Wireless monitoring of the longitudinal displacement of the Tamar Suspension Bridge deck under changing environmental conditions Nicky de Battista*a, Robert Westgatea, Ki Young Kooa, James Brownjohna Department of Civil & Structural Engineering, The University of Sheffield, Mappin Street, Sheffield S1 3JD, United Kingdom
a
ABSTRACT In order to be able to monitor the performance and health of a civil structure it is essential to understand how it behaves under different environmental conditions. It is a well documented fact that the structural performance of bridges can be altered considerably when they are subjected to changes in environmental conditions. This paper presents a study investigating the longitudinal movement of the road deck on Tamar Suspension Bridge in Plymouth in the UK over six months. The expansion joint of the bridge deck was instrumented with pull-wire type extensometers. The data were transmitted wirelessly using commercial wireless sensor nodes and collected at a data acquisition laptop computer, which was accessible online for remote monitoring. In addition, position data of various locations on the bridge deck were collected using a Robotic Total Station (RTS). Environmental data, such as the temperature, and structural data, such as cable tension, were acquired from other monitoring systems. Conclusions drawn from a fusion of the bridge deck’s longitudinal displacement with other structural and environmental data are discussed in this paper. Keywords: Structural health monitoring, suspension bridge, environmental effects, wireless sensor networks, data fusion
1. INTRODUCTION Structural health monitoring (SHM) has been carried out on a number of vehicular and pedestrian bridges[1,2,3,4,5] in order to understand the behavior of the structures, in particular the load–response relationship. Suspension bridges are often very complex structures that still tend to present surprises[6,7] every now and again, despite advances in computational technology and design procedures such as wind tunnel testing. Monitoring exercises can provide an important source of information which both engineers and the bridge management can use to gauge the performance of the structure and identify any anomalies that might be present. The Vibration Engineering Section, VES, at The University of Sheffield has monitored a number of bridges in the UK and overseas. One of these is the Tamar Bridge (Figure 1), a suspension bridge in the south west of the UK, from which a wealth of knowledge has been gained from a number of short and long term SHM exercises carried out by various members of the research group and others[8,9,10,11,12,13]. In this paper we present the latest addition to the monitoring systems installed on Tamar Bridge, consisting of a wireless sensor network (WSN) that has been used to collect data about the longitudinal movement of the bridge deck at the expansion joint. The next section provides some background information about the structure of the Tamar Bridge and the various monitoring systems deployed on it. Next, the hardware and software of the wireless longitudinal displacement monitoring system that is the focus of this paper is described. The data obtained from this monitoring system are then presented, together with a discussion about the deductions that can be made from a fusion of this data with environmental and structural information. Finally the main points of this paper are summarized in the conclusion.
*
[email protected]; http://vibration.shef.ac.uk/
Figure 1. Tamar Suspenssion Bridge in Plymouth, P UK.
Figure 2. B Bird’s eye view w of the Tamar Bridge.
