Stress Intensification Factors for Small Branch Connections

[auzie5]

Given

Service: High Pressure Steam
Header: NPS 36"
Branch connection: NPS 2” (unrestrained low point drain)
Type of Branch connection: Weldolet


Background Information

An unrestrained drain is installed at a low point on a steam line. However, this location is close to an expansion loop and is subject to high bending stresses (Sb) in the header. Stress calculations performed using AutoPipe show that the computed displacement stress range (SE) exceeds the allowable stress range (SA). Note that AutoPipe results are as per B31.3 para. 319.4.4 (Flexibility Analysis) and Appendix D (to determine the stress intensification factor, SIF).

A hand check of the equations found in para. 319.4.4 shows that SE is proportional to the SIF when the torsional stress (St) is negligible (i.e. SE=Sb).

However, the SIF taken from Appendix D (Table D300) is independent of the branch to header diameter ratio. This means that a 10" branch and a 2" branch both have the same SIF (3.3 in this case)!!

I do not feel it is necessary to use such a high SIF in this situation since my unrestrained drain line is only 2".

Question:

1. Are small vents and drains (i.e. 2", 3") subject to B31.3 para. 319.4.4 (Flexibility Analysis)?

2. Is there a specified branch to header diameter ratio that is not subject to B31.3 para. 319.4.4 (Flexibility Analysis)?

3. If branch connections smaller that NPS 3” are in fact subject to B31.3 para. 319.4.4, can the SIF factor be adjusted to reflect the branch to header diameter ratio?


Here are some reference paragraphs from B31.3 pertaining to small size weldolet branch connections:

para. 304.3.2 (Strength of Branch Connections): ...There are, however, certain branch connections which have adequate pressure strength or reinforcement as constructed. It may be assumed without calculation that a branch connection has adequate strength to sustain the internal and external pressure which will be applied to it if:

(b) the branch connection is made by welding a threaded or socket welding coupling or half coupling directly to the run in accordance with para. 328.5.4, provided the size of the branch does not exceed DN (NPS 2) nor one-fourth the nominal size of the run.

However, according to para. 304.3.5 (Additional Design Considerations): The requirements of paras. 304.3.1 through 304.3.4 are intended to ensure satisfactory performance of a branch connection subject only to pressure. The designer shall also consider the following:

(a) In addition to pressure loadings, external forces and movements are applied to a branch connection by thermal expansion and contraction, dead and live loads, and movement of piping terminals and supports. Special consideration shall be given to the design of a branch connection to withstand these forces and movements.


As I mentioned above, this location is close to an expansion loop which is subject to high bending stresses (Sb) in the header due to large movements.

Thanks,

 

[BigInch]

Correction: The code does not require small diameter lines be subject to stress analysis, however the more astute engineers recognize that codes only state the absolute minimum requirements and an engineer will overrule those whenever he thinks it is necessary to do so.

Its always the 2" that breaks, hence the reason many company standards require XS pipe for a 2" branch off a 36", even if its not required by a stress analysis.

Is 2" XS too expensive or something?

**********************
"The problem isn't working out the equation,
its finding the answer to the real question." BigInch

http://virtualpipeline.spaces.live.com/

 

[LSThill]

auzie5 (Mechanical)

Question: will your 2" ASME B31.3 be in HIGH CYCLE FATIGUE ASSESMENT OF PIPING SYSTEMS??



L S THILL

 

[DSB123]

auzie5,
If you are the design engineer then it is up to tou to use your experience/knowledge to ensure the system you are 2designing" is fit for service. If you are unable to do this suggest you employ an engineering company that can make such a judgement. After all you can receive different interpretations on the forum but in the end it's your "call" of what to do!!!

 

[simplemath2]

I am just curious what is the thickness of header. For this size of pipe, D/T is normally more than 100 thus out of scope of appendix D.

What is the SIF you get from appendix D? If you think it unreasonale, try to use FE/Pipe or ANSYS to get a closer to reality value.

