table 1. Summary of experiments discussed in this article

Cryogenic operability and survivability experiments

Specimen Number of submersions Method

Unidirectional composite strip 10 Impedance1 Pitch–catch

Cryogenic damage detection experiments

Specimen Damage Environment Load

Lap joint Impact damage Room temperatureCryogenic temperature

Thick plate Delamination Room temperatureCryogenic temperature

Cylindrical specimen Unknown Cycling from room to cryogenic temperature Cyclic pressure




Figure 4 shows pitch–catch wave propagationbefore, during, and after submersion of the specimen inliquid nitrogen. The data collected showed that the

PWAS was able to send and receive signal at CTs.However, when the specimen was submerged in theliquid nitrogen, the amplitude of the wave packetdecreased. This behavior is consistent with studies per-formed elsewhere, which showed that fluid couplingcan reduce the Lamb wave propagation velocity andamplitude.31 In this study, it is apparent that the effectof submerging the specimen into liquid N2 is twofold:(a) effects that are due to fluid loading and Lamb waveleakage and (b) effects that are due to CT. In our arti-cle, the focus has been on the latter effects, that is, onhow the CT might influence the performance of thePWAS transducers. The data shown in Figure 4 indi-cate the following: (a) PWAS transducers are still activeat CTs and are able to transmit and receive guidedLamb waves into the specimen and (b) nonetheless, aspecimen with pitch–catch PWAS attached to itbehaves differently at RT after exposure to CT by sub-mersion in liquid N2 than before submersion. Figure 4shows that the wave amplitude after submersion islower than before submersion. (When submerged inliquid N2, the wave amplitude is even lower because ofthe additional wave leakage effect.) We are not sure atthis stage why this reduction in performance happensafter submersion in liquid N2 and return to RT, but webelieve that it is not due to a degradation in piezoelec-tric properties of the piezoceramic material becauseseparate tests performed on free PWAS transducersindicated that there is no performance degradation tothe piezoelectric material after cryogenic exposurethrough submersion in liquid N2.

30 Hence, we believethat the decrease in performance could be attributed toa degradation of the adhesive bond between the PWASand the specimen due to the differential coefficient ofthermal expansion (CTE) between the two materials.This aspect needs further investigation; we plan to do it

(b)(a)

Figure 2. Cryogenic survivability specimens: (a) carbon/epoxycomposite specimen having several PWAS transducers forpitch–catch wave propagation testing and (b) liquid N2

immersion setup for wave propagation testing showing thespecimen being readied for submersion in liquid N2.PWAS: piezoelectric wafer active sensor.

0

500

1000

1500

2000

2500

3000

5 10 15 20 25 30 35 40Frequency (kHz)

Impe

danc

e (Ω

)

Baseline 1 Cycle2 Cycles 3 Cycles4 Cycles 5 Cycles6 Cycles 7 Cycles8 Cycles 9 Cycles10 Cycles

Figure 3. Indication of survivability through resumption ofresonant properties after submersion in liquid nitrogen (PWAS,AE-15, RT).PWAS: piezoelectric wafer active sensor.

-25

-20

-15

-10

-5

0

5

10

15

20

25

0 20 40 60 80 100

Time (μs)

Am

plitu

de (m

V)

BaselineCryogenic1 Cycle

Figure 4. Wave propagation in composite for various thermalenvironments; comparison of a wave packet before, during, andafter submersion in liquid N2.




as soon as practically possible if further funding isavailable; we will report such new results in a futurecommunication.

In brief, the amplitude of the wave packets at RTafter submersion in liquid nitrogen did not return tothe original amplitudes. However, the amplitudes ofthe wave packets were greater than those of the speci-men submerged in liquid N2, because the wave excitedby the PWAS leaked into the liquid when the specimenwas submerged in liquid N2.

Damage detection experiments onlaboratory specimens

Composite specimen types

The PWAS ability to perform damage detection SHMon composite specimens under cryogenic conditionswas verified on three test specimen types. The first typeof specimen was a composite lap joint. The geometryof the specimen is shown in Figure 5. The specimenwas made of two composite plates of 305 3 230 mm2

(12$ 3 9$), and the plates were bonded together withan overlap of 100 mm (;4$). Two PWAS pairs wereinstalled as indicated in Figure 5. The following testswere performed: (a) damage detection at RT withPWAS pair 1 and (b) damage detection at CT withPWAS pair 2.

The second type of specimen was a thick compositeplate (Figure 6). The composite plate had a dimensionof 305 3 230 mm2 (12$ 3 9$). To simulate delamina-tion damage, the specimen was fabricated with 16Teflon patches of different sizes located insertedbetween various plies. The four patches of interest inthe experiments are marked as A, B, C, and D inFigure 6.

