Artificial Neural Network Modelling, Simulation and Prediction of Gas Production

Abstract

Recently, exploration, drilling and production in oil and gas sources have become challenging owing to the complexity of the system. Industry developers established these sources several years ago, and their production histories now differ. To that end, the production and administration of oil and gas resources have necessitated the application of an advanced method of data processing, referred to as an artificial neural network, to solve these challenges. An Artificial Neural Network (ANN), a type of Artificial Intelligence (AI), is a network of interconnected nodes inspired by the structure of neurons in the brain. One primary goal of neural networks is to solve the complex problems of the oil and gas industry that cannot be easily addressed using traditional modelling tools. This method typically helps decision-makers improve choices and reduce non-productive time and costs. In this research, we developed an artificial neural network (ANN) model of a gas production system in MATLAB to model, simulate, and predict gas production. We trained the network on field gas-production data, using temperature and pressure as input parameters, and applied various training algorithms. We varied the number of hidden-layer neurons and the delays in the model, which produced 35 distinct outputs. The predicted outputs demonstrated excellent performance, achieving a correlation value of 0.98 and a mean squared error of less than 2%. Furthermore, the statistical error metrics showed excellent agreement between the ANN predictions and field report data. Thus, the results indicated that the ANN model could be applied to predict gas production accurately from a flow station.