Probabilistic modeling of offshore deep cement mixing improved ground

Abstract

Deep cement mixing (DCM) is an effective technique for ground improvement by injecting dry cement or cement slurry into the ground with continuous mixing and finally forming cemented soils with improved mechanical properties. A large-scale spatial distribution model is established in this study to investigate the characteristics of spatial variation in the unconfined compressive strength (UCS) of DCM clusters in a marine clay area. Variograms were used to determine the spatial variability and scale of fluctuations in the UCS data of DCM coring samples and the UCS of the natural ground from cone penetration tests. Referring to the layering characteristics of the study site, the auto-correlation structure was determined using random field modeling. Based on the generated spatial distribution and probabilistic assessment of UCS for the DCM construction site, this study compares the variations in spatial magnitude, uniformity, and variability of site-wide UCS before and after the construction of DCM clusters.