Question Regarding ASME Section VIII Division 1 and 2. 4

[IdanPV]

Dear Experts,

Few questions regarding ASME Section VIII Division 1 and 2.

Proof Test in accordance with ASME Section VIII Division 1, UG-101(m)
1. Is there any requirement in the Code regarding the number of the pressure gauges in use during the test?
The A.I says that at least 2 pressure gauges shall be contacted to the vessel and both shall show the same pressure. I wonder if the Code has any requirement regarding that.

2. Is there any requirement in the Code regarding the calibration date of the pressure gauges in use during the test?

3. From ASME VIII-1, UG-101(m)(2)(-a) the calculated burst pressure requirement for a vessel is 50BAR. during the test, this pressure was achieved but the vessel didn't burst, so the test was stopped and a burst pressure of 50BAR will use in calculate the MAWP for that vessel.[sup][/sup]
The A.I said that he have to see if there are leakage from the vessel otherwise he won't approve the test. I wonder if there are any requirements regarding that issue. This test is not a standard hydrostatic test per UG-99, so why we have to worry about leakage.

Design by Analysis Requirements in ASME Section VIII Division 2
Few questions regarding Part 5 of ASME VIII-2.

4. Protection Against Local Failure – There are two options for evaluating protection against local failure.
As per 5.3.1.2 "When protection against plastic collapse is satisfied by the method in 5.2.3, either method listed below is acceptable".
Assuming I'm using 5.2.3 to satisfied protection against plastic collapse and I want to use the Local Strain Limit method, what is the local criteria for this case? Table 5.5 is applicable only for an Elastic-Plastic Analysis and Table 5.4 stated the Local Criteria is per 5.3.2.1 which does not seem to contain any factor for the loads.
Shall I apply loads factored by 1.7 as per Table 5.5 even though this factor is larger than the factor needed to meet the protection against plastic collapse criteria in 5.2.3?

5. Fatigue Assessment – using the Elastic-Plastic Stress Analysis and Equivalent Strains with the Twice-Yield Method as per 5.5.4. Why there is no requirement regarding the use of Weld Surface Fatigue-Strength-Reduction Factors as in 5.5.3?

6. When designing a pressure vessel for a Full Vacuum condition using 5.2.4 (Elastic-Plastic) when the atmospheric pressure is applied to the outside surfaces of the vessel.
Does the atmospheric pressure need to multiply by the ß factor in accordance with Table 5.5 for the global criteria?
The vessel is rectangular in shape with external reinforcements, the Aspect Ratio of the vessel is less than 2. Therefore, I am assuming that buckling of the vessel is not a failure mode in this case.

 

[DriveMeNuts]

1. Look at UG-102 for Gauge requirements.
2. Look at UG-102(c)
3. This seems odd. The proof test is for structural integrity, not leakage. Perhaps if water is leaking, then it will be difficult to get an accurate pressure. A standard Pneumatic test (UG-100(d)) says that any leaks detected, should be corrected and the test repeated, however this doesn't apply to a proof test.

4.
5.
6. If you are referring to a Type 3 buckling analysis, ß needs to be applied. I understand that ßcr does not need to be applied because the FEM model includes imperfections in geometry. Your model does include deviations of geometry such as ovality so that Type 3 analysis can be used? My understanding is that if your model can replicate the exact geometry (including imperfections) of the application then the buckling failure pressure will be accurate enough to consider the analysis as essentially a standard elastic-plastic analysis (5.2.4) with a standard table 5.5 margin of ß. Of course, if Part 4 has rules for your geometry, then you must use those rules.

 

[TGS4]

4. If you are using 5.2.3 for your analysis of Protection Against Plastic Collapse, then you need to consider Protection Against Local Failure as a completely separate analysis. If you are doing elastic analysis (5.3.2), then follow the rules in 5.3.2. If you are doing an elastic-plastic analysis, then follow the rules in 5.3.3, including the factored loads from Table 5.5. Either way, it is completely independent of the analysis for Protection Against Plastic Collapse.

5. In my opinion, there ought to be. Nevertheless, the ASME Code Committee could not come to a consensus on how to handle welds in the elastic-plastic fatigue analysis. I would recommend using the method in 5.5.5 for the welds.

6. Yes, the pressure differential needs to be multiplied by the β factor. I would highly recommend that you see what is contained in PVP2020-21112. This is where the Code is going, and represents the best technology in the industry. You may need to register for the current PVP2020 virtual conference to access the paper.

