Effect of Densification on the Random Field Model Parameters of Liquefiable Soil and their Use in Estimating Spatially-Distributed Liquefaction-Induced Settlement

Abstract

Densification-type ground improvement has gained wide acceptance as an effective technique to improve the static and cyclic strength and seismic performance of liquefiable soils. However, there has been little attempt to quantify the degree of change in certain measures of soil spatial variability following densification. This study uses the random field model (RFM) framework to quantify changes in the trend, inherent variability, and the autocorrelation of soil following various densification-based ground improvement methods using case history data. Ground improvement technologies investigated include driven displacement piles, vibro-replacement (i.e., stone columns), and deep dynamic compaction. Vertical RFM parameters are shown to change significantly following densification, with reductions in the coefficient of inherent variability, and increases in the scale of fluctuation, the latter of which appears related to the initial relative density. One case history is used to illustrate the utility of the RFM framework to link spatial variability of the post-shaking settlements to the results of a geostatistical model of the subsurface, including the three-dimensional distribution of CPT measurements and fines content. It is shown that the spatial distribution of differential surface settlements is correlated to those spatially-varying subsurface characteristics that are known to lead to larger liquefaction-induced settlement.