Quantifying plasticity

[trainguy]

All,

When designing with FEA to a requirement of "no plastic deformation" or "no yielding" under proof, one time loading, how would you utilize the results of a nonlinear static analysis?

Would you look at plastic strains and keep them under 0.2%? That's what I usually do when working with steel, for example.

I have come across some colleagues who look at stress distributions after such analyses and they identify that most of the hot spots show stresses less than yield, without examining strains. Is this valid?

Is there some accepted quantified value of plastic strain that is typically allowed?

tg

 

[rb1957]

if your requirement is zero plasticity then a linear analysis will be sufficient.

if you are referring to FAR25.305 the requirement is no "detrimental" yielding at limit (not quite the same as no yielding at limit).

if you are working to a company/customer spec, and you have any sort of holes (rivets, bolts) in your structure then you'll be limited to something like yield/3 ("something like" 'cause the Kt for a filled hole is less than 3 (for rivets) but an open hole (screws, bolts) is of course 3).

 

[trainguy]

rb,

This is for non-aerospace applications. I work on rail vehicles, and I suspect it's closer to your notion of "no detrimental yielding".

How do you confirm "no detrimental yielding" by NL static FEA in your industry?

tg

 

[rb1957]

highly localised yielding (like near a fastener hole) would be judged acceptable 'cause after loading the part would look the same as it did before the test. what you don't want is plasticity that distorts the part; an example would be a linkage in a mechanism that would "spring" if disassembled ... ie the linkage had significant plastic deformation from limit load such that the part was distorted).

the real test/confirmation for us is our static tests.

 

[DanStro]

It's rare that I need to deal with plastic deformation but I was taught that, except at extremely low loads, there is always some plastic deformation and that this is why the 0.2% guideline is usually used. If that is true then is a requirement for "no plastic deformation" realistic? If it isn't then I'll learn something today.

 

[EdR]

Guess I would add that "no Yielding or Plastic Deformation" would be very difficult to justify any yielding anywhere if it ever came to court, etc.....

I would be very careful with this one....we might know (or think we do) what is safe, etc. but I'm sure any explanations would quickly become very difficult if it became a legal matter......

Ed.R.

 

[trainguy]

Guys,

I was hoping, maybe simplistically, that the engineers who perform FEA have certain guidelines for evaluating the results of nonlinear static analyses, before performing redesign.


If so, do you evaluate strains? Or do you look at stress distributions? I am mostly referring to structures without explicitly modeled connections, so there are no holes. The only real stress raisers are corners, for example.

tg

 

[rb1957]

it depends ... if i'm dealing with an ultimate design load and a beam cap is showing a "sliver" of high stress, and the remainder of the section is "well in" i might just say "local plasticity", or do a full cozzone analysis if i think it's getting closer than that.

a stress peak at a corner ... how well modelled is the corner ? again, a sliver of high stress would probably be "clearly" acceptable.

but your spec raises the question. if it says "no yielding at limit/service loads" then that's what you have to work with; you know fty and fcy, so they'll limit your allowable stresses.

 

[corus]

If you're using non-linear analysis with some regions experiencing plasticity then cycle the loads to see if gross deformation occurs. Secondary stresses may exceed yield on the first loading, but shakedown so that stresses remain linear in subsequent load cycles. Stresses that exceed yield in a local area, such as a stress concentration won't cause gross deformation anyway, but could be a source of fatigue damage.

Tara

http://tinyurl.com/4ydjg7m