Acoustic emission monitoring of bridges: Review and case studies

11
Engineering Structures 32 (2010) 1704–1714 Contents lists available at ScienceDirect Engineering Structures journal homepage: www.elsevier.com/locate/engstruct Acoustic emission monitoring of bridges: Review and case studies Archana Nair, C.S. Cai * Department of Civil and Environ. Engineering, Louisiana State University, Baton Rouge, LA 70803, United States article info Article history: Received 12 January 2008 Received in revised form 11 December 2009 Accepted 3 February 2010 Available online 6 March 2010 Keywords: Acoustic emission Bridge monitoring Bridge test Non destructive evaluation abstract This paper gives a brief review of the acoustic emission technique and its applications to bridge health monitoring. Emphasis is given to the discussion of available techniques of AE data processing, both qualitative and quantitative. An assessment of the statistical quantitative analysis technique, intensity analysis, is illustrated through two case studies. This technique of damage quantification is applied to AE data collected from two genres of bridges in Louisiana: a prestressed concrete slab-on-girder bridge and a steel bridge with a concrete deck. Although there were limitations concerning the number and type of sensors used, much information was collected and useful inferences were made that may help better diagnose the health of bridges monitored in the future using this technique. © 2010 Elsevier Ltd. All rights reserved. 1. Introduction The current state of bridges in the United States calls for the implementation of a continuous bridge monitoring system that can aid in timely damage detection and help extend the service life of these structures. A typical monitoring system would be one which enables non-invasive, continuous monitoring of the structure. The passive nature of the acoustic emission (AE) evaluation technique makes it an ideal choice to serve this purpose. Although the technique has been successfully used for decades for damage detection in other fields, its potential in bridge monitoring has not yet been fully exploited. Be it for quality control of bridges under construction or struc- tural integrity assessment and monitoring of existing bridges, the versatility of nondestructive evaluation (NDE) justifies its use in these structures. Of the many passive NDE techniques available today, AE was found to be the most widely used for highway structural assessment [1]. AE testing is a powerful nondestructive testing tool for real time examination of the behavior of materials deforming under stress. Load conditions that exist in bridges have been known to cause materials like concrete and steel to emit en- ergy in the form of elastic waves due to various material-relevant damage mechanisms. These waves are picked up by sensors attached to the surface of the material. Further evaluation of the collected information gives us an overall picture as to the health of the bridge and helps prioritize repair and maintenance. This review primarily focuses on the role of AE in bridge mon- itoring. In the context of bridges, a few merits and limitations of * Corresponding author. E-mail addresses: [email protected] (A. Nair), [email protected] (C.S. Cai). the AE technique have also been listed. The first part mainly dis- cusses the basics of AE, covering the topics of equipment require- ments and advances, measurement methods, and various available data processing techniques. This is followed by a brief overview of the relevant research work completed to date, including both lab and field tests carried out on concrete, steel, and fiber reinforced polymer (FRP) bridges or bridge components. Future prospects of enhancing the capabilities of AE in bridge monitoring systems are also discussed briefly. Finally, the possibilities of one of the quan- titative processing techniques have also been illustrated through two case studies conducted by the writers. Although a limited amount of information regarding AE in bridge monitoring is available, previous reviews presented by au- thors such as Carter and Holford [2], Holford and Lark [3], and the ASNT NDT handbook [4] provide quite a comprehensive resource. Basics of AE pertaining to bridge monitoring are introduced below for the convenience of the readers. 2. Methodology of acoustic emission 2.1. Basics of acoustic emission AE is the class of phenomena whereby transient elastic waves are generated by the rapid release of energy from a localized source or sources within a material, or the transient elastic wave(s) so gen- erated (ANSI/[5]). Thus, an acoustic monitoring system essentially requires two integral components: a material deformation that be- comes the source, and transducers that receive the stress waves that are generated from the source. The schematic shown in Fig. 1 represents the general work- ing principle of an acoustic monitoring system. A developing flaw emits bursts of energy in the form of high frequency sound waves which propagate within the material and are received by sensors. 0141-0296/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.engstruct.2010.02.020

Transcript of Acoustic emission monitoring of bridges: Review and case studies

Page 1: Acoustic emission monitoring of bridges: Review and case studies

Engineering Structures 32 (2010) 1704–1714

Contents lists available at ScienceDirect

Engineering Structures

journal homepage: www.elsevier.com/locate/engstruct

Acoustic emission monitoring of bridges: Review and case studiesArchana Nair, C.S. Cai ∗Department of Civil and Environ. Engineering, Louisiana State University, Baton Rouge, LA 70803, United States

a r t i c l e i n f o

Article history:Received 12 January 2008Received in revised form11 December 2009Accepted 3 February 2010Available online 6 March 2010

Keywords:Acoustic emissionBridge monitoringBridge testNon destructive evaluation

a b s t r a c t

This paper gives a brief review of the acoustic emission technique and its applications to bridge healthmonitoring. Emphasis is given to the discussion of available techniques of AE data processing, bothqualitative and quantitative. An assessment of the statistical quantitative analysis technique, intensityanalysis, is illustrated through two case studies. This technique of damage quantification is applied to AEdata collected from two genres of bridges in Louisiana: a prestressed concrete slab-on-girder bridge anda steel bridge with a concrete deck. Although there were limitations concerning the number and typeof sensors used, much information was collected and useful inferences were made that may help betterdiagnose the health of bridges monitored in the future using this technique.

© 2010 Elsevier Ltd. All rights reserved.

1. Introduction

The current state of bridges in the United States calls for theimplementation of a continuous bridgemonitoring system that canaid in timely damage detection and help extend the service life ofthese structures. A typical monitoring systemwould be one whichenables non-invasive, continuous monitoring of the structure. Thepassive nature of the acoustic emission (AE) evaluation techniquemakes it an ideal choice to serve this purpose. Although thetechnique has been successfully used for decades for damagedetection in other fields, its potential in bridge monitoring has notyet been fully exploited.Be it for quality control of bridges under construction or struc-

tural integrity assessment and monitoring of existing bridges, theversatility of nondestructive evaluation (NDE) justifies its use inthese structures. Of the many passive NDE techniques availabletoday, AE was found to be the most widely used for highwaystructural assessment [1]. AE testing is a powerful nondestructivetesting tool for real time examination of the behavior of materialsdeforming under stress. Load conditions that exist in bridges havebeen known to cause materials like concrete and steel to emit en-ergy in the form of elastic waves due to various material-relevantdamage mechanisms. These waves are picked up by sensorsattached to the surface of the material. Further evaluation of thecollected information gives us an overall picture as to the health ofthe bridge and helps prioritize repair and maintenance.This review primarily focuses on the role of AE in bridge mon-

itoring. In the context of bridges, a few merits and limitations of

∗ Corresponding author.E-mail addresses: [email protected] (A. Nair), [email protected] (C.S. Cai).

0141-0296/$ – see front matter© 2010 Elsevier Ltd. All rights reserved.doi:10.1016/j.engstruct.2010.02.020

the AE technique have also been listed. The first part mainly dis-cusses the basics of AE, covering the topics of equipment require-ments and advances, measurementmethods, and various availabledata processing techniques. This is followed by a brief overview ofthe relevant research work completed to date, including both laband field tests carried out on concrete, steel, and fiber reinforcedpolymer (FRP) bridges or bridge components. Future prospects ofenhancing the capabilities of AE in bridge monitoring systems arealso discussed briefly. Finally, the possibilities of one of the quan-titative processing techniques have also been illustrated throughtwo case studies conducted by the writers.Although a limited amount of information regarding AE in

bridge monitoring is available, previous reviews presented by au-thors such as Carter and Holford [2], Holford and Lark [3], and theASNT NDT handbook [4] provide quite a comprehensive resource.Basics of AE pertaining to bridge monitoring are introduced belowfor the convenience of the readers.