2. THE E TAMAR BRIDGE B The Tamar Bridge was opeened in 1961 to t provide a vvital transportaation link overr the River Taamar, between n Saltash to thee west and Plym mouth to the east. e At the tim me it was thee longest suspeension bridge in the UK, caarrying approx ximately 40000 vehicles everyy day with a maximum m gross weight of 224 tons. The original o structu ure was a sym mmetrical susp pension bridgee, with a main sspan of 335m and side span ns of 114m onn either side (Figure ( 2). Reinforced conccrete towers riise 73m abovee their caisson foundations at a either end of the main sspan, with thee bridge deck suspended att half their heeight. The twoo main suspenssion cables haave a diameterr of 350mm aand each conssist of 31 lock ked coil wire rropes. They support s lockedd coil hangers aat 9.1m intervvals, from whiich the deck iss suspended. The T deck structure consistss of 5.5m-deep p trusses madee of welded hollow-box steeel sections. Orriginally thesee carried a 3-laane compositee deck 150mm m deep and a mastic asphallt top layer. Thee truss bearinngs on either side s of the Salltash Tower are a separated and a an expanssion joint is in ncorporated inn the roadway iin order to alllow differentiaal longitudinaal movement between b the Saltash S side sppan and the rest of the deckk. The bearings at the Plymouuth Tower were linked togeether in the oriiginal design. In the late 19990s the Tam mar Bridge waas strengtheneed and widen ned in order to t come in linne with a Eu uropean Unionn Directive thatt bridges shouuld be capablee of carrying vvehicles weigh hing up to 40 tons, t a proces s which was completed c in 3 phases by December 2001. Details off the upgrade exercise are provided by y Fish and Giill[14]. The up pgrade mainlyy consisted of replacing thee composite deck d with a llightweight orthotropic o steeel deck and constructing an additionaal
vehicular lanne and an addditional pedeestrian walkw way on the north n and sou uth side of tthe original deck d structuree respectively. In order to support the increased i weiight and carry ying capacity y of the bridgge, a numberr of structuraal modificationss were necessary. Sixteen additional sstay cables were w installed between the towers and the deck andd abutments (F Figure 3); the deck trusses were strengthhened with thee addition of structural meembers, plates and cables at a key locationss (Figure 4a); orthotropic stteel panels weere welded to o form the add ditional side laanes cantileveering 6 meterss out of the exiisting deck strructure and su upported by ouutrigger trusses (Figure 4a)); and the linkk joining the bearings b at thee Plymouth Tow wer was severred (Figure 4b b), with longittudinal restrain nt being proviided by the neew cantilevereed side lanes.
Figure 3. P Position of addditional stay cab bles (identical oon north and sou uth sides).
Figure 4. Structural moddifications mad de to the deck trusses: (a) streengthening plates and outriggger trusses supp porting the cantiilevered orthotrropic steel side decks d (left); andd (b) the severeed bearing link at the Plymouthh Tower (right)).
s 2.1 Existing monitoring systems As part of thhe upgrade prrocess, a perm manent enviroonmental and structural mo onitoring syste tem was instaalled by Fugroo Structural Monitoring in order o to prov vide the projeect team and the bridge management m w with informattion about thee behavior of tthe bridge duuring and afteer the upgradde works[8]. This T monitorin ng system con onsisted of an nemometers too measure windd speed and direction d abov ve the deck aand on the to owers; a fluid pressure bassed level senssing system too measure the vvertical displaacement of thee deck; sensorss measuring th he temperaturre of the main suspension caable, the deckk, the trusses annd the air; reesistance strain gauges to m measure the load l on the additional a stayy cables; and d an electronicc distance meassuring (EDM)) device measu uring the relattive displacem ment between the tops of thee two towers. In 2006, VES S installed a loong-term mon nitoring system m to study the dynamic behavior of the bbridge deck an nd a number of the additionaal stay cablees. The instrrumentation cconsisted of uni-axial QA700 and Q QA750 force-balance typee accelerometerrs (Figure 5). Accelerometter pairs (in thhe two perpen ndicular directions) were innstalled on fo our stay cabless and three acccelerometers (two ( vertical and a one transsverse) were installed i on th he bridge deckk, approximattely 36 meters
away from thhe center of thhe main span n. In addition to collecting and storing data, d the dataa acquisition (DAQ) ( system m which was prrogrammed ussing LabVIEW W and MATLA AB, also carried out real-time automaticc system identtification[11]. A number of obbservations maade after fusin ng the data froom the Sheffieeld and the Fu ugro monitorinng systems aree discussed byy Koo et al.[15]. In addition too the accelerometers, this monitoring m sysstem originallly included three pull-wire type extensom meters locatedd across the exppansion joint at the Saltash Tower. Thes e were intend ded to provide insight into th the deformatio on of the deckk, which could tthen be comppared with env vironmental annd modal paraameter inform mation. Howevver, the extensometers weree connected to the same wiring system ass the accelerom meters and th his turned out to be an inadeequate solutio on. Soon afterr, the extensom meters stoppedd functioning correctly annd were remo oved. Withoutt any knowled edge about the longitudinaal deformation oof the bridge deck, the pictture remained incomplete until, u in 2008, a Leica robottic total station (RTS) usingg total positionning system (TPS) ( techno ology was insstalled[16] (Fig gure 6). Thiss collected deeflection dataa from fifteenn reflectors located on the soouth side of th he bridge deckk, as well as on the tops off the towers (F (Figure 7). Ho owever, due too logistical reassons, the RTS cannot record d the deck moovement at thee Saltash Tower expansion jjoint[15].
