Showing results for 
Search instead for 
Do you mean 
Solved! Go to solution

Update direction for triangles

Hello all,



I need to update direction of my elements to be along the weld. For quad plate it works fine and for triangles I have an issue. Original direction is 1 - 2, it should be modified to 1 - 3.


I use function Modify -> Update Elements -> Orient Plate Normal/First edge (or Api App.feModifyElemReverse):

Direction is updated when I selected from point 3 to 1 and does not work when first point 1 and then 3 is selected. 


Maybe someone can explain why it happens and how can I be sure I choose correct sequence.




Accepted by topic author yarko
‎02-02-2016 03:41 PM

Re: Update direction for triangles

[ Edited ]



The problem may be that the "First Edge" mode of this command is intended not to reverse the element normal (top / bottom face).  For a triangular element the positive side of the element is defined by the side from which the nodes are viewed anti-clockwise as listed on the Nastran CTRIA entry (editing the element in Femap shows the same listing order) .  This means that if you try to align the first edge in the way you describe, then the element can only use that numbering order if the element normal is reversed  - and for most models it is preferable not to have randomly reversed element normals, in order to have consistent top face or bottom face results.  Thus, that specific command would only allow you to end up with a first edge which goes either from 1 to 2, or 2 to 3, or 3 to 1.  Anything else reverses the top and bottom face of the element, which the "First Edge" mode of the command will not allow.


CQUAD elements also use the anti-clockwise listing / positive side rule, so at least it should be possible for you to end up with both consistent normals and aligned first edges in a mixed mesh of tria and quad elements.


I hope that helps.


Re: Update direction for triangles

Dear EndZ,



Thank you for detailed explanation. Now it's clear. 

We need to change the direction for testing purpose. We have a tool which finds welds on the model and later convert stresses into weld direction.  (Element X along the weld).

The idea was to orient all welds in femap and compare with our results. 



Re: Update direction for triangles



That tool sounds useful for speeding up weld fatigue assessment... does it address specific weld fatigue codes / standards?


Anyway, I believe if you identified a weld line (let's say via a non-manifold curve, like a T-junction), and then identified the surfaces for each distinct "side / zone" of the Tee Junction (tricky, but perhaps via a combination of surface CG location and Direction Cosine of the surface normal), then you could unify the surface normals of the elements on each of the three distinct "sides" which would finally allow you to align the first edge directions of the elements in the 3 zones.

But just a couple of cautions (and I hope I am not stepping on your toes here, as you are clearly a serious expert with Femap!):

(1) the X direction of a Nastran CQUAD element is the diagonal bisection of the node locations.  The first edge will be exactly parallel with X direction of the element for QUADs where (I think this covers it) the 3rd edge is parallel with the 1st edge.  Otherwise, First Edge and Element X cannot be parallel.

(2) For triangular elements you are OK. except for any where a point of the triangle is on your weld line, instead of an edge of the triangle being on the weld line.  Only those CTRIAs with edges on the weld line can have their element X direction aligned with the weld line.  Triangles with only one corner touching the weld line cannot have their X directions aligned with the weld.  Even if their first edge is parallel with the weld line, the X direction will be 180 degrees reversed compared to ones with an edge on the weld line, IF the element normals are all aligned.


Thus, if the objective is to have results parallel and perpendicular to the weld... rather than trying to align the X direction, couldn't you use a results transformation instead?  Then once the element normals are aligned (for consistency with top / bottom results), you can get results for any alignment you want, no matter what the mesh looks like adjacent to the weld.


Lastly, just be careful if you reverse element normals on a non-manifold sheet solid... if any re-meshing occurs, the element normals revert to the top/buttom direction of the underlying surface, whose normal cannot (at current version) be flipped once it belongs to a non-manifold sheet solid.

Re: Update direction for triangles

Dear EndZ,



Thank you very much for your time and sharing your experience. 

I have a lot of experience in Femap Api but in Femap average user (hope so). 


For now we use weld stresses for DIN15018/FEM 1.001 and Eurocode3 standards. Later we plan to cover also DNV RP C203, EN13001.


In our tool we have own algorithm to recognize welds and their direction. According to your comments:

1) This we already know. When we started checking procedure we've noticed difference for quad which are not perfect rectangle. We modified our conversion function to handle this.

2) Weld direction is recognized by 2 nodes of element which is on the weld. Triangles with one node is skipped:

I'm really appreciated for comments, it helped me to make correct verification with Femap.