DESIGN AND CONSTRUCTION OF THE RION ANTIRION BRIDGEFOUNDATIONSA. PeckerGéodynamique et Structure, Bagneux, FranceAbstractThe choice of a design concept for a bridge foundation is guided by various factors; severalof these factors are indeed of technical origin, like the environmental conditions in a broadsense, but others non technical factors may also have a profound impact on the final designconcept. The foundations solutions adopted for the Rion Antirion bridge are described andan attempt is made to pinpoint the major factors that have guided the final choices. TheRion Antirion bridge is exemplar in that respect: the foundation concept combines thesimplicity of capacity design, the conceptual facility of construction and enhances thefoundation safety. The design of these foundations was a very challenging task whichrequired full cooperation and close interaction with all the parties involved: concessionaire,contractor, designers and design checker.INTRODUCTIONThe Rion-Antirion bridge project is a BOT contractgranted by the Greek Government to a consortiumled by the French company Vinci Construction(formerly Dumez-GTM). It is located in Greece, nearPatras, will constitute a fixed link between thePeloponese and the Continent across the westernend of the gulf of Corinth and is intended to replacean existing ferry system (Fig. 1). The solutionadopted for this bridge is a multiple spans cablestayed bridge with four main piers; the three centralspans are 560 m long each and are extended by twoadjacent spans (one on each side) 286 m long. Thetotal length of the bridge, with the approach viaducts,is approximately 2.9 kilometers (Fig. 2). The call fortender was launched in 1992, the contract, awardedto the consortium in 1996, took effect in December1997. Construction started in 1998 and has beencompleted in summer 2004 and opened to traffic onthe 11thof August, five months ahead of schedule.The bridge has to be designed for severeenvironmental conditions (e.g., Teyssandier et al.2000; Teyssandier 2002): weak alluvium deposits,high water depth, highly seismic area, possibleoccurrence of large tectonic displacements. Veryearly in the design process it turned out that designwas mainly controlled by the seismic demandimposed to the foundations.Fig. 1: Location of the bridgeIn order to alleviate potential damage to the structuredue to the above adverse conditions and to carry thelarge earthquake forces brought to the foundation(shear force of the order of 500 MN and overturningmoment of the order of 18 000 MNm for a verticalbuoyant pier weight of 750 MN), an innovativefoundation concept was finally adopted whichconsists of a gravity caisson (90 m in diameter at thesea bed level) resting on top of the reinforced naturalground (Combault et al. 2000). The groundreinforcement is composed of steel tubular pipes, 2m in diameter, 20 mm thick, 25 to 30 m long driven1
at a grid of 7 m x 7 m below and outside thefoundation, covering a circular area of approximately8 000 m2. The total number of inclusions under eachfoundation is of the order of 150 to 200. In addition,the safety of the foundation is greatly enhanced byinterposing a gravel bed layer, 2.8 m thick, on top ofthe inclusions just below the foundation raft with nostructural connection between the raft and theinclusions heads. This concept (inclusions plusgravel layer), in addition to minimizing the hazardsrelated to differential settlements, enforces acapacity design philosophy in the seismic foundationdesign, (Pecker 1998).In the following we will focus on the foundations ofthe main bridge, because the foundations of theapproach viaducts (pile foundations) do not deservemuch comments.Fig. 2: Bridge elevationGEOLOGICAL AND GEOTECHNICALENVIRONMENTThe site has been subjected to extensive offshoresoil investigations performed either from floatingbarges or from a ship, controlled with a dynamicpositioning system; these investigations includedcored boreholes, static Cone Penetration Tests withpore pressure measurements (CPTU), StandardPenetration tests (SPT), vane tests and dilatometertests, seismic cone tests and sampling of intact soilsamples for laboratory testing. All the boringsreached depths ranging from 60 m to 100 m belowthe sea bed. Under each of the main bridge pierthree continuous boreholes, three CPTU, twoseismic cones, one SPT/dilatometer boring havebeen drilled. Approximately 300 samples have beenretrieved and subjected to advanced laboratorytesting. Based on the results of these investigationsrepresentative soil profiles have been defined at thelocations of the main bridge piers and ranges of soilmechanical characteristics have been derived.The water depth in the middle of the strait reaches65 m. The soil profile consists of weak alluvial stratadeposited in alternate layers, with individualthickness of a few meters, of silty sands, sandy claysand medium plasticity clays. In the top hundredmeters investigated by the soil survey, the clay, orsilty clay, layers predominate (Fig. 3). No bedrockwas encountered during the investigations andbased on geological studies and geophysicalsurveys its depth is believed to be greater than500 m.65mRIONANTION65m65 mRIONANTIONCLAYSILTSAND-GRAVELFig. 