PERFORMANCE EVALUATION OF GAS LIFT VALVES USING NUMERICAL SIMULATION

Abstract

This work aimed to assess the performance of two numerical simulation methods to replicate flow capacity test (FCT) results and predict the flow coefficient (Cv) and critical pressure ratio (Rcp) of gas lift valves (GLVs). FCT's use a modified GLV with an adjustable stem positioning system to obtain pressure as a function of flow rate for different stem positions to calculate Cv and Rcp. Therefore, this study used both a one-dimensional (1D) mechanistic model and computational fluid dynamics (CFD) to predict the same variable from the FCT without tests using modified GLVs. This methodology demonstrates accurate results, which were compared to benchmarking information from tests and an extensive GLV database. This study developed a 1D model and mechanistic flow equation representing the restrictions of GLV internal flow domains. This model considers the equivalent open area at the orifice and check valve and uses experimental data from the dynamic test of the GLV performance test to calibrate the model before running for multiple stem positions. Twelve different GLVs were modeled and simulated using this approach. Similarly, a CFD model of the full GLV including all internal features was built to assess the potential for using CFD to predict the correlation coefficients for flow coefficients (Cv) and critical pressure ration Rcp. The results were also compared against experimental data from the Valve Performance Clearinghouse (VPC) database. The 1D and CFD simulation results show strong consistency and accuracy when compared to experimental results and data from the VPC database. CFD results show greater accuracy than the less complex 1D model results. Most flow coefficients (Cv) and critical pressure ration Rcp datapoints from CFD simulations for different stem positions are within a 15% error range. While the 1D model shows higher variability than the CFD methodology, 9 out of the 12 valve configurations evaluated show the majority of data points for Cv and Rcp within a 15% error range of the experimental results. VPC correlation is the best available correlation and predicts flow rates with +/-20% accuracy over the pressure range.