Use of global optimisation technique in groundwater pumping strategy for plume removal

The remediation of groundwater contamination by pumping and injection is generally a long-term strategy and requires a huge amount of money. Aquifer cleanup time is a highly non-linear function of pumping rates and well locations, and there exist multiple local minima. The objective of a cleanup strategy often involves minimising or constraining the cleanup rime or cleanup cost. Linear programming and non-linear optimisation techniques used in practice cannot guarantee a global solution. In this study, a Global Optimisation (GO) technique is used to minimise both cleanup time and cleanup cost, taking pumping rates and/or well locations as decision variables. Two contaminated aquifers are considered - one real and one hypothetical. The models, MODFLOW and MODPATH, developed by the United States Geological Survey (USGS) are used for groundwater flow and particle tracking, and the code, GLOBE, developed by IHE, Delft is used as a GO tool. The technique essentially involves the coupling of MODFLOW and MODPATH with GLOBE so as to run them in a loop. Results are compared using four algorithms, namely Genetic Algorithm (GA), Adaptive Cluster Covering (ACCO), Controlled Random Search (CRS4) and Multis. Results, in all cases, show that both cleanup time and cleanup cost can be reduced significantly using this technique.