Numerical and experimental dynamic soil response studies

Abstract

This report presents both an analytical and experimental investigation of the dynamic response of soil sites. The focus of this work was to develop a suitable numerical method to analyse the multi-dimensional response of soil deposits and to gain an insight into the characteristics of the dynamic response of volcanic soils. Although soils of volcanic origin are commonly found in New Zealand, very little data has been published to date in the literature on this subject. A three dimensional finite element effective stress analysis computer program was developed to investigate the response of nonlinear soil deposits and soil structures to multi-directional seismic loading. Validations with closed form elastic solutions, existing nonlinear site response methods and a simple case study of an earth dam indicated this program was capable of modelling the dynamic response of soil deposits and earth structures. The liquefaction of pumice sands during the 1987 Edgecumbe earthquake and the high seismicity and occurrence of volcanic soil deposits in central North Island areas Iead to the instigation a dynamic testing programme on a pumice sand. An aim of this work was to quantify the differences and similarities of the pumice sand to sands of quartzitic origin on which much of the data and correlations in the literature is based. Cyclic triaxial tests, dynamic torsion tests and bender element tests of sand samples were undertaken. The results indicated that the pumice sand differed significantly from the behaviour of quartz sands, particularly with respect to empirical correlation factors that are widely used in engineering practice. It was found that the geometry of a site could have a significant effect on both the total and effective stress response of a soil valley, which indicates that the one dimensional assumption in particular is not valid for all cases. Both two dimensional in plane and out of plane analyses are affected by the in plane boundary conditions, although this is significantly Iess so for the out of plane case. Three dimensional dynamic analyses showed that multi directional earthquake loading influences the behaviour of a soil for both total stress and effective stress cases. The numerical requirements of three dimensional analyses were found to be very significant, which still limits these types of analyses.