Towards the efficient turbulence closure for mixing phenomena in the core outlet of a nuclear reactor

Proper mixing in the core outlet region of a nuclear reactor is a crucial issue for its safety assessment. Because of the high temperatures involved, it is especially important for the further development of high temperature reactors. Computational fluid dynamic (CFD) simulations are very useful for the assessment of such mixing phenomena, although they have to be accurate and cost efficient. Advanced approaches for turbulent modeling, such as large eddy simulation (LES), are usually more accurate and also more computationally expensive. On the other hand, the gain in accuracy with respect to the traditional models, such as Reynolds averaged Navier-Stokes (RANS) and its unsteady counterpart (URANS), is not quantified. The present study provides a quantitative assessment of different approaches for turbulence modeling (RANS, URANS and LES) in terms of accuracy and computational costs. The accuracy is evaluated by direct comparison of the numerical results to the experimental data and the costs are assessed from the processor time spent for simulation. The results show that the average cost of unsteady simulations is higher than RANS almost by two orders of magnitude. Although the increase in the accuracy is not very big, the RANS simulations seem to be more efficient for the considered flow.