Chemically induced flow in contaminated unsaturated soil

Laboratory experiments and numerical modeling were conducted to assess the effects of glycerol concentration gradients on water flow through an unsaturated loamy sand (Yellow bush sand, Australia) at 46% glycerol mass fraction and 12.5% volumetric water content. When the contaminated and uncontaminated experimental soil sections were placed into contact, two opposite mechanisms were found to drive the flow: (i) the matric potential gradient affected by concentration-dependent surface tension, viscosity, and density repelled contaminated water; and (ii) the osmotic potential gradient had an opposite effect. Experiments demonstrated that the tested sand did not exhibit the complete semipermeable characteristic necessary to induce pure osmotic flow. Rather, physicochemical effects caused by dissolved glycerol were more relevant for water flow than those induced by the osmotic potential. We confirmed our experimental finding by numerical modeling that explicitly accounted for these effects compared with our experiments and earlier experiments conducted with 7% butanol–water mixtures as benchmarks. Our results were finally confirmed by a thermodynamic interpretation of matric and osmotic potentials. Experiments suggest that a rapid transient condition occurred on the first day and that equilibrium was recovered only after >60 d. The results support the hypothesis that chemically induced water flow in the vadose zone is contributed mainly by matric and only secondarily by osmotic effects, which have displaced 3 to 10% water during the initial transient condition. Predictive tools of contaminant hydrology in unsaturated soil may have to account for the osmotic effects in particular applications that involve soil with high osmotic efficiency.