The four patches had different dimensions and dif-ferent thickness-wise locations (Table 2). Patches A and

B had a diameter of ;6.35 mm (0.25$), patch C had adiameter of ;19 mm (0.75$), and patch D had a dia-meter of ;13 mm (0.5$). Patches A and C were locatedclose to one of the surfaces while patches B and D werecloser to the mid-plane. Two experiments were per-formed on the specimen: (a) patch damage detection atRT and (b) patch damage detection at CT.

The third specimen type was a full-scale compositespecimen of cylindrical shape. The specimen wasthermomechanically cycled to CT of about �1858C(�3008F) and peak strains around 7000 microstrain.This section discusses the laboratory tests performedon the first and second specimen types. The tests per-formed on the third specimen type are discussed in thenext section.

Damage index calculation

The data collected in the experiments were analyzedusing a damage index (DI). The DI is a scalar numberthat reveals the difference between sequential readings(impedance spectrum or wave packets) due to the dam-age accumulation. Ideally, the DI should only capturethe changes in the signal that are directly related tothe damage presence, while not being influenced by

Figure 5. Lap joint: PWAS location (circles) and impactlocation.PWAS: piezoelectric wafer active sensor.

Figure 6. Schematic of thick composite specimen and locationof Teflon inserts (crosses A–D).

Table 2. Data collection notation

Step Patchname

Patchdimension(mm)

Patch location PWAS

1 04–0104–07

2 A 6.35 Close to surface 05–00C 19 05–08

3 B 6.35 Close to the mid-plane 03–02D 13 03–06

PWAS: piezoelectric wafer active sensor.




variations due to normal operation conditions (i.e. sta-tistical difference within a population of specimens,and expected changes in temperature, pressure, ambi-ent vibrations, etc.). To date, several simple DI formu-lae have been proposed, for example, root mean squaredeviation (RMSD), mean absolute percentage devia-tion (MAPD), and correlation coefficient deviation(CCD).26 In this work, we have used RMSD defined as

RMSD =

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPn

Re Sið Þ �Re S0ið Þf g2

Pn

Re S0ið Þf g2

vuuuut ð3Þ

where Si, i = 1, . . . , n, is the sampled signal and super-script 0 denotes the initial (baseline) state. Thus, thesymbol S0

i represents the baseline signal. The RMSDequation (3) yields a scalar number, which is zero ini-tially and increases as the signal becomes more andmore different. The advantage of using a simple DI for-mula is that no data preprocessing is needed, that is,the data obtained from the measurement equipmentcan be directly used to calculate the DI. The DI wasused to assess the severity of the damage in each testrun. Several readings were taken of the undamaged(baseline) condition and of each damaged configura-tion in order to assess the statistical spread. The result-ing DIs were plotted as ‘‘box plots’’ representing themean value plus/minus a standard deviation.

Impact detection on composite lap joint

The composite lap joint was subjected to impact dam-age. The impact damage was applied to the plate usingthe impactor shown in Figure 7.

The impactor had a hemispherical tip of ;13 mm indiameter and its weight was 391 g. The impactor weightcould be increased by adding barrels (Figure 7(b)).Each barrel weighted 500 g; a total of three barrelscould be added to the impactor.

The impactor used in our experiments had a totalweight of 1391 g. Two different impact damage stateswere created at each site by dropping the impactor fromdifferent heights. The first impact damage state had animpact energy level of 6 ft lb (;8 J) and hit the plate atabout 11 ft/s (3.35 m/s); the second impact damage

state had an impact energy level of 12 ft lb (;16 J) andhit the plate at about 16 ft/s (;5 m/s).

A total of 11 readings were taken in the undamagedbaseline configuration, 10 readings were taken after theimpact with energy level of 6 ft lb, and 10 readings wererecorded after the impact at 12 ft lb (see Table 3). Thefirst reading in the set of baseline readings was used asthe reference reading for the DI analysis.

The location of the PWAS and the impact sites onthe lap joint are shown in Figure 5. Two pairs of PWASwere installed on one side of the lap joint; each pair ofPWAS was bonded close to one of the edges of thejoint. The distance between the PWAS transducers was;200 mm (8$). We used the Lamb wave tuning methodto select the frequencies at which there was only onemode present. We found such conditions at 54 and 318kHz. The wave speed at 54 kHz was 1175 m/s, withwavelength of 21.8 mm. The wave speed at 318 kHzwas 3065 m/s, with wavelength of 10 mm.

