6TH INTERNATIONAL CONFERENCE ON DAM ENGINEERING C.Pina, E.Portela, J.Gomes (ed.

) Lisbon, Portugal, February 15-17, 2011

TRIDIMENSIONAL GPR MAPPING IN THE UPSTREAM FACE OF AN RCC DAM

Isabella F. R. Figueira*, Luiz Alkimin de Lacerda*, Ernesto Goldfarb Figueira, Emerson Luiz Alberti
Instituto de Tecnologia para o Desenvolvimento (LACTEC) Br-116, Km 98 s/n Centro Politcnico da UFPR, 81531-980 Curitiba, Brasil e-mail: isabella.figueira@lactec.org.br, webpage: http://www.lactec.org.br/pt/
*

Keywords: Dams, Concrete, Ground Penetrating Radar, Non Destructive Tests. Abstract. Mapping with geophysical techniques such as the Ground Penetrating Radar (GPR) is based on the propagation of electromagnetic waves and has a wide range of applications in civil engineering. This work presents a non-destructive investigation with GPR in the conventional concrete face of a roller compacted concrete (RCC) dam. The objective is the identification of subsurface anomalies, such as niches of higher porosity in the conventional concrete, whose thickness is not greater than 50 cm. The acquisition of radargrams was carried out with a 2 GHz antenna with dual polarization, which allowed assembling a three-dimensional subsurface image from a single direction reading path. The images were treated with a specific package of filters for concrete surveys. After this post-processing, the group of images was integrated to form a block diagram. The interpretation of the whole block was performed with a controlled thickness slicing procedure for identification of anomalies within sound concrete. A general view of the investigated area is presented with the indication of anomalous regions which may require a localized repair. 1 INTRODUCTION The integrity, load capacity and service life of concrete dams is directly correlated to the quality of constituent materials and construction methods. The common practice of investigation of concrete in civil structures uses invasive methods to extract samples for laboratory tests. Besides being destructive, these methods cannot consistently reveal abnormalities within the analyzed structure. Civil engineering has recently started using non destructive geophysical techniques for the inspection of hydraulic structures. However, investigation of anomalies such as niches of higher porosity, cracks and damaged areas within concrete is not trivial. With geophysical surveys one can determine concrete conditions, detecting anomalous areas and mitigate the damage. The Ground Penetrating Radar (GPR) is a geophysical tool whose principle is based on the propagation of electromagnetic (EM) waves. It is a non destructive method of investigation used for different purposes, such as to detect the thickness of slabs1, the integrity of masonry structures2, the integrity of reinforcement in bridge decks3, buried

Esteio Engenharia e Aerolevantamentos S.A. Elejor Centrais Eltricas do Rio Jordo S.A.

Isabella Figu ueira, Luiz Alkimin A de La acerda, Ernest to Goldfarb and a Emerson L Luiz Alberti.

pipes4, archaeolog gical surve eys5 and o others. In particular, it has bee en used to o detect 6 embedd ded objects and voids in concrete e with high h resolution n. Alth hough the use u of GPR is widespr read in geological and d geotechni ical areas, this t tool is not y yet a usual practice in n concrete d dams or oth her hydraulic concrete e structures. Thus, the present study is also an initial i surv vey, which will form the t basis fo or the defin nition of EM par rameters for a better acquisition i in future stu udies. This s paper pres sents a tridi imensional high resolu ution GPR mapping of f a selected d area in the ups stream face e of the RC CC dam of f Santa Clara in Para an state - Brazil. Th he study aimed to identify y the possi ible presen nce of ano omalies in the subsur rface layer r of the conven ntional conc crete, whos se thickness s is less th han 50cm. The T mappin ng was carried out with a 2 GHz dua al polarizati ion antenna a allowing the direct formation f o of a 3D sub bsurface image f from single e direction reading r pat th. The trid dimensional l images we ere integrat ted after post-processing th he read sect tions with a specific package of filters f for co oncrete sur rveys. rpretation of o the image es was perf formed afte er the analy ysis of each h section rad dargram Inter with centimetric slicing, se earching f for possible subsurfa ace anomal lies within n sound concret te. 2 GPR R GROU UND PENE ETRATING G RADAR R

