Reducing the risk of hydrocarbon exploration and lithology characterization — using a neural network approach in the West Delta Deep Marine Concession Area, Offshore Nile Delta, Egypt

Abstract

There are many approaches and concepts for the exploration and development of hydrocarbon reservoirs. In this study, we develop both a lithology classification and a gas chimney prediction workflow. We classify the entire seismic volume into its most likely facies domains and transform the seismic cube into a gas chimney probability cube. Both these approaches utilize a neural network algorithm, which we find is an efficient method to integrate the multiple input attributes required to extract the target log, or to isolate target features from the seismic background data. A combination of a set of seismic attributes (obtained from multilinear regression) with the nonlinearity of the artificial neural network technique is used to develop effective workflows to classify and explore the reservoirs. Analysis of the results helps to quantify the chance of success of the exploration prospects and also, in conjunction with the petroleum system of the prospects, leads to a reduced risk of the hydrocarbon exploration and development wells.