3: Soil profileThe mechanical characteristics of the offshore layersare rather poor with undrained shear strengths of thecohesive strata increasing slowly with depth fromapproximately 30-50 kN/m2at the sea bed level to80-100 kN/m2at 50 m depth (Fig. 4). A majordifference occurs in the cohesionless strata: some ofthese layers are prone to significant pore pressurebuildup, even possibly to liquefaction, under thedesign earthquake. Accordingly, the undrainedstrengths of the cohesionless layers are taken equaleither to their cyclic undrained shear strengths or totheir residual strengths.Based on the results of the laboratory onedimensional compressibility tests and of correlationswith CPT results or with the undrained shearstrengths, a slight overconsolidation, of the order of150 kN/m2, of the upper strata was evidenced.The shear wave velocities are also small, increasingfrom 100-150 m/s at the ground surface to 350-400 m/s at 100 m depth (Fig. 5).It is worth noting that for design, as shown in Fig.4and Fig.5, two sets of soil characteristics have beenused instead of a single set based on characteristicvalues; this decision is guided by the fundamentalnecessity to maintain compatibility in the wholeseismic design process: the seismic demand iscalculated assuming one set of properties(successively lower bound and upper bound) andthe foundation capacity is checked with theassociated strengths (respectively lower bound andupper bound); the capacity is never checked withlow properties when the forces are calculated withhigh properties and vice versa. In addition, given thelarge foundation dimensions, special attention hasbeen paid to the spatial variability of the soilproperties across anyone foundation; this variability
may have an impact on the differential settlementand tilt of a pylon. Fig. 6 gives an example of thevariability of the cone point resistance across afoundation, variability which is obviously reflected byvariability in the shear strength and compressibilityof the soil strata.Fig. 4: Undrained shear strengthENVIRONMENTAL CONDITIONSThe environmental conditions are defined by threemajor possible events likely to occur during thebridge life time: ship impact on a main bridge pier,occurrence of a major earthquake in the vicinity ofthe bridge, long term tectonic movements.Ship impactThe hazard represented by this impact correspondsto 160 000 Mg tanker hitting one pier at a speed of16 knots (8.2 m/s). This impact induces anhorizontal shear force of 480 MN acting at 70 mabove the foundation level; at the foundation levelthecorresponding forces are: shear force of 480 MNand overturning moment of 34 000 MNm.Earthquake eventThe bridge is located in one of the most seismic areain Europe. In the past 35 years three earthquakesexceeding 6.5 on the Richter scale have occurred inthe Gulf of Corinth. The 1995 Aigion earthquaketook place less than 30 km east of the site. Fig. 7presents the epicenters of the major earthquakes feltin the Gulf of Corinth along with the major tectonicfaults. The contract fixed for the design motion areturn period of 2 000 years. A comprehensiveseismic hazard analysis has defined the governingevent as a 7.0 surface wave magnitude earthquakeoriginating on the Psathopyrgos fault (shown with anarrow in Fig. 7) only 8.5 km east of the site (circle inFig. 7).Fig. 5 : Shear wave velocity0204060801000 100 200 300 400 500 600Shear wave velocity (m/s)Depthbelowgroundsurface(m)DepthbelowseabedDistance-60-80-12040-60 600 206555452515535-100-40 -200204060801000 100 200 300 400 500Undrained shear strength (kN/m2)Depthbelowgroundsurface(m)Fig. 6 : Spatial variability of cone resistanceIn recognition of the influence of the soilcharacteristics on the ground surface motion, thedesign response spectrum at the sea bed elevationis defined from specific site response analysesbased on the actual soil characteristics and on thedesign rock motion defined by the seismic hazardanalysis. The 5% damped response spectrum isshown in Fig. 8: the peak ground acceleration isequal to 0.5g, the plateau at 1.2 g is extending from0.2 s to 1.1 s and at 2 s the spectral acceleration isstill high, equal to 0.62 g.Tectonic movementsThe seismic threat arises from the prehistoric drift inthe earths crust that shifted the Peloponese awayfrom mainland Greece. The peninsula continues tomove away from the mainland by a few millimeterseach year. As a result the bridge mustaccommodate a 2 m differential tectonic
