|
[Sponsors] |
June 28, 2016, 08:42 |
2D problem with Quasi-3D mesh
|
#1 |
New Member
Join Date: Jun 2016
Location: Germany
Posts: 3
Rep Power: 9 |
Hi everyone!
I'm running a quasi-3D simulation (2D mesh with 1 cell width in third direction z). I chose inviscid walls for the boundaries in z direction and did some study with varying the physical dimension of the width in z. Depending on this width I get different boundary layer thicknesses what confuses me. Anyone with experience/expertise in this area? Or can give a hint where to get some (scientifcally) sound information about Quasi-3D setups? I searched the web but wasn't successful in finding any well-founded theoretical approach. |
|
June 28, 2016, 09:38 |
|
#2 |
Senior Member
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,769
Rep Power: 71 |
what do you mean for "boundary layer"? you applied inviscid bc...
Generally, the periodic boundary condition is used for such test |
|
June 28, 2016, 10:29 |
|
#3 |
Super Moderator
Alex
Join Date: Jun 2012
Location: Germany
Posts: 3,399
Rep Power: 46 |
Either use periodic or symmetry boundary conditions. Any boundary condition that involves a "wall" can give rise to some unexpected behavior in this case. E.g. when using turbulence models where k=0 is imposed at wall boundary conditions.
|
|
June 28, 2016, 12:00 |
|
#4 |
New Member
Join Date: Jun 2016
Location: Germany
Posts: 3
Rep Power: 9 |
Ok, let me be more specific (see also attached pic of my reduced test setup):
-structured mesh of a duct profile (in the reduced test setup it is a straight line, no curvature), ijk-ordered, i-j-k orientation is equivalent to x-y-z axis -flow direction is x=0 -> x=inf -in y direction there are viscid walls -in z direction I have my "one cell height" BCs for z=min and z=max is inviscid What I did is vary the z coordinate of k=2 I hope it became clear enough what I did? I took a real 2D solver and I analytically calculated an estimate for a turbulent boundary layer. Both of them match. For my Q3D problem: If I choose the "right" cell width in z direction, I also get those results. If the z width becomes to low or high, the boundary layer thickness shrinks (?). I also did this with a symmetry instead of a inviscid BC, but to the same effect. |
|
June 28, 2016, 12:30 |
|
#5 |
Senior Member
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,769
Rep Power: 71 |
not sure to understand you final goal ... however, you are checking the Mach number, I suppose your problem is compressible (I dont'know if you are solving laminar or turbulent, steady or not), therefore you have also influence of the energy and density equations (Mach profile is a combination of several variables). I suggest to check always for the resolved variables, for example the velocity field and the temperature field, not the Mach number.
Furthermore, you are changing the global volume of the domain and I suppose that the consequence in inflow is in a greater mass and kinetic energy flowing into the duct. Finally, try setting periodic boundary conditions |
|
June 28, 2016, 12:55 |
|
#6 |
New Member
Join Date: Jun 2016
Location: Germany
Posts: 3
Rep Power: 9 |
Thanks for your answers so far.
Sure, I can check for velocity, Mach number was just still activated. However, the profile looks the same for velocity in x direction. The phenomena with changing cell volume sizes still exists. The final goal is to simulate a duct with a more complex geometry than this little example. The problem is steady, turbulent, compressible. The one you see is just to understand how numeric parameters influence my physical solution (i.e. cell width, BC etc.). However, I'm still looking for some literature that can explain the Quasi-3D approach. Someone must have thought about that before? |
|
June 28, 2016, 13:01 |
|
#7 | |
Senior Member
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,769
Rep Power: 71 |
Quote:
First of all, the test must be performed for steady laminar condition, no turbulence modelling must be used. Then, there is no need to have a reference in the literature, this test is used for example when a 2d analytical solution is used as reference in a 3d code. If you set walls (even if with inviscid BC.s) you are forcing a real strong 3d effect. |
||
March 17, 2019, 11:22 |
|
#8 | |
New Member
Mehmet
Join Date: Feb 2019
Location: mhmt.ozden@yandex.com
Posts: 23
Rep Power: 7 |
Quote:
I would like to apply quasi 3D simulation for radial outflow turbine, but I don't know how to determine the geometry and set up. If you figure it out, could you help me, please? Do I have to use some code for quasi 3D and what I need to do with the growing end-wall boundary layers? "If the area of the flow-path changes significantly in the axial direction it might be necessary to instead make a quasi-3D simulation. A quasi-3D simulation is a 2D simulation in which extra source terms are used to account for the acceleration/deceleration caused by a changing channel height or growing end-wall boundary layers. Codes focused on turbomachinery applications often have the possibility to perform quasi-3D simulations" |
||
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Dynamic Mesh Problem. | Tom Clark | FLUENT | 10 | June 21, 2021 04:27 |
Overset MESH problem DFBI 6DOF | Ale85 | STAR-CCM+ | 0 | October 1, 2013 12:25 |
[ICEM] ICEM Structured Mesh Problem | OMJT | ANSYS Meshing & Geometry | 3 | March 22, 2013 10:06 |
[GAMBIT] SOFC crossflow mesh possible problem??? | epc1 | ANSYS Meshing & Geometry | 1 | February 4, 2013 01:57 |
[snappyHexMesh] snappyHexMesh won't work - zeros everywhere! | sc298 | OpenFOAM Meshing & Mesh Conversion | 2 | March 27, 2011 21:11 |