Figures 8 to 11 show the plot of the RMSD valuesfor the baseline (step 1), the first damaged condition(step 2), and the second damaged condition (step 3).Figure 8 shows the plots of the DI values for two dif-ferent frequencies. Both low and high frequencies wereable to detect impacts at 6 and 12 ft lb with a signifi-cance level of 99%.

From Figure 8, we see that the absolute difference atlow frequency between steps 1 and 2 is much higherthan at high frequency. This indicates that lower fre-quencies are more sensitive to impact damage at RTthan the high-frequency excitations. The explanationfor this phenomenon is as follows. For impact damage,the structural damage is a delamination in the compo-site layup; the presence of the delamination changessignificantly the local flexural stiffness of the specimen.As indicated by the tuning principle, lower frequenciestend to excite more predominantly the pseudo-flexuralantisymmetric wave mode, which is more sensitive tothis type of damage than the pseudo-axial symmetricwave mode excited at higher frequencies. This explainsour observation that the low frequencies are more sen-sitive to this type of damage than the high frequencies.

Similar tests were performed using the PWAS systemfor damage detection on the lap joint at CTs. Two fre-quencies were selected through tuning: 60 and 318 kHz.The wave speed at 60 kHz was 1410 m/s with wave-length of 23 mm. The wave speed at 318 kHz was 3065

Figure 7. Impactor for impact damage: (a) base impactor with hemispherical tip, (b) barrel, and (c) impactor assembled.




m/s with wavelength of 10 mm. The data were collectedat temperatures below �1508C.

Figure 9 shows the plots of the DI values for the twodifferent frequencies. Both low and high frequencieswere able to detect the impacts with a significance level

of 99%. At CTs, the low and high frequencies show asimilar sensitivity to the impact damage. We do nothave at this stage an explanation to why the differencein detection sensibility between different frequencies, asobserved at RT, did not also appear at CTs.

In brief, the PWAS-based pitch–catch method wasable to detect impact damage in composite lap-jointspecimen at CTs with reasonable sensitivity.

Simulated delamination detection on thick platecomposite specimen

Nine PWAS transducers were installed on the thick spe-cimen (Figure 6). The experiments were performed to

Table 3. Summary of impact test parameters on lap-jointspecimen

Readings Energy ft lb (J) Velocity ft/s (m/s) Step

00–10 111–20 6 (~8) 11 (~3.35) 221–30 12 (~16) 16 (~5) 3

Figure 8. DI values for different damage levels (PWAS pair 00–02) at RTon the composite lap-joint specimen: (a) excitationfrequency at 54 kHz and (b) excitation frequency at 318 kHz.DI: damage index; PWAS: piezoelectric wafer active sensor; RT: room temperature.

Figure 9. DI values for different damage levels (PWAS pair 00–02) at CTon the composite lap-joint specimen: (a) excitationfrequency at 54 kHz and (b) excitation frequency at 318 kHz.DI: damage index; PWAS: piezoelectric wafer active sensor; CT: cryogenic temperature.




detect the presence of simulated delamination (Teflonpatches) inserted during specimen manufacture. Thereadings taken with PWAS pairs P04–P01 and P04–P07were used as baseline readings because there were nopatches in the wave path between these PWAS pairs.Two experiments were performed on the specimen: (a)patch detection at RT and (b) patch detection at CT.

The thick plate specimen was scanned at two differ-ent frequencies: 60 and 318 kHz. The wave speed at 60kHz was ;2680 m/s with the wavelength of ;45 mm.The wave speed at the high frequency of 318 kHz is notavailable.

Figure 10 shows the DI values for the thick plate spe-cimen at RT. From the analysis of the DI values, we seethat the low frequency is more sensitive to the patchthickness-wise location, especially when the patches arelarge (Figure 10(c)). The high frequency was more

sensitive to the patch presence, but it is not affected bythe patch dimension.

A similar experiment was conducted at CT.Readings were taken with temperatures below 2150�C.We used a different low frequency (75 kHz) becausethe CT caused a shift in the frequency of the maximumamplitude of the A0 mode. Figure 11 shows how DIindex changed with the different steps. We found thatthere was significant difference between the steps; thePWASs were able to detect the presence of the patches.From the DI values, we determined that both frequen-cies could detect the presence of the patches; however,at 318 kHz, there was greater sensitivity to patch pres-ence. The DI difference between steps 2 and 3 is alwayssmaller than between steps 1 and 2. We can concludethat the depth or the dimension of the patches did notaffect significantly the DI values at CT.

