Secondary Porosity Evaluation on Permeability Forecasting Using Fuzzy Logic and Neural Networks in Heterogenic Carbonate

Direct measurements of permeability through core analysis and well test are expensive and may not be possible for all boreholes, whereas all wells are almost petrophysicaly logged which permeability from open-hole logs can be estimated. One of the major challenges in carbonate formations is to analyze and quantify non-effective pores like secondary porosity. If this can be achieved, permeability evaluation will be more accurate. For this purpose, three wells of the Iran offshore Balal oil field were chosen; the focus of interest was Surmeh (Arab) formation. The permeability estimation model was built in well BL-1P through core and well logs data using two methods fuzzy logic and back propagation neural network (BP-ANN). Moreover, secondary porosity index (SPI) was calculated and imported in the model. This model was applied to predict permeability in control wells 3I and 3W-2. The outputs were cross checked against core data. The results of this study showed that fuzzy logic prediction was a bit better than ANN and in accordance with elimination the effect of discontinuous porosity after adding SPI, the prediction became better.