Doubt about the Hot Spot Method in Shell Elements

[victor6397]

Hello, I have some doubts about some concepts, maybe very basic, but I hope you can help me.
I know that hot spot is a method to eliminate peak stress, normally from the weld toe to evaluate fatigue in S-N curves that do not require this component, but I have also seen this method used in shell geometries for, for example, bolt holes.
Is this correct?
It is my understanding that in shell elements there is no peak stress, since they linearize the stresses directly through the thickness, so a method like the one described in ASME Sec VIII Div 2 does not make sense to eliminate this stress component.
But what about the hot spot method, which linearizes in the meridional or hoop direction? Obviously the stress distribution will not be linear if we apply this method in a bolt hole for example, where there will be a stress concentration.
Is it correct to assume in that case that the stress above the linearized maximum stress at the edge of the hole is a peak stress, and therefore should be eliminated for fatigue analyses that require them?

Thanks in advance

 

[rb1957]

sorry but what the heck are you asking ?

you start with "hot spot" method ... which could be mesh refinement to address a localised stress peak or it could be using nueber's analysis to adjust for local plasticity.

you say "shell elements there is no peak stress" ... what on earth do you mean ? Of course there are peak stresses in shell elements, like at point loads. Do meshes reflect stress peaks due to details not modelled ? Of course not. Yes they linear stress through the thickness ... this is "just" the linear stress due to bending.

can you explain "the hot spot method, which linearizes in the meridional or hoop direction" ? Do you mean averaging elements ?

The problem is using linear FEA beyond the proportional limit.

Does this "Is it correct to assume in that case that the stress above the linearized maximum stress at the edge of the hole is a peak stress, and therefore should be eliminated for fatigue analyses that require them?" mean not modelling holes, but then using Kt = 3 to account for the hole ?? If so, really poor wording ...

 

[victor6397]

sorry but what the heck are you asking ?

you start with "hot spot" method ... which could be mesh refinement to address a localised stress peak or it could be using nueber's analysis to adjust for local plasticity.

you say "shell elements there is no peak stress" ... what on earth do you mean ? Of course there are peak stresses in shell elements, like at point loads. Do meshes reflect stress peaks due to details not modelled ? Of course not. Yes they linear stress through the thickness ... this is "just" the linear stress due to bending.

can you explain "the hot spot method, which linearizes in the meridional or hoop direction" ? Do you mean averaging elements ?

The problem is using linear FEA beyond the proportional limit.

Does this "Is it correct to assume in that case that the stress above the linearized maximum stress at the edge of the hole is a peak stress, and therefore should be eliminated for fatigue analyses that require them?" mean not modelling holes, but then using Kt = 3 to account for the hole ?? If so, really poor wording ...

With the hot spot method you get the structural stress (membrane plus bending), eliminating the peak component (pressure vessel analysis terms).

In shell geometries, you obtain the stress on the mid-surface, and in the post-processing the stress is extrapolated linearly to the theoretical outer and inner surfaces, obtaining membrane plus linearised bending, without peak stress.

In codes such as ASME Sec. VIII, the linearisation is done through the thickness, not with the hot spot method, so in shell elements this procedure is not necessary, as the peak stress does not exist (that added to the primary and secondary stress due to the effect of gross structural discontinuities).

My question is whether it is appropriate to use the hot spot method on shell elements, where it is not linearised through the thickness, but along the surface, and consequently the stress measured at these nodes will not be linear.
By linearising you will be assuming a peak or notch stress.

Sorry if I have not expressed myself well, I am still learning about this subject and English is not my main language.

 

[3DDave]

Background on "hot spot analysis."

FEA Weld Evaluation: Nominal vs Hot Spot Stress | SDC Verifier

There are different approaches to defining weld stresses in FEA. The nominal and hot spot stress methods are the most popular among them.

sdcverifier.com

As far as I can tell the hot spot analysis is to focus on known areas of high rates of change in the stress distribution rather than assuming a uniform stress distribution.

 

[rb1957]

I don't doubt there's a body of knowledge behind this idea ... but I don't see it. Never no mind ...