Experimental and Numerical Investigation of Liquid-Assisted Gas-Lift Unloading

Abstract

The case for a unique form of gas-lift unloading, termed “Liquid-Assisted Gas-Lift (LAGL),” is presented. This work demonstrates that the injection of a gas-liquid mixture allows transport of gas to a deep injection point utilizing injection pressure considerably lower than single-phase gas injection. The LAGL is demonstrated in a 2,880 ft deep test well. The test well is kicked-off using an injection pressure that would normally be lower than the pressure for single-point single-phase gas injection at this depth. Experimental results indicate that the LAGL can lower the injection pressure by up to 75%. This work breaks the LAGL system in three sub-systems: two-phase downward flow in annulus, two-phase flow through orifice Gas-Lift Valves (GLVs), and upward two-phase flow in pipes. The last sub-system is well described in the literature and will not be investigated in this work. However, there is a lack of studies on two-phase downward flow in annulus and through GLVs. Therefore, these two topics are investigated in this work. An experimental and numerical study on two-phase flow through orifice GLVs is presented. The experimental results are compared to numerical models published in the literature for two-phase flow through restrictions. It was observed that some models can successfully characterize two-phase flow thorough gas-lift valves with errors lower than 10%. Experimental characterization is performed for downward two-phase flow in the annulus and the result is compared with downward flow in pipes. The comparison shows differences between downward two-phase flow in annulus and pipes for liquid holdup and flow regimes. The experimental results show that the liquid holdup is consistently higher for two-phase downward flow in annulus than in pipes for the annular flow regime. After the experimental validation of LAGL unloading and the characterization of two-phase flow through GLVs and annulus, a simulation model is built using a commercial transient flow simulator. The model is initially validated with experimental data from the field-scale test well. Out of 15 validation cases, only two cases showed average error higher than 15%. After the model validation, the simulation model is used to simulate the complete unloading of the well.