Figure 5. A Accelerometerss mounted on th he stay cables ( left) and on thee bridge deck (right).
Figure 6. The Leica robootic total station n mounted on tthe roof of the control room (left) ( and one oof the fifteen reeflectors on the bbridge (right).
Figure 7. L Location of thee RTS reflectorss.
2.2 Finite element modell A 3D finite eelement (FE) model of thee upgraded brridge was dev veloped as paart of our reseearch, to prov vide analyticaal explanations for the responnses encounteered in the m monitored data by applying external loadding, such as traffic t and thee measured tem The FE modeel was created mperature of the t structural elements[17]. T d in ANSYS, as shown in Figure 8. Thee majority of ellements were modeled as SHELL63 S for plates, BEAM M4 for beamss and LINK100 for cable and d tension-onlyy elements. Mooving elemennts, such as th he rockers annd the lateral thrust girder,, were simulaated with hinged BEAM444 elements. For thermal anaalyses, plate elements werre replaced with w SHELL57 and beamss and cables replaced withh LINK33 elem ments. In total,, the model co ontains nearly 5200 nodes for f the majoritty of the analyyses.
Figure 8. F Finite element model of the Taamar Suspensioon Bridge.
33. WIREL LESS LONG GITUDINAL L DISPLACEMENT MONITORIN NG SYSTEM M Extensive annalysis of the data being continuously ggathered from m the various monitoring ssystems in place on Tamaar Bridge has bbeen (and is still being) carried out. H However, thee lack of info ormation aboout the deck’s longitudinaal displacement at the expanssion joint wass crucial. Of pparticular interrest was the reelationship beetween the exp pansion of thee deck and struuctural (cable tension, natu ural frequencyy, vertical defo formation, etc.) and environnmental (temp perature, windd speed, etc.) chhanges. Thereefore in June 2010 2 the samee pull-wire ex xtensometers which w had beeen previously removed weree re-installed att the Saltash Tower T expansion joint (Figuure 9). These consisted of ASM A WS12 liinear potentio ometers havingg a measuremennt range of 0 to 500mm, wiith the pull-ouut distance being directly prroportional too an analogue output currennt varying from 4 to 20mA. One O extensometer was instaalled on the no orth side and another a on thee south side off the deck. Thee extensometerrs themselves were fixed to o the Saltash Tower structu ure and the pull-out p cabless were attached to bracketss fixed to the ddeck trusses, thhus measuring g the relative ddisplacement between b the deck d trusses annd the Saltash h Tower acrosss the expansionn joint. A thirrd extensometer was installled at the cen nter of the deeck to measurre the relativee displacemennt between the m main and Saltaash side spanss, but this senssor turned outt to be defectiv ve and was dissregarded.
Figure 9. The ASM WS112 pull-wire exttensometers moounted across th he deck expansiion joint at the SSaltash Tower.