1 2 3

0.0

0.2

0.4

0.6

0.8

Step

DI

1 2 3

0.0

0.2

0.4

0.6

0.8

Step

1 2 3

0.0

0.2

0.4

0.6

0.8

1 2 3

0.0

0.2

0.4

0.6

0.8

DI

(b)(a)

(d)(c)

A

B

AB

C

D

C D

Figure 10. Thick composite plate at RT: (a) excitation frequency at 60 kHz, delaminations A and B; (b) excitation frequency at 318kHz, delaminations A and B; (c) excitation frequency at 60 kHz, delaminations C and D; and (d) excitation frequency at 318 kHz,delaminations C and D.DI: damage index.




Damage detection on full-scale compositespecimen subjected to cryogenic andpressure cycling

The cylindrical specimen used for this experiment was acontainer that could be filled with liquid nitrogen andthen emptied cyclically during the test. The specimencould also be cyclically subject to internal pressure ofabout 2500 lbf/in2.

A PWAS transducers network was installed on thespecimen as shown in Figure 12. The specimen areapermitted for sensor installation was limited to fourlongitudinal columns located at 90� increments aroundthe tank. Based upon this constraint, the PWAS trans-ducers were installed as shown in Figure 12: 16 pairs ofPWAS were installed along four rows at 90� incre-ments; electrical ground connections were installedclose to PWAS 16, PWAS 7, and PWAS 31.

The sensors were installed using a vacuum curingblanket. The adhesive used was the two-componentVishay AE-15, which has an operating temperaturerange down to �2708C (�4208F) and an elongationcapability of 2% (20,000 microstrain) at �1958C(�3208F), which is far more than the expected maxi-mum test strain (9000 microstrain). SWG 34 copper wirewas used to connect the PWAS; the solder was 97In3Ag,which is known to behave well at low temperatures. Thetest lasted 4 days; impedance readings for each PWASwere taken at the beginning and end of each day.Table 4 reports the scans taken and the test environmentfor both pitch–catch and impedance data collection.

There were a total of six cycles with strain above4000 mm/m and temperature below or equal to �1858C(�3008F). There were a total of seven cycles with strainabove 4000 mm/m, and there were a total of three cycleswith temperature below �1858C.

1 2 3

0.0

0.2

0.4

0.6

0.8

1.0

DI

1 2 3

0.0

0.2

0.4

0.6

0.8

1.0

1 2 3

0.0

0.2

0.4

0.6

0.8

1.0

Step

DI

1 2 3

0.0

0.2

0.4

0.6

0.8

1.0

Step

(b)(a)

(d)(c)

A B

A B

C D

C D

Figure 11. Thick composite plate at cryogenic temperature: (a) excitation frequency at 75 kHz, delaminations A and B;(b) excitation frequency at 318 kHz, delaminations A and B; (c) excitation frequency 75 kHz, delaminations C and D; and(d) excitation frequency 318 kHz, delaminations C and D.DI: damage index.




The status of each PWAS was visually checkedbefore the test started; it was found that all PWAStransducers were properly bonded to the specimen.Capacitance and impedance readings were taken tocheck the integrity of the PWAS transducers and wir-ing. The capacitance readings were all within the rangerequired. The impedance readings (Figure 13) showedPWAS 16 had a problem in the solder connection.Repair action was taken, and a new wire–PWAS con-nection was put into place.

After the recording of the last data (Reading 29 inTable 4), visual inspection was performed on the SHMsystem. Of the 32 PWAS transducers installed, five pre-sented a wire disconnection due to the solder disconnec-tion from the PWAS (Figure 14(a)); one was brokenwith the wire attached to the detached part of the

PWAS (Figure 14(b)); one was broken with the wirestill on the part of the sensor attached to the structure(see Figure 14(c)). The tube exploded a few pressurecycles after these images were taken.

Impedance readings

For each PWAS, six impedance readings were taken inthe frequency range of 1–500 kHz at each stage duringthe test. During postprocessing, plots of the real part ofE/M impedance were drawn. The real part of E/Mimpedance, Re(Z), measured at the PWAS terminalsreflects with fidelity the mechanical impedance of thestructure at the PWAS location.26

Hereunder, we report the graph of the six impedancereadings for PWAS 0. Impedance 0 corresponds to theimpedance taken before the test was started. Impedance1 was taken after one cycle at 5000 mm/m and one cycleat temperature below �1858C. There is no much differ-ence between these two readings. Impedance 2 wastaken when the specimen was at ambient temperatureand without load. The reading was taken the day afterimpedance 1 was taken; there was no loading change inbetween. Impedance 3 was taken with the specimen atabout �1698C and after three cycles at about 6000 mm/m. Impedance 4 was taken while the tank was fillingwith liquid nitrogen and without load. (This readingwas taken the next day after impedance 3 was taken).Impedance 5 was taken at the end of the test at RT andwith no load. (Note that the tube was brought to RTbetween impedance readings 3 and 4.) Between impe-dance curves 4 and 5, the SHM system has withstoodother three cycles with strain above 6000 mm/m.

