do so through contâct sjth ùe Cooper¿tive Research Prog,rams Stall, Transponation Research Board, 2l0l Constitution Ave., N.V.. Wash¡ngton, D.C. 204IU.

Bridge Foundations-Results of NCHRP Project 2I'i5

This NCHRP digest presents the findings of NCHRP Project 2I-5, "Determinatíon of 'Unknown Subsurface Bridge Foundations, " which

evaLuates existing and new technologies for use in determining unknown subsurface bridge foundation characteristics. A continuationproject (NCHRP Project 21-5(2) is undenuay, which will research and develop equipment, field technologies, and analysis methods forthose new technologies with the most promising application to nondestructive testing of unknown subsurface bridge foundations. This

digest was prepared by the staff of Olson Engineering, Inc. The Principal Investigatorfor the project ís Larry D. Olson of OlsonEngineering, Inc.

Subject Areas: IIC Bridges, Other Structures, and Hydraulicsand Hydrology; IIIA Soils, Geology and Foundations; and IIIBMaterials and Construction

INTRODUCTION

This digest contains information about thefeasibility of using nondestructive test (NDT)methods for the determination of unknown depthsof bridge foundations. This will be of interest tobridge and other structural engineers; soils,geology, and foundation engineers; and materialsand construction engineers.

Of the approximately 580,000 highwaybridges in the National Bridge Inventory, many ofthe older, non-federal-aid bridges have no designplans available. Therefore, no information isavailable regarding the type, depth, geometry, ormaterial incorporated in the foundations (Elias,1992; Watson, 1990; Baguelin, et al. 1980). Thecurrent best estimate of the population of bridgesover water with unknown foundations is 106,000,with 25,000 of the bridges being on-state systemsand 81,000 bridges being off-state systems. Theseunknown bridge foundations pose a significantproblem to the departments of transportation/ì1-õdi-\.^4 +!^+ _-^_-:^___ ^-^.^^ L.^ ¿t^.^ n:,-l:---_rl\l,r

=r-l-t]-u !--t!l!:--v ð"l]!¡J uù-ù!¿]ùçù=uççdLtÞ-Ga!!tEi,-Ir truçl:al-

Nondestructive Testing of Unkno\ryn Subsurface

Responsible Senior Program Officer: LloydR. Crowther

.: ; : :;::

{-<

Highway Administration (FHWA) is requiring stateDOTs to screen and evaluate all bridges todetermine their susceptibility to scour. Foundationdepth information, in particular, is needed forperforming an accurate scour evaluation at each

bridge site, along with as much other informationon foundation type, geometry, materials, andsubsurface conditions as can be obtained.

NCHRP Project 21-5, "Determination ofUnknown Subsurface Bridge Foundations," wasintroduced to evaluate and develop existing andnew technologies that could determine subsurfacebridge foundation characteristics, where suchinformation was unavailable. The project wascarried out in two stages. The first stage consistedof the review and evaluation of existing andproposed technologies having promise for use indetermining unknown subsurface bridge foundationcharacteristics such as depth, type, geometry, andmaterials; this was followed by development of a

research plan. The second stage of the project_ _,^^:,-L.^,1 .._4- _-,^l_--¿i,- - ---s.-L-_L: __ -¡ rt.Llull¡i l5 Lçu- u F tr v. al ua ulltB-- aI rera tE U !:ll ig- as. i t lafiJ/: o r-. tirg-

TR¡NsrrontnTroN REsEARCT-I BoÂRD

NA'I'IONAI- RESEARCH COT JNCII-

2

recommended concepts, methods, and equipment as

was feasible under the remaining project budget.Nine technologies were selected for the

second stage of research work. They included fivesurface techniques (i.e., sonic echo/impulse re-sponse, bending [flexural] wave, ultraseismic, spec-tral analysis of surface waves [SAS\ry], anddynamic foundation response tests) and four bore-hole tectniques (i.e., parallel seismic, boreholesonic, borehole radar, and induction field tests).The surface techniques require access to the expos-ed parts of the bridge substructure elements; theborehole methods require access through a nearbyboring. The major objective of the research was toevaluate the capabilities of the various NDT meth-ods that indicate depth and other information on un-known bridge foundation characteristics for widelyvarying known bridge substructure conditions.

Nondestructive tests, coupled with theoret-ical modeling, were performed at seven bridge sites(four in Colorado, two in Texas, and one inAlabama) under NCHRP Projecr 21-5. Casehistories were also reported from independent NDTconsulting investigations to determine unknownfoundation conditions.

Since this project was completed, a con-tinuation project, 2l-5(2), "Unknown SubsurfaceBridge Foundation Testing," has been awarded toOlson Engineering, Inc. That project is for furtherdevelopment, instrumentation, and validation of thesurface and borehole NDT technologies with thegreatest application range as determined in thisresearch project.

NCHRP PROJECT 21.5 RESEARCHAPPROACH AND INTERPRETATION

Discussions of the NDT methods used inthe research and their applications follow, alongwith a summary of the research findings. Dataexample results are shown below only for the mostpromising NDT methods.

Sonic Echo/Impulse Response Method andResults

The sonic echo/impulse response methodwas developed for testing the integrity and lengthof-singlg:rodlike: .crdummar.:=dæn,fisrmdatft¡r¡sirnr*¡*-Ë--¡_.æås=-dfillËdi'strft- and' diivcn pítæ (Kilten ard

Middendorp, 1981; Davis and Dunn, 1975). Thismethod (Figure 1) involves hitting the top of a deepfoundation with a hammer to generate a downward-traveling compressional wave; this compressionalwave reflects back to the surface from changes instiffness, cross-sectional area, and densþ (i.e.,acoustic impedance). The arrival of the reflectedcompressional wave energy is sensed by a receiver(i.e., an accelerometer or a vertical geophone).Analyses are done in the time domain for the sonicecho test and in the frequency domain (mobilitytransfer function, that is, velocþ/force) for theimpulse response test. Where top access is notavailable, or cap beams or superstructure arepresent that would greatly complicate theinterpretation of reflections, testing can beperformed on the sides of accessible substructure.

In the sonic echo test, exponentialamplification with time of the receiver trace (invelocity units) is usually required to enhance weakechoes and compensate for the damping of thewave energy as it travels down and back up afoundation. For wavelengths that are long rela-tive to the rodlike foundation diameter, the barcompressional wave velocity, V, is equal to thesquare root of Young's modulus, E, divided by

mass density, p (V = rfø t p¡. The depth of the

reflector is calculated as D : Lt .V/Z; where D isthe reflector depth and Âr is the time intervalbetween two echoes.

In the impulse response test (also called thetransient dynamic response), the transfer and thecoherence functions and the spectrum of the re-ceiver are typically recorded to calculate the depthof reflectors. The coherence function is used tojudge the quality of data. The depth of reflector iscalculated as D : Vl(2,Afl; where {/ is the fre-quency interval between two or more evenly spacedresonant peaks in the transfer function or the autopower spectrum plots.

The sonic echo/impulse response method ismost applicable to columnar substructures on drill-ed shafts or other columnar deep foundations thatare exposed above the ground or water. For con-crete and steel piles, lengths can be predicted towithin 5 to 10 percenr. Brooks et al. (1992) es-timate that the lengths of timber piles can beco¡se¡v¿tively-- p¡edicfed^ within, t5 percent-bec,aüseof+ tlle- greater- ve{É¡rfigrr e*, wlpwioad; wevê-

velocity in wood than in concrete. When embeddedlength to diameter ratios are greater than 20:1 to30:1 in stiffer soils, there will be no identifiablebottom echoes because of excessive damping of the

compressional wave energy in the sonicecho/impulse response tests. This problem is evenworse for steel pipe and H-piles, which have alarger surface area per unit length than solidconcrete, and timber piles, which are typicallysquare and round in cross section.

Sonic Echo/Impulse Response Example Results

Figure 2 (top) shows results from sonicecho tests from a timber pile of a FranktownCounty, Colorado, timber bridge. On the basis ofthe calculations shown in the figure, a length of29.8 ft was calculated for this pile, which agreedreasonably well with the 28-ft length from theengineering drawings. The length of the pile iscalculated on the basis of a compressional wavevelocity of 17,000 ftlsec measured from longitud-inal ultrasonic pulse velocity (UPV) tests from a

source on the pile top to a receiver on the pile side.The auto power spectrum of the accelerometer re-ceiver used in impulse response testing of the pileindicated a frequency interval of 305 Hz betweenthe peaks as shown at the bottom of Figure 2. Thiscorresponds to a pile length of 27.9 ft, which isalso in agreement with the actual length of 28 ft.

The sonic echo method also showed poten-tial for identifying the depths of shallower, moremassive abutments and piers, although not as con-clusively as the ultraseismic method. Becausesonic echo/impulse response testing has been exten-sively used for evaluating the integrity and lengthof concrete piles, drilled shafts, and timber piles(which are all columnar, rodlike elements), the testwas expected to work best on columnar bridgesubstrucfure, and this was generally the case.

