Transient Multi-Phase Wellbore Flow Simulation to Accurately Predict End of Well Life and Enhance Recovery

Abstract

Abstract The field is a Pliocene, shallow-marine, gas reservoir offshore North Trinidad with one producer well and is believed to be supported by a moderate aquifer. In February 2014, formation water production was observed and with near term shut-downs in plan, there was a natural risk of losing the well during a shut-down due to liquid loading. Field remaining reserves were also at risk due to its single well development nature. To optimally manage the well and mitigate these risks, a better understanding of liquid loading and a more accurate end of well life prediction were identified as requisites. In our experience, reservoir simulation models are unable to accurately predict well life post water breakthrough as they typically use steady state vertical lift performance curves whereas loading, shut-in and start-up operations are inherently transient conditions. Steady state lift curves are derived using nodal analysis and flow regime specific correlations and in case of liquid loading, are valid only for specific liquid film and droplet conditions – which cannot be guaranteed in practice. Thus, a transient multi-phase well flow simulator was used to model with higher accuracy lift performance at higher water cuts and reservoir pressures. This enabled a more robust calculation of liquid hold ups, pressure drops and allowed better prediction of liquid loading. A tuned well model was developed by matching historical data points and a sensitivity study performed using a range of reservoir pressures, chokes, gas rates, water influx volumes and WGRs. Production conditions potentially initiating liquid loading were predicted and were used to identify any near term shut-in periods during which the well could load. Transient modelling results indicated that the well could die during a shut-in when reservoir pressure dropped below 235 pressure units. Production trend curves were used to forecast these conditions accurately in the near future. A combination of these techniques estimated the end of well life to be any time after Dec 2014, and in reality in Feb 2015 post a shut-down, the well was unable to unload and flow. This closely validated the predictability and robustness of the analysis. By building a better understanding of the high risk periods, more informed well management decisions were made during potential shut-in or choke back periods. This resulted in an extension of the well life and an incremental recovery of c. 2 – 3%. For gas wells producing water, this is suggested as a robust workflow to closely predict end of well life conditions and timing and hence, support better management of the reservoir and operating conditions during those periods.