2. Methodology of acoustic emission

2.1. Basics of acoustic emission

AE is the class of phenomena whereby transient elastic wavesare generated by the rapid release of energy froma localized sourceor sourceswithin amaterial, or the transient elasticwave(s) so gen-erated (ANSI/[5]). Thus, an acoustic monitoring system essentiallyrequires two integral components: amaterial deformation that be-comes the source, and transducers that receive the stress wavesthat are generated from the source.The schematic shown in Fig. 1 represents the general work-

ing principle of an acoustic monitoring system. A developing flawemits bursts of energy in the form of high frequency sound waveswhich propagate within the material and are received by sensors.

Page 2: Acoustic emission monitoring of bridges: Review and case studies

A. Nair, C.S. Cai / Engineering Structures 32 (2010) 1704–1714 1705

Signal

One or more Sensors

stimulus(force)

stimulus(force)

Electronics

Fig. 1. Principle of acoustic emission [6].

In general, acoustic emissions can be classified into primary andsecondary emissions. Primary emissions are those originating fromwithin the material of interest, while secondary emissions referto all other emissions generated from external sources. Acousticevent detection is closely related to the characteristics of stresswaves, such as wave mode, attenuation, effects of multiple paths,and source location algorithm criterion [3].Basically, there are two types of AE monitoring strategies that

can be adopted, global and local. A global monitoring helps assessan entire structure’s integrity, while local monitoring addresses agiven specific area of damage [2,7]. Based on the duration forwhichmonitoring is required, AE bridge monitoring may also be cate-gorized into long term and short term. Long term monitoring ismostly applicable in scenarioswhere the bridges are relatively newor require monitoring as a follow up to a regular bridge inspection.Short term monitoring is more specific, usually the outcome of ashort-term study to update the structure’s integrity status.The AE technique has both advantages and disadvantages. The

advantages of the AE technique in the context of bridgemonitoringmay be listed as:

1. Damage growth essentially generates AE and is the outcome ofthe load history experienced by the structure [8].

2. AE is applicable for local, global, remote, and continuous moni-toring purposes without hindering traffic over the bridge struc-ture.

3. Source detection and location algorithms have been improvedto a great extent, assuring reliable analyses.

4. Dynamics of the material are observable in real time due to thetechnological advancements made in acquisition systems.

The disadvantages of this method are:

1. Although the issue of background noise discrimination hasbeen fairly addressed; real-time separation requires severaltrial monitoring sessions and experienced personnel.

2. Quantitative AE analyses are still difficult for applications to ac-tual bridge structures.

3. Standardized procedures are not available for all types ofbridges, as most recommendations cater for bridges underunique conditions of loading, materials, etc.

The measurement of AE essentially involves three basic compo-nents: the generated AE wave, the detection equipment used tocapture AE signals, and the processing and interpretation of thecollected data. Understanding the propagation characteristics ofAE waves is vital in distinguishing meaningful data from the un-wanted information. AE is an elastic wave that usually has a broad-band frequency at the origin. Thus, a typical AE wave detected byan AE sensor is a combination of longitudinal, transverse, reflectedwaves, etc. [9]. The knowledge of wave modes and their charac-teristics is necessary to implement accurate source location algo-rithms. The details regarding the theoretical aspect of AE wavepropagation is addressed in numerous works (such as [3], etc.).Most traditional AE sensors utilized consist of piezoelectric ele-

ments undergoing transduction. Their primary function is to de-tect transient elastic mechanical waves and convert them into

Fig. 2. Common types of AE sensors (pacndt.com).

electrical AE signals. Today, a wide variety of these sensors areavailable commercially (Fig. 2). The appropriate transducersneeded for data acquisition are chosen based on the purpose andsensitivity required for the investigation. Most researchers rec-ommend the use of resonant sensors, as they are highly sensi-tive to typical AE sources. For bridge monitoring, unidirectionalsensors and sensors sensitive more to in-plane wave modes mayprove beneficial in differentiating AE sources in various bridgecomponents [2]. The sensors are usually mounted on to the struc-ture, by using adhesives such as epoxy and hot melt glue, or byholders. Preamplifiers available either separately, or integratedwith the sensor, are integral components that improve the signal-to-noise ratio. The signals that are received by the sensors arecollected, stored, and processed in a data acquisition system.Commercial systems provide customized software that facilitatesqualitative real-time assessment of the collected AE data.

2.2. Interpretation of AE signals

Both the substructure and superstructure of bridges exhibit typ-ical damage modes, which may include corrosion, cracking, andphysical damage due to impact, fire, or fatigue cracking. Whilemonitoring the superstructure, it must be assessed thoroughlyin critical sections such as shear zones, tension zones, bearingregions, corrosion prone areas, etc. [10].Prior to any monitoring procedure, it is essential to understand

all the factors that may influence the AE signature, such as trans-ducer sensitivity, background noise, etc. Special attention must bepaid to the attenuation and wave velocity properties of materials.Higher frequency sensors tend to exhibit greater attenuation withdistance in steelmembers [2], while in compositematerials such asconcrete and FRP the influence of attenuation is much greater andplays a crucial role in determining sensor placement and sourcelocation.Typically, the signals collected can be represented by charac-

teristic parameters such as amplitude, duration, etc., as shown inFig. 3. There are numerous qualitative as well as quantitative waysto interpret these signal parameters or waveforms. For example,parametric analysis of the AE signal resulted in evaluation criteriasuch as: (i) A concrete beam integrity (CBI) ratio, defined as the ra-tio of the load at the onset of new AE in a subsequent load cycle tothe maximum prior load [11] and (ii) Calm and load ratios of re-inforced concrete beams; where the calm ratio is the cumulativeAE activities ratio during the unloading process to the maximumof the last loading cycle and the load ratio is the ratio between theload at onset of AE to the prior load [12], etc.A material under load is known to emit acoustic waves only

after a primary load level is exceeded. This characteristic, namedthe ‘‘Kaiser’’ effect, was first investigated by Joseph Kaiser in 1950.This effect, portrayed along plot points 1-2-3 in Fig. 4, has beenshown to exist at 70 to 85% of ultimate strength in concrete mate-rial. Meanwhile, the Felicity effect (absence of the Kaiser effect) in

Page 3: Acoustic emission monitoring of bridges: Review and case studies

1706 A. Nair, C.S. Cai / Engineering Structures 32 (2010) 1704–1714

VoltsRiseTime

Time

Time

Energy

Amplitude

Threshold

Threshold Crossing

Counts

Duration

Fig. 3. A typical AE signal [13].

