Sigma is derived from the decay rate of pulse neutron capture, which has specific value for rock, vapor or gas, and fluid. The specific values for gas and matrix enable us to predict the steam saturation and sigma matrix. This paper focuses on estimating sigma matrix model in Area 10 and its implications on the steam saturation and remaining oil saturation calculation. The variable data used to generate the sigma matrix model are gamma ray, neutron, and density log from open hole data, and sigma formation log in cool condition from observation wells as ground truth. Previously, Area 10 sigma matrix model was generated using simple multiple-regression. This paper proposes a new method using facimage (multi-resolution graph based clustering) tool that gives more accurate prediction on sigma matrix, because it clusters the data into classes of electrofacies and utilizes the statistical and neural network to give the best log prediction. The accuracy of both sigma matrix methods is justified from the coefficient correlation cross plot between the predicted sigma matrix and the cased hole sigma matrix in a cool condition as ground truth. The result shows that coefficient correlation from multiple-regression method is 0.83, and facimage method is 0.95. The high correlation is achieved after selecting the proper input data for the modeling and separating sigma matrix model between Rindu and Pertama sand. Sigma matrix directly impacts on the steam saturation value, which than ultimately impacts on the oil saturation. The different sigma matrix values from both methods give significant difference on gas and oil saturation values, which will be important for steam injection optimization and development strategy.
Author : Henrikus Panjaitan, I Gede Putu P. Adnyana