Saddle Allowable Stresses in VIII-2

[Paulettea]

Dear All,

I have some questions regarding allowable stresses as per ASME BPVC VIII-2 when it comes to saddles stresses.
According to 4.15.3.5, the circumferential compressive membrane plus bending stress is denoted by σ7. The allowable stress for this stress is mentioned in 4.15.3.5 (f) to be 1.25*S.
My question is that why is this value of allowable stress less than that given in Equation 4.1.2: Pm+Pb<1.5S ??
The second question is that what is the allowable stress in hydrotest condition? Is it again 1.25*S? Can I use the Equations 4.1.4 and 4.1.5 as allowable stress in hydrotest condition? I mean since the allowable for normal operating condition is a little lower than equation 4.1.2, should I reduce the allowable stresses in 4.1.4 and 4.1.5 accordingly?
The third question is what values shall I consider as allowable stresses when the saddle is meant for shipping only?

Warm Regards

 

[TGS4]

I can't answer your question about stresses in the shell due to the saddle.

However, I can answer your question about the allowable addresses in the test condition. There are no limits on localized stresses at all. The only limits for the test condition are Pm (and for flat plates Pm+Pb) given in the equations that you listed.

 

[Paulettea]

Why TGS4?
Is it not possible to have excessive plastic deformation under local membrane loads?

 

[TGS4]

In a word - no. Provided that the primary membrane limits are achieved, then the localized stresses will take care of themselves.

This was made MUCH more clear in the 2017 edition of VIII-2.

However, I am always open to being demonstrated to be wrong. If you have an example where this is not the case, then please submit it to the Code Committee. (or you could share it here)

 

[Paulettea]

TGS4, It seems that VIII-2 and VIII-1 are the same when they want to give rules to determine required thickness of a component in the sense that they do not consider primary local loads in the rules. Nonetheless, both of them state clearly that if rules are not provided to cover all details of design it shall not be interpreted as they are not important or can be neglected. I know U-2 (g) in VIII-1 clearly mentioning this issue. I also saw the following that could be some equivalent for U-2 (g) in VIII-2.

4.1.1.2.1 Class 1. When design rules are not provided in Part 4 for a vessel or vessel part, the Manufacturer shall either perform a stress analysis in accordance with Part 5 considering all of the loadings specified in the User’s Design Specification, or, with acceptance by the Authorized Inspector, use a recognized and accepted design-by-rule method that meets the applicable design allowable stress criteria given in 4.1.6. If the design cannot be performed using Part 5 or a design-by-rule method (e.g., creep-fatigue), a design method consistent with the overall design philosophy of Class 1 and acceptable to the Authorized Inspector shall be used.
4.1.1.2.2 Class 2. When design rules are not provided for a vessel or vessel part, the Manufacturer shall perform a stress analysis in accordance with Part 5 considering all of the loadings specified in the User’s Design Specification.

I see in 4.1.6 there is no mention of a limit on local primary membrane stress whether it is test condition or design condition. I am not sure but this may be due to the fact that the equations of Part 4 does not calculate the local primary stresses.

What I do not understand is that if a vessel has to be safe regarding local primary stresses then why should I even care the source of those stresses are from design condition or test condition. I think the vessel does not understand that the loads are from design or test condition. If excessive plastic deformation is to occur due to local primary stresses then it will happen regardless of test or design condition.

 

[TGS4]

The key word in all of your last post is "excessive" plastic deformation. Excessive, how?

When we look at the acceptability of stresses, we do so in the context of a certain design margin. The margin for design conditions is different from the test condition. So, if you demonstrate that the local stresses are acceptable, with the appropriate design margin, at the design conditions, then provided that you achieve the test conditions general membrane stress limits (which have substantially lower margins) for the actual test pressure, then it is a given that your test condition local stresses will be acceptable. Certainly, small amounts of plasticity will occur. But provided that the general membrane stress is below yield (the margin is Beta_T), then you're going to be fine.

Take a read through 5.2.2.5. Specifically the sentence

...It is not required to evaluate any stress categories not listed in 4.1.6.2 (e.g. primary local, secondary, or peak) in the test condition....

If you are somehow uncomfortable with that, then I would recommend using 5.2.4.5 to evaluate the test condition.

 

[Paulettea]

TGS4, I know that the word "excessive" here has to be some measurable quantity if we want to talk in an engineering sense. But my justification of using this word is based on the ASME PTB 1 discussion on categorizing stresses.