Sonic E cho /Impuls e Resp ons e The oretical M odelingSummary

Theoretical modeling studies were per-formed for the NCHRP 21-5 project by Dr. JoseRoesset and his sfudent, Shu-Tao Liao, of the Uni-versity of Texas at Austin, of sonic echo/impulserespÞns+tstÈ=oËpilæ=rstlrg=p¡2:=ard+3_:diineå:sional (t-,2:, and 3-D) fiiäte-element programs to

3

model the application of a vertical force at the top-center of a drilled shaft and the receiver response

toward the top-edge of a shaft (Liao, 1994\. Thebasic difference between the three approaches isthat the 3-D solution allows the energy to spread

out into the larger radius top slab, and changes inimpedance (cross-sectional area changes in theFigure 3 models) stand out sharply. Thus, the 3-Daxisymmetric model (all elements are circular incross section) much more accurately reflects thereal 3-D wave propagation, and its results areshown here.

Two bridge substructure and superstructuremodels used for the theoretical modeling of pilesystems are shown in Figure 3 (top a and b).Example results of a theoretical sonic echo plot arepresented in Figure 3 (bottom a and b) for the caseof the top slab having the same diameter as thecolumn (top a) and the case of a 2-m-radius slab ontop of a 0.4-m-radius column (top b). Comparisonof velocity plots for the two geometries indicatessignificant differences in the recorded responses.

The Figure 3 bottom a record was theoret-ically generated (for the straight column case top a)by applying an impact to the top of the drilled shaft(Point A) and recording its response as waves trav-eled up and down the slab-column-shaftsubstructure. The letters in Figure 3 refer to wavereflection events from the various impedancechanges (i.e., cross-sectional area changes).Examination of the 3-D velocity plot (bottom a)shows the initial downward break at time 0 thatbecomes positive at the time duration of the impact,Id at 1.5 ms. The record is then quiet until thestrongest reflection event, ACA, occurs at about4.2 ms. This is the time required for the generatedtensile wave (first particle motion is tension) totravel 8 m upward at an assumed velocity of 3,793m/sec for concrete to the top of the column, reflectalmost 100 percent at the free column end andtravel 8 m downward as a compressional wave tothe receiver on top of the drilled shaft. The reflec-tion from the bottom of the l2-m-long shaft occursat about 6.3 ms as indicated by the ADA label.The bottom reflection event started downward as acompressional wave (first particle motion iscompression) and was only partially reflected backup the shaft as a tensile wave-some of the energyrrrac-.fffi @¡{ú¡d ¡in+d -.+lid -^;l ì-'*ÃirâJ.:å* *Jå *'ai-.4¡lvvqor-!r4rN¡ultws=rlrLv-tl¡EùV¡r-ùgl:t_V.Urru_f Àf 6Èlf f\¿?ùf ¡A,lL-

and at the bottom of the shafr by radiatioidamping.

4

Examination of the Figure 3 bottom brecord (which is the velocity record for the 2-m-radius slab on top of the O.4-m-radius column) nowshows a positive break at ABA, which correspondsto the (reflected) upward traveling tensile waveencountering an increase in impedance, or fixedcondition, at the column/slab (now larger than thecolumn) interface. This is followed by a downwardbreak corresponding to the reflection from the slab-air interface. Also of interest is that the bottomreflection is now completely obscured in the 3-Dvelocity plot at the ADA event. This is because somuch energy was reflected back down the columnfrom the larger area slab.

The theoretical finite element modelingclearly shows the complex vibrations that resultfrom column to beam interactions can mask thedesired echoes from foundation bottoms in sonicecho tests. Impulse response tests can be evenmore complicated, particularly if the testing is donein the middle of a column rather than toward theend.

Feasibility of Neural Networks for SonicEcho /Impuls e Resp ons e Analy s es

Even experts find it can be very difficult,even impossible, to analyze the sonic echo/impulseresponse test results for most bridge substructuresbecause of the highly variable geometry. Dr.Glenn Rix of Georgia Tech conducted a study (aspart of this project) to evaluare the feasibility ofusing neural networks as a means of analyzing themore complex data produced by sonic echo/impulseresponse tests.

An artificial neural network is a highlyinterconnected collection of relatively simpleprocessing elements that excels at patternrecognition tasks, including classification,forecasting, and mapping. In this context,determining the depth of an unknown foundationfrom nondestructive test results is viewed as amapping problem-personnel can train a neuralnetwork to associate the length of pile with NDTresults by presenting it with examples of therelationship. Neural networks are fast in theircomputations, learn from example and experience,and deal well with noisy input data. The extent towhich1"e.m¡gl_:.ne&mmku¡-ilLl_@=serße; byttre bredhTof'tlúe trainirg exrmrples and

how similar a particular input is to those trainingexamples.

Two synthetically trained networks (basedon l-D finite element modeling for raw [i.e.,normall data) and enhanced impulse responsemobility plots were used to predict pile lengths forsteel H-piles in air and in the pilecap and for timberpiles. Good agreement between predicted andactual depths was obtained for the steel pile in airand the timber piles. In both cases, theexperimental mobility curves matched thetheoretical mobility curves reasonably well. Theseare also the two cases fo¡ which l-D modelingwould predict the response accurately. In morecomplex cases (e.g., the pile in a pilecap and thepile with a beam on top of it) a l-D model wouldnot predict the response accurately. Notsurprisingly, the neural network did not predict thedepths of these piles well.

The study did demonstrate the feasibility ofusing artificial neural networks in analyzing thesonic echoiimpulse response data for simplerrodlike shapes. Solid, broad training is important;for more complex substructure shapes, 3-Dmodeling is beneficial. Ideally, the best trainingwould be on experimental results obtained from abroad spectrum of bridges with known foundationdepths; this would allow experience to beincorporated into the neural network, but largeamounts of experimental data would be requiredand this training is expensive.

Bending Wave Method and Results

The bending wave test is based on thereflection of dispersive (i.e., velocity is a functionof wavelength) bending (flexural) waves in a timberpile resulting from the horizontal impacts of small-to large-sized hammers (Douglas and Holt, 1993).The passage of the flexural wave energy up anddown the timber piles is monitored by twoaccelerometer receivers positioned a few feet apartand mounted on the heads of roofing nails drivenradially into the pile. The two receivers are alsoused in determining the bending (flexural) wavevelocity propagating up and down the pile. Thereceivers are in-plane with the hammer blow on theopposite side of the pile as shown in Figure 4.

Düugþr:anü=f {blt+G1993FflN+coneeived.lof¿and'' ussd tlle' sinrt kemel' rrrctM to anaþe

bending wave data for timber piles. The method is

similar to narrowband cross-correlation proceduresbetween the input (the hammer blow) and theoutput (receiver response). However, instead ofmeasuring the hammer blow, a periodic function of0.5 to I or more cycles is used as a "kernel seed,"and several seeds of frequencies ranging from 500to 4,000 Hz may be cross-correlated with thereceiver responses. The short kernel methodcorrelation procedure amplifies flexural waveenergy responses with the selected seed frequency.The dispersion of the bending wave velocity is thusaccounted for by calculating the bending wavevelocity for each kernel seed frequency. Thekernel frequencies are selected after transformingthe receiver outputs to the frequency domain toidentify the predominant frequencies in the receiverresponses.

Echoes of bending wave energy from thetoe and the head of the pile are manifested in thereceiver outputs. By knowing the velocity of thebending waves, the length of the pile can be easilycalculated (using a procedure similar to that for thesonic echo test) as D : Vt . Ltl2; where D is thelength of pile, V¡ is the velocity of flexural(bending) waves at a given seed frequency, and Âfis time interval between two return echoes (or thetime between first arrival and first echo withconsideration of the receiver location).

Bending Wave Example Results

The same timber pile from the Franktown,Colorado, timber bridge was also tested with thebending wave method. This pile has an exposedsurface approximately 9 ft long above the groundsurface with a free end at top. Two accelerometerswere placed in the transverse direction on the testedpile at distances of 42 in. and 84 in. from the top ofthe pile. The pile was excited in the transversedirection with a hammer hit at the top and theresponse of the pile was measured by the tworeceivers. The time records measured by the tworeceivers were then optimally cross-correlated with1 cycle of a 50O-Hz-frequency signal to determinethe flexural wave velocity at that frequency toaccount for the dispersive effect. Fig,ure 5 (bottom)sho:vs=thæresBonse+oe:ir*reeciver=iserfÊf Ëat=4æ

5

from top with echoes separated by 22 ms. Figure 5

(top) shows the time records of the two receiversafter cross-correlating them with the 500-Hz, 1-

cycle wave from which a bending wave velocity of2,480 ftlsec was calculated. A pile length of 27.3ft was calculated; this compares well with theactual length of the pile (28 ft on the designdrawing). The sonic echo results also showed a

length of 29.8 ft, and the impulse response resultsshowed a length of 27.9 ft, which are comparableto the value determined by the short kernel methodanalysis.

