本项目针对DNAPLs污染含水层修复方案优选分析过程中尚待解决的理论和技术问题开展研究。在建立多相流数值模拟模型的基础上，研究应用灵敏度分析法，识别确定出对SEAR修复过程和效果影响较大的变量，以减小分析过程中的计算负荷并提高计算精度。选用优化的拉丁超立方采样(OLHS)法在多相流模拟模型可控输入变量的可行域内采样，分析其对采样空间总体覆盖度和样点分布均匀性的改善效果。研究应用克里格(Kriging)法建立多相流数值模拟模型的替代模型，分析其可行性和优缺点。采用交叉验证法对不同的替代模型进行整体性检验，分析不同替代模型对多相流模拟模型本身的逼近程度，从中甄选出精度最优的替代模型。最后在综合分析各种因素的基础上，建立DNAPLs污染含水层修复方案决策优选的优化模型，从而获得关于修复方案的最优决策。丰富和拓展DNAPLs污染含水层修复方案优选分析过程中的理论与技术内涵。

Some challenging problems which have yet not been solved on the optimal remediation scenarios for the aquifer contaminated by DNAPLs will be studied in this project. On the basis of multiphase flow numerical simulation model, the variables that possess obvious influence to the SEAR effects will be identified aiming to improve analysis precision and reduce computation time. Optimal Latin hypercube sampling (OLHS) will be applied to sample specimens in the feasible domain of input variables of the multiphase flow simulation model. Then the coverage and uniformity degree of OLHS sample results will be appreciated. Kriging method is to be adopted to formulate the surrogate model of the multiphase flow simulation model. The merits and demerits of the surrogate model will be evaluated. Cross validation will be applied for the overall verification of different surrogate models evaluating the approximation degree of surrogate models towards the multiphase flow simulation model itself and the optimal surrogate model with the highest fidelity will be selected for the further use. At last, an optmization model will be formulated to obtain optimal decisions on remediation scenarios to the aquifer contaminated by DNAPLs. This research project will enrich and develop the theoretical and technical connotation for the analysis of remediation scenario optimization of the aquifer contaminated by DNAPLs.