Branch Connections in Recip Compressor Piping

[StressGuy]

Need a pointer. I'm facing a pipe spec that is being considered for recip compressor piping and it is allowing unreinforced nozzle welds for branch connections on some of the sizes.

Now, I know this is a bad idea due to the highly cyclic nature of the system and I thought that it was either strongly discouraged or flat out forbidden by some kind of code or industry standard. I'm digging through my references but I've not come across it yet. Can someone point me in the direction of a document that addresses this?

Thanks,
--ED Edward L. Klein
Pipe Stress Engineer
Houston, Texas

All opinions expressed here are my own and not my company's.

 

[abeltio]

The code under which the piping is spec'd should have a reinforcement calculation. If the pipe is small enough and the system design conditions are not at the high end (i.e. temp/press) the reinforcement requirements may be fulfilled by the inherent strength of the pipes involved (i.e. the required wall thickness is so small that the excess wall thickness considered as reinforcement is enough to cover the code requirements).
Happened to me many years ago with an ASME Sec VIII div I calculation.
Hope this helps.
Saludos.
a.

 

[StressGuy]

Looks like I finally found what I was looking for:

B31.3 304.3.5 Additional Design Considerations

(b)Branch connections made by welding the branch pipe directly to the run pipe should be avoided under the following circumstances:

(2) where repetitive stresses may be imposed on the connection by vibration, pulsating pressure, temperature cycling, etc. Edward L. Klein
Pipe Stress Engineer
Houston, Texas

All opinions expressed here are my own and not my company's.

 

[MJCronin]

Stressguy,

Excellent question.....

I would imagine that some piping specs developed by a large engineering company have already considered this. I am sure that this issue has been evaluated by the natural gas distribution pipeline companies....

There may also be an ASME paper or Welding Researce Council bulletin developed on this subject..... (I found one on branch reinforcement under external pressure loadings)

Consider contacting/e-mailing John Breen ( a frequent contributer to these forums)

Good Luck

MJC

 

[KernOily]

I've not seen any papers but the issues are several, and are well-documented, which is why most owner's specs, in my experience, disallow unreinforced branch connections ("stub-ins&quot en toto due to the several fabrication and operating issues extant with this type of connection (fatigue, unrepeatable fabrication processes, unreliable or incomplete NDT, etc.).

I don't see how you could justify the use of a stub-in in this service especially given the minimal cost savings of this type connection, for only a few connections on small pipe, vs. an o-let or a B16.9 tee.

Note that the Code said "... should be avoided...". That means you COULD justify the use of a stub-in in this service, but most prudent operators/pipe designers/analysts wouldn't use it. Thanks!
Pete

 

[StressGuy]

Unfortunately, I've seen many owner pipe specs that allow stub-ins in a variety of services, usually in class 150 service. I don't think I've had to deal with stub-in in class 300 and higher.

However, in most of these cases, the piping system cannot afford the SIF of a stub-in and I find that I have to override the pipe spec and call for a reinforced connection anyway.

I agree, I generally at least groan, if not cringe when I see stub-ins on any line that I have to review. But, to see it in this kind of service is just mind boggling. It does seem strange to me that the code actually allows this kind of connection in cyclic service. Of course, it is ultimately up to the designer to do it correctly, but having that opening in the code can make it harder to make an uninformed client understand why a reinforced connection is necessary. Edward L. Klein
Pipe Stress Engineer
Houston, Texas

All opinions expressed here are my own and not my company's.

 

[PAN]

StressGuy,
How about the half coupling for small branch? Is this acceptable in your pipe specification? Please comment.

 

[StressGuy]

I must confess, I'm not familiar with a half coupling. I did a search for them, but the only hits I got were for an inline coupling to join socket weld pipe to threaded pipe, but I don't think that's what you're talking about.

Edward L. Klein
Pipe Stress Engineer
Houston, Texas

All opinions expressed here are my own and not my company's.

 

[inspector03]

Hello StressGuy. Given the fact you don't list main or branch pipe sizes for your project, you may want to visit ASME B31.1-2001 code for your answer. See page 19, Para. 104.3.1 (Br. Conn). See Para. "C" (Pg. 20) Br. Conn. not Requiring Reinforcement. Also, you should consider using Weld-o-lets, given the fact reinforcement is already factored into the design of the fitting. *See Bonney Forge for complete line of Weld-o-lets, Thread-o-lets and Coup-o-lets. They will provide you with installation and welding instructions. Pay particular attention to the welding instructions as, many welders often apply excessive filler metal to these joints if not properly instructed. Weld to the "weld line" not, the "rib line". Hope this helps.
Inspector03 Bill

 

[butelja]

I'd vote for weldolets and then additional plate gussets, especially if it is an instrument connection.

 

[DSB123]

inspector03,
You say that many welders apply excessive weld to Weldolets and the like. This is contrary to what I've come across. In many instances welders do not know what a "weld line" is on an olet and probably 70-80 % of welds have not been welded out to the weld line with the result that there is an inherent "crack" at the toe of the weld. Suggest that if an olet is used for this service then a competent welding inspector is employed to check out the weld profile around the olet to main.

 

Hello all,

The problem here is that we are discussing two very different issues.

