Challenges in Well Testing Data from Multi-layered Reservoirs and Improving Nonlinear Regression: A Gas filed case

Abstract

Well test analysis of multi-layer reservoir comprises several parts. The first part concerns the estimation of parameters values and next considers finding an appropriate method to determine the unknown reservoir parameters. If the initial estimations are less accurate and weak, the final assessment may lead to incorrect results, which are totally different from the reality. Utilizing Automated Type-Curve Matching method by means of computers is effective and successful. In addition, it has priority over the ordinary analysis methods, and numerical calculations are performed with higher speed thereby causing fewer errors. In this study, first, a numerical simulation of reservoir was conducted by means of Eclipse 100 software and then its production and pressure results were introduced to Ecrin software of well test analysis. Different states were investigated in the analyzer to obtain the minimum error, and suggestions, including effect of regression point selection, initial guess, and effect of using each layer production were made to minimize the error. The final analysis results were compared with numerical simulation data to determine the performance of Ecrin software, and then a real well test was investigated in the gas field of South Pars by using this method. Furthermore, parameters of each layer were obtained so that applying the mentioned method caused the analyzer error to decrease to a favorable desired value.