This article was downloaded by: [University of Connecticut]On: 11 October 2014, At: 08:32Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Structure and Infrastructure Engineering:Maintenance, Management, Life-Cycle Design andPerformancePublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/nsie20

Chloride profiles in a coastal bridgeManuela Salta a , Ana Melo a , João Ricardo a & Afonso Póvoa ba Laboratório Nacional de Engenharia Civil , Avenido do Brasil 101, 1700-066 , Lisboa ,Portugalb Estradas de Portugal, S.A., Ed. Quinta das Varandas, Avenida Cónego Urbano Duarte ,3030-215 , Coimbra , PortugalPublished online: 18 Aug 2010.

To cite this article: Manuela Salta , Ana Melo , João Ricardo & Afonso Póvoa (2012) Chloride profiles in a coastal bridge,Structure and Infrastructure Engineering: Maintenance, Management, Life-Cycle Design and Performance, 8:6, 583-594, DOI:10.1080/15732479.2010.505378

To link to this article: http://dx.doi.org/10.1080/15732479.2010.505378

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Chloride profiles in a coastal bridge

Manuela Saltaa*, Ana Meloa, Joao Ricardoa and Afonso Povoab

aLaboratorio Nacional de Engenharia Civil, Avenido do Brasil 101, 1700-066 Lisboa, Portugal; bEstradas de Portugal, S.A.,Ed. Quinta das Varandas, Avenida Conego Urbano Duarte, 3030-215 Coimbra, Portugal

(Received 2 November 2008; accepted 1 June 2010; published online 18 August 2010)

This article addresses the results obtained in the field investigation performed on Barra Bridge, on the PortugueseAtlantic coast, to determine the degree of chloride transport into the concrete. A critical analysis is performed toquantify (i) the influence of the degree of exposure to the chloride environment, (ii) the average values of theapparent chloride diffusion coefficients, (iii) the surface chloride content estimated from the chloride profiles and (iv)the uncertainties associated to these concrete variables, which are expressed in terms of their covariance.Considerations are also given to the relevance of these results for (i) corrosion modelling of reinforced concrete,which aims to predict the time to corrosion initiation in concrete structures in marine environments and (ii)improving maintenance planning.

Keywords: chloride threshold; chloride transportation; corrosion initiation; reinforced concrete

1. Introduction

Corrosion due to chloride penetration is the mostimportant degradation process in reinforced concretestructures, in both coastal and marine environments.The time for corrosion initiation in reinforced concretestructures will depend on the rate of penetration ofchlorides into the concrete and on the chloridethreshold value. The chloride threshold value in theconcrete depends mainly on the type of binder in theconcrete and also on the access of oxygen. Increasedcover depth and the level of saturation in the concretecan limit the oxygen access to the steel surface givinghigher threshold values. For the same concretecomposition, different threshold values are expectedto be reached, either in permanently saturated concreteor in concrete under airborne exposure (Glass andBuenfeld 1995).

Several factors can affect the rate of chloridepenetration into concrete; some are related to theproperties of the materials, mainly the porous structureof concrete and the binding capacity of the cementpaste, whereas others are related to local parameters,such as, the heterogeneity of concrete cover and themicroclimatic environmental conditions. The extrememicroclimatic environmental conditions, given bydifferent concrete saturation conditions, by tempera-ture, by location as regards sea winds and rainfall andby the distance from seashore, which can change in

different parts of the marine concrete structures,mainly govern the chloride transport behaviour(Bamforth and Price 2000, Song et al. 2008). However,comparatively few data has been reported in theliterature concerning the effect of location onthe chloride transport into the concrete (Wood andCrerar 1997, Fluge 1997), for structures in coastalenvironments.

For marine concrete structures, it seems that thesplash zone near sea level is the easiest location forchlorides to accumulate on the concrete surface, due towetting/drying cycles. However, other contradictoryresults have shown higher values of chloride surfacecontents (Cs) in tidal exposure due to water absorption(Uji et al. 1990). In general, a build-up of Cs is expectedto take place at sea level. For structures exposed toairborne conditions, it is common to observe a build-up of Cs over time, but the values reached will dependnot only on the distance to water level or to seashore,but also on sea wind orientation and on the protectionof the concrete surface against rainfall (Fluge 1997,Meira et al. 2007).

