In situ tests for liquefaction potential evaluation: Application to California data including data from the 1989 Loma Prieta earthquake

The main objective of this dissertation is directed towards developing relationships to evaluate liquefaction potential using in-situ tests, in particular the Dilatometer and Cone Penetrometer. A large set of field data has been collected from various sources for sites where liquefaction and no liquefaction were observed during earthquakes. In particular, field data was obtained in the Imperial Valley and San Francisco, California, at locations where strong motion records were available. The levels of variability of the dilatometer and cone parameters are studied using a statistical approach. The study shows that, in general, the cone and dilatometer data exhibit levels of variability that are essentially the same. Hence, there is no apparent advantage from the statistical standpoint of preferring one test with respect to the other to evaluate liquefaction potential of sands. Based on one dimensional wave propagation analyses at different sites in San Francisco, a criterion to subdivide soft sites in dynamic analyses is proposed. This study also shows that the proposed techniques to evaluate liquefaction potential predict the performance of the San Francisco sites reasonably well. Comparison between pre-earthquake and post-earthquake cone data shows that the relative density of loose to medium dense deposits increased during the Loma Prieta event. The change in relative density of a medium to dense deposit was negligible. An analysis of post-earthquake field data from the San Francisco sites validates the previously suggested boundary curve (Reyna and Chameau, 1991a) between cyclic stress ratio and the horizontal stress index, K$\sb{\rm d}$, to evaluate liquefaction potential. Similar boundary curves for different earthquakes magnitudes are developed. An alternative index using the dilatometer to evaluate liquefaction potential which takes advantage of driven and pushed dilatometer tests is also presented. The usual limitation in the amount of data available for making a direct correlation between soil liquefaction characteristics and CPT values has been largely overcome in this study. The proposed curves are substantiated using data from Japan, Romania, U.S.A., China and more recently Imperial Valley and San Francisco, California. The statistical discriminant analysis approach is used to confirm the validity of the proposed boundary curves using the CPT to evaluate liquefaction potential