In order to understand the discrepancy in stratification observed between experimental and numerical studies of natural convection at high Rayleigh number in air-filled differentially heated cavities, we developed three different but complementary approaches: experimental measurements, 3D DNS and 3D LES.
An experiment of a cavity of height 1meter, width 1m and depth 0.32m was set up: velocity profiles are given by Laser Doppler Anemometry, temperature profiles on the top and bottom walls are measured using micro-thermocouples and temperature distribution on the front and rear walls are captured by an infrared camera. Using the measured temperature profiles as the thermal boundary conditions on the four passive walls, 3D LES yielded results in good agreement with experimental data. An excellent agreement is also obtained for 3D DNS, when attentions are paid to heat conduction in the thermal insulation and surface radiation between cavity interior surfaces.
The present work shows that the discrepancy in stratification observed is due to the fact that thermal boundary conditions used in previous numerical simulations were different from the experimental ones. The weak thermal stratification in experiments comes mainly from the coupling of natural convection with surface radiation and cannot be correctly predicted without considering this coupling.