A comparative analysis of meshless based simulation optimization models with metaheuristic algorithms for groundwater...

A robust Simulation-Optimization (SO) framework is proposed for the cost-effective design of groundwater remediation schemes in contaminated aquifers. The simulation involves the solution of coupled groundwater flow and transport phenomena using the Meshless Local Petrov Galerkin (MLPG) method, selected due to its high stability, truly meshless nature and independence from complex meshing process. The MLPG simulator is integrated with four metaheuristic optimization techniques: the emerging nature-inspired Whale Optimization Algorithm (WOA), Aquila Optimization (AO), Golden Jackal Optimization (GJO) and the widely used Differential Evolution (DE), forming MLPG-WOA, MLPG-AO, MLPG-GJO and MLPG-DE models. These SO models advance existing approaches by minimizing remediation costs while simultaneously optimizing extraction rates and remediation well locations in Pump and Treat (PAT) remediation schemes. Additionally, the proposed models have several advantages including minimal sensitivity to initial estimates, simplified fine tuning, rapid convergence and reliable designs. The performance of the metaheuristic algorithms is investigated through two case studies of hypothetical and field-type aquifers. All the models effectively design remediation strategies that reduce contamination within permissible limits within the remediation period. In the hypothetical case, a single well extracting 1368 m³/day for 1000 days is identified by MLPG-WOA as the least-cost solution of INR 4,115,238. For the field-type aquifer, nine-well strategy with the lowest cost of INR 145,072,081 is identified by MLPG-DE. Optimal well placement zones are identified in both the case studies. The proposed SO models are thus found to be efficient in providing reliable remediation designs and can serve as alternative to the existing PAT remediation models.