The placing in the primary category of local membrane stress produced by mechanical loads, however, requires some explanation because this type of stress really has the basic characteristics of a secondary stress. It is self-limiting and when it exceeds yield, the external load will be resisted by other parts of the structure, but this shift may involve intolerable distortion and it was felt that it must be limited to a lower value than other secondary stresses, such as discontinuity bending stress and thermal stress.
Secondary stress could be divided into membrane and bending components, just as was done for primary stress, but after the removal of local membrane stress to the primary category, it appeared that all the remaining secondary stresses could be controlled by the same limit and this division was unnecessary.

You see the root of this discussion is the word "intolerable distortion". As it can be understood the limit on the primary local membrane stress is only and only due to the fact that we want to prevent excessive plastic deformation otherwise it was not even necessary to categorize this stress as primary. Therefore, it is a concern in design to limit excessive plastic deformation.

Regarding the design margin that you mention I think I have some doubts. You assign a design margin on the stress limits of primary membrane stress so that in the real life you feel safe by that margin. However, this margin seems good when the loads and the resultant stresses are related to each other in a linear way. For primary stresses in linear stress region the relation between the load (e.g. pressure) and the stress is almost linear ( especially for lower thicknesses). In elastic stress evaluation there will be an almost linear relation between load and local stresses. But if local yielding occurs near discontinuities, the relation between the load and the stress will not be linear in the real world.

What I am trying to say is that if you increase the pressure by a factor of 1.2 you may have primary stresses due to pressure increase by a factor of almost 1.2. But can you guarantee that the local deformations will increase by a factor of 1.2? I cannot say.

 

[TGS4]

If you are truly concerned about deflections, then you need to perform an elastic-plastic analysis. The deflections there are real (as opposed to the limit load analysis where the displacements are fictitious). No more hand-waving at un-definable terms such as "intolerable" or "excessive". And trying to manage this in the context of pseudo-elastic stresses in an elastic analysis is madness.

 

[Paulettea]

TGS4, I have another question about hydrotest which is not related to previous posts. One of the methods of increasing fatigue life of a component is to create some compressive residual stress on the maximum stress locations. Sometimes, hydrotest is called leak test but I think it can also be a way of increasing fatigue life since after an overpressure, there will be compressive residual stresses in local discontinuities. It is interesting that these residual stresses occur upon unloading in the locations which are critical for fatigue (where peak stresses are maximum). So could this be another reason why we perform hydrotest or it is just done as a leak test? Is there any credit given to compressive residual stresses in evaluating fatigue life?

 

[TGS4]

In the very high pressure regime (VIII-3), this is called autofrettage, and it certainly is used to improve fatigue life.

But this brings up an issue of comparing crack initiation and crack growth. Crack initiation is governed by stress range without (much) reference to the residual stress field. However, crack growth is very sensitive to a residual stress field - cracks cannot propagate if the stress is compressive.

There are many thoughts about the purpose of a pressure test. Certainly the effect that you mention is there. Also it is to check the weld strength, as well as being a leak test.

 

[NRP99]

1. Regarding why 1.25*S is used and not 1.5*S for Pm+Pb(Compressive bending)-"for design purposes Zick recommends σ[sub]7[/sub]<=1.25*S(Brownell and Young's Process equipment design )". ASME code calculations are based on Zick analysis. Hence it is better to search the reason behind the lower allowable in the Zick's work. I am not expert in Zick analysis so I cant say anything other than this.

2. Regarding allowable stress in hydro test condition- Refer ASME Sec. VIII Div-1 para 4.1.6.2. The only allowable not mentioned is local membrane stress. The membrane stress for the hydrotest pressure will be higher at the corner locations of the vessel or at the sharp changes in geometry configurations which we can term as local primary membrane. Does it really needed to check P[sub]L[/sub] is LESS than (1.5 or 1.25)*S since we kind of know it is certainly crossed the yield for that pressure? I think the code has restricted the overall stress levels of the vessel thickness away from discontinuity/sharp corners to less than yield to avoid plastic collapse of the vessel which is not the case for local stresses. The local stresses increase up to point where the corners/part of the vessel yields and there by changing the stress distributions from only corner to other areas and reducing overall stress levels. Now this corner yielding may be tolerable or intolerable depending on the actual distortion happened. If you think that this can be significant then it can be calculated based on the elastic-plastic analysis procedure-5.2.4 as suggested by TGS4. By doing elastic plastic analysis, you are simulating actual(not actual per say but approximately actual) behavior of vessel for hydrotest and you can see the strain hardening taking the control of sharp corner and giving actual(approximately actual...phew..)deformations.
My logic behind why ASME not included the local limit- Corner stress adjust itself by redistributing around whole area and reducing overall stress level by yielding the corner. Most of the time yielding that happens at the surface are so small that may not necessarily cause any intolerable distortion and yielding happens for very intricate/deep surface part of thickness, you don't see the actual distortions on the surface. Also strain hardening play its role of limiting deformations. The more important stress level remaining is Pm and Pm+Pb which increase as you increase the pressure and can cause through thickness yielding which can cause visible distortion of the vessel and hence can not be allowed beyond yield.

