Flow prediction using well log data and intelligent techniques

Abstract

ABSTRACTSpatial distribution of petrophysical properties within the reservoirs is important to provide a reliable reservoir model. The most complete reservoir description is generally provided through the identification of flow units. Then, the constructed model can be used for reservoir simulation of different production scenarios within the reservoir. However, before performing modeling and simulation jobs, accurate prediction of flow units is an important task and to reach a reliable petrophysical description first of all one should be able to predict flow units with high precision. Achieving this goal will be discussed in this article.This paper suggests intelligent techniques using artificial neural network (ANN), fuzzy logic (FL), and neuro-fuzzy (NF) to determine flow units from well log data. For this purpose, available routine core and wireline log data from five wells in Ilam Formation at the Ahwaz super giant oil field are used to find the best intelligent formulation between core-derived flow units and well log data. Validation of the predictive capability of the models was evaluated in one separate cored well (blind test well). The results show that FL-derived flow units had better accuracy with respect to other techniques. This comparison showed that FL can be used as the most reliable intelligent technique for flow unit prediction from well log data in uncored wells.