Design with Yield or Ultimate stress?

[321GO]

Hello Guys,

i my attempts to get practical experience with FEM, i reverse engineered an excisting part and used actual (fairly accurate) worst case loading.

After refinement the stress in the tooth fillets are considerabel above yield. The stress is actualy around 390MPa for ductile iron, which has a yield of 320MPa.

Basically this would mean the part would yield in the filled in service, but i'm reluctant to accept this, since the actual part has been in service for years without any failure in the fillet with the "too high" stresses.

Simply put, i'm confused how to deal with this..


Now my question: do you guys "accept" parts for prototype testing even if the FEM is above yield (but below ultimate)?


p.s. again, the stress is real and not a hot spot. The high stress is the location one would expect it to be in the actual part. Furthermore i double checked the loading with actual test report data.

 

[TGS4]

Is this loading static or cyclic?

 

[321GO]

TGS4,

yes it it cyclic.

The stress (P1) in the fillet fluctuates per revolution of the part.

The lowest value of P1 is near zero, so the fluctuation is almost equal to the max. value of P1, the 390MPa.

 

[corus]

A stress at a fillet radius is a localised (peak) stress where stresses above yield would be acceptable for static cases. As the stress is cyclic then failure would probably be by fatigue. You'd need to check the stress against the SN curve for that material to see how many cycles it could be designed for.

Tata

 

[321GO]

Since it was a truck wheel end, i was kind of surprised that it seems to be designed for fatigue.

Is it common to design for fatigue and not for yield for such critical components?

It seems to me the risks are a lot higher this way(what will happen in service?,loads not underestimated?,subpar material casted?). Furthermore extensive fatigue testing should also be done, which is all but cheap.

Just curious.


Regards,

 

[corus]

Even if the stress was just under yield it may fail from fatigue damage, depending on the number of cycles. There is a general rule of thumb though that if the stress range is below half UTS then it has 'infinite' life. For a wheel I would have thought such a limit would apply. Best to check on the actual value for that material.

Tata

 

[321GO]

I looked it up,

Yield strenght: 320MPa
Bending fatigue limit(non-symmetric): 380MPa

Since the stress cycles from near zero to Max. P1(320MPa), it seems to be designed for fatigue after all.

So, I was wrong in my initial assumption.

Thank you all!

 

[flash3780]

It might not matter in your particular case, but in general you should Walkerize the component stresses from FEA to determine mean stress and stress amplitude for life calculations.
Components are often designed to meet a fatigue life, rather than a simple yield criteria, especially when considering concentrated stresses, so your situation is not uncommon.