Centrifuge and numerical modelling of twin-propped retaining walls.

Abstract

A series of tests on flexible model diaphragm walls embedded in an overconsolidated clay was carried out at the London Geotechnical Centrifuge Centre. The walls were propped at the crest and, following the simulation of excavation, were propped at formation level. Although a retained height of 10m was modelled in all tests, the depth of embedment below dredge level varied between Sm, 10m and 15m. A 'softer' propping sequence was also investigated with excavation to 5m below the retained surface prior to the installation of the crest level prop then, following further excavation to dredge level, the bottom prop was installed. The pre-excavation lateral earth pressure was also investigated. The background and use of twin-propped retaining walls is discussed together 'with the design of the centrifuge model and modelling procedure. The results of the tests are presented and the effects of embedment depth, construction sequence and pre-excavation lateral earth pressure coefficient is discussed. A series of finite element analyses using the critical state soils program CRISP was undertaken in which it was attempted to model the centrifuge models. Generally, results were in reasonable agreement, although it was discovered that the calculated wall movements and prop loads were sensitive to the slope of the Hvorslev surface required for the Schofield soil model used in the analyses. The prop loads from the centrifuge tests and finite element analyses were compared with prop loads calculated using popular empirical methods and with prop loads observed on site. Generally, prop loads were underpredicted using the empirical methods which are unable to account for construction sequence effects and probably overestimate the degree of lateral stress reduction that takes place during excavation.