Effects of near-wall Reynolds-stress modeling on the calculation of the turbulent thermal field

The effects of near-wall Reynolds-stress modeling on the calculation of inc ompressible wall-bounded turbulent flow with heat transfer were investigate d. Two near-wall heat transfer models, a two-equation model and a Reynolds heat flux model, and five different near-wall Reynolds-stress models were c onsidered. All models examined satisfied the wall boundary conditions exact ly. Some were asymptotically consistent near the wall, others were not. Whe n the Reynolds-stress models are used in conjunction with the two-equation model to calculate the thermal field, the predicted mean and fluctuating te mperature characteristics are found to be essentially the same when the nea r-wall asymptotes are predicted correctly by the models. However, some disc repancies are noted in a region very near the wall where the calculated dis sipation rate of temperature variance differs for the different Reynolds-st ress models considered. On the other hand, the calculated mean temperature differs significantly in the outer region when the Reynolds-stress models a re used in conjunction with the Reynolds heat flux model and when the near- wall asymptotes are not estimated correctly by the two-equation model. In t he former situation, the major source of error can be traced to the modeled equation for the normal heat flux. This equation, in its current modeled f orm, is very sensitive to the prediction of the leading terms in the near-w all expansions for the turbulent kinetic energy, k, and its dissipation rat e, epsilon. Small differences in these predictions call lead to large discr epancies in the calculations of the normal heat Bur in the near-wall region and the mean temperature in the outer region. In the latter situation, the discrepancies can be traced to an incorrect estimate of the near-wall asym ptotes of k, epsilon and the temperature variance and its dissipation rate. This suggests that the temperature field model has to be at least one leve l lower than the velocity field model. Even then, the models should be able to predict the near-wall asymptotes correctly. (C) 2000 Elsevier Science I nc. All rights reserved.