1. The computational simulation of fluid dynamics and heat transfer, in aeronautics and other fields, has become widespread because it serves as an effective tool for designing and investigating various engineering processes. In m.ost practical circumstances, these processes involve complicated nonlinear phenomena, and as a result, significant efforts have been recently directed towards the credibility of the simulations. A special section of a recent issue in the AIAA Journal (vol. 36, no. 5, May, 1998) has been devoted to this topic. Since significant efforts are often devoted.to mesh refinement studies, and design decisions may be based on the numerical simulations, the results should represent physically plau-Sible flow .behaviour. However, most simulations typically involve only the mass, momentum and energy equations; thus, nonp.hysical results may still arise because little attention is given to the Second Law of Thermodynamics.
2. In this problem, thermal buoyancy leads to a clockwise recirculation cell within the cavity. Figures 2 - 3 illustrate the temperature and vvelocity results after 2 timesteps (At = 0.155) have elapsed. The timestep complies with the earlier guideline At M 0.`5 - 5At0022. In Fig. 2, the temperature contours are skewed as a result of the recirculating fluid flow within the cavity. Figure 3 illustrates that fluid ascends along the hot wall. It then turns rightward near the top boundary as it encounters an adverse pressure gradient in the upper left corner. Then the fluid accelerates along the upper boundary due to the favorable pressure gradient in this direction. Another adverse pressure gradient in the upper right corner turns he fluid down and cooler air descends along the right cold wall. It soon returns back to the left boundary to complete the recirculation cell.
3. Figure 5: U-velocity (m/s) - steady state Figures 4 - 6 illustrate the steady state contours for temperature, u-velocity and v-velocity. Several aspects of the computations agree well with the benchmark solution. For example, the predicted position of the maximum horizontal velocity along the vertical midplane agrees within 1% of the corresponding result from the benchmark solution.