CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (http://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ANSYS Meshing] How to improve skewness? (http://www.cfd-online.com/Forums/ansys-meshing/77832-how-improve-skewness.html)

em02131 July 5, 2010 19:33

How to improve skewness?
 
Hello guys, I am a student finalist in Portugal that is doing the master's thesis which is about CFD.

In my case, I am trying to simulate numerically the ventilation of an amphitheater at my university. The problem is that I'm already working on this for weeks now and had to stop.

What is happening is the following:

I want to simulate the ventilation for several turbulence models (standard ke, RNG ke, realizable ke, kw and standard RSM) and what happens is i get results that converge to "first order upwind discretization methods" but not for them " second order upwind". I alredy diminishes the parameters of "under-relaxation" and nothing new...:mad:

There is something that intrigues me and I can not in any way solve!

When creating the mesh (a mesh created by ANSYS WORKBENCH v.12.0) happens to me when doing a "report quality" appears to me this:

Applying Quality criteria for tetrahedra / mixed cells.
Maximum cell squish = 9.76386e-001
Maximum cell skewness = 9.99855e-001
Warning: Exceeds maximum cell skewness 0.98.
Maximum aspect ratio = 2.02305e +002

Is for this reason that my calculations do not converge?!?!

Can someone help me in this?!

I'm really desperate ...:confused:

Thanks in advance.

Rui Alves

bob12 July 6, 2010 04:05

yes, the simulation depend on quality of mesh and in your case are too low so you can try to improve it.
for improve the quality you can used the size function or you change the size from coarse to medium or fine. Look with this operation you improve the number of the mesh and consenquently the phisics memory of PC.

em02131 July 6, 2010 08:28

Quote:

Originally Posted by bob12 (Post 265879)
yes, the simulation depend on quality of mesh and in your case are too low so you can try to improve it.
for improve the quality you can used the size function or you change the size from coarse to medium or fine. Look with this operation you improve the number of the mesh and consenquently the phisics memory of PC.

hi bob.

i'm alredy do that procedure and no improvements...:-S

any other sugestions?

the geometry of the problem have influence in that?

bob12 July 6, 2010 10:33

yes, also the geometry influence the mesh quality, in particular the angle too small are critical generally when it's possible a semplification at the geometry are well accepted.

srr November 23, 2010 12:04

Sometime it is posible to improve doing some local densities where the maximum skewness is located, for example if you have a corner you should make a density there because that geometrical feature is critical.

hope this helps

srr November 23, 2010 12:07

I also recomend to change the mesher, ICEM meshes way better than the ANSYS workbench

PSYMN November 24, 2010 11:34

To counter (or at least temper) srr's suggestion...

ANSYS Meshing is perfectly good and can often create a better mesh. It has many ICEM CFD methods integrated under the hood along with technologies from Fluent and other ANSYS brand technologies.

For instance, the ANSYS Meshing Sizing Function, combined with the Patch Conforming Tetra and Pre prism algorithm can usually generate a much prettier mesh tetra prism mesh than ICEM CFD can, and usually in a fraction of the time.

However, you should go to ICEM CFD if your model is dirty or you need to start from a mixture of geometry, mesh and/or faceted geoemtry. If you really need Patch independence or if you need to generate a really large model, or if you need ICEM CFD Hexa, particularly for Structured Hexa meshing, or if you really need mesh editing or scripting output to 3rd party solvers, etc.

ICEM CFD is more flexible than ANSYS Meshing, but for new users, that can also mean harder to learn. ANSYS Meshing, especially version 13.0, can generate very nice mesh.

Perhaps EM02131 can post some images of the geometry and mesh so we can suggest improvements or perhaps a different mesh method. For instance, if he were using the Hexa Dominant method (designed for FEA), he should definitely expect issues with his CFD solver.

Nicci January 31, 2013 03:44

Improving accuracy of results for skew meshes
 
Hallo,

I would just like to know if there are any interpolation methods that can be implemented in OpenFOAM to improve the accuracy of the results for skew meshes?

Thank you.

PSYMN January 31, 2013 10:25

Nicci, that would be a great question for the Open Foam forum...

mausam24 June 5, 2013 05:19

Can we generate fine mesh for the 3D rotating turbines using Hex dominant mesh? The size of the turbine is huge and I have been trying to solve the whole at once. I tried using Tetrahedral but the size of the element were extreme. So I tried with Hex mesh, It didnt get any result

sanjeetlimbu April 7, 2015 11:34

4 Attachment(s)
Dear sir

I am trying to do a 2D planer simulation for a compression stroke , using in cylinder dynamic mesh- with profile notepad- data point. I am able to move the piston part for the named boundary piston as per requirement. but I am getting error Negative volume -dynamic mesh update error in run at fluent. The issue seems to be due to skweness as i tried to improve the orthogonal quality.
I am using Quad mesh tppe. I am new in CFD and I need to do axisymetric mode once this runs for the planer mode. My aim is to analyse combustion at end of compression


Details: Stroke length 254mm (travel by piston edge), bore 25.4mm (breath of mesh rectangle)
In cylinder motion: time step 0.000041s (default).
Compression stroke time = 30 ms , + 30ms to see combustion event
Veloity profile= start from zero to peak 14 m/s and abrupt deceleration to zero at 254mm stroke travel end.
Mesh : body size 0.001m

Pl tell how to get good mesh and how to get it for axisymetric.

sanjeetlimbu April 7, 2015 12:22

4 Attachment(s)
I tried edge sizing but unable to replicate the grid made in the referance photo shared earlier
Pl share any suggestion if how to get the mesh as in the piston geometry photo shown in before post- in that the grid are distributed very perfect square size with inflattion


All times are GMT -4. The time now is 12:11.