A fast natural convection algorithm based on dividing fluid development stages

We develop a numerical method for fast computation of natural convection, which proposes a new dimensionless number (Fs) to characterize the degree of influence of convection on the temperature field in the flow field and determines the moment of pause for the updating of the flow field by designing judgmental conditions to delineate the stage of development of the flow field, where the loosely coupled computation is turned on to improve the efficiency of the transient temperature field computation. The accuracy of the algorithm is verified using an experimental case of a standard model of natural convection, and the robustness of the algorithm is verified by specifying different monitoring boundaries and setting different numbers of monitoring steps, and the algorithm is applied to a model of natural convection in the equipment in the cabin of the vehicle. The results show that the computational speed is increased by 8.8, 6.4, and 3.5 times after turning on the loosely coupled computation in the first, second, and third development phases, respectively, and the average errors of the monitored point temperatures are 0.7%, 0.1%, and 0.028%, respectively. By monitoring the change in Fs during the loosely coupled computation, the error of the first development stage under the variable boundary is reduced by 95.1%, and the computation speed is 2.2 times faster than that of the second development stage.