A Compositional Simulation Evaluation of the Brassey Artex B Pool, British Columbia, Canada

Abstract

Abstract A compositional reservoir simulation evaluation1 of the Brassey Artex B Pool, located in British Columbia, Canada, was undertaken to predict hydrocarbon recoveries under a variety of development schemes. The subject reservoir is a thin aeolian sand containing a volatile, highly undersaturated oil. A miscible flood was initiated in 1989 immediately following delineation drilling. The numerical model was constructed using the most up-to-date geological, geophysical, and petrophysical data. Surface facilities were incorporated into the model, through the use of individual well test separators for production tests performed prior to implementation of a miscible flood, and by a four-stage separator after implementation of the miscible flood. This model was then calibrated by history matching three years of volumetric and compositional data. Tracer survey results were also used in the model calibration phase. Reservoir fluid compositions were represented in the model with a nine pseudo-component Peng-Robinson equation of state. The equation of state was calibrated to laboratory constant composition expansion, differential liberation, swelling test, and slim tube data to provide representative PVT properties of the reservoir fluid and injected solvent. The history matched model was then used to forecast the production performance for a variety of production and injection schemes, including infill drilling and sensitivities to injected solvent composition. The reservoir fluid characterization phase of the study verified that the recovery process was one of first-contact miscibility at the 4000 psia operating pressure. Analysis of the pressure, saturation, and fluid composition distributions demonstrated that an effective first-contact miscible displacement was occurring in the reservoir. While the majority of the pool will be swept with the existing well configuration, two infill drilling locations have been identified. Phase behaviour analysis indicated that the current injection gas is richer than required to achieve first contact miscibility; therefore, liquids can be extracted and sold from the injection gas stream. History matching of the compositional model suggested an initial oil in place volume approximately one-third less than that derived from previous pseudo-miscible black oil studies. However, the initial analytical material balance estimates derived from production tests conducted prior to the miscible flood showed an oil in place volume midway between the two model estimates. This comparison highlighted the difficulties of characterizing the volumetric behaviour of near-critical fluids using either compositional or black oil techniques.