Fatigue analysis maxi principal stress

[marulanda]

HI, I'm Alfonso.
I need explanation on a fatigue analysis using Max principal method.
I've applied 8 fatigue load condition on a fem model, with a static analysis.
I've found the critical location on the model, and allowable delta sigma from en-1993.1.9 .
Now I've 8 max pirncipal and 8 min principal stresses in this location, but I don't know how use those.

Is it correct subtract the maximum MaxPrinc and minimum MaxPrinc and compare the result with the allowable (do the same with MinPrinc)?

I've read that there are a lot of different approach.
What are the most popular methods?

thanks.

 

[corus]

The general method is to calculate the maximum principal stress range, ie. not simply the difference between the maxima but the difference between the stress components from which you calculate the principal stresses. Then it's common to use the rainflow method. I think it's all detailed in the standard you quote.

 

[realtime2]

Generally if one has a bunch of possible max principal stresses it
is best to create a plot of their values versus
their angle. As Corus recommends one must do this for both tensile and compression
principal stresses. The 2D plot of stress vs angle will usually show two
large critical angles. One for each principal stress (assuming planar problems).
One needs to pick load cases that give both large tensile principals and
large compressive principals.

When there are multiple channels of load inputs scan each channel
for max and min values, and at those times also record the values of the
other channels. i.e.: things may be out of phase. Then plot the principals
for each of these "simultaneous max mins". To save FEA time it is probably
best to run an FEA for a unit of load for each channel independently. Then use
superposition to recombine the loads for each set of simultaneous max mins.

After picking the critical direction from the stress vs angle
plot run all the load cases with stress output for this critical direction
and then rainflow count the result. This assumes no multiaxial stress conditions
where the principals and angles are all over the place.

 

[marulanda]

Thanks all

Is it possible use von mises stress insted of max princ,if you have a multiaxial stress?

 

[rb1957]

you can use von Mises but fatigue is more dependent on max principal (von Mises is IMO more of static stress, combining the multi-axis stress state into a single value).

you say you have 8 stress cases. do your 8 cases happen sequentially ? are the cases assumed to cycle, +ve to -ve, when they occur ? does the stress state reduce to zero between loads ? would this be a conservative assumption ?? someone has mention "rainflow", look it up (as it applies to fatigue analysis). you need to understand how the max. principal stress can cycle at the critical location. It is not simple and difficult to write (in a blog like this).

another day in paradise, or is paradise one day closer ?

 

[realtime2]

In elements that are severely multiaxial, as observed in a
max/min principal vs angle plot, one cannot use von Mises or just a principal.
Best method is to divide the possible angles into 10 or 20 representative
cuts and then resolve the element stress history to stresses
perpendicular (plus shear) to each cut. Then Rainflow these resulting stress
"on a cut" histories to find the critical angle. Probably also best to use
a damage parameter that includes shear. The one by Findley is probably ok, but
there are a number of others.

 

[rb1957]

maybe a post-FEA code (like "ncode") could help ?

another day in paradise, or is paradise one day closer ?

 

[realtime2]

Yes ncode, or any of the other two(?) use a similar scheme.
But its expensive per seat for the commercial codes,
and the process isn't really that difficult for a few elements.
I'm tempted to write an open source version.

 

[rb1957]

sounds like you should draw mohr's circle for each load case.

but load sequence and cycles is vital (I mean I'd use the max principal stress for each load case as being conservative and worry about the possible load cycles before I tried to "perfect" the stress story by saying in this particular direction I have max principal stress for case 1, some normal stress and some shear stress for case 2, etc).

another day in paradise, or is paradise one day closer ?