A New Approach to Quantify the Wellhead Performance for Gas Condensate Reservoirs Using Artificial Intelligent Techniques

Abstract

Abstract Selecting the optimum choke size that can deliver the optimum gas flow rate is extremely complex in gas condensate wells because of the high unpredictability of two-phase flow behavior and the changes in gas-liquid ratio with pressure and temperature. There are many analytical and empirical correlations in the literature, that describe the two-phase flow through wellhead chokes in critical conditions. However, the nature of the flow in most gas condensate wells is sub-critical, and using these correlations causes severe errors in the estimation of flow parameters. Also, the models available for sub-critical flow are not providing satisfactory accuracy for prediction and are often difficult to use. The objective of this work is to develop reliable models for predicting choke performance. Artificial intelligent (AI) based models were developed to accurately design the optimum wellhead choke size, based on the required flow rate, pressure difference, and gas to liquid ratio. The newly developed models are accurate and easy to be used, comparing to the existing predictive models. In this work, artificial neural networks (ANN) and adaptive neuro-fuzzy inference systems (ANFIS) were applied to estimate the gas flow rate from gas condensate wells. Among the developed models, ANFIS with a subtractive clustering approach performed the best with a percentage error of 2.4% and coefficient of determination (R2) of 0.981. The newly developed models can improve production management by proposing the optimum flow parameters that help in regulating the flow rate of two phases from gas condensate wells. Also, they can help in stabilizing the flowing pressure downstream the choke and prevent downhole reservoir formation damage, by providing the required back pressure to avoid excessive drawdown. Overall, the presented models provide an easy