Joint Efficiency with FEA 4

[Eng822]

Thank you in advance for taking to time to read through my questions. How do you define your allowable stress for a finite element analysis with a joint efficiency of 0.7?

Code: ASME Sec VIII Div. 1 2010 (No ed) with guidance from Div. 2 for FEA
Customer specific seismic requirement with reference to ANSI 360-05 and N690

I have completed my code calculations using COMPRESS however, due to our customers specific seismic requirements we are also analyzing the vessel and it's supports with FEA. I am familiar with the different levels of stress classification (general membrane, general membrane + bending, etc.) however, in the past when I have done this type of analysis the design was for a joint efficiency of 1. This time the design does not require any radiography therefore my joint efficient is 0.7. I believe a conservative approach would be to just take my allowable and multiply it by my joint efficiency but is this common practice for FEA?

Here is my specific example.

General membrane allowable limit: 20 ksi * 0.7= 14 ksi
Local and general membrane plus bending limit: 30 ksi * 0.7 = 21 ksi
general membrane + secondary stress limit: 60 ksi * 0.7 = 42 ksi

Below is a stress plot of the membrane stress. I realize in this example I am well below my allowable limit even after multiplying by a joint efficiency of 0.7 but that is not always the case. lets say the localize stress around the support is greater than 21 ksi. Because the stress crosses the circumferential seam it makes sense to me that the design would be unacceptable. However, say my deign uses support lugs and the localized stress area does not cross over a circumferential or longitudinal seam, would it still fall under the same allowable limit of 21 ksi or would the limit be 30 ksi?

 

[chevproe]

In absence of any clear directive from Code on this issue, I would go the conservative way and consider 21 Ksi.
I would request the stalwarts in this forum to share their opinions.

 

[pdiculous963]

Brian822, you are correct in multiplying the allowable stress by the joint efficiency. From my understanding, the upcoming 2015 version of API 579 will require that the allowable stress be multiplied by the joint efficiency at welded connections when performing design by analysis.

 

[TGS4]

I can speak rather authoritatively about this issue, as I am the Technical Project Manager for shepherding this topic through on the VIII-2 Part 5 side, and have had significant input into the changes that have been proposed for API-579.

The current proposal (subject to change...) for VIII-2 Part 5 is:
Modify 5.2.2.4, Step 5 (a) to say “1.5 times the tabulated allowable stress for the material from Annex 3-A. If the point on the vessel being evaluated is located at or adjacent to (less than 1*t away) from a welded joint, the allowable stress shall be multiplied by the weld joint factor for that weld as defined in Table 7.2.”

Modify 5.2.2.4, Step 5 (b) to say “Sy for the material from Annex 3-A, except that the value from paragraph (a) shall be used when the ratio of the minimum specified yield strength to ultimate tensile strength exceeds 0.70, or the value of S is governed by time-dependent properties as indicated in Annex 3-A. If the point on the vessel being evaluated is located at or adjacent to (less than 1*t away) from a welded joint, Sy shall be multiplied by the weld joint factor for that weld as defined in Table 7.2.”

In order to be consistent with the rules elsewhere in VIII-2, as well as VIII-1, the joint efficiency is ONLY applicable for Protection Against Plastic Collapse. Consideration of the weld joint efficiency is not required for Protection Against Local Failure, Protection Against Collapse From Buckling, or Protection Against Failure From Cyclic Loading: Ratcheting. The presence of the weld is considered in Protection Against Failure From Cyclic Loading: Fatigue through either the use of a FSRF or the Structural Stress Method.

It is important to note that this factoring of the allowable stress by the joint efficiency factor is ONLY required in the immediate vicinity of a weld joint. Furthermore, even though a general weld seam may not require NDE (and hence have a non-unity weld joint efficiency), it is common for fabricators to perform NDE on weld seams that will underlie reinforcement pads.

Brian822 - you allowable stress for the stresses classified as primary local membrane would be (assuming SA516-70 at room temperature) 38ksi * 0.7 = 26.6 ksi. This would be true for all of the Design Load Combinations listed in Table 5.3 (which should all be checked, regardless of whether your vessel is VIII-1 or VIII-2, as I discussed in my blog post -

http://becht.com/blog/performing-an...ion-1-to-qualify-an-article-u-2-g-component).

