Inter-Code Comparison of Variably Saturated Fluid Flow and Prediction of Percolation through a Tailings Impoundment in Southeast Nevada, USA.

Abstract

Hypothetical infiltration columns were simulated to evaluate the performance of 6 different modeling codes (HYDRUS 1D, HYDRUS 2D/3D, HELP, SV FLUX, UNSAT-H, and VS2D) at simulating near ground surface water balance processes occurring at mine process components (e.g. tailings pile or heap leach pile) under a simplistic scenario. Model results suggest that various codes for unsaturated flow arrive at similar solutions for identical models of well-defined problems. A literature review of several previous model comparison studies was conducted in order to provide insight into considerations for more complicated and realistic modeling scenarios. Unsaturated zone hydrologic modeling, field investigations, and laboratory analysis were conducted for a legacy tailings impoundment in Caselton, Lincoln County, Nevada to assess the potential for percolation of meteoric water through the tailings material. Field investigations consisted of field-saturated hydraulic conductivity testing, in-situ density measurement, and collection of tailings samples for laboratory analysis. Laboratory analysis included determination of field moisture content, bulk (dry) density, and water retention curve measurement by hanging column, pressure plate, and chilled mirror hygrometer methods. Hydrologic models of the tailings were performed using a variety of conceptual models, hydraulic property models, finite element discretizations, temporal boundary condition data resolutions, tailings hydraulic property descriptions, and lower boundary conditions. Numerical, mechanistic hydrologic flow models predict percolation rates ranging from 51 - 0 mm/yr. through the tailings, while previous efforts utilizing non-mechanistic water balance models predict a rate of 3.2 mm/yr., suggesting that significant variability exists in the model results depending on model approach and assumptions. Some simulations utilizing more complex and physically representative models resulted in higher percolation rates than the HELP simulation, which is counter to common assumptions of overprediction of drainage for simple water balance approaches.