Using sequential self-calibration and genetic algorithin methods to optimally design tracer test for estimation of conductivity distribution

Limiting the quantity of field test data needed to obtain an accurate estimate of a hydraulic conductivity field is a continuing challenge for hydrogeologists. A gradient-based inverse method, the sequential self-calibration (SSC) method, conditioned using tracer test data is presented as a means for estimation of hydraulic conductivity fields. To improve the calculation efficiency of sensitivity coefficients, a fast streamline-based approach was applied to compute the derivative of concentration with respect to the changes of hydraulic conductivity. The performance of SSC method was tested using a synthetic aquifer with a sandwich-like geologic structure, in which hypothetical tracer tests were conducted. The SSC method was also used to assess the impact of sampling well locations and the number of sampling wells on the estimation accuracy of the hydraulic conductivity field. A genetic algorithm, combined with the SSC method, was applied to estimate the optimal tracer test design plan. We found that the estimation accuracy of the hydraulic conductivity field increased with an increase in the number of the sampling wells, but the rate of increase in the estimation accuracy decreased as the number of sampling wells increased. The estimation accuracy, was also significantly influenced by the locations of the sampling wells. The optimal sampling well locations were dependent on the geologic structure.