Predicting Carbonate Reservoir Rock Types and Permeability with High Success Ratio Using Data Mining Approach, a Case from the UAE
M. Amin1, A. Al Khoori1, M. Adib1, M. Benlacheheb1, and P. Bhatt2
This work presents the results of a study to predict the Reservoir Rock Types (RRTs) and the permeability for none cored wells. The established reservoir rock types scheme; 8 RRTs were based on depositional facies sequences, diagenetic overprints and Petrophysical properties, including pore throat size distribution, porosity and permeability for 47 cored wells.
Rock typing prediction was based on the available log data for 150 wells without core data. The workflows started by selecting key wells for building the prediction model. Different data mining approaches were applied for rock typing prediction; Unsupervised-cluster analysis, Supervised–Neural Network, and Fuzzy logic study. However, a “Combined” including Fuzzy Logic and Neural Network model was generated and successfully able to predict all the 8 Reservoir Rock types with different matching degree between the predicted and defined rock types. The model was applied on two blind-test wells to predict the rock types. The blind test results showed the range of success ratio from 65-75% which gave the confidence to apply the model for distributing Reservoir Rock types on all un-cored wells.
AAPG Search and Discovery Article #90188 ©GEO-2014, 11th Middle East Geosciences Conference and Exhibition, 10-12 March 2014, Manama, Bahrain