A Stochastic Optimization Model Based on Adaptive Feedback Correction Process and Surrogate Model Uncertainty for DNAPL-contaminated Groundwater Remediation Design
Xue Jiang,Wenxi Lu,Jin Na,Zeyu Hou,Yanxin Wang,Baoming Chi
DOI: https://doi.org/10.1007/s00477-018-1559-4
IF: 3.821
2018-01-01
Stochastic Environmental Research and Risk Assessment
Abstract:A stochastic optimization model based on an adaptive feedback correction process and surrogate model uncertainty was proposed and applied for remediation strategy design at a dense non-aqueous phase liquids (DNAPL)-contaminated groundwater site. One hundred initial training samples were obtained using the Latin hypercube sampling method. A surrogate model of a multiphase flow simulation model was constructed based on these samples employing the self-adaptive particle swarm optimization kriging (SAPSOKRG) method. An optimization model was built, using the SAPSOKRG surrogate model as a constraint. Then, an adaptive feedback correction process was designed and applied to iteratively update the training samples, surrogate model, and optimization model. Results showed that the training samples, the surrogate model, and the optimization model were effectively ameliorated. However, the surrogate model is an approximation of the simulation model, and some degree of uncertainty exists even though the surrogate model was ameliorated. Therefore, residuals between the surrogate model and the simulation model were calculated, and an uncertainty analysis was conducted. Based on the uncertainty analysis results, a stochastic optimization model was constructed and solved to obtain optimal remediation strategies at different confidence levels (60, 70, 80, 90, 95%) and under different remediation objectives (average DNAPL removal rate ≥ 70, ≥ 75, ≥ 80, ≥ 85, ≥ 90%). The optimization results demonstrated that the higher the confidence level and remediation objective, the more expensive was remediation. Therefore, decision makers can weigh remediation costs, confidence levels, and remediation objectives to make an informed choice. This also allows decision makers to determine the reliability of a selected strategy and provides a new tool for DNAPL-contaminated groundwater remediation design.