LAUSR

Abstract The swelling prediction of bentonite–sand mixtures due to wetting is very important in evaluating the long term performance of the engineered barrier in the high level radioactive waste disposal system and the hydraulic barriers in geoenvironmental constructions. Sun et al. (2015) proposed the swelling prediction model of bentonite-sand mixtures due to full hydration, and predicted the swelling of different types of bentonite-sand mixtures, which was verified, to be consistent with the swelling test results. However, the predicted swelling results of bentonite-sand mixtures with extremely high sand content obtained by the original swelling prediction model have a large deviation from the test results. The reason is that the original model is based on an assumption that all the pores and available water are only associated with bentonite/montmorillonite fraction. However, for mixtures with extremely high sand content, the sand skeleton exists and resists the external stress from the very beginning. At the same time, the bentonite, dispersing in the sand skeleton, in contact with the pore fluid, swells freely to fully saturated state, however, still fails to fill the sand skeleton voids completely. In this research, the original swelling prediction model is extended by introducing the concept of critical filling sand content and critical contact stress. When sand content is more than the critical filling sand content, the deformation of the mixture is mainly due to the sand skeleton deformation. After the stress increases greater than the critical contact stress, the saturated bentonite fills the sand skeleton voids completely, and the swelling can still be predicted by the original swelling prediction model. In the extended model, the swelling prediction can be divided into three zones according to the two limit values of critical sand content and critical filling sand content. In different zones, the distribution ratio of the vertical stress that is borne by montmorillonite and sand skeleton is different. The extended swelling prediction model is applicable to predict the swelling due to full hydration for mixtures within the full range of sand content, and is verified by the swelling tests on bentonite–sand mixtures with various sand contents.