Maximizing efficiency and uniformity in SAGD steam circulation through effect of heat convection

The thermal recovery is facing the challenge of improving quality and efficiency. Steam-assisted gravity drainage (SAGD) steam circulation has a profound implications for the formal production. There is a lack of research on the situation when the steam injection pressure of injection well is lower than that of production well during startup stage. In this paper, the effects of diverse injection pressure difference on SAGD steam circulation and production stage are investigated by analytical modeling and numerical simulation, especially when the steam injection pressure of producer well is larger than that of injection well, and reliable results are obtained by temperature falloff test and model comparison validation. Meanwhile, in order to minimize the factors affecting the simulation accuracy, a sensitivity analysis of the temperature prediction model for the startup stage is carried out using Monte Carlo method and the finest possible mesh is used in the numerical simulation. The results show that:①The preheating results are faster and more uniform than the conventional preheating method when steam injection pressure of producer is greater than that of injector, and the subsequent production indexes are also superior to those of the conventional preheating method. ②The injected steam temperature had the greatest effect on the prediction accuracy of the analytical model; The finer the numerical simulation grid division, the lower the midpoint temperature of the horizontal well pair; ③An optimal range of injection pressure differences that achieves the best balance between preheating efficiency and thermal recovery effectiveness is achieved with Pprod-Pinj in the range of 400–500 kPa. ④The preheating method investigated in this paper minimizes the effect by unfavorable factors such as reservoir non-homogeneity, which holds the potential for more uniform, time-saving preheating and without the addition of field equipment.