Centrifuge modelling tests show clearly the phase change in lateral acceleration in the backfill behind a retaining wall as shear waves propagate from the base of the model towards the ground surface. However, design calculations for the dynamic lateral earth pressure on a retaining wall which use a pseudo-static approach assume that the backfill experiences a uniform acceleration throughout. Researchers have agreed that the total lateral earth pressure calculated using this approach is approximately correct, but have disagreed over the distribution of the dynamic increment of pressure. The Paper presents an analysis which takes into account a finite shear wave velocity in the backtill, thus allowing for the phase change in a prototype structure. The phase change does not have a significant influence on the magnitude of the total earth pressure, but it has a marked effect on the distribution of the dynamic increment. The resultant pressure is seen to act at a point above one third of the height of the wall. The maximum dynamic earth pressure and the peak bending moment on the wall are approximately in phase with the acceleration at mid-depth, and therefore this acceleration may be the most appropriate value to nse for design. The effect of a non-uniform shear modulus distribution is considered, as is the effect of amplification of acceleration on the distribution and magnitude of earth pressure. Ampliication of acceleration has an influence similar in character to the effect of increasing the acceleration coefftcient in a uniform acceleration field. Centrifuge model test data analysed using this approach show good agreement if the amplification of motion is taken into account. KEYWORDS:analysis; centrifuge modelling; earth pressure; ear&quakes; retaining walls; soilstructure interaction.Des es&s de modele i la centrifugeuse indiquent clairemeat le changement de phase de l'acceleration laterale dans le remblai derriere un mur de soutenement pendant que les ondes de cisaillement se propagent 1 partir de la base du modhle vers la surface du terrain. Les calculs classiques pour la pression laterale dynamique des terres sur un mur de soutenement qui emploient un methode pseudostatique admettent cependant que le remblai entier subit une acceleration uniforme. Les rechercheurs reconnaissent que la pression laterale totale des terres calculb par cette methode est approximativement correcte, mais ne sont pas d'accord sur la distribution de l'incremeot dynamique de pression. L'article prbente une analyse qui inclut une velocite finie de l'onde de cisaillement dans le remblai. Le changement de phase darts une structure prototype eat ainsi pris en compte. Le changement de phase est pratiquement sans influence sur la valeur totale de la pression des terres, tandis qu'il prod& un effet marque sur la distribution de l'increment dynamique. On observe que la pression r6sultante s'applique au-dessus du tiers de la hauteur du mur. La pression maximale dynamique des terres et le moment maximal de flexion sur le mur se trouvent appro...

The failure of quay walls induced by earthquakes has been widely observed in the field. A study of the seismic response of anchored flexible quay walls based on the data of dynamic centrifuge model tests is reported, and the applicability and limitations of the widely used pseudo-static calculations are investigated. It is found that amplification of motion due to resonance may cause a greatly increased earthquake response over that predicted by a pseudo-static analysis. Resonant vibration may be a consequence of the deterioration of soil stiffness in the backfill, which is itself a result of cyclic loading and excess pore pressures induced by base shaking. This analysis is strongly supported by the data of centrifuge model tests. Principles for design are proposed, which by a model test are shown to be effective.La rupture des murs de quai provoqu&e par des Gsmes a eti largement observi?e in-situ. L'article prbente une etude de la reponse sismique de murs de quai flexibles et ancrb reposant sur les rbultats obtenus au tours d'essais de modklisation par centrifugation dynamique. Les limites d'applicabiliti! des calculs pseudo-statisques sont Ctudi&es. II apparait qu'une augmentation du mouvement di Z+ la Gsonance peut entrainer une Gponse sismique bien sup&rieure P celle que pri?voit I'analyse pseudo-statique. La vibration de r(?sonance pourrait avoir pour origine une diminution de la raideur du remblai, qui rbulterait elle-m&me du chargement cyclique et de I'exc&s de pression interstitielle c&s lors de l'bbranlement de la base du mur. Cette analyse est confor& par les essais de modSsation. Des principes de conception d'ouvrages, vbrifiis sur un essai de modClisation, sont ensuite proposb.

No abstract