Order of nodes for elements

[ckay]

Hi All,

When exporting my mesh from GiD into a tcl file, I found out that the order of nodes for forming the elements is different within the same model. E.g., sometimes numbering starts at the top right corner for some elements and for other elements in the same model at the bottom right corner. They are all in counter-clockwise order though.
I run a quick check on a smaller model with SSPquad elements and found that the above does seem to have an effect on the results. While the difference is very minor for the case I checked, I was wondering if you could help with the following two questions:

1. How important it is that the numbering follows the same pattern for all elements, e.g. all start cumbering from bottom right node.
2. If it does matter, is this particular for the SSPquad and what is the reasoning behind it?

Looking forward to hear from you.
Kind regards,
Ckay

[fmk]

1. theoertically it should not matter, numerically it may. This is because internally the element determines constitution matrix and stress and strains does by looping over nodes and gauss points individually doing stuff like forming jacobians, .... . if computer worked in exact arithmetic this order would not matter, but computers don't and roundoff happens and so order can matter. it usually doesn't or differences are so small as not noticable, however situations do arise where it can matter. i don't know of any program that internally would worry about these numerical issues, you could though rewrite an element that would be aware of such things, i.e. it might form the resiuaul contribution from each gauss pt and add them from smallest to largest based on norm, .....