Whereas in the 2006 settup the exten nsometers werre wired to the same DA AQ system aas the dynam mic monitoringg instrumentation, this time they were keept separate. 220V DC pow wer was suppliied to the sennsors via existting cables. A National Instrruments (NI) wireless senssor network (W WSN) was in nstalled to tran nsmit the dataa from the exttensometers too the control chhamber locateed at the abutment on the Plymouth sid de (Figure 10). The NI WSSN system op perates on thee 2.4GHz frequuency range using u the IEEE E 802.15.4 prrotocol. The extensometers e were conneccted to an NI WSN-3202 4channel, 16-bbit analogue innput node. At the other end a laptop com mputer located in the controll chamber serv ved as the dataa sink and this was connecteed to an NI WSN-9791 Etheernet gateway y node. Althou ugh by using dduty cycling itt is possible too operate the N NI WSN inputt node using four f AA batteeries, the pow wer limitation which hampeers many WSN N installations was overcom me by supplyinng both nodes with DC pow wer from existiing junction bo oxes availablee near both loccations. The distance between the sensing s locatio on (the Saltassh Tower expaansion joint) and a the data siink (the Plymouth abutmennt control cham mber) was appproximately 450m. 4 This w was greater than t the transsmission rangge of the NI WSN nodess. Therefore an intermediate NI WSN-320 02 node was iinstalled and configured c to act as a routeer so as to transmit the dataa wo-hop networrk. The locatio on of the routter node had too be chosen carefully sincee, from the end node to the siink using a tw unlike the ennd node, the NI N WSN router cannot usee duty cycling g and is consttrained to rem main awake co ontinuously too listen for trannsmissions. Thhis necessitatees an externall power supplly since AA batteries b woulld be depleted d within a few w hours. Thereffore the routeer node was in nstalled on thee main span near n the Plym mouth Tower, where there was w a readilyyavailable DC power supplyy. With this co onfiguration, the wireless network n consisted of an appproximately 330m 3 hop over the main spann between thee end and routter nodes andd an approxim mately 120m ho op over the PPlymouth side span betweenn the router annd gateway noodes (Figure 11). 1 The form mer was still longer l than th he specified ooperational raange of the NI WSN nodes, thus necessiitating a diffeerent antennaa to increase the transmisssion range. T This was an unprecedented u d approach usinng the NI WS SN equipmentt and thereforee a number of tests using different d thirdd-party high-g gain omni- andd uni-directionaal antennae were w carried out. The link qquality was allso found to improve i consi siderably when n line-of-sighht conditions weere available between b nodees. A 10dBi hiigh-gain, omn ni-directional antenna was eeventually insstalled on eachh of the three nnodes and thiis increased th he transmissioon range enou ugh to operatte the networkk reliably (Fiigure 12). Thee antenna at eaach node locattion was fixed d on top of thee crash barrieer running along the entire length of the bridge. Whilee this did not prrovide compleete line-of-sig ght due to the ppositive camb ber of the deck k, it was the neext best solutiion available.
Figure 100. The NI WSN N-3202 analogu ue input node (left) and the NI WSN-9791 1 gateway nodee (right) in weeatherproof encloosures and secuured to the struccture.
Figure 11. Schematic layyout of the wireless sensor netw work.
Figure 12. A 10dBi high-gain antenna mounted m on the crash barrier an nd connected to o a wireless nodde.
Since the exttensometer reaadings were not n expected tto change rapiidly over timee, they were ssampled and transmitted t byy the end node at a rate of 0.2Hz. 0 The daata acquisitionn was managed using a virrtual instrumeent (VI) progrrammed in NI LabVIEW runnning on the laptop l acting as a the data sinnk (Figure 13)). The VI also served as a reeal-time system observationn monitor, show wing network link quality, power p status oof the nodes and a a numericcal and graphic ical visualizatiion of the dataa being collected. The laptoop was contin nually accessibble remotely via a broadbaand connectioon and could be queried ass necessary. Thhe data were stored s on the laptop l in binar ary files each spanning s 24 hours. h Each fille was automaatically copiedd to an off-site database systeem and processed to removve any significcantly noisy daata.
Figure 13. Screenshot off the Tamar Brid dge extensometter monitor VI programmed p in n NI LabVIEW..
