Estimation of Laminated Reservoir Properties Using Artificial Neural Network Integration

Due to current economic constraints on companies' capital expenses, the lack of reservoir characterization information has become a challenging issue. As a result, reservoir engineers must make the most of the available data and estimate the reservoir characteristics that are not available. Porosity is typically used to estimate permeability through classical correlations during core analysis. However, due to the effects of complex lithology and pore geometry, it becomes unreliable to predict permeability solely from porosity and using classical relationships. The objective of this study is to enhance the Tortonian reservoir description in Gamma oil field by testing the integration of Flow Zone Indicator (FZI), Artificial Neural Network (ANN), and Convergent Interpolation (CI) techniques. The study utilized data from one exploratory well and four appraisal wells to model the non-linear relationship between the Tortonian reservoir properties, calculate the effective porosity, estimate the permeability of uncured wells, and create a permeability map for the Tortonian oil reservoir. The results showed that there are three rock types in the Tortonian reservoirs, and effective porosity and permeability logs were successfully estimated. The permeability map created showed a direct relationship with the porosity map, validating the methodology. The reliability of the porosity/permeability relationship increased to 90% after using the integrated techniques presented in this study. By integrating FZI, ANN, and CI techniques, this study has successfully modeled the non-linear relationship between the porosity and permeability of the Tortonian reservoir, enabling an economical improvement in the complex reservoir description with minimum capital budget and available data.