MODELLING STUDIES BY APPLICATION OF ARTIFICIAL NEURAL NETWORK USING MATLAB

Abstract

Four ANN models to estimate Bubble point pressure (Pb), Oil Formation Volume Factor (Bob), Bubble point solution Gas Oil Ratio (Rsob) and Stock Tank Vent GOR (RST) in the absence of Pressure, Volume and Temperature (PVT) analysis, were proposed as a function of readily available field data. The estimated Rsob and RST values from the proposed models can be used as a basic input variable in many PVT correlations in order to estimate other fluid properties such as the Pb and Bob. Another proposed ANN model has the ability to predict and interpolate average reservoir pressure accurately by employing oil, water and gas production rates and number of producers are used as four inputs for the proposed model without the wells having to be closed. Another ANN model proposed is to predict the performance of oil production within water injection reservoirs, which can be utilized to find the most economical scenario of water injection to maximize ultimate oil recovery. It has reasonable accuracy, requires little data and can forecast quickly. ANN approach to solving the identified pipeline damage problem gives satisfactory results as the error between the ANN output and the target is very tolerable. The results conclusively proved with error 0.0027 that it has the ability to accurately predict the pipeline damage probability by employing the model data obtained in this study.

Keywords

• Artificial neural network
• PVT models
• Reservoir pressure; Oil production; Pipeline damage
• Reservoir pressure
• Oil production
• Pipeline damage