In the natural convection tutorial cases gravity is set to -6.94E-5 m/s^2, rather than -9.8. My cases I had been running used that also (for a similar physical problem). I am trying a couple of variations with -9.8 instead. On one of them, I got floating point errors when I made that change (the ONLY change to an otherwise converging case). All it says in the literature is that gravity was adjusted to make the Prandtl and Raleigh numbers come out to the desired values. I don't really understand how this was done. Am I missing something? Any help you can give would be appreciated.

If you've changed gravity form 6.94e-05 to 9.8, it means that Rayleigh number was 140000 times increased. Then, the Rayleigh number may be too large and may cause numerical divergence. So, first of all, please check your Rayleigh number.

Sincerely, Jinwook

Thanks for answering. I've been thinking about it and asking other people, but I have not really gotten any where. The problem is, for what I am solving, I don't really know what the Raleigh number is. I would have to know the surface temperature, which I don't. It is one of the things I am solving for. What I am really wanting to know is, in the tutorial, did they force a desired answer by changing actual physical properties? I would like to be able to have the model be physically real. Is it possible that in reality I must turn on the turbulence model?