Dispersivity variations of solute transport in heterogeneous sediments: numerical and experimental study

Heterogeneity significantly effects the accuracy of flow and contaminant transport prediction in subsurface formations. The spatial correlation structure of hydraulic conductivity (K) is a crucial factor to characterize the heterogeneous architecture. In presented study, the relationship between the spatial correlation structure of K and plume dispersion is analyzed through the integration of experimental and numerical simulation approaches. A detailed description on the sedimentary facies types in a column experiment is obtained to ensure the accuracy of the heterogeneous characterization. The spatial correlation structure of K is analyzed with the components of ln(K) covariance and facies transition probability structures. Lagrangian-based models are developed to estimate solute dispersion in nonreactive tracer injection experiments. The results show that the model can predict plume spreading accurately when the spatial correlation structure is well defined. The dispersivities calculated by the Lagrangian-based model are slightly higher than those obtained from the solute transport experiments. Further, the upscaled dispersivity derived from the transition probability is dominated by the cross-transition probability structure, while the contribution of the auto-transition terms is quite small. The numerical modeling results confirm that the upscaled dispersivity can reproduce the solute breakthrough in the heterogeneous sediment well. The scale dependence of dispersion is strengthened when the flow direction is perpendicular to the bedding plane where the conductivity dramatically changes along the flow path in a layered bedding sediment.

Automated summary

The accuracy of flow and contaminant transport prediction in subsurface formations is significantly affected by heterogeneity. The study found that the model can accurately predict plume spreading when the spatial correlation structure is well defined. Moreover, the upscaled dispersivity is mainly influenced by the cross-transition probability structure.