The Ground Pe enetrating Radar non n destructiv ve techniqu ue consists in the con ntinuous emissio on of electr romagnetic c waves and d receiving g reflected signals to obtain sub bsurface images of the structure. The equipment t used in th he present survey is shown s in Figure 1. It has an interesting d polariz zed antenna a (see Figu ure 2), whic ch allows p erforming the data feature such as a dual acquisit tion in only y one direc ction. The 2 GHz freq quency ante enna allows s a high resolution investig gation, but with limite ed range, up p to 50 cm in concrete e. An e electronic distance d meter EDM was attach hed to the antenna, a wh hich is bas sically a wheel a acting as a trigger for the emissi ion of equid distant elec ctromagneti ic pulses, allowing a a regula ar data acqu uisition.

Fi igure 1: GPR R with dipole 2 GHz antenn na.

Isabella Figu ueira, Luiz Alkimin A de La acerda, Ernest to Goldfarb and a Emerson L Luiz Alberti.

adar wave propagatio p on in concr rete 2.1 Ra One of the mos st important t points in a an investigation with EM radar w waves is the e choice of freq quency. This definition n is compl lex and inv volves a "c choice" be etween capa acity of penetra ating the ma aterial and signal reso lution for identifying characteris stic targets7.

Figure e 2: Layout of f receivers (R RX) and transmitters (TX) in a dipole an ntenna.

Thre ee major factors f det termine th he choice of frequen ncy for th he survey: spatial resoluti ion, clutter ring and depth d of pe enetration. These fact tors are inf nfluenced by b basic charact teristics of the t materia al as conduc ctivity (), permittivit ty () and h heterogeneit ty. Conc crete is a dielectric d material m an nd can beha ave as a co onductor o or insulator r of EM waves. The dielec ctric (constant) perm mittivity of concrete varies from m about 5 to 12. Normal lly, concret te is not con nductive7 a and the atten nuation of EM signals s is given by,

1 640 0,5 (dB d / m)

(1)

The conductivi ity is direct tly proporti ional to fre equency (f) of the wav ves. Theref fore, the attenua ation is high her as the fr requency in ncreases. Sinc ce the radar r provides the time of the EM wa ave propaga ation betwe een transmi itter and receive er, it is imp portant to know the wave spee ed in order r to estima ate the dep pth of a detected target. Th he equation n of the EM M wave velo ocity is give en by7, v = c -0,5 (2) where c = 0,3 m/ns m is the speed of l light. For radar freque encies wav ve velocity can be conside ered constan nt. Assuming a permit ttivity = 8, 8 one has v = 0,106 m m/ns. It is s argued th hat for a good g vertic cal resolut tion (r) for r a reflecte ed EM pu ulse, the 7 wavelen ngth () sh hould be r > /4 As = v / f, then, f>0 0,0264 / r (GHz) ( (4) refore, theo oretically, to t have a re esolution of 2,64 cm in i the imag ge, frequency must Ther be grea ater than 1 GHz. G To minimize m no ise in the ra adar image, the domin nant wavele ength of the EM M pulse sh hould be at a least 4 times larg ger than th he characte eristic size e (h) of 3 (3)

Isabella Figu ueira, Luiz Alkimin A de La acerda, Ernest to Goldfarb and a Emerson L Luiz Alberti.

heterog geneity7, > 4h Thus, f < 0, ,0264 / h (GHz) (6) For example, for f h = 1,2 2 cm, the f frequency must m be les ss than 2.2 2 GHz. Fin nally, an aspect t that should d not be ove erlooked is that when the antenna is brough ht into contact with a surfac ce area for r survey the e pattern o f the EM field f changes consider rably, and reduces the freq quency of th he transmit tted signal. The equati ion below was w sugges sted to estim mate the effectiv ve frequenc cy (fr) signa al in a mate erial such as s concrete, fr = f [2/(1+)] 0,5 3 ONTROLLE ED FIELD D TEST CO (7) (5)

