Abstract
This paper shows the importance of Artificial Intelligence (AI) techniques as a practical engineering tool for predicting and estimating the gas flow rate through chokes. Studying the single gas flow through wellhead chokes is vital to the oil industry, not only to ensure the accurate estimation of gas flow rate but also to keep equipments protected from damage due to high gas flow rate. It also has the potential to avoid sand problems. Many studies have investigated the predictability of gas flow through chokes. In this paper, we reviewed, evaluated and compared the predictive performance of the available choke correlations in literature with five AI techniques.
162 data points were used to develop five AI models for predicting the gas flow rate. The data were fed to the five AI techniques Artificial Neural Network (ANN), Fuzzy Logic (FL), Support Vector Machine (SVM), Functional Network (FN) and Decision Tree (DT) (ANN, FL, SVM, FN and DT) and the results were optimized for each technique. The new models were found to perform better than the correlation and give the lowest error, with a mean absolute percentage error of 0.83%. Because of these reduced errors, the proposed AI-based models can improve gas flow rate prediction through chokes.
The results of this paper will provide a better alternative to predictive modeling of petroleum reservoir properties. It will also open windows of opportunity for researchers and engineers to explore advanced machine learning techniques such as hybrids and ensembles for continued improvement of petroleum exploration and production.