Assessing a Response Surface-Based Optimization Approach for Soil Vapor Extraction System Design

Abstract

This paper presents a study on a response surface-based optimization approach for soil vapor extraction system (SVES) design. The SVES design involves extraction rate determination for a number of wells at fixed locations. Two optimization problems were investigated in this study, including: (1) minimizing total cost while achieving the required cleanup goal; and (2) maximizing percentage of contaminant mass removal under the constraint of total extraction rate. The response surface-based optimization approach integrates a multiphase flow and transport simulator, a genetic algorithm and a response surface method that employs regression analysis for constructing an approximation function of mass removal percentage in terms of extraction rate. A symmetric Latin hypercube design (SLHD) or factorial design plan is employed to obtain a moderate number of design points for different optimization problems. Besides, an iterative optimization procedure is implemented in the approach to enhance the accuracy of the approximated response function by gradually adding more design points to the design space. Results show that the response surface-based optimization approach is superior to a simulation/optimization approach in terms of optimal solution and computation time.