|
[Sponsors] |
April 23, 2020, 13:01 |
Inflation layer problem
|
#1 |
Member
Join Date: Mar 2020
Posts: 33
Rep Power: 6 |
Hello.
I am trying to run steady state simulation of compressor in CFX. I have created an initial mesh which results in convergence. However, If I try to generate inflation layer on blades surfaces with appropriate inital layer thickness, I get very very large aspect ratio and expansion ratios and simulation diverges in CFX. The inflation layer looks fine but why aspect ratios are that high? How can I reduce this? Is it due to large element size? I have attached the quality check report from ICEM. |
|
April 24, 2020, 05:10 |
|
#2 |
Senior Member
Gert-Jan
Join Date: Oct 2012
Location: Europe
Posts: 1,827
Rep Power: 27 |
Difficult to say. We need geometry and mesh.
Looking at the numbers it are the 2454 elements with very sharp angles (<6°). If you click on the bar of this group in the histogram using your Right mouse button, you can create a subset. Then in the tree on the left hand side in the mesh menu, you can make these visible and it will show you where the problem is. |
|
April 24, 2020, 12:43 |
|
#3 |
Member
Join Date: Mar 2020
Posts: 33
Rep Power: 6 |
I have attached the geometry and mesh images. You can see inflation layer on the surface of blades with zoom in in 3rd picture.
|
|
April 25, 2020, 07:10 |
|
#4 |
Senior Member
Gert-Jan
Join Date: Oct 2012
Location: Europe
Posts: 1,827
Rep Power: 27 |
Well, what do you think yourself?
Think about the volume ratio between the last inflation layer and the first tetrahedral element. This should be closer to 1. You need to increase the number of layers, increase the growth ratio, decrease the tet size. There are so many options. Try out a few to see what it brings. |
|
April 25, 2020, 11:25 |
I can not find the option, could you send me an image?
|
#5 | |
Senior Member
Join Date: Dec 2017
Posts: 384
Rep Power: 9 |
Quote:
|
||
April 26, 2020, 09:44 |
|
#6 |
Senior Member
Gert-Jan
Join Date: Oct 2012
Location: Europe
Posts: 1,827
Rep Power: 27 |
This only works in ICEM to which UsamaQ refers. Do you use that?
|
|
April 26, 2020, 15:09 |
Yes, I use that
|
#7 |
Senior Member
Join Date: Dec 2017
Posts: 384
Rep Power: 9 |
||
April 28, 2020, 17:01 |
|
#8 |
Senior Member
Gert-Jan
Join Date: Oct 2012
Location: Europe
Posts: 1,827
Rep Power: 27 |
You cannot grab this in a picture. It is a procedure which requires a animation. Too much work for this free helpdesk.
So, again, if you plot the quality (edit mesh -> Quality), you get a histogram on the right hand side. If you click on one of the bars of the histogram with your right mouse button, you get the option to create a subset. Do this. Then, in the tree on the left hand side, open the mesh tree, go to subsets, and tick the just created subset to visualise the elements in the subset. |
|
April 28, 2020, 19:22 |
|
#9 |
Member
Join Date: Mar 2020
Posts: 33
Rep Power: 6 |
Hello Gert-Jan.
As mentioned above, to reduce the aspect ratios, I increased number of layers, growth ratio and reduced the tetra size (number of elements increased from 2M to 46M) but still aspect ratio is of the order e8. I don't understand why it is happening. Also, I checked Y plus in post for non inflation case and it is quite high and is in e3 to e4 range. What can I do now? How to improve the mesh further. I am using patch conforming tetra mesh and already sitting at 2M elements. Thank you |
|
April 29, 2020, 02:54 |
|
#10 |
Senior Member
Gert-Jan
Join Date: Oct 2012
Location: Europe
Posts: 1,827
Rep Power: 27 |
In Icem you can locate the nasty elements. Also in Post you can ask for mesh statistics, create variables to plot volume ratios etc.
Based on that you need to refine the mesh locally. And as long as you aren't specific, I can only direct you to general options. E.g., in ICEM you can create a volume density to define local refinement. It is also possible to refine along a curve. Alternatively, split surfaces and add a refinement on the smaller nasty surfaces. |
|
April 29, 2020, 05:57 |
|
#11 |
Member
Join Date: Mar 2020
Posts: 33
Rep Power: 6 |
Thanks for the information. Can I do local refinement by creating a volume density in workbench meshing? Because I only used ICEM for mesh quality report.
|
|
April 29, 2020, 11:56 |
Thank you, got it. Do you know how to use Mesh software?
|
#12 | |
Senior Member
Join Date: Dec 2017
Posts: 384
Rep Power: 9 |
Quote:
|
||
April 29, 2020, 13:45 |
|
#13 |
Senior Member
Gert-Jan
Join Date: Oct 2012
Location: Europe
Posts: 1,827
Rep Power: 27 |
Yes. I know how to use Mesh software.
|
|
April 29, 2020, 13:46 |
|
#14 |
Senior Member
Gert-Jan
Join Date: Oct 2012
Location: Europe
Posts: 1,827
Rep Power: 27 |
Not that I know. You can create a refinement (Insert Sizing in Mesh tree) on a surface or edge and then add a region with constant size (radius of influence, or similar).
|
|
April 29, 2020, 16:59 |
I do this, but I find there is a discontinuous for the inflation and the face mesh c
|
#15 |
Senior Member
Join Date: Dec 2017
Posts: 384
Rep Power: 9 |
I do this, but I find there is a discontinuous for the inflation and the face mesh connected to it, how to decide the curvature normal angle in the mesh size?
|
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Boundary Layer - Inflation Layer | dlkadk | FLUENT | 0 | September 16, 2017 11:11 |
Inflation/Prism layer for rough surface | Dhairya@CFD | ANSYS | 0 | February 23, 2017 00:00 |
[snappyHexMesh] Snapping problem, non orthogonal faces | Zephiro88 | OpenFOAM Meshing & Mesh Conversion | 7 | April 17, 2015 02:56 |
[ANSYS Meshing] Inflation Layer Problem | pyroknife | ANSYS Meshing & Geometry | 0 | May 20, 2013 14:05 |
[ANSYS Meshing] Inflation layer _ Boundary Problem | elina | ANSYS Meshing & Geometry | 4 | March 31, 2013 08:47 |