DEEPWATER PRODUCTION FORECASTING USING COMPUTATIONAL INTELLIGENCE TECHNIQUES

Abstract

Precise and consistent production forecasting is indeed an important step for the management and planning of petroleum reservoirs. This research work is aimed at developing a new neural approach to forecast cumulative oil production.  The methodology employed is using a neural network with its weight optimized with genetic algorithm (NN-GA). NN-GA overcomes the limitation of the conventional neural networks of settling in a local minima by optimizing the weights and biases to be used to initialize a Nonlinear Autoregressive Neural Network with Exogenous Input (NARX). Thus, NN-GA possesses a great potential in forecasting petroleum reservoir productions without sufficient training data. A pre-processing procedure was employed in order to reduce measurement noise in the production data from the oilfield by normalizing the raw data using z-core normalization technique and optimal network topology selection using correlation functions (CCF).  Simulation studies were carried out on a deepwater reservoir located off the coast of the Niger-Delta in Nigeria, to prove the efficacy of the NN-GA in forecasting cumulative oil production of the field with data available. The results of these simulation studies indicate that the NN-GA model has a good forecasting capability with high accuracy to predict cumulative oil production and this was compared to the widely used Decline Curve Analysis technique.  It was concluded that the NN-GA procedure has a very good potential to be applied to petroleum forecasting and the possibility of including more reservoir production parameters to the model to help for better accuracy and precision in the forecasting process.