The area replacement rules apply to the internal pressure design of branch connections ONLY. These rules do not directly address the fatigue issue that Ed is talking about (there is some S-N consideration in the equation for allowable displacement stress range). If all fabricated branch connections were perfectly welded with a minimum of geometric discontinuities, no metalurgical discontinuities and NO undercut then one would just consult Appendix "D" to compare the relative resistance to fatigue (B16.9 welding Tee, vrs. unreinforced, vrs. reinforced, et. al.). Because of the crotch radii and the minimized geometric notches, the welding tee is the right choice for vibration prone systems. Any fabrication welds will muddy the water. The fabrication of a branch connection by making a "stub-in" will be imperfect and will have a short fatigue life. Remember that each peak-to-peak vibration cycle is a "partial cycle" for calculating fatigue stresses (e.g., displacement stress range). You will get many "partial cycles" very soon when your piping system vibrates.

Look at the B31.3 guidance regarding severely clclic service. And remember the Code is not a design guide, and it never will be.

Best regards from sunny Aruba

John.

 

[DSB123]

Good Point John,
There are many people in the Piping Game who beleive that if you follow the Code then that's all you need to do to ensure a safe pressure system. I also agree about any weld irregularities resulting in weakness. I was trying to make that point in my last post whereby olets are very seldom welded out correctly with a smooth contour between weld and fitting/pipe

 

[helmley]

It is correct that the area branch replacement method is only for internal pressure design. B31.3 requires a stress reduction factor for cyclic loading. If the compressor loads and unloads once a day, consider it one cycle per day, 365 cycles per year and 7,300 cycles for a system life of 20 years. If the compressor turns on/off more than once a day the number of cycle will increase and so will the stress reduction factor. Please note that vibration due to machine operation (rpm) is not considered as cyclic loading, but as a dynamic load external to the piping system.

Please consider the stiffness of the branch connection in question. Consider also the natural frequency response of the piping system. In the dynamic analysis of a piping system, any frequency response beyond 33 Hz is considered rigid. Below this cut-off frequency the piping system shall be dynamically analyzed. Consider also the stiffness of the pipe support, they are to be rigid just the same. Again, any natural frequency lower than 33 Hz is considered not rigid. With all this analysis to be done, it may cheaper in the end to reinforce the nozzle. The spec does not prevent one from doing it.

My understanding is that the spec allows you to use a unreinforce nozzle but does not prevent you from using a reinforce nozzle. It does not say..."Thou shall not".

 

Ed, I'm new at this "forum stuff". However, I have 32 years of piping/fabrication experience which, doesn't make me the best of anything. That said, if you are performing code piping work, I feel that not following the minimum requirements of one of the ASME B31 codes would present you with a great deal of liability or at best, an extensive amount of testing on behalf of your company. I agree, codes are not design manuals. However, they are based on accepted industry practice and proven safety procedures from hundreds of competant construction personnel such as Hartford Ins., Exxon Research & Egr., Fluor Daniel, Bectel, etc.

All ASME B31 CODES allow the use of weld-o-lets for branch connections. *see ASME B31.3, 304.3.5, Para (e)key word being integral reinforcement. Also note that you have one cut to make for weld-o-lets vs. two cuts and preps for tees. One weld vs. two for tees etc. Cost for tees is about double also. As far as welding o-lets, the instructions are very simple to understand, as most manufacturers will provide written as well as graphics on how to install them. Using them prevents us from going through extensive piping system analysis in most cases. The opinions expressed in any and all forums are my own, and not my company's. Thank you, Bill

 

Hi all,

Interesting discussion. Bill, thank you for your input, please come back and enter more discussions.

Well, to summarize what we have learned in this thread, the B31 Codes are not design manuals. However they are based upon theory AND experience and they try to reflect good CURRENT industry practice (not the cutting edge of technology). The B31 Codes DO provide some warning (and in B31.3, rules) about cyclic service, but no specific guidance regarding pressure pulsation or pressure cycling.

Looking at Appendices "D" of B31.1 and B31.3 we can see equations to calculate the Code "stress intensification factors". The SIF's are in the Codes specifically to "adjust" the calculated (by beam theory) BENDING (NOT internal pressure) stresses to address fatigue. The SIF's came from testing that was done in the early 1950's and there have been some updates in the form of "tweaks" added to speak to pressure stiffening et. al. It can be seen from perusing Appendices "D" that as far as BENDING fatigue is concerned, the B16.9 welding TEE is the best choice, followed by the sweep-o-let, followed by the weld-o-let (or approved equal products), followed by the pad reinforced fabricated branch connection, with the unreinforced fabricated branch connection bringing up the rear.

There is currently additional being done for the B31 Pressure Piping Code Committee. This work is under the direction of Glynn Woods and Everett Rodabaugh (Ev was one of the guys who was involved in the 1950's testing). Glynn reported to the Committee that regarding fabricated branch connections, the quality of the fabricating welds was crucial - the fatigue life of a component was VERY dependent upon the quality of the welds. So it would seem that if you must use a fabricated branch connection, construction surveillance is very important - look over the welder's shoulder. Also we should be aware that in fabricated branch connections, some branch size-to-header size ratios are more prone to fatigue failures than are others. Branch to header ratios between 0.5 and 0.7 are in need of special (more rigorous) construction and NDE.

For additional reading see the books by Glynn Woods (CASTI Publishing), and Dr,. Chuck Becht (ASME Publications). Also, you might be interested in these sites:

http://www.hydra-stop.com/pressure/PresEngr.html

http://www.wfi-intl.com/tech1.html

Also be aware that recently Bonney Forge Corporation Acquired WFI International

Best regards, John.