Introduction
Located between the Peloponnese and the continent, at the entry of the Gulf of Corinth in Western Greece, the Rion-Antirion Bridge (Fig. 1) is intended to replace an existing ferry system.
The crossing consists of a main bridge, 2252 m long and 27.20 m wide, connected to the land by two approaches, respectively 392 m and 239 m long, on each side of the gulf.
An exceptional combination of environmental and physical conditions made the project quite complex:
1) Large water depth;
2) Deep soil strata of weak alluviums;
3) Strong seismic activity;
4) Possible tectonic movements;
5) Possible ship impacts.
Design features
Indeed, the structure spans a stretch of water about 2500 m long. The seabed presents fairly steep slopes on each side and a long horizontal plateau at a depth of 60 to 70 m.
No bedrock has been encountered during soil investigations down to a depth of 100 m. Based on geological studies, it is believed that the thickness of sediments is greater than 500 m.
General trends identified through soils surveys are the following:
1) A cohesionless layer is present at mud line level consisting of sand and gravel to a thickness of 4 m to 7 m, except in one location (near the Antirion side), where its thickness reaches 25 m.
2) Underneath this layer, the soil profile, rather erratic and heterogeneous, presents strata of sand, silty sand and silty clay.
3) Below 30 m, the soils are more homogeneous and mainly consist of clays or silty clays.
In view of the nature of the soil, liquefaction did not appear to be a problem except on the north shore, where the first 20 m are susceptible of liquefaction.
The seismic conditions to be taken into account were presented in the form of a response spectrum at seabed level (Fig. 2).
The peak ground acceleration is equal to 0.48 g and the maximum spectral acceleration is equal to 1.2 g between 0.2 s and 1.0 s.
This spectrum is supposed to correspond to a 2000 year return period.
In addition, the Peloponnese being drifting away from mainland Greece by a few millimeters per year, contractual specifications required the bridge to accommodate possible fault movements up to 2 m in any direction, horizontally and/or vertically, between two adjacent supports which could simultaneously be the subject of small inclinations due to the corresponding rearrangement of the sea bed below the foundations.
Finally, the bridge supports had to be capable to withstand the impact from an 180000 dwt tanker sailing at 16 knots.
Clearly, these site conditions should have favored the design of a suspension bridge but a major slope stability problem on the Antirion side eliminated such a solution from the very beginning of the conceptual design stage.
Main bridge concept
Instead, the bridge type and the span lengths had to be selected to simply make the bridge feasible and the global cost of the project acceptable by limiting the number of supports located in the strait and, finally, an exceptional cable stayed bridge (Fig. 3) made of three central spans 560 m in length and two side spans 286 m long was selected [
1].
Of course, the foundation of any support under more than 60 m of water was a major point of concern:
1) As the seismic activity in the area is severe, it is clear that an earthquake will unavoidably lead to high soil structure interaction forces at any bridge support location.
2) As these high forces have to be resisted by weak layers of soil, no classical foundation type was available.
Foundations
The four pylons of the main bridge simply rest on the seabed through a large concrete substructure foundation, 90m in diameter, 65m high at the deepest location (Fig. 4).
To provide sufficient shear strength to the top 20 m of soils, which are rather heterogeneous and of low mechanical characteristics, the upper soil layer of the seabed is reinforced by inclusions to resist large structural inertia forces and hydrodynamic water pressures.
These inclusions (Fig. 5) are hollow steel pipes, 25 to 30 m long, 2 m in diameter, driven into the upper layer at a regular spacing of 7 to 8 m (depending on the pier).
About 150 to 200 pipes were driven in at each pier location. They are topped by a 3 m thick, properly levelled gravel layer, on which the foundations rest. Due to the presence of a thick gravel layer, these inclusions are not required under one pylon.
Pylons
The pylon bases consist of a 1 m thick bottom slab and 32 peripheral cells enclosed in a 9 m high perimeter wall and covered by a top slab slightly sloping up to a conical shaft. For the deepest pier, this cone, 38 m in diameter at the bottom, 27 m at the top, rises 65 m over the gravel bed up to 3 m above sea level.
These huge bases support, through vertical octagonal pylon shafts, 24 m wide and nearly 29 m high, a 15.8 m high pyramidal capital which spreads to form the 40.5 m wide square base of four concrete legs.
