History Match of a Mature Cyclic Steam Stimulation Process at Cold Lake

Abstract

Abstract The Imperial Oil Limited, Cold Lake commercial bitumen recovery project has been in operation for over ten years. Over 500 pilot area and 1900 commercial area wells have been drilled and produced using a cyclic steam stimulation (CSS) process. The bitumen, with a live in-situ viscosity of 65,000 cp, is essentially immobile before heating with steam. Initial steam injectivity is only achieved with pressures exceeding the formation fracture pressure of 10 MPa. The CSS process is a complex interplay of recovery mechanisms including bitumen viscosity reduction by heating, gas drive, recompaction drive and gravity drainage. Multi-well numerical simulation history matches of several pads in this mature cyclic steam operation have been completed. The current study utilized 9 cycles, or 10 years of performance data in which up to 24% of the original bitumen in place has been recovered. This is almost double the 5 cycles (5 years) and 15% bitumen recovery used in previous Cold Lake CSS history match work. Significant changes from previous Cold Lake modeling work are the inclusion of a match of gas production, pressures and temperatures at six observation well (OB well) locations and process conformance as measured by seismic. Several new simulation modeling techniques were incorporated to achieve the field data matches. One example is that the match of OB well temperatures required that steam injection create an isotropic horizontal fracture near the well perforation depth. Another is that matching of gas production trends required a "foamy oil" phase, or higher critical gas saturation, in the warmer regions of the reservoir as well as a representation of in-situ generation of CO2 from steam-rock reactions. The resulting modeling techniques are used to assess improvements to operating strategies, future development and CSS follow-up recovery options. The study also identifies several areas that warrant further improvement. Introduction Background. Imperial Oil's Cold Lake bitumen recovery project is located 260 km northeast of Edmonton, Alberta. This commercial cyclic steam stimulation (CSS) project has been in operation for over ten years and now involves 1900 wells currently producing 14500 m3/day of bitumen. Twenty or thirty wells are drilled on a 1.62 ha (4.0 acre) spacing from a single surface location called a "pad". Throughout the piloting and commercial implementation of CSS at Cold Lake, reservoir simulation has been utilized to guide development and operating practices. The current work is the fourth major history match that has been conducted on Cold Lake CSS. The first work in this area was completed in 1987 and focused on the average behavior of L pad wells in the Cold Lake Lemming pilot area. The single well, 2D radial L pad model demonstrated the need to account for geomechanical effects including matrix dilation and fracturing. These mechanisms were necessary to achieve a steam injection pressure match and an acceptable representation of areal heat distribution. References 1 and 2 discuss these modeling techniques. In 1989, a multi-well history match was completed which focused on W pad in the Lemming pilot area. W pad is a prototype of the commercial 1.62 ha, 1.7:1 aspect ratio well configuration. The principle objective was matching typical inter-well fluid movements. The study concluded that to match five cycles of CSS, a model containing a minimum of three wells was needed in order to match the middle well in the model. The study also concluded that horizontal fractures were narrow features which, on W pad, were oriented at N43 E. In 1990, a study was undertaken to match the performance of wells drilled on close spacing of 0.94 ha (2.33 acre). This study concluded that modeling parameters used in the W pad work were not suitable for the closer spacing. The current study was undertaken to improve confidence in simulation based predictions of CSS and secondary or "follow-up" processes to CSS by extending previous history match work and clarifying the role of solution gas as a drive mechanism. Available field data spanned more than ten years, double that of the earlier work. P. 307^