Pedro's comments got me to thinking about a fatigue question that comes up in my company
from time to time.  (My background has been static FEA analysis -- over the past few years
in my present job I've had to add fatigue analysis to my bag of analysis tools, though I'm
very much still learning.)

Normally our analysis is done with an S/N approach to fatigue with a Miner's rule
summation for the various fatigue loads/cycles that are given.  The stresses are obtained
from an FEA analysis (linear elastic) and are almost always below yield.  The question
arises when we have a set of high-cycle, low-stress fatigue loads given (several hundred
thousand to a few million cycles) combined with 1-2 cycles of a certain load that causes
stresses to go above yield.

Does the fact that we have 1 cycle of stresses above yield negate the using of an S/N
curve that was generated from testing in the elastic region?

Another place where this has come into play is when a small area of the component exceeds
yield during factory acceptance hydrostatic testing (according to analysis).  Can we still
use S/N curves on this part for calculations of fatigue life or should we use a crack
growth method?

I know this only partially relates to FEA, but I'd love to get some insight from some of
you who do/have done fatigue life/crack growth analysis more than I.  Thanks.

Greg Cruse


----- Original Message -----
From: marcalpv <[log in to unmask]>
To: Fea-l List Member <[log in to unmask]>
Sent: Wednesday, October 20, 1999 10:00 AM
Subject: [Fea-l] Failure Criterion {01}


Hi Karl H.,
The various failure criterion you describe and elaborated on by others here,
describe various points on the stress strain curves of different types of
materials. As an analyst I prefer to think of failure as systemic. That is
to say governed by our overall structural conditions ie loadings,boundary
conditions and etc. Your failure conditions are then conditions for failure
initiation. (Sometimes these are used as failure criterion by the codes as
part of acceptable practice.). In order to classify the various systemic
failures ,  I divide the material behavior into three parts. The elastic
(opening game), plastic(middle game) and material failure(end game). Then we
can perform various analysis or tests to assess systemic failure according
to the type of material behavior that we expect.

material       Systemic type of failure

elastic          large deflection, buckling, elastic fracture, resonance,
                    high cycle fatigue
plastic          limit loads and other unconstrained flow, low cycle
                    fatigue, nonlinear versions of the above.
mat. failure  propagation of material failure, release of stored energy,
                    creep, corrosion.
                    (Our analysis models do a poor job of this third phase)

It is important to note here that there are extensive approximate methods to
help us assess systemic failure. If this is the case you might well ask "why
are there failures?"  I am sure the experience of this group can help us
understand this better. Perhaps a consistent application of all the above
failure modes will lead us to sigma six land.
regards,
Pedro,
[log in to unmask]





%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%