Rigidly embedded in the capital to form a monolithic structure, the four legs (4.00 m × 4.00 m), made of high strength concrete, are 78 m high; they converge at their tops to impart the rigidity necessary to support asymmetrical service loads and seismic forces.
They are topped by a pylon head, 35m high, comprising a steel core embedded in two concrete walls, where stay cables are anchored.
From sea bottom to pylon top, the pylons are up to 230 m high (Fig. 6).
Deck and cables
The deck is a composite steel-concrete structure, 27.20 m wide, made of a concrete slab, 25 to 35 cm thick, connected to twin longitudinal steel I girders, 2.20 m high, braced every 4 m by transverse cross beams (Figs. 7 and 8).
It is fully suspended from 8 sets of 23 pairs of cables and continuous over its total length of 2252 m, with expansion joints at both ends.
In the longitudinal direction, the deck is free to accommodate all thermal and tectonic movements and the joints are designed to accommodate 2.5 m displacements under service conditions and movements up to 5.0 m under an extreme seismic event.
In the transverse direction it is connected to each pylon with 4 hydraulic dampers of 3500 KN capacity each and a horizontal metallic strut of 10000 KN capacity (Fig. 8).
The stay cables are arranged in two inclined planes according to a semi-fan shape. They are made of 43 to 73 parallel galvanised strands individually protected.
Design process and philosophy
Initially, the huge pylon bases supported, through a sophisticated set of bearing devices, P.T tendons and powerful spring dampers, a concrete block which was part of the deck and the base of the concrete legs converging at the top of the pylons and giving them the appropriate rigidity (Fig. 9); the deck was made of 510 m long double cantilevers supported by the stay cables anchored in the corresponding pylon head and each cantilever was connected to the adjacent one or to the approaches by a simply supported drop-in span 50 m long. This was deemed to be the only way for the structure to accommodate impressive tectonic movements.
As long as the signature of the contract was delayed by the banks and the bridge was really an enormous undertaking, it was decided to spend about one year to proceed with sophisticated parametric studies in view of optimizing the concept and the structure as well.
Clearly, the design of the Main Bridge was mainly governed by the capability of the whole structure to resist the major seismic events including a possible fault movement but this meant that the structure had to be first designed to resist what will be the main actions during its span life (i.e. for the classical Serviceability Limit States and the corresponding Ultimate Limit States) and that the main components of the structure had to be adjusted to the demand during a given design earthquake in terms of acceptable damage. This was the best way to get the most flexible structure and therefore the most favorable concept from a seismic behavior point of view.
Bearing capacity of the reinforced soil and parametric studies
The foundations are a typical example of a major part of a structure where the performance of the concept had to be evaluated through the capacity of the soil, to resist the soil-structure interaction during the earthquake event, and the ability of the structure to be the subject of exceptional displacements (generated by the ground motion) with a controlled damage considered as acceptable.
In the case of the Rion-Antirion Main Bridge, the foundations of the structure consist of two separate parts (Fig. 5):
1) The reinforced soil, which is a clay-steel composite 3D volume.
2) The pylon bases, which are rigid bodies not subject to any unusual strength problems.
These parts are made partially independent through the gravel layer which was designed to transfer a range of horizontal forces compatible with the strength of the reinforced soil, the global stability of the structure and acceptable permanent displacements of the pylons. Although the foundations look like piled foundations, they do not at all behave as such: no connection exists between the inclusions and the pylon rafts. The pylon bases are therefore allowed to uplift or to slide with respect to the reinforced soil.
The capacity design philosophy, introduced in foundation engineering for the evaluation of the seismic bearing capacity of shallow foundations through the yield design theory, was then extended to this innovative foundation concept in seismic areas [
2]. Using the yield design theory, through a set of appropriate kinematic mechanisms (Fig. 10), it was possible to derive an upper bound estimation of the global bearing capacity of the reinforced soil (Fig. 11).
For this purpose the reinforced soil was modeled as a two-dimensional continuum appropriately connected to beams simulating the stiff inclusions. Consequently, the calculations included the contribution of the inclusions to the overall resistance of this new concept. The simplicity of such calculations allowed optimizing the size and the spacing of the inclusions. A set of centrifuge tests was run to validate the concept and the theoretical approaches.
