CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > Software User Forums > ANSYS > ANSYS Meshing & Geometry

[ICEM] Unexplained changes of mesh quality and blocking approach

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

Reply
 
LinkBack Thread Tools Search this Thread Display Modes
Old   November 10, 2016, 09:10
Default Unexplained changes of mesh quality and blocking approach
  #1
New Member
 
Simon
Join Date: Oct 2016
Posts: 16
Rep Power: 7
salumi is on a distinguished road
Hey everyone

My post consists of two main questions.

1) I have created a mesh for a certain design of a gravity separato, as seen in attached pictures. The mesh quality seems relatively okay. Now i need to assign some blocks to a different part, since they belong to the solid reaching into the fluid domain and shouldn't be part of "fluid".
When i assign these blocks to a new part, without changing anything else, the skewness, quality, etc. drop significantly (e.g. skewness from 0.36 to 0.17), which i cannot explain?
Then when i convert either one of the meshes to unstructured mesh, export from ICEM and import in Fluent and check the quality there, it is really bad, so i believe i can't use it.

Fluent quality says:
Mesh Quality:

Minimum Orthogonal Quality = 6.43867e-02
(Orthogonal Quality ranges from 0 to 1, where values close to 0 correspond to low quality.)

Maximum Ortho Skew = 9.35613e-01
(Ortho Skew ranges from 0 to 1, where values close to 1 correspond to low quality.)

Maximum Aspect Ratio = 5.97472e+02


How can this be?
Can anyone explain why this happens? Can i avoid this somehow?
This probably means i cannot use this mesh, which icem told me was okay, right?

I added some pictures and i am adding a link to google drive containing the uns mesh and the msh mesh for fluent. If anyone wants to have a look and needs more files, let me know.
Link:
https://drive.google.com/drive/folde...mc?usp=sharing


2)The second question is about the blocking approach for a different geometry.

I attached a picture so it is easier to see It is basically a pipe junction of different diameters with two small pipes coming in at different locations at the bottom. I have no idea how i shall realize a good joining up of the ogrid that i have to create in the small pipe off the center to the main mesh of the pipe junction.
Any ideas for approach would be appreciated?

Thanks

Let me know if anything more is needed
Attached Images
File Type: jpg mesh before assigned to new part.jpg (97.4 KB, 60 views)
File Type: jpg mesh after assigned to new part.jpg (127.3 KB, 50 views)
File Type: jpg newgeometry.JPG (48.0 KB, 32 views)
salumi is offline   Reply With Quote

Old   November 13, 2016, 11:41
Default
  #2
Senior Member
 
shereez234's Avatar
 
MS
Join Date: Jan 2014
Location: England
Posts: 324
Blog Entries: 1
Rep Power: 11
shereez234 is on a distinguished road
ICEM mesh quality and Fluent quality is a bit different. First of all you have to understand what orthogonality means. A cell which is 90 degrees i.e. |_ has a quality of 1/1 orthogonal quality. A cell which is \__ (45 degrees) have 0.5/1 orthogonal quality. Why I am saying this? Because visualization of your mesh and observing on your own where the orthogonality is bad will be helpful for you to do this. Specially if you have high aspect ratio cells where the Dx length is much greater than Dy length then orthogonality and skew issues arise. Especially if you have a near wall region which is refined and has a very small cell spacing and then you extend it all the way to the far field it will create a lot of orthogonality issues.

Move your edges/ Vertices around and try to make the cells as straight as you can. Get a combination of this:

Far(Sijal's comments on this):
Important thing to remember:

If you are using Fluent then make the mesh which has orthogonal quality greater than 0.01. To ensure this you need to have the

1. Min quality greater than 0.3
2. Angle greater than 18
3. Smooth cell size change


For CFX, you need to ensure this in ICEM:

1. Min quality greater than 0.3
2. Min angle 18
3. Smooth cell size change.

Hope some one helps you out with the blocking alternative strategy.

Cheers
Shereez
shereez234 is offline   Reply With Quote

Old   November 14, 2016, 11:36
Default
  #3
New Member
 
Simon
Join Date: Oct 2016
Posts: 16
Rep Power: 7
salumi is on a distinguished road
Thank you very much for your reply
Unfortunately my blocking is relatively complicated (ICEM says 380 blocks) so it is hard to get the cells any "straighter", i have tried a lot but the current result is the best i could get....


