An evaluation of boundary conditions for one‐dimensional solute transport: 2. Column experiments

Abstract

As the result of a theoretical comparison of analytical models for one‐dimensional solute transport (Novakowski, this issue), it has been found that to reconcile the substantial differences observed between the models under conditions of large dispersion, a physical modeling study of the processes of solute transport in the vicinity of boundaries must be undertaken. The physical modeling is conducted using columns ranging in diameter from 76 to 352 mm and length from 300 to 400 mm. Geological materials of either large or small coefficient of dispersion are employed as packing for the columns. Reservoirs of finite volume are located at the inlet and outlet boundaries of each column. Using a conservative fluorescent tracer, experiments are conducted to investigate the use of macroscopic continuity in concentration at the boundaries, and the use of the flux‐averaged transformation for this boundary value problem. Concentration of the tracer was determined noninvasively from both the inlet and outlet reservoirs and, for some experiments, resident concentration was determined from within the interior of the column by excavation. Results of the experiments conducted using different volumes of the outlet reservoir show that the analytical model for flux concentration accounting for macroscopic continuity in concentration at the boundaries only poorly simulates the physical mixing process in the outlet reservoir. In addition, the results of the experiments conducted in which resident concentrations were determined from the interior of the column show that the concept of macroscopic continuity is not supported by physical evidence at either the inlet or outlet boundary. Thus, the analytical model in which concentration at the boundaries is macroscopically discontinuous best simulates the solute transport processes for this boundary value problem. Unfortunately, the solutions for resident and flux concentration with these boundary conditions are identical and further distinction between these models cannot be undertaken. Analytical inversion of the Laplace domain solution is also presented.