A Water Saturation Prediction Using Artificial Neural Networks and an Investigation on Cementation Factors and Saturation Exponent Variations in an Iranian Oil Well

Abstract In this article, two approaches are presented to calculate three crucial petrophysical parameters (water saturation [Sw], cementation factor [m], and saturation exponent [n]). In the first approach, an artificial neural network (ANN) was implemented to calculate water saturation in Sarvak limestone formation in an Iranian oilfield (Azadegan). Core data from three wells (AZN01, AZN03, and AZN04) have been used to train and test the artificial neural network. Conventional wireline logs such as formation resistivity log, density log, sonic log, gamma ray, and porosity recorded from cores to be used as input to the ANN and water saturation measured in laboratory as target. Among four generated networks, network 3 showed the best correlation coefficient of 0.969 between ANN-predicted Sw and target Sw. Using this network, water saturation of the Sarvak formation in well AZN01 was predicted that showed good conformity with results of a dual water saturation model considering m = 2.0 and n = 2.0 in the upper parts of the formation, whereas in the lower parts of the formation the dual water saturation model calculated higher values, proving the variations of both m and n parameters along the formation. In the second approach, using Archie's equation's components (ANN-predicted water saturation, resistivity log, porosity log, and aRw), a system of two equations and two unknowns was developed to calculate the cementation factor (m) and saturation exponent (n) simultaneously for every two consecutive wireline tools measurements all over the formation. Then the variations of m and n, for the entire depth of the reservoir pay zone, versus depth were plotted. These plots prove that m and n should not be considered constant in petrophysical interpretations that conventionally assume constant in the water saturation models.

[1]  Water Saturation Computation From Laboratory, 3-D Regression , 2002 .

[2]  Djebbar Tiab,et al.  Petrophysics: Theory and Practice of Measuring Reservoir Rock and Fluid Transport Properties , 1996 .

[3]  E. Dean,et al.  A Convenient Method for the Determination of Water in Petroleum and Other Organic Emulsions. , 1920 .

[4]  Terrilyn M. Olson,et al.  Porosity and Permeability Prediction in Low-Permeability Gas Reservoirs From Well Logs Using Neural Networks , 1998 .

[5]  Hans B. Helle,et al.  Fluid saturation from well logs using committee neural networks , 2002, Petroleum Geoscience.

[6]  Paul F. Worthington,et al.  The Evolution Of Shaly-sand Concepts In Reservoir Evaluation , 1985 .

[7]  G. E. Archie.  The electrical resistivity log as an aid in determining some reservoir characteristics , 1942 .

[8]  M. R. Rezaee,et al.  Petrophysical Evaluation of Kaolinite-bearing Sandstones: I: Water saturation (Sw), an example of the Tirrawarra Sandstone reservoir, Cooper Basin, Australia , 1996 .

[9]  A. K. Rigler,et al.  Accelerating the convergence of the back-propagation method , 1988, Biological Cybernetics.

[10]  J. K. Ali,et al.  Neural Networks: A New Tool for the Petroleum Industry? , 1994 .

[11]  Robert Desbrandes,et al.  In Situ Rock Wettability Determination With Wireline Formation Tester Data , 1988 .

[12]  J. L. Dumanoir,et al.  Theoretical and Experimental Bases for the Dual-Water Model for Interpretation of Shaly Sands , 1984 .

[13]  D. J. O'Meara,et al.  Centrifuge Measurements of Capillary Pressure: Part 1-Outflow Boundary Condition , 1992 .

[14]  M. Al-Awad,et al.  Water Saturation Computation from Laboratory 3d Regression , 2002 .

[15]  W. D. Lyle,et al.  Improved Data-Analysis Method Determines Archie Parameters From Core Data (includes associated paper 24964 ) , 1992 .

[16]  Z. Bassiouni,et al.  Log-derived cation exchange capacity of shaly sands : application to hydrocarbon detection and drilling optimization , 2002 .