Strangely my mesh fulfills the given criteria for min quality and angle in ICEM, but in Fluent it is still lower. Probably this will have to do with the "smooth cell size change"? Is there a measurement for it?
I think the problem might be that in my geometry the smallest diameter (round inlet orifice) for which i have to resolve the mesh is 1mm, while the biggest one is 40mm. Therefore i always end up with tiny cells in the innermost ogrid when the outer ones are just getting okay in size. That makes a smooth cell size change very hard, as well as the aspect ratio for the small cells in the center is very high (~200).
Cell count is already 850k now.


Quality ICEM:
min. quality: 0.496
min. angle: 29.7 deg
min. equiangle skewness: 0.33
max. aspect ratio: 210

Quality Fluent
Minimum Orthogonal Quality = 8.94068e-02
(Orthogonal Quality ranges from 0 to 1, where values close to 0 correspond to low quality.)

Maximum Ortho Skew = 9.10593e-01
(Ortho Skew ranges from 0 to 1, where values close to 1 correspond to low quality.)

Maximum Aspect Ratio = 2.11326e+02



I attached some pictures of what the mesh and blocking currently looks like.
I think due to the fact with very big changes in diameters of round pipes and additional difficulties because of pipes coming in from the side, it seems like the approach of a structured mesh wont be able to give me satisfactory quality?

Do you think unstructured mesh would be better in this case?

Thank you!

BR,
salumi
Attached Images
File Type: jpg blocking.jpg (88.7 KB, 50 views)
File Type: jpg grid.jpg (143.8 KB, 43 views)
File Type: jpg grid2.jpg (119.1 KB, 36 views)
File Type: jpg grid3.JPG (126.1 KB, 39 views)
salumi is offline   Reply With Quote

Old   November 14, 2016, 12:33
Default
  #4
Senior Member
 
shereez234's Avatar
 
MS
Join Date: Jan 2014
Location: England
Posts: 324
Blog Entries: 1
Rep Power: 11
shereez234 is on a distinguished road
I could take a look at your files I have a free evening today. PM or drop the link
shereez234 is offline   Reply With Quote

Old   November 14, 2016, 16:51
Default
  #5
New Member
 
Simon
Join Date: Oct 2016
Posts: 16
Rep Power: 7
salumi is on a distinguished road
Hey,

that is very kind and would be much appreciated.

I dropped the geometry, blocking, ICEM project, unstruct mesh and fluent mesh here:
https://drive.google.com/drive/folde...mc?usp=sharing
I hope this is working and contains all the files needed.

Let me know what you think

Again, thank you very much!
salumi is offline   Reply With Quote

Old   November 15, 2016, 11:33
Default
  #6
New Member
 
Simon
Join Date: Oct 2016
Posts: 16
Rep Power: 7
salumi is on a distinguished road
Now somebody told me that it might be possible to reduce cell count in the middle area by "paving" these parts while leaving the outside area.
I am not sure how that works in ICEM though.

I tried to change the middle blocks to "swept", which looks kind of what he described now... however, when i change the number of nodes on the middle blocks, it still changes it for all the outer blocks as well....
i attached a picture and added the project file (middleswept) in the drive link.

Any ideas? Is this the right approach? Can i somehow "deconnect" the cellcount/node count in the middle swept blocks from the outer blocks with ogrids?

Getting desperate :/

https://drive.google.com/drive/folde...mc?usp=sharing

Thanks!

BR,
salumi
Attached Images
File Type: jpg middleswept.JPG (165.1 KB, 28 views)
salumi is offline   Reply With Quote

Old   November 17, 2016, 06:51
Default
  #7
Senior Member
 
shereez234's Avatar
 
MS
Join Date: Jan 2014
Location: England
Posts: 324
Blog Entries: 1
Rep Power: 11
shereez234 is on a distinguished road
Extremely sorry for the late reply. I have been busy. But this morning I had a look at your mesh. I found a few problems here and there but over all its a very good blocking approach I must say! Some of the things I changed are:

- The O grid block in the center of the pipe ( just moved the vertices from a diamond shape to a more square shape).

- Some edges did not have any nodes. So I put in atleast 5-6 nodes for them.

- Also your geometry is kind of weak. It's not airtight especially at the connections between two pipes there are holes. You might want to build topology of geometry with a suitable tolerance before you build the mesh if you are going to do it again.

Anyways after these few alterations the mesh Orthogonal quality for fluent is 0.25 and aspect ratio is 211.6. Here is the link:

https://www.dropbox.com/sh/bexh2yf7j...AR5D6xe2a?dl=0

Besides that, I believe that even with Orthogonal quality just above 0.01 with correct numerical setup the solution will converge.