Fig. 4. Kaiser and Felicity effects [16].

composites has led to the use of the Felicity ratio in tracking dam-age progression in this material [14,15].Meanwhile, quantification by statistical analysis of parameters

gave rise to the use of Historic and Severity indices in assessingstructural members [17]. This technique has already been success-fully applied to FRP and metal piping system evaluations [18]. Thetechnique’s applicability to concrete bridges has previously beenreported byGolaski et al. [17]. A typical intensity analysis evaluatesthe structural significance of an AE event by tracking the changesover time of two indices known as:(a) The Historic index, which is defined as a measure of the

change in signal strength through the loading phase of the test, and(b) The Severity index, which is defined as the average signal

strength among the largest numerical values of the signal.The indices are calculated using the following formulas [19]:

H(I) =N

N − K·

N∑

i=K+1Soi

N∑i=1Soi

(1)

Sr =1J·

(J∑m=1

Som

)(2)

where, H(I) — Historic index; N — number of hits up to time t; Soi— signal strength of the ith hit ; K — empirically derived constant

1000

100

10

101

1

Historic index

Major

Follow-up

Intermediate

Minor

Insignificant

Sev

erit

y

Fig. 5. Typical intensity chart for FRP material [18].

based on material; Sr — Severity index; J — empirically derivedconstant based on material; Som — signal strength of the mth hit,where the order of m is based on the magnitude of the signalstrength.For concrete, K values are related toN by the relations:N ≤ 50,

K = 0; 51 ≤ N ≤ 200, K = N − 30; 201 ≤ N ≤ 500, K = 0.85N;and N ≥ 501, K = N − 75 as well as J values for N < 50, J = 0and N ≥ 50, J = 50 [20,17].For metals, K values are related to N by the relations: N ≤ 15,

K = 0; 16 ≤ N ≤ 75, K = N − 15; 76 ≤ N ≤ 1000, K = 0.8N;and N ≥ 1001, K = N − 200 as well as J values for N < 10, J = 0and N ≥ 10, J = 10 [20].These indices are evaluated from the signal strength data col-

lected by each sensor. The severity index and maximum valueof historic index is plotted on an intensity chart, which may bedivided into the zones of damage shown in Fig. 5 [18]. The de-pendence of the technique on the number of data points andempirically derived constantsmay be considered limitations of thetechnique.Quantitative analysis of AE signals is mostly carried out with

AE waveforms. Appropriate interpretations of the collected wave-forms may be done by subjecting them to any typical waveformprocessing tool. ‘‘b-value’’ analysis of AE signals was yet anotherquantitative analysis proposition put forward by Colombo et al.[21]. Although numerous techniques of AE data assessment havebeen proposed and proven useful in laboratory trials, very rarelyhave any of them been reported viable for practical onsite moni-toring.Though a standardized procedure is not yet available for mon-

itoring of all bridge structures, the ASTM E1932-02 [22] serves asa guide for local area short term monitoring of bridge structures.Recommendations for bridge and/or bridge component monitor-ing have been compiled by Lozev et al. [23] for steel bridge mem-bers, Yuyama et al. [11] for RC beam evaluation, and in 2001 theTexas Department of Transportation developed the procedure formonitoring prestressed concrete girders.

3. State-of-the-art of AE bridge monitoring

The primary goal of AE monitoring in structures is to detect,locate, and assess the intensity of damage [3]. Thus, indicators ofstructural damage such as cracks, corrosion, and delamination thatwarn of an impending failure have become the focal point of anyAE study. The exceptional features of thismonitoring technique arethat there is no issue of traffic interruption during in-service mon-itoring of the bridges and the use of sensors with small surface ar-eas eliminates any concern about the contact surface profile [1].Investigations into materials that constitute the civil infrastruc-ture, such as concrete and steel, took place much later com-pared to other fields where AE is a well-established standard NDEtechnique. In the following paragraphs AE monitoring researchconducted on conventional bridge component materials such assteel and concrete are discussed alongwith the newgeneration FRPbridge components.

Page 4: Acoustic emission monitoring of bridges: Review and case studies

A. Nair, C.S. Cai / Engineering Structures 32 (2010) 1704–1714 1707

3.1. Crack monitoring

Numerous laboratory studies have been conducted to demon-strate the ability of AE to detect cracks prematurely. Morton et al.[24], Holford et al. [25], Hamstad and McColskey [26], etc., havefocused on monitoring fatigue crack development and its correla-tion with AE activity in steel members. A summary of fracture AEin metals can be found in [27]. Similarly, cracking in concrete hadbeen the interest of Yuyama and Ohtsu [28] who primarily usedmoment tensor analysis to characterize fracture mechanisms inRC beams reinforced with reinforcing bars and fiber plastic sheets.They reported that the breakdown of the Kaiser effect occurredonce shear cracking started to set in, and high AE activity in un-loading phases implied serious damage. A detailed study of theAE waveforms revealed that signals produced as an outcome ofshear and flexural cracking had larger amplitudes and durationthanmicro cracking phases of damage [29]. Use of conventional AEinstrumentation has led researchers to come out with conflictingremarks with regards to trends observed in fiber reinforced poly-mer (FRP) fracture characteristics. A group of researchers, includ-ing Shippen and Adams [30], Gostautas et al. [15], etc. reportedthat matrix cracking in FRP produced low amplitude signals, whileValentin [31], Jamison [32], etc. claimed that matrix cracking wasthe higher amplitude source mechanism.

3.2. Bridge monitoring applications

In addition to the numerous laboratory studies conducted, AEtechnology has been used for source location and damage intensitypredictions in numerous field bridge testing applications. Bothshort termand long termmonitoring instanceswill be summarizedin the following paragraphs.One of the pioneering works in AE bridge monitoring was car-

ried out by Pollock and Smith in the 1970s [33]. They monitored aportable military bridge subjected to proof testing, and reportedamplitude distribution analysis and source location results. The1980s saw the advent of long-term continuous monitoring imple-mented on a bridge. The project was spearheaded by the DuneganCorporation andwas carried out for about 10months. The purposeof this study was to check out the feasibility of long-termmonitor-ing. The issue of background noise discrimination led researcherslike Miller et al. [34] to assess both time and frequency domainAE signals to distinguish various sources. Prine and Hopwood [35]contributed to this field using an AE weld monitoring system tolocate crack activity in steel bridges. The significance of usingguard sensors to eliminate irrelevant AE events was the outcomeof numerous studies carried out on steel bridges by the PhysicalAcoustics Corporation (PAC). Guidelines meant for local and globalmonitoring of steel bridges were developed thanks to the contri-butions of Pollock and Carlyle [36], Carter and Holford [2], Pullinet al. [37], etc.Reinforcement corrosion and the resulting cracking are consid-

ered the main damage mechanisms that require the need for long-term condition assessment of concrete bridges. Various universi-ties and highway agencies have begun working towards achiev-ing the goal of implementing the AE method for concrete bridges[38]. Work on concrete structures has been primarily carriedout by Yuyama and Ohtsu [28]. They applied AE to study frac-ture characteristics, quantifying micro fractures and evaluatingthe damage intensity in concrete. Attenuation trends in concretebridgeswere the subject investigated by researchers such as Landisand Shah [39] and Beck et al. [40]. The NDE Validation Center(NDEVC) in Virginia conducted acoustic emission tests on concretebridges from 1996–2000. They used the monitoring technique todetect cracks in bridges by passing high experimental overloads.

1000

100

10

101

1

Sev

erit

y

Historic index

A

B

C

D

E

Fig. 6. Intensity plot for a whole bridge [17].