For permanently immersed or saturated concrete,chloride penetration occurs mainly by diffusion and,hence, models based on Fick’s second law have beenapplied. Considering that the diffusion coefficient andthe surface chloride content are time independent andthat for a semi-infinite surface, where Cranks

*Corresponding author. Email: msalta@lnec.pt

Structure and Infrastructure Engineering

Vol. 8, No. 6, June 2012, 583–594

ISSN 1573-2479 print/ISSN 1744-8980 online

� 2012 Taylor & Francis

http://dx.doi.org/10.1080/15732479.2010.505378

http://www.tandfonline.com

Dow

nloa

ded

by [

Uni

vers

ity o

f C

onne

ctic

ut]

at 0

8:32

11

Oct

ober

201

4

integration of Fick’s second law can be applied toestimate chloride penetration into the concrete, thetime for corrosion initiation will be given by:

T ¼ d 2

4Derf �1

Cth � C0

Cs � C0

� ��2ð1Þ

where D is the chloride diffusion coefficient, Cth is thechloride threshold for depassivation of the reinforcingsteel, d is the concrete cover, Cs is the equilibriumconcrete surface chloride content and C0 is the initialchloride content in the concrete. It follows from thisequation that the time necessary for corrosion initia-tion is inversely proportional to D, and it is thereforeof great importance to obtain a reliable estimate of Dfor modelling the corrosion initiation period inreinforced concrete.

In non-permanently saturated concrete, the chlor-ide penetration into the concrete usually takes place bya multi-stage physical process, i.e. by absorption andconvection in the external concrete layers and bydiffusion in the internal concrete layers. Therefore, it isessential that the two physical processes are consideredto accurately model the chloride penetration.

It is apparent that the uncertainties associated withmodelling D and Cs will have a significant influence onthe accuracy of the estimated corrosion initiationperiod, and by implication with (i) determination ofconcrete service life, (ii) prescription of concretedurability specifications for the design of new struc-tures and (iii) prediction of maintenance costs. Inaddition, the quality of construction in a structure, andthe associated consequences regarding uncertainty inactual cover depth (d) values, will be decisive inachieving a design service lifetime with minimisedmaintenance costs.

In an attempt to quantify the aforementioneduncertainties, it was considered necessary to performextensive field measurements on a structure in a marineenvironment. In this regard, the 36-year old BarraBridge, located on the Atlantic coast of Portugal, wasselected for extensive investigation prior to the execu-tion of major rehabilitation works to repair andrehabilitate the reinforced concrete and to give thebridge the necessary structural capacity.

The aim of the investigation was to evaluate thespatial variability of D and Cs and to correlate thedistribution of these values with the visual localiseddamage condition in the bridge and with informationcollected from non-destructive tests (NDT) for con-crete corrosion condition evaluation based on thecorrosion rate and potential measurement.

This article presents the results obtained by theNDT performed at the zones identified for evaluatingthe corrosion state of the reinforcement and for

measuring the concrete cover, and also the chlorideprofiles in the concrete obtained in the same zones.From the total chloride profiles obtained, the valuesfor D and Cs were estimated in each concrete areainspected. The average values and the associateduncertainties given by covariance were calculated forthe different zones of the bridge. Finally, considerationwas given to the expected chloride threshold values –Cth, for the different location on the bridge based on (i)chloride profiles and cover depth measurements, (ii)the evaluation of the corrosion state of the reinforcingsteel by NDT tests and (iii) visual observation of steelcondition after removal of the concrete cover.

2. General description of Barra Bridge

Barra Bridge, which was designed by the Portuguesebridge designer Prof Edgar Cardoso, began operationin 1975. The bridge is located in the north of Portugalover Canal Mira in Aveiro, with an NE to SWorientation, and is integrated in highway E.N.109-7,Figure 1. It is made up of reinforced concrete and has alength of 578.00 m, between support axes on theabutments, Figure 2. The central span is 80.00-m longand the access viaducts, which are symmetric about thecentral span, are 249.00-m long. The viaducts arecomposed of eight spans, i.e. seven internal spans of32.00-m and one external span of 25.00-m long, Figure2. The central spans are formed by two 34.00-m-longcantilevers, which are connected by an iso-staticcentral span. In the central span, the deck consists ofa variable-height double-box girder with constantthickness webs, Figure 3, with a slab of variablethickness, which is transversally pre-stressed.

The deck of the viaducts consists of a 16.60-m wideT-beam slab with four longitudinal pre-stressed beamsof constant web thickness equal to 0.40 m, Figure 4.The beams, which are pre-stressed longitudinally, areof variable height, between 1.30 m in the span and2.00 m at the supports. The transversally pre-stressedtop slab has a variable thickness ranging from 0.18 to0.25 m. The reinforced concrete piers are constitutedby two portal frames, Figure 5. The first framesupporting the deck consists of a cross beam (with2 m height) and two shafts (with variable heights from7.77 to 10.33 m), both with a rectangular cross section.The second frame consists of a lintel (with 4.00 mheight), where the shafts of the first frame are built-in,and of two cylindrical caisson foundations. Theabutments are hollow boxes.

The nominal concrete cover depth prescribed in thedesign was 40 mm on lintels and 30 mm on the otherpier zones and on the deck. The concrete class ofresistance defined in the design was B35 for deckbeams and B30 for all other structural elements.

