Site classification using equivalent soil profiles for building-liquefaction interaction

Abstract

The seismic behaviour of a building on a liquefiable deposit is a complex interaction which involves quantifying both shaking induced damage and permanent ground deformation-related damage. In this paper the key parameters that influence both surface shaking and foundation settlements have been identified as the depth, thickness and liquefaction resistance of an equivalent liquefiable layer. These parameters can be used to develop an 'equivalent soil profile' that is analogous to the equivalent single degree-of-freedom that reduces the complexity of the dynamic response of a building into comparable and easily understood quantities. The equivalent soil profile is quantified independent of the seismic hazard, making it compatible with performance based design and assessment frameworks such that the building and soil profile can be directly assessed at different levels of seismic hazard. Several numerical studies are presented that demonstrate the influence of these key parameters on the ground surface shaking and foundation settlement. A set of criteria are proposed for classifying soil profiles into 22 different soil classes for regional loss assessment. An algorithm was developed for automatically fitting the equivalent soil profile to a cone penetration test trace and issues with the fitting are discussed. Field reconnaissance was undertaken to collect additional data to support existing datasets on the performance of buildings in Adapazari, during the 1999 Kocaeli, Turkey, earthquake (Mw = 7.4). The field case history data was used to investigate the correlation between the depth, thickness and liquefaction resistance of an equivalent liquefiable layer, on the extent of foundation permanent deformation. The case history data showed that in general a shallow, thick and weak liquefiable layer near the surface results in significant settlement but a lack of data for buildings on non-liquefiable deposits and the additional complexities involved with real buildings and soil deposits, meant that the trends observed in the idealised numerical models could not identified in the field case history data set.