CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > Main CFD Forum

Negative Jacobian

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

Reply
 
LinkBack Thread Tools Display Modes
Old   July 27, 2006, 16:31
Default Negative Jacobian
  #1
PC
Guest
 
Posts: n/a
Hi,

I wonder if you can help me here. I have created a body-fitted mesh inside a circular nozzle where I have some negative jacobians, I dont undertand how I can remove these negative jacobian. These happens if I increase the mesh points.

Thanks in advance
  Reply With Quote

Old   July 27, 2006, 16:42
Default Re: Negative Jacobian
  #2
Renato.
Guest
 
Posts: n/a
You can work out it by changing the nodal connectivity of the negative elements, e.g.:

A triangle with nodes i, j, k can be described by:

i-j-k or i-k-j

one will have positive Jacobian while the other will have negative. You only need to set the nodal mapping accordingly. In finite element method this ordering is usually counter-clockwise for positive Jacobians.

ps: It can also be due distorted/invalid elements

Cheers

Renato.

  Reply With Quote

Old   July 28, 2006, 01:19
Default Re: Negative Jacobian
  #3
Dr.Nick
Guest
 
Posts: n/a
Renato's right. it depends on the vector product and normal, i.e. say Normal=(V1xV2), angle(Normal,i)=+R1. Hence, Normal=(V2xV1), angle(Normal,i)=-R1; where i={1,0,0}
  Reply With Quote

Old   July 28, 2006, 08:59
Default Re: Negative Jacobian
  #4
Arambatzis George
Guest
 
Posts: n/a
I am also writing a 3D code using tetrahedrals. My problem is that some elements have negative Jacobian.

Have you any ideas to solve this prolem? Or any books or lecture notes or links?

  Reply With Quote

Old   July 28, 2006, 10:16
Default Re: Negative Jacobian
  #5
Renato.
Guest
 
Posts: n/a
Your problem can be solved in the same way described previously. For a tetrahedra formed by i-j-k-l nodes you might have the following conectivities:

(assuming FEM standard mapping) I-J-K-L = positive Jacobian I-K-J-L = negative Jacobian

You can use the following piece of Fortran code to compute the element volume (x6) V6 = x21*(z31*y14-y31*z14) + x31*(z12*y14-y12*z14) + x41*(z12*y31-y12*z31)

where: xij = xi - xj (regarding the nodal coordinates)

Regards

Renato.

  Reply With Quote

Old   July 28, 2006, 10:31
Default Re: Negative Jacobian (NOTE)
  #6
Renato.
Guest
 
Posts: n/a
I should have also emphasized that negative Jacobians can be due to distorted and invalid elements!!! In this cases, you can't do anything else remeshing your model or fixing the bad elements locally (usually made by hand, picking and moving nodes). Another possibility can be some float point problem related to nodal coordinates. You can always be facing the range the machine number representation -- usually IEEE for PCs (*). In this case, changing the scale of your model can helpful.

Mesh problems usually happen when you import/export your model through several softwares due to loss of accuracy when representing the model entities (nodes, edges, faces, etc...).

Regards

Renato.

(*) (unfortunately or fortunately) computers have limited floating point representation.

  Reply With Quote

Old   July 28, 2006, 12:46
Default Re: Negative Jacobian (NOTE)
  #7
Mani
Guest
 
Posts: n/a
if you get negative Jacobians in "some" cells of your structured grid, it will hardly be a problem of coordinate orientation.

the fact that it happens only with fine grids suggests small inaccuracies in your grid that lead to cross-overs at small cells, maybe related to floating point precision, or simply incomplete convergence of your grid generator. you should inspect the grid.
  Reply With Quote

Old   July 30, 2006, 05:44
Default Re: Negative Jacobian (NOTE)
  #8
R Andom.
Guest
 
Posts: n/a
If all your jacobians are negative, then you are fine also.
  Reply With Quote

Old   January 22, 2012, 15:34
Default
  #9
New Member
 
Join Date: Jan 2012
Posts: 2
Rep Power: 0
Vijayendra is on a distinguished road
Quote:
Originally Posted by Renato.
;44158
You can work out it by changing the nodal connectivity of the negative elements, e.g.:

A triangle with nodes i, j, k can be described by:

i-j-k or i-k-j

one will have positive Jacobian while the other will have negative. You only need to set the nodal mapping accordingly. In finite element method this ordering is usually counter-clockwise for positive Jacobians.

ps: It can also be due distorted/invalid elements

Cheers

Renato.
Hello,
I found this message regarding the naegative Jacobian helpful as I also have a similar problem. I have the following question regarding the triangular elements:
Considering i-j k if I get negative jacobian, I considered i-k-j. In most of the cases it did work. But in few cases, even this did not work. In that case, I just permutated like:
1-2-3
1-3-2
2-1-3
2-3-1
3-1-2
3-2-1
one of the permutation did give a positive jacobian. Can I consider this permutation and proceed with the further calculation?
Any clarification will be highly appreciated.
Thanking you in advance
Vijayendra
Vijayendra is offline   Reply With Quote

Reply

Thread Tools
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 On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
negative element volume (CFX-10.0) CFDworker CFX 8 September 27, 2011 18:16
channelFoam for a 3D pipe AlmostSurelyRob OpenFOAM 3 June 24, 2011 13:06
blockMesh error ... balkrishna OpenFOAM Pre-Processing 0 August 17, 2010 02:39
BlockMesh FOAM warning gaottino OpenFOAM Native Meshers: blockMesh 7 July 19, 2010 14:11
Axisymmetrical mesh Rasmus Gjesing (Gjesing) OpenFOAM Native Meshers: blockMesh 10 April 2, 2007 14:00


All times are GMT -4. The time now is 06:59.