Application of Simulated Annealing and Adaptive Simulated Annealing in Search for Efficient Optimal Solutions of a Groundwater Contamination related Problem

Abstract

Characterization of groundwater contamination sources is a complex inverse problem. This inverse problem becomes complicated, due to the nonlinear nature of the groundwater flow and transport processes and the associated natural uncertainties. The mathematical challenges arise due to the nonunique characteristics of this problem resulting from the nonunique response of the aquifer system to a set of stresses and the possibility of instead locating only local optimal solutions. The linked simulation-optimization model is an efficient approach to identifying groundwater contamination source characteristics. Efficiency and accuracy of the search for optimum solutions of a linked simulation-optimization depend on the utilized optimization algorithm. This limited study focuses on the application and efficiency of simulated annealing (SA) as the optimization algorithm for solving the source characterization problem. The advantages in using adaptive simulated algorithm (ASA) as an alternative are then evaluated. The possibility of identifying a local optimal solution rather than a global optimal solution when using SA implies failure to solve the source characterization inverse problem. The cost of such inaccurate characterization may be enormous when a remediation strategy is based on the model inferences. ASA is shown to provide a reliable and acceptable alternative for solving this challenging aquifer contamination problem.