Bending Wave Theoretical Modeling Results

Theoretical modeling on propagation offlexural waves was done by Dr. Jose Roesset andhis student, Chih-Peng Yu, of the University ofTexas at Austin for this project (Olson et al.,1995). Their work indicated that flexural wavereflections do not follow the same pile-endboundary condition rules (i.e., for free to fixedconditions) as body wave reflections (i.e.,compressional and shear waves) do. Body wavesshow the wave form changing sign (i.e., fromcompression to tension) after reflecting from a freeend and show no sign change after rêflecting froma fixed end (i.e., from compression tocompression). However, flexural waves do notfollow these rules. Thus, under some pile-endboundary conditions, picking the same sign peaksin the short kernel method (as recommended in theexperimental research of Douglas and Holt t19931)can introduce error into reflector depth predictions.Using a half-cycle kernel seed helps make the signsof the peaks more obvious so that changes in signdue to differences in end conditions are taken intoaccount. The effect of embedment depths forconcrete and timber piles was also examined byRoesset and Yu, and it was found that the bendingwave method may be unable to detect lengths ofdeeply embedded piles. For shallowly embeddedtimber piles, the reflection from the soil surfacemay be as significant as that from the bottom, andwith the shallow embedment, the reflection signalswill overlap and make bottom depth identificationdifriÐig

6

Ultraseismic Test Method and Results

As described in the previous two sections,the nondestructive sonic echo/impulse response(Davis and Dunn, 1975) and the bending wavemethods (Douglas and Holt, 1993) are best used todetermine the integrity and/or unknown depths offoundation elements of columnar, rodlikegeometries with minimal overlying superstructure.For columnar, complex-shaped bridge substructureelements with large changes in cross-sectional area(impedance boundaries), however, the results ofthese tests are hard to interpret because of multiplereflections that occur at each impedance boundary.

In response to the difficulties encounteredby the sonic echo/impulse response and the bendingwave test methods for a large subset of non-columnar bridges or other complex civil structures,the ultraseismic test methodt was conceived as abroader application of these methods during theNCHRP 21-5 srudy. The method can be used morebroadly on bridge substructures because it isapplicable to both columnar and more massivesubstructures. The ultraseismic test method usesmultichannel recording of acoustic data, followedby digital filtering techniques adapted from rheseismic exploration method, to process the data.Seismogram records are collected by using impulsehammers or vibrators (as the source) andaccelerometers mounted on the accessible structureelement at intervals of 1 ft or less (as receivers).The structure element itself is used as the mediumfor transmitting the seismic energy. All wavemodes traveling down or reflected back (i.e.,echoes from the bottom) are recorded by thismethod. For concrete structure elements, usefulwave frequencies up to 4 to 5 kHz are commonlyrecorded.

Multichannel recording of ultraseismic datais used to isolate and enhance "desired signal" inthe recorded seismograms. Desired signal oftenconsists of reflection echoes from major defects orthe bottom depth of the foundation. Dataprocessing is used to isolate and enhance thissignal, using some characteristic property, from

I The ultraseismic method (ultraseismic refers to high-frequencyseismic waves) was researched and developed by.Farrokh Jalinoos, ascoarinoipai'iirvesiigaroroi.NrSÊRÞ?rõjeot4i=5i

other sources of coherent and incoherent (i.e.,random noise) energy.

Although certain similarities exist betweenprocessing data from extended (i.e., geological)and bounded media, different processing schemesand philosophies must be adopted. For example,because of the access requirements, wavepropagation velocity and source wave shape canoften be deterministically measured rather thanstatistically estimated. The source of coherentnoise contamination, like multiple reflection frommany reflecting boundaries, is far more acute; andnormal travel time moveouts are often small.

Moreover, the underlying physicalprinciples for propagating stress waves in boundedmedia are different; this makes (algorithm)requirements for forward modeling, inversionanalyses, and some filter designs different. For amedium with a bounded geometry, such as a bridgecolumn, three types of stress waves are generated.In each type, longitudinal, torsional, and flexuralwaves are included (Kolsþ, 1963). In longitudinalvibration, each element of the column extends andcontracts along the direction of wave motion (i.e.,along the column axis). In torsional vibration, eachtransverse section of the column remains in its ownplane and rotates about its center. In flexuralvibration, the axis of the column moves laterally ina direction perpendicular to the axis of the column.Although each wave type independently canprovide information about the depth of thefoundation or the presence of significant flawswithin the bridge substructure, longirudinal (i.e., P-wave and compressional) and flexural (i.e.,bending) waves are much easier to generate thantorsional. Consequently, compressional andflexural wave energy is generated by orientingimpacts to tested structures vertically andhorizontally, respectively.

The following types of ultraseismic testgeometrieS have been developed specifically forthis problem:

¡ For a l-D image of the foundation depthand for tracking the upgoing and downgoingevents, the vertical profiling test method is used.Data appearance is similar to that of the verticalseismic profile method from seismic exploration.I¡.- tËì ¡.^qa+!¡¡ á -./EtÈ' -^ - ¿l a rrå- -L"-:i¡ ¡-- - ^ --rtu- - ^ -¡¡r-ul$r_=r¡le-tlrv-r¡=\IérË-ul:çrrfl FrlrçÉur;ruBçÉürr¡ifilllr=(,r---

abutment is hit from the top or bottom (bothvertically and horizontally), and the resulting wavemotion is recorded at regular intervals down thebridge substructure element. Typically, three-component recording of the wave field is taken inorder to analyze all types of ensuing wave motion.A vertical profiling line can be run in both acolumnar (e.9., a bridge pier or pile foundation)and a tabular (e.9., a bridge abutment) structure.

¡ For a2-D image of the foundation depth,the horizontal profiling test method is used. Withthis method, data appearance is similar to that ofthe surface seismic reflection method. In thismethod (Figure 7), the reflection echoes from thebottom are analyzed to compute the depth of thefoundation. The source and receiver or receiversare located on the top or at any accessible lengthalong the side of the substructure element, and afull survey is taken.

An important distinction between the twotests is that, in a vertical profiling test, the target isperpendicular to the survey line; in a horizontalprofiling test, the target is parallel to the surveyline. The vertical profiling lines are used todifferentiate downgoing events from upgoing eventson the basis of their characteristic time moveoutand to measure their velocity accurately. A verticalprofiling line is also used to link reflection eventsfrom the bottom to a corresponding horizon in ahorizontal profiling section.

In the verrical profiling rest (Figure 6), theimpact point can be either at the top or the bottomof the ¡eceiver line. Vertical impacts to thesubstructure are comparatively rich incompressional wave energy, although considerabledispersive flexural/Rayleigh surface modes are alsogenerated. Horizontal impacts are rich in flexuralwave energy when the impacts generatewavelengths longer than the thickness of thesubstructure element. Impacts that generatewavelengths shorter than the thickness willpropagate with Rayleigh wave velocity, and longerwavelengths will propagate at a slower flexuralwave velocity.

'In the horizontal profiling test (Figure 7),the source and receiver can be close to each otheron the top at an optimum offset to reduceint*re¡reei¡om'=k¿yicigtF;s:¿dii'çs=rvayeeTiÉ

7

source and receiver are then marched together inorder to obtain a conrmon offset dataset. A fullsurvey using all combinations of source andreceiver locations can also be taken, but this ismore data intensive. Alternative survey lines canalso be obtained by moving the source or receiverlines to a different depth within the structure asdictated by accessibility requirements. In thealternative survey line shown in Figure 7, reflectionevents from the bottom (or top) as well as thebackside of the wall can be analyzed to give usefulinformation about the structure.

Vertical Profiling Example Results

Figure 8 shows the source/receiver layoutfor the vertical profiling testing from a pier (i.e.,columnar) foundation of a bridge. The l}-lbhammer source was located at the bottom of thepier near the grade and was struck both verticallyand horizontally (i.e., perpendicular to the axis ofthe column). Two horizontal hammer hits wereused from opposite sides of the column. Three-component accelerometers were mounted on theside of the column at 0.5-ft intervals above thesource location. Thus, a full six-component datasetfrom two source directions and a three-componentrecording was obtained. This allows for a near-fullstudy of the dynamics of wave motion for alllongitudinal, torsional, and flexural wave modes inthis type of semicolumnar geometry.

The field recordings from a horizontal 12_lb hammer hit and horizontal receiver responses(perpendicular to the axis of the column) are shownin Figure 9. Each trace indicates the record from adifferent survey location on the column-with trace2 corresponding to 0.5 ft and trace 21 at 10 ftabove the source location as indicated in Figure g.

