Abstract
The scaling approach presented in this paper has proven effective in modeling the performance of West Texas CO2 projects. This methodology was developed, and has been utilized successfully, to model the performance of the Wasson Denver Unit, which is the world's largest miscible CO2 project. Methods are described for predicting injection and production rates for tertiary enhanced oil recovery (EOR) projects using dimensionless recovery curves. The primary use of the program has been for miscible CO2 flooding; however, the same methods could also be used for other tertiary processes such as alkaline surfactant flooding.
This program, which has been titled "DUSCALE," is based upon dimensionless recovery curves that can be either simulation-based or derived from analog actual project performance. One key attribute of the program is the ability to model the changes in reservoir processing rate that are common in these projects. Such variation impacts both the performance of the underlying waterflood and also the performance of the tertiary EOR project. Underlying waterflood production is forecasted using a throughput-based regression fit of historical pattern-level performance. Predictions are generated for all injected and produced fluids (oil, hydrocarbon gas, NGL yield, water, and CO2).
The program has also proven valuable as a surveillance tool for comparing actual versus predicted performance of an EOR project at the pattern level. Such analysis has identified project areas with good performance that are worthy of larger injected CO2 slugs, and also patterns within projects that are not meeting tertiary recovery targets. The ability to adjust expectations for processing rate variations is critical in such evaluations, as accelerated waterflood production (due to improved reservoir throughput during a CO2 project) can be incorrectly attributed to the CO2 displacement process.
This paper presents the methodology employed in the program and demonstrates, with examples from the Permian Basin, how the program can be used to generate forecasts for field-scale EOR projects. Examples highlighting surveillance applications are also be included.