Befo ore applyin ng the GPR system in Santa Clar ra dam, a controlled c t test was per rformed to evalu uate its cap pacity for de etecting ano omalies of different si izes embed dded in conc crete. The constructio on plan con nsisted of a concrete block b built in two step ps with a minimum m size for r the GPR readings. The T block w was 5,20 m long, 1 m wide and d had thick kness of 0,50 m m. Eighteen n anomalies s were inse erted in a symmetric c configura ation, as sh hown in Figure 3. The an nomalies ha ad differen nt areas an nd thicknes sses and co onsisted of f empty spaces filled with h aggregate es size 1 a and 2, whic ch were co overed with h a plastic film to ensure its higher porosity. p The T sizes of f the anomalies were printed in the concret te using regular molds, as shown in Figure F 4: Pla acement of f the anoma alies in the concrete bl lock. .

Figu ure 3: Schema atic plan of th he concrete bl lock with the anomalies la ayout.

Isabella Figu ueira, Luiz Alkimin A de La acerda, Ernest to Goldfarb and a Emerson L Luiz Alberti.

a)

b)

Figure 4: Placement P of f the anomalie es in the conc crete block.

er the GPR readings, the t radargra ams were post-proces p sed with a package of f filters: Afte high an nd low pas ss bands in n the vertic cal and horizontal dire ections, ba ackground removal r and rec covery gains. Two o radargram ms are show wn in Figur re 5 and Fig gure 6 alon ng with a s section view w of the block a at the read ding line. The presen nce of all anomalies could be detected and are indicate ed by yellow arrows in n each figu ure. Larg ger anomali ies show ve ery strong h hyperbolic signals in the images s. Small an nomalies are still l detectable e, but the signals are n not so evid dent, especially from t the less thic ck ones. In gene eral, thicker r anomalies s show stron nger signal ls. Alth hough in th he initial plan all a anomalies were place ed 10 cm in depth, during constru uction some e deviations s occurred, as indicate ed by the ra adargrams.

F Figure 5: Rada argram of sec ction A with 5 cm thick an nomalies placed at a depth h near to 10 cm m.

Isabella Figu ueira, Luiz Alkimin A de La acerda, Ernest to Goldfarb and a Emerson L Luiz Alberti.

F Figure 6: Rad dargram of se ection B with h 10 cm thick anomalies placed at a dep pth near to 10 0 cm.

4 APP PLICATIO ON OF GP PR IN THE E UPSTRE EAM FACE E OF SANT TA CLARA RCC DAM ey in Santa Clara was p performed in a limited d area of th he upstream m face of The GPR surve m (Figure 7), in order to t evaluate the conditi ion of the conventiona c al concrete. . the dam

Figure 7 7: Location in n the upstream m face where the GPR sur rvey was carri ied out read ding paths in the right side.

The selected ar rea was dis scretized w with 75 lines, 2,20 m long l each o one, with a 0,10 m g. Once the e lines were e marked, t the GPR readings wer re done swe eeping the antenna spacing over ea ach line, alw ways in the same direc ction (Figur re 8). The first GPR profile p (0,00 m) was r read at the level l of the crest of the e dam, in th he same height of the concrete pavem ment. The last profile e (7,50 m) was read 2,20 m be elow the normal reservoir water w level. .

Isabella Figu ueira, Luiz Alkimin A de La acerda, Ernest to Goldfarb and a Emerson L Luiz Alberti.

Figu ure 8: GPR su urvey in the u upstream face e with a 2 GH Hz dipole ante enna.

4.1 Results Alm most all prof files showed d the presen nce of faint t irregular and a truncate ed reflectors s, which mally assoc ciated with surface irr regularities in the conc crete surfac ce and the medium are norm heterog geneity. The first two profiles p at th he levels 0 0.0 and 0.10 0 m show striking s refl flectors in the form of regu ularly space ed hyperbol les. This is s a clear ev vidence of the t concret te reinforce ement at the pav vement leve el (Figure 9) ).