displacement in any direction and between any twopiers.Fig. 7: Locations of major earthquakes0.000.250.500.751.001.251.50- 1.0 2.0 3.0 4.0 5.0Period(s)Pseudoacceleration(g)Fig. 8: Design ground surface responsespectrumMAIN BRIDGE FOUNDATIONSVery soon during the design stage it appeared thatthe foundations design will be a major issue. On theone hand, the unfavorable geotechnical conditionswith no competent layer at shallow depth, the largewater depth, typical of depths currently encounteredin offshore engineering and the high seismicenvironment represent a combination of challengingtasks. It was also realized that the earthquakedemand will govern the concept and dimensioning ofthe foundations. On the other hand, the timeallowed for design turned out to be a key factor:thanks to the contractor who decided to anticipatethe difficulties, the design studies started one yearahead of the official effective date. Advantage wastaken of this time lapse to fully investigate alternativefoundation solutions, to develop and to validate theinnovative concept that was finally implemented.The amount of time spent initially for thedevelopment of the design concept was worthwhileand resulted in a substantial saving for thefoundation. In addition, the close cooperation thatexisted from the beginning within the design teambetween structural and geotechnical engineers,between the design team and the construction teamon one hand, and between the design team and thedesign checker on the other hand, was a key to thesuccess.Investigated foundation solutionsAfter a careful examination of all the environmentalfactors listed above, no solution seems to dominate.Several solutions were investigated: piledfoundation, caisson foundation, surface foundation.Piles were quickly abandoned for two reasons: thedifficulty to realize the structural connection betweenthe slab and the piles in a deep water depth, and therather poor behavior of floating piles in seismic areasas observed in Mexico city during the 1985Michoacan earthquake. Caissons foundations werehazardous due to the presence of a gravel layer atthe ground surface (Fig. 3), which may induce somedifficulties during penetration of the caisson.Surface foundation was clearly impossible in view ofthe poor foundation bearing capacity and of the highanticipated settlements. However, it was quicklyrealized that surface foundation was the only viablealternative from a construction point of view:construction techniques used for offshore gravitybase structures are well proven and could easily beimplemented for the Rion-Antirion bridge.The problem posed by the poor soil conditions stillremains to be solved. Soil improvement is requiredin order to ensure an adequate bearing capacity andto limit the settlements to acceptable values for thesuperstructure. Several techniques werecontemplated from soil dredging and backfilling (soilsubstitution), to in situ treatment with stone columns,grouted stone columns, lime columns. The need fora significant high shear resistance of the improvedsoil and for a good quality control of the achievedtreatment led to the use of driven steel pipes, atechnique derived from offshore engineering, toreinforce the soil beneath the foundations. Toprevent any confusion with piles foundations, whichbehave differently than steel pipes, those are namedinclusions.
Adopted foundation conceptIn order to alleviate potential damage to the structuredue to the adverse environmental conditions and tocarry the large earthquake forces brought to thefoundation (shear force of the order of 500 MN andoverturning moment of the order of 18 000 MNm fora vertical buoyant pier weight of 750 MN), theinnovative foundation design concept finally adopted(Fig. 9) consists of a gravity caisson (90 m indiameter at the sea bed level) resting on top of thereinforced natural ground (Teyssandier 2002,Teyssandier 2003). The ground reinforcement (Fig.10) is composed of steel tubular pipes, 2 m indiameter, 20 mm thick, 25 to 30 m long driven at agrid of 7 m x 7 m below and outside the foundationcovering a circular area of approximately 8 000 m2The total number of inclusions under eachfoundation is therefore of the order of 150 to 200. Inaddition, as shown below, the safety of thefoundation is greatly enhanced by interposing agravel bed layer, 2.8 m thick, on top of the inclusionsjust below the foundation raft with no structuralconnection between the raft and the inclusionsheads (Fig. 10). The reinforcement scheme isimplemented under three of the four piers: M1, M2and M3. Under pier M4, which rests on a 20 m thickgravel layer, soil reinforcement is not required andthe caisson rests directly on a 1 m thick ballast layerwithout inclusions.65m90m230m65m90m230mFig. 9: View of one pylon of the Rion-AntirionbridgeThe concept (inclusions plus gravel layer) enforces acapacity design philosophy in the foundation design(Pecker 1998). The gravel layer is equivalent to the"plastic hinge" where inelastic deformation andenergy dissipation take place and the "overstrength"is provided by the ground reinforcement whichprevents the development of deep