Modelling Approach to Reservoir Characterization of the Camal Field-Masilah Basin-Yemen

Abstract

The significant impact of reservoir characteristics including porosity, permeability, and saturation on the well productivity calls for concerted research effort towards developing promising models for high accuracy prediction. In this work, a new model for estimating rock properties of Camal field was developed using various approaches of Artificial Intelligence (AI) such as neural networks and hybrid neuro-fuzzy techniques.

In order to achieve the objective of this study, actual porosity, permeability and saturation data collected from laboratory analysis of the core samples obtained from eight wells in Camal field. In addition, the logs data (bulk density, gamma ray, induction) was also gathered for the same wells. Among different AI approaches were applied in this study to estimate reservoir characteristics. Hybrid method are also used such as adaptive neuro-fuzzy inference system (ANFIS) for estimation reservoir properties for Camal field. The proposed models were validated through the graphical and statistical error analysis.

An AI models to estimate different rock properties such as porosity, permeability and saturation were presented using actual field data in this work.  The obtained results showed that the hybrid neuro-fuzzy method had a great potential in predicting reservoir properties among the other AI approaches. Statistical analysis and comparative study showed that the performance of proposed hybrid neuro-fuzzy model is the best one with lower root mean square error (0.05) and higher accuracy of correlation coefficient (0.98) than those obtained with other models. It was observed that artificial intelligence modeling is reliable and accurate for prediction reservoir rock properties applied for this field.

The modeling approach presented in this work can be used as alternatives methods for determining petrophysical properties. These methods are accurate and less expensive techniques of reservoir description.