Study On Identification Of Groundwater Pollution Sources And Optimization Method Of Pump-treat Scheme Based On Uncertainty Theory

Identification of groundwater contamination source and optimization of pump-treat scheme is an important research content of groundwater restoration.The uncertainty of model parameters and the uncertainty of surrogate model error were taken into consideration,the optimization design of groundwater pollution monitoring scheme and identification of contamination source conditions based on Bayes'theorem were carried out respectively,and the optimization of pump-treat scheme for polluted groundwater based on stochastic optimization model was studied.In view of the shortcomings of monitoring frequency of carbon tetrachloride concentration in Qiligou groundwater source and the lack of verification data of pump-treat scheme,this paper first constructed an ideal model for theoretical research and analysis,and then relevant results were applied to the identification of pollution sources and the optimization of pump-treat scheme in Qiligou groundwater source.The main research results are as follows:1.Research on the identification of pollution sources in the case of hypothetical cases.?1?Optimized design of single well monitoring scheme and the identification of pollution source:Under the condition of pre-set single well monitoring and 10 times of monitoring,the single-target optimized monitoring scheme with the minimum information entropy was 3#well,two monitoring intervals?35?t?28?37d.The multi-objective optimized monitoring scheme with the minimum information entropy and the shortest monitoring time was 3#well,?35?t?28?11d.Compared with the inversion results of the single-objective monitoring scheme,the average value of the relative error of the posterior mean of the five pollution source parameters by the multi-objective optimized monitoring scheme increased by 0.31%,but the monitoring time was reduced from333d to 99d,and by 70.27%.Therefore,the monitoring scheme based on multi-objective optimization was of more practical significance for the rapid identification of pollution sources.?2?Optimized design of multi-well monitoring program and identification of pollution sources:the numbers of monitoring wells was larger,but not the information entropy was smaller.Under each monitoring type,there was an optimal monitoring scheme with the smallest information entropy.And under the optimal scheme conditions of each monitoring type,the information entropy decreases by the number of wells increased.?3?Whether by the single well monitoring scheme or the multi-well monitoring scheme,the inversion results of parametersX_S and _SY were well.But the inversion results of ₁T,₂T and C_s were not ideal,and the posterior distribution ranges were consistent with the prior ranges,whih was mainly due to that ₁T,₂T,and C_s were not the insensitive parameters.In the parameter inversion process,the insensitive parameters could be taken as the intermediate value of the prior distribution ranges,and only the sensitivity parameters were inversed,which can not only reduce the dimension of the inversion parameters,but also reduce the workload,and improve the accuracy of the inversion.?4?DREAM algorithm can realize the synchronous inversion of pollution source and aquifer parameters,and progressive addition of wells with minimum information entropy significantly reduced the amount of calculation in the process of monitoring program optimization.2.Research on the optimum design of pump-treat scheme for the polluted groundwater.The groundwater level of the stochastic optimization scheme under the two modes of“pump-treat-discharge”and“pump-treat-recharge balance”were within the limited values,but the concentration of polluted groundwater in some wells could not reach the target of treatment,which was caused by the error of the surrogate model in the stochastic optimization model.However,due to the migration of pollution plumes and the pumping wells that could not cover the studied area?or the pumping wells cannot be evenly distributed in each of the split grids?,there was still a phenomenon of excessive concentration of pollutants in some areas of the studied area.Compared with the deterministic simulation optimization model,the stochastic optimization model had better results,and the stochastic optimization model considered the error of the surrogate model,which improved the reliability of the model solution.According to the treatment effect,the“pump-treat-discharge”mode was preferred as the pump-treat scheme,and the pumping wells were preferentially arranged on the central axis of the pollution plume and its both sides,showing a partial“+”shape distribution.The problem of"trailing"or"rebounding"of pollutants in pumping Wells can be solved by increasing pumping Wells or prolonging pumping time on the central axis of pollution plume.3.The identification of pollution source and pump-treat scheme optimization design in Qiligou groundwater source.A conceptual model of Qiligou groundwater source was constructed according to the geological and hydrogeological conditions.Then the conceptual model could be generalized into three-dimensional pressure-free fractured karst unsteady flow aqueous systems with heterogeneous anisotropy.And the water flow simulation model parameters?including conductivity,specific storage,specific yied,etc.?were carried out by using the PEST tool provided by GMS software,which was taken as the basis for the subsequent pollution source inversion after the parameter identification and verification.Then the solute transport simulation model was constructed.And the Kriging surrogate models of the simulation models were established.The DREAM algorithm was used to perform parameter inversion,and 26stable Markov chains were obtained,so 26 sets of solutions?parameter combinations?were obtained.Finally,root mean square error was used as the evaluation index to select the optimal solute transport model parameters.According to the situation of pollution plume in the studied area,combined with the distribution and the use of wells,19 wells were selected as alternative pumping wells.And the Kriging surrogate models of groundwater level and carbon tetrachloride concentration of each well were established.The results of stochastic optimization model under the pump-treat-discharge mode showed that the carbon tetrachloride concentration in the studied area could not be reduced to less than 2?g/L before 2030.And it is necessary to implement the treatment of the pollution by Sub-area sub-target.Under the conditions that the treatment target setting the carbon tetrachloride in the peripheral and the central areas of the pollution plume as 4?g/L and 7?g/L,the final pump-treat cost was 605140 yuan.