584 M. Salta et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

onne

ctic

ut]

at 0

8:32

11

Oct

ober

201

4

3. Investigation programme

The investigation programme developed for this studyaimed to evaluate primarily the influence of thedifferent microclimatic conditions of the bridge onthe concrete performance to chloride penetration,expressed in terms of apparent chloride diffusioncoefficients and chloride surface content. It was alsointended to identify the other most significant environ-mental factors affecting the steel reinforcement corro-sion development.

3.1. Selection of inspection locals

After combining the existing information from theevaluation of data obtained during previous

inspections performed at the bridge, throughout itsservice life, with the information obtained from a newgeneral visual inspection of the bridge, several sites onthe structure were selected for performing the tests toreach the intended objectives of this study. Two pierson land, five piers in water and five spans of the deckwere selected for investigation. Six locations on eachselected pier on land (4 on the shafts and 2 on the crossbeam), 10 on each selected pier in water (4 sites on thelintel, 2 sites on each shaft and 2 on the cross beam)and 5 on each span of the deck have been evaluated,Figure 6.

In each selected pier, two concrete areas in thecolumns (one at 4.50 m and the other at 12.00 m abovethe mean water level) were located at the same levels

Figure 2. General view of Barra Bridge.

Figure 1. Location of the bridge.

Structure and Infrastructure Engineering 585

Dow

nloa

ded

by [

Uni

vers

ity o

f C

onne

ctic

ut]

at 0

8:32

11

Oct

ober

201

4

on both the seaward face (i.e. east face) of theupstream column (JE) and the opposite face (i.e. westface) of the downstream column (DM). In the crossbeam, another two areas were selected in the centralzone of the two vertical faces (NE and SW), Figure 7.

From among each of the five selected locations onthe deck, four were located on the viaducts and one onthe central span, all of which were over the water. Ateach location, three areas were inspected. Two areaswere located on the vertical external concrete face ofthe two external longitudinal beams and the third on avertical face of an internal longitudinal beam.

Each selected area had a 2 m length 6 1 m heightand all the NDT and destructive tests were performedin each of the inspected areas.

3.2. Tests performed

In each area selected for testing, the concrete surfacewas submitted to a detailed visual inspection to detectexternal defects on the concrete resulting from possiblecorrosion of the reinforcing steel and to identify otherconcrete damage due to chemical reactions or con-struction defects. Following the visual observation ofthe concrete surface, NDT was performed to locate theposition of the reinforcement and to evaluate concretecover depths, as well as for measuring the corrosionpotential and the corrosion rate of steel reinforcement.The concrete cover depths of steel reinforcements inthe different zones were measured using the coverdepth measurement equipment from Hilti. The map-ping of corrosion potential was performed with Ger-man Instrument GD-2000 Mini Great Daneequipment, and the determination of the corrosionrates was done by the polarisation resistance methodusing the equipment GeCor 6.

Subsequently, taking into consideration the resultsfrom the NDT recorded for each tested area, aminimum of two concrete cores (with 100 mmdiameter and 150 mm in length) were drilled (i) forthe determination of concrete carbonation and chlor-ide profiles, (ii) for microscopic tests to evaluate theconcrete microstructure and (iii) for carrying out othercomplementary lab tests on concrete (capillary absorp-tion and migration chloride coefficients). From thechloride profiles obtained, the concrete chloride diffu-sion coefficients, the chloride surface content and thechloride content at the steel surface were estimated.After NDT and core sampling, the cover concrete inthe inspected area was removed, which made itpossible to perform the direct visual inspection of thereinforcement steel surface condition, so as to evaluatealso the conformity of electrochemical measurements.

From some cores, concrete samples were obtainedfor microscopic tests, using polarising and fluorescence

Figure 3. Deck of the central span.

Figure 4. Deck of the viaducts.

Figure 5. Reinforced concrete piers.

586 M. Salta et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

onne

ctic

ut]

at 0

8:32

11

Oct

ober

201

4

optical microscope and scanning electron microscope,for evaluating the concrete microstructure. For mea-suring the concrete capillary absorption and themigration coefficients, samples were also extractedfrom the most internal zone of the cores where theconcrete is not contaminated with chlorides. Data forconcrete compressive strength was obtained fromrecords of tests performed on concrete samplescollected in previous inspections at the bridge, duringthe years 2000–2001.

For characterising the environmental corrosivity,the environmental data concerning relative humidity(RH) and temperature (T), in the region where thestructure is located, have been collected from thedatabase of the Portuguese Meteorology Institute.Considering the atmospheric corrodibility and inaccordance with the corrosion map for Portugal, thisregion could be classified as C3/C4 corrosivity classaccording to ISO 9223. To complement this informa-tion, during the execution of the repair works on thebridge, three chloride intake systems (wet candles)

were also installed, with a view to collect data onpollution by airborne salinity relative to chlorides.Systems were installed on the west viaduct deck at theNE abutment, as well as on a pier in water and on apier on land at the intermediate level of the shaft.