Figure 9 shows the first 15 msec of data at a 4-psecsampling interval. All data were de-biased (i.e.,direct current shift was removed and data werebandpass zero-phase-filtered [using a 0-0.5-3-4 kHztrapezoidal filterl and automatic gain controlled) toenhance weaker arrivals. The following events,similar to a typical vertical seismic profile record,are evident in Figure 9:

_ 1. Upgoing flexural events. (including_ref läcrionefr$ìnr.tlæ¡footirg:þw,ithea*tinea*posir,iv+-.

8

moveout with travel time increasing with distanceand

2. Downgoing flexural events (includingreflections from the beam) with a linear negativemoveout with travel time decreasing with distance.

The reflector event indicates a flexuralwave reflection at 4.9 ft. below the source point,which corresponds to the (bottom) depth of thepilecap footing. This gives a depth below the topof the beam of 30.2 ft, which compares well withthe 31.l-ft distance shown on the bridge plans.

Because of the geometry of the pier at itstop, both the downgoing and upgoing events arecyclical (ringy). In the frequency domain, themultiple reflection of the ringy seismic waveletgives rise to characteristic resonant peaks of acolumn. By calculating the distance between theresonant peaks (Âl) and knowing the velocity (Vr)of compressional wave motion, the unknown depthof the foundation (d) can be calculated by d :Vcl(z\f). This is the basis for the impulse responseNDT method for a depth (length) determination ofa column as previously described.

No evidence of the steel piles underlyingthe concrete pilecap footing was apparent in thevertical profiling results. Vertical profiling testlines can also be run from the side of a massiveabutment and other tabular type structures.

Horizontal ProfiIing Examp le Results

This method was developed for use onmassive abutment and wall substructure elements;they typically have greater widths of top or sidesurface access, which permits a line of receivers tobe placed at the same elevation. The source andreceiver locations used on one side of a pier of aWeld County, Colorado, bridge are shown inFigure 10, which shows a wall-type piersubstructure on a strip footing on steel H-piles.This pier wall was a good site for the initialresearch on the horizontal profiling method. Anautomated and repeatable protoiype tool wasdeveloped, which allowed for signal averaging ofthe received signal. The tool consists of anelectromechanical solenoid source generating animp¡lse* siggals b¡4 tap_p_ing;. with-= a=0 2.lb : forcÈ.tr¿rmducer; Tf¡E autnnatd¡O2ålbr sosfceìwðs fîffi"

horizontally (perpendicular to the wall surface)from one side of the wall (placed at 3.5 ft from thetop), and horizontal accelerometers were placed onthe opposite side of the wall directly across fromthe source locations. The source and receiverswere then moved simultaneously at l-ft intervalsalong the width of the wall foundation.

Figure 11 shows the data for 32 stationlocations across the pier length for a single-hitsource. These data were corrected for directcurrent bias, bandpass-filtered (0-0.7-4-5 kHztrapezoidal), and automatic gain controlled to aidwith the display of the weâker, late (i.e., deeper)arrivals. The approximate locations of the bottomreflections are indicated in the figure. Data qualityis comparable to a set obtained with manual(hammer) hitting. The bonom depths in Figure llwere obtained after comparing the horizontalprofiling line with a vertical profiling line (Olson etal., 1995) obtained from the corner of this pier. Ina vertical profiling section, events coming fromabove the survey line (like the superstructure) aredistinguished from those coming from below (likethe foundation bottom) on the basis of the slope oftheir time moveout curve. This dataset indicatedweak, bottom P-reflections, which seriouslyreduced the predominance of surface wave energyin the data. Optimum survey geometries andeffective means for suppressing the Rayleigh wavesare the subject of current continued research; thisresearch may lead to use of techniques differentfrom those employed in surface seismicexploration.

SASW Method and Results

The SASW method, which measures thevariation in surface wave velocity with depth inhorizontally layered media, was initiated at theUniversity of Texas at Austin in the early l9g0swith funding from the Texas Department ofTransportation (Heisey et al., 1982; Nazarian andStokoe, 1984). The method has been successfullyapplied to determine the shear wave velocityprofiles for soils (Stokoe et al.,19Sg) andpavement systems (Rix, 1988; Aouad, lg93).Testing is per-formed by impacting the surface andrecordingatlu pass ag.t> of:plqd@(smfæe)r'u,"åye eræ-rgyT rsirg. tço' recciver5: Æ.-

series of receiver spacings is used, and testing isperformed in for-ward and reverse directions ateach receiver spacing. A plot of the surface wavevelocity versus wavelength is called a dispersioncurve. Once the dispersion curve is determined,one can obtain the shear wave velocity profile ofthe structure or soil being tested.

In the application of this test to bridgesubstructures (Figure l2), one source and tworeceivers are placed on top of the abutment so thatthe distance from the source to the first receiver isequal to the distance between the two receivers. Adynamic signal analyzer is used to capture andprocess the receivers' outputs [denoted by .r(f) andy(f) for receivers I and2, respectivelyl. Then,.x(r)and y(/) are transformed to the frequency domain[X(fl and Í(fl] through the employment of a FastFourier Transform. X(fl and Y(fl are used tocalculate the cross power spectrum between the tworece*ivers, denoted by GxvfGxv: X(f)*. f(l) wherethe ' symbol denotes the complex conjugate)1. Thesurface wave velocity and wavelength, Vn(Í) and /¡(fl, are determined as: V¡ (f) : Dlt (D; and InA :Vnfllf, where tA : time delay between receiversas a function of frequency ( - phase shift of thecross power spectrum in degrees divided byfrequency), D is the distance between the two re-ceivers. A plot of surface wave velocity versuswavelength is called a dispersion curve and is usedin determining the depth of the foundation.

SASW Example Results

The SASW can determine the thickness ordepth of bridge abutments as long as certainphysical conditions are met: the foundation must bemassive and must have an exposed, fairly flat ledgeor top surface on which impacts are applied and apair of receivers is placed. As an example of data(Aouad et al., 1996), an SASW test array for thewest abutment of a bridge in Connecticut is shownin Figure 13. Three receiver spacings of 3, 6, and12 ft were used. The composite curve from thethree receiver spacings is shown in Figure 14. Thedispersion curve shows a two-layer system withtwo different velocities. The first part, up to awavelength of 9 ft, shows a surface wave velocityof 7,700 ftlsec which is indicative of concreteveioeitieð=-ThiÞseeÐãd-.-paliãshio:ùûsÈvelo"oitiËs-of+

9

approximately 4,000 ft/sec for wavelength greaterthan 9 ft, which is indicative of velocities ofmedium-hard rock. Therefore, it can be inferredfrom the SASW measurements that the depth of theabutment is approximately 9 ft.

Dynamic Foundation Response Method

This test method (Figure 15) was proposedas a means of differentiating between shallowfoundations and foundations with piles (or otherdeep foundations) underlying the visible bridgesubstructure. Although the method is unproven forthis use in bridges, it is based on the dynamicanalysis theory for vibration design of foundations(soil dynamics) and geotechnical analyses offoundations subjected to earthquake loading. Thedynamic analysis theory and the geotechnicalanalyses are based on the theoretical work ofNovak (1976) and Novak and Aboul-Ella (1978).Novak analyzed the problem of a simple shallowfooting foundation with and without piles forvertical and horizontal modes of vibrations on theground surface and embedded foundation. Pilefoundations underlying a pilecap are more rigid andexhibit higher vibration amplitudes and resonancefrequencies than shallow footing foundations in thesame soils. In NCHRP Project 21-5, dynamicfoundation response measurements to determineresonances were performed on all applicablebridges (i.e., nonvisible pile bridges). Theoreticalmodeling of the specific vibration response tomatch field tests as well as free-vibration analyseswere performed for selected bridges to becompared with the measured results. Somedifferences in the dynamic foundation responseswere found between the shallow and deepfoundations. However, bridge substructureresponses were influenced by the effects of thevisible substructure and the superstructure. Enoughpotential was seen that an FHWA research projectis being conducted (Olson et al., 1996).

Borehole Parallel Seismic Test Method andResults

The borehole parallel seismic method wasresearched and developed specifically to determinetliædepf huoÊu¡dñ¡rew-¡¡=f o'¡nd#toæ'blr+tîF€EBçe

10

research organization headquartered in Paris,France (Stain, 1982). This method (Figure 16)consists of impacting the exposed foundation top(or other accessible substructure element) eithervertically or horizontally with an impulsive hammersource to generate compressional or shear rwaves

that travel down the foundation and are transmittedinto the surrounding soil. The wave arrival is

tracked at regular intervals by a hydrophonereceiver suspended in a water-filled, cased borehole(the past procedure) or by a clamped, three-component, geophone receiver (the new procedure)in a cased or uncased borehole. The boring isdrilled typically within 3 to 5 ft of the foundationedge and should extend at least 10 to 15 ft deeperthan the anticipated and/or minimum requiredfoundation depth. The foundation bottom, or a

major defect in the foundation, will appear as a lowvelocity zone with reduced signal amplitude. Thefoundation depth (or defect depth) is determined byplotting initial arrival times of the wave energy tothe downhole receiver against test depths andobserving the depth where change of slope in thetravel time plot occurs.