seated failuremechanisms involving rotational failure modes of thefoundation. These rotational failure modes would bevery detrimental to the high rise pylon (230 m). If thedesign seismic forces were exceeded the "failure"mode would be pure sliding at the gravel-foundationinterface; this "failure mechanism" can beaccommodated by the bridge, which is designed formuch larger tectonic displacements than thepermanent seismically induced ones. The conceptis also somehow similar to a base isolation systemwith a limitation of the forces transmitted to thesuperstructure whenever sliding occurs.Fig. 10: Foundation reinforcementJUSTIFICATION OF THE FOUNDATIONSThe foundations have to be designed with respect tostatic loads (dead loads, live loads and tectonicdifferential movements) and with respect to dynamicloads (ship impact and earthquake). With regards tothe static loads, the main issue is the settlementsevaluation: absolute settlement and differentialsettlements inducing tilt, to which the 230 m highpylon is very sensitive. With regards to theearthquake loads, several problems have to besolved: evaluation of the seismic demand, i.e. theforces transmitted to the foundation during anearthquake, and check of the foundation capacity,i.e. calculation of the ultimate bearing capacity andearthquake induced permanent displacements.Static loadsThe immediate settlement of the foundation is not amajor concern since approximately 80% of the totalpermanent load (750 MN) is brought by the pierbelow the deck level. More important are theconsolidation settlements of the clay strata whichinduces differed settlements. The settlements arecomputed with the classical one dimensionalconsolidation theory but the stress distribution in thesoil is computed from a three dimensional finiteelement analysis in which the inclusions aremodeled. It is found that the total load is sharedbetween the inclusions for approximately 35 to 45 %
and the soil for the remaining 55 to 65 %. Theaverage total settlement varies from 17 cm to 28 cmfor Piers M1 to M3. For Pier M4 the computedsettlement is much smaller, of the order of 2cm. Thefigures given above are the maximum settlementspredicted before construction; they were computedassuming the most conservative soil characteristics,neglecting for instance the slight overconsolidationof the upper strata. Consolidation settlements wereestimated to last for 6 to 8 months after completionof the pier.The differential settlements are evaluated on thebasis of the spatial variability of the compressibilitycharacteristics assessed from the variability of thepoint cone resistances (Fig. 6). The computedfoundation tilts range from 7. 10-4to 1.8 10-3radians,smaller than those that would have been computedwithout inclusions: the inclusions homogenize thesoil properties in the top 25 m and transfer part ofthe loads to the deeper, stiffer, strata.020040060080010000 200 400 600 800Fréquence (Hz)TotalLoad(MN)-12.0-10.0-8.0-6.0-4.0-2.00.02.0Elapsed time (days)Settlement(cm)Fig. 11: Settlement versus time – Pier M3Although the previous figures were acceptable, itwas decided, in order to increase the safety of thestructure, to preload the pier during construction.This preloading was achieved by filling the centralpart of the cone pier (Fig. 9) with water duringconstruction to a total weight slightly larger than thefinal pier weight. Deballasting was carried out asconstruction proceeded, maintaining the total weightapproximately at its final value. The beneficial effectof this preloading is readily apparent in Fig. 11,which shows the recorded settlement and appliedload versus time during construction. The finalsettlement amounts to 8 cm as compared to 22 cmoriginally anticipated, and occurred more rapidly.This is clearly a beneficial effect of theoverconsolidation, as back calculations have shown.More important the foundation tilts are almostnegligible, less than 3 10-4.Seismic loadsThe evaluation of the seismic behavior of thefoundation requires that the seismic demand and theseismic capacity be calculated. However, as for anyseismic design, it is important to first define therequired performance of the structure. For the RionAntirion bridge this performance criterion was clearlystated in the technical specifications: "Thefoundation performance under seismic loading ischecked on the basis of induced displacements androtations being acceptable for ensuring reusability ofthe bridge after the seismic event". In other words, itis accepted that the foundation experiencespermanent displacements after a seismic event,provided the induced displacements remain limitedand do not impede future use of the bridge. Fulladvantage of this allowance was taken in thedefinition of the design concept: sliding of the pier onthe gravel layer is possible (and tolerated) butrotational mode of failures are prevented (andforbidden).Seismic capacityBecause of the innovative concept and therefore ofits lack of precedence in seismic areas, itsjustification calls for the development of new designtools and extensive validation. A very efficient threestages process was implemented to this end:• Development of design tools based on a limitanalysis theory to estimate the ultimate capacityof the foundation system and to define theinclusions layout: length and spacing. As anylimit analysis method, this tool cannot give anyindication on the induced displacements androtations.