Injection pressure is a key control parameter, which determines the amount of solvent and steam mixture consumed and its optimum level dominates the economic efficiency of solvent-aided thermal recovery in oil sands reservoirs. The authors determine the optimal strategy of determining operating pressures to achieve the maximum economic value; the scheme is based on an artificial neural network (ANN). The multilayer perceptron using backpropagation minimizes the objective value including bitumen production, steam injection, solvent retention, commodity price, and manufacturing cost. The numerical approach integrating with ANN results in accurate prediction similar to the time-consuming reservoir simulation. The application to the Athabasca oil sands reservoir confirms the enhanced results compared with the constant injection scenario and proposes the optimal schedule of injection pressures that repeats the increment and the decrement until reaching the same pressure level between the injection and the production well. This model maintains the consistency of the peak production rate on account of latent heat despite reducing the cycle amplitude. The developed model could be applicable to economically designed injection scenarios without any modification of production facilities.
- artificial neural network
- injection pressures
- oil sands
- optimal strategy
- solvent-aided thermal recovery