Production optimization using gas lift incorporated with artificial neural network

Abstract

This work investigates the impact of artificial neural networks on gas lift optimization to improve the Crude oil production. Data were collected from two wells which was trained in neural network using MATLAB 7.9 neural network toolbox. The data was divided into three parts; training (60%), validation (20%) and testing (20%), and was done on 20 hidden neuron multi-layer feed-forward neural network. The result indicated that the data from the two wells were well trained for accurate optimal prediction as the R-value was close to 1 and the Mean Square Error (MSE) was very low. The average error comparing the actual and predicted data was minimal using few data from the general data. The behavior for both wells follow same pattern and can be concluded that for optimal production to be achieved, it will involve a reduced Well head parameter and Gas Compressor Suction pressure and a higher Gas compressor Injection pressure. It can be concluded that artificial intelligence have a very high impact on the optimal Crude oil production using gas lift.