In my opinion, there would be no location nor stresses in your vessel that would be classified as primary local bending. Furthermore, the checks that you should be doing for the stresses classified as secondary relates to Protection Against Failure From Cyclic Loading: Ratcheting, and should follow the rules in that section. There, you will be evaluating the operating load ranges, not the Design Load Combinations (as I have discussed in my blog post -

http://becht.com/blog/asme-section-...-analysis-discussion-collapse-vs-ratcheting).

And, as I stated above, there is no need to penalize the allowable stresses for this failure mode (to be consistent with the rest of VIII-1 and VIII-2).

One final note: there are similar proposals for Protection Against Plastic Collapse using the Limit Load analysis method and the Elastic-Plastic analysis method. However, based on some sample problems that I am currently running (and problems/challenges therein), I am not yet ready to share my thoughts on how to approach this issue for those methods.

 

[Eng822]

Thank you all for your responses. I am getting some great information here.

TGS4 - I realize that it was probably confusing that I posted a plot of my general membrane stress but talked about general membrane plus bending. However, the stress plot of the inner and outer surface of the vessel wall would be classified as general membrane plus bending, correct? But now I am curious as to why you do not classify the area around the legs as localized stress areas? Section VIII-2 1998 defines the area as localized bending stress if the region does not exceed sqr(R*t) in any meridional direction. I know later additions no longer define it this way but I was wondering if you could expand on why this is the case?

I almost feel like this next question might deserve its own separate thread on classifying stress... Anyway, On the topic of secondary stress limits, I've considered classifying thermal induce stress or bending stress at finite points such as a 90 deg corner or an abrupt edge as secondary because it is self-limiting and only slight local yielding or distortion can satisfy the developed stress area. If this is not the case, how do you justify the areas around a boundary condition or fixed constrain where the stress is often vary high?

Just a heads up I have not read your blog post yet but intend to next so my apologies if you already covered my questions.

 

[TGS4]

Brian822 - to keep things in this thread coherent and relevant only to weld joint efficiency, please start another thread to discuss the stress classification issues. There is much that I would like to discuss with you on that topic, however.

My final comment on stress classification for this thread is (and those who know me well know how much I really like this paragraph)

For components with a complex geometry and/or complex loading, the categorization of stresses requires significant knowledge and judgment. This is especially true for three-dimensional stress fields. Application of the limit load or elastic-plastic analysis methods in 5.2.3 and 5.2.4, respectively, is recommended for cases where the categorization process may produce ambiguous results.

 

[jte]

Hey TGS4-

Oh yes, we know you like 5.2.1.2! Would you agree to the following modification of one of your above statements:

It is important to note that this factoring of the allowable stress by the joint efficiency factor is ONLY required in the immediate vicinity of a weld joint and only when the stress is tensile and crossing the weld seam.

The fun here is that if we are looking at von Mises stresses without a directional component. Yet in theory, I could have a high longitudinal stress right at a longitudinal seam with the primary driver of the high von Mises being the third principal stress which happens to be perfectly aligned with the longitudinal seam. If I were performing manual calc's in a Div. 1 principal stress based mindset I would ignore the JE. Similarly, if the stress is compressive, I would reasonably ignore the JE even if its direction crosses the weld seam.

Your thoughts?

 

[TGS4]

jte - I would agree to that modification. That would be consistent with there being no modification for Protection Against Buckling Failure due to a JE. Although I'm not sure what shear across such a weld would do...

That brings up the issue of whether or not to modify the calculated stress (or component stress in this case) or the material resistance. And that ends up being the factor that is complicating my work for the limit load and elastic-plastic analysis. Although we generally assume our materials to behave isotropically, in the presence of a flaw (undetected because the examination was not performed), we know that is not the case; which is the situation that we are trying to approximate with a joint efficiency factor. But rather than address the anisotropic nature, we stick with the isotropic assumption and penalize the material strength. Again, probably not the best solution, but it might be all that we have with the blunt tool that is joint efficiency.