Diplay template... Contour or Element? 1

[umit]

Hello all,

I did an FEA on a vehicle suspension part using Ideas 9. When I postproces the results. There are couple of options in the display template menu. I first selected the contour and I got about 10 ksi maximum stress and 1.19E-2 in max deflection.
Then I selected the element on the display template menu and corresponding results were 22 ksi and 1.19E-2 in.

What is the difference between the contour and element display. Does contour display average the stress values?

Any help will be appreciated.

Thank you.

umit

 

[corus]

If you look at the picture it produces you may see that the contour values for the element are in block colours to represent the stresses at the element gauss point. Stresses at the nodes are an average of these and produce smooth contours.

corus

 

[umit]

Corus,
thank you for your reply.
If the contour shows the average stresses at the nodes.
What does element display show? The actual stresses on the elements?

Thanks.

Umit

 

[corus]

Stresses for the elements will be those calculated at the gauss points. For a linear element (four noded or 3 noded triangle) there is just a single gauss point.

A measure of the accuracy of the solution, and how fine your mesh needs to be, can be made by comparing adjacent element stresses and the element nodal stresses before averaging. For a linear element the stresses are constant across the element and hence the element nodal stresses are the element stress.

corus

 

[umit]

Corus,

How close should be the stresses at the adjacent elements?
( 75 percent, 80 percent)?

Thank you for your reponse.

umit

 

[MToft]

Hi,
I generally reckon that a minimum accuracy level is with two Gauss points within 10% of each other at the stress peak (eg 300 and 330MPa) - ideally agreeing to within 1% (or 2 significant figures eg 300 and 304MPa). In these cases the element averaged ("contour") value should be pretty close to the raw Gauss point data.

A further guide can be that a 10-colour scale should probably not have more than two colours in an second-order element of interest, since this means the stress is changing within that element by more than 10% of the overall stress range, which is probably too much for accuracy.

For information, Abaqus CAE has a slider which can adjust the averaging - VERY dangerous; I've seen it used to "pass" components!

The most important thing is probably consistency - you can often get away with much worse for a purely comparative analysis.

Good luck!