An Analysis Method of Tracer Tests by Using Genetic Algorithms and Its Application

Abstract

In the case of a tracer test analysis using a nonlinear least squares method, we should notice that the final estimates are influenced by the initial values of the reservoir properties for the estimation. In this study, we apply the genetic algorithm and that of hybridized with a nonlinear least squares method as a local search to tracer test analysis. Genetic algorithms are stochastic optimization and search algorithms. Compared with a nonlinear least squares method, genetic algorithms have some remarkable features as follows: the initial guess for the values of the parameters is not required; they can search simultaneously many estimates in an identical search space. This paper shows that the hybrid genetic algorithm is effective to estimate the reservoir properties on the tracer test analysis resulting from the numerical simulations and the analysis of the field data. In general, the increase in the number of estimates at the same time can improve the convergence speed. It is more effective to use the local search algorithm for the rapid convergence and for the accuracy of the estimates.