Integrated Res Characterization Tool to Construct High Resolution Geological Model in MR FM of DF Field

This paper establishes the approach of finding a relationship between reservoir rock typing (RRT)-derived from core and well log data to generate full field continuous RRT models that can be used to predict RRT in undrilled locations and predicting blind wells In the present study, a total of 16 wells were analyzed uses a unique technology that integrates all data using artificial intelligence where a neural network is trained and tested using existing data. Multiple realizations are created, analyzed and validated through blind well testing.The RRT prediction was carried out using 'Ipsom' module in TECHLOG and the module is based on supervised neural network technology.This module using the core RRT as desired log and well logs as an input curves for the neural network training, RRT in the cored intervals were used as training set to obtain the neural network engine which will be used to predict the un-cored intervals and wells.Once the RRT were defined, the attempt was made to generate permeability and saturation height function for each of these RRTs. For permeability computation, FZI (Flow Zone Indicator) and Simple Geometric Regression Type Equations were tested. Having compared both methods of regression, it was observed that there was a minimal difference between both methods using FZI or using a simple regression type in permeability prediction. However, FZI gave a slightly better result when compared with the simple regression method within the range of data in the field.