Hydraulic Parameter Inversion And Solute Transport Simulation In Discountinuous Heterogeneous Porous Media

The industrialization and urbanization process has led to galloping groundwater pollution caused by human activities.In order to ensure public health and national water safety,the remediation of groundwater contaminated sites was explicitly included in the Action Plan for Prevention and Control of Water Pollution and Action Plan for Prevention and Control of Soil Pollution in 2015.However,the effectiness of the remediation has been compromised in many groundwater or soil remediation cases.Comparing to hydraulic head,solute transport is considerable more sensitive to the spatial variability of hydraulic parameters.Hydraulic tomography(HT)can accurately predict the hydraulic heterogeneity in porous media.Sim SLE(Simultaneous successive linear estimator)is adopted in this technique.The natural logarithm of hydraulic parameters are assumed follow Gaussian distribution and conceptualized into random fields described by mean,variance and correlation scale.Spatial statistics are used as prior information,and then state variables(such as hydraulic head)are used to predict the heterogeneity of aquifer media.Discontinuous heterogeneous porous media are rules rather than exceptions in real world practice,which do not necessarily follow the Gaussian distribution assumption.Moreover,site-specific geological information(such as the spatial location and the mean hydraulic parameters of sedimentary facies,faults and paleochannels)is usually unknown.Geologic maps and geophysical surveys can indeed provide the general idea about the spatial distribution of large-scale geological features,but such large-scale geologic information often contains considerable uncertainty and can only be used as qualitative data.These shortcomings lead to modest performance of the inversion model using rough spatial statistics as initial values.Some other algorithms are also used to improve this problem,but there are still certain limitations.To address the issues mentioned above,four research sections are carried out:(1)aquafer tests,well logs and hydrogeologic survey in a coastal area in Tianjin.Analyzing the impact of accurate prior geologic information in solving the inversion problem by using cross-correlation analysis algorithm and the concept of redundant/nonredundant information.(2)Development of an iterative method based on hierarchical clustering is developed to predict large-scale heterogeneity boundary,initial mean,update covariance matrix and integrate it into the next iteration.Subsequently,the inversion performance of this method and original Sim SLE was validated against a set of numerical solute transport experiments in synthetic aquifers.The prediction of solute transport is significantly improved.(3)Monte Carlo simulation of 100 discontinuous heterogeneous stochastic fields and a set of sandbox experiments were carried out to validate the improvements of the proposed scheme.Such inversion improvements on solute transport prediction could provide scientific foundation for P&T configuration optimization.