These preliminary analyses of the behavior of the reinforced soil and resulting improvements of this innovative concept lead to give up the initial static scheme of the main bridge and to definitely move toward a much more efficient structure with a continuous pylon (from sea bed to pylon head) and a continuous deck fully suspended and therefore isolated as much as it could be. This made it also possible to reduce the depth of the deck and therefore the wind effects on the bridge.
Nonlinear finite element analyses of the reinforced soil
Nonlinear finite element analyses could be then carried out (Fig. 12). They lead to the constitutive laws of the reinforced soil which were used in the general calculations of the structure.
All these calculations, adequately combined with a global dynamic analysis, allowed checking that the effect of the coupled gravel layer and soil reinforcement was to improve the bearing capacity of the whole foundation system while controlling the failure mode:
1) The fuse provided by the gravel layer limits the maximum shear force at the interface, dissipates energy by sliding and forces the foundation “to yield” according to a mode which is compatible with an acceptable behavior of the structure.
2) The stiff inclusion reinforcement increases the strength capacity of the soil in order to eliminate undesirable failures modes, like rotational failure which would compromise dangerously the global stability of the structure, and dissipates an important amount of energy as it could be anticipated from the Force-Displacement Diagram.
General calculation of the bridge (dynamic analyses)
All the previous calculations and results were used to carry out detailed and careful 3D dynamic analyses of the whole structure [
3]. Thanks to the development of a certain number of calculation tools on the basis of an existing powerful computer software (ANSYS), the following very important properties were taken into account:
1) Non linear hysteretic behavior of the reinforced soil;
2) Possible sliding of the pylon bases on the gravel beds precisely adjusted to the accompanying vertical force;
3) Non linear behavior of the reinforced concrete of the pylon legs (including cracking and stiffening of concrete due to confinement);
4) Non linear behavior of the cable-stays;
5) Non linear behavior of the composite bridge deck (including yielding of steel and cracking of the reinforced concrete slab);
6) Second order effects (or large displacements if any).
Several sets of independent artificial accelerograms conforming to the seismic design spectrum for the three components of the ground motion (the vertical one being scaled to 70%) were used.
From these calculations, the way the reinforced soil behaves and the bases slide could be carefully checked.
Behavior of the reinforced soil
The general calculation of the bridge, including a lumped parameter model of the reinforced soil, allowed checking all the components of the software specifically developed for this bridge. The results were consistent with the assumptions. They showed that the forces and overturning moments applied to the soil are always remaining within the bounding surface. They confirmed the very good behavior of the fully suspended deck which is isolated as much as it can be. The relative displacement of the pylon bases with respect to the gravel layer evidenced some sliding which remains nevertheless acceptable and if, for any reason, this sliding could not occur it has been checked that this was not a major point of concern. Under the most severe seismic event, the substructure of the bridge will slide (Fig. 13); the pylon bases will slightly rotate; but all this will happen without any detrimental effect for the structure of the bridge as the fully suspended and flexible deck is able to align automatically and lately to be adjusted to an acceptable geometry through a re-tensioning of the stay cables.
Behavior of the structure
Because the stability of the fully suspended multi cable-stayed span deck is secured by the stiffness of the pylons which consist therefore of four legs converging at mid height of the anchorage zone, the pylons were the most critical parts of the structure. The dynamic analyses evidenced that pylons and shortest cable-stays are indeed heavily loaded during the earthquake event. Clearly, from this point of view, there is a contradiction between what is required for the normal operation of the bridge and the demand when a severe earthquake occurs. Indeed, the pylons are too stiff and the shortest cables as designed for serviceability are not flexible enough.
Dynamic calculations (Fig. 14) have shown that the extreme shakes generate various crack patterns, distributed along the legs, coming from both bending and tension. On the one hand, it could be observed that this cracking is favorable as it generates the necessary flexibility of the legs without leading to unacceptable strains in the materials (i.e. non acceptable damages). On the other hand, it was not an easy task to get a global view of the behavior of the pylon as the information produced by a sophisticated analysis is too impressive. Time steps being 0.02 s, i.e., 2500 steps for a 50 s event-the number of cross-sections in the model of one pylon leg being 13-this means that there would be 130.000 configurations of reinforced concrete cross-sections to be checked for each pylon in order to evaluate the global behavior of the structure at any time.
To face this voluminous quantity of information, the option was to check, for the duration of the earthquake, that the strains of the materials (concrete and steel) in each cross-section are not exceeding the acceptable limits which guarantee a controlled damage of the pylons while the general consistency of these sophisticated calculations through the corresponding deflection shapes of the legs, axial shear forces and bending moments generated in each cross-section, can be verified for time history peak values of these parameters.
