Assessment of a parallel evolutionary optimization approach for efficient management of coastal aquifers

Abstract

This study presents a parallel evolutionary optimization approach to determine optimal management strategies of large-scale coastal groundwater problems. The population loops of evolutionary algorithms (EA) are parallelized using shared memory parallelism to address the high computational demands of such applications. This methodology is applied to solve the management problems in an aquifer system in Kish Island, Iran using a three-dimensional density-dependent groundwater numerical model. EAs of continuous ant colony optimization (CACO), particle swarm optimization, and genetic algorithm are utilized to solve the optimization problems. By implementing the parallelization strategy, a speedup ratio of up to 3.53 on an 8-core processor is achieved in comparison with serial model. Based on solution quality and computational time criteria, the CACO robustness is observed in comparison to other EAs. Moreover, the optimization solution of the case study for a scenario of sea-level-rise indicates that a reduction of 20% in groundwater extraction rate is mainly due to the land-surface inundation caused by sea-level rise.