4. Results

4.1. Visual inspection of the concrete surface

The visual inspection showed the existence of dis-organised cracking in the concrete at many locations,particularly, on the concrete surfaces most exposed todirect contact with water, on lintels, or to rainfall, onpier columns (at the lowest zones of the shafts), bothon land or in water, Figure 8. In general, gel productscan be observed in the cracks, Figure 9. On lintels, itwas also possible to observe disorganised cracks andsome cracks with a preferential vertical orientation,reaching, in many situations, 4 1 mm width, Figure10. Only in the splash zone was active localisedcorrosion of the reinforcement observed along theselarge cracks, Figures 8 and 10.

All over the bridge, it was possible to observeconcrete areas with concrete delamination or oxidespots. These were the most significant and extensive onthe bottom flange of the longitudinal external beams ofthe deck, Figure 11 and on the splash zone of lintels,Figures 8 and 10. They were less significant on thecross beams and the columns of the piers, Figure 12.The zones with extensive concrete delamination,mainly on the bottom flange of external beams of thedeck, were already observed to demonstrate significantreduction in reinforcement cross section. At thelocations where the corrosion was more significantand extensive, i.e. on the bottom flange of thelongitudinal beams of the deck, low concrete cover(10 to 15 mm) and concrete segregation were observed.

4.2. In situ tests

4.2.1. NDT tests

The distributions of measured parameters, concretecover, corrosion potential and corrosion rate on

Figure 6. Longitudinal plan of Barra Bridge with the locations of inspection.

Figure 7. Location of the inspected zones in piers.

Structure and Infrastructure Engineering 587

Dow

nloa

ded

by [

Uni

vers

ity o

f C

onne

ctic

ut]

at 0

8:32

11

Oct

ober

201

4

inspected areas were determined to identify correla-tions with the environmental exposure conditions.Different distributions have been obtained with datacollected from several inspected zones, such as, piercrossbeams, longitudinal beams of the deck and at thetwo height levels on the two faces of columns of piersin water and on land. From each group of piers inwater or piers on land, only one distribution wasobtained at each height level, suggesting that the twocolumn faces with different orientations to the seashoreexhibited a similar performance to chloride damage.After this evaluation, both the averages and coeffi-cients of variation (COV), for cover depth, corrosionrate and potential measurements were calculated forthe zones of the structure presenting a distinctperformance to chlorides, as presented in Table 1.

The most significant cover depth variability wasobserved on columns of piers on land and on deckbeams. Only on lintels (mainly at the tidal zone) wasthe measured cover depth lower than the design covervalue of 40 mm. In other bridges zones, the coverdepth generally fulfilled the nominal values specified inthe bridge drawings, excluding the flanges of thelongitudinal beams of the decks where reduced coverdepths were measured.

Concerning the corrosion potential measurements,the lowest variability was obtained on lintels where theaverage measured values represented a high probabil-ity of steel corrosion state.

The corrosion rate of concrete reinforcing steel isoften regarded as being significant when the values aresystematically higher than 0.2 mA/cm2 (Cigna et al.1997). Nevertheless, when the concrete is watersaturated, a decrease in corrosion rate is expected tooccur due to the reduced access of oxygen in theconcrete. The corrosion rate measurements showed, on

Figure 8. Disorganised cracks on lintels and columns.

Figure 9. Cracked concrete with gel products.

Figure 10. Large cracks in the lintel with and without corrosion.

588 M. Salta et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

onne

ctic

ut]

at 0

8:32

11

Oct

ober

201

4

the tidal and splash zones of lintels, the highest averagevalues, between 0.48 mA/cm2 and 0.39 mA/cm2, mean-ing that the corrosion is active. Mention must be madeabout the fact that the measurements were performedafter installation of cofferdams for the execution ofrepair works at pier foundations. As a result, theconcrete cover is not water saturated and the access ofoxygen is easier. In the other zones, the corrosionaverage values are between 0.15 mA/cm2 and 0.26 mA/cm2, which means that, in general, corrosion is notactive and only at a few localised points could steeldepassivation be initiated.

After the conclusion of NDT tests, two or threecores each with a diameter of 100 mm and a length of150 mm were extracted from each inspected concretearea.

