A new workflow to improve the carbonate reservoir types discrimination combing the empirical model decomposition and energy entropy classification methods

Abstract

Classification and well-logging evaluation of carbonate reservoir rock is very difficult. On one side, there are many reservoir pore spaces developed in carbonate reservoirs, including large karst caves, dissolved pores, fractures, intergranular dissolved pores, intragranular dissolved pores, and micropores. On the other side, conventional well-logging response characteristics of the various pore systems can be similar, making it difficult to identify the type of pore systems. We have developed a new reservoir rock-type characterization workflow. First, outcrop observations, cores, well logs, and multiscale data were used to clarify the carbonate reservoir types in the Ordovician carbonates of the Tahe Oilfield. Three reservoir rock types were divided based on outcrop, core observation, and thin section analysis. Microscopic and macroscopic characteristics of various rock types and their corresponding well-log responses were evaluated. Second, conventional well-log data were decomposed into multiple band sets of intrinsic mode functions using empirical mode decomposition method. The energy entropy of each log curve was then investigated. Based on the decomposition results, the characteristics of each reservoir type were summarized. Finally, by using the Fisher discriminant, the rock types of the carbonate reservoirs could be identified reliably. Comparing with conventional rock type identification methods based on conventional well-log responses only, the new workflow proposed in this paper can effectively cluster data within each rock types and increase the accuracy of reservoir type-based hydrocarbon production prediction. The workflow was applied to 213 reservoir intervals from 146 wells in the Tahe Oilfield. The results can improve the accuracy of oil-production interval prediction using well logs over conventional methods.