A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
AbstractTo limit the high cost of surfactant-enhanced aquifer remediation (SEAR) for clearing dense nonaqueous phase liquids (DNAPLs), the simulation-optimization technique is generally adopted for determining the optimal remediation strategy in advance. The simulation model requires an uncertainty analysis, and incorporating the results into the SEAR strategy optimization process is critical. However, previous studies have rarely involved corresponding problems. In the present study, an uncertainty analysis is performed by combining a Monte Carlo random simulation with the Sobol’ global sensitivity analysis to assess the contribution of different parameters to the remediation efficiency and distribution characteristics of the simulation model outputs. The surrogate model technique based on Kriging was used to reduce the high computational load of the sensitivity and uncertainty analyses. The results of the sensitivity analysis showed that the porosity is the most important parameter with the largest influence on the remediation efficiency at a weight of 70%, followed by the oleic phase dispersity at a weight of 27%; the influence from variations in other parameters can be neglected. The rebuilt surrogate model with fewer input variables performed significantly better than the one built before the sensitivity analysis for all performance evaluation indices. Uncertainty in the aquifer parameters resulted in clear variations in the simulation model outputs. Output fluctuations from the average were nearly 2.5%. The results of this study showed that the failure risk of a given remediation strategy can be obtained based on the distribution of model outputs.
AbstractTo limit the high cost of surfactant-enhanced aquifer remediation (SEAR) for clearing dense nonaqueous phase liquids (DNAPLs), the simulation-optimization technique is generally adopted for determining the optimal remediation strategy in advance. The simulation model requires an uncertainty analysis, and incorporating the results into the SEAR strategy optimization process is critical. However, previous studies have rarely involved corresponding problems. In the present study, an uncertainty analysis is performed by combining a Monte Carlo random simulation with the Sobol’ global sensitivity analysis to assess the contribution of different parameters to the remediation efficiency and distribution characteristics of the simulation model outputs. The surrogate model technique based on Kriging was used to reduce the high computational load of the sensitivity and uncertainty analyses. The results of the sensitivity analysis showed that the porosity is the most important parameter with the largest influence on the remediation efficiency at a weight of 70%, followed by the oleic phase dispersity at a weight of 27%; the influence from variations in other parameters can be neglected. The rebuilt surrogate model with fewer input variables performed significantly better than the one built before the sensitivity analysis for all performance evaluation indices. Uncertainty in the aquifer parameters resulted in clear variations in the simulation model outputs. Output fluctuations from the average were nearly 2.5%. The results of this study showed that the failure risk of a given remediation strategy can be obtained based on the distribution of model outputs.
Surrogate-Based Sensitivity Analysis and Uncertainty Analysis for DNAPL-Contaminated Aquifer Remediation
2016
Article (Journal)
English
| British Library Online Contents | 2016
| British Library Online Contents | 2018