4.2.2. Visual inspection of steel surface condition

After carrying out the planned NDT and after drillingthe concrete cores, the concrete cover was removed inthe zones where electrochemical tests showed differentresults in each selected area, so as to evaluate theeffective corrosion state of the steel reinforcement bydirect visual observation and to measure the possiblereduction in the steel cross section. Only at thelocations inspected on the lintels, was it possible toobserve steel corrosion initiation. However, even inthese locations, steel corrosion only occurred wheresome heterogeneities or cracks were present on theconcrete cover. In the other inspected locations underairborne exposure, no corrosion initiation has beenobserved on the reinforcement even when small crackswere present on the concrete surface. Steel corrosion isonly significant, with a high reduction in the crosssection of reinforcements, when concrete delaminationis already present. This is the case for the bottomflanges of some longitudinal beams of the deck where(i) the cover depth is lower than the design

Table 1. Average and coefficient of variation of the concrete cover depth, corrosion rate and corrosion potential values.

Structural element

Cover (d) (mm)aPotential (mV vs.

Cu/CuSO4)Corrosion rate

(mA/cm2)

Average COV (%) Average COV (%) Average COV (%)

Piers in water Columns, at 4.5 m 36 19 7157 43 0.18 86Columns, at 12 m 33 16 7154 43 0.26 58Cross beams 34 12 7279 26 0.15 98Lintels at splash 40 15 7428 22 0.48 87Lintels at tidal 35 13 7454 8 0.39 62

Piers on land Columns, at 4.5 m 33 27 774 76 0.21 60Columns, at 12 m 34 26 768 45 0.19 55Cross beams 31 12 7186 18 0.15 60

Deck Longitudinal beams 29 21 7166 42 0.24 68

aThe concrete cover specified in the original drawings was 40 mm on lintels and 30 mm on all other parts of piers and on decks.

Figure 11. Extensive concrete delamination and corrosionon the longitudinal beams of the deck.

Figure 12. Localised concrete delamination on columns.

Structure and Infrastructure Engineering 589

Dow

nloa

ded

by [

Uni

vers

ity o

f C

onne

ctic

ut]

at 0

8:32

11

Oct

ober

201

4

specification, (ii) the chloride penetration into theconcrete is bi-directional and (iii) concrete segregationexists.

4.2.3. Environmental corrosivity

Based on the data collected from the records of thePortuguese Meteorology Institute, the annual averagevalues of temperature and RH of the air around thebridge are 158C and 75–80%, respectively, and thepreferential wind direction is NE to SW. Taking intoconsideration the chloride deposition measured on wetcandles, the airborne pollution to chlorides wasclassified as S1 on inland piers and in the deck andas S2 on piers in the sea, according to EN 9223. Asalready reported, the class of corrosivity on the bridgelocation is C3/C4.

4.3. Lab tests

4.3.1. Chloride content and carbonation in concrete

For the cores drilled from the different zones inspected,the carbonation depths were measured by spraying theconcrete with a phenolphthalein solution. The averagevalues of carbonation measured in the concrete fromaerial zones are between 10 mm and 19 mm. On thetidal and splash zones no significant carbonationvalues were measured.

Powder samples were collected from the concretecores by drilling at successive depths of 5 mm from thefree concrete surface to 50 or 60 mm from the exposedconcrete surface. After dissolution of the powdersamples obtained at each concrete depth in hot nitricacid, the total chloride content in the concrete wasobtained by direct potentiometry using a chlorideselective electrode. Two chloride profiles were obtainedin each selected zone. The apparent values of chloridediffusion coefficients and chloride surface contentswere estimated by adjusting the experimental profilesto Cranks integration of Fick’s second law.

The equivalent locations in the several piers inwater demonstrated similar performance with respectto chloride penetration and corrosion initiation. Asimilar performance was observed between the equiva-lent locations of the inland piers.

Figures 13–16 show typical chloride profiles ob-tained at the different locations inspected on piers inwater (P10) and on land (P15).

For comparison, Figures 13 and 15 show resultsobtained from the concrete areas tested on the seawardfaces of the different pier elements: the four concreteareas at the two levels tested in the columns (i.e. at4.5 m and 12 m), one area in the cross beam and alsothe two areas on tidal and splash exposure of the lintel,

note this last condition is presented only for pier P10 inwater. Profile data obtained on downstream andupstream columns, in the faces with the two orienta-tions to the seashore, seaside face (JE) and the oppositeface (MD) at two levels (4.5 m and 12 m above themedium sea level), are presented in Figures 14 and 16for piers P10 and P15, respectively.

Figure 13. Pier P10 in water. Chloride profiles at thedifferent locations of the pier: lintel, column and crossbeam.

Figure 14. Piers P10 in water. Chloride profiles at thedifferent locations in the two columns.

Figure 15. Pier P15 in land. Chloride profiles at thedifferent locations of the pier: column and crossbeam.

