Get your copy of the code and read it. Also discuss this with your pipe fabrication shop engineering manager. The shop might have some limitations or be able to offer cost savings with one or the other.
Thanks for responding, I have gone through ASME B31.3 and don't see where it explains the reasons for choosing one style over the other (if you know a specific paragraph that would be helpful)
Is it due to strength, vibration, external forces, SIF's, ease of installation, ....?
Most B31.1/B31.3 piping line specifications I have seen only allow these type of connections in cases where weldolets and ANSI tees cannot be used.
Both stub-on and stub-in require more skill of the fabricator and welder.
The ASME B31.3 calculated SIF of stub-ins and stub-ons is much higher than weldolets and ANSI tees, which is significant when a detailed stress analysis is done on the system.
For piping systems, I would use them only in cases of very low pressure service(under 50 psig) or when there is no other alternative.
You are using slang terms. I do not mean that to be disrespectful rather, you should use the terminology of the Code. That way everybody know what you are talking about.
Extruded welding Tee (don't even think about it) Unreinforced fabricated branch connection = stub in Reinforced fabricate branch connection = stub in Branch weld on fitting = weld-o-let (weld on fitting) Welded in contour insert = sweep-o-let (weld in fitting) B16.9 welding Tee fitting = Tee
The further down this list that you go the better is the fatigue life with the B16.9 Tee being the best. If you can't use a TEE and the sweep-o-let is too labor intensive then use a weld-o-let (they are self reinforcing so you do not need to do "area replacement" calculations). For field fabrication of any kind it is important to do good nondestructive examination. Look for weld undercut on weld-o-lets.
Fabricated branch connections (reinforced or unreinforced) are stress risers and it is very important to have a very skilled welder do the work (and NDE is essential). Getting it right with minimal warping takes a steady experienced hand. I would not even think about this route unless it is a good grade of carbon steel (no "exotics").
John: (I was hoping you would reply - I am aware of your credentials/experience - thanks :)
But I guess to look at it from another approach:
ASME VIII-1 Fig. UW-16.1(a) {"stub-on"} versus Fig. UW-16.1(b) {"stub-in"}
Both are exempt from strength calculations as per UW-15(b), so is it safe to say that, as a designer it does not matter, and the question is then really only a matter of convience for the welder?
I think since we assume all the weld metal (that lies within the "reinforcing zone") is part of the reinforcement there is no advantage if the branch pipe protrudes into the hole in the run pipe. I cannot remember reading anything about the relative merits from a fatigue standpoint. Contouring the protruding branch pipe (to keep it out of the flow path) could be a chore. I think I would like to have a reinforcing pad even if not required for pressure design just to have the advantage of the lesser SIF. So the technical issue seems to be moot and the fabricator's opinion on what will facilitate good welding should carry some weight.
(as a side, after discussing with some pipefitters - they said the fit-up of the branch, if left on the outside of the header, would need to be extra careful if the branch was to be used for an insertion instrument or such so that there was no interference on the entire ID of the branch)
we had this problem before, as our fabrication drawings are geared towards stub in rather than stub out. it is really a matter of convenience to the fabricator, welder and to the quality control engineers depending on your non-destructive test requirements. we are now using stub-out. both stub styles are recommended by the Code.
I agree the lowest SIF branch connection is a welding tee, followed by sweepolet, then weldolet, but there are circumstances where a "stub-in" or "stub-on" branch may be selected, such as instrumentation take-offs, hot tap, etc. (Obviously a hot tap has to be "stub on" - unless a saddle or full circle type fitting is being used). A stub-in can be an economical solution for large line sizes, (e.g. in our plants, we have a lot of lines larger than 30 inch diameter, up to 8 feet diameter. On large branches (say 10 inch), a stub-in design may perform better with respect to thermal stresses than for example, a weldolet. But whenever I do a stub-in connection, I try to design it where I have access to get inside and do back weld and radius grind etc. to get better SIF. (I know - not possible on small line sizes!!!)
One thing to consider re: stub-in versus stub-on, is the thickness of the branch connection. In some cases, the design may call for a heavy wall branch connection (such as a "Long Weld Neck" flange). In that case, a stub-on design may require more weld deposit, and therefore more heat input, than a stub-in design.
mayda, If I read your response correctly you seem to be stating two different reasonings. Initially you say "I agree the lowest SIF branch connection is a welding tee, followed by sweepolet, then weldolet..." which suggests that the weldolet has a lower SIF than a stub-in/set-on branch. Then later you say "On large branches (say 10 inch), a stub-in design may perform better with respect to thermal stresses than for example, a weldolet." These two statemnts are conflicting as I see it unless I am mistaken.