4. D DATA ANAL LYSIS meter data werre analyzed allongside the ddata collected from the Fug gro and RTS m monitoring systems in order The extensom to investigatee the relationnship between the longituudinal movem ment of the deck and envvironmental and structuraal parameters. T This analysis was w carried ou ut for data colllected over siix months betw ween July andd December 2010. 2 The timee histories obtaained from alll the monitorring systems were reduced d to four-minu ute averages every 30 min nutes so as too facilitate com mparisons of the t different data d sets. Sincce the data obtained from the extensom meters are a measure m of thee distance betw ween the mainn span deck and a the suppoorting Saltash Tower, a possitive change in the readin ngs indicates a movement off the deck to thhe east (away from the Tow wer) and vice-v versa. Figure 14 shows a time history h of the extensometerr data recordeed over the entire period tthat has been analyzed andd Figure 15 shoows the samee data during a typical weeek in July 201 10. As expectted, the longittudinal displaacement of thee bridge deck hhad a clear diuurnal pattern as a well as a seeasonal trend. The deck wass closer to thee Tower during the day timee and during thhe warmer summer monthss. Also noticeeable are sections of missin ng data from both extenso ometers. Thesee were due to a recurring prooblem with th he power suppply to the mon nitoring system ms from the B Bridge’s electtrical circuitryy. There were aalso a number of instances where the datta from the so outh extensom meter were cleearly erroneou us, most likelyy due to the exttensometer caable getting stu uck and not bbeing pulled back b by the spring mechanissm. These seg gments of dataa were deleted before any annalysis was caarried out andd the time hisstories shown in the figuress are for the ‘cleaned’ dataa. The longitudiinal displacem ment data used d throughout tthe analyses described d in th his section aree the average of the signals from the northh and south exxtensometers.
Figure 14. Time history of o the extensom meter readings ffor 1st July to 31 1st December 20 010.
Figure 15. Time history of o the extensom meter readings ffor a week durin ng July 2010.
4.1 Longitud dinal displaceement vs. tem mperature Figure 16 shoows the longiitudinal deck displacement at the expanssion joint plottted against thhe temperaturre measured inn the main spann’s suspension cable by the Fugro systeem. The plot shows s a lineaar trend betweeen the tempeerature and thee displacement of the bridge. The calcullated correlatiion coefficien nt between these two sets of data was -0.988, whichh demonstrates that the statiic configuration of the briddge was closeely related to the temperatture of the caable. This was confirmed byy the FE modeel which show wed a similar ttrend when the thermal exp pansion coefficcient for the deck d and cablee steel was assuumed to be 122×10-6/°C. Figure 17 shhows the same extensometter data plotteed against thee temperaturee monitored bby sensors mo ounted on thee underside of the deck and on the surface of the truss beneath the deck. d For tem mperatures beloow approximaately 15°C thee trends were ssimilar to thaat shown in Figure F 16. Hoowever, abovee this temperaature thresholld both sets of o temperaturee
readings exhiibited a non-llinear relation nship with thee deck displaccement. This is i clearer in FFigure 18 whiich shows twoo one-week portions of thiss data during July and Deecember when n temperatures ranged from m approximattely +12°C too +30°C and -88°C to +6°C reespectively. The trend in Deecember was linear for both h the deck andd truss temperrature data. Onn the other hand, the non-linnear trend was evident in Juuly, mostly in the truss temp perature compparison which h indicates thaat the deck exppanded more than expecteed for particuular temperatu ures mostly above a 15°C. Therefore the nonlinearityy between the ddeck expansioon and the trusss and deck tem mperatures occcurred season nally, and not throughout th he day. This phenomenon was accounted for in the FE modell by adopting a bi-linear rellationship betw tween the cablle temperaturee and the deckk and truss tem mperatures. Below B 15°C alll the nodes in i the model were w given thhe same temp perature as thee suspension caable, Tcable, whhich was taken as the variaable parameterr. Above 15°C C the deck tem mperature wass set as Tdeck = (0.433Tcable + 7.877) and thhe truss tempeerature was seet as Ttruss = (1 1.544Tcable – 8.798).
Figure 16. Longitudinal deck d displacem ment plotted agaainst the suspension cable temp perature.
Figure 17. Longitudinal deck d displacem ment plotted agaainst the truss an nd deck temperatures.
