Implementation of waterflooding, which is one of the most common techniques for enhancing oil recovery, entails a number of issues related to difference in mobility of injected water and heavy hydrocarbon phase. Polymer flooding allows to make the properties of injected fluid more similar to the ones of oil by using a polymer solution. The study suggests artificial neural network as an effective forecasting tool for oil recovery factor in polymer flooding at heavy oil fields. To train the network, a backpropagation algorithm was combined with seven input parameters to predict one output parameter through a hidden layer of 12 neurons. Trained model was validated using independent experimental data. The study has shown that suggested technique provides high accuracy and time-efficiency in forecasting waterflooding parameters with limited experimental input data.
Keywords: NEURAL NETWORKS, FORECAST, POLYMER FLOODING, HEAVY OIL, OIL RECOVERY FACTOR.