These kinds of special overloads were used as they could not ob-serve any acoustic activity prior to that point. Drafting of a recom-mended practice in Poland, for testing reinforced and prestressedconcrete structures by AE, culminated from the research work thatwas carried out by Golaski et al. [17]. They reported qualitative re-sults from the testing of five different concrete bridges situated inPoland at different intensities of damage. Shown in Fig. 6 is thequantified AE result they plotted from testing a new prestressedconcrete bridge, wherein each dot plotted represents AE signalscharacterized by analogous features. Since all points lie in the Azone, implying no serious deterioration within monitored zones,the plot aptly portrays the structural health of the new bridge.In-service RC bridges were also assessed by Beck et al. [40] andPullin et al. [41], who reported that the reliability of the AE tech-nique of monitoring bridges was in need of an upgrade.Bridge cable monitoring using AE had been set up since the

1970s. The successful application of AE in monitoring prestressedstructures [42] inspired Paulson and Cullington [43] to adapt thetechnique for continuous monitoring of suspension and cable staybridge cables. From several trials, they concluded that AE moni-toring is indeed suitable for detecting and locating wire breaks incable structures. A similar prestressed concrete bridge applicationwas reported by Brevet et al. [44]. They observed the effectivenessof wire fracture monitoring, in a prestressed concrete bridge opento regular traffic, on cables that cannot be inspected otherwise. AE-based health monitoring approaches for bridge stay cables wereextensively studied by numerous researchers such as Rizzo and DiScalea [45], Kretz et al. [46], Fricker and Vogel [47], Li and Ou [48],Jin et al. [49], etc. Gaillet et al. [50] and Zejli et al. [51] assessedcable anchorages using AE.Monitoring of prestressed concrete bridges was also reported

by Vogel et al. [52]. Prestressed concrete structures are known tohave almost no cracks at their initial phase of service life; thus,Vogel et al. [52] suggested that AE might prove more beneficial inmonitoring new cracks that may develop during their service life.Owing to the various advantages possessed by Fiber Reinforced

Polymers (FRP) it has become an emerging alternative to the tra-ditional materials that constitute bridge components. Since thismaterial is still not conventionally used in bridges, the literatureavailable on AEmonitoring of bridge componentsmade of thisma-terial is limited. However, themethod has been used in various lab-oratory investigations of FRP bridge decks to study theAE signatureand make valuable correlations. One such study was conductedby Ziehl and Bane [8], who reported their qualitative approachto testing a sinusoidal sandwich FRP bridge deck. They deviseda cyclic load profile to enable study of acoustic events generated

Page 5: Acoustic emission monitoring of bridges: Review and case studies

1708 A. Nair, C.S. Cai / Engineering Structures 32 (2010) 1704–1714

wireless RF datatransmission

transmitter and receiver (internet)

Fig. 7. Wireless sensing of bridges using radio frequency transmission [60].

at load holds, and traced damage progression from variations ob-served in the Felicity ratio. Another successful qualitative assess-mentwas conducted byKalny et al. [53]. They evaluated the changein AE signature exhibited by a specimen, before and after repair,under static loading conditions. They concluded that AE activitywas clearly distinguishable prior to repair, and that pre-existingdamage detection was possible by observing AE activity trends.Historic and severity indices were the basis on which six full-scaleFRP bridge decks in both original and repaired conditions wereevaluated by Gostautas et al. [15]. Although a clear intensity grada-tion was not achieved for this unique instance, they reported thatthe intensity analysis was useful to identify the onset of damageand subsequent calculation of the Felicity ratio.

3.3. Advancements in AE equipment technology

Since there has been no one system that has been confirmed asan ideal bridge monitoring system, various issues with regards tolimited sensitivity of available sensors, practical difficulties facedduring onsite installation, and remote access capabilities have beeninvestigated over the years. The shortcomings observed while us-ing traditional sensors for structural monitoring have been ad-dressed with proposals for new generation AE sensors that aremuchmore compact, sensitive, and economically viable. The intro-duction of micro-electro mechanical systems (MEMS) AE sensorsby Ozevin et al. [54] is one such innovation. These sensors showpromising applications for use in bridge monitoring in conjunc-tion with artificial intelligence networks [55]. Similarly, fiber op-tics technology is also being explored to develop a new generationof AE sensors [56].Obviously, one of the key features desirable in a bridge mon-

itoring situation is remote monitoring. This ability for existingcommercial systems has already been incorporated by corpora-tions, such as PAC, which provide on-line remote web monitoringfacilities. The Local Area Monitoring (LAM) is one such AE mon-itoring instrument developed by PAC in collaboration with theFederal Highway Administration (FHWA). Stryk and Pospisisl [57]proposed developing amonitoring system that identifies rebar cor-rosion, a crucial concern in concrete bridges. A Canadian company,Pure Technologies Ltd., has developed ‘SoundPrint’, which locateswire breaks in prestressing tendons [58]. Vallen systems has intro-duced AMSY4 and AMSY-5 that had continuous sampling rates of10 MHz for the feature extraction required for real-time data pro-cessing [59]. Implementation of wireless AE sensors is yet anotherinnovation proposed by Grosse et al. [60]. Fig. 7 provides the basicconcept behind remote monitoring intended for AE using wire-less technology. Using this technology, along with performance-enhanced sensors based on MEMS, makes this technology moreeconomic for huge structures such as bridges [60].

16.77m 16.77m16.77m

Instrumented span

Fig. 8. Sketch of the tested prestressed concrete bridge.

In the following sections, the feasibility of intensity chartsto evaluate the health of bridges is explored by analyzing datacollected from two bridge sites in Louisiana. A prestressed concretebridge and a steel bridge under live load conditions were tested.Both qualitative and quantitative assessments of the AE datacollected from the bridge sites will be presented. Although thereare no standard intensity curves specific to reinforced/prestressedconcrete and steel materials, the general trends observed in anintensity plot, and the structural assessments that can be made,are reported here.

4. Case study of a prestressed concrete bridge

This typical prestressed concrete slab-on-girder bridge is lo-cated over the Cypress Bayou in District 61 on LA 408 East,Louisiana. It was built in 1984 and designed to be comprised ofthree straight simple spans each 16.77 m (55 ft) in length (Fig. 8).Each span was supported by 7 AASHTO Type II prestressed con-crete girders, with girders spaced 2.13 m (7 ft center to center)apart. The roadway had a width of 14.33 m (47 ft) and a bridgedeck of 203 mm (8 in) thickness. Each span has one intermediatediaphragm (ID) located at its mid-span which is not connected tothe deck. Only the third span of the bridge was instrumented.The bridge was tested for 3 consecutive days in Feb 2006. Both

static and dynamic live load tests were carried out using twosimilarly weighing dump trucks. All four of the acoustic sensorsused were R6I (55 kHz resonant frequency) in conjunction with aDiSP-16 outdoor acquisition work station, both manufactured byPhysical Acoustics Corporation (PAC).

4.1. Loading schedule and equipment setup

One of the main objectives of this test was to assess thenecessity of intermediate diaphragms in prestressed concretebridge structures. For this bridge test, acoustic emission monitor-ing was conducted as a supplementary NDE technique in an at-tempt only to assess whether there was any additional damageat the girder–diaphragm connection when subjected to live loads.Although other gauges such as strain gauges, accelerometers, and

Page 6: Acoustic emission monitoring of bridges: Review and case studies

A. Nair, C.S. Cai / Engineering Structures 32 (2010) 1704–1714 1709

(a) SR_T1Sh_P1. (b) SS_T1L1_P1.

(c) SS_T2L2_P1. (d) SR_T1L1_T2L2_P1.

Fig. 9. Truck position for various load cases.

Table 1Load case nomenclature.

