Prediction of steam distillation efficiency during steam injection process by Artificial Neural Network

Steam distillation mechanism is one of the important and effective mechanisms during steam injection process in fractured heavy oil. Due to its important effect in oil recovery, several attempts have been made to simulate this process experimentally and theoretically. Because of limitations in implementing experiments, various models have been studied to predict the distillation effect with minimum entry parameters. So, in this study, a Multi-Layer Perceptron (MLP) neural network is used as an effective method to simulate the distillate recovery, so that the parameters such as API, viscosity, characterization factor and steam distillation factor are the input parameters and distillate yield is the output of the model. After gathering our data from the literature, 77 data of 128 input data were used for training, 33 data for testing, and 18 data for cross validation. Then, the results of one-layer and two-layer networks with various neurons were compared to the experimental data and other models.