3. Regarding allowable stress values for the shipping saddles- I guess for vessel stresses you can use the same allowable given in 4.15.

4. Regarding the effect of compressive stress on fatigue life and hydrotest reasons- TGS4 has explained it clearly about residual compressive stresses and reasons of test. The effect of residual compressive stress on the fatigue life is to increase the fatigue life since the some amount of tensile stress is required to overcome the residual compression and hence reducing overall effective stress required to initiate crack and further its growth.

This is the reason why shot peening is used for increasing the residual compressive stress and hence to increase the fatigue life of component. In the shot peening process , high speed metal balls creates the small indents of very small depth on the surface by yielding the surface but the below surface offers resistance to this yielding and thereby creating compressive stress field around the indent. This compressive field then reduce the ability of surface to crack initiation and propagation by offering resistance to tensile stress range.

 

[Paulettea]

Thank you NRP99 for your reply.
1-Can you tell me how the hydrotest is different from the operating condition of equipment in terms of their allowable stresses? I mean why the hydrotest condition has higher allowable stresses in comparison with the design condition? Is it because the hydrotest is just an occasional load and the vessel throughout its whole service may not experience that load again? If this is the reason then I can tell you that the stresses during shipping of equipment are not going to happen for the rest of the equipment life so the allowables can be considered higher than the design condition.
2-For the case of local primary membrane stresses I have to say that even in the design condition there is no guarantee that these stresses do not exceed the yield stress. These stresses can be higher than the yield stress when the general primary membrane stresses are well below the allowable stress. Note that this is not going to happen in the real world but it may be the case when you are considering a linear elastic model to describe the resultant stresses.
There are two options for design of equipment against plastic collapse:
-design it as per the rules provided in Part.4 and there will be no limit on the local primary membrane stress
-design it as per the procedures in Part.5 and there will be a limit on the local primary membrane stress. (Pl<Spl)
Why does it all of a sudden become important to put a limit on the local primary membrane stress in Part.5?
3-If I want to have a judgement regarding the allowable stress of local primary membrane stress when my design is based on Part.4 I use the article 4.1.6.3. Here it does not mention local primary stress specifically however, it says that primary plus secondary stress which includes Pl also. And the limit is Sps. Meaning in the absence of secondary stresses the local primary membrane stress can reach the level of Sps. Again it is different from the limit in Part.5 which is Spl.
4-It is very difficult to develop rules of designing equipment as per non-linear models so I can understand that DBR be governed by linear-elastic formulas. It would not look a manageable task to work with elastic-plastic models to develop rules for designing equipment. However, for numerical methods like FE it is not impossible to work with elastic plastic models. If using linear-elastic methods to analyze stresses higher than yield is so much troublesome why not putting it aside totally and forget about it. We can always use elastic-plastic models these days the systems are much powerful and can give results much faster.

 

[TGS4]

1. The pressure test takes place under controlled circumstances, and typically only once at the beginning of the vessel life. The ASME Code Committee (a group of engineers just like us) decided that it would be acceptable to realize a lower design margin against plastic collapse for this condition. As far as stresses during transport, the Code does not address this, because it is not an operating loading condition (you will not find transportation in Table 5.1, for example). That some engineers choose to, in their judgement, apply the same design margins to the various failure modes is on them, but it is completely outside of the Code's jurisdiction.

2. The design rules in Part 4 are derived (generally) from hand calculations and finite element analysis (4.5.3, for example, is derived entirely from FEA - see WRC 521). These rules may not explicitly require a primary local membrane stress check, but I can assure you that it is generally baked into the rules. It IS important.

3. Regarding 4.1.6.3, the limit on primary-plus-secondary stresses is based on a COMPLETELY different failure mode than the primary local membrane stress limit. This is discussed in PTB-1.

