Conditioning clayey soils with a dispersant agent for Deep Soil Mixing application: laboratory experiments and artificial neural network interpretation

Plasticity of clays makes Deep Soil Mixing (DSM) problematic due to the tendency of the material to congest the rotating blades, reduce mixing efficiency and remain clustered in lumps which affect the mechanical behavior of the cemented soil. The paper investigates this problem systematically with a comprehensive experimental campaign that shows the efficacy of a clay dispersant in scattering soil particles and making soil more workable. The performed laboratory investigation combines consistency, micro-structural, vane, rheological and uniaxial compression tests on two reference soils, a kaolin and a bentonite, treated with various proportions of cement, water and chemical additive to quantify the effects on workability, homogeneity and strength of the material. The variety of investigated conditions enables to understand the principles of chemical modification and infer a quantitative dependency of viscosity and uniaxial compression strength on the material composition. Observation is interpreted for the sake of generality with artificial neural networks, merging the role of the different components into a novel definition of the soil consistency index, accounting for the presence of the additive. The inferred empirical relations, expressed with charts, are proposed to optimize soil conditioning for DSM.