UCLA

Pile foundations have suffered extensive damage in liquefiable soils during earthquakes around the world, particularly when a nonliquefied crust layer spreads laterally on top of underlying liquefied layers. Recent research has begun to clarify certain aspects of the problem, but many fundamental mechanisms have not yet been explored. As a result, designers are faced with making uninformed assumptions that could result in structural damage and the associated loss of life and money, or excessive conservatism in the design and the associated unnecessary high cost of construction. A series of dynamic centrifuge model experiments was performed on a 9-m radius centrifuge to improve our understanding of these phenomena. One model contained a number of single piles and a two-pile group while the other models each contained a six-pile group. Soil profiles consisted of a nonliquefiable clay crust overlying liquefiable loose sand over dense sand. While the models spun on the centrifuge to simulate a large prototype, a servo-hydraulic shaker applied a sequence of realistic earthquake motions. Data was collected from dense arrays, and was subsequently archived and made publicly available. The raw recorded data was processed to obtain time series of fundamental engineering behaviors, some of which had never before been measured or anticipated. The observed subgrade reaction loads between piles and liquefied sand, phasing of lateral spreading loads and inertia loads, and load transfer between pile groups and laterally spreading nonliquefiable crusts were all shown to be more complex than commonly assumed in design practice, and represent a challenge to our expanding numerical modeling capabilities. The complex observations were distilled into a simplified design guidelines and recommendations that can directly be implemented at a standard-of-practice computational level. Suites of hundreds of simplified analyses were performed to identify the potential accuracy of standard-of-practice methods, the influence of alternative assumptions and approximations, and the sensitivity of predicted bending moments to variations in the properties of the p-y materials, magnitude of inertia forces, magnitude and shape of ground displacements, axial capacities, flexibility of connections between piles and pile caps, and stiffness of the load transfer relation between the spreading crust and pile foundation.