Online Journal for E&P Geoscientists

Kumar, Arvind ^*1
(1) Petroleum Engineering, Indian School of Mines, Dhanbad, India, Dhanbad, India.

Due to the increasing number of complicated problems and time consuming analysis, the applications of advanced information technologies like fuzzy Logic, pattern recognition, intelligent networks and artificial neural network have gained momentum. Among all of them, Artificial Neural Network (ANN) proves to be having an edge on other computing applications for all types of data interpretations and analysis work related to petroleum exploration as well as exploitation. Nowadays, ANN has been widely accepted as the most powerful and efficient tool especially for reservoir characterization. Reservoir characterization mainly includes prediction of porosity, permeability, lithology, sand thickness, and well log data. This paper focuses on the application of ANN in the prediction of permeability and porosity of a reservoir for a given well log data and seismic data. This paper discusses many examples which highlight the efficiency of ANN in obtaining nonlinear systems and models for reservoir characterization problems. Well log data and seismic data are the parameters which have been used in the prediction of porosity and permeability using ANN in a carbonate reservoir.

This work used a method which combines different well log attributes to predict reservoir properties like Porosity and Permeability. This method uses multi-attribute regression technique to obtain an optimum ordering of the well log attributes and neural networks to increase the resolution of the final result. The regression curves explain the extent to which the predicted reservoir properties are more uniform and usable than that of given core properties in the field of reservoir characterization. The methodology was demonstrated using four different well logs for training: the gamma-ray log, the depth log, the core porosity log and the core permeability log