[ICEM] Prism Quality mesh problem

bharath · April 9, 2014, 01:27

Hi to all,

I am doing a tet/prism mesh of a simple geometry with octree -> delauny -> Prism approach.

I had two cases of first cell height.

1) 1e-6 [m] as initial height(First cell height), i got good quality mesh after smoothing and other standard methods.

2) 1e-7 [m] as inital height and i am not able to get good quality (> 0.05) after so many trials (like one layer approch,Redistribute prism , combination of multilayer split and redistribute).

Is this because of initial height (1e-7 [m]) is too small? Could anybody suggest some methods to get good quality.(Tried simon's prism mesh presentation also)

thanks in advance

praveen@cfd-online.com · April 9, 2014, 02:24

Hi Bharath,

I guess the issue is with the units tolerance, as the fist cell height is very low (more than 6 digits after decimal point).

Please scale the geometry to mm and then create the mesh. In that case, your first cell height becomes 1e-4mm in place of 1e-7m. Later when you take this mesh to FLUENT you can scale the mesh saying 'Mesh was created in mm'

Happy Learning!!!
Praveen Kumar R

bharath · April 9, 2014, 02:34

Hi Praveen,

Thanks for your reply.

But i already doing the same what you told about scaling however it is not a success yet.

what else i can do?

praveen@cfd-online.com · April 9, 2014, 02:55

OK.

As you said, you have tried one-layer approach, What is the min. quality of mesh with one layer?

I am assuming the single layer height you used is the total height of prism layers. Typically we calculate the total height based on the initial height, growth ratio and number of layers. Once a single prism layer is created (with the calculated total height as the initial height) we smooth the mesh to a good quality and then split the prism in to number of layers, with desired initial height and growth ratio.

I am interested to know, whether the quality issue was also in the single prism layer too.

Happy Learning!!!
Praveen Kumar R

bharath · April 9, 2014, 05:05

Yes Praveen.I got bad quality prism with single layer approach also (Total layer height = 6 mm,Tetsize = 8 mm).After split it throws bad quality prism.

With single layer i got 0.03 as prism quality and after smoothing it becomes 0.07.

Then i tried to make 21 layer split with initial height (0.000162 mm).This leads to 2e-5 as quality.Then, i started smoothing prism it is not helpful as the quality remains same after no of trials.

kad · April 14, 2014, 09:22

Maybe this one http://www.cfd-online.com/Forums/ans...lity-hexa.html can help you as well.

April 9, 2014, 01:27	Prism Quality mesh problem	#1
bharath Senior Member Bharath kumar Join Date: Apr 2009 Posts: 169 Rep Power: 17	Hi to all, I am doing a tet/prism mesh of a simple geometry with octree -> delauny -> Prism approach. I had two cases of first cell height. 1) 1e-6 [m] as initial height(First cell height), i got good quality mesh after smoothing and other standard methods. 2) 1e-7 [m] as inital height and i am not able to get good quality (> 0.05) after so many trials (like one layer approch,Redistribute prism , combination of multilayer split and redistribute). Is this because of initial height (1e-7 [m]) is too small? Could anybody suggest some methods to get good quality.(Tried simon's prism mesh presentation also) thanks in advance

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[ICEM] Low quality Tet prism mesh	saisanthoshm88	ANSYS Meshing & Geometry	17	September 28, 2016 16:50
[ICEM] Negative volume error in hybrid mesh	siw	ANSYS Meshing & Geometry	4	September 3, 2014 05:25
[Gmsh] 2D Mesh Generation Tutorial for GMSH	aeroslacker	OpenFOAM Meshing & Mesh Conversion	12	January 19, 2012 03:52
[ICEM] Tetra mesh quality before and after prism layer	Chander	ANSYS Meshing & Geometry	0	December 25, 2011 22:04
ICEM - Prism Tool Problem	carpe85	CFX	0	February 10, 2009 12:25

April 9, 2014, 02:24		#2
praveen@cfd-online.com New Member Praveen Kumar R Join Date: Apr 2014 Location: Pune, India Posts: 19 Rep Power: 12	Hi Bharath, I guess the issue is with the units tolerance, as the fist cell height is very low (more than 6 digits after decimal point). Please scale the geometry to mm and then create the mesh. In that case, your first cell height becomes 1e-4mm in place of 1e-7m. Later when you take this mesh to FLUENT you can scale the mesh saying 'Mesh was created in mm' Happy Learning!!! Praveen Kumar R

April 9, 2014, 02:34		#3
bharath Senior Member Bharath kumar Join Date: Apr 2009 Posts: 169 Rep Power: 17	Hi Praveen, Thanks for your reply. But i already doing the same what you told about scaling however it is not a success yet. what else i can do?

April 9, 2014, 02:55		#4
praveen@cfd-online.com New Member Praveen Kumar R Join Date: Apr 2014 Location: Pune, India Posts: 19 Rep Power: 12	OK. As you said, you have tried one-layer approach, What is the min. quality of mesh with one layer? I am assuming the single layer height you used is the total height of prism layers. Typically we calculate the total height based on the initial height, growth ratio and number of layers. Once a single prism layer is created (with the calculated total height as the initial height) we smooth the mesh to a good quality and then split the prism in to number of layers, with desired initial height and growth ratio. I am interested to know, whether the quality issue was also in the single prism layer too. Happy Learning!!! Praveen Kumar R

April 9, 2014, 05:05		#5
bharath Senior Member Bharath kumar Join Date: Apr 2009 Posts: 169 Rep Power: 17	Yes Praveen.I got bad quality prism with single layer approach also (Total layer height = 6 mm,Tetsize = 8 mm).After split it throws bad quality prism. With single layer i got 0.03 as prism quality and after smoothing it becomes 0.07. Then i tried to make 21 layer split with initial height (0.000162 mm).This leads to 2e-5 as quality.Then, i started smoothing prism it is not helpful as the quality remains same after no of trials.

April 14, 2014, 09:22		#6
kad Senior Member Join Date: Feb 2013 Location: Germany Posts: 200 Rep Power: 24	Maybe this one http://www.cfd-online.com/Forums/ans...lity-hexa.html can help you as well.