Type of live load Meaning

SR Static rolling, truck speed< 5 mphSS Static stopping, truck mid axle located at mid-spanD30 Dynamic, number following designation

represents the speed of truck (mph)Truck designationT1 Truck 1T2 Truck 2Roadway designationL1 Lane 1L2 Lane 2Sh Shoulder laneLoad case repetitionP1 Pass 1P2 Pass 2

deflection gauges were an integral part of this test, the resultspertaining to the acoustic sensors alone will be discussed here.Since there were limitations on the choice of sensors and theirnumber, the sensor configurations were chosen based on criticalregions previously predicted from the finite-element model of thebridge.The live load tests were carried out with two dump trucks

weighing 271.8 kN (61.1 kips) for Truck 1 and 272.7 kN (61.3 kips)for Truck 2. Both trucks had a single front axle and two-axles tan-dem at the rear. The static wheel loads for the first, second, andthird axles of both trucks were about 40.0, 47.8, and 47.8 kN (9,10.75 and 10.75 kips), respectively.Since the test consisted of many load cases, a systematic label-

ing system was developed for identifying the details of each loadcase, as shown in Table 1. Thus, a load case named SR_T1Sh_P1willtranslate into a static rolling test case (SR, moving at a speed lessthan 5 mph) with truck 1 (T1) over the shoulder lane (Sh) on thefirst passage (P1). A few illustrations of load cases accompanied bytheir names and cross-sectional details of the monitored span areshown in Fig. 9.On the first day of testing, the sensors were placed on the

intermediate diaphragm around girder # 6 (G6 in Fig. 10). Theintention of this configuration was to detect and/or locate any sig-nificant changes in stress or the presence of damage at the girder–diaphragm connection region.The second and third day sensor array consisted of two sensors

(Sensors 1 and 2) being placed 0.61 m (2 ft) apart around the mid-span of girder # 4, and the other twowere placed on opposite facesof a section of the intermediate diaphragmclose to the same girder.The chosen sensor arrangements for both days of testing are shownin Fig. 11.The data acquisition was carried out with a convenient outdoor

DiSP-workstationunit placed at a location close to the bridge. Real-time monitoring was enabled by the available AEWin software

Sensor 3

Sensor 1 Sensor 2

Sensor 4

G5G6G7

Fig. 10. Sensor arrangement for test Day 1.

45.10mSensor 1

Sensor 2 Sensor 3

Sensor 4

GIRDER #4

Diaphragm

0.61 m

Fig. 11. Sensor locations for Day 2 and Day 3 (Elevation view).

along with the acquisition system. Prior to acquiring any actuallive load test data, pencil lead break (PLB) tests were carried outclose to all four sensors to ensure their sensitivity. This procedurealso helps in evaluating the attenuation properties in the regionof interest. The suitable sensor spacing is usually determined at adistance within which the AE amplitude attenuates to about 30 dB[12]. A suitable threshold level of 45 dB is also chosen at this stage,based on the background noise level existing at the bridge site.

4.2. Qualitative results

Customized qualitative results are generated by the providedAEWin software. Only the load cases that generated the mostacoustic activity on each day are discussed in the following section.The AE activity observed on Day 1, when two trucks were stati-

callymoving over the shoulder and lane 1 of the bridge, is indicatedin Fig. 12. The plot reveals that activities recorded by the sensorslocated at the lower part of the diaphragm (sensors 1 and 2 inFig. 10) generated more acoustic activities relative to the sensorsplaced close to the deck. This trend may be attributed to the stressgenerated at themonitored joint due to the relativemovement be-tween the discontinuous joint at the beam–diaphragm connectioncreated during loading. Since the observed signal amplitudes werelow, they are not associated with the presence of any live cracks.Thus, from the amplitude information shown in Fig. 13, one caninfer that the monitored region had no serious structural damage.

Page 7: Acoustic emission monitoring of bridges: Review and case studies

1710 A. Nair, C.S. Cai / Engineering Structures 32 (2010) 1704–1714

300

240

180

120

60

90

60

30

0

Eve

nts

X position 08

1624

32

40

Y position

Fig. 12. Events versus location of sensors forDay1 configuration (SS_T1Sh_T2L1_P1).

0 20 40

40

60

60

80

80

100

100

Time (sec)

Am

plit

ude

(dB

)

Fig. 13. Amplitude versus time of sensor 2 for load case SS_T1Sh_T2L1_P1.

Results for a single load case in the following paragraph are dis-cussed individually for each pair of sensors due to their locationon different parts of the bridge. The Day 2 amplitude plot seen inFig. 14 is generatedby the two truck load cases,where the first clus-ter of data pointswere generated during the trucks’ rolling phase ofthe test (load case SR_T1L1_T2L2_P1) and after a short time lapse(approximately 220 s), the second cluster of data points were gen-erated after the trucks were backed up and stopped at themidspanof the bridge (load case SS_T1L1_T2L2_P1). Careful examination ofthe amplitude plots in Fig. 14 also reveals the existence of a fewhigh amplitude events. This may have been caused by secondaryAE sources originating from the structure due to load effects.Although the sensor configuration on Day 3 was similar to the

previous day (Day 2), on this day the dynamic load case gener-ated themost acoustic activity among all the load cases. This resultcould have occurred because of the existence of the Kaiser effect inconcrete mentioned earlier. Since most of the acoustic signal am-plitudes lie in the 60 dB range, the activity may not be a result ofany crack-related damage. Relatively higher acoustic activity wasobserved at sensors 2 and 3, placed on the diaphragm, than theother sensors 1 and 4. This could be attributed to the presence ofdiscontinuities at the girder–diaphragm connection. Upon close vi-sual observation, the bridge girders appear to be in fairly good con-dition,with virtually crack-free surfaces. This condition is expectedfor prestressed concrete bridges. Thus, even though a few high am-plitude events were recorded, these may have been contributed tosecondary sources of AE such as relative displacement of the mon-itored regions due to load effects and concrete-reinforcement in-teractions at the interface. It was also noted that a better acousticresponse from the bridge was observable when the structure wassubjected to quasi-static loads rather than dynamic loads. A simi-lar observation was also reported by Golaski et al. [17]. A detailedanalysis of the AE waveforms can help to distinguish the variousAE sources (see Fig. 15).

0 50 100 150

40

200

60

250

80

350300 400

100

Time (sec)

Am

plit

ude

(dB

)

0 50 100 150

40

200

60

250

80

350300 400

100

Time (sec)A

mpl

itud

e (d

B)

Fig. 14. Amplitude versus timeplot for load case SR_T1L1_T2L2_P1/SS_T1L1_T2L2_P1(Top: sensors 1 and 4; bottom: sensors 2 and 3).

0 5 10 15

40

20

60

80

100

Time (sec)

Am

plit

ude

(dB

)

0 5 10 15

40

20

60

80

100

Time (sec)

Am

plit

ude

(dB

)

Fig. 15. Amplitude versus time plot for load case D40_T1L2_P1 (Top: sensors 1 and4; bottom: sensors 2 and 3).

4.3. Damage quantification

To get a better insight into the significance of the AE data col-lected, quantification of the data is attempted here using the inten-sity analysis technique. This method requires the accumulation ofAE data obtained from successive load cycles. This AE data is thenused to determine the indices given in Eqs. (1) and (2), which aresummarized in Table 2.The maximum historic index value and severity indices calcu-

lated for each load case shown in Table 2 are plotted on an intensity

Page 8: Acoustic emission monitoring of bridges: Review and case studies

A. Nair, C.S. Cai / Engineering Structures 32 (2010) 1704–1714 1711

Intensity chart for Day 1

11

10

10

Seve

rity SR_T1Sh_P1

SR_T1L1_T2Sh_P1SS_T1L1_T2Sh_P1

4

2

2

Historic Index

Fig. 16. Intensity chart for load cases of Day 1 (Numbers within the plot representsensor #).

Table 2Summary of results from intensity analysis.

