Classification of ordovician tight reservoir facies in Algeria by using Neuro-Fuzzy algorithm

Abstract

The Tight reservoirs in Algeria are generally characterized by their complex nature and their degree of heterogeneity. Wherein, the quantitative evaluation of such type of reservoirs necessitate the determination of facies in order to estimate the in-situ hydrocarbons and their nature. However, the classical methods of determining facies are essentially based on core data and carrots, which are not always technically available. Artificial neural network (ANN) is one of the recent developed methods being used to provide facies classification with a minimum available core data and by using well logs. Even though, the ANN results are acceptable, it determines only the dominant facies at each depth point off logs, no information can be provided for the secondary facies. For that reason, the main objective of this study is to develop a Neuro-fuzzy algorithm that allows the determination of secondary facies in addition to dominant facies. Indeed, the algorithm has been trained by using core data at wells’ scale in the Ordovician reservoir located in an Algerian southern Petroleum field. Moreover, the Neuro-fuzzy classifier has been tested in near wells, for which, the obtained results has demonstrated the effectiveness of the proposed approach to improve tight reservoir characterization in the studied field. Hence, the designed algorithm is highly recommended for other petroleum systems in Middle East and North Africa region