On horizontal variability in lime-cement columns in deep mixing

Abstract

The variability with respect to results from hand-operated penetrometer tests on lime-cement column cross-sections is investigated by means of spatial correlation structures. Experimental directional semi-variograms are evaluated for a large number of column cross-sections from two field tests. Analyses of variance (ANOVA) are performed in order to investigate the influence of various factors in the installation process on the inherent variability. The paper also briefly discusses the concept of sufficient mixing: the link between the mixing process and the mechanical system. The results have been interpreted within the framework of variance reduction based on an averaging model for axially loaded columns. The results from the two field tests show that the radius is the dominating variable in the variability, and different types of correlation structure can be obtained in spite of the fact that they have been equally installed. The mechanical mixing work does not significantly influence the type of spatial correlation structure. Furthermore, a simple design consideration shows that the variance reduction has a major influence on the determination of the partial factor of safety.