A training trajectory random walk model for upscaling colloid transport under favorable and unfavorable conditions

Abstract

In this study, we present a general random walk model for upscaling colloid transport and retention in two-dimensional porous media. The model is based on direct sampling from spatial and temporal jump distributions of single-interceptors, colloids that first enter the near-surface zone within 200 nm of a collector surface, derived from mechanistic pore-assembly trajectory simulations. This framework facilitates the connection and transition between the interception space and physical space, thereby enabling the upscaling of spatial and temporal distributions of single interceptors to full retention profiles and total residence time distributions. Additionally, the proposed random walk model has been tested across a range of advection and diffusion scenarios, demonstrating both accuracy and efficiency in predicting retention profiles and total residence time distributions. Overall, with the appropriate inputs, this model provides a reliable and efficient framework for predicting colloid transport and retention in porous media without the need for extensive computational sources.