Hy,

I want to try a transient analysis with natural convection, including a solid body (steel) and I have some questions.

1. What cell-sizes should I use in the solid and at the boundary to the fluid ? Do I have to refine the cells at the solid-fluid-boundary in the solid-region too or is it better to make them large ? And in the solid itself ?

2. Does the segregated solver with simple work ?

3. How do change boundary conditions with time ? I want to increase the temperature of a velocity inlet for delta-T=200 in 2 hours. Is the only way to write a journal-file, that changes temperature each timestep and iterates one timestep ?

4. If Ra is high enough, which turbulence-model and which wall-modell do you recommend ? How do you handle the change from laminar to turbulent natural convection ?

Thanks a lot

Christof

Hi, I know some of your answers! 1) A segragated solver with SIMPLE works for natural convection problems. 2) It is not possible to solve the transition regime flow of any kind (in FLUENT). 3) The cell sizes will have effect on accuracy only but not convergence!

Hy, thanks for your answers. You said, cells-size (solid-cells) will have effect only on accuracy, not on convergence. But I don't understand this really. If you make heat transfer calculations by hand (for example heat transfer with several layers) you don't have to discretisize the area/layers, you only take the thickness. And isn't heat-transfer in a solid a linear problem (something like (d2T)/(dx^2)+(d2T)/(dy^2)=0) which could be solved directly ? Do you have for example different heatfluxes from the fluid to the solid, if you have different sizes of the "first" solid-cell ?

Thanks christof

You want to make sure that you don't try to cram too many cells into the boundary layer. Estimate the thickness of your boundary layer, and try having one or two cells in that thickness against the wall. We've had good results using this approach for solid/water interfaces.