590 M. Salta et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

onne

ctic

ut]

at 0

8:32

11

Oct

ober

201

4

The most relevant aspects shown in these figuresare the different shapes of the chloride profilesobtained in the tidal and splash zones of lintels andin the airborne exposure zones, at the two levels, oncolumns of piers in water and on cross beams. A skineffect, penetrating deep in the cover depth (until20 mm), was observed in almost all the chlorideprofiles in concrete under airborne exposure, whichwas more significant in the zones of columns at 12-mlevel and in the cross beams. The highest D and Cs

were achieved under airborne exposure on the crossbeams of the piers, both in water and inland. Oneexplanation can be the well-protected exposure of crossbeams to rainfall, which minimises the surface concretewashing effects. No significant effect due to orientationof the concrete surface in relation to the seashore wasobserved on chloride penetration parameters D and Cs,Figures 13 and 15. On columns of piers in water, the Cs

values are, in general, higher at the lowest height leveltested (4.5 m) and, in general, the skin effects at thislocation is not significant, Figure 13. On piers on land,no significant variation in Cs was obtained at the twolevels in the columns, Figure 14.

The distribution of the measured parameters D,and Cs, on similar zones of the bridge have been

calculated to establish whether correlations exist withthe environmental exposure conditions. Figure 17shows an example of the histograms of values takenby D, Cs and the cover depth (d) collected from all thezones of piers in water inspected at the 4.5-m level withthe two orientations in relation to the seashore.

Based on the information collected from thehistograms obtained, the bridge zones with differentperformances of the concrete to chloride penetrationwere identified and the average values for D and Cs, foreach distinct zone, were estimated. Table 2 presentsthose values and the average estimated values forchloride content in the concrete on the reinforcementsteel surface, which were estimated by the integratedequation of second Fick’s law using the values for Dand Cs and the cover depth (d) obtained in each zone.

4.3.2. Observation of concrete microstructure andother tests

Observation of the concrete microstructure underpolarising and fluorescence optical microscopy andscanning electronic microscopy was conducted onconcrete slices obtained from the concrete cores. Themicrostructure observations had shown a homoge-neous and compact Portland cement paste in eachconcrete sample. The samples obtained from coresdrilled in zones where the concrete is cracked demon-strated products of alkali silica reaction, Figure 18.These products, occurring in the concrete paste arealso associated with compact ettringite formation. Theaggregates used are mainly calcareous; however, insome samples, even those obtained from coresextracted in different concrete batches on a samecolumn, the use of siliceous rolled stone aggregates wasobserved. From the petrographic analysis of the slices,the W/C ratio in the concrete was estimated. On lintels,columns and deck beams, the average values were 0.55,which is more or less in accordance with the concreteclass specified in the design, but a slightly higher valueof 0.60 was estimated for the cross beams.

Figure 16. Pier P15 in land. Chloride profiles at thedifferent locations in the two columns.

Figure 17. Histograms of D, Cs and d obtained in piers in water (zones at 4.5 m level on columns).

Structure and Infrastructure Engineering 591

Dow

nloa

ded

by [

Uni

vers

ity o

f C

onne

ctic

ut]

at 0

8:32

11

Oct

ober

201

4

From the most internal zones of the concrete cores(where concrete is not contaminated with chlorides),concrete specimens were obtained for the determina-tion of concrete capillary absorption and migrationchloride diffusion coefficient. The mean value ofcapillary absorption is 0.80 kgm72 h71/2 and thechloride migration coefficients of concrete are in therange of 15 to 25 6 10712 m2 s71.

5. Discussion

The NDT results have shown that corrosion wasinitiated only where large defects were present in theconcrete cover, in localised areas where the concretecover is very low (around 15–10 mm of cover) andmore extensively where bi-directional chloride pene-tration occurs. The splash zone of lintels where theconcrete is cracked due to the existence of internalexpansive reactions (ASR) also exhibited the existenceof active corrosion. Other cracked concrete zones

without direct contact with sea water or submersed didnot present significant corrosion initiation even at lowheight levels in the columns.

As regards the concrete performance to chloridepenetration, the most relevant exposure conditions arethe existence of contact with sea water, the distancefrom the sea water and the protection from rainfall.The location according to the wind preferentialorientation is not relevant. This is not in accordancewith the results in other bridges (Fluge 1997), but apossible explanation can be due to the fact that thepredominant wind direction is the same as the bridgeimplantation. The highest average D values, 1.6–2.1 m2/s, have been obtained on lintels. At the otherzones of the structure, the D values achieved a rangefrom 0.4 to 1.2 m2/s. The lowest D values wereestimated on the deck beams, where a concrete mixwith increased compressive resistance has been speci-fied in the design and on the lowest level of columns,where better compaction of concrete is usually reached

Table 2. Average and coefficient of variation values of D, Cs and total chloride content (Cl7)st near the steel surface, expressedin % relatively to concrete weight.