• Verification of the final layout with a non lineartwo, or three, dimensional finite element analysis.These analyses provide non only a check of theultimate capacity but also the non linear stressstrain behavior of the foundation that will beincluded in the structural model for the seismiccalculations of the bridge.• Experimental verification of the design toolsdeveloped in step 1 with centrifuge model tests.As shown below all three approaches give resultswhich are within ±15% of each other, whichincreases the confidence in the analyses performedfor design.The design tools are based on the Yield Designtheory (Salençon 1983), further extended toreinforced earth media, (de Buhan and Salençon1990). The kinematic approach of the Yield Designtheory applied to mechanisms of the type shown inFig. 12, (Pecker and Salençon 1999), permits thedetermination, for a wrench of forces (N verticalforce, V horizontal force and M overturning moment)applied to the foundation, of the sets of allowable
loads; this set defines in the loading spaceparameters a bounding surface with an equation:( , , ) 0N V MΦ = (1)Any set of loads located within the bounding surfacecan be safely supported by the foundation, whereasany set located outside corresponds to an unsafesituation, at least for permanently imposed loads.ΩωBλΒεα µδFFig. 12: Kinematic mechanismFig. 13 presents a cross section of the boundingsurface by a plane N=constant, corresponding to thevertical weight of one pier. Two domains arerepresented in the figure: the smallest onecorrespond to the soil without the inclusions, thelargest one to the soil reinforced with the inclusions.The increase in capacity due to the inclusions isobvious and allows the foundation to supportsignificantly larger loads (V and M). Furthermore,the vertical ascending branch on the boundingsurface, on the right of the figure, corresponds tosliding at the soil-foundation interface in the gravelbed layer. When moving on the bounding surfacefrom the point (M=0, V=560 MN), sliding at theinterface is the governing failure mechanism until theoverturning moment reaches a value ofapproximately 20 000 MN; for larger values of theoverturning moment, rotational mechanisms tend tobe the governing mechanisms and the maximumallowable horizontal force decreases. The height ofthe vertical segment, corresponding to a slidingmechanism, is controlled by the inclusions layoutand can therefore be adjusted as necessary. Thedesign philosophy is based on that feature: for astructure like the pylon with a response governed bythe fundamental mode, there is proportionalitybetween M and V, the coefficient of proportionalitybeing equal to the height of the center of gravityabove the foundation. When V increases the pointrepresentative of the loads moves in the plane ofFig. 13 along a straight line passing through theorigin, assuming that the vertical force is constant; iteventually reaches the bounding surface definingfoundation failure.050001000015000200002500030000350000 100 200 300 400 500 600 700Horizontal force at foundation level (MN)Overturningmoment(MN-m)Fig. 13: Cross section of the boundingsurface: Doted line without inclusions; solid line withinclusionsThe inclusions layout is then determined in a waysuch that this point be located on the verticalascending branch of the bounding surface.The previous analysis provides the ultimate capacityof the foundation but no information on thedisplacements developed at that stage. Thesedisplacements are calculated from a non linear finiteelement analysis. Most of the calculations areperformed with a 2D model, but some additionalchecks are made with a 3D model. For thosecalculations, the parameters entering the non linearelastoplastic soil constitutive model are determinedfrom the laboratory tests carried out on theundisturbed soil samples; interface elements withlimiting shear resistance and zero tensile capacityare introduced between, on one hand, eachinclusion and the soil and, on the other hand, the raftand the soil. These models are loaded to failureunder increasing monotonic loads such thatM/V=constant (Fig. 14). Results are compared tothose obtained from the Yield Design theory in Fig.13; a very good agreement is achieved withdifferences of the order of ±12% for all thecalculations made for the project. However, it isworth noting that the finite element analyses,because of the large computer time demand, couldnot have been used to make the preliminary design;the Yield Design theory is, in that respect, a moreefficient tool: to establish the full cross section of thebounding surface represented in Fig. 10 requiresonly 10 to 15 minutes on a PC; a non linear finiteelement analysis, i.e. a single point of the curve,requires more than 4 hours of computing in 2D and15 hours in 3D on a workstation with 4 parallelprocessors.This totally innovative concept, at least in seismicareas, clearly calls for extensive theoretical analysesand experimental validation. As sophisticated as