Load case Ch H(I) Sr

Day 1

SR_T1Sh_P11 1.69 0.412 4.13 2.814 1.79 1.18

SR_T1L1_T2Sh_P1 1 1.72 0.72

SS_T1L1_T2Sh_P1 2 4.77 4.664 1.80 2.02

Day 2

SR_T1L1_P1

1 3.57 1.822 1.53 0.133 1.93 0.224 2.50 1.13

SR_T1L1_T2Sh_P1

1 3.62 2.812 2.54 2.113 3.34 2.764 3.28 1.92

D40_T1L1

1 3.62 3.312 2.08 2.203 2.49 2.864 3.28 2.23

Day 3

SR_T1L1_P1 1 1.68 0.38

SS-T1L1_P12 2.18 0.553 2.40 0.914 2.14 0.50

SR_T1L1_T2L2_P1

1 1.68 0.462 2.18 0.733 2.40 1.134 2.14 0.60

D40_T1L2_P1

1 1.68 0.792 2.35 0.953 2.40 1.424 2.25 1.04

chart for each day of testing, as shown in Figs. 16–18. For thesecond and third days of testing, results from sensors 1 & 4, lo-cated under the girder, are separated from those of sensors 2and 3, placed across the thickness of the diaphragm close to thegirder–diaphragm joint. There is no data for some sensors due toeither too small numbers or malfunctions.The dependence of this analysis technique on aminimumnum-

ber of data points inhibits the representation of data from everysensor used for monitoring on the intensity chart. Each intensitychart has been developed for each day of testing and consecutiveload cases, as the technique requires cumulative data assessment.Here again, the Day 1 results plotted in Fig. 16 show that sen-sor 2 seems to have acquired AE events of higher intensities thanall the other sensors. Pre-existing cracks at the girder–diaphragminterface around the two sides of the observed Girder #6 mighthave led to the generation of such acoustic activities. Incremental

Intensity Chart for Day2 (Ch# 1 & 4)

11

10

10 Historic index

Seve

rity SR_T1L1_P1

SR_T1L1_T2Sh_P1

D40_T1L1

4

1

4 1

1

4

Intensity chart for Day 2 (Ch# 2 & 3)

11

10

10Historic index

Seve

rity SR_T1L1_P1

SR_T1L1_T2Sh_P1

D40_T1L12233

Fig. 17. Intensity charts for load cases onDay 2 (Numberswithin the plot representsensor #).

Intensity Chart for Day 3 (Ch# 1 & 4)

0.1

1

10

Historic index

Seve

rity

SR_T1L1_P1 SS-T1L1_P1

SR_T1L1_T2L2_P1

D40_T1L2_P114

Intensity chart for Day 3 (Ch# 2 & 3)

0.1

1

10

Historic index

Seve

rity

SR_T1L1_P1 SS-T1L1_P1

SR_T1L1_T2L2_P1

D40_T1L2_P12

3

10

101

1

Fig. 18. Intensity charts for load cases onDay 3 (Numberswithin the plot representsensor #).

loading leads to consequent intensity points which are plotted to-wards the right corner of the chart.The intensity charts plotted for the second and third days

seem to correspond to the qualitative evaluationsmade previously.Higher loads generate AE with higher intensities, which in turnhelp reflect the intensity of crack-related damage in themonitoredstructure. From the plots shown in Figs. 16–18, we see that thepoints of lower loads plot to the left corner of the chart, while ahigher load causes the intensity point to shift towards the right endof the chart.

5. Case study of a steel bridge

The bridge that was monitored is located along highway LA-1 over the Intracoastal Waterway in Port Allen, Louisiana. The

Page 9: Acoustic emission monitoring of bridges: Review and case studies

1712 A. Nair, C.S. Cai / Engineering Structures 32 (2010) 1704–1714

Sensor Sensor 1

Sensor 3

Sensor 4

Column1.54m 2

Fig. 19. Acoustic sensor locations on the steel bridge.

Fig. 20. Oversize load on bridge.

bridge consists of multiple spans with varying span lengths. Thespan that was tested is 17.99 m (59) ft long with four steel girders(W36X182) supporting a concrete deck. The girders were spacedat 2.64 m (8 ft 8 in center to center). The girders were bolted toa cross-girder which provides support for them; the cross-girderwas also bolted to the columns. This steel bridge was tested underoverload conditions in July 2006.

5.1. Loading schedule and equipment setup

The objective of this test was to assess the structural behaviorof the monitored span when subjected to overloads. The plan wasto compare the acoustic data collected from normal traffic on thebridge to that of the overload passage. Since the highway structureis located near a port, the normal traffic also included heavy trucks.Potential damage regions such as the mid-span of the girders andbeam–column joints were chosen to be monitored. Thus, two ofthe sensors were placed under a girder around the mid-span andthe other two were located at the beam–column joint, as shown inFig. 19.The optimal threshold level for data acquisition was set at

40 dB. Under the chosen threshold level, no acoustic activity wasobserved in the absence of vehicles on the bridge. The oversizeload comprised of two trucks and two trailers, weighing a total of2401.9 kN (540 kips). The truck was 6.10m (20 ft) wide and 70.3m(230.5 ft) long. The truck’s configuration is shown in Fig. 20.

5.2. Qualitative results

AEmonitoringwas carried out for two phases of loading: duringnormal traffic and overload passage. Since stronger AE activity wasrecorded by sensors 3 and 4 located at the beam–column jointfor both loading phases, their cumulative hits are represented inFig. 21. Upon comparing the signal intensities obtained during bothphases of loading, the following observations may be made:

(a) The signal amplitudes under both loading conditions rarely ex-ceeded 60–70 dB. Events in this amplitude range are usuallynot associated with any significant structural damage.

(b) The signals obtained from sensors 3 and 4, located at thebeam–column joint, were much stronger in comparison with

0 20 40 60 80 100 120 140 160 180 200

400

200

0

600

800

1000

Time (sec)

Time (sec)

a

b

Hit

sH

its

400

200

0

600

800

1000

1200

0 20 40 60 80 100 120 140 160 180 200

Fig. 21. Cumulative AE hit rate — (a) normal traffic phase and (b) overload phase.

100

10

101

1

Sr

H(I)

3

3

4

Ch # 3

Ch # 4

Normal traffic

Normal traffic

Overlaod

Overlaod

Fig. 22. Intensity chart for acoustic activity from sensors 3 and 4.

the signals recorded by the sensors placed on the girder.Again, this observation can currently be attributed only as asource originating from some relative displacement betweenthe members (such as slip of connection bolts) since no physi-cal damagewas observed at thatmonitored joint after the test.

(c) As expected, the acoustic activity due to the overload is greaterthan the normal traffic, which consisted of both light vehi-cles and heavy trucks. However, the increase (Fig. 21(b)) wasnot significant enough to justify more scrutiny in assessing themonitored region.

5.3. Damage quantification

Intensity charts developed for metal piping systems [15] wereused for the quantitative analysis of data collected from the steelbridge. Upon analyzing the signal strengths obtained from this testin both loading conditions, only the data from sensor 3, located onthe beam–column joint, during both load conditions could be plot-ted on the intensity chart. All the other signal intensity values havea severity value below 1, and thus, were not represented in Fig. 22.Most of data points fell into the insignificant damage region, ex-cept the signal intensities from sensor 3 during overload. Although

Page 10: Acoustic emission monitoring of bridges: Review and case studies

A. Nair, C.S. Cai / Engineering Structures 32 (2010) 1704–1714 1713

a signal intensity value lying in zone C implies a defect that requiresa follow-up evaluation, here we cannot assign the high signal in-tensity to any actual defect, due to various local uncertainties suchas themovement of bolts, proximity to the beam–column joint, etc.in the monitored region.

