Results in Hyperview

[oppera]

Hi guys, I've done a static analysis of a mechanical component but when I displayed results (VonMises) I've noted this:

at left all elements are in the same component, at right elements are in two different component

at left seems to be continous material while at right seems to be discontinuity... but are the same model!!

I've meshed dragging 2D element on the tube and the plate, after nodes of solid elements are equivalenced and have the same property and mat (only different components: elements of tube in one and the plate in other)

I've analyzed using Nastran and Radioss and the results are the same..
Naturally I thing that the right result is on the left
Where I'd wrong? How it's possible this?

Thanks in advance

 

[PaulRomanos]

Hello,

For me it's a matter of post-processing related to stress averaging.

It seems in the model on the left the stress is averaged at the 2 components interface hence the smoother result while on the right it isn't hence the discontinuity at the common edge between the tube and the plate. This explains the higher value for stress in the latter case.

Usually in every post-processor you have different parameters and criteria for specifying how and where the stress is averaged (through different materials, different properties, different element sets,above certain feature angle, etc...)

If you want to obtain the same result on the left in the 2nd model, you need to enable averaging at elements boundaries between the tube and the plate. I know this can be done in Abaqus Viewer in Results -> Options -> Averaging. I hope someone could provide the equivalent in Hyperview.

As for the validity of either result, it depends on how the real parts are assembled. It is intuitive to think that smoother results are more accurate however this is not always the case.
If the stress discontinuity between the tube and the plate is real (i.e. welding, sharp 90 deg angle, etc...) consider non-averaged results (right-side results). If the 2 components form a single part with a chamfer or fillet that you simplified in the FE model than it's safe to consider the averaged results (left-side results).

As a rule of thumb, non-averaged stress results are always higher and hence more conservative.

Regards,
Paul