Search for a command to run...
In this article, we consider the analysis of heat transfer in corrugated channels, a typical geometry found in plate heat exchangers. Using finely resolved large eddy simulations along with the periodic resolution method for the temperature transport equation, we analyze the structure of the temperature and of the wall heat flux for various corrugation patterns, with corrugation angle ranging from 15° to 45° and for a range of Reynolds numbers from Re = 4721 to 47025. We highlight the importance of the contact points between the upper and lower walls, characteristic of our geometrical setting, for the intensification of the heat flux and of the pressure drop and the turbulence generation. We report that the generalized Lévêque equation gives a very good estimation of the mean wall heat flux over the full range of Reynolds numbers and geometries, once the head loss coefficient is known accurately, confirming the close connection between head loss and heat transfer mechanisms.The LES results are further used to assess the performances of the modeling of the heat flux with RANS approach. We compare the predictions of three commonly used algebraic heat flux models (AFM), namely the models that rely on the Simple Gradient Diffusion Hypothesis (SGDH), the Generalized Gradient Diffusion Hypothesis (GGDH) and the Higher Order Gradient Diffusion Hypothesis (HGDH). We show that for such complex geometries, as well as for the plane channel flow, the predictions of the various AFM with the standard value of their parameters are quite far from the reference. We also demonstrate that the performances of the RANS predictions of the Nusselt number can be greatly improved by adjusting the model parameter. However, once optimized, it is shown that all those AFM give exactly the same prediction for the wall heat flux. Nevertheless, even if good predictions of the heat transfer coefficient can be achieved, significant discrepancies, remain in the RANS prediction of the turbulent heat flux.