Enhancing GOR Estimation in Gas-lift Oil Production: A Comparative Study of Support Vector Machines and Neural Networks

Abstract

The accurate prediction of GOR (Gas-Oil Ratio) in oil production operations is crucial for optimizing production and ensuring the longevity of oil wells. In this study, we compare the performance of SVR and Neural Networks for predicting GOR in oil wells operating with gas lift elevation production. Our results show that neural networks outperform SVRs in terms of accuracy and speed, making them a more suitable approach for these predictions. To validate our findings, we collected and analyzed data from real-world oil wells, using both SVR and neural networks to make predictions. We found that the neural network approach was able to accurately predict GOR within a reasonable timeframe, whereas the SVR approach was slower and produced less accurate results. Our study provides valuable insights for oil well operators and engineers seeking to optimize their GOR predictions.