Applying Optimized ANN Models to Estimate Dew Point Pressure of Gas Condensates

Abstract

It is economically and technically essential to promptly and accurately estimate the dew point pressure (DPP) of gas condensate to, for example, characterize fluids, evaluate the performance of reservoirs, plan and develop reservoirs for gas condensates, and design/optimize a production system. Indeed, it is difficult to experimentally explore the DPP. Furthermore, experimental tests are time-consuming and complicated. Therefore, it is required to develop an accurate, reliable DPP estimation framework. This paper introduces artificial neural network (ANN) models coupled with optimization algorithms, including a genetic algorithm (GA) and particle swarm optimization (PSO), for DPP estimation. A total of 721 data points were employed to train and test the algorithm. In addition, the outlier data were identified and excluded. The root-mean-squared error (RMSE) and the coefficient of determination (R2) were calculated to be 230.42 and 0.982 for the PSO-ANN model and 0.0022 and 0.997 for the GA-ANN model, respectively. The model estimates were found to be in good agreement with the experimental dataset. Therefore, it can be said that the proposed method is efficient and effective.