6. Conclusions and comments

This paper presents a brief review of the research and technol-ogy prevalent in acoustic emission monitoring of bridges. Threematerials used in bridge construction: concrete, steel, and FRPhavebeen considered for this discussion of the research in bridge moni-toring. Currently available interpretations of the acquired AE data,both qualitative and quantitative, have also been discussed. A com-mendable effort is being made in the direction of improving AEsystems, addressing the practicality and economic issues of imple-menting the technique for monitoring purposes.All in all, the applications of AE in bridge monitoring reveal the

potential of this technique’s versatility. The technological advancesmade in recent years have made the method more suitable for on-sitemonitoring situations. Althoughmore researchmay be neededto implement the current ideas, the future looks promising for theapplication of this technology in efficient continuous bridge mon-itoring scenarios.The observations and results obtained from the acoustic emis-

sion data of two field test cases under live load conditionswere alsodiscussed in this paper. The following conclusions are drawn basedon the observations made through both qualitative and quantita-tive analyses of the collected AE data.1. The overall trend seen is that, even with the limited coverage ofthe structure, a credible amount of information was collectedand the analysis of this information gives an insight into thestructural response of the local area monitored under live loadconditions.

2. It may also be noted that almost all intensity points plotted onthe data charts for the prestressed concrete bridge had severityvalues below 10. This range is considered to represent insignifi-cant emissions inmost previously defined intensity charts; and,thus, leads one to infer that the monitored regions had not de-veloped any significant structural damage during the course ofour testing.

3. The use of intensity chartsmay help to better estimate the dam-age severity, although clearly marked zones of damage are notyet prescribed for certain materials such as concrete and steel.

4. Since the intensity analysis technique assesses cumulative AEdata over successive loads, continuedmonitoring can help tracethe health of a bridge.

The results obtained from both bridge sites seem to indicatethat the monitored regions had no real issues with their structuralintegrity. In spite of the practical challenges faced for use in thefield, continued efforts show that the technique has a promisingfuture in becoming an integral part of any structural health moni-toring system.

Acknowledgements

The authors would like to thank the Louisiana DOTD andLouisiana Transportation Research Center (LTRC) for making thisstudy possible. The Louisiana DOTD crew helped conduct thebridge field tests. Special thanks go toMr.Walid Alaywan andArturD’Andrea’s group. Many graduate students and visiting scholars atLSU also helped prepare and carry out the bridge test.

References[1] Rens KL, Wipf TJ, Klaiber FW. Review of nondestructive evaluation techniquesof civil infrastructure. J Perform Constructed Facil, ASCE 1997;11(4):152–60.

[2] Carter DC, Holford KM. Strategic consideration for AE monitoring of bridges:A discussion and Case study. INSIGHT - J British Inst NDT 1998;40(2):112–6.

[3] Holford KM, Lark RJ. In: Gongkang Fu, editor. Acoustic Emission testing ofbridges: Inspection and monitoring techniques for bridges and structures.Cambridge (UK): Woodhead Publishing Ltd; 2005. p. 183–215.

[4] American Society of Nondestructive Testing (ASNT). Nondestructive testinghandbook, third edition: vol. 6. Acoustic emission testing, Columbus, OH. 2005p. 1–25.

[5] ASTM E1316-07b. Standard terminology for nondestructive examinations.West Conshohocken (PA): ASTM international; 2007.

[6] Grosse CU.www.NDT.net—Editorial: Special issue on acoustic emission. 2002.[7] Carlos MF, Vahaviolos SJ, Cole PT, Halkyard T. Acoustic emission bridgeinspection/monitoring strategies. In: Structural materials technology IV — AnNDT conference. 2000. p. 179–83.

[8] Ziehl P, Bane WS. Nondestructive evaluation of fiber reinforced polymerbridges anddecks. FHWA/LA03/376. LA:Department of Civil and Environmen-tal Engineering, Tulane University. 2003.

[9] Kawamoto S, Williams RS. Acoustic emission and acousto-ultrasonic tech-niques for wood and wood-based composites: A review. Gen. tech. rep. FPL-GTR-134. US Department of Agriculture, Forest Service, Forest Products Labo-ratory Madison, WI. 2002.

[10] Ghorbanpoor A, Benish N. Non-destructive testing of highway bridges.Final report # 0092-00-15. Madison (WI): Wisconsin Department ofTransportation; 2003.

[11] Yuyama S, Okamoto T, Shigeishi M, Ohtsu M, Kishi T. A proposed standardfor evaluating structural integrity of reinforced concrete beams by acousticemission. In: Vahaviolos SJ, editor. Acoustic emission: Standards andTechnology update, ASTM STP 1353. West Conshohocken (PA): AmericanSociety for testing and materials; 1999.

[12] Ohtsu M, Uchida M, Okamoto T, Yuyama S. Damage assessment of reinforcedconcrete beams qualified by acoustic emission. ACI Struct J 2002;99(4):411–7.

[13] Huang M, Jiang L, Liaw PK, Brooks CR, Seeley R, Klarstrom DL. Using acousticemission in fatigue and fracture materials research. J Mater 1998;50(11).

[14] Ziehl P, Lamanna AJ. Monitoring of the Bonnet carre Spillway bridge duringextreme overload. LTRC project no. 03-6ST. LA: Department of Civil andEnvironmental Engineering, Tulane University; 2003.

[15] Gostautas RS, Ramirez G, Peterman RJ, Meggers D. Acoustic emissionmonitoring and analysis of glass fiber reinforced composites bridge decks. JBridge Eng ASCE 2005;10(6):713–21.

[16] Grandt AF. Fundamentals of structural integrity, damage tolerant design andnondestructive evaluation. NJ: John Wiley and Sons Ltd.; 2003. p. 426–8.

[17] Golaski L, Gebski P, Ono K. Diagnostics of reinforced concrete bridges byacoustic emission. J Acoust Emiss 2002;20:83–98.

[18] Committee on Acoustic Emission from Reinforced Plastics (CARP). Recom-mended practice for acoustic emission of fiberglass reinforced plastic resin(RP) tanks/vessels. New York: Composites Institute, Society of the Plastics In-dustry; 1987.

[19] Blessing JA, Fowler TJ, Strauser FE. Intensity analysis. In: Proc., 4th int symp.on acoustic emission from composite materials. Columbus (Ohio): AmericanSociety for Nondestructive testing; 1992.

[20] Chotickai P. Acoustic emission monitoring of prestressed bridge girders withpremature concrete deterioration. Masters thesis. Austin (Texas): Universityof Texas; 2001.

[21] Colombo IS,Main IG, FordeMC. Assessing damage of reinforced concrete beamusing b-value analysis of acoustic emission signals. J Mater Civil Eng, ASCE2003;15(3):280–6.

[22] ASTME1932-97. e1, Standard guide for acoustic emission examination of smallparts. West Conshohocken (PA): ASTM International; 2002.

[23] Lozev MG, Clemena GG, Duke JC, Sison MF, Horne MR. Acoustic emissionmonitoring of steel bridge members. Final report, FHWA/VTRC 97-R13.Charlottesville (VA): Virginia Transportation Research Council; 1997.

[24] Morton TM, Harrington RM, Bjeletich JG. Acoustic emissions of fatigue crackgrowth. Eng Fract Mech 1973;5:691–7.