 

[IdanPV]

Thank you DriveMeNuts & TGS4, I appreciate your help.
@ TGS4,

4. Assuming I am using 5.2.3 (Limit Load Analysis Method) for my analysis of Protection Against Plastic Collapse. As you said, I need to consdier Protection Against Local Failure as well.

When protection against plastic collapse is satisfied by the method in 5.2.3, either method listed below is acceptabe.

The thing is, I want to use 5.3.3 - The Total Strain Limit Method, but I can't find the required factor in the Code.
Table 5.4 only mention that for the Local Criteria one shall "See 5.3.1.2" but that paragraph doesn't contain any factor.
Table 5.5 stated that for the Local Criteria, the load shall be multiplied by 1.7 but that seem wierd since the factor for the Global Criteria in the Limit-Load Analysis is only 1.5.

5. You mean using 5.2.4 for analysis of Protection Against Plastic Collapse but for the fatigue assessemnt of the welded joint I shall use Tables 5.11 & 5.12 reduction factors?

6. Thank you.

 

[TGS4]

4. You use a different factor because you are using the full elastic-plastic stress-strain curve. The Limit Load uses an elastic-perfectly-plastic stress-strain curve. Completely disregard what you had done in Protection Against Plastic Collapse.

That said, since you need to develop the full stress-strain curve for 5.3.3, you might as well use 5.2.4 for demonstrating Protection Against Plastic Collapse.

5. No. I said that if you are using the elastic-plastic fatigue method in 5.5.4, then for evaluating the welds you should use the Structural Stress Method in 5.5.6.

 

[IdanPV]

4. According to what you wrote, there is no way to fulfil the requirement of 5.3.3.1 using the results from a Limit-Load Analysis?
I came to conclusion that using the Elastic-Plastic Method is the best opition for DBA.

5.

Structural Stress Method in 5.5.6.

5.5.6 gives the procedure for Ratcheting Assessment, am I missing something?

 

[TGS4]

4. Indeed. Elastic-Plastic is the best option for demonstrating Protection Against Plastic Collapse. It gives you the opportunity to also evaluate Protection Against Local Failure using the same model/material property. And you can also use the same model/material property to evaluate Protection Against Collapse from Buckling.

However, you need to ensure that each assessment is separate. My approach is to treat each failure mode as separate and unique. Even if you use the same model and material deck, it is highly advisable to treat each failure mode separate. In that context, then, I studiously avoid saying that there is an elastic-plastic "methodology" in design-by-analysis. There are elastic-plastic methods in each of the various failure modes. It's a small distinction, but when you get to elastic-plastic ratcheting or fatigue, the approaches are completely different.

5. My bad - I meant 5.5.5. Mistyped. Sorry about that. Definitely use the Structural Stress Method for welds.

 

[IdanPV]

@TGS4, Thank you.

5. 5.5.5 Uses the results from Elastic Analysis and required the elastically calculated membrane and bending stress. As far as I know, there is no way to get these values from Elastic-Plastic Analysis, am I right?
Regarding the structural stress range and the structural strain, can I get these values directly from the analysis results?

P.S
Regarding the PVP2020-21112, is it possible to get it from the ASME Digital Collection?

 

[TGS4]

5. Correct. There is no way to get that info from an elastic-plastic fatigue analysis. Unfortunately.

You cannot get PVP2020-21112 from the Digital Collection yet. You would need to register for the PVP2020 Virtual Conference. It's rather cheap at only US$299. And you get copies of ALL of the PVP2020 papers.

 

[IdanPV]

So, how should one perform an fatigue assessment of the weld with the approach of 5.5.5 without doing an elastic analysis?
Do you any reference or an article regarding the use of an elastic-plastic analysis and fatigue assessment of weld using 5.5.5?
How shall I use the results of an analysis made using the Twice-Yield Method for a fatigue assessment of the welds?

This topic because some kind of "private lesson" so if it is too much I will appreciate an article or some sort of reading material regarding that issue.

 

[DriveMeNuts]

TGS4,

6. The paragraph for ASME VIII-2 for Type 3 analysis says "It should be noted that a collapse analysis can be performed using elastic or plastic material behavior. If the structure remains elastic when subject to the applied loads, the elastic–plastic material model will provide the required elastic behavior, and the collapse load will be computed based on this behavior."

What is mean by "a collapse analysis can be performed using elastic or plastic material behavior"?

My interpretation of this is that you apply design loads to an elastic analysis (with a linear elastic material model) a turn on large deflections, and then linearise the membrane and bending stresses, ensuring that they don't exceed 1.5xS, however the Type 3 paragraph explicitly says use 5.4.2?

