Experimental Investigation and Modeling of the Hydraulic Conductivity of Saturated Bentonite–Claystone Mixture

Abstract

Mixtures composed of MX80 bentonite and crushed Callovo-Oxfordian (COx) claystone have been considered as one of the promising sealing/backfilling materials in French deep geological disposal for radioactive waste. One of the key issues about these mixtures is hydraulic conductivity. In this study, the swelling pressure and hydraulic conductivity of bentonite–claystone mixtures with different bentonite fractions and dry densities were determined by carrying out infiltration tests under constant volume conditions. The results indicated that as the bentonite fraction and dry density increased, the overall swelling pressure increased and the hydraulic conductivity decreased. Based on the assumption that the stress between bentonite and claystone grains was equal to the overall swelling pressure, the void ratios and volumetric fractions of bentonite and claystone in the mixture were determined. According to the relationships between hydraulic conductivity and void ratio for pure bentonite and claystone, the overall hydraulic conductivities of the bentonite–claystone mixtures were then calculated using five analytical models (parallel, series, Maxwell–Eucken 1, Maxwell–Eucken 2, and effective medium theory models) corresponding to different potential composite structures, allowing appreciation of the composite structures of bentonite–claystone mixtures with various bentonite fractions. For specimens with ≥70% bentonite, the bentonite formed a matrix with the claystone dispersed within it, for specimens with <40% bentonite, the matrix was formed by swollen claystone with the bentonite dispersed into the matrix, and for specimens with 40%–70% bentonite, the bentonite and claystone were randomly distributed, without being continuous or dispersed. Based on the defined composite structure of the bentonite–claystone mixture, a new method is proposed for estimating the overall hydraulic conductivity. A comparison between estimation and measurement showed the relevance of the proposed method.