Permeability Estimation Using HFU Method Enhanced by Bootstrap Forest AI-Approach for FZI Prediction in Mishrif Reservoir, Southern Iraq

Exploitation of the artificial intelligence (AI) with hydraulic flow unit technique (HFU) to enhance permeability prediction is an important step in reservoir characterization. HFU classifies the reservoir into distinct petrophysical properties, which are then used in permeability estimation. They are a function of flow zone indicator (FZI), which is determined based on core permeability and porosity. Core measurements are often costly, time-consuming and only cover limited intervals; therefore, it is necessary to use other techniques to predict FZI in un-sampled intervals and wells. Thus, an advanced artificial intelligence method called Bootstrap Forest is employed to predict the FZI value from well logs. In this research, 889 core samples of six wells from the Buzurgan Oilfield's Mishrif Formation are used to identify the number of HFUs based on three techniques; probability plot, RQI versus Φz analysis and histogram analysis. Their utilization reveals that 10 HFUs represent this formation with a correlation coefficient ranging between 0.950 to 0.996. Core measurement and logs are used to construct the Bootstrap Forest model in cored wells, which yielded a high coefficient of determination (R2) of 0.999 and a low root mean squared error (RMSE) of 0.01 indicating that it is an effective model for predicting FZI based just on log data for uncored wells. Finally, the effective porosity is used with the predictive FZI to estimate the permeability resulting in a perfect matching between estimated and core permeability.