Figure 15 indicates that after a few cycles at highmicrostrain levels, the E/M impedance spectrum revealsa new resonance frequency at about 100 kHz. This reso-nance frequency is evident only when the structure isnot subjected to load and critical temperatures (impe-dance curves 4 and 5). The same behavior could be seenin all the PWASs that were still active.

The impedance curves 1 and 3 are similar to the base-line curve 0. The impedance curves 4 and 5 are very dif-ferent because they show a new resonance frequency ataround 100 kHz. The resonance peak increases consid-erably between impedance curves 4 and 5. In a previousstudy,32 we noticed that new frequency resonance in theimpedance spectrum corresponds to structural changesand disbonds. We believe that this new resonance fre-quency can be attributed to a structural change in thetube after the application of extreme loading cycles.

Pitch–catch readings

Pitch–catch data were collected at two frequencies. Thefrequencies were determined through PWAS–Lamb

Figure 12. Sensors layout on the cylindrical specimen(projection view).

−1000

−900

−800

−700

−600

−500

−400

−300

−200

−100

00 50 100 150 200 250 300 350 400 450 500

P_00 P_01P_02 P_03

P_04 P_05P_06 P_07P_08 P_09P_10 P_11P_12 P_13P_14 P_15P_16 P_17P_18 P_19

P_20 P_21P_22 P_23P_24 P_25P_26 P_27P_28 P_29

P_30 P_31

PWAS 16

Frequency (kHz)

Im (z

)

P_00 P_01P_02 P_03P_04 P_05P_06 P_07

P_08 P_09P_10 P_11P_12 P_13P_14 P_15P_16 P_17P_18 P_19P_20 P_21P_22 P_23P_24 P_25P_26 P_27P_28 P_29

P_30 P_31

Figure 13. Imaginary part of the E/M impedance before the test.PWAS: piezoelectric wafer active sensor.




wave tuning. We found the maximum of the A0 modeat 45 kHz. For the S0 mode, we selected the frequency,165 kHz, at which the S0 mode achieves a maximumwhile the A0 and SH0 modes stay at low values.

We noticed that if the tube was under high loadingand at CT, then the pitch–catch analysis could not beperformed because the wave amplitude was too attenu-ated and not visible.

Figure 16 shows the pitch–catch results betweenPWAS 2 and PWAS 4 with at ambient temperatureand with the tube at rest. Due to the composite mate-rial properties and layup, the waves propagating longi-tudinally along the cylindrical specimen have higherwave speeds than those propagating circumferentiallyor obliquely. For this reason, the first wave packet inFigure 16 is very close to the initial burst. For pitch–catch along the circumferential direction (e.g. PWAS 2! PWAS 10), the wave speed was smaller by ;40%.For oblique pitch–catch (e.g. PWAS 2 ! PWAS 12),the wave speed decreased even more by ;46%. Hence,in these cases, the first wave packet was easier to iden-tify and process for DI calculation.

Since all the analyzed data appeared qualitativelysimilar, we discuss here only a couple of cases: (a) thecircumferential pitch–catch PWAS 2 ! PWAS 10(Figure 17), and (b) the oblique pitch–catch PWAS 2!PWAS 12 (Figure 18). Figure 17(a) shows the circum-ferential pitch–catch with the wave propagating fromPWAS 2 to PWAS 10. The wave amplitude of the A0

mode decreases as the number of fatigue and thermalcycles increases. While the wave amplitude of the base-line reading (Reading 0) is about 0.14 mV, the waveamplitude of the last reading (Reading 29) is about

Table 4. Test sequence for impedance

Reading Strain (mm/m) Temperature (�C) Reading Strain (mm/m) Temperature (�C)

Impedance 0 31 18–19 Few ;�18500–04 31 20 6000/7000 ;�16805 31/2183 21 ;�16906–07 .�178 Impedance 3 ;�16908 ;2400 ;�177 Impedance 4 ;�11509 Few ;�177 22 ;�12910 4000/5000 ;�188 23 6000/7000 ;�189Impedance 1 ;�154 24 Few ;�189Impedance 2 31 25 ;6000 ;�18911–12 31 26 ;300 ;�18813 Few ;�190 27 6000/7000 ;�18814 ;5000 ;�185 28 ;�18715 ;600 ;�185 Impedance 5 3116–17 ;6000 ;�185 29 31

Figure 14. Visual inspection of the broken PWAS or connection after the test (after Reading 29 in Table 4). (a) PWAS 1 broken, (b)PWAS 12 disconnected, and (c) PWAS 18 broken and PWAS 19 disconnected.PWAS: piezoelectric wafer active sensor.