they can be, the theoretical and numerical tools donot have the capacity for modeling all the details ofthe behavior of this complex scheme during anearthquake. Centrifuge model tests were thereforeundertaken with a three-fold objective:• To validate the theoretical predictions of theultimate bearing capacity of the foundation undermonotonically increasing shear force andoverturning moment,• To identify the failure mechanism of thefoundation under these combined loads,• To assess the behavior of the foundation undervarious cyclic load paths.01002003004005006007000.00 0.05 0.10 0.15 0.20 0.25 0.30Displacement at foundation level (m)Horizontalforce(MN)M = 50 VM = 30 V0500010000150002000025000300000.0E+00 1.0E-03 2.0E-03 3.0E-03 4.0E-03 5.0E-03Rotation at foundation level (rd)Overturningmoment(MN-m)M = 50 VM = 30 VFig. 14: Finite element analyses: force-displacement curvesFour tests have been performed in the 200 g-tongeotechnical centrifuge at the LCPC Nantes center,(Pecker and Garnier 1999). It is designed to carry a2000 kg payload to 100 g accelerations, the bucketsurface is at a radius of 5.5 m and the platform has aworking space of 1.4 m by 1.1 m. All tests havebeen carried out at 100 g on models at a scale of1/100. The dimensions of the correspondingprototype are as follows:• radius of the circular footing: Bf = 30 m,• inclusions length and diameter: L = 8.5 m andB=0.67 m,• Wall thickness t = 6.7 mm (steel), Stiffness EI =158 MN.m2,• Thickness of the ballast layer: 1.2 m.The soil material has been sampled at the location ofpier N17 of the Antirion approach viaduct and sent tothe laboratory where it was reconsolidated prior tothe tests to reproduce the in situ shear strengthprofile. The loads applied to the foundation consistof a constant vertical force and of a cyclic shearforce and overturning moment; at the end of thetests the specimens are loaded to failure undermonotonically increasing loads with a constant ratioM/V. The main findings of the test are summarizedin Fig. 15 which compares the theoretical predictionsof the failure loads (class A prediction according tothe terminology introduced by Whitman) to themeasured failure loads. Disregarding thepreliminary tests that were carried out with anotherequipment (CESTA centrifuge in Bordeaux), all fourtests yield values within ±15% of the predictionsobtained with the Yield Design theory.The centrifuge tests not only provide the ultimateloads but also valuable information on factors thateither could not be easily apprehended by theanalysis or need experimental verification. Ofprimary interest is the fact that, even under severalcycles of loading at amplitudes representing 75% ofthe failure load, the foundation system does notdegrade; no tendency for increased displacementwith the number of cycles is noted. The equivalentdamping ratios calculated from the hysteresis loopsrecorded during the tests are significant with valuesas high as 20%; it is interesting to note that thesevalues have been confirmed by numerical analysesand can be attributed, to a large extent, to thepresence of the inclusions (Dobry et al. 2003).Finally, a more quantitative information is given bythe failure mechanisms observed in the centrifuge,which compare favorably either with themechanisms assumed a priori in the Yield Designtheory (Fig. 12), or with those computed from thenon linear finite element analyses (Fig. 16).The approach explained above takes care of thejustification of the inclusions. Another importantissue is the behavior of the gravel bed layer; afundamental requirement for this layer is to exhibit awell defined and controlled shear resistance, whichdefines the ultimate capacity for the sliding modes offailure. This requirement can only be achieved if nopore pressure buildup occurs in the layer during theearthquake; the development of any excess porepressure means that the shear resistance isgoverned by the soil undrained cyclic shear strength,a parameter highly variable and difficult to assesswith accuracy. The grain size distribution of thegravel bed layer (10-80 mm) is therefore chosen toensure a fully drained behavior (Pecker and al2001). Furthermore, the friction coefficient betweenthe slab and the gravel has been measured on sitewith friction tests using a concrete block pushed ontop of the gravel bed layer; a rather stable value(0.53 to 0.57) has been measured for thatcoefficient.
