Determining hydraulic characteristics of production wells using genetic algorithm

Proper well management requires the determination of characteristic hydraulic parameters of production wells such as well loss coefficient (C) and aquifer loss coefficient (B), which are conventionally determined by the graphical analysis ofstep-drawdown test data. However, in the present study, the efficacy of a non-conventional optimization technique called Genetic Algorithm (GA), which ensures near-optimal or optimal solutions, is assessed in determining well parameters from step-drawdown test data. Computer programs were developed to optimize the well parameters by GA technique for two cases: (i) optimization of B and C only, and (ii) optimization of B, C and p (exponent) as well as to evaluate the well condition. The reliability and robustness of the developed computer programs were tested using nine sets of published and unpublished step-drawdown data from varying hydrogeologic conditions. The well parameters obtained by the GA technique were compared with those obtained by the conventional graphical method in terms of root mean square error (RMSE) and visual inspection. It was revealed that the GA technique yielded more reliable well parameters with significantly low values of RMSE for almost all the datasets, especially in case of three-variable optimization. The optimal values of the parameters B, C and p for the nine datasets were found to range from 0.382 to 2.292 min m-2, 0.091 to 3.262, and 1.8 to 3.6, respectively. Because of a wide variation of p, the GA technique resulted in considerably different but dependable and robust well parameters as well as well specific capacity and well efficiency compared to the graphical method. The condition of three wells was found to be good, one well bad and that of the remaining five wells satisfactory. The performance evaluation of the developed GA code indicated that a proper selection of generation number and population size is essential to ensure efficient optimization. Furthermore,a sensitivity analysis of the obtained optimal parameters demonstrated that the GA technique resulted in a unique set of the parameters for all the nine datasets. It is concluded that the GA technique is an effective and reliable numerical tool for determining the characteristic hydraulic parameters of production wells.