0

50

100

150

25 75 125 175 225 275

Re (z

)

Frequency (kHz)

Impedance 5

Impedance 4

Impedance 3

Impedance 2

Impedance 1

Impedance 0

Figure 15. Impedance change collected with PWAS 0 duringthe thermal and pressure cycling experiment.PWAS: piezoelectric wafer active sensor.




0.06 mV. The variation in amplitude is also shown bythe change in DI values (Figure 17(b)). Similar resultswere obtained for oblique pitch–catch PWAS 2 !PWAS 12 (Figure 18).

The decrease in wave amplitude with number ofcycles was observed in all PWAS readings. This phe-nomenon could, in principle, be due to either (a) astructural change of the tube material or (b) a degrada-tion of the PWAS capability to transmit and/or receivesignals. However, we believe that the first assumptionapplies because both the pitch–catch signals and impe-dance readings indicate that the PWAS transducersseem to be functioning well. Hence, we believe that theobserved changes are due to degradation of the compo-site material under thermomechanical cycling, thusindicating that the PWAS transducers are able to per-form SHM functions at CT.

-0.1

-0.06

-0.02

0.02

0.06

0.1

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Reading 0Reading 11Reading 12Reading 28Reading 29

V (m

V)

Time (μs)

Figure 16. Pitch–catch at ambient temperature and zero loadfor PWAS 2 transmitter and PWAS 4 receiver (longitudinal wavepropagation) at different history times (A0 tuning at 45 kHz).PWAS: piezoelectric wafer active sensor.

-0.1

-0.06

-0.02

0.02

0.06

0.1

0.00015 0.0002 0.00025 0.0003

Reading 0Reading 11Reading 12Reading 29

V (m

V)

Time (μs)

DI

(a) (b)

0

0.2

0.4

0.6

0.8

1

R0 R11 R12 R29

Figure 17. Pitch–catch at ambient temperature and no load for PWAS 2 transmitter and PWAS 10 receiver (A0 tuning at 45 kHz):(a) wave packet (circumferential wave propagation) at different history times and (b) plot of DI values.PWAS: piezoelectric wafer active sensor.

PWAS: piezoelectric wafer active sensor; DI: damage index.

-0.06

-0.02

0.02

0.06

0.0002 0.00025 0.0003


V (m

V)

Time (μs)

0

0.2

0.4

0.6

0.8

1

R0 R11 R12 R29

DI

(a) (b)

Figure 18. Pitch–catch at ambient temperature and no load for PWAS 2 transmitter and PWAS 12 receiver (A0 tuning at 45 kHz):(a) wave packet (oblique wave propagation) at different history times and (b) plot of DI values.PWAS: piezoelectric wafer active sensor; DI: damage index.




Similar results were obtained for S0 tuning at 165kHz. Figure 19 presents the results for circumferentialpitch–catch PWAS 2! PWAS 10 at 165 kHz. Althoughthe wave amplitude is small, the noise-to-signal ratio isstill quite acceptable; hence, it was possible to recognizethe same amplitude reduction with the increase in thenumber of cycles as for A0 tuning at 45 kHz.

Summary and conclusions

This article has presented how PWAS can be used todetect damage in composite structures at CT. PWAStransducers are lightweight and inexpensive; they enablea large class of SHM applications such as embeddedguided-wave ultrasonics, that is, pitch–catch, pulse–echo, phased arrays, and high-frequency modal sensing,that is, the E/M impedance method. The focus of thisarticle has been on the ability of PWAS-based SHMsystem to survive and operate at low temperatures andsustain the thermal and mechanical cycling. This articleillustrates several experiments that prove how thePWAS transducers are effective for detecting multipletypes of damage (impact damage and delaminations) incomplex composite materials. In particular, results wereshown for damage detection of impact damage on acomposite lap-joint specimen at room and CT, anddetection of simulated delaminations in a compositethick plate at room and CT. These results indicate thata PWAS transducers would be effective and reliable forSHM of composite structures at cryogenic environmen-tal conditions.