4. It is only recently that elastic-plastic analysis has become readily available at reasonable speeds. Going back to the infancy of FEA (in the 1960's) and before, what mechanical engineers had available was elastic analysis and pseudo-elastic stresses. The elastic rules certainly aren't perfect and are extremely challenging to implement, but for 30-40 years it was the best that they could do. Personally, I use elastic analysis about 5-10% of the time and elastic-plastic the other 90-95% for demonstrating Protection Against Plastic Collapse. I use elastic-plastic analysis 100% of the time for demonstrating Protection Against Local Failure and Protection Against Collapse From Buckling. I use elastic analysis 98% of the time for Protection Against Failure From Cyclic Loading: Fatigue, and about 75% for Protection Against Failure From Cyclic Loading: Ratcheting. And I use elastic-plastic ~*5% of the time when investigating problems in the creep regime (using the rules in API 579-1/ASME FFS-1).

 

[NRP99]

Paulettea
1. Apart from hydro test taking under controlled circumstances only once in its lifetime as commented by TGS4, the vessel is undergoing high pressure loading during testing. The logical reason seems to be to allow higher allowable than design condition since we are increasing the pressure to ~43% from design/MAWP pressure to limit anyway below yield to avoid plastic collapse. Will you be able to control the circumstances (in this case pressure) once vessel is put to its REAL test?
Do you want lower allowable for design/operating condition when there is lot of loads like pressure, TEMPERATURE, wind, seismic acting together than for the shipping loads where only load acting is gravitational/acceleration based load?

2. Is there any reliable method of calculation for local stress than FEA? Same reasoning in my earlier post of why its not included in hydrotest applies here along with reason of reliable method of calculating the local stress. This doesen't necessarily mean locals stresses can be excluded from the analysis.

When design rules are not provided for a vessel or vessel part, the Manufacturer shall perform a stress analysis in accordance with Part 5 considering all of the loadings specified in the User’s Design Specification.

3. If you want to have a judgement regarding the allowable stress of local primary membrane stress when your design is based on Part.4-Use ASME Section VIII Div 2-4.1.1.2.2-Part 5.
Whatever you written about primary and secondary stress seems you need to better understand the subject of Stress classification. Start here-

http://www.eng-tips.com/search.cfm?q=primary+and+secondary+stress&action=search

4. I guess, TGS4 has made it more clear for the elastic plastic analysis use. Our general practice is however to use elastic analysis to the extent possible but with caution of accurately classifying the stresses.

 

[Paulettea]

NRP99, I did not say that the allowable stress for hydrotest being higher than the design condition is a problem. On the contrary I said since it is applied just once in the vessel life it can have higher allowable stresses. My question was if the allowables for hydrotest can be higher since it occurs just once then the allowables for shipping can be higher than those for design condition since that loading happens only once in the whole life of the equipment.

3. Regarding allowable stress values for the shipping saddles- I guess for vessel stresses you can use the same allowable given in 4.15.

What you said above looked like I should use allowables for design condition in the shipping condition also. Again in the last post you said something that looks paradoxical to previous post.

Do you want lower allowable for design/operating condition when there is lot of loads like pressure, TEMPERATURE, wind, seismic acting together than for the shipping loads where only load acting is gravitational/acceleration based load?

So what is the decision? can I use higher allowable stresses for shipping like hydrotest or not?

Unfortunately you may have misunderstood my point about putting aside linear-elastic method for local stresses. I did not say that FE is not appropriate. It is very good but for local stresses linear-elastic analysis is quite troublesome and sometimes could give misleading results if stress categorization is not performed correctly. What I was trying to say is that now that more powerful systems are available it is better to move toward solving these problems by using elastic-plastic method. I agree with TGS4 for using elastic-plastic method for protection against plastic collapse since this way you will not get trapped by wrong stress classification. For the case of protection against local failure again I prefer to go for elastic plastic method since I think the procedure in VIII-2 for protection against local failure using elastic method is very wrong and shall be avoided. for the case of fatigue analysis since there is no need of stress classification and separation of primary stresses from secondary ones the easiest way is to consider elastic analysis. What TGS4 has mentioned above regarding using elastic or elastic-plastic methods looks reasonable to my eyes but may be his reasoning is different.

TGS4, I know that the limit on the primary plus secondary stress is about ratcheting failure. But there is no other limit in Part.4 on the local primary membrane stress. This does not sound good practice.

These rules may not explicitly require a primary local membrane stress check, but I can assure you that it is generally baked into the rules. It IS important.

If I want to follow the rules in Part.4 then I am certain that the thicknesses obtained from those rules are such that the limit on general membrane and bending stresses and secondary stresses are hidden in the formulas provided by the rules. Nevertheless, In article 4.1.6 the allowable stresses are introduced. If the local primary membrane stress is baked into the rules it is better to put its limit in 4.1.6.

