Computationally Efficient Optimization of Groundwater Remediation

A computationally efficient algorithm for time-varying optimization of pump and treat groundwater remediation is presented. An enforced sparsity approach is applied to the derivative based optimization algorithm within a dynamic optimal control model. This approach exploits the sparsity structure of finite element simulation model derivative matrices in order to reduce the computational demand for each optimization iteration. Hence, the required computational demand for solving the optimization problem decreases. Two different sparsity approaches are applied and expected computational savings show the efficiency of the developed approach. The method presented is applied to a hypothetical contaminated groundwater aquifer. It is shown in our example that the computational demand can be significantly reduced.