Push-over analyses of the pylons
Under these conditions, it made sense to carry out a push-over analysis of the pylons to evaluate their global behavior and compare their performance to the demand, in terms of displacements, during the extreme seismic event. It can be pointed out that such a push-over analysis has become usual. Moreover it is extremely simple for a high pier of a bridge which behaves as a single degree of freedom system and is therefore loaded by a shear force acting at the level of the center of gravity of the bridge deck. It is not that simple anymore when the pier has become a pylon group made of four legs converging in a zone where a large number of cables are generating many forces at various levels.
In this case, one way of performing such a push-over analysis consists in reproducing the state of equilibrium at a stage of the dynamic analysis which can be considered as the most unfavorable situation during the 50 s event, i.e., when forces, bending and displacements are the most severe. This approach allowed assessing the effect of the displacement demand on a pylon as well as its displacement capacity as estimated from the 3D dynamic analysis.
In a static analysis on a precise model of the pylon, inertial forces coming from the deck through the cables and from the pylon concrete mass acceleration were gradually increased by a magnification factor while gravity or initially applied forces (permanent loads) were not.
The diagram showing the displacement D at the top of the pylon legs versus the magnification factor A allowed to make a clear differentiation of the various steps characterizing the behavior of a whole pylon group (Fig. 15). As the displacement is mainly diagonal, these steps are as follows:
1) Step 1 (0<A<0.4) -elastic behavior 0<D<0.10 m.
2) Step 2 (0.4<A<1.2)-axial Cracking in the tension leg, hinges forming at the top of this leg then at the top of the middle legs (0.10 m<D<0.45 m).
3) Step 3 (1.2<A<1.4)-yielding of steel in the tension leg (0.45 m<D<0.60 m).
4) Step 4 (1.4<A<1.6) -hinge forming at the top of the compression leg (0.60 m<D<0.90 m).
Such a push-over analysis showed that the displacement demand (D = 0.36 m for A = 1) is far under the displacement capacity of the pylon legs which is of the order of 0.90 m at maximum and, therefore, either that the damage should be limited in case of an extreme event or that any deviation with regard to the input motion should not have any bad consequences.
Construction
The main bridge concept has undergone a spectacular evolution, which took into account all the aspects of the project economy and that has been the result of the close interaction between design and study of realistic construction methods.
Unusual aspects of construction methods
As already mentioned, construction of the main bridge was facing the major difficulty of an unusual water depth, which reaches 65m for central piers, and bad quality of the seabed. In relation with this, foundation works, including not only dredging and steel pipe driving but also exceptional works like precision spreading of an 8000 m2 gravel bed, were forming an impressive package requiring unusual skill and equipment. To achieve this challenging task, a combination of the latest technologies available in the construction of concrete off-shore oil drilling platforms, immersed tunnels and large cable stayed bridges was extensively used.
Pylon base construction
Pylon bases were built in two stages near Antirion: the footings were cast first in a dry dock 230 m long and 100 m wide; the conical shafts were completed later in a wet dock where water deep enough was available.
In the dry dock, two cellular pylon footings were cast at a time (Figs. 16 and 17). In fact, two different levels in the dock provided 12 m of water for casting a leading footing and 8 m for another one behind. When the first footing, including a leading 3.2 m lift of the tapering shaft, was complete, the dock was flooded and the nearly 17 m tall structure was towed a few hundred meters to deep water (Fig. 18).
A very original idea allowed saving a lot of time in the production cycle of the pylon bases. Before the first tow-out, the dry dock was closed by a classical sheet piled protection dyke which had to be completely extracted. Clearly, build and remove such a dyke again would have been terribly time consuming.
In fact, as long as the less advanced second footing was floated forward into the deeper dock and sunk by flooding, it was possible to use it as a gate, providing that everything was designed to do so. Instead of building a dyke again, temporary steel walls around the base top slab and sheet piles projecting from its sides could easily seal the dock mouth, allowing it to be de-watered (Fig. 19).