4.1.1.2.1 Class 1. When design rules are not provided in Part 4 for a vessel or vessel part, the Manufacturer shall either perform a stress analysis in accordance with Part 5 considering all of the loadings specified in the User’s Design Specification, or, with acceptance by the Authorized Inspector, use a recognized and accepted design-by-rule method that meets the applicable design allowable stress criteria given in 4.1.6. If the design cannot be performed using Part 5 or a design-by-rule method (e.g., creep-fatigue), a design method consistent with the overall design philosophy of Class 1 and acceptable to the Authorized Inspector shall be used.

So suppose that I want to use a design by rule method other than VIII-2 for some detail which is not covered by Part.4. What should I use as the limit on the local primary membrane stress? There is no mention of local primary membrane stress in 4.1.6. The only limiting condition on this stress is that given in 4.1.6.3 which is good for protection against ratchet not plastic collapse.

 

[MrPDes]

My assumption is that a relatively large saddle attachment causes wide ranging 'regional' primary stresses that are not really 'Local' and are also not 'General'. They are somewhere in between. Therefore 1.5×S is too non-conservative and 1×S is too conservative. So they came up with an arbitrary number in between. As with the entire Elastic Method, it is not an exact science.

Pauletea,

Most design methods in Part 4 that I have used have a limit for Local Membrane Stress.
When reinforcing a nozzle, the entire purpose of the nozzle design method is to calculate the Local Primary Stress. This stress is then limited to 1.5×S.

The reason that the saddle design doesn't assess local primary membrane stress is because PL is less than Pm. A saddle consists of a large pad welded to a flawless shell. The pad provides reinforcement to the shell, reducing the primary membrane stress locally (Why would you bother assessing PL around a saddle when you know it is less than the General Membrane Stress?). I would expect an element of Secondary Membrane stress to be present, but PL will be less than Pm. The local primary stresses will be almost entirely Pb (while the vessel is full and unpressurised). When internal pressure is then applied the vessel inflates, causing a decrease in bending stresses and a corresponding increase in and Secondary membrane stresses and PL (where PL < Pm).

All assumptions.

I don't see why stresses during Hydrotest are being compared to Transport stresses. They are completely different circumstances. It is like comparing buckling design to Plastic collapse design. During transport of a vessel, stresses should be calculated as though it is a structure with corresponding allowable stresses for a structure.

 

[Paulettea]

MrPDes, I have never heard of regional primary stress. The code in Part.5 5.2.2.2 (b) introduces a procedure to separate Pl from Pm. Nevertheless, I agree that in some cases the state of stress is a mix of different categories.
If design procedures in Part.4 put a limit on the Pl then it would be better to introduce that in 4.1.6. Why is it absent there?
What you say about Pl being less than Pm is a little strange.
Can you help me understand what a "sustained load" means as opposed to an "occasional load"? What are the implications for each of these terms in designing a structure? Is shipping load an occasional load or is it a sustained one? What about hydrotest load?

 

[MrPDes]

I just invented the term 'regional' stress. I think the size and nature of the saddle stresses are quite different from a typical nozzle 'local' primary stress. As stated, elastic stress classification is not an exact science. There are hotly debated interpretations and exceptions. I may be wrong with my assumptions.

Think of placing a thick Re-pad around a nozzle. If you make it think enough, the Div 2 nozzle reinforcement method can calculate a PL local to the nozzle less than Pm.
If you have a flawless shell with a General Membrane stress of 100MPa and weld a doubler Pad it, then this will act as reinforcement to the flawless shell and therefore reduce the primary membrane stresses around the pad to less than 100MPa. Therefore around the edge of the doubler pad, PL < Pm. (If the doubler pad is the same thickness as the shell then underneath the doubler pad PL =~ 0.5×Pm.)
The presence of the pad will cause local membrane stretching of the shell around the pad, however this stretching will redistribute and is therefore Secondary Membrane stress.
All of those clips and lugs and stiffening rings that are welded to a flawless shell create local areas of PL < Pm.

I would say shipping of a vessel has little or nothing to do with Pressure Vessel design and is therefore neither "sustained" or "occasional" (in a pressure vessel sense anyway).

"Occasional" is a loading condition with a set off allowable stresses. "Hydro Test" is another completely different loading condition with its own set off allowable stresses. See table 5.1.

"sustained" is not a loading condition, it is the nature of a load. Consult an Oxford dictionary for the definition of "sustained". I would associate "sustained" with a non-relenting "load-controlled" load, such as internal pressure during "hydrotest" or "normal operation". When you then add a cyclic load to this there is potential for ratcheting.

 

[r6155]

Please, can you tell us shell diameter, length TL-TL, shell thickness and material?

Regards
r6155