Determination of aquifer parameters and parameter structures by using simulation-optimization models

In this study, an integrated simulation-optimization approach is proposed to simultaneously determine aquifer hydrogeological parameters and their parameter structures. In the simulation part of the proposed approach, the numerical simulation of groundwater flow processes is performed using the MODFLOW model. Within this model, the zone structures of hydrogeological parameters are defined by partitioning the flow domain into a finite number of zones using the general form of the Voronoi Diagram (VD). The developed MODFLOW and VD-based simulation model is subsequently integrated into an optimization model utilizing the Ant Lion Optimization (ALO) approach. ALO is a population-based optimization technique that numerically simulates the hunting strategies of antlions in nature. While ALO has been widely applied to solve various problems in the literature, it has been reported that the method often requires a high number of iterations for complex or large-scale problems and tends to converge to local optimum solutions. To overcome these limitations, two significant improvements were made to ALO in this study, resulting in the development of the Shuffled Ant Lion Optimization (SHALO) approach. The applicability of the proposed SHALO approach was tested on several unconstrained and constrained optimization test functions in the literature. After validating the optimization model, the performance of the proposed simulation-optimization approach was evaluated using a synthetic aquifer model given in the literature. This evaluation involved solving the problem under identical conditions using the original ALO, self-adaptive ALO (saALO), and SHALO approaches. Additionally, the results of the proposed approach were compared with those obtained using Harmony Search (HS) and hybrid Genetic Algorithm (GA)-based solution approaches. The results demonstrate that the proposed simulation-optimization approach outperforms ALO, saALO, HS, and hybrid GA approaches in terms of various evaluation metrics.