I have an open question regarding RSS combining of VonMises stress.
I don't understand how FEMAP's approach can make sense mathematically/physically.
When combining complete outputsets FEMAP computes the RSS of individual components, then computes VonMises of these RSS components.
This seems false, the counter example is tri-axial stress: if you pull a brick along X, Y and Z independantly, then RSS combine these 3 loadcases, FEMAP will give VonMises = 0!
I am biaised by the fact that I only use RSS combinations in probabilistic terms: RSS combining for me is a way to say "I have a series of unit equiprobable and independant loadcases, I'm not exactly sure what combination I'll get but the RSS is mathematicallythe most representative without being too pessimistic". So in the example above, I would expect to get Svm*sqrt(3) if Svm if the vonMises Stress for each subcase. Not 0!
My questions are:
- what is the mathematical justification for FEMAP's approach when doing RSS combination of VonMises and principal stress?
For linear operations like sum however I perfectly agree that FEMAP operates on individual components, then recomputes VonMises. That is correct. But not for RSS.
- does anyone use this functionnality and this particular approach in mechanical calculations?
Thanks for your help,
Even though the answer seems implausible, I believe it is correct for the special case you describe. If you inspect the equation for Von mises Stress
then I think you can see how this happens. We have the same stress in each direction and zero shear, so the answer really is zero for this case. I have attached the test model that I used. It has the individual load/constraint cases set up so you can use Femap to RSS, and it also has a combined case where I let Nastran combine the 3 loads and you can see Nastran also gets zero for the von mises stress in this special case. If you have a different boundary conditions than what I used, please send the model or describe in more detail.
Nice timing, I was writing a response as I recieved you response.
Yes I agree with you, tri-axial stress gives a 0 VonMises stress - and I had the same test model than you.
But unless I'm mistaken in your test file you don't ask NASTRAN for the RSS: you ask for the SUM via a LOAD card, and this bounces back to what I said earlier:
- for SUM it is correct to sum individual components, then recompute vonmises on these combined components. It is the ONLY physical/mathematically correct method. This is what you do in your test model
- but for RSS it seems completely false to me. To me RSS combining is purely a probabilistic approach, it makes no physical sense, therefore the resulting combined individual components make no sense and neither does computing a criteria using them.
- however doing the RSS of the criteria makes sense. (this is the response i was writing). The correct operation is Model > Output > Process > RSS > One or More Selected Vectors + Combine Each Vector in All Sets. On the tri-axial stress this will give you a non-zero result
To sum up there are 2 methods in FEMAP for RSS combining VonMises stress:
- RSS combine entire sets => FEMAP RSS combines individual components and recomputes VonMises
- RSS combine each vector in all sets => FEMAP sees VonMises as a scalar, and directly RSS combines VonMises stress
In my opinion Method 1 should be banished! Non advanced users can very easily make the mistake (because Method 1 is the easiest to do in FEMAP)
But that is where I am limited by my own field, perhaps other people have cases where this method makes sense and I would be interested to learn about it.
Thanks for your interest!
I am now dealing with RSS and you are right . I do not know which use the method has.
Normally if you have two load cases the procedure to get, let's say, von mises is to combine the stress tensors and then get von mises. Here RSS ,as it is by default, does the combination of von mises of each load case and get von mises total=(von mises1^2+von mises^2)^0.5.
(I think I did it right in FEMAP.)