Is it possible to perform an B31.3 SIF determination via FEA shell or brick analysis of a group of fittings? For example, if the branch of a tee is welded directly to the header of a second tee, there will be four end conditions. I am told that WRC 392 (I don't have a copy of it to review yet) only describes SIF determination when loading individual fittings with a maximum of two end conditions altered.

Is there an approach other than WRC 392 for SIF determination?

Can individual SIFs be determined for individual fittings in fitting groups, or must the the fitting group be analyzed by FEA under BPVC VIII, Div. 2 as a whole for each load case? With dozens of load cases the use of developed SIFs in Caesar seems much preferred.

You can calculate an SIF of anything with an FEA program. But calculation is only a small part of engineering.

The fitting combination you describe has a lot of potential flaws.

1. Flow losses in this junction are going to be horrid. If you're trying to divide flow evenly among an inlet and three outlets, you won't.

2. Transient events in any one line are going to interact with joints for all the others. Thus, the accumulated cycles for any of the branches here will be the sum of the cycles accumulated by each of the four lines. This is not going to be pretty if this stuff cycles at all.

3. Depending on the sizes of the lines, you may be making it difficult to execute one or more of the girth butt welds effectively. This can only make (2.) that much worse.

If you have inherited an existing situation and are trying to justify it, please be very cautious. Signing off on something that has been in operation for X hours/days/weeks/months/years simply because it has been in operation for that period of time is a good way to start a second career as an overqualified auto mechanic. (Not that I have anything against auto mechanics.) As numerous other posters have mentioned here over the time I have been a visitor, past performance is no guarantee of future results. Signing your name to such a guarantee is not a wise career move.

If you are designing this and trying to economize on pipe, I sympathize. I am right now trying to explain to a client that yes, it is necessary to provide for the effects of 8" of thermal expansion. He objects because both expansion loops and expansion joints cost money, and is apparently concerned that this may reduce the size of his bonus when the project is complete.

Please, if at all possible, do it right. Have a long, large-bore header with appropriately spaced branch connections. The codes I am familiar with generally assume that the piping layouts they are concerned with will be reasonably sensible.

This sounds to me like it's the piping equivalent of plugging two 3-way plugs into a wall outlet so you can run your TV, DVD player, stereo, computer, printer, and monitor off of the same outlet. It might work, but why try it?

Good advice and both of you I strongly empathize with you!!!

Ken, FEPIPE, Ansys and others all can offer what you are searching for numerically. But at the end of the trip you still may not have the absolutely 100% guaranteed answer!

Why???? Your work may not match any or all imperfections and bumps in the real world that exist... all the way to granular boundaries of the metal.

Ken,

You will find lot of discussion on SIF computation via FEA is this forum. You will also find response of Dr. Charles Becht to this question.

Regards

Craig -

An existing piping system brought these questions to mind. Your admonishments regarding the evaluation of existing piping such as this is appreciated. Thank you for telling me where you have seen the evaluation of this type of situation lead.

"Why try it?" To try to find a solution to a space problem. It is undesirable to install piping this way, but it is also undesirable to the owner to spend hundreds of thousands of dollars to relocate the system from where it has outgrown the space, if the undesirable flow characteristics can still produce acceptable flow and pressure.

John -

Thanks too. But if I had to make a 100% guarantee I couldn't continue my work, it's not perfect. But I think your point is that the FEA shell/brick analysis carries more risk than the beam type analysis of Caesar because the latter has stood more of the test of time within the constraints of the Code and inherently considers more of the imperfections and bumps of the real world than FEA shell/brick analysis.

Anindya -

Thanks for the suggestion, and I had already search and found Dr. Becht input offered in this forum, which was helpful but doesn't get at the concept I am trying to understand as a non-user of FEA shell/beam.

------

So, assuming the old piping is gutted, but the new piping tee arrangement won't allow compliance with B31.3, D300, (13), and a way to evaluate the group of fittings is FEA shell/brick my question is:

Does a proper FEA analysis attach 2 PDs of straight pipe to say, one header of the subject tee and anchor that end, then attach 2 PDs of straight pipe to any other end of the attached fittings in the group and load that end, to obtain a header SIF of the subject tee? And does it not matter which "other" end of the group is loaded, the SIF of the subject tee will be nearly the same?

How can a non-FEA shell/brick user learn the concept of how this is performed?

Ahhh the heart...
"Does a proper FEA analysis attach 2 PDs of straight pipe to say, one header of the subject tee and anchor that end, then attach 2 PDs of straight pipe to any other end of the attached fittings in the group and load that end, to obtain a header SIF of the subject tee? And does it not matter which "other" end of the group is loaded, the SIF of the subject tee will be nearly the same?

How can a non-FEA shell/brick user learn the concept of how this is performed?"

In the original fatiuge tests there was always a piece of straight pipe attached to the tee in the machine I'm sure by now you have a copy of Markls original paper showing this. Why did he do this???? I will venture to guess that he felt that the majority of designs would use tees in this manner... and later on the "infamous" note was added that started down this path.

The increase in stiffness of a flange welded directly to one or more ends of a tee or another fitting which might decrease the tees flexibilty would I believe decrease the SIF.... in general stiffer = lower SIF.

To explore this interaction you would construct your fea model based on your layout and then solve for the "theoretcal" FEA SIF. This you could then I suppose put back into your beam element CAESAR II model.

Tony Paulin as I recall has done some studies along this line with flanges on mitered elbows using his program FEPIPE.

My point I was trying to make to you was... you can get your SIF apply it to your beam element model and still then (if you have my luck) be bitten by a bad tee, bad weld, or some other insane variable (don't even ask me about the so-called temperatures I have been given... safety shower lines at 650F)!

John Breen wonders why I swear so much! Pursue your SIFS via FEA keep us posted.... You could contact Tony I hope COADE doesn't mind but here is his web site.... http://www.paulin.com/prg/News-03-17-06.htm


say Hi for me as well as John Breen!



Post Script.... When a trunion or lug is welded onto a elbow it stiffens the elbow, this does not decrease the SIF but rather the discontinuity between the attachment and the elbow actually increases the SIF rather dramatically.

So this is one exception to my general statemnt above.