Figure 188. Longitudinal deck displacem ment plotted aggainst the truss and deck temp peratures for a week in July (left) ( and a weekk in December (right). (
The non-lineaar relationshipp between the longitudinal deck displaceement and tem mperature, whiich only appliied to the deckk and truss tem mperatures butt not to that of the suspensiion cable, was probably du ue to differenttial heating off the structuraal elements as a result of direect solar radiaation. On a cllear sunny day y, the thin ortthotropic deckk which has a large surfacee area heats up more than the bulkier suspension ccable. Conversely, the tru uss which is constantly shaded by thee cantilevered ddeck can onlyy be heated by y convection aand therefore would be at the t same temp mperature as th he surroundingg air. Thus, durring a sunny day, the temp perature of thhe suspension cable would be lower than an the deck teemperature buut higher than thhe truss temperature. This is best explaiined with a ty ypical example. Figure 19 sshows two webcam images taken from thhe top of the Plymouth P Tow wer at 11:59am m on two con nsecutive dayss in October 22009. The imaage on the lefft shows the Taamar Bridge on o a fairly cleaar, sunny day whereas durin ng the next daay the bridge was surround ded by fog andd the city in thee distance is barely b visible in i the image. Figure 20 sho ows a time hisstory of the tem mperatures reecorded duringg these two dayys and the variation from on ne day to the nnext is clear. On O the first (ssunny) day thee differences in i temperaturee between the ddeck, the suspension cable and a the truss iis noticeable. However, on the second daay, when theree was no direcct sunlight, there were very liittle differencees between thee temperatures of the structtural elementss. These concluusions demonsstrate that not only does thee temperaturee have a signifficant influencce on the stattic response of the bridge, buut also that annalyses should d account for tthe influencess of direct solar radiation an and any possib ble differentiaal temperatures this may givee rise to. Further investigattions will attempt to disting guish betweenn clear and ov vercast days inn order to refine the mathematical models of the Bridgee.
Figure 19. Webcam imagges taken at 11:59am on 5th (leeft) and 6th (righ ht) October 200 09 from the top of the Plymoutth Tower.
o cable, truss and a deck tempeeratures recordeed during 5th and d 6th October 20009. Figure 20. Time history of
4.2 Longitud dinal displaceement vs. RT TS readings Figure 21 shoows the longiitudinal deck displacement at the expanssion joint plottted against thhe longitudinaal and verticaal positions of thhe RTS reflecctors located along a the mainn span deck. The T locations of these reflecctors are show wn in Figure 7. 7 The northerlyy, transverse, motion m of thee deck is not ppresented heree since it has no n clear correelation with th he longitudinaal displacement. It is no surpriise that the exxtensometer reeadings had a linear relation nship with thee longitudinall movement reecorded by thee RTS system ssince they werre essentially measuring thee deck motion n in the same plane. In addiition, position ns closer to thee expansion joiint (RTS refleector number 044) went thhrough a wid der movement range since they were farrther from thee restraint at thhe Plymouth Tower. T The height h of the ddeck also chaanged linearly with the exteensometer reaadings. As thee deck became colder and coontracted (thee extensometeer reading incrreased), so did the suspenssion cables. This caused thee deck to be puulled up, resultting in an incrrease in heighht, with the ch hange being more m evident clloser to the ex xpansion jointt. The opposite happened whhen the deck and suspensionn cables heateed up and expaanded. This coould also be observed o in thee FE model. Fiigure 22 and Figure F 23 sho ow the deflectted shapes off the model when it was suubjected to -5°°C and +30°C C respectively. The exaggeraated vertical motion m is show wn in the figu ures while the longitudinal ddisplacement is representedd by the color contours which maintain the t same signn convention as a the extenso ometer readinngs, that is, ellements whichh move to the rright (east) haave a positive displacementt and vice-versa. By compaaring the two ccontour plots it can be seenn that as the teemperature deecreased the bridge deck contracted an nd curved upw wards while an increase in temperaturee resulted in aan expansion and sagging g of the deckk. The FE model m also deemonstrated th that the maxiimum verticaal displacement of the bridgee deck was in n fact greater than the long gitudinal mov vement. This w was confirmeed by the RTS S readings as shhown in Figurre 21.
