Pipe stress analysis and work hardening of wrought copper elbow 1

[lewist57]

A 90 deg wrought copper elbow had a full circumferential break in the middle of the elbow following the sudden loss of a piping hanger. It has been proposed that the wrought copper elbow had experienced work hardening due to low cycle stresses induced by the copper pipe being too flexible and the piping system experiencing periodic movement events over 10 years. The copper pipe and its support system was to be designed to ASME B31.9 standard.

The simplistic question I have is - what should value or characteristic should the original designer looked for in the original piping stress analysis and support system layout to flag for the potential of work hardening fatigue of the elbow due to low cycle pipe movement over a long period of time? It has been a long time since I have done piping stress analysis, and have never analyzed an ambient temperature copper (Type L) pipe system, just carbon steel steam piping to B31.1 or B31.3.

 

[racookpe1978]

I have not worked with copper in a vibrating pipe stress system.

It will be interesting to what other people's experiences are. I HAVE had copper hot water pipe erode through with a pin-point leak due to corrosion where the copper pipe was likely touching an adjacent steel support. 28 year service history at the time of the leak. To date, no copper pipes in any system I am familiar with has failed due to cyclic stress.

 

[RoboCop16]

Hmmmmm, interesting case, but I think we will need a bit more information.

- What makes you think that this is low cycle fatigue? How did you rule out other possible causes, e.g. high cycle fatigue or sub-standard (brittle) material?

- What kind of cyclic load do you have in mind? You mention that the system might be too flexible. But that would mean that displacement stresses from temperature expansion or other types of forced displacement will be low, not high. Low flexibility can lead to vibration, but that would normally be high cycle, not low.

- Is the system exposed to external forces, e.g. high wind or safety relief valve kick forces?

- Is it confirmed that the system has been stress analyzed? If the system is ambient temperature as you mention and for building services as indicated by the use of the ASME B31.9 code, the system may have been considered low criticality and not analyzed.

As to your question regarding stress analysis: Low cycle fatigue would most likely be caused by temperature expansion or forced displacement in a system that is too restrained to absorb the displacement. Or alternatively, by repeated application of some external force. In any case, it is standard part of the stress analysis to identify to such loads, and make sure that the stresses stay below code allowable so that fatigue cannot occur.

 

[RoboCop16]

Forgot to ask what service the pipe is in and whether it might be subject to surge or water hammer?

 

[MikeHalloran]

Aren't wrought copper elbows already work hardened just by the forming process, before they're installed?



Mike Halloran
Pembroke Pines, FL, USA

 

[lewist57]

OK, all good questions. Here's more details.

1) The fluid service is chilled water at 40 degF. System internal pressure is about 60 psi, ambient temperature is 72 degF. There are no known conditions that would cause significant thermal cycling of the line.
2) It is unknown, but assumed that a piping analysis was done for the pipe segment 10 years ago.
3) This is indoors, so no wind, no snow, etc. No relief valve. The 1.5" pipe is a fill line, and provides chilled water to a tank. The fill operation is not constant, and is known to be very infrequent, say once a quarter, as this is a manual operation.
4) Water hammer in the entire system is an assumption that we are trying to determine if it exists or not. We have some trend data that implies water hammer spikes in the mains, but we are running that down in a separate project. My personal assumption is not that the water hammer pressure spike blew out the elbow, but rather it rattled the line, which caused a nut to fall off the clevis hanger, which then dropped the line about a foot and created enough force on the elbow to cause it to fail.
5) It is my limited understanding that thermal or mechanical cyclic stress will cause work hardening of the copper. For 90 deg elbows, this shows up as the inner radius being harder than the outer radius. We are confirming this by Rockwell hardness test on the elbow.
6) Our working theory is the elbow was exposed to mechanical cyclic stress (I don't know if it was high or low). The pipe was supported only by clevis hangers, which allowed it to move in the horizontal plane. The elbow in question was pretty much anchored, in that it was less than 1' off the top of the 6" horizontal main CS pipe. The other end of the pipe is anchored as it is attached to a tank. It had been reported by plant personnel that the line would "shake" when the tank was being filled. The line is in a part of the plant that is not normally occupied, so we can not prove or disprove how much or how often the pipe would move over a 10 year period.
7) No evidence of erosion, corrosion, stress cracking, wall thinning, mechanical damage, or manufacturing defect in the elbow. No embrittlement of the material.