F Figure 9: Rada argrams at the e pavement le evel in the crest of the dam m.

It is noticeable e that beyon nd the dep th of 0,50 m the sign nal quality is poor du ue to the high at ttenuation of the 2.0 GHz EM waves in the concre ete. Beyond d this dept th noise masks the presen nce of refle ectors, mak king data interpretati ion even m more difficult (see Figure 10). Coincidently, th his depth is also ass sociated to o the trans sition betw ween the conven ntional conc crete and RC CC.

Isabella Figu ueira, Luiz Alkimin A de La acerda, Ernest to Goldfarb and a Emerson L Luiz Alberti.

Fig gure 10: Rada argram showi ing noisy data a beyond the depth of 0,50 0 m.

ome profile es it was no oticed the p presence of f isolated hy yperboles o of small am mplitude, In so which a are associa ated with th he occurren nce of met tallic fragm ments inside e the conventional concret te as shown n in Figure 11.

Figure 11: Presence P of m metal fragmen nts in the face e of the dam.

In specific rad dargrams th here was t the occurre ence of str riking hype erboles of greater magnitu ude, which h are relat ted to the presence of metal elements e u used to fas sten the climbin ng forms du uring the co oncrete wor rk. The form ms are show wn in Figur re 12 along gside the radargr ram of the correspondi c ing level.

Isabella Figu ueira, Luiz Alkimin A de La acerda, Ernest to Goldfarb and a Emerson L Luiz Alberti.

Figure 1 2: Example of o striking hy yperboles (rig ht) resulting from metal elements used d to fix the for rms (left) left embedded d in concrete. which were l

The GPR prof files along g the lines just below the nor rmal reserv voir level showed increased presenc ce of irregu ular and tr runcated re eflectors, due d to the irregular concrete c surface e with show wing aggreg gates. Wear r in this ar rea is partic cularly high h due to th he water level fluctuations (Figure 13). None of these hiper rboles detected in the radargram ms are an indicati ion of inter rnal flaws.

Figure 13: Normal wat ter level zone e showing gre eater wear.

For a better vi isualization n of the pro ofiles at di ifferent dep pths from t the mapped d area a specific c processin ng procedu ure called C-Scan was w performed8, con nsisting of a data interpolation base ed on posi itive and n negative peaks of th he electrom magnetic re eflectors generat ted by the GPR. G

Isabella Figu ueira, Luiz Alkimin A de La acerda, Ernest to Goldfarb and a Emerson L Luiz Alberti.

The product of f this inter rpolation is s a block, which after a 5 cm s slicing, allo ows the ication of highly h cond ductive are eas, depicte ed in warm m colors, an nd highly resistive r identifi areas in n blue (Figu ure 14). The metal elem ments used for fasteni ing the clim mbing form ms are clea arly shown at well defined d and equidistant spots s of high co onductivity between 40 cm and 4 45 cm depth h.

Figu ure 14: 5 cm C-Scan C of the e mapped area a. The circled d areas indica ate anomalies at various de epths.

Void ds or niche es of higher r porosity with high water quan ntity would d also be in ndicated with red colors. In n the surve eyed volum me the circle ed red spot ts detected from depth h 20 cm to 35 cm m, slightly above the 5,00 5 m leve el, is an evi idence of one or more e anomalies. Look king at the e radargram ms between n levels 4,60 m and d 4,80 m, in Figure 15, the presenc ce of a cont tinuous pla ane anomaly y in a diagonal positio on with res spect to the e face of the dam m can be ob bserved. Th he anomaly y goes from m 15 cm to 35 cm dep pth (Figure e 14 and Figure 16 detail).

Figure 15: Radargrams R at a levels 4,60 m, 4,70 m an nd 4,80 m wit th the anomal ly indicated.

a the C-sca an shows th he anomaly y in detail, which w is ov ver 80 cm long and A closer look at ranges from depth h 15 cm to 35 3 cm.