This procedure works well for purelycolumnar foundation elements and where thesurrounding soil velocity is constant (as is the casein fully saturated soils). For soils with varyingvelocities, the traces in the section must bestatically shifted to correct for the variation in the(measured) soil velocities. Typically, at the testedsites, the soil compressional velocities increasedfrom about 1,000 ftlsec to 5,000 ftlsec(corresponding to 100 percent saturated soil, whichis the nominal velocity of water at the water table)as well as at bedrock. Moreover, the bridgefoundation shapes were mostly noncolumnar (incross section), because of the existence of a footingor pilecaps. In these cases, the conventionalparallel seismic analyses do not always work.

However, it was observed that the bottomof the foundation can act as a strong (secondary)source of energy, especially in more massivefoundations. The foundation tip acts as a pointdiffractor in emitting both upward and downwardtraveling waves into the borehole. This diffractionevent is best observed by using three-component..-,.;h*sa.-^ ic- - ^^ ^^.1 * ^-^,.+^ 'l ¡L-^-^liñ I Ä-.^ -{ E--, ^ - --õ!¿-vyfrvBw= rrf4Ëvqù_!¡sitt:J<!:vJl-çré*-t \tr:ttra\'lqgattlt¡=w aù-not as readily observable in hytlrophone sections

because of high contamination from the tube lvavesenergy. The diffraction results in a steep, V-shaped, hyperbolic event in the recorded seismicsection. The bottom of the foundation is thenidentified by noting the depth where the peak of thehyperbolic event occurs.

Parallel Seismic Example Results.

As a typical example of the parallel seismictest with geophones, Figures 17 and 18 show testgeometry and field data respectively from theconcrete pile pier of a highway bridge in Texas.The wave energy generated by the horizontal blowis mostly propagated in the flexural wave mode inthe pile caisson; this becomes refracted as a shearwave in the borehole. The SP positions shownacross the tops of the figures refer to depths of therecords in feet from the top of borehole.Therefore, SP 2 refers to a receiver location of 2 ftbelow the top of the borehole. At this site, soilshear velocity increased from 850 ft/sec to about1500 ftlsec at the bedrock. A sharp point diffractorevent in the flexural wave energy from thefoundation tip is indicated at a depth of 32.5 ft fromthe top of the borehole; this predicts a depth of33.2 ft below the pilecap for the pile bottom. Itagrees fairly well with the plan depth of 34 ft.

Parallel seismic data were also obtainedfrom the same highway bridge in Texas usinghydrophone receivers. Figure 19 shows thehydrophone results after automatic gain controlledamplification. Review of this figure shows a fastvelocity (corresponding to the concrete pile belowabout 16 ft þelow the water table, saturatedl) andthen a slower velocity (corresponding to the shale).The 33-ft depth of the pile below the top of the

borehole is clearly apparent in the figure. Thispredicts a depth of 33.7 ft to the pile tip below thepilecap, which also agrees very well with the plandepth of 34 ft. Note the predominance of bothupgoing and downgoing tube wave events (with avelocity of 1,660 ft/sec) in the hydrophone section.

The three-component parallel seismicmethod provided foundation depth data for thewidest range of bridge substructures and g'eologicalconditions by using geophone receivers in agrouted, cased boring. By using three-component

-:- -^^^^-l

LL-_^--l- -.------!rtq - - - -iLëçup¡rurrsÈrrrlr-- uasËuãuurdilB¡¡FwrrrtFBu-uuF col ttaçti-between the casing and soils, improved quality of

parallel seismic results was obtained at sites withvariable soil velocity conditions.

Borehole Sonic Test Method

The borehole sonic method was proposedfor research as a promising, but unproven,approach for determining unknown bridgefoundation depths (and perhaps even foundationgeometry). This method ideally involves loweringa source and a receiver unit in the same boreholeand measuring the reflections of compressional orshear waves from the side of the bridgesubstructure foundation using essentially horizontalraypaths (Figure 20). No borehole sonic toolsuitable for determining unknown bridgefoundation conditions exists; conventional soniclogging tools have limited depth of penetrationaround the borehole. A three-component downholesource and a three-component geophone receiverfor crosshole seismic tests were used in two pairsof separate borings at a wall-shaped caissonfoundation in Texas (the largest bridge substructureelement tested, and consequently the best target).The results of these tests showed promise for theborehole sonic method because the reflectionsobtained indicated, by their timing, the presence ofthe caisson foundation. However, no clear reflec-tions were obtained with a commercial singleholesonic logging tool for measurement of compres-sional and shear wave velocities around theborehole. The limited tests showed promise, butconsiderable research would be required to developa borehole sonic tool for use in determiningunknown foundation conditions.

Borehole Radar Test Method and Results

As with the borehole sonic test, boreholeradar had not previously been used to determineunknown foundation conditions. This test uses atransmitter/receiver radar antenna to measure thereflection of radar echoes from the foundation(Figure 21). The radar signal is propagated as anelectromagnetic wave in the soil in response to thedisplacement currents generated from the appliedelectric field. Displacement currents, generatedfrom intrinsic dþole moment distribution in a

eiieleetri_enuleriaþare-=i¡rtcontËsltr_tpiûeeoff rtçJionåcurrents created from the movement of free

11

electrons in a conducting material. Reflectionsoccur in the subsurface at interfaces with a contrastin dielectric properties (i.e., changes in electricalimpedance). Electrical impedance changes aregoverned by changes in relative permittivity ordielectric constant in the ground. For a verticallyincident plane wave impinging on a boundarybetween two materials of dielectric constant K1 andK2, a reflected wave is returned upward with anamplitude governed by the reflection coefficient, .it,

given by: ^

: (Jt' - J K' ) t(J *' *..Æ ) .

The reflected portion of the signal is recorded bythe receiver antenna and processed with variousgains and filters.

For the borehole radar test (Figure 21), thereflection from the soil/foundation boundary (i.e.,the abutment) is of interest. The soil dielectricconstant of a soil is a direct function of the watercontent-that is, as the water content increases, sodoes the dielectric constant. This results in adecrease of radar velocity or longer travel pathsthrough the same thickness of soil. The increase ofwater content also increases the electricalconductivity in soil, which increases the intrinsicattenuation of the radar signal. This results inreduced depth of penetration of the radar signal.Typical radar records include a constant transmitpulse at early parts of the record as well as thedeeper reflected events from the subsurface or thebridge abutment. Thus, environmental factors(such as salt water, conductive soils, groundmoisture conditions, and buried electrical powerlines) can critically limit and/or confuse radarsignals.

Borehole Radar Example Results

Borehole radar did not work at sites withconductive clayey soils. For example, Figure 22shows the field testing geometry of a battered steelBP pile of an Alabama. Figure 23 shows anexample borehole radar result from the batteredpile. The pile rip and the angled reflectioncorresponding to the battered pile are relativelyclear in Figure 23. Steel is an extremely goodreflector of the radar signal; therefore, it serves asan ideal target for this method. The horizontal axisdisp.lays'depth-mëasur€ddioi-rr-tr-rrrEioprof +tiiebor,iry=with depth increasing from right to left. The small

12

tick marks on the top of the figure indicate depthmarks in l-ft increments. In this figure, the depthmarks increase from 0 to 50 ft. The vertical axisdisplays time in nanoseconds (ns), which increasefrom top to bottom.

Several events are discernible in these typesof sections. The first event to note in Figure 23 isa constant amplitude and travel time "transmitpulse" event at about 10 ns, which is the result ofthe direct electrical coupling between thetransmitter and receiver antennae. A second seriesof events are believed to result from the effect ofsoil saturation at water-table depth. This ringysignal is indicated at about 17 ns where a linearevent is observed, which increases to about 20 ns at28 ft (the water table) from the top of the boring.The third series of events in this figure results fromthe normal incidence primary reflection events (andtheir multiples) from the steel pile, which arepassed through the lower-velocity, saturatedsediments below the water table and higher-velocity, dry soils above. The slanted reflectionevent increases in time at shallow depths because ofthe tilt of the battered pile. The botrom depth ofthe steel pile is identified where the reflectionevents are substantially reduced in amplitude at 35ft from the top of the boring. This would predict apile length of 31 ft, which is 8 ft shorter than the39-ft pile lengths from driving records. This effectis more apparent in the color video monitor displayin the field or from a color plot section.

Induction Field Method

In the induction field method (Wright,1979; Beattie, 1982), an alternating current flow isimpressed into a steel pile (or the reinforcing bar ina reinforced concrete pile) from which the currentcouples into the subsurface and finally to a returnelectrode (Figure 24). The return electrode can beanother pile, or it can be a pipe or piece ofreinforcing bar driven into the ground. A receivercoil suspended in a nearby boring is then used as asensor of the magnetic field induced by thealternating current flow in the pile. By plotting themagnitude of the induced voltage against the depthof the search coil, an indication of the length of thepile is provided, because the field is weaker belowth o:ry iies#ìrê'oasiee-i iinitaiion:of:i,iúS=me_iho-rÞis=rhaË

electrically continuous steel for its entire length,and the steel used must be accessible at the top toallow the electrical connection. Steel pile and(electrically continuous) reinforced concrete piledepths can be obtained from this type of survey.This method was not examined in this study butwill be investigated in the Phase I[ research.