[25] Holford KM, Davies HW, Sammarco A. Analysis of fatigue crack growth instructural steels by classification by acoustic emission signals. Eng Syst DesignAnal 1994;8:349–53.

[26] HamstadMA,McColskey JD. Detectability of slow crack growth in bridge steelsby acoustic emission. Mater Eval 1999;57(11):1165–74.

[27] Ono K. New goals for AE inmaterials research. In: Acoustic emission — Beyondthe millennium. UK: Elsevier; 2000. p. 87–190.

[28] Yuyama S, Ohtsu M. Acoustic emission evaluation in concrete. In: Kishi T,Ohtsu M, Yuyama S, editors. Acoustic emission-beyond the millennium.Elsevier Science Ltd.; 2000. p. 187–213.

[29] YoonDJ,WeissWJ, Shah SP. AssessingDamage in corroded reinforced concreteusing acoustic emission. J Eng Mech, ASCE 2000;126(3):273–83.

[30] Shippen NC, Adams DF. Acoustic emission monitoring of damage progressionin graphite/epoxy laminates. J Reinf Plast Compos 1985;4:242–61.

[31] Valentin D. A critical analysis of amplitude histograms obtained duringacoustic emission tests on unidirectional composites with an epoxy and a PSPmatrix. Composites 1985;16(3):225–30.

[32] Jamison RD. Microscopic techniques for damage assessment in laminatedcomposites. Louthan Jr MR, LeMay I, Vander Voort GF, editors.Microstructural science, vol. 14. American Society forMetals; 1987. p. 539–59.

[33] Pollock AA, Smith B. Acoustic emission monitoring of a military bridge.Nondestr Test 1972;5(6):164–86.

Page 11: Acoustic emission monitoring of bridges: Review and case studies

1714 A. Nair, C.S. Cai / Engineering Structures 32 (2010) 1704–1714

[34] Miller RK, Ringermacher HI, Williams RS, Zwicke PE. Characterizationof acoustic emission signals. Report no. R83-996043-2. East Hartford(Connecticut): United Technologies Research Center; 1983.

[35] Prine DW, Hopwood T. Detection of fatigue cracks in highway bridges withacoustic emission. J Acoust Emiss 1985;4(2–3):S304–6.

[36] Pollock AA, Carlyle JM. Acoustic emission for bridge inspection: Applicationguidelines. Final report, Contract DTFH61-90-C-0049. Washington (DC):Federal Highway Administration; 1995.

[37] Pullin R, Carter DC, Holford KM. Damage assessment in steel bridges. In: Keyengineering materials. Switzerland: Trans Tech Publications; 1999. p. 167–8;335–42.

[38] Watson JR, Yuyama S, Pullin R, Ing M. Acoustic emission monitoringapplications for civil structures, BridgeManagement, Thomas Telford, London,2005. p. 563–70.

[39] Landis EN, Shah SP. Frequency-dependent stress wave attenuation in cement-based materials. J Eng Mech 1995;121(6):737–43.

[40] Beck P, Bradshaw TP, Lark RJ, Holford KM. A quantitative study of therelationship between concrete crack parameters and acoustic emission energyreleased during failure. In: Key engineeringmaterials. Switzerland: Trans TechPublications; 2003. p. 245–6; 461–6.

[41] Pullin R, Holford KM, Lark RJ, Beck P. Acoustic emission assessment ofconcrete hinge joints. In: Key engineering materials. Switzerland: Trans TechPublications; 2003. p. 245–6; 323–30.

[42] Elliot JF. Monitoring of prestressed structures. Civil Eng, ASCE 1996;66(7):61–3.

[43] Paulson PO, Cullington DW. Evaluation of continuous acoustic monitoring asmeans of detecting failures in posttensioned and suspension bridges. In: XIIIFIP congress & exhibition. 1998.

[44] Brevet P, Robert JL, Aubaagnac A. Acoustic emission monitoring of Bridgecables: Application to a Pre-stressed Concrete bridge. DEStech Publications,First European Workshop on Structural Health Monitoring, SHM 2002, ENSCachan – France. 2002. p. 287–93.

[45] Rizzo P, Di Scalea FL. Acoustic emission monitoring of CFRP cables for cable-stayed bridges. In: Proceedings of SPIE — The International Society for OpticalEngineering, vol. 4337. 2001. p. 129–38.

[46] Kretz T, Brevet P, Cremona C, Godart B, Paillusseau P. Continuous monitoringand structural assessment of the Aquitaine suspension bridge. Bull LPC 2006;13–32.

[47] Fricker S, Vogel T. Site installation and testing of a continuous acousticmonitoring. Constr Build Mater 2007;21(3):501–10.

[48] Li D, Ou J. Acoustic emission monitoring and critical failure identificationof bridge cable damage. In: Nondestructive characterization for compositematerials, aerospace engineering, civil infrastructure, and homeland security2008, Proceedings of SPIE - The International Society for Optical Engineering,vol. 6934. 2008. p. 1–5.

[49] Jin T, Sun Z, Sun L. Acoustic emission monitoring of stayed cables basedon wavelet analysis. In: Sensors and smart structures technologies forcivil, mechanical, and aerospace systems 2008, Proceedings of SPIE — TheInternational Society for Optical Engineering, vol. 6932. 2008. 1-7.

[50] Gaillet L, Tessier C, Bruhat D, Michel R. Diagnostic assessment of multi-layercable anchorages bymeans of acoustic emissions. Bull LPC 2004;250–1; 55–63.

[51] Zejli H, Laksimi A, Tessier C, Gaillet L, Benmedakhen S. Detection of the brokenwires in the cables hiddenparts (anchorings) by acoustic emission. In: Acousticemission testing — Proceedings of the 27th European conference on acousticemission testing. Advanced materials research. 2006. p. 13–4; 345–50.

[52] Vogel T, Schechinger B, Fricker S. Acoustic emission analysis as a monitoringmethod for prestressed concrete structures. In: Proceedings EC NDT 9thEuropean conference on NDT. 2006.

[53] Kalny O, Peterman RJ, Ramirez G. Performance evaluation of repair techniquefor damaged fiber-reinforced polymer honeycomb bridge deck panels. J BridgeEng ASCE 2004;9(1):75–86.

[54] Ozevin D, Greve DW, Oppenheim IJ, Pessiki S. Steel plate coupled behaviour ofMEMS transducers developed for acoustic emission testing. In: 26th Europeanconference of acoustic emission testing. 2004. p. 557–64.

[55] Hay TR, Hay DR, Hay JR, Greve DW, Oppenheim IJ. Transforming bridgemonitoring from time-based to predictive maintenance using acousticemission MEMS sensors and artificial intelligence. In: 7th world congress onrailway research. 2006.

[56] Spillman Jr WB, Claus RO. Optical-fiber sensors for the detection of acousticemission. MRS Bull 2002;396–9.

[57] Stryk J, Pospisisl K. Rebar corrosion in concrete bridges and its detection byacoustic emission method. CDV- Transport research center, Czech Ministry ofTransportation and Communications. 2001. p. 1–6.

[58] Paulson PO, Elliott JF, Youdan DG. SoundPrint r©acousticmonitoring to confirmintegrity of stressed wire in bridges, structures and water pipelines. In: 15thworld conference on nondestructive testing. 2004.

[59] Vallen DIH. AE testing fundamentals, equipment, applications. NDT Net 2002;7(9).

[60] Grosse CU, Finck F, Kurz JH, Reinhardt HW. Monitoring techniques based onwireless AE sensors for large structures in civil engineering. In: Proc. EWGAE2004 symposium. 2004. p. 843–56.