I cannot think where a stub-in design has a lower SIF, and hence stresses, than a weldolet - please educate me!
DSB123, Every system is different, and it is not possible to make a statement that will apply to every branch connection. However, I can provide some discussion on considerations which go into the design, and why I have found in some circumstances, a stub-in provides better performance than a weldolet. By performance, I mean safety, operability, maintainability, and overall life cycle cost.
In systems which operate at high temperature, geometry and differential mass play a big role in thermal fatigue life. In those cases, branch connections made with weldolets can be prone to thermal cracking. Example: 48 inch outside diameter piping run, with a 20" OD branch. Design Pressure = 30 psig, Design Temperature = 1500°F (operating at up to 1480°F), material 304H stainless steel. The run could be as thin as about 0.6 inch thick, depending on support system, thermal expansion stresses, mechanical loads, etc.
To meet reinforcement at the branch, one could use a weldolet. It will weigh about 120 lbs., and the material thickness of the weldolet plus weld, at the junction to the run, will be about 1.3 inches.
If a Class 300 Long Weld Neck ("LWN")flange is used for the branch, and the run thickness is increased to 0.75 inch, then the MAWP of the connection is about the same as above. Granted, we have increased the run thickness and therefore the cost, but the 0.6 was a minimum, and would probably need to be increased anyway, to meet all the other loading conditions. In this case, the thickness at the joint will be about 2 inches.
This is simplistic, but as a comparison, the ratio of branch thickness to run for the weldolet example is about 2.17, while for the LWN, about 2.7. Weight of the weldolet is about 120 lbs. It is about 4 inches high. Weight of the LWN, for the same 4 inch distance, is about 30 lbs.
You see the geometry is fairly similar. In both cases, if I have access to the inside of the run, I can radius the corner to get a better stress distribution there. If no access to the inside, then I can radius the LWN before I insert it. The LWN can also be contoured to the ID of the run, so the inside is flush. In the case of the weldolet, as long as it is big enough (in this case it is), I can reach in with a grinder and do some radius work on the run opening, but the results may not be as good as on the LWN. On openings smaller than about 4 inches, it will be quite difficult to radius the run opening for the weldolet design, but again, the LWN can be radiused before insertion.
But the main impact on this particular example, is the difference in weight. If there is any thermal cycling, the weldolet will take longer to heat up, and hold the heat for a longer time, as compared to the run, than the LWN will, due to the 4 times difference in mass.
Over time, and depending on number and severity of cycles, the result can be thermal cracking of the weld attaching the weldolet to the run. The LWN design may also eventually crack, but it will have a much longer service life than the weldolet design. This has been seen in practice, so even though I have not performed FEA on this design, I am pretty confident that the LWN will perform better.
But that does not mean I would never use a weldolet, or only LWN. Each case has to be considered individually.
If there is any thermal cycling, the weldolet will take longer to heat up, and hold the heat for a longer time...
That reminded my of my cold box (low pressure, really low temp) days where we'd use Sch. 10s, stub-ins and slip-on flanges (flanges only where absolutely necessary) for just that reason. I always wondered why sweepolet-type branches weren't preferred for LOX service as that would avoid the eventual hydrocarbon collection in stub-in "sharp" edges, crevices and snag points. Perhaps the additional welding/fitting hassle outweighed the risk after considering the typical strict maintenance requirements for liquid oxygen service.
We have a piping design which has weld'o'let located onto eccentric and concentric reducung butt weld fittings. As this is not sound engineering practice,and B31.3 has no referenct to this type of design has anyone found any written document which does not allow.
Think of a Reducer as just a piece of pipe that changes size from one end to the other. If you can weld an "O-Let" fitting to a piece of straight pipe then you can also weld it to a Reducer.
Actually, the reducer is a fitting. The reducer is not designed for pressure the same way as a piece of pipe (wall thickness per paragraph 304). Fittings do not have a wall thickness specified in B16.9 - they are required to pass a pressure test or be shown to be able to pass a pressure test. So you cannot do "area replacement" calculations for a branch connection. If you take a B16.9 fitting (a piping "component") and alter its design (punch a hole in it for a branch attachment) you have created an "unlisted component" and you must then follow the rules of paragraph 304.7.2 (Unlisted Components).
So, BillyS, the component that you describe is an "unlisted component" and I would suggest reading ALL of paragraph 304.7 and its sub-paragraphs.
Regards, John.
Start A New Thread
Posting in the Eng-Tips forums is a member-only feature.
DESCRIPTION: Pipelines, Piping and Fluid Mechanics engineering technical support forum and mutual help system for engineering professionals. Selling and recruiting forbidden.
Branch Connections