Structural element

D (610712 m2/s) Cs (%) (Cl7)st (%)

Average COV (%) Average COV (%) Average COV (%)

Piers in water Columns, 4.5 m 0.44 29 0.27 27 0.06 34Columns, 12 m 1.04 62 0.08 64 0.04 88Crossbeams 1.11 80 0.31 25 0.15 55Lintels – tidal 2.07 51 0.47 31 0.27 48Lintels – splash 1.59 59 0.38 32 0.11 79

Piers on land Columns, 4.5 m 0.89 55 0.14 25 0.05 35Columns, 12 m 0.59 23 0.17 36 0.05 75Crossbeams 1.23 41 0.32 16 0.14 17

Deck beams 0.37 70 0.39 38 0.11 44

Figure 18. Concrete microstructure with ASR gel products.

592 M. Salta et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

onne

ctic

ut]

at 0

8:32

11

Oct

ober

201

4

due to the compressive effects on the lower parts ofpier columns.

The estimated Cs values obtained, expressed in %of total chloride by cement weight, range from 0.60 to3.5%. Values for Cs from 0.2 to 2.43%, by weight ofcement, have also been reported for structures inmarine environments at 50–1000 m away from sea andafter 30 years of exposure (Ha-Won et al. 2007). Theuncertainties evaluated by the COV associated to Dand Cs values are lower for lintels where high values ofchloride content exist in the concrete. Lindvall (2007)reported COV values, for D, between 8.6% and 34.9%,and for Cs, between 8.0% and 33.0%, for the sameconcrete mix with specimens prepared in the lab andexposed in sea immersion conditions, in several warmand cold countries. In the present investigation, similaruncertainty values have been obtained for Cs in thebridge. Nevertheless, for D, a higher uncertainty valuethan those reported was obtained, which can resultfrom the concrete heterogeneity in different batchesobserved in the bridge.

The estimated values for concrete chloride contentat the reinforcement steel surface, expressed in %relative to concrete weight, are 0.15%, in the cross-beams of piers on land, 0.14% in the cross beams ofpiers in water, 0.11% in the longitudinal beams of thedeck, 0.27% and 0.11% in the tidal and splash zones oflintels and less than 0.06% in columns for piers inwater or on land. By combining these results with thesteel corrosion state evaluation given by NDT and bythe visual inspection of concrete and steel reinforce-ments after removing the concrete cover, it seems thatthe chloride threshold for corrosion initiation, Cth, onbridge zones with atmospheric exposure must behigher than 0.15% (relative to concrete weight), whichcorresponds to 1.2% (by weight of cement). Forconcrete zones with a high water-saturation level(mainly in tidal and splash zones of lintels) Cth valuesmust be higher. In fact, the chloride content in thereinforcing steel reached values of 2.1% (rel. to weightof cement) in the splash zones but the corrosion wasonly initiated in specific zones where the concrete coverhas large defects, large cracks in the concrete or othersignificant defects like honeycombs.

Using the equation presented in Section 1 and theminimum achieved value of D, as well as the maximumCs value obtained in the airborne salt zones, theestimated time for corrosion initiation, using thechloride threshold of 0.15% (rel. to concrete weight)is around 50 years, which is in agreement with the lowcorrosion damage observed in these zones. If theinitiation time is also estimated for the splash zone,using the same limit value for Cth, the estimated timefor corrosion initiation is 10 years. This value disagreeswith the observation of the good condition of steel

reinforcement in these zones of the bridge if hetero-geneities are not present in the concrete cover, meaningthat the threshold value of 0.15% is actually veryconservative for these exposure conditions.

The values of the concrete chloride migrationcoefficient measured in the lab have an order ofmagnitude 10 times higher than the achieved D valuesin the bridge estimated by the chloride profiles. Thisresult agrees with other studies involving naturalexposure of concrete specimens in a marine environ-ment during several years (Salta et al. 2006). Thesignificant reduction obtained in the achieved D valuescompared with the non–steady state chloride diffusioncoefficient obtained in lab under migration can beexplained by two main reasons. The first is theinfluence of the environmental exposure conditionson the chloride transport process. The second is due tothe fact that the D coefficients achieved in the structurecorrespond to the average values reached during theservice life of the concrete, which will reflect the effectof the variation of exposure conditions and also someprogress in Portland cement hydration in the concreteduring the life of the bridge. Other aspects, which mustbe considered to understand the different values ofachieved D and the migration coefficients, are thedifferent concrete porosities, in the bulk concrete (onwhich migration tests have been performed) and in thecover of the concrete in the several structural elements.All of these aspects are relevant for modelling the timeto corrosion initiation in reinforced concrete structuresand particular attention must be paid when the Fick’ssecond law is used for corrosion modelling withconcrete diffusion or migration coefficients estimatedby accelerated tests performed in the laboratory.

