2 Replies Latest reply on Oct 27, 2016 11:12 AM by Jesse Powers

    Iso Clipping and "Important" Percent Volumes

    Jesse Powers

      Hey, FEA wizards, etc


      I've been doing FEA for quite some time-- using iso clippings...  As a rule of thumb, when I do look at Iso Clippings, I generally ignore any stress elements that take up less than 0.02% geometric volume of the model (or less than 0.10% element "volume").  I don't have any technical understanding of this, however, because it's something I took at face value many moons ago.


      As you may know, sometimes a sim (simulation) can give you wild (or at least "loose") max stress values, but only in VERY few elements.  And as you move downward on the iso value ever so slightly, the stress drops dramatically***.


      Topic 1)  Do you have any general rules (or even specific rules with specific examples) when analyzing an Iso Clipping?  Even ones that don't refer to "Important" Percent Volumes?


      Topic 2)  Do you prefer Standard or Curvature-Based Meshes when dealing with Iso Clippings?  I generally use Curvature-Based meshes when dealing with Iso Clippings, but not always.


      Topic 3)  As you understand it, should you look at the failure mode of a part (based on the inherent initial crack size associated with both the material and production method) when you're analyzing Iso Clippings?**



      ** The way I look at it, the Iso Clipping has to be above the yield strength AND go through the body of the part; the Iso Clipping (above or very close to the yield stress) has to touch both opposing surfaces, or be so close as to cause concern, before a re-design-- or double-checking your sim set-up-- is necessary.


      *** I've noticed this: say you have a model that you simplify for FEA, and you use different fixturing and load schemes (with varying degrees of complexity to mimic roughly, at first, the real-life stressing, and then, finally, with greater accuracy).  While the different set-ups may report quite different max stress values (e.g. 48 MPa, for one, and 31 MPa, on the next one), the Iso Clipping values at the same element volume (or geometric volume), will be much closer (e.g. 19 MPa, and 17 MPa, at 0.10% element volume for both).  Those might be exaggerated examples, but I hope it illustrates what I'm curious about discussing.


      I looked online for a bit, and couldn't find anything of this sort.  So, I'd like to start the discussion.  Do feel free to post links if you've found anything relevant.





        • Re: Iso Clipping and "Important" Percent Volumes
          Bill McEachern

          Well, at least you are not asking for much .  I will cover a few things and see where it goes.

          1. You need to remember you are typically doing a linear analysis so what happens beyond yield is not representative of what is going on in reality. In an element in a post yield stress condition the stiffness in reality essentially goes to zero for ductile materials. So you could have situations where the first indication of yield can result in catastrophic runaway failure. Situations might be rare but I happened to be working on one right now and you have to recognize it by looking at it and thinking about it. Consider a high speed rotating part that is out of balance. Once a location on the inside bore goes plastic the section is now effectively reduced. And, if can't stay below yield with the whole section the situation is unlikely to bleed out with the section going down hence catastrophic failure is almost assured. Also the loading will get worse by it bending and then you get runaway out of balance....further aggravating the situation.
          2. Small volume stress riser bleed out in fillets - in most situation this is no big deal as the radius increases and the stress goes down or is picked up by other structure/elements - however this is a non-linear response and you won't see any of this in a linear analysis. You just see a high stress and you assume it away as in consequential in the big scheme of things. If durability is the object though you can't do this - the high stress is real at least to yield.
          3. it doesn't matter what mesher you use. What does matter is the shape of the elements in the area of interest. You don't want badly shaped elements in areas of high stress as you will get misleading indications of what the stress is. A red element in a sea of blue is not too difficult to assume away but when they are where you are keenly interested then it is a problem and iso plots don't really matter. I have made a request o SWX a long time a go that they should iso plots for aspect ratio and jacobian under the mesh plot options so it is easy to see where the bad elements are but nobody else, in their view, wants them so they don't put it in.  It is a bit of a problem, in my view,  when your user based in highly populated with less experienced users and you base your product enhancements on the volume or voting for a given request. C'est la vie.
          4. the failure mode issue in linear static at any rate, reduces to what stress component you look at. If brittle failure is the expected mode then P1 is the stress component to look at as it is the most tensile. They typically fail first in tension. For ductile materials Von Mises is the way to go. Obviously, if its buckling, then you need to think its a possibility and do an analysis for it.

          That's a start...