high skewness in meshing

  • 198 Views
  • Last Post 4 weeks ago
  • Topic Is Solved
omarjumaah posted this 18 January 2018

 Hello everyone,

I have fluid flow domain consists of 4 parts , the two circles are face split on the disk. I got high skewness on surface (circle). Any hints or suggestions to improve the mesh quality please. Thank you in advance

Order By: Standard | Newest | Votes
peteroznewman posted this 18 January 2018

The best (and most work) to reduce skewness is to split the solids into sweepable bodies. If you want to see how to do that on your geometry, attach a project archive to a post and I will demonstrate. In Meshing, click on Mesh and Clear Generated Data to delete the mesh, save the project, then in Workbench, File, Archive... to make a small file size.  You can attach a .wbpz file to your post as long as it is less than 120 MB.

Here is an example.

 

  • Liked by
  • omarjumaah
omarjumaah posted this 18 January 2018

Thank you peter for replying,

I had hard time to improve the mesh quality. Thank you in advance for any help may you provide,

 

peteroznewman posted this 18 January 2018

I got good at slicing. Your model is now 100% sweepable bodies. I slightly changed your element sizes.

You should have mentioned what version of ANSYS you were using. I did this work in 18.2 and the archive is attached, but I could have done it in 17.2.

Attached Files

  • Liked by
  • omarjumaah
omarjumaah posted this 18 January 2018

Thank you so much dear peter for such a nice effort. I really appreciate your kind help.  Unfortunately I couldn't open the file as I use ansys 18.0. could resend it again please . Thanks

peteroznewman posted this 18 January 2018

Sorry, I don't have 18.0 available. Can you upgrade to 18.2? Please reply on that question first. If you cannot upgrade, I will document the steps I used to achieve this mesh in a video or Powerpoint and reply with that later today. You can show your appreciation by clicking "Like" on the posts that are helpful.

There are only 384 elements out of 520312 that have skewness worse than 0.6 and no elements are worse than a skewness of 0.70

  • Liked by
  • omarjumaah
omarjumaah posted this 18 January 2018

Dear peter, 

thank you so much. I am using ansys 18.0, for now I cannot upgrade that to 18.2. I appreciate if you give me at least the steps to re-mesh the model. Kindly reminder there is a curvature at the corner did not appear in the mesh pic. I am sending you the file again. Thanks

omarjumaah posted this 19 January 2018

Dear Peter,

any update ..

thanks

peteroznewman posted this 19 January 2018

Can you find another computer to install ANSYS 18.2 to examine how I created the high quality mesh above? You may even be able to install it on your current computer and still have ANSYS 18.0 work. I have done this with multiple versions of ANSYS on my desktop, but they were all commercial licensed. I have both ANSYS Student 17.2 and 18.2 on my laptop.

I tried to loft to slice up above the circles thinking would be better than cutting straight up as I did in the first geometry without the blend, but it's not yet good enough.

I was recording a video, then the Cybersecurity software my company uses prevented the mesher from working! 

peteroznewman posted this 19 January 2018

Dear Omar,

You have Extrude 1 (left) that cuts a thin disk off the circular inner face. Why is that a separate solid?  If it is just to achieve a quality inflation layer, that could be accomplished without that thin cut. Please let me know if I can suppress Extrude 1 (right).

               

Since you are using Periodic BCs, why do you have a 90 degree slice with two disks?
Could you have a 45 degree slice with one disk?

  • Liked by
  • omarjumaah
omarjumaah posted this 19 January 2018

Hello peter, 

thank you for replying. Actually I added that cut trying to improve the mesh.

you are absolutely right one disk can work, but I put two disks to check if the there is a difference in results, and also to check disk position related to the results. anyhow this such a good idea I will simulating another model with one disk only.  

here is a simple model without modifications. You can see the mesh to get an idea about each part of the model. Thanks

peteroznewman posted this 19 January 2018

How about the 45 degree slice?

omarjumaah posted this 19 January 2018

So, is it possible to get a high quality mesh with two disks. 

peteroznewman posted this 19 January 2018

If I get a high quality mesh with one disk, then just rotate the solids 45 degrees to get two disks.
I might even take a 22.5 degree slice, get a good mesh on that, then mirror the bodies to get back to 45 degrees.

The point is, you are already assuming a periodic BC over a 90 degree angle, why not a 45 degree angle?

Here is a couple of slices that get the center core sweepable.

