To thin boundary layer?

 Register Blogs Members List Search Today's Posts Mark Forums Read

 February 13, 2009, 08:15 To thin boundary layer? #1 Luk Guest   Posts: n/a Hi all, I want to ask what are the consequences of using to thin boundary layer in meshig of the domain. I perform simulaiton of axial fan (I simulate 1 blade) with Re=10^6 and diameter of 1.5 m. While doing mesh independent studies I noticed following observation. I meshsed domains 2 times, one time with inflation layer first node placed for 1 m dimension (which is suitable for geometry) and one time for 0.3 m (what gives relatively thinner first layer of the mesh). I perform the simulation in ansys 11 and ansys 10 cfx version having following results: a) cfx ver. 11 and "thick" first element of boundary layer - mass flow of 2.85m3/s and nice convergence b) cfx ver. 11 and "thin" first element of boundary layer - almost the same mass flow and convergence c) cfx ver. 10 and "thick" first element of boundary layer - mass flow of 1.3m3/s and big problems with convergence d) cfx ver. 10 and "thin" first element of boundary layer - the same situation as for version 11,the same flow, everything ok. e) cfx ver. 10 and "thick" first element of boundary layer and 4 times denser mesh - mass flow was again in range of 1.3m3/s and big problems with convergence Every other parameters are exactly the same. The simulaitons vere performed few times to ensure. Can anyone explain it? Luk

 February 13, 2009, 09:47 Re: To thin boundary layer? #2 Timon Guest   Posts: n/a I hope you mean mm's for your first node height . It all depends on your turbulence model and whether or not you wish to apply wall modeling. What are the y+ values for both simulations and what turbulence model are you using? In general your solution should become better if you refine your boundary layer area (when your solving up to the wall), but if you're using wall modeling this isn't necessarily true.

 February 13, 2009, 11:06 Re: To thin boundary layer? #3 Luk Guest   Posts: n/a Hi, Yes,it is in mm. My problem is that every other parameter of the mesh and simulation is the same (Y+ is set to 11 in both cases, I use k-e model with scalable wall functions). The strange thing is that there are differences between cfx versions and overall differenfce in flow image. I am not yet sure about other mesh densities. Luk

 February 27, 2009, 03:22 Re: To thin boundary layer? #4 radionline Guest   Posts: n/a Hi Luk, I can't imagine that the thickness of boundary layer is the reason of your convergence problem?! Which conditions do you apply on Boundaries (Inlet/Opening/Outlet)? I had made the experience the mass flow has to be set an Inlet, and Outlet Total pressure, or sometimes like better to implement as Opening (Re-Flows!!) If I switched the boundary conditions i had also various problems with the convergence...