Hy! I'm trying to model a pool with water entering at one temperature and going out at a lower temperature. The pool is loosing energy through the walls and mainly by the surface where I fixed the heat flux. I'm getting a huge transfer coefficient and almost a constant temperature in the vertical direction. I imagine it´s because of the model I used (k-E) and the mesh I had. I'm using now k-w model. Will this fix the problem? What should I do when I want to calculate the heat through a wall? Bye. I'll be waiting for answers.

This is a tough problem. Is it turbulent? Only locally I guess. Have you turned on buoyancy? If so, what is your time step? I bet you have a very large timestep to speed up the distribution of the temperature but this is generally not applicable for buoyant flows. Stick to the buoyancy timestep, or increase by a factor of 2 max. And then wait until eternity to get a proper balance for the energy equation.

Moreover, if it is turbulent, how good is your grid near the wall. This has to be very fine..........

Good luck, Gert-Jan

www.bunova.nl

Yes, I have turbulent flow locally. Bouyancy is turned off becaused the change in density is very small. I supose that the boussinesq model is working (works automatically??). Now I have a result with the k-w model and the vertical profile is almost constant but next to the lowest temperature (it seems a problem of the initial value). The timestep is automatic by the moment and even the residuals (heat transfer) are of less than 10-3 the energy balance is still away from the being satisfied. Next to the upper wall I used inflation and the rest of the mesh is unstructured (tetrahedral). Should I change the model? I thought that using k-w it wasn´t neccesary to refine the mesh next to the wall. I'll be waiting for your answer. Bye.

You have to turn on buoyancy. It won't switch on automaically. Otherwise you will be modelling unrealistic physics. Unstructured with Inflation is fine. However, what is your Prantl Number? I guess it is 7. Then all alarm bells should ring. What does this number mean and what is the effect on the mesh density near a non-adiabatic wall in a turbulent region.........

Gert-Jan

www.bunova.nl

Hello. Thanks for the answers and sorry for taking so much time to answer you. My Prandtl number is very close to 7.

Why is that I don´t have problems at the others walls? If I'm using a Eddy-viscosity model or a reynold stress with wall functions, why should I have a finer mesh? Isn't is enough with y+>30? Wouldn't I get a wring result with a finer mesh? I will be waiting for your answer. Thank you very much.

What does Pr=7 mean? Did you look it up? When you have done this, try to figure out what the consequences will be for the near wall mesh.

If you don't understand just do the simulation and check the temperature profile near the wall.

Regs, Gert-Jan

www.bunova.nl

I think it means that transfer trough convection will be more or less 7 times more important than tranfer by conduction. So, all the temperature profile will be in the hydrodynamic boundary layer. Then, I'm overestimating h at the wall. I think I will have to use a low Reynolds model and refine the mesh until y+=1. Is this ok? Thanks for the help.

Bingo. Use SST.

Good luck, Gert-Jan

www.bunova.nl

Thank you very much. I'll do it. Bye.