So now, the overall theory is that the elbow had been work hardened via some cyclic mechanical stress over 10+ years. Then, a clevis hanger in the middle of the pipe failed (perhaps a water hammer event occurred in the main that shook the nut off the hanger), dropping the pipe and causing a large, sudden force in the same plane as the elbow, causing it to snap. If this is true, then the physical properties (ie - hardness) of the elbow changed over 10 years. Therefore, if I were doing the piping stress analysis 10 years ago, what calculated value would have warned me of the potential that the elbow would be exposed to enough cyclic movement to cause work hardening, and thus the change in the physical properties of the elbow, leading to its failure after the failure of a pipe hanger? My initial reaction is that this sequence of events was too unlikely for the original designer to consider to be a credible scenario. Thoughts?

 

[RoboCop16]

Thanks for the additional info, Lewist

I think we need to get the terminology right:

Work hardening is an increase in yield strength caused by plastic deformation. Loads that results in a stress higher than the original yield strength will work harden the material. Since the yield is increased and the ultimate tensile strength remains the same, work hardening cannot in itself be an explanation for the failure, but the material will be more brittle (less ductile). Work hardening will happen also for a one-off application of the load – there is no need for the load to be cyclic.
It seems to me that you are confusing work hardening with fatigue.

Fatigue is crack propagation in the material and is caused by cyclic loading. If the fatigue is caused by stress above yield, it is called low cycle fatigue, otherwise it is high cycle. Yes, in low cycle fatigue there will be work hardening of the material during the first cycles, but the work hardening is not what causes the material to fail, it is the gradual crack propagation.

As for the original stress calculation: Unless there are some extraordinary circumstances, a 1.5” water line working at near ambient temperature and at low pressure will be considered uncritical and not be stress analyzed. And even if it was, it is improbable that the shaking (surge?) load that you mention would have been included in the analysis. Also the sudden impact from a support failure and subsequent drop of the line would not have been considered.

I agree that it is surprising that the elbow failed, as copper alloys are normally ductile. So maybe the ductility of the original material was sub-standard (it is not clear to me how you have ruled this out), or the elbow was weakened by low cycle fatigue, or maybe the increased brittleness from work hardening was a factor, as you suggest - or maybe there is some other cause. If it is important for you to get to the bottom of this, I suggest that you obtain a metallurgical examination of the cracked material.

 

[lewist57]

OK, update:
1) Thanks for your clarification of terminology. I am a Mech Engineer, not a materials person. I remember taking a materials class in college and making an "A", but obviously I have forgotten more than I have remembered in the intervening 30+ years.
2) Yes a metallurgical analysis of the elbows is on going. The initial report was "no sign of fatigue" but we have asked them to take a look for work hardening, or any other degradation of the material. Erosion and corrosion have been ruled out.
4) No, a stress analysis was not done for this piping system, and was not required by B31.9. No need to beat up the original engineer because he was not compelled to do a stress analysis, or anticipate any of the potential failure mechanisms that may or may not have affected the elbows.
5) Our focus now is the metallurgical analysis, we want to either confirm or rule out that the elbows had not degraded or had been compromised from what you would expect for a 'new' B16.22 elbow. If the elbow was somehow compromised, then we will have to look at all of the other similar fittings in the system to see if they are potentially compromised as well.

Thanks for everyone's input to date!

 

[racookpe1978]

Dropping a pipe 12" due to a hanger failure (lost nut, or busted hanger embed or busted hanger itself, doesn't matter) means the single elbow now in question is trying to "hold up" the entire previously hung pipe system AND all of it hangers and supports!

It "could of" held up under normal conditions - probably would have! - if nothing else happened and the rest of the pipe network was held up by other hanger and members. But, as usual, something else did happen and the single elbow was left trying to support the whole pipe.

Too much load.