In the last part of this article, a full-scale experimentwas presented. A full-scale composite specimen ofcylindrical shape was subjected to high-strain cycles

and CT cycles. The test finished with the specimenbursting under pressure after a considerable number ofhigh-pressure low-temperature load cycles. Most of thePWAS transducers proved to be working till the end ofthe test. However, they indicated damage initiation andprogression during the test. The impedance readingsshowed a new resonance peak at about 100 kHz, afterthe first two cycles at high strains. It is not possible toassess with certainty whether the new resonance is dueto a structural change in the tube or to a PWAS—bonddegradation. However, we believe that it was due to astructural change in the tube during the test because (a)a new resonance peak appeared in the impedance read-ings after several high load–low temperature cycles and(b) the pitch–catch readings showed consistently a DIincrease with the increase in the number of high load–low temperature cycles due to decrease in wave ampli-tude and appearance of phase shifts. The PWAS systemhas detected a structural change after four high strainscycles and three cryogenic cycles. The experimentconducted also shown good survivability of the PWAS-based system under both harsh environmental condi-tion (extreme cold temperatures, aging, and liquidcontact) and extreme loads.

This article has only presented a preliminary investi-gation aimed at determining whether PWAS transdu-cers can be used for damage detection at CT. In thisinvestigation, we have shown that a PWAS-based SHMsystem could be used with the specimen at CT andunder cycling between RT and CT. We have shown thatthe signal processing algorithms developed for damagedetection at RT can be also used for damage detectionat CT. However, further investigation is needed becausethis study has not yet addressed some fundamental

-0.05

-0.03

-0.01

0.01

0.03

0.05

9.00E-05 1.10E-04 1.30E-04 1.50E-04 1.70


V (m

V)

Time (μs)

0

0.2

0.4

0.6

0.8

1

1.2

R0 R11 R12 R29

DI

(a) (b)

Figure 19. Pitch–catch at ambient temperature and no load for PWAS 2 transmitter and PWAS 10 receiver (S0 tuning at 165 kHz):(a) wave packet (circumferential wave propagation) at different history times and (b) plot of DI values.PWAS: piezoelectric wafer active sensor; DI: damage index.




issues associated with CT. For example, are the piezoelec-tric behavior of PWAS, the mechanical performance ofcomposite structures, and the bonding between PWASand composite structures sensitive to temperature fluctua-tion down to CT? If yes, how much and how severe is thisdependence? In principle, if the changes are negligible,the algorithms developed at RT would be expected toalso function at CT. If the changes are not negligible,then how large are they? When did the changes happensubstantially on the route from RT down to CT? Theseare common issues that affect the applications of variousSHM techniques based on piezoelectric elements down toCTs (and also up to elevated temperatures). Such studiesneed to make the object of further research and futurecommunications.

Funding

This research was partially supported by NSF grant CMS0528873 and NASA STTR T7-02.

Acknowledgment

The authors thank program director Dr Shih Chi Liu.

References

1. Doebling SW, Farrar CR, Prime MB, et al. Damage

identification and health monitoring of structural and

mechanical systems from changes in their vibration char-

acteristics: a literature review. Report LA-13070-MS, Los

Alamos, NM: National Laboratory, 1996.2. Sohn H, Farrar CR, Hemez FM, et al. A review of struc-

tural health monitoring literature form 1996–2001. ReportLA-13976-MS, Los Alamos, NM: National Laboratory,

2004.3. Farrar CR, Sohn H, Hemez FM, et al. Damage prog-

nosis: current status and future needs. Report LA-14051-

MS, Los Alamos, NM: National Laboratory, 2004.4. Rose JL and Soley L. Ultrasonic guided waves for the

detection of anomalies in aircraft components. Mater

Eval 2000; 50(9): 1080–1086.5. Giurgiutiu V. Lamb wave generation with piezoelectric

wafer active sensors for structural health monitoring. In:SPIE 2003 International Symposium on Smart Structures

and Materials and Nondestructive Evaluation for Health

Monitoring and Diagnostics, San Diego, CA, 3–6 March

2003, paper no. 5056–17. Society of Photo-Optical

Instrumentation Engineers, Bellingham, WA.6. Giurgiutiu V, Zagrai A, Bao JJ, et al. Active sensors for

health monitoring of aging aerospace structures. Int J

COMADEM 2003; 6(1): 3–21.7. Giurgiutiu V, Zagrai A and Bao J. Damage identification

in aging aircraft structures with piezoelectric wafer active

sensors. J Intel Mat Syst Str 2003; 15(9–10): 673–687.8. Cuc A, Tidwell Z, Giurgiutiu V, et al. Non-destructive

evaluation (NDE) of space application panels using

piezoelectric wafer active sensors. In: Proceedings of

IMECE2005, Orlando, FL, 5–11 November 2005, paper

no. IMECE2005-81721. American Society of Mechanical

Engineers, New York, NY.9. Kessler SS, Spearing MS and Soutis C. Damage detection

in composite materials using Lamb wave methods. Smart

Mater Struct 2002; 11(2): 269–278.10. Saravanos DA, Birman V and Hopkins DA. Detection

of delaminations in composite beams using piezoelectric

sensors. In: Proceedings of the 35th structures, structural

dynamics and materials conference (AIAA), 18–20 April,

1994, Hilton Head, SC.11. Keilers CH and Chang FK. Identifying delamination in

composite beams using built-in piezoceramics: part I—

experiments and analysis. J Intel Mat Syst Struct 1995; 6:

649–663.12. Zhongqing S and Ye L. Lamb wave propagation-based

damage identification for quasi-isotropic CF/EP compo-

site laminates using artificial neural algorithm: part II—

implementation and validation. J Intel Mat Syst Struct

2005; 16: 664–672.13. Giurgiutiu V and Zagrai AN. Embedded self-sensing

piezoelectric active sensors for online structural identifi-

cation. J Vib Acoust 2001; 124: 116–125.14. Giurgiutiu V and Zagrai A. Damage detection in thin

plates and aerospace structures with the electro-

mechanical impedance method. Struct Health Monit

2005; 4(2): 99–1183.15. Achenbach JD. Wave propagation in elastic solids, North-

Holland series in applied mathematics and mechanics,

vol. 16. Amsterdam, The Netherlands: Elsevier Pub.,

1973.16. Graff KF. Wave motion in elastic solids. Mineola, NY:

Oxford University Press, 1991.17. Viktorov IA. Rayleigh and Lamb waves. New York: Ple-

num Press, 1970.18. Thomson WT. Transmission of elastic waves through a

stratified solid medium. J Appl Phys 1950; 21: 89–93.19. Haskell NA. The dispersion of surface waves on multi-

layered media. Bull Seismol Soc Am 1953; 43 (1): 17–34.20. Nayfeh AH. The general problem of elastic wave propa-

gation in multilayered anisotropic media J Acoust Soc

Am 1991; 89(4): 1521–1531.21. Nayfeh AH. Wave propagation in layered anisotropic

media with application to composites. Amsterdam: Else-

vier, 1995.22. Kausel E. Wave propagation in anisotropic layered

media. Int J Numer Methods Eng 1986; 23: 1567–1578.23. Wang L and Rokhlin SI. Stable reformulation of transfer

matrix method for wave propagation in layered anisotro-

pic media. Ultrasonics 2001; 39: 413–424.24. Rokhlin SI and Wang L. Stable recursive algorithm for

elastic wave propagation in layered anisotropic media:

stiffness matrix method. J Acoust Soc Am 2002; 112: 822.25. Giurgiutiu V. Tuned Lamb-wave excitation and detection

with piezoelectric wafer active sensors for structural

health monitoring. J Intel Mat Syst Struct 2005; 16(4):

291–306.26. Giurgiutiu V. Structural health monitoring with piezoelec-

tric wafer active sensors. New York: Elsevier Academic

Press, 2008, pp.239–502.




27. Raghavan A and Cesnik CES. Modeling of piezoelectric-based Lamb-wave generation and sensing for structuralhealth monitoring. In: Proceedings of SPIE 2004 Interna-

tional Symposium on Smart Structures and Materials and

Nondestructive Evaluation for Health Monitoring and

Diagnostics, 14–18 March 2004, San Diego, CA, paperno. 5391–42, pp. 419–430. Society of Photo-OpticalInstrumentation Engineers, Bellingham, WA.

28. Bottai GS and Giurgiutiu V. Lamb wave interactionbetween piezoelectric wafer active sensors and hoststructure in a composite material. In: 5th IWSHM5, Stan-ford University, CA, 12–14 September 2005, pp.1777–1784. DEStech Pub., Lancaster, PA.

29. Santoni-Bottai G. Fundamental studies in the Lamb-wave

interaction between piezoelectric wafer active sensor and

host structure during structure health monitoring. PhD Dis-

sertation, Department of Mechanical Engineering, Uni-

versity of South Carolina, Columbia, SC, 2010.30. Lin B, Giurgiutiu V, Pollock P, et al. Durability and sur-

vivability of piezoelectric wafer active sensors on metallic

structures. AIAA J 2010; 48(3): 635–643.31. Cheeke JDN, Shannon K and Wang Z. Loading effects

on A0 Lamb-like waves in full and partially filled thin-

walled tubes. Sens Actuators B 1999; 59: 180–183.32. Giurgiutiu V, Harries K, Petrou M, et al. Disbond detec-

tion with piezoelectric wafer active sensors in RC struc-

tures strengthened with FRP composite overlays.

Earthquake Eng Eng Vib 2003; 2(2): 213–223.




SHM442441 1. - University of South Carolina

Documents

Transcript of SHM442441 1. - University of South Carolina