Analytical Method for Forecasting ROZ Production in a Commingled MOC and ROZ CO2 Flood

Abstract Residual Oil Zone (ROZ) refers to a formation whose discovery saturation equals the rock's residual oil saturation. The ROZ makes up an excellent target for CO2 flooding since this oil is immovable by primary and secondary production processes. The San Andres ROZ has been recognized as an extensive residual oil saturated fairway in the Permian created during the Leonardian uplift, which caused spillage from the natural traps (Melzer, 2006). It has been developed through CO2 flooding in several fields across the Permian, including the Denver Unit in the Wasson San Andres formation, where it was developed years after the Main Oil Column (MOC). Both zones are producing from commingled producers and are flooded by commingled or dedicated injectors. This commingled configuration presents a challenge in discerning the production coming from each zone. In this paper, we will present an analytical approach to distinguish between MOC and ROZ production without the need for numerical simulation or costly well interventions such as production logging or zonal isolation. A sector of the Denver Unit's CO2 flood was used as an example in this paper. Dimensionless analysis, which entails normalizing production and injection to the target pore volume, was used along with the Pulser process (Liu, Sahni, and Hsu, 2014; informal communication with Deepak Gupta, 2019) to history-match MOC production and then extrapolate it using zonal injection obtained from injection profile logs. This calculated MOC production is then subtracted from the total production to calculate ROZ production, with its dimensionless response function fitted with Pulser for forecasting. Additionally, a fully compositional numerical simulation of the same area was history-matched and used to validate the approach mentioned above. The results of the analytical approach showed excellent agreement with the numerical simulation results and with historical performance through multiple years. A few challenges presented themselves, such as pattern-to-pattern interference, the quality of injection profile logs, and pattern reconfigurations, which we will discuss below along with limitations and assumptions that must be considered when using this approach. The methodology presented in this paper presents a simple method to allocate and forecast MOC and ROZ performance individually despite changes in injection throughput, based on injection distribution without the need for complex simulation or costly well configuration. This approach could also be applied to any commingled flood that meets the criteria outlined in this paper.