0204060801001200 50 100Measured horizontal failure load (MN)Computedhorizontalfailureload(MN)Fig. 15: Computed versus measured failureloads in the centrifuge tests:Preliminary tests (diamonds); final tests (triangles)Fig. 16: Failure mechanisms: centrifuge test(upper left), FE analysis (lower right)Seismic demandFor the seismic analyses it is mandatory to take intoaccount soil structure interaction, which is obviouslysignificant given the soft soil conditions and largepier mass. In the state of practice, the action of theunderlying soil is represented with the so-calledimpedance functions, the simplified version of themconsisting, for each degree of freedom, of constantsprings and dashpots. In the calculation of thesesprings and dashpots the soil modulus is calibratedon the free field strains, neglecting any further nonlinearity developed in the vicinity of the foundation(Pecker and Pender 2000). Such an approach maybe inadequate because it implicitly assumes that theforces generated on the foundation are independentof its yielding. Therefore there is someinconsistency in checking the foundation capacity forthose forces. Recent studies, (e.g., Paoluci 1997,Pedretti 1998, Cremer et al. 2001, Cremer et al.2002), throw some light on this assumption andclearly show that close to the ultimate capacity it isno longer valid. This is typically the situation facedfor the foundation of the Rion-Antirion bridge.Therefore it was decided to implement a morerealistic approach which closely reflects the physicsof the interaction. Since numerous parametricstudies are necessary for the design, a dynamic nonlinear soil finite element model is not the proper tool.In the same spirit as the impedance functions, theaction of the soil (and inclusions) below thefoundation slab is represented by what is called amacro-element (Cremer and al 2001, Cremer and al2002). The concept of macro-element is based on apartitioning of the soil foundation into (Fig. 17):• A near field in which all the non linearities linkedto the interaction between the soil, the inclusionsand the slab are lumped; these non linearities aregeometrical ones (sliding, uplift of the foundation)or material ones (soil yielding). The energydissipation is of hysteretic nature.• A far field in which the non linearities aregoverned by the propagation of the seismicwaves; the energy dissipation is essentially of aviscous type.In its more complete form the macro-element modelcouples all the degrees of freedom (verticaldisplacement, horizontal displacement and rotation)(Cremer et al. 2001). For the design studies of thebridge a simplified version of it, in which the degreesof freedom are uncoupled, is used. Conceptuallythis element can be represented by the rheologicalmodel shown at the bottom of Fig.17; it consists ofan assemblage of springs and Coulomb sliders forthe near field and of a spring and a dashpot for thefar field. One such model is connected in eachdirection at the base of the structural model; theparameters defining the rheological model arecalibrated on the monotonic force-displacement andmoment-rotation curves computed from the nonlinear static finite element analyses (Fig. 14). Forunloading-reloading a Masing type behavior isassumed; the validity of this assumption is backedup by the results of the centrifuge tests (Pecker andGarnier 1999) and of cyclic finite element analyses(Dobry et al. 2003). The adequacy of the macro-element to correctly account for the complex nonlinear soil structure interaction is checked bycomparison with few dynamic non linear finiteelement analyses including a spatial modeling of thefoundation soil and inclusions.Fig. 18 compares the overturning moment at thefoundation elevation calculated with both models; avery good agreement is achieved, not only in termsof amplitudes but also in terms of phases.
Aλ4λ3λ2λ1C0FoundationNear FieldFar FieldK0K1 K2 K3NEAR FIELDMFAR FIELDθK0 C0Fig. 17: Macro element for soil structureinteractionThis model has been subsequently used by thestructural design team for all the seismic analyses ofthe bridge. It allows for the calculations of the cyclicand permanent earthquake displacements; Fig.19shows the displacement of the foundation center(bottom) and the variation of the forces in the (N-V)plane (top); sliding in the gravel bed layer occurs atthose time steps when tanV N φ= .-16000-800008000160000 5 10 15 20 25 30 35 40 45 50Time (s)Overturningmoment(MN-m)Simplified modelFinite elementFig. 18: Time history of foundationoverturning momentCONSTRUCTION METHODSThe construction methods for the foundations,described in details by Teyssandier (2002 and2003), are those commonly used for the constructionof offshore concrete gravity base structures:• construction of the foundation footings in a drydock up to a height of 15 m in order to providesufficient buoyancy;• towing and mooring of these footings at a wetdock site;• construction of the conical part of the foundationsat the wet dock site;• towing and immersion of the foundations at itsfinal position.However some features of this project make theconstruction process of its foundations quiteexceptional.The dry dock has been established near the site. Itwas 200 m long, 100 m wide, 14m deep, and couldaccommodate the simultaneous construction of twofoundations. It had an unusual closure system: thefirst foundation was built behind the protection of adyke, but once towed out, the second foundation,the construction of which had already started, wasfloated to the front place and used as a dock gate.Verticalforce(MN)12008004000 2000 60040000VH = V tgφHorizontal force (MN)tanV N φ=Uy(m)-- 0.15-- 0.10-- 0.0500.0500.1000.15-0.35 -0.25 -0.15 -0.05UxyUx (m)Fig. 19: Time history of foundationdisplacementTop – variation of forces in the (N-V) planeBottom - displacement of foundation centerDredging the seabed, driving of inclusions, placingand levelling the gravel layer on the top, with a waterdepth reaching 65 m, was a major marine operationwhich necessitated special equipment andprocedures. In fact, a tension-leg barge has beencustom-made, based on the well known concept oftension-leg platforms but used for the first time formovable equipment.