6. Conclusions

The study provided valuable data about the effects ofexposure conditions on chloride penetration intoconcrete and about their consequences on parametersneeded for predicting the corrosion initiation time. Thedistance from the sea water level is of majorimportance for the environmental chloride loadrepresented by the concrete surface chloride content.Concrete surface protection from rainfall is also veryimportant as regards chloride loads, although thisaspect is not usually considered in design specificationsfor reinforced concrete in marine environments. Dueto the specific bridge location, no significant influenceon the concrete chloride transport was observed on thecolumn faces with different orientations relative to theseashore and to the preferential wind direction. Thesurface chloride content ranges from 0.10 to 0.50%(relative to concrete weight) and the COV valuesobtained are very uniform in the range of 20–40%. The

Structure and Infrastructure Engineering 593

Dow

nloa

ded

by [

Uni

vers

ity o

f C

onne

ctic

ut]

at 0

8:32

11

Oct

ober

201

4

highest values of surface chloride content have beenobtained in tidal zones and in the zones protected fromrainfall.

Chloride diffusion coefficients in the concreteranged from 0.37 to 1.23 6 10712 m2 s71 in the zonesof the bridge without direct contact with sea water andfrom 1.59 6 10712 to 1.21 6 10712 m2 s71 in thezones in contact with sea water. The COV valuesreached 80%, which means that concrete quality is notuniform, as confirmed by the microscopic test. This isalso an important factor that must be considered whenmodelling the service life of structures.

The estimated chloride threshold values must behigher than 1.2% relative to cement weight, which isthree times higher than the 0.4% chloride reported instandards and used in the modelling of the concreteservice life.

This study also confirms the relevance of qualitycontrol to fulfil the design specifications, mainly asregards the concrete quality and the specified coverdepth value.

In a marine structure, the existence of both a highvariability and the associated uncertainties in the mostrelevant parameters for chloride ingress into theconcrete also confirms the relevance of the use ofprobabilistic models for predicting the service life andfor establishing the maintenance strategies of infra-structural elements.

Acknowledgements

This study was developed under MEDACHS project (MarineEnvironment Damage to Atlantic Coast Historical andTransport Structures) from the EU INTERREG- AtlanticSpace Program. The authors thank Estradas de Portugal,S.A. for the permission given to perform this study and forthe local facilities made available during the in situ tests.

References

Bamforth, P.B. and Price, W.F., 2000. Factors influencingchloride ingress into marine structures. In: R.K. Dhir andM.R. Jones, eds. Concrete 2000. Vol. 2. Dundee: E&FNSpon, 1105–1118.

Cigna, R., et al., 1997. Monitoring corrosion in concrete,Corrosion and Protection of Metals in Contact withConcrete, In: R.N. Cox et al., eds. Final Report COST509 Action, Brussels, EUR17608, 65–79.

Fluge, F., 1997. Environmental loads on coastal bridges. In:A. Blankuoll, ed. International Conference Repair ofConcrete Structures. From Theory to Practice in a MarineEnvironment. Svolvaer: Norwegian Road ResearchLaboratory, 89–98.

Glass, G. and Buenfeld, N., 1995. Chloride threshold levelsfor corrosion induced deterioration of steel in concrete.In: L.-O. Nilsson and J.-P. Ollivier, eds. InternationalRilem Workshop on Chloride Penetration in Concrete, Oct1995. St Remy-les-Chevreuse: Rilem Publications SARL,429–238.

Lindvall, A., 2007. Chloride ingress data from field andlaboratory exposure. Influence of salinity and tempera-ture. Cement and Concrete Composites, 29, 88–93.

Meira, G., et al., 2007. Chloride penetration into concretestructures in the marine atmosphere zone – relationshipbetween deposition of chlorides on the wet candle andchlorides accumulated into concrete. Cement and Con-crete Composites, 29, 667–676.

Salta, M.M., et al., 2006. Chloride transport in concrete fromLab tests to in situ performance. In: V. Baroghel-Bounyet al., eds. PRO 047-Performance based evaluation andindicators for concrete durability. Bagneux: Rilem RilemPublications SARL, 229–240.

Song, H-W., et al., 2008. Factors influencing chloridetransport in concrete structures exposed to marineenvironment. Cement and Concrete Composites, 30,113–121.

Uji, K., Matsuoka, Y., and Maruya, T., 1990. Formation ofan equation for surface chloride content of concrete dueto permeation of chloride. In: C.L. Page, K.W.J. Tread-away, and P.B. Bamforth, eds. Corrosion of reinforce-ments in concrete, Barking: Elsevier Applied SciencePublishers Ltd, 258–267.

Wood, J. and Crerar, J., 1997. Tay road bridge: analysis ofchloride ingress variability & prediction of long termdeterioration. Construction and Building Materials, 11,249–254.

594 M. Salta et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f C

onne

ctic

ut]

at 0

8:32

11

Oct

ober

201

4