Prediction of the residual strength of liquefied soils

Soil liquefaction has led to extensive damage to buildings, bridge, and lifelines in past earthquakes. While the effects of liquefaction can be manifested in many ways, the most devastating are in the form of flow slides. Flow slides can move at high speeds for distances of hundreds of meters, and destroy everything in their path. They have been observed in both natural and man-made soils, and can be triggered by many types of events of which earthquake shaking is only one.; Evaluation of the potential for flow slide development requires evaluation of the residual strength of liquefied soil. This problem has been one of the most difficult in geotechnical earthquake engineering because the residual strength is difficult to be measured accurately. Empirical procedures based on back-calculation of residual strength from flow slide failures have been developed and used widely. These approaches suffer from (1) only a limited number of flow slide case histories available, (2) some cases have little quantitative data, and (3) the process involves assumptions. These factors introduce considerable unquantified uncertainty into the predicted residual strengths.; A new approach to the back-calculation of residual strength from flow slide case histories has been developed. The approach objectively and consistently identify and quantify the most significant uncertainties in the parameters through the back-calculations by the Monte Carlo simulation. Each flow slide case history is represented by a joint probability density function of penetration resistance and residual strength, rather than by a single deterministic data point as in previous analyses. By applying this methodology to some well-documented case histories, and applying its basic principles to other case histories, a database of residual strength data was developed. Using basic principles of soil mechanics, the approach that residual strength is related to penetration resistance was found to be consistent with field data.; A predictive model for residual strength was developed. It allows estimation of the conditional probability distribution of residual strength given corrected standard penetration test resistance. It allows engineers to determine the probability of a given residual strength value being exceeded for a given average standard penetration resistance.