A black‐box automated approach to calibrate numerical simulations and optimize cover design: Application to a flow control layer constructed on an experimental waste rock pile

Abstract

Mining operations often produce large volumes of waste rock to access economically valuable mineralized zones. Waste rock is usually stored in surface piles, the construction and reclamation of which represent a challenge for the industry. A flow control layer (FCL) made of crushed waste rock or sand and constructed on top of each waste rock bench could contribute to control water infiltration, thus improving waste rock pile stability and limiting contamination. An experimental waste rock pile was built and instrumented at the Tio mine (Rio Tinto Fer et Titane, Canada) to evaluate the performance of an FCL in field conditions. Large infiltration tests and rainfall monitoring were carried out, and measured outflow and water contents were used to calibrate numerical simulations. However, data were noisy and sometimes incomplete, and the models were difficult to calibrate. A new automated calibration approach was therefore proposed. An algorithm was developed to automate the numerical simulation calibration, using a black-box method that involves solving an optimization problem on a function without an analytic form. The approach was applied on measurements obtained from large-scale infiltration tests and validated using 2 yr of field monitoring data. Finally, the automated approach was adapted to optimize the design of the FCL, and an optimal design (material properties and layer thickness) was recommended based on local climate conditions. The proposed automated method could contribute to reduce the bias induced by manual calibration and allows for rapid multivariable calibration and optimization for a broad spectrum of mine waste cover system applications. INTRODUCTIONMining operations produce large volumes of waste rock to access economically valuable mineralized zones. Waste rock Abbreviations: AMD, acid mine drainage; CND, contaminated neutral drainage; FCL, flow control layer; HGS, HydroGeoSphere; MADS, mesh adaptative direct search.This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.