cosmosexpress problem

[vscid]

I was trying to do basic stress analysis with cosmosexpress.
I realize that it is flawed.
I tried to apply a 2 N load and then later a 700 N load to my object,
tried to compare the deformation models with the 2 forces.
It showed the same deformation model for both,even though the von mises stresses were
considerably different.
Am I missing something here?

Sid

 

[GBor]

Yes...CosmosWorks, or CosmosProfessional...

 

[vscid]

Uh i dint exactly understand your reply.....

 

[MrMyers]

I am guessing there is some automatic magnification factor being applied to your deformed shape for visualisation purposes, such that they look the same in your post processor.

You will likely get the results you want, if you find how to manually set this to a number you specify as the same for each model.

Check the difference you should expect for the field-output displacement results.

 

[cbrn]

Hi,
vscid, depending what you are trying to do, Cosmosexpress works exactly as it should.
CEx DOES have an "auto-rescale" for the deformed image, which you can adjust by right-clicking on the plot object in the tree and selecting "properties" (or whatever it is, I haven't been seeing a CEx since 2004...).
Do always look at the NUMERICAL values in the scales: FEM is not a matter of nice plots, neither in the very simple cases allowed by CEx... And also make sure you don't confuse the "strain" plot with the "displacements" plot (in some languages, they can be called the same way, for ex in Italian where in v.2004 the weird translation "deformazione" was used for both!)

Regards

 

[vscid]

dear cbrn

Can you tell me where that option to change the scale is?
I cant seem to find it.

 

[GBor]

vscid,

Sorry, I was being rude. Are you missing something? Yes, a real FEA pacakge, which COSMOS does sell, just not in their "Express package". There have been plenty of rants about COSMOS Express, so I won't add to them here. Good Luck!

 

[cbrn]

Hi,
I can't remember it right now. I'll try to figure out but things may have changed a lot since when I last used Cosmos family of products.
I understand GBor's point but I have to partially disagree: IF the case analyzed is within the appliability of CEx, and IF the analysis is set up properly, then the results will be nothing different as those obtained with CosmosWorks (the solver is the same).
Of course, the key of the problem is to know whether you are still in the appliability range of CEx or not...

Instruments like CEx are "sold" (it's a word misuse, since you don't have to pay anything for CEx...) by the Salesmen as easy and quick ways for the designer to check his work, but in reality the simple knowledge of where you must stop using them (i.e. the knowledge of the appliability range) is not a simple matter.

Regards

 

[cbrn]

Hi,
I checked at home with the old v.2004 which is still "alive" though definitively "sleeping". You're right, there is no such option as to control the scales. In fact, the deformation plot is really a "displaced shape" plot and not a deformation plot, so whatever the load value it will always display EXACTLY identical!
But don't get despaired... There's a trick...

Look into the directory where the CEx results are stored. You will find a file with extension ".out". This is exactly a -- numerical -- results report, where the values of stress AND deformation (as well as reaction forces, which also are extremely important in my opinion... ;-) ) are listed, together with the coordinates of their maxima. It's really rough as an output, but sufficient (or maybe not...) for what CEx has been thought for...

Regards

 

[vscid]

Thanks cbrn.
I did find the .out file.

 

[FEAHappy]

It is important to check a FEA solution in several ways.
Does the deflection look correct to you?
Do the reaction forces make sense?
Do the stress values change with a finer mesh?
Is there a smooth gradient in the stresses?

The difficulty in using xpress is that the tools needed to interrogate the solution are not there.

Having no way to check the solution, there is no way to
believe the results.