New to fatigue analysis and have a few fundamental questions of anyone would care to comment. Based on what I think I know is that a (S-N) stress based approach is best for high cycle fatigue and a (E-N) strain based approach for low cycle fatigue. The cut off generally being around 100K cycles.
Stress Based Approach
For high cycle fatigue I am using a (S-N) approach because the linear stresses are below 70% of yield and in the elastic part of the material curve. I clearly understand this no problem.
Strain Based Approach
When the linear stresses are over yield this is where I am still unclear and have questions.
Any comments would be appreciated.
Solved! Go to Solution.
You should be using strain life when the maximum stress exceeds the yield strength of your material.
You can use either linear and nonlinear results, and in both cases the software will apply an elastic-plastic correction. You need to check on 'Use Non-Linear Results' in the Solve Options tabs for the nonlinear results to be considered.
If you elect to use nonlinear results, you should not be defining a notch factor.
Thanks for the feedback. When I use the non linear stress option should i use the transient durbaility option? If so how does the force get fully reversed?
I could run the load non linear and reverse the load to get a full cycle, however, I have 5 load points so it is very complex?
Would you just take the peak stress from the non linear and full reverse it kind of like a linear analysis?
You can extract stresses/strains from either a static or a transient solution. You would define a transient or static durability event accordingly.
In a transient event, the reversals are determined by the response history and by the subsequent rainflow counting.
In a static event, you define the reversals by selecting a pattern type, full or half cycle. The stresses (linear or nonlinear) follow the same pattern.
Dear cadguy, hi Phillippe,
let me add a little on the background of strain-life approaches.
First, everything about the limitation of the range of usage of stress-life approach is correct.
People found some 50-70 years ago that indeed the damage correlates better with the local strain history than with the local stress history, so above yieling one should have a look at strain-life approaches.
But for application, people also see that the yielding is typically only local (as one has does not design parts that see permanent plastic deformation in the range of fatigue loads). Therefore the strain-life methods also include approximation formulae (typically Neuber or ESED) that locally give a good approximation for the plastic behavior. In most cases this means elasic FE is enough. From our experience transient results are needed whenever
play a role. I.e. effects that can not be handled by locallized approximation.
Hope this helps
Let me also invite you to our parallel forum
Dr. Michael Hack
Product Line Manager Durability Solutions
Simulation & Testing Solution
Siemens PLM Software