 

  • Liked by
  • omarjumaah
peteroznewman posted this 19 January 2018

 Omar,

I have a fairly low skew mesh everywhere except in the center region that needs more work. All bodies need mesh controls to get the element sizes you want. I will explain how to create these shape in DM, but I have a full time job (not at ANSYS) and I volunteer here for fun, so please be patient.

  • Liked by
  • omarjumaah
omarjumaah posted this 19 January 2018

Thank you peter,  

I really appreciate your time and effort

peteroznewman posted this 20 January 2018

Omar, are you using the Student license which is limited to 50 bodies or are you on a full license?

omarjumaah posted this 20 January 2018

I have student license, but I think I can access to ansys with full license at the school lab..

peteroznewman posted this 20 January 2018

Starting with the simple geometry attached to this post.

 

 

 

 

 

  • Liked by
  • omarjumaah
  • Fran
omarjumaah posted this 20 January 2018

Dear Peter,

words cannon express my thanks for all your kindness. I really appreciate your valuable help and effort.

Fran posted this 5 weeks ago

Very good videos Peter, they are very helpful

Thank you very much!

Regards

Fran

omarjumaah posted this 5 weeks ago

Dear peter,

thank you for your help. unfortunately I did not get as your results. 

- Can the version of ansys affect the results  also,

-should I share the topology between the parts to avoid connections in meshing. Thanks

 

peteroznewman posted this 5 weeks ago

Dear Omar,

Did you see video #5 above?  I found there was a defect after video #4.

Yes, each version of ANSYS has software changes that affect meshing results.

When all the bodies are put in a single part in DM, share topology is automatically used.

Here is another video on slicing geometry in DM.

omarjumaah posted this 5 weeks ago

Hello Peter,,

thank you for replying. yes I saw it many times. I followed it literally, but I got poor mesh. I could not figure out my mistakes. here is the file, I hope you can get a look. Thanks 

peteroznewman posted this 5 weeks ago

 Good Morning Omar,

This video shows how to create a Multibody part, which is required. And shows the first attempt at meshing. I'm working on Part 2.

 

 

  • Liked by
  • omarjumaah
peteroznewman posted this 5 weeks ago

Topology problems being exposed in these videos. 

 

 

  • Liked by
  • omarjumaah
peteroznewman posted this 4 weeks ago

Omar, have you considered using a tet mesh with inflation? I just learned this by reading the ANSYS Help Viewer.

If in DM you have a named selection for the inlet and the outlet and the periodic boundaries, all the other outside faces will be treated as walls and will get an inflation layer before it starts filling the volume if you set Use Automatic Inflation to Program Controlled.

If you do, you can achieve your desired skewness quality target of < 0.7 with no effort!

Mesh metrics are stable at a 2 mm Max Face Size...

 

...and at a 4.5 mm face size.

I wonder what the CFD experts have to say on tet vs hex meshing?

  • Liked by
  • omarjumaah
omarjumaah posted this 4 weeks ago

Dear Peter,

thank you of considering my problem. I learned from your posts a lot, I appreciate that. It looks very good idea to include the inflation with the mesh. Actually, I tried this kind of mesh without inflation, unfortunately I got difficulty in convergence of continuity equation. And also, according to some articles is recommended to simulate deposition thin film on the surface using structure mesh. Anyhow, I updated the mesh depends on your suggestions and here is the mesh results. I just start the run in fluent and it takes couple hours to finish. I will inform you if there is a progress. Meanwhile, I am still working to improve the previous model with structure mesh. Thanks   

Attached Files

peteroznewman posted this 4 weeks ago

Dear Omar,

If you are making a Periodic BC assumption, why not just build an Axisymmetric model?  That way you can get an awesome structured mesh with an excellent inflation layer and amazingly low values of skewness. This model will solve so fast, it is ideal for optimization! And when the optimizer changes the parameters in the shape of the fluid domain, there are no slices to break and cause the whole optimization process to grind to a halt. Think about the benefits.

                                                                 

 

I did this in ANSYS 17.2 so you can open the archive and try it out.

Attached Files

  • Liked by
  • omarjumaah
omarjumaah posted this 4 weeks ago

Dear Peter,

thank you for your efforts, I really appreciate that. My point is to get the result using 3D geometry. I tried the one part with tetrahedral mesh  and inflation. Also I modified the mesh depends on your discussion, I didn't get good results. As the continuity equation didn't converge more than 3e-03 after 300, then it rises up. I'll try to check the BCs. Thanks

Attached Files

Close