Porosity and Permeability Prediction Based on Computational Intelligences as Artificial Neural Networks (ANNs) and Adaptive Neuro-Fuzzy Inference Systems (ANFIS) in Southern Carbonate Reservoir of Iran

Reservoir characterization is a hard-to-do task because of the extremely heterogeneous nature of petroleum bearing formations. Studying different sources of data obtained from underground formations shows that abrupt changes in reservoir rock properties are very commonplace, especially in carbonate formations. Overcoming heterogeneity of reservoir is seemed to be impossible at least with current practices. In addition, obtaining reliable data from every foot for all wells is not feasible because of its high cost as well as being very time-consuming. Porosity and permeability distribution are essential reservoir rock properties to be determined in order to build a reservoir model with acceptable accuracy. Analyzing well test and core data are two reliable sources of porosity and permeability determination. Due to the additional time and cost, coring from all points of formation is not feasible. Therefore another way of defining porosity and permeability distribution should be sought in which a more available source of data is used. The fact that geophysical well logs are routinely run for every wells makes researcher to find a way to predict porosity and permeability for uncored wells by correlating well logs and core data. Computational intelligence are intelligent approach that be used to estimate permeability and porosity in cored zone and related the results to uncored zone. Fuzzy logic, genetic algorithm, artificial neural network (ANN), and adaptive neuro-fuzzy inference system (ANFIS) are some kind of computational intelligence. The authors used ANN and ANFIS as two models to estimate of permeability and porosity in Southern Carbonate reservoir of Iran and results compared with each other to have good decision about methods of computational intelligence.