one element different

[navigator30]

Hi guys,

I've got a question on something I noticed after doing a calculation. When I look at the results in the contour plot at one location one element displays entirely different than the surrounding ones.

To make myself clearer: I have calculated a contact problem with friction. Then the stresses are displayed and at one spot there is an element that is for example blue (e.g. stress=10N/mm²) while the surrounding elements are red (e.g. stress=20N/mm²). How can that "jump" happen? Do I have to mesh my model finer?

Hope for an answer!
Adios!

 

[corus]

Some kind of averageing method is used which you can alter. By default it's set at 75% but you can alter that to remove these apparent discontinuities. What the percentage refers to I'm not sure.

corus

 

[brep]

With the default averaging scheme the results are not being averaged across between certain elements when the discontiuity is above a certain threshold (75% is the default). Therefore you see a jump in results.

There are many options for postprocessing in Viewer that you should explore. Have a look at the menu Results - Options (There's also an toolbox icon for this). You'll see you can turn off averaging, control the different regions used in the averaging and so on. There also an alternative whereby the order of compution of the scalars (before or after averaging) can be set. This will give people an option to make results look more like ABAQUS/POST (!). BTW if there are significant differences in the unaveraged, /POST-like and default /Viewer schemes then you've likely got a mesh that is not refined enough. There is no right or wrong in averaging schemes - it is, after all, averaging!

This is yet another compelling reason to look at the documentation. It is very complete, and will save you from having to say "I'm not sure". I don't believe non-linear FE analysts should ever be satisfied with an "I'm not sure"

 

[corus]

Once upon a time you asked yourself the same question and read the book. However, once satisfying yourself with the explanation these matters tend to leave the mind until someone, someday, asks that same question, which rekindles that old memory. Unfortunately as the days pass and the memory gets filled with all kinds of junk, the exact transcript from the book eludes you and you can only give a partial, but hopefully helpful answer. Why I bother answering these questions "I'm not sure". Clearly the answer to every questioner is 'go read the book'.

corus

 

[navigator30]

well thanx for your answers everybody!

 

[navigator30]

so corus Ihave read the "book" og abaqus documentation a couple of times now, but haven't quite gotten it wether to use hybrid elements for an elastomeric material or rather reduced integration elements. The material is subjected to shear forces, would that do some hourglassing effect, or does hourglassing only refer to bending shear??????

I can' get that quite clear.

Hope you can answer me that one too.

navigator