Cheers
Shereez
shereez234 is offline   Reply With Quote

Old   November 17, 2016, 10:43
Default
  #8
New Member
 
Simon
Join Date: Oct 2016
Posts: 16
Rep Power: 7
salumi is on a distinguished road
No worries at all! Thanks for having a look and helping!

Looks good, i'll try and throw it into fluent to see if i can get results to converge.

I wasn't aware of the weak geometry.
Do you know where that originates from?
The part was modelled in solidworks and imported in icem directly from the solidworks part (no conversion)
Is is too late to build the topology now?
Might the weak geometry cause any problems for the simulation?


Do you think the big aspect ratio is not critical?
Would it be possible to change the type of the inner blocks from mapped to swept or mapped with one free face (dont really know what is the difference between the two?) and therefore reduce cell density in the middle? I tried that but somehow it wouldn't work.
With this very small cell size in the middle i feel it would be impossible to have a "smooth change in cell size" like many people say is necessary and the aspect ratio is really high.


Again, thanks very much!
salumi is offline   Reply With Quote

Old   November 22, 2016, 21:30
Default
  #9
Far
Super Moderator
 
Sijal
Join Date: Mar 2009
Location: Islamabad
Posts: 4,552
Blog Entries: 6
Rep Power: 52
Far has a spectacular aura aboutFar has a spectacular aura about
Send a message via Skype™ to Far
did you try to run your case?
Far is offline   Reply With Quote

Old   November 23, 2016, 04:14
Default
  #10
Senior Member
 
Sebastian Engel
Join Date: Jun 2011
Location: Germany
Posts: 566
Rep Power: 19
bluebase will become famous soon enough
Hi salumi,

> I wasn't aware of the weak geometry.
> Do you know where that originates from?
imported geometries have some issues sometimes. I don't know a cad-format which completely avoids conversion problems.

> Is is too late to build the topology now?
> Might the weak geometry cause any problems for the simulation?
Regarding build topology, yes you can use it at any step. Though save the project before you use it. Sometimes you have to play around with the settings to get a satisfying result.
For the structured blocking it is not absolutely necessary to have a water tight geometry. If the elements are bigger than any gap, then you won't notice any problems. However, when you refine your mesh, then gaps could become a problem. In general it's better to have a water tight geometry.
The geometry itself is only used to generate the mesh. A solver will only see the mesh. It it will not know anything about the geometry.

> Do you think the big aspect ratio is not critical?
Big aspect ratios are only a problem if you have strong gradients in the direction of the longest edge. Imagine a boundary layer mesh. In the flow direction, we accept high aspect ratio. But, perpendicular to the wall we have a very high density. So, resolve areas where you expect some changes of the flow or its direction.

> I tried that but somehow it wouldn't work.
It isn't such a bad idea to have an unstructured block in the middle in case the element amount is crucial for your simulation.
I believe there is a tutorial for hybrid meshing coming with the icem help. Have a look at it. For hybrid meshes you need a interface mesh or a matching mesh at each side of the interface. It's described in the tutorial.

> "smooth change in cell size"
Have you tried you way into edge parameters? You can change the bunching law to make a smoother transition. The standard setting for the ratio in each direction is 2. This is way to high, reduce it to 1.2 whereever possible. A second feature which is usefull in this context is the edge matching. It matches the element sizes on the ends of two edges.


When element count is really crucial you could try this mesh approach:
Meshapproach.png
Though it might not yield the highest element qualities.

With regard,
Sebastian
bluebase is offline   Reply With Quote

Reply

Tags
icem, icem to fluent, mesh quality, pipe junction

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
[ICEM] Tips on how to improve the quality of this mesh crimsoncabbages ANSYS Meshing & Geometry 0 September 12, 2014 06:26
[ICEM] Blocking Strategy And Quality FJSJ ANSYS Meshing & Geometry 14 April 30, 2013 11:21
[ICEM] Problem making structured mesh on a surface froztbear ANSYS Meshing & Geometry 4 November 10, 2011 08:52
[ICEM] Problem making structural mesh on a surface froztbear ANSYS Meshing & Geometry 1 November 10, 2011 08:52
[ICEM] Hex refinement negative quality afterwards Anorky ANSYS Meshing & Geometry 8 March 22, 2010 03:25


All times are GMT -4. The time now is 01:42.