Figure 21. Longitudinal deck displacem ment plotted aggainst the longiitudinal (left) and a vertical (rigght) positions of the RTS reflecctors.
Figure 22. Contour plot of o the FE modeel when subjecteed to a temperaature of -5°C.
Figure 23. Contour plot of o the FE modeel when subjecteed to a temperaature of +30°C.
4.3 Longitud dinal displaceement vs. stay cable tensioon at the expanssion joint plotted against thhe tension in the Figure 24 shoows the longittudinal deck displacement d t stay cables supporting thhe main span monitored m by the Fugro syystem. As the bridge deck expanded e tow wards the Saltaash Tower andd the extensom meter readings decreased, th he stay cables connected to o the Plymouth h Tower (P2 aand P4) weree stretched andd their tension iincreased. At the same timee, the stay cabbles connected d to the Saltash Tower (S2 aand S4) slack kened and their tension decreeased. This is demonstrated d by a negatiive gradient for f P2 and P4 4 and a positivve gradient for fo S2 and S44. While the FE E model resultts related reassonably well ffor the Plymo outh Tower caables, they didd not match as a well for thee Saltash Toweer cables. Thee complex beh havior of the bridge deck as a it sags while expanding towards the Saltash S Tower requires furthher study in orrder to understtand this discrrepancy betterr.
Figure 24. Longitudinal deck d displacem ment plotted agaainst the tension n in the main span stay cables.
5. CONCLUS SION In this paperr we have preesented a wireless sensor nnetwork whicch was installled on the Taamar Suspenssion Bridge too monitor the llongitudinal displacement d of o the main sppan deck usin ng pull-wire extensometers fitted across the expansionn joint. The daata obtained from f this mon nitoring systeem were fused d with enviro onmental and structural daata in order too further understand the behaavior of the structure. It hass been shown that below a temperature t thhreshold of aro ound 15°C thee longitudinal ddeck displacem ment changed d linearly withh the temperatture measured d on the mainn span suspenssion cable, thee truss and the deck itself. However H for temperatures t hhigher than th his threshold the t relationshiip between th he longitudinaal deck displaceement and the truss and deck temperaturees was non-lin near, probably y due to differeential heating of the various structural elements. This behavior b was replicated in the FE modeel by adopting g bi-linear tem mperature graadients for thee truss and decck. A compariison of the ex xtensometer rreadings with the data from m an RTS moonitoring systeem has shownn that, as the ddeck and susppension cables cooled dow wn and contraacted, the deck k rose, with tthe change in n height beingg greater than tthe longitudinnal displacemeent. Converseely, when the deck and suspension cablees heated up and a expandedd, the deck saggged. This was also confirmeed from the F FE model. Thee third comparrison was madde between th he longitudinaal deck displaceement and the tension in thee stay cables ssupporting thee main deck. As A the deck exxpanded and moved m towards the Saltash Toower, the tenssion in the stay y cables connnected to the Plymouth P Tow wer increased w while the tenssion in the stayy cables conneccted to the Salltash Tower decreased. One recurringg observation throughout th hese analyses was that the static configu uration and thhe structural behavior of thee bridge are farr from straighhtforward. Thee performancee of a compleex structure lik ke the Tamarr Bridge depen nds on severaal factors whichh should be taaken into acccount when deesigning and monitoring itt, some of whhich might no ot be obviouss. Hence the neccessity to carrry out long-terrm SHM exerrcises. Future work w will focu us on refiningg the temperatture-dependennt behavior of tthe FE modeel to take into o account thee heating effeect of direct solar s radiationn, investigatin ng further thee combined verrtical and longgitudinal displlacement of thhe bridge deck k as it expandss and contractts in order to understand u thee changes in tennsion of the sttay cables, and studying thee dynamic pro operties of the Bridge in relaation to the diisplacement of the deck.
ACKNOWLEDGEMENTS The authors would like to acknowledge the assistance given by National Instruments in providing some of the equipment used in this investigation and the on-site support given by Steve Rimmer, Richard Cole and David List from the Tamar Bridge and Torpoint Ferry Joint Committee.