CONCLUSIONS AND SUGGESTEDRESEARCH

The results of this research indicate that theultraseismic test has the broadest application of thesurface tests (no boring required) for determiningthe depths of unknown bridge foundations. Thesonic echo/impulse response tests were also usefulfor predicting depths of columnar (for example,partially exposed piles) foundation substructure; thesonic echo test was effective on more massiveabutments, but was not as conclusive as theultraseismic test. The bending wave methodshowed good agreement with sonic echo/impulseresponse tests of the slender timber piles, but use ofthe bending wave method is limited to very slenderfoundations. The dynamic foundation responsemethod showed only minor changes in vibrationresonances for footings with and without piles.Another method, SASW, was attempted onConnecticut DOT bridges, where access permitted.This method worked well for these shallow, moremassive wall abutments. The major limitation ofthe surface methods is that none of them can detectthe presence of piles underlying a buried pilecap ora more massive abutment or pier wall.Nevertheless, the new ultraseismic method was ableto determine the depths to the fïrst significantsubstructure change (i.e., footing on soils orpilecaps, bottoms of abutments, bottoms of piers),especially for shallow footings. However, noindication of secondary weak echoes from steelpilings underlying a buried pilecap or a moremassive abutment or pier wall was observable inthe data.

In the case of buried pile foundations, theparallel seismic test was found to have the broadestapplication for varying substructure and geologicalconditions, but it reqgires drilling- boreholes.äigþ:= quaiþ= <Íaa= may=

-ne*' obtaineri= b-i/F usragÞ

clamped geophone receivers in grouted, casedthe foundation substructure contain

borings instead of using only hydrophone receivers

that are suspended in a water-filled borehole. The

commercial borehole radar tool also workedrelatively well at sites where soils were

nonconductive. The borehole sonic feasibility tests

showed promise, but much research remains before

a fully evaluated commercial tool can be

developed.NCHRP Project 2l-5 (2)-researching and

developing equipment, field techniques, and

analysis methods for the technologies with the mostpromising immediate applications-is underway.The specific NDT methods to be further researchedand developed are the surface method of the

ultraseismic (including sonic echo/impulse response

and the bending wave methods) and the boreholemethods of parallel seismic and induction fields.Another FHWA study is being conducted (Olson et

al., 1996) in which the modal response of bridgesubstructures from dynamic excitations is beinginvestigated. This study will build on the dynamicfoundation response tests.

Summary evaluations of the promisingNDT methods in Phase I are presented in Tables 1

and 2 for the surface and borehole tests,

respectively.

ACKNO\ryLEDGMENTS

This research was supported by NCHRPProject 21-5 for the determination of unknown

13

depth of the bridge foundations. Mr. FarrokhJalinoos was the co-principal investigator duringthe original study. Dr. Marwan Aouad was

involved in the original sfudy and is the co-principal investigator for the continuation project.Olson Engineering, Inc., would like to thankNCHRP for this research support. Many helpfuldiscussions, recommendations, and technical

contributions were made during research by the

following individuals: Dr. Alfred H. Balch(Department of Geophysics, Colorado School ofMines) on the ultraseismic and the parallel seismicmethods; Drs. Stokoe and Roesset with graduateassistance from Dr. Shu-Tao Liao and Mr. Chih-Peng Yu and Messrs. Michael Kalinski and James

Bay (University of Texas at Austin) on theoreticalfinite element modeling and field assistance in NDTresearch. Dr. Glenn J. Rix (Georgia Institute ofTechnology, Atlanta) served as a consultant for a

feasibility evaluation of applying neural networkanalyses to the research problem. Mr. Peter J.

Loris (Loris and Associates, Boulder, Colorado)consulted in the selection of bridges and analysis oftheir dynamic response with the assistance fromMessrs. Donald Shosþ and John Ben Herrera(Loris and Associates); and Messrs. Dean Peterson(Range Engineering) and Ron Akin, an independentconsultant. Olson Engineering, Inc., would alsolike to thank the Connecticut DOT and Greiner,Inc. for their permission to publish the SASW data.

T4

Receiver

Defects andlntrusions

Gracks andB rea ks

Receiver

Bulbs and LengthD ete rm in atio n

ReceiverHammer Hammer Hammer

R efle ctio nfrom Defects

R efle ctio nfrom B u lbs

IIIIIIl

Figure 1. Sonic echo/impulse response test methods.

15

Iængth Calculation:Ât = 3.5 msV = 17,000 ftlsecLength of Pile = V x Ltl 2= 17,000 x 3.5 x Xi3 tZ =29.8 ft

U)II()o0)

1.000

0.500

0.000

-0.5002nd,3rd and 4th Echoes

-1.0Q0-¿x

Length Calculation:Af = 305 HzV = 17,000 flsec

(tt

Ë:ooooo.Êg?

8 200000

o! rooooo()ooo

1000 1500Frequency, Hz

Figure 2. Sonic echo test resulls (top) afier integration and exponential amplification, and auto power spectrum ofaecelerome,letoutp-u.t:from:imp-ulÍeerespols-e:te$Í:(botþn):fram:auexp.ostd:tir.tb-eæpjJc:o$a:btidggrin:E+ank¡plynr&-lçtado-,-

Length of Pile =Y I (2 x Âf) = 17,000 I (2x 305) = 27.9 ft

305 Hz , 30-5 Hz , 305 Hz

16

It

cenrcrline

SO

8õ-)

-3x10{

z4¿åt'oi 16x10-3

fbl 7ett

à0ë-1{)

¿4-t

-3x10{0246810 12 l6xt0'i

Time (sec)

Figure 3. Bridge substructure model for finite element analysis of sonic echo tests from the top of the drtiled shafi (top aand b) and theorelical resulÍs þottom a and-b).

lal

rfs

ccnterline

Time (sec)

texio'3

l7

Receivers

Reflectionfrom Defects

Bendine Wave Testfor Tim6er Piles

Figwe 4, B ending-wate test,tnethod*

t8

Velocity Calcularion:

^t = 1.41 ms

Rl-R2 = 42 inches = 3.5 ttBending lVave Velociry = Ri-R2 / Ar = 2,4g0 f/sec

a: Output of Accelerometer 1

1200

€ 600

s04.)(J

å -600

-1200

-4x1

2000

€ rooo

s0o()å-1000

-2000

-202Time,

t of Accelerometer 2

2Time, Sec

-4xL

1200

600

0

-600

Length Calcularion:At=22msV = 2,480 ft/sec

Length of Pile = V x Lr I 2 = 2,480 x 22 x tO3 tZ = 27.3 ft

o

(L¡

oI

-1200

-1xi l2Time, Sec

Figyre 5. Bending wave pile results for a timber pile of a Frar*town County, Colordo, timber bridgg.

19

flexurøl

3-G Accelerometers

HamrnerI:" " "*: cornpressíonøl wavegcnerøúíon

iltilt,iltlt,ilntnililt/I,II ABUTMENT/ililt PIERt,ilt

I I,n

ì r Ittt 1 I tlt

1\ 1t I lrrlt,,lrrrÌ{,rrrr i,r, r I I

ir{lilt1 1l\t\t ilt{lVllìltl,Át I

tl Uìt\t IllÌtrtt,t,ìt\r

t,uUitU

Direct SignaArrivals

Reflections From

Figryre' 6:' Ultrasetsnic: method;. v erticat,prtfrltng(V p_)- t est;,

20

SignalAnalyzer

Hammer accelerometers

tlcxurøl wøvcgcnerøtlon

øltcrnøtíve tcstItnc

Fig_wd;.-Abras-eiatßiemef Ind*-hartzar.xølsprofi ling:QlEJ_=test+

2t

N

-l>2t'.O"

22'

ELEVATION VIEW

IZBP 53 Steel PilesEstimated length = 25 ft

ULTRASEISMIC TESTSPILE FOUNDATIONPIER 4COORS BRIDGE

Notes:1) 1 Ultraseismic line2) 22 receiver locations3) Receiver 1 is at 21 ft below top of column4) Receiver 22 is at 10.5 ft below top of column4) Distance between receivers is equal to 6 inches5) Horizontal hits are at 20 ft below top of column(on the opposite side of the receiver locations)

Figure 8. Source/receiver layout for ultraseismic VP tests with impacts at the bottom of a Colorado bridge pier.