Or do you do a 5.4.2 analysis with a linear elastic material model and large deflections turned on, and load the model until convergence can't be reached? However, I would have thought that as the strain exceeds yield into the pseudo-elastic region, the material becomes too unnaturally stiff and therefore the result will be non-conservative.

 

[TGS4]

IdanPV For a fatigue analysis of welds using the Structural Stress Method, you are obligated to perform an elastic analysis. That's the entire premise of the method.

DriveMeNuts the current method for demonstrating Protection Against Collapse From Buckling is broke and effectively incorrect. As such there is effectively no point in trying to explain or justify the current words in the Code. The method described in PVP2020-21112 is what is currently in front of the ASME Code Committees for review. It is my hope that it will pass shortly so that it will be incorporated into the 2021 Edition.

In my Part 5 training course, I have been teaching this "new" method for several years now. The long and short of it is that you ought to be using an elastic-plastic buckling analysis, and explicitly imposing a perturbation. The shape of that perturbation should come from an eigenvalue buckling analysis, and the magnitude should be equal to the fabrication tolerances that match the mode shape. Then, the design margin should be equal to 2.0 (which matches the Design By Rules design margin).

 

[DriveMeNuts]

Thanks TGS4,
That makes perfect sense. I have been experimenting with something similar relating to the perturbation step in ANSYS. I'm currently just applying a perturbation force on a few nodes. I'm trying to figure out the ANSYS process of creating an eigenvalue result with deflection magnitudes that reflect fabrication tolerances and then saving the result as an unstressed 3D model and then running an Elastic-Plastic buckling Analysis on that model.
It looks like it requires a lot of manual steps in ANSYS, if it is even yet possible.
Perhaps as the method becomes more formalised, ANSYS will streamline the software process.
Or I am getting it wrong and just need to read PVP2020-21112.
So the new rules use a design margin of 2.0, rather than a β of 2.4. I guess Appendix 46 of Div 1 may need to be amended if it is deemed necessary to apply a higher design margin.

 

[mskds545]

DriveMeNuts -- See this article for a rundown on how to import the shape from an eigenvalue buckling analysis into a static analysis and scale the deformation in ansys:

http://www.padtinc.com/blog/eigenvalue-buckling-and-post-buckling-analysis-in-ansys-mechanical/.

IdanPV -- I'll second TGS4's recommendation for fatigue analysis of welds. A lot of great engineering went into the structural stress method for welds and it is based on a broad test database. The PD5500 structural stress method was based on a completely different database with very different weld details; yet, if you strip out the different design margins from the two methods and calculate the stresses on the same basis, the mean fatigue curves are nearly coincident. There is a large amount of very varied data taken around the world over half a century to back up the structural stress method. For elastic-plastic analysis of welds, there's basically no data available. Even if the welds are ground smooth so the geometry is well defined and matches your model, you need to deal with the stress concentrating effects of code-allowable defects. The structural stress method accounts for acceptable defects because it is based on tests of full-scale as-welded joints.

-mskds545

 

[IdanPV]

Thank you all for your detailed response, I appreciate it!

@TGS4, Thanks, all clear now. Do you know if there will be any update regarding the use of Elastic-Plastic Analysis for the fatigue assessment of welded joint in the upcoming 2021 Edition?
Regarding the PVP 2020, I will consider to register as I saw a lot of interesting papers in the event list.

@mskds545 Thank you very much, the code stated that

Weld joints with controlled smooth profiles may be evaluated using 5.5.3 or 5.5.4

So I guess that for the case of machined fillet welds I may be used 5.5.4? is that correct?

 

[mskds545]

Yes, the text is clear that welds machined to smooth profiles may be evaluated using 5.5.4. There isn't a prohibition against using 5.5.4 for non-machined welds either, as far as I am aware, although I'm not sure how you would analyze the weld toe accurately with no dressing at all. Even with machined welds, I would spend some time thinking about the defects that your welds might contain and how those defects might impact the fatigue strength of your weld. While 5.5.5 would be conservative for machined welds, I believe that 5.5.4 can be non-conservative if your weld contains code-acceptable defects. It is telling that the word weld doesn't even appear in 5.5.4.

-mskds545

 

[TGS4]

No changes to 5.5.4 will be forthcoming in the 2021 Edition. I hope that we will be able to have charges for the 2023 Edition.

I hope you will join the virtual conference. Lots of good papers.