Material testing validation

[wort]

Hi,

When modelling a material testing in ABAQUS, the mesh is obviously of great importance. I want to model a specimen of the size 1x1x10 cm, and I tried with two meshes, according to the picture (the first one with 100 elements lengthwise and the second one with 500 elements). The results differ significantly - and the specimen even bends in different directions in the two cases.

The material model is a J2 pasticity model. The top surface is restricted in z-direction, and a point on the top surface is chosen to be restricted in the xy-directions. The bottom surface is then displaced 3 cm upwards.

Since I'm new to FE-modeling, does anybody have any comments on this behavior? How should I proceed in order to get a valid model?

Thanks,
wort
http://indiewanttobe.angelfire.com/abaqus/abaqus.GIF

[wort]

Sorry, if the image doesn't show, it can be found on this url:

http://indiewanttobe.angelfire.com/abaqus/abaqus.GIF

[Serchu]

Well, to me the two images shows almost the same thing! I don't know if you can be sure if the different values are really important, is a rubber part, so the tolerances normally are very high.

See the stress, for the minimun is 2 against 1.93, these numbers are very similar. On the other way, for the maximun values you have 4.45 against 3.75, here you have something. Normally if you decrease the size of the elements, stress should be higher, but this is no the case in your example. I can see that in the second picture the part is rotating a little bit, maybe you had selected another point (node) to fix in X-Y thus giving you some misalignament between load and restriction.

[wort]

Thanks Serchu for the comments.

The node fixed in xy is the same for the two analyses - the only thing I changed was the mesh. I actually expected the fine mesh to be softer than the coarse one (as opposed to what you suggested) but I didn't think that the ratio would be as big as 4.45/3.75.

I used explicit integration. Could this be of importance?

[ashu28]

Hello,
If it doesnt take too much time, why dont you go ahead and further increase the mesh refinement. Then you may compare whether you are getting results similar to your 2nd simulation. Lets see how your results vary then.

/Ashu