22

(/,:t

3000

4000

5000

ó000

7000

8000

7000

8000

9000

tmoo

I ì000

uta

tU

&"l**4ooo

-\"lI

g***_

t¡J

i*o*;f"Þ

ì 4000

15000

Figure 9. Ultraseismic VP records of flexural waves from horizontal impacts at the top of a bridge pier with a 3-tbhammer and 22 horizontal receiver locations in the same vertical plane on the south column (see Figure I for testgeometry).

S*e,r 4000

I 5000

23

N<-

PLAN VIEWo Source Locationso Receiver Locations

Steel pilingHP 12x53

ELEVATION VIE\ry

00 000()0()000()00000000000000000000Notes:1) I Ultraseismic line2) 32 receiver locations3) Source and receiveron opposite sides ofthewall4) Distance between receivers is equal to 1 ft5) Horizontal hits from automated source6) Source and receiver are moving together.

Figare- lA Sourceheceiver- Iayoat.þr-HP tesl,on-sideoe,of.a.pier.wall,of øeoJa-r.ada--Co_unty,bridggt

24

tions from thc top of the

f\øø

P reflection from the footing

25ØØ

C, \\.JELDg\L I \EV4. SX3COHHON SHOT GATHERSsHoïs, 1.Ø - 32.øGAIN, TRACE MAX29-SEP-94 16, t7

Figure lI. Example datafrom the HP dataset with the source and receivers moving simultaneousty atong the width of aColorad.o Caunty bridge.pierwall at l-fi iwenah.

Uìa

u_J

:l-at-

it

25

Jr.r__

---/-rr-

A ACQUISITION

TEST ELEMENT

Figure I 2, Source /receivet: arrøy- used,in SASN rneasaremenls'.

26

west AbuEnentBridge No.4896Flamden, Connecticut

PLAN VIEW

ELEVATION VIE\ry

SASW TestLocations:IR1-R2=3fttr R1-R2=6ftE R1-R2 = 12 ft

Fi gur e- 1 3. SASW test ar.røy -from. an- abutmsû- of- a- b ridg ø in. eorueetieut,

27

oEx

R1-R2 = 12R1-R2 = 6 ftR1-R2 = 3 ft

Velocity = 7700 fps

.tto.q-r'

*Ðc)o-()q)

(tt

FC)o(tlf)U)

Figure 14. S,,{SW test resubs fiom an abutment of a bridge in Connecticut, boltom depth : 9 rt, feference = top oÍ the

abutment).

28

Pier Abutment Columnar Substructure

Figure 15. Dynamic Foundation Response (DFR) test method.

29

SiqnalAn-alyzer

Superstructure

Geophone incased boring

F$p4æe1;&Ebg.llãJí¡ffi&¡nig'(ffiètewrærW

30

Concrete piles, L4" square,32 ftlong,l ft in cap

w

-rl>PARALLBL SEISMICBOREHOLE 4PILES FOLINDATIONBASTROPBRIDGE

vVertical hit location

Horizontal hit location

First receiver location forparallel seismic is at 2 ftfrom top ofborehole

Parallel seismic wasperformed at 1 ft intervalbetween receivers

Borehole 4

Last receiver location forparallel seismic is at 48 ftfrom top of borehole

2'

3'0"

T_

PLAN VIEW

Figur-elT.Sotr,celreceiyer.løyaui,þr.PS.test-from.a.highway"bridget

31

SP0

3øøø

6øøø

9øøø

t2øøø

r 5ø0ø

r gøøø

2røøø

24øøø

27øøø

3øøøø

33øøø

36øøø

39ø6ø

ø

3øøø

6øøø

9øOø

t2øøø

I 500€

Uìrgoøø )UIf

UJ

=l-

1500 ftlsec"2tøQø l_rJ

=24øøø l-

27øøø

3øøøø

33øøø

36øøø

39øø€

Soil velocity profilc - 850 ft/sec

C. \TEXAS9\HH SX3COI'IIION SHOT GATHERS

sHoTS 2ø- 47øGâtN AcC - t/lNDou 2øøøø s2t-JUN-94 tø 39

Figure 18. Parallel seismic fieM records (AGQ from a concrete pile foundation (see Figure 17 for geometry) lrom a 12-lbhorizontal hammer hit and horizontal component geophone recording.

Pile tip at 32 f t

' ilIl:,ll

32

fast P-wave velociEy

L4in ptle below water-tabl e -, t1¿ólJð i.le cip

at 33 fr.SP,

3øøø

6øøø

9øøø

L2øøø

r5øøø

L8øøø

2Løøø

24øøø

27øøø

3øøøø

33øøø

36øøø

39øøø

haleøøø

ressionalfile

fc/secL2øøø

2Løøø

24øøø

27øøø

lJ't

l

LrlEt-

36øøø

39øøøbe waves

1660'ft/sec

C. \PS4\VH. SX3COMI1ON SHOT GÊTHERSsHoTs, 2.ø - 48.ØGAIN. TRâCE MâX27-HÊR-95 ø8,32

Figure 19. Parallel seismic field records (AGC) from a concrete pile fourúation (see Figure 17 for geometry) from a l2-tb.vertical hammer hit and hydrophone receivers.

33

Dillling/Orientolion/rConlrolSyslem

/

Receivel

Figure 20. Borehole Sonic (BHS) test method.

34

SIGNALANALYZER

RADAR ANTENNA

ReflectedRadar Signals

UnreflectedRadar Signals

ABUTMENT

Figwe2 l . Bo re hole--Radør (B$R) :tett:ttrythod;

35

\ryrr*

€Èrc)q

r,ñ i*.lçlú)t6ülÉl

12 BP 53 Steel PilesEstimated length =39 ft (from groundsurface)

120 MhzMonostaticRadar Antenna

BOREHOLE RADAREAST STEEL PILEALABAMA BRIDGE

ELEVATION VIEW

Figure 22. Field test layout lor the BHR test lrom the east battered BP steel pile of a steel bridgç. in Alahatna.

36

Depth (ft)o'or--:

.â

(.)

18.7

37.5

56.2-Ì* tr

t

pile end battered pile

l'r.¡..iol'. 3l'. -:' :l*:tì¡ruEr+ añ:----tì .-ê. -, ii'

Figur+8,- BHß=fuÍn recards--from.thøeastr bollør¿d-steelB,P piløof.a,stcelbridgøinAloþøø-

37

CHOOSE TAPPING TO

MAXIMISE cuBRENr A

0'5m

-l

RETURNELECTROOE

PILE OFINTERESÍ

SEARCH

COI L

PV.C. TUBE AS

DRILL HOLE LINING.

F igprc+24+ Itútøian:ßields$ß) rt e$nurhadE(ø-fræBeeie+, 1982) x-.

38

TABLE 1 Summary Evaluation of the Applicable Surface NDT Methods

Ability to ldentifyFoundation Parameters

Sonic Echo (SE)/ImpulseResponse (IR) Test(Compressional Echo)

Bending'Wave (BW)Test(Flexural Echo)

Foundation Parameters:Depth of Exposed PilesDepth of Footing/CapPiles Exist Under Cap?Depth of Pile below Cap?Geometry of SubstructureMaterial ldentification

Fair-ExcellentPoor-Good

N/AN/ANiAN/A

Fair-ExcellentPoor-Fair?

N/AN/AN/AN/A

Access Requirements:Bridge SubstructureBorehole

YesNo

YesNo

Subsurface Complications :Effect of soils on response Low-Medium Medium-High

Relative Cost Range:Operational CoslSSU-Equipment Cost

$1,000-$1,500$1s,000-$20.000

$1,000-$1,500$15,000-$20,000

Required expertise:Field AcquisitionData Analysis

TechnicianEngineer

TechnicianEngineer

Limitations: Most useful for columnar ortabular structures. Responsecomplicated by bridgesuperstructure elements.Stiff soils and rock limitpenetration.

Only useful for purelycolumnar substructure.Response complicated byvarious bridgesuperstructure elements,and stiff soils may showonly depth to stiff soillayer.

Advantages: Lower cost equipment andinexpensive testing. Datainterpretation for pilefoundations may be able tobe automated using neuralnetwork. Theoreticalmodeling should be used toplan field tests.

Lower cost equipment andinexpensive testing.Theoretical modelingshould be used to planfield tests. The horizontalimpacts are easy to apply.

"SSU = Substructure Unit cost is for consultant cosf. only - DOT to supply !-2-people,. N/A=Not.App-licable.

39

TABLE I Summary Evaluation of the Applicable Surface NDT Methods (Cont.)

Ultraseismic (US)Test (Compressional andFlexural Echo)

Spectral Analysis ofSurface Wave (SASW) Test

Surface Ground PenetratingRadar (GPR) Test

Fair-ExcellentFair-Excellent

N/AN/AFairN/A

N/AFair-Good

N/AN/A

Poor-GoodGood

N/APoor

Fair-PoorPoor

Poor-GoodPoor-Fair

YesNo

YesNo

YesNo

Low-High Low High

$1,000-$1,500$20,000-$2s,000

$1,000-$1,500$15,000-$20,000

$1,000-$1,500$30,000+

TechnicianEngineer

Technician-EngineerEngineer

Technician-EngineerEngineer

Cannot image piles belowcap. Difficult to obtainfoundation bottomreflections in stiff soils.