10

Isabella Figu ueira, Luiz Alkimin A de La acerda, Ernest to Goldfarb and a Emerson L Luiz Alberti.

Figure 16: Detail of th he C-scan wi ith identified anomaly.

CO ONCLUSIO ONS

The GPR has proved to o be a use eful comple ementary non n destruc uctive tool for the investig fication of internal gation of th he upstream m face of co oncrete dam ms, allowing g the identif anomal lies with hig gh precisio on and resol lution. It is a relatively y quick me ethod of da ata acquisiti ion, which can be app plied to larg ge areas or for in nvestigation of specifi fic areas wit th single or r periodic assessments a s. In th he examine ed area of th he upstream m face of Santa S Clara a RCC dam m, the GPR profiles reveale ed the prese ence of a pl lane anoma aly inside th he conventional concr rete. Its loc cation is above the norma al reservoir r level and d the implementation n of a refi ined inspec ction or correcti ive interven ntion was not n justified d. The GPR rad dargrams also reveal ed the presence of metal ele ements ins side the conven ntional conc crete zone, regularly s spaced in vertical v and d horizontal l directions s, which were us sed for fastening the climbing c for rms during the dam co onstruction n. 6 ACKNOWLE EDGEMEN NTS

The authors wo ould like to o thank ELE EJOR Ce entrais Hidr reltricas d do Rio Jord do S.A. for the financial su upport for carrying ou ut this resea arch, and CNPq C for th he research support (LI 10/ 1233240-8) ) through la aw 8010/90 0. 7 REF FERENCE ES [1] PER REZ-GRAC CIA V., DI CAPUA D D., GONZA ALEZ-DRIG GO R., PUJ JADES L. G.. G GPR Res solution in NDT Stud dies of Str ructural Ele ements: Ex xperimental Methodolo ogy and Exa amples. ND DTCE09 Non-Destruc N ctive Testin ng in Civel l Engineerin ng Nantes, France, 200 09. [2] HA AMROUCH HE R.; KLY YSZ1 G.; BA ALAYSSAC C J.; AURE ENS S.; RH HAZI J.; BA ALLIVY G.; ARLIGUI IE G.. Num merical Mod deling of Ground-pen G netrating Ra adar (GPR) for the Inv vestigation of Jointing g Defects in Brick Masonry Structures. S NDTCE0 09 NonDestructive Te esting in Civ vel Enginee ering Nantes, France 20 009. [3] MA ASER K. R.. R Integratio on of Groun nd Penetrati ing Radar and Infrared d Thermogra aphy for Bri idge Deck Condition Evaluation n. NDTCE E09 Non-D Destructive Testing in Civel Eng gineering Nantes, N Fran nce 2009. [4] PORSANI J.L L.; BORGES S W.R.; RO ODRIGUES S.I.; HIOD DO F.Y.. O S Stio Contro olado de a do IAG/U USP: Instala ao e Resu ultados GPR R 2D-3D. R Revista Brasileira de Geofsica Rasa Geofsica 24(1 1): 49-61 20 006. [5] CO OLLA C.. Diagnosi D di d Muratura a, Calcestru uzzo e Leg gno Tramite te Georadar r per la Val lutazione e Conservazi ione di Cos struzioni e Infrastruttur I re. II Giorna ale delle Pr rove non Dis struttive Mo onitoraggio Diagnostica a, Itlia 200 09.

11

Isabella Figueira, Luiz Alkimin de Lacerda, Ernesto Goldfarb and Emerson Luiz Alberti.

[6] SPORER A.. Inspeo no Destrutiva de Estruturas de Concreto por Georadar (GPR). Revista Abendi n 32. Artigo Tcnico, Brasil 2009. [7] PADARATZ, I.J.; FORDE, M.C.. Influence of antenna frequency on impulse radar surveys of concrete structures. 6th Int. Conf. on Structural Faults and Repair, London, 2, 331-336. 1995. [8] GRED 3D Copyright 2008

12