A Production Performance Model of the Cyclic Steam Stimulation Process in Multilayer Heavy Oil Reservoirs

Cyclic steam stimulation (CSS) is a typical enhanced oil recovery method for heavy oil reservoirs. In this paper, a new model for the productivity of a CSS well in multilayer heavy oil reservoirs is proposed. First, for the steam volume of each formation layer, it is proposed that the total steam injection volume will be split by the formation factor (Kh) for the commingled steam injection mode. Then, based on the equivalent flow resistance principle, the productivity model can be derived. In this model, the heavy oil reservoir is composed of a cold zone, a hot water zone, and a steam zone. Next, using the energy conservation law, the equivalent heating radius can be calculated with the consideration of the steam overlay. Simultaneously, a correlation between the threshold pressure gradient (TPG) and oil mobility is also applied for the productivity formula in the cold zone and the hot water zone. Afterward, this model is validated by comparing the simulation results with the results of an actual CNOOC CSS well. A good agreement is observed, and the relative error of the cumulative oil production is about 2.20%. The sensitivity analysis results indicate that the effect of the bottom hole pressure is the most significant, followed by the TPG, and the effect of the steam overlay is relatively slight. The formation factor can affect the splitting of the steam volume in each layer; thus, the oil production rate will be impacted. The proposed mathematical model in this paper provides an effective method for the prediction of preliminary productivity of a CSS well in a multilayer heavy oil reservoir.

Automated summary

In this paper, a new model for the productivity of a CSS well in multilayer heavy oil reservoirs is proposed, which can accurately predict the preliminary productivity of CSS well by analyzing the formation factor and equivalent flow resistance, etc. The accuracy of the model is verified by comparing with actual results.