Evolutionary neural networks for estimating viscosity and gas/oil ratio curves

In oil and gas industry, prior prediction of certain properties is needed ahead facility design. Some of these properties, e.g. viscosity and gas/oil ratio (GOR), are described as curves with their values varying over a specific range of reservoir pressures. However, the usual prediction approach could result into curves that are not consistent, exhibiting scattered behavior as compared to the real curves. In the proposed work, a new approach is implemented using a hybrid artificial neural network with differential evolution (DE+ANN) optimization technique. Inputs into the developed models include hydrocarbon and non-hydrocarbon crude oil compositions and other strongly correlating reservoir parameters. Graphical plots and statistical error measures, including root mean square error (RMSE) and average absolute percent relative error (AAPRE) have been used to evaluate the performance of the models. For both viscosity and gas/oil ratio curves, the prediction by DE+ANN has outperformed significantly the standalone ANN. The predicted curves are consistent with the shapes of the actual curves and closely replicate the field data.