As work resumed in the dock, the conical shaft of the leading base anchored with chains in the wet dock was progressively cast with classical jump forms atop the footing. Cells in the base were progressively flooded to sink the structure and maintain a constant height above water (Fig. 20). When a pylon base was tall enough to stand a few meters above water, tugs were taking it to its prepared bed where it was ballasted down and placed at its final location. It was then pre-loaded by being filled with water, to speed up and anticipate settlements (between 20 and 30 cm) during pylon shaft and capital construction, thus allowing a correction for differential settlements when erecting the pylon legs.
Foundation works and tension leg platform
Foundation work began in October 1999 by dredging the seabed at pylon locations, laying a 90 cm thick sand layer, driving the inclusions and leaving them projecting 1.5 m above the sand to be finally covered by a 1.6 m to 2.3 m thick layer of rounded river gravel and a 50 cm thick layer of crushed gravel. Gravel were laid in parallel berms, 2 m wide, separated by V shaped cuts around 30 cm deep to provide some flexibility when placing the pylon bases.
All these marine works were performed, step by step, from a 60 m long and 40 m wide tensioned leg platform anchored in movable concrete blocks by adjustable chains (Fig. 21). Equipment for driving soil reinforcing tubes and preparing the seabed was mounted on submersible pontoons anchored to one end of the platform with steel arms. A movable steel tube, reaching nearly to the sea floor, guided piling equipment and deposited sand and gravel on the pre-dredged bed. This equipment permitted to perform the necessary works on a 14 m wide, 28 m long area and the platform had to be moved from one area to the next one by a barge equipped with a dynamic positioning system. A permanent Sonar scanning of the finished gravel bed allowed precise controlling of the achieved foundation level from the platform and showed it was generally within a 5 cm tolerance. Forty displacements of the platform, taking about five months, were therefore necessary to prepare the sea bed under each pylon base.
Construction of the upper part of the pylons
For the remaining sections of the pylons, all materials, concrete, reinforcement, Post Tensioning and specific devices were provisioned through a support barge, used as a fixed base, and a roll-on roll-off barge transporting the truck mixers and the reinforcement from the coast to the pylons. The octagonal shafts of three pylons were cast in place with self climbing formworks.
The huge inverted pyramidal capitals are key elements of the pylon structures; they have to withstand the tremendous forces coming from the legs, mainly during a seismic event, and to transfer them to the shafts. This is the reason why they are heavily reinforced and pre-stressed. The construction of these components, which were cast in place, took seven months and required 4000 m3 of concrete, 1750 t of steel reinforcement and 30000 m2 of external forms as well as impressive equipment.
The construction of the pylon legs was progressing step by step, in 4.8 m long sections, up to the point where they merge to support the cable anchoring zone. A heavy temporary bracing allowed them to resist earthquakes during the construction (Fig. 22).
The steel core of the pylon head was made of two units fabricated in a factory and transported to the site. They were placed at their final location by a huge floating crane able to reach a height of 170 m above sea level (Fig. 23).
Erection of the deck
The construction method of the composite steel-concrete deck was similar to the one successfully used on the second Severn crossing. Deck elements, 12 m long and including the concrete slab, were prefabricated on a preassembly yard.
They were placed at their final location by the floating crane and bolted to the previously assembled segments, using the classical balanced cantilever erection method (Fig. 24). Only small joints providing enough space for an appropriate steel reinforcement overlapping had to be cast in place.
The deck was erected from two pylons at the same time. Five to seven deck segments were put in place each week. In total the deck erection took 13 months.
Conclusions
The Rion-Antirion bridge is clearly a major and impressive link when compared to the second Severn cable-stayed bridge and even to the Normandy bridge. The design and construction of this US $800 million project undertaken under a private BOT scheme could overcome an exceptional combination of adverse environmental conditions thanks to the choice of an appropriate concept and seismic design philosophy: the pylons rest directly on the sea bed, through a gravel layer allowing them to be the subject of controlled displacements under the most severe earthquake and, based on an innovative concept, the top 20 m of soils located under the large diameter bases (90 m) of the pylons are reinforced by means of steel inclusions to resist high soil-structure interaction loads; the 2252 m long deck of the cable-stayed bridge is continuous deck, fully suspended and therefore isolated as much as possible from the worst seismic shakes. If small damages may be experienced in the pylon legs after the big seismic event, the whole bridge will be safe and still opened to the emergency traffic if necessary.
Completed in August 2004, the Rion-Antirion Bridge has been opened to traffic four months before the contractual deadline.
Higher Education Press and Springer-Verlag Berlin Heidelberg
PDF
(1272KB)
