hi,all:

I learn LES recently.But I still confuse with the difference for heat transfer in LES simulation and RANS simulation.What is the advantage and disadvantage in the two method to simulate the heat transfer?

Roughly, RANS claims that the all the fluctuating parts of (u,p,T) are random and need to be included in the RANS-turbulence model. LES calims that only the smaller scales of the fluctuating pats of (u,p,T) are random, and that all the larger scales of the fluctuations should be calculated directly(i.e. resolved). LES uses its SGS-model to model the effect of smaller scales. Then when you use RANS, you computer gives you a steady flow, while with LES your computer gives you a fluctuating flow that contains the resolved eddies(the larger scales as mentioned above). For heat transfer, the RANS-turbulence model of the heat equation has greater diffusion coeffients, while the SGS-model of LES has smaller diffusion coeffients. Because the resolved LES eddies also make contribution to heat diffution, so the SGS-model diffusion plus the resolved-eddy diffusion should be at leat theoretically equal to the RANS diffusion! How to divide the two scales of LES is an art actually. Most LES in practice uses the grid size as a natural threshold scale.

Because the large scales of turbulence depends much on the boundary. To expect a RANS-turbulence model to be accurate seems too naive. But RANS is easy to compute with a coarses grid system. There is a trouble with the LES mesh resolution. Very fine mesh is often required for LES. Which is better? LES or RANS? No exact answer. Too many factors there.