Cannot image piles belowcap. Use restricted tobridges with flat, longeraccess for testing.

Signal quality is highlycontrolled by environmentalfactors. Adjacentsubstructure reflectionscomplicate data analysis.Higher cost equipment.

Lower equipment and

testing costs. Can identifythe bottom depth offoundation inexpensivelyfpr a large class ofbridges. Combinescompressional and flexuralwave reflection tests forcomplex substructures.

Lower equipment and

testing costs. Also showsvariation of bridge materialand subsurface velocities(stiffnesses) vs. depth and

thicknesses of accessibleelements.

Fast testing times. Canindicate geometry ofaccessible elements and

bedrock depths. Lowertesting costs.

40TABLE 2 Summary Evaluation of the Applicable Borehole NDT Methods

Ability to ldentifyFoundation Parameters

Parallel Seismic (PS)Test

Borehole Radar (BHR)Test

Induction Ficld (IÐTest

Foundation Parameters:Depth of Exposed pilesDepth of Footing/CapPiles Exist Under Cap?Depth of Pile below cap

Geometry of SubstructureMaterial Identification

Good-ExcellentGoodGood

Good-ExcellentFair

Poor-Fair

Poor-ExcellentPoor-GoodFair-GoodFair-Good

Fair-ExcellentPoor-Fair

None-ExcellentN/A

None-ExcellentNone-Excellent

N/APoor-Fair

Access Requirements:Bridge SubstructureBorehole

YesYes

NoYes

YesYes

Subsurface Complications:Effect of soils on response Medium High Medium-High

Relative Cost Range:Operational CoslSSU*Equipment Cost

$1,000-$1,500$15.000-$25.000

$1,000-$1,500$35,000+

$1,000-$1,500$10,000

Required expertise:Field Acquisitior/SSU'Data Analysis

Technician-EngineerEngineer

EngineerEngineer

TechnicianEngineer

Limitations: Difficult to transmitlarge amount ofseismic energy frompile caps to smaller(area) piles.

Radar response is highlysite dependent (verylimited response inconductive, clayey, salt-water saturated soils).

It requires thereinforcement in thecolumns to be

electrically connectedto the piles underneaththe footing. Onlyapplicable to steel orreinforced substructure.

Advantages: Lower equipment andtesting costs. Candetect foundationdepths for largestclass of bridges andsubsurface conditions.

Commercial testingequipment is nowbecoming available forthis purpose. Relativelyeasy to identifyreflections from thefoundation; however,imaging requires carefulprocessing.

Low equipment costsand easy to test. Couldwork well tocomplement PS testsand help determine piletype.

-SSU = Substructure Unit cost is for consultant cost only - DOT to supply 1-2 people + does not

include drilling costs.N/A = Not Applicable.

BIBLIOGRAPHY

Aouad, M. F., "Evaluation of Flexible Pavementsand Subgrades using the Spectral Analysis ofSurface Waves Method," Dissertation submitted inpartial fulfillment of the Doctor of Philosophydegree, University of Texas at Austin (1993).

Aouad, M. F., Olson, L. D., and Jalinoos, F.,"Determination of Unknown Depth of BridgeAbutments using the Spectral Analysis of SurfaceWaves (SASW) and Parallel Seismic (PS) TestMethods," Second International Conference onNDT of Concrete in the Infrastructure, Nashville,TN (1e96).

Baguelin, F., Corte, J.F., and Levillain, J. p.,"Methods of inspection and non-destructive testingof existing bridge piers and foundations, "Fédération Nationale des Trav Publ & des SyndÃff , Travaux N544 (1980) (in French) pp. 48-59.

Beattie, G. J., "Pile Length DeterminationInduction Field Method," Technical Note 5-3,Works Consultancy Services, Central Laboratories,Gracefield, Lower Hutt, New Zealand (1932).

Brooks, R., "Determining In-Situ Timber pileLength Using Stress Wave," Timber BridgeInitiation Special Projects Program, USDA ForestService, Morgantown, WV (1992).

Chang, D. W., "Nonlinear Effects on DynamicResponse of Pavements Using the NondestructiveTesting Techniques," Dissertation submitted inpartial fulfillment of the Doctor of Philosophydegree, University of Texas at Austin (1991).

Davis, A. G. and Dunn, C. S., "From Theory toField Experience with the Non-Destructive Vibra-tion Testing of Piles," Proceedings, Institute ofCivil Engineers, Part 2 (Dec. 1975) pp. 571-593.

Douglas, R. A. and Holt J. D., "DeterminingLength of Installed Timber Pilings by Dispersive\\?avu=PropagaiionÞiilêìhuds#Repnn*€ehtp,r-.-f(tìr:

4L

Transp ortation Engine ering Studies, North CarolinaState University (June 1993).

Elias, V. 4., "Strategies for Managing UnknownBridge Foundations," Report FHWA-RD092-030,Washington, D.C. (January 1992).

Heisey, J. S., Stokoe, K. H., Hudson, W. R., andMeyer, A. H., "Determination of In Situ ShearWave Velocity from Spectral Analysis of SurfaceWaves," Research Report 256-2, Center forTransportation Research, University of Texas atAustin (1982).

Jalinoos, F. and Olson, L.D., "Determination ofUnknown Depth of Bridge Foundations UsingNondestructive Testing Methods," Structural,Materials Technology, NDT Conference, SanDiego, CA (1996) pp.9l-97.

Jalinoos, F., Olson, L.D., and Aouad, M.F.,"Determination of Unknown Depth of BridgeFoundations Using Two Nondestructive SeismicMethods," SAGEEP 96 Conference, Keystone, CO(1996) pp. 91,-97.

Jalinoos, F., Aouad, M.F., and Olson, L.D.,"Three Stress-Wave Methods for the Determinationof Unknown Pile Depths," .srress Waves '96Conference, Orlando, FL (1996).

Kolsky, H., Stress Waves in Solids, DoverPublishers., Inc., NY (1963).

Koten, H. van and Middendorp, P., "Testing ofFoundation Piles," Heron, joint publication of theDepartment of Civil Engineering, Delft Universityof Technology, Delft, The Netherlands, andInstitute TNO for Building Materials and Sciences,Rigswijk (ZH), The Netherlands, vol. 26, no. 4(1981).

Liao, S. T., "Nondestructive Testing of Piles,"Dissertation submitted in partial fulfillment of theDoctor of Philosophy degree, University of Texas^+:.¡l $à+¡L* / 1 aì¡ì¡lS+4 LF.r t¡ ù_Lilli\'I-7 7-¡rt r

42

Nazarian, S. and Stokoe, K. H., "NondestructiveTesting of Pavements Using Surface Waves,"Transportation Research Record 993,Transportation Research Board, Washington, D. C.(1e84).

Novak, M., "Dynamic Stiffness and Damping ofPiles," Canndian Geotechnical Journal, Vol II(1976) p. 574.

Novak, M. and Aboul-Ella, F., "ImpedanceFunctions of Piles in Layered Media," Journal ofEngineering Mechanics, ASCE, Vol. 104, No. 3(1978) pp. 643-661.

Olson, L.D., Jalinoos F., and Aouad, M.F.,"Determination of Unknown Subsurface BridgeFoundations," NCHRP Project 21-5, UnpublishedFinal Report, Transportation Research Board,National Research Council, Washington, D.C.(lees).

Olson, L.D. and Aouad, M.F., "Dynamic BridgeSubstructure Evaluation and Monitoring System,"approved FHWA contracr No. DTFH61-96-C-00030, Washington, D. C. (1996).

Rix, G. J., "Experimental Study of FactorsAffecting the Spectral Analysis of Surface WavesMethod," Dissertation submitted in partialfulfillment of the Doctor of Philosophy degree,University of Texas at Austin (19S8).

Stokoe, K. H., II, Nazarian, S., Rix, G. J.,Sanchez-Salinero, I., Sheu, J.C., and Mok, y.J.,"In Situ Seismic Testing of Hard-to-Sample Soilsby Surface Wave Method," proceedings,

Earthquake Engineering and Soit Dynamics II,ASCE Specialty Conference, UT (l9BB) pp. 26a-278.

Stain, R. T., "Integrity Testing," Civil Engineering(1982) pp. 53-72.

Watson, R., "Hundreds of Bridges to UndergoScour Tests," New Civil Engineer, Telford(Thomas) Limited, London, England (1990).

Wright P.D., "Rangitaiki River Bridge, Te Teko,Pile Length Investigatiorts," Cental LaboratoriesReport 5-79/9, Ministry of Works andDevelopment, Lower Hutt, New Zealand (1979).

TRANSFORTATION RESEARCXI BOARDNational Resea¡ch Courcil

21O1 Constitution Arænr¡c, N.W.Washington,DCZX1S