A simplified, yet rigorous forecasting technique was developed to predict the recovery performance of a hydrocarbon miscible flood in the Judy Creek Beaverhill Lake (BHL) ‘A’ pool in Alberta. This technique incorporates the physics of the hydrocarbon miscible displacement process as well as the complex geological reservoir description across the entire pool. The advantages of the technique over conventional three-dimensional reservoir simulation include cost effectiveness, speed, and flexibility of parametric evaluations.
A real reservoir simulation model was used to describe the areal sweep; oil recovery performance as a function of injection and solvent bank size; and solvent, chase gas and post-miscible water breakthrough times. The effect of gravity override was determined by a cross-sectional reservoir model. A geological model provided the porosity and permeability distribution for every 0.6 meter interval in each injection pattern as well as the continuous pore volume; i.e., the volume that is available for miscible displacement. The results of these models were combined in a simplified algorithm (the Forecast Model) to predict the tertiary recovery. The Forecast Model adds the tertiary recovery forecast to the waterflood forecast, which was developed from decline analysis, to generate a total oil forecast under miscible flooding. The miscible flood forecast derived from this model is physically reasonable and predicts recovery levels consistent with observed performance of large-scale miscible projects. This technique is being extended for use in other reservoirs which are candidates for miscible flooding.
References and illustrations at end of paper.
In designing a miscible flood project, it is necessary to predict the production and recovery performance of the pool. It is desirable to be able to do this quickly and cost effectively, to optimize on the various critical miscible flood parameters such as solvent composition and bank size, injection rate, water-alternating-gas ratio (WAG) and pattern staging as well as incorporate complex reservoir geology.
The tool most often used to predict miscible flood performance is the three-dimensional compositional reservoir simulator. However, this tool is costly to run and is plagued with numerical dispersion that makes the results suspect. Consequently, only small portions of a reservoir are usually evaluated, and as a result, the pool wide geology is not rigorously included.
This paper describes a forecasting technique that was developed to evaluate quickly and cost effectively (less than 4 CPU minutes per run) miscible flood product ion and recovery performance. This technique incorporates the key parameters of miscible displacement and applies it to the complex and varying reservoir geology throughout the vertical and areal extent of the entire reservoir.
Esso has employed this forecasting technique, termed the Forecast Model, to evaluate the miscible flood performance for both the Judy Creek BHL ‘A’ and ‘B’ pools1, 2, 3. The ‘A’ pool results will be used for illustrative purposes in this paper to describe this model.
Judy Creek BHL ‘A’ pool, discovered in 1959, is located 200 km northwest of Edmonton, Alberta, Canada (Figure 1). Solution gas drive is the primary depletion mechanism in this reservoir, and as a result, a peripheral waterflood was initiated in 1962 to maintain pressure.