A Sensitivity Study Of Cyclic Steam Stimulation Using A 2-D Mathematical Model

Abstract

Abstract INTERCOMP's steamflood model was used to carry out a sensitivity study of cyclic steam stimulation as applied to the Cold Lake Oil Sands. A 2-D radial system was used and parameters were based on knowledge of the Mannville "C" unit. No history matches were attempted in this study. Variables investigated included steam-treatment size, length of soak, steam injection rate, steam quality, completion interval, effect of natural gas injection and strategy for multiple cycles. It was found that the benefits of a soak period were outweighed by the resulting drop in calendar-day oil rate and that, for the same reason steam should be injected at the greatest feasible rates. Gas injection was shown to be of limited value. The optimum strategy for scheduling cycles involves a continuous increase in treatment size in order to contact unheated bitumen on each cycle. Runs to investigate the effect of completion interval are less easy to interpret, but restricted intervals are beneficial in the short term. Introduction Cyclic steam stimulation has proved to be a technically feasible method of recovering bitumen from the Cold Lake Oil Sands of north-eastern Alberta and several operators are currently conducting field trials of the process. BP's pilot work in the area consisted of two steam-floods carried out between 1966–70 on adjacent legal subdivisions in Oil Sands Lease 64 in the northern part of the Cold Lake deposit as indicated in Figure 1. The second pilot included a few steam stimulation cycles which were, however, insufficient both in size and number to constitute a reasonable test of the applicability of the procedure. This paper reports on a study, using the INTERCOMP steamflood simulator described in Ref. 1, to model cyclic steam stimulation for an oil sands situation. Since many reservoir and fluid parameters were unknown and there was no field data to history-match, the study is regarded as being very preliminary in nature. By using "best-estimate" parameters it was hoped to investigate the problems involved in carrying out such simulations and to perhaps gain some insight into the process mechanisms. In these respects, the study was judged to be quite successful. General levels of performance calculated by the model were reasonable and changes seen in temperature pressure and saturation distributions throughout a cycle are believed to be realistic. A number of runs were carried out, varying certain of the operating parameters, and the resulting changes in performance were analyzed. Representation of the Reservoir-Fluid System The INTERCOMP model has the capability of handling 3-phase flow in a three-dimensional system but, in this study, the grid-block configuration represented two-dimensional radial flow into a single well. As shown in Figure 2, the reservoir was divided into six layers. Parameters for each layer are shown in Table I. Layer 1 has high gas and water saturations and is not typical of the total lease area. A further feature is the presence of a number of hard impermeable layers which are not, however sufficiently continuous to represent a serious impediment to vertical communication.