 

[auzie5]

Getting an SIF of 2.96. D/T is not greater than 100 but I'm interested where you read that D/T>100 falls outside of the Appendix D scope?

My main issue with the code is that it makes no provision to adjust the SIF in Appendix D based on branch-to-run diameter ratio or to the location of the drain opening with respect to the neutral plane (bending is my concern during thermal expansion).

Drains are located on the bottom of the pipe and are therefore centered about the neutral axis with respect to the significant bending moment during thermal expansion. Yet according to Appendix D there is no difference between this example versus a 10" opening located on either side of the pipe (which experience the maximum bending stresses - in compression and tension respectively)!!!

The material in the vacinity of the drain opening (not including the notch sensitive zone) experience only a fraction (16 times less)of the bending stress experienced at at the furthest point from the neutral axis (occurs at the pipe surface on an axis perpendicular to the drain axis/neutral axis)yet I apply the same stress concentration factor (2.96) if I make in opening centered about either axis???

Thanks for the feedback so far...great food for thought

 

[auzie5]

Reply to BigInch,

I think you misunderstand the problem. There are no stress issues in the branch, the problem occurs in the header when I'm forced to apply the SIF from Appendix D at the intersection of branch and run axis. This results is an inflated stress result at that location that doesn't make physical sense since the stress value used for header analysis is taken as the maximum bending stress(felt a point on the pipe 90 degrees away from the drain axis).

Worst of all is that this drain is unrestrained so by adding a weldolet and drain line the header has actually increased in stiffness!

But you're right...an astute engineer should not need to follow the code blindly.

Thanks,

 

[auzie5]

The question of fatigue did come up. But fatigue analysis is typically for 10,000 plus loading cycles. However, I do agree that a fatigue safety factor reflective of the pipe life cycle should be considered.

 

[rneill]

You might want to get a copy of WRC 329 "Accuracy of Stress Intensification Factors for Branch Connections". I don't have this myself, and it is quite old, but I believe provides a good discussion on these issues.

 

[BigInch]

Yes, I did think the stress issue was with the branch. I don't think you actually ever said which component was overstressed.

The result then is caused by too much stiffness of the branch. Increasing stiffness always increases stress, without an accompanying increase in area; no way around that. And a notch always causes a stress intensification, so stresses could increase in both components.

Likewise, I also think that the code cannot be applied "rigidly" to ALL situations and that with a more detailed stress analysis by FEM, for example, showing point values all around the cross section are less than maximum, the standard SIFs do not need to be employed. If due note of the stress intensification at the branch is taken, but bending stress is the problem at a location 90 degrees away, there is no need to combine all stresses at all points comprizing the cross section. I would consider only the corresponding stress combinations at each point. And if that is also the judgment of the responsible and experienced piping engineer in charge, IMO, it can be done.

**********************
"The problem isn't working out the equation,
its finding the answer to the real question." BigInch

http://virtualpipeline.spaces.live.com/

 

[DSB123]

auzie5,
Remember that the Appendix D SIF's are based on full size branches. As you have a small branch on a large header then the SIF's will be different to the Appendix D conservative values especially for the header. You could always model the situation using FE-PIPE if you had a copy and get more accurate values for the header and branch SIF's.
As Simplemath mentions, the closer you are to a D/t of 100 the more inaccuracies you may have with your stress levels. Most pipe stress analysis packages warn of this. Have a look in your Autopipe documentation. Remember you are using "beam" elements for the Stress Analysis!!!

 

[ColonelSanders83]

I would concur with DSB123's assesment, especially the fact that the SIF you are applying are for FULL sized branch connections.

You could rectify the issue by calculating the actual SIF for your exact joint configeration. Additionally to FE-PIPE you could use Paulin research group's FESIF, the cost is reasonalbe and it gives you a rock to stand on when you justify your lowered SIF's.

(no I don't work for them)

Just my two cents worth

A question properly stated is a problem half solved.

