Application of an Evolutionary Algorithm in Well Test Characterization of Naturally Fractured Vuggy Reservoirs

Abstract

TX 75083-3836, U.S.A., fax 01-972-952-9435. AbstractDuring the last years, several efforts have been done focused on developing the automatic computer-aided well test analysis.Traditionally, conventional gradient-based algorithms such as Levenberg-Marquardt have been used for parameters estimations. However, recently the use of new approaches based on global optimization methods has been introduced in the literature, mainly attempting to eliminate the influence of the initial guess on the overall regression, and overcoming the uniqueness problems.The aim of this work is to study the characterization of naturally fractured vuggy reservoirs via the automatic well test analysis by using a triple porosity model recently developed. Thus, an Evolutionary Algorithm (EA) was designed, taking into account the important fact that the triple porosity case may yield multiple solutions. Therefore, several aspects such as the parameters estimations with high precision and the necessity of developing analysis techniques computationally efficient are taken into account in the design of algorithms.According to the above, it is demonstrated that the algorithm developed in this work allows finding the multiple optimal solutions, with the desired precision. Advantages of the EA over the classical Levenberg-Marquardt method for this specific well test analysis problem are also discussed.