ASME VIII DBA Local Failure

[pvuk]

I've searched the forum and found multiple threads on this topic, but not one that directly answers my query.

Within Div. 2 there are two assessment methods for local failure (elastic and elastic-plastic).

The elastic method is a check against 4S. For the elastic-plastic method the equivalent plastic strain is compared with a calculated strain limit. ASME PTB-1 recommends that the elastic-plastic method is applied to all applications.

My understanding is that if performing an elastic-plastic check, to calculate the plastic strain and strain limit, solid elements really should be used. However, this adds time to model build, meshing and solve processes.

Am I missing something, or does this section of the code and the PTB document push us down the route of using solid elements and not shells for all ASME VIII DBA checks?

Thanks,

pvuk

 

[TGS4]

This particular failure mode generally doesn't occur in thin shells (that you would typically use shell elements for), because it relies on a state of high tri-axiality. But for a thin shell, the third principal direction generally has a stress of negative internal pressure or zero.

Take advantage of the exemption provided at the start of 5.3.

 

[pvuk]

Thanks for the reply TGS4.

Section 5.3.1.1 states ‘it is not necessary to evaluate protection against local failure (5.3), if the component design is in accordance with Part 4 (e.g., component wall thickness and weld detail per 4.2)’.
When referring to component wall thickness, is this only referencing the minimum thickness requirements of Section 4.1.2? I am assuming it does not require the user to carry out the DBR calculations given in Part 4, as presumably this would make the DBA check redundant?

I am actually looking at this from an integrity perspective, so API 579 is the code of interest. Section 2D.3.1 of API 579 states that ‘the strain limit criterion does not typically need to be evaluated if the component design is in accordance with the applicable construction code and the presence of a flaw does not result in a significant strain concentration’.

I think this may also give me a local failure / solid element FEA model get-out clause as I am typically concerned with vessel features which were designed to DBR, but due to deterioration can no longer be deemed FFS via these methods. However, I do not think typical deterioration such as corrosion or pitting would be expected to provide a strain concentration or have any effect on the tri-axiality.

 

[TGS4]

I agree with your assessment.

 

[BJI]

Out of interest, what would you classify as a significant strain concentration?

 

[TGS4]

A sharp notch or square inside corner, or even a re-entrant corner.

 

[BJI]

So no there is no magnitude of equivalent plastic strain, or distance over which it extends, that would be considered a significant vs insignificant concentration? Would that also mean that any strain not in the vicinity of a notch is insignificant for local failure assessment? I know most of the testing behind the development of the fracture damage equations were based on notched bar samples but I don't think any of this was ever published by WRC.

Theoretically in a purely triaxial stress state you couldn't support any strain but I am yet to see a real FFS application that comes close, or many that even exceed the triaxial strain limit (definitely not at low strain levels). Could a low level strain limit be used, one that could be considered conservative for any conceivable PV application?

I have come across the significant strain concentration debate a few times, it would be good to have it quantified.

 

[TGS4]

The issue with Local Failure is that at a state of high triaxiality causes the material to act in a brittle manner - which is why the triaxial strain limit plummets when the state of triaxiality increases - there is less and less room for plastic strain. What's most important is not the specific magnitude of the plastic strain, but the ratio of the factored plastic strain to the triaxial strain limit. I call that quantity the Strain Limit Damage Ratio - see my paper here.

The other aspect of this failure mode is that in a state of pure triaxiality, regardless of the magnitude of the principal stresses, the von Mises stress is always zero (if S1=S2=S3, then the equivalent stress is always zero).