[Stormbringer505]

Hello everyone,

I am working on my master thesis which is about CFD modeling and improving a woodchip boiler's water jacket.

I've already created a mesh, mostly hex cells where heat transfer is relevant, and now I am running some simulations to determine the flow field within the water jacket as a start before I add the combustion/energy equation.

I am having trouble simulating the flow (using Ansys Fluent 15.0). On the in- and outlet pipes the Re-number is about 27.000 so this flow is fully turbulent but there are areas in the waterjacket that are going to be laminar. I can't determine exactly where they are so I can't just slice them out and add a laminar zone to them.

Some of the fluid zones are large but thin, for example 1.2 x 1.2 meters but only about 30mm thick (or 3x3 feet with a thickness of one inch in imperial). I have created five to seven cells across the thickness to keep the cell number low, it is already about five million.

Now, there's the problem: I am really not shure which turbulence model can be used in this case. I have thought about some possibilities but I want to hear the opinion of further experienced people on them:

a) Do fully laminar calculations?
I guess the results won't be very good.

b) Add a turbulence model with wall functions?
Right now I am running a simulation using the realizable k-epsilon-model with enhanced wall treatment. Only looking at the residual monitors it seems to work fine but I know that low residuals do not necessarily mean that the solution is correct. I will examine this tomorrow and let Fluent do some more iterations on it. I am running on the pressure based coupled solver with a CFL of 200 and 2nd order discretisation (1st order is unstable as I've seen so far).

In the beginning I was running on the SST k-omega-model with low Re-correction but the residuals for continuity wouldn't go below 1e-1 which is, in my opinion, not very good in this case. I've done some reading since them and I've learned that I need to set a finer mesh to resolve the boundary layer with this model, am I right?

c) Use a transition model
I haven't tried this yet and it is completely new to me. I've read that transition models also need a fine mesh to resolve the boundary layer, is that true?

I hope you guys can give me some advice. Even though I am working on a pretty powerful computer at the university calculation times are long and since literature does not simply say "use this or that model" I hope some experience might help.

Thank you!