Always remember, free advice is worth exactly what you pay for it!

 

[simplemath2]

App D note 1 D/(Tbar)<100

 

[auzie5]

Again, thanks for the continued comments everyone. And my apologizes for not clarifying that my analysis is focused on the header (since the significant bending moment during thermal expansion applies as a torsional moment when analyzing the drain. Torsional stress is not subject to the stress multiplier SIF - and therefore is not an issue).

I am modeling the situation in FEPipe to verify my predictions. My hand calc. shows that the stress in the material around the opening is 16 times less than at the furthest point form the neutral surface. Also, standard notch sensitivity tables recommend an SIF of 2.4 for a d/D of 32"/2" (and this is when the opening is located on the "critical" axis where the highest bending moments occur - opposite of the neutral axis). But even if I did take the SIF of 2.4 (conservative), the local stress in the notch sensitive zone would still only be a fraction of the bending moment felt on the surface of the pipe on the critical axis (which is the value I'm using in my flexibility analysis). This means that the local stress at the opening (multiplied by the stress concentration factor for notch sensitivity) is still less than the stress on the pipe on the critical axis during bending (which should be used as a max stress and should not be subject to an SIF due to the opening). If the local stress at the opening turned out to be the max (after being multiplied by SIF), then I would use that value as the max bending stress in my flexibility analysis.

Does anyone have a take on Note 11 from Appendix D Table D300? The note is 3 sentences long. The first sentence states that for d/D > 0.5 the table may not be conservative. The second sentence discuses a trend of lowering the SIF from the table by increasing the quality of the connection (concave weld vs as-built weld). The third sentence states that selection of the SIF is the responsibility of the designer.

I've argued that the spirit of this note is to allow the designer to adjust the SIF from the table to suit the application. However, others have suggested that if one does not use the SIF as calculated using Appendix D then they have not meeting code.

Any thoughts?

 

[simplemath2]

Very interesting question.

I normally do not include vent or drain this size into CAESAR II model. I might get the same stress amplification or even overstress if I did.

I will go check and let you know.

 

[auzie5]

Thanks simplemath. We typically don't either but since one of our low point drains was being routed to a condensate tank we threw it in to check if we've allowed enough expansion in the drain line to the tank. That's when AutoPipe started red-flagging the opening.

Appreciate the interest.

 

[ColonelSanders83]

The porpose of the lines in the code you have quoted is to give you the lattitude to remove conservatism and/or increase the accuracy of your model. In order to do this you MUST BE ABLE TO JUSTIFY THE USE OF ALTERNATIVE NUMBERS. The code allows us to be engineers, to calculate the actual values, and use them accordingly; although this is not often cost effective, there are always exceptions. I have overridden such requirments before, when given the lattitude to do so by the code. That being said your justification for doing so needs to have a rock solid foundation based on sound engineering priciples and data.

A question properly stated is a problem half solved.

Always remember, free advice is worth exactly what you pay for it!

 

[auzie5]

ColonelSanders83:

I couldn't agree more. Very well expressed.

All my understanding of this problem and the numbers I have manually produced have convinced me that I can justifiably adjust the SIF to suit. However, I will reserve final decision based on validation of my predictions using a FEA program (set up by an independent party).

What I predict is that the model will show an increase stress around the opening due to the stiffness increase introduced by the reinforcement weldolet and the geometry of the drain pipe. The SIF to account for built up stresses in the material adjacent to the opening should not be required since the stress disturbance in the material disappeared once we added the drain assembly. However, this same drain assembly adds stiffness (flexural rigidity = EI) and thus increases the bending stress at the opening in the header. Therefore, a SIF to account for increased stiffness in the vicinity of the weldolet-to-header junction is required. Further analysis is required to develop an accurate approach to calculate this stiffness SIF.

 

[LSThill]

ENGINEERING TOOL TO DO YOUR WORK

http://www.pclgold.com