Obtaining an optimal gas injection rate for an oil production system via neural networks

Using a model-based optimization, a neural network model is developed to calculate the optimal values of gas injection rate and oil rate of a gas lift production system. Two cases are analyzed: a) A single well production system and b) A production system composed by two gas lifted wells. For both cases minimizing the objective function the proposed strategy shows the ability of the neural networks to approximate the behavior of an oil production system and to solve optimization problems when a mathematical model is not available.