This concept was based on active verticalanchorage to dead weights lying on the seabed (Fig.20). The tension in these vertical anchor lines wasadjusted in order to give the required stability to thebarge with respect to sea movements and loadshandled by the crane fixed on its deck. By increasingthe tension in the anchor lines, the buoyancy of thebarge allowed the anchor weights to be lifted fromthe seabed, then the barge, including its weights,could be floated away to a new position.As already stated, once completed the foundationsare towed then sunk at their final position.Compartments created in the footings by the radialbeams can be used to control tilt by differentialballasting. Then the foundations are filled with waterto accelerate settlements. This pre-loading wasmaintained during pier shaft and pier headconstruction, thus allowing a correction for potentialdifferential settlements before erecting pylons.The deck of the main bridge is erected using thebalance cantilever technique, with prefabricateddeck elements 12 m long comprising also theirconcrete slab (another unusual feature).Fig. 20: Tension-leg bargeACKNOWLEDGMENTSThe development of this original foundation conceptwas made possible thanks to the full cooperationbetween all the parties involved in the project: theConcessionaire (Gefyra SA), the Contractor (GefyraKinopraxia), the Design JV and the Design Checker(Buckland & Taylor Ltd). The author is indebted toMr. Teyssandier from Gefyra SA, Mr De Maublancand Morand from Gefyra Kinopraxia, Mr Tourtoisfrom the Design JV and Dr. Taylor for their long termcollaboration and confidence. He also wishes tospecially thank Dr Peck, Dr. Dobry and Dr Gohl,consultants to the Checker, for their invaluablecontribution through their constant willingness tohelp him clarify and develop the concept.More information on the project can be found on theweb site www.gefyra.gr.REFERENCESCombault, J., Morand, P., Pecker, A. (2000)."Structural response of the Rion Antirion Bridge."Proc. of the 12th World Conf. on Earthq. Eng.,Auckland, Australia.Cremer, C., Pecker, A., Davenne L. (2001). "Cyclicmacro-element for soil structure interaction - Materialand geometrical non linearities." Num. Methods inGeomech., 25, 1257-1284.Cremer, C., Pecker, A., Davenne L. (2002)."Modelling of non linear dynamic behaviour of ashallow strip foundation with macro-element." J. ofEarthq. Eng., 6(2), 175-212.de Buhan, P., Salençon, J. (1990). "Yield strength ofreinforced soils as anisotropic media." in Yielding,Damage and Failure of Anisotropic Solids; EGF5,Mechanical Engineering Publications, London.Dobry, T., Abdoun, T., ORourke, T.D., Goh, S.H.(2003). "Single piles in lateral spreads: Field bendingmoment evaluation". J. Geotech. and Geoenv. Eng.,ASCE, 129(110), 879-889.Dobry, R., Pecker, A., Mavroeidis, G., Zeghal, M.,Gohl, B., Yang, D. (2003). "Damping/Global EnergyBalance in FE Model of Bridge Foundation LateralResponse." J. of Soil Dynamics and Earthq. Eng.,23(6), 483-495.Paolucci, R. (1997). "Simplified evaluation ofearthquake induced permanent displacement ofshallow foundations." J. of Earthq. Eng., 1(3), 563-579.Pecker, A. (1998). "Capacity Design Principles ForShallow Foundations In Seismic Areas." Proc. 11thEuropean. Conf.. on Earthq. Eng., A.A. BalkemaPublishing.Pecker, A., Salençon, J. (1999). "GroundReinforcement In Seismic Area." Proc. of the XIPanamerican Conf. on Soil Mech. and Geotech.Eng., Iguasu, 799-808.
Pecker, A., Garnier, J. (1999). "Use of CentrifugeTests for the Validation of Innovative Concepts inFoundation Engineering." Proc. 2ndInt. Conf. onEarthq. Geotech. Eng., Lisbon, 433-439.Pecker, A., Pender, M.J. (2000). "Earthquakeresistant design of foundations : new construction."GeoEng2000 , 1, 313-332.Pecker, A., Prevost, J.H., Dormieux, L. (2001)."Analysis of pore pressure generation anddissipation in cohesionless materials during seismicloading." J. of Earthq. Eng., 5(4), 441-464.Pedretti, S. (1998). "Nonlinear seismic soil-foundation interaction: analysis and modelingmethod." PhD Thesis, Politecnico di Milano, Milan.Salençon, J. (1983). "Introduction to the yielddesign theory." European Journal of MechanicsA/Solids 9(5), 477-500.Teyssandier, J.P., Combault, J., Pecker, A. (2000)."Rion Antirion: le pont qui défie les séismes." LaRecherche, 334, 42-46.Teyssandier, J.P. (2002). "Corinthian Crossing." CivilEngineering, 72(10), 43 -49.Teyssandier, J.P. (2003). "The Rion-Antirion bridge,Design and construction." ACI InternationalConference on Seismic Bridge Design and Retrofit.La Jolla, California.