REFERENCES [1] Brownjohn, J. M. W., Dumanoglu, A. A. and Severn, R. T., "Ambient vibration survey of the Fatih Sultan Mehmet (Second Bosporus) Suspension Bridge," Earthquake Engineering and Structural Dynamics 21, 907924 (1992). [2] Dumanoglu, A. A., Brownjohn, J. M. W. and Severn R.T., "Seismic analysis of the Fatih Sultan Mehmet (Second Bosporus) Suspension Bridge," Earthquake Engineering and Structural Dynamics 21, 881-906 (1992). [3] Kwong, H. S., Lau, C. K. and Wong, K. Y., "Monitoring System For Tsing Ma Bridge," Proceedings of the ASCE Structures Congress, 264-267 (1995). [4] Ko, J. M., Ni, Y. Q., Zhou, H. F., Wang, J. Y. and Zhou, X. T., "Investigation concerning structural health monitoring of an instrumented cable-stayed bridge," Structure and Infrastructure Engineering 5(6), 497-513 (2009). [5] Brownjohn, J. M. W., Magalhaes, F., Caetano, E. and Cunha, A., "Ambient vibration re-testing and operational modal analysis of the Humber Bridge," Engineering Structures 32(8), 2003-2018 (2010). [6] University of Washington, "Aerodynamic stability of suspension bridges with special reference to the Tacoma Narrows bridge," Bulletin No. 116, University of Washington Engineering Experiment Station (1954). [7] Larsen, A., Esdahl, S., Andersen, J. E. and Vejrum, T., "Storebaelt suspension bridge - vortex shedding excitation and mitigation by guide vanes," Journal of Wind Engineering and Industrial Aerodynamics 88(2-3), 283-296 (2000). [8] List, D., Cole, R., Wood, T. and Brownjohn, J. M. W., "Monitoring performance of the Tamar Suspension Bridge," Proceedings of the Third International Conference on Bridge Maintenance, Safety and Management (IABMAS'06) (2006). [9] Brownjohn, J. M. W. and Carden, E. P., "Tracking the effects of changing environmental conditions on the modal parameters of Tamar Bridge," Proceedings of the 3rd International Conference on Structural Health Monitoring of Intelligent Infrastructure (SHMII-3) (2007). [10] Brownjohn, J. M. W., Pavic, A., Carden, E. P. and Middleton, C. J., "Modal testing of Tamar suspension bridge," Proceedings of the 25th International Modal Analysis Conference (IMAC XXV) (2007). [11] Brownjohn, J. M. W. and Carden, P., "Real-time operation modal analysis of Tamar Bridge," Proceedings of the 26th International Modal Analysis Conference (IMAC XXVI) (2008). [12] Brownjohn, J. M. W., Worden, K., Cross, E., List, D., Cole, R. and Wood, T., "Thermal effects on performance on Tamar Bridge," Proceedings of the 4th International Conference on Structural Health Monitoring of Intelligent Infrastructure (SHMII-4) (2009). [13] Cross, E., Worden, K., Koo, K. and Brownjohn, J. M. W., "Modelling environmental effects on the dynamic characteristics of the Tamar suspension bridge," Proceedings of the 26th International Modal Analysis Conference (IMAC XXVIII) (2010). [14] Fish, R. and Gill, J., "Tamar suspension bridge - strengthening and capacity enhancement," Proceedings of the International Conference organized by the Institution of Civil Engineers (1997). [15] Koo, K. Y., Brownjohn, J. M. W., List, D. I. and Cole, R., "Structural health monitoring of the Tamar Suspension Bridge," Structural Control and Health Monitoring (submitted for publication) (2011). [16] Koo, K. Y., Brownjohn, J. M. W., Cole, R. and List, D., "Robotic total station for long-term structural health monitoring of the Tamar Suspension Bridge," Structural Health Monitoring (submitted for publication) (2011). [17] Westgate, R. J. and Brownjohn, J. M. W., "Development of a Tamar Bridge finite element model," Proceedings of the 26th International Modal Analysis Conference (IMAC XXVIII) (2010).
