Hybrid-Optimization Approach for Estimating Parameters of a Virus Transport Process in Aquifer

Abstract

Simulation of a virus transport process in a groundwater aquifer is necessary for predicting the movement of viruses in an aquifer. It is also necessary in implementing remedial measures to inactivate viruses present in groundwater. For simulating the virus transport process in a groundwater aquifer, the parameters responsible for movement of viruses in an aquifer have to be estimated. In the case of saturated porous media, the transport parameters are linear distribution coefficient, hydrodynamic dispersion coefficient, and inactivation coefficient for aqueous and sorbed viruses. These parameters can be estimated using an inverse optimization procedure that minimizes an error function represented by the difference between the experimentally observed and simulated virus concentration. Gradient-based classical optimization methods can be applied to minimize the error function for estimating the parameters of the transport process. It is reported that the error function is nonlinear and nonconvex in nature when all four parameters are considered in the model. As such, classical optimization methods are not effective when all four parameters are considered. Global optimization methods, such as genetic algorithms, may be efficient to solve the inverse problem. However, in many cases, these algorithms yield a near global optimal solution. As such, a hybrid optimization model using genetic algorithms and the simplex method is proposed for estimating the parameters. Example problems have been solved to evaluate the efficiency of the proposed method. The evaluation shows that the proposed procedure is efficient and can be applied in the field for estimating parameters of the virus transport process in a groundwater aquifer.