GiD forum

Hi,

I have a variable (SDV2G) that is calculated at Gauss point level and its value is between 0 and 1. GiD extrapolates from Gauss point to nodes but when I visualise it, there are some elements that are greater than one or even negative, see the attached image. Post.res file has the following structure:

------------------------------------
GiD Post Results File 1.0

GaussPoints GP_HEX08 Elemtype Hexahedra
Number of Gauss Points: 8
Natural Coordinates: given
-0.57735026918962595 -0.57735026918962595 -0.57735026918962595
0.57735026918962595 -0.57735026918962595 -0.57735026918962595
0.57735026918962595 0.57735026918962595 -0.57735026918962595
-0.57735026918962595 0.57735026918962595 -0.57735026918962595
-0.57735026918962595 -0.57735026918962595 0.57735026918962595
0.57735026918962595 -0.57735026918962595 0.57735026918962595
0.57735026918962595 0.57735026918962595 0.57735026918962595
-0.57735026918962595 0.57735026918962595 0.57735026918962595
End GaussPoints

Result SDV2G ALYA 0.30000000E+02 Scalar OnGaussPoints GP_HEX08
Values
element1 value_gp1
value_gp2
value_gp3
value_gp4
value_gp5
value_gp6
value_gp7
value_gp8
element2 value_gp1
....
...
...
End Values

---------------------

Thank you very much,

Gerard

It is normal that extrapolating the values from 'gauss points' to the corner nodes its value is bigger or lower that the range of values on the inner gauss points.

for example imagine a 2D line element of 2 nodes (linear line element), using two equi-spaced gauss points for example with values 0.5 and 1.0 extrapolating them linearly to the corners will have values 0.0 and 1.5 (outside the range [0.5-1.0])

on the other hand maybe in your case the problem is another. attach your full case (mesh + results)

It seems that your are using hexahedra of 8 nodes, and writing results on 8 gauss points, then we are extrapolating to corners using the 8-noded hexahedra shape functions (bilinear). Basically we solve a linear system that has 8 unknown values determined forcing to interpolate your 8 gauss point values.

Hi escolano,

Now I see, you are completely right. Find attached the files of this example.

I have another doubt, i've seen that you can do a "smooth". If i understand correctly this smooth consist of a kind of averaging. I've checked on the internet and there are different ways to do that:

(I) Unweighted averaging: assign same weight to all elements that meet at a node;
(II) Weighted averaging: the weight assigned to element contributions depends on the stress
component and the element geometry and possibly the element type.

What kind of averaging Gid does on that case?

Thank you very much,

Gerard

For 'smoothed' result nowadays GiD is doing (I), unweighted average.
If do you want other kind of weightings (e.g. averaged with the 'volume' of the element) could directly calculate in your program and write a 'nodal' result instead of a dicontinuous elemental,'on gauss points', result.

About your example, a comment,
your 'user defined' gauss point is this
It exactly match our 'GiD internale definition' (the same location and in the same order as our 8-noded hexahedra gauss points,
then it is not necessary to explictily define the natural coordinates location. you can use the word
Natural Coordinates: Internal
and also more, you can completely avoid define this block GaussPoints/End GaussPoints in the results file, and user our keyword gauss point name: GP_HEXAHEDRA_8, that is implictly defined.

Result SDV2G ALYA 0.30000000E+02 Scalar OnGaussPoints GP_HEX08
->
Result SDV2G ALYA 0.30000000E+02 Scalar OnGaussPoints GP_HEXAHEDRA_8

Great! Thanks a lot for your helpful comments and corrections.