linearized stress question

[TxAg78]

Background:
I have a utility that calculates the linearized stress distribution from FEA results (component by component). It outputs a table at each end end of the SCL that includes: 1) membrane and bending values for each component, 2) vonMises and Tresca (S.I.) calculated from Sm+Sb components. I do the math; classification of the curves (into primary, secondary, etc) is left to the engineer. I also have an option to plot the data: 3 curves are available for each component: input FEA distribution, calculated Sm, and Sm+Sb. The last 2 are straight lines, primarily to compare linearized vales to FEA as a sanity check.

I've been asked to provide a plot of linearized vonMises distribution across the section. (In other words, calculate the Sm+Sb for each component at points thru the section, then calculate vonMises.) The math and data management isn't hard (but complexity can always add undetected errors). So far I have resisted, as I don't see any engineering value, and fear an engineer would be confused, misuse (or abuse) these results. I'm told this can be done with ANSYS and SolidWorks Simulation. (Don't know if this is true, but think it spurred the request.)

Questions:
1) Does the vonMises pot have any value to someone doing ASME Sec VIII Div 2 DBA?
2) If so, please explain.
3) If not, why would other FEA codes include this (or do they)?
4) In other words, is this a useful feature, or a distraction to the user?

Thanks in advance for your comments.

 

[prex]

Of course you can calculate the von Mises stress at each point through the section, then you can linearize this distribution. However the vM stresses at the section ends will be different from those that you correctly calculate as per the first paragraph of your post. Hence this result is of no use in Div.2.
I've never used the linearization feature of Ansys, as I found it, a long time ago, to be not in accordance with code rules. So I don't know what Ansys can presently do in this respect.
Could you review the faq794-982 to see if it is in accordance with your math?

prex
[URL unfurl="true"]http://www.xcalcs.com[/url] : Online engineering calculations
[URL unfurl="true"]http://www.megamag.it[/url] : Magnetic brakes and launchers for fun rides
[URL unfurl="true"]http://www.levitans.com[/url] : Air bearing pads

 

[TxAg78]

Prex,
Thanks for your comments and the link to the procedure. Yes, that's exactly what I do. I can integrate constant or axisymmetric sections. (There's a flag to request the appropriate integration.) I output all linearized values (for each component, plus combined results at both ends of the SCL, along with invariants). The user doesn't really need all of them, but it helps track down potential mistakes (at the risk of inducing a few also!). I'd insert an image if I could figure out how.
Note: The request was to plot the intermediate vonMises distribution (but not "re-linearize" them). Some feel it's possible that an intermediate point can have a larger value than the ends. I'd be VERY surprised if that ever happened.

 

[TGS4]

When you calculate the Von Mises stress of the linearized components, are you following the rules of 5-A.4.1.2 for the bending stresses?

Note that the membrane stress and any bending that would be classified as primary is all that is needed for satisfying Protection Against Plastic Collapse. Secondary bending is only applicable to Protection Against Failurr From Cyclic Loading: Ratcheting - and only in the context of stress ranges between operating conditions.

To answer your questions:
1) No
2) because it is not a quantity requested by Part 5
3) Because they are not primarily concerned about the BPV code or complying with it
4) Not useful

Even if an intermediate had a higher Von Mises stress - it doesn't matter in the context of a Part 5 evaluation.

 

[TxAg78]

TGS4,
Thanks for the responses. That's what I thought (that plotting intermediate values of Von Mises is a useless exercise, with no real value when using the BPV code). That said, it's interesting to hear the request justified as "FEA vendor XXX can do it". What can I say?

I first learned and applied stress linearization methods long before ASME formalized the rules (or FEA codes provided the data). I'm vaguely familiar with the 5-A.4.1.2 rules for the bending stresses. As I recall, Sb is only used for some of the components (shear is ignored).

While we're on this thread, given the DBA alternatives, how often is linearization used vs nonlinear analyses (90% of the time, 50%, less?)

Thanks!

 

[TGS4]

I, personally, use the nonlinear 85-90% of the time. Selecting valid SCLs and then classifying the resultant linearized stresses is just too complicated. However, I suspect that the industry experience is more a 1-X relationship to what I am doing.

 

[TxAg78]

Selecting valid SCLs and then classifying the resultant linearized stresses is just too complicated.

You can say that again! Given the maturity of FEA software, and speed of modern computers, I wonder why more don't use nonlinear methods. But not for me to question why.

Thanks for the feedback!

 

[TGS4]

I wonder why more don't use nonlinear methods

Education, experience (mentoring) and general lack of familiarity with LRFD methods.