Caesar II capabilities to model pipe hanger failure

[lewist57]

Simple question: Can Caesar II model the loss of pipe hanger, which causes a pipe section to drop, generating forces on an elbow without FEA add-ons? And I assume this would be a dynamic analysis, could it predict the amount the pipe would drop due to the loss of the hanger?

 

[curtis2004]

Lewist57,

Why you need this?
If you are afraid that your piping will collapse if some of hangers break, you are underestimating rigidity and stiffness of pipe.

Caesar is only a tool which can be used to model various operational conditions. You have to review appropriate output and will be able to interpret it. For instance, if you see restrains report (whether static or dynamic analysis) that some hanger locations have excessive forces and moments (for example, 1,000,000 lbs force), you can reasonably expect that these hangers will be done during this loading. conditions.

 

[RoboCop16]

Lewist57,

I am sure there are ways to do it dynamically in Caesar II, e.g. by using time history analysis and modelling the support as a force equal to the support force in the corresponding static load case and then suddenly letting the force drop to zero.

A simpler way would be do two static weight-only calculations, one with the support and one without the support. You then take the difference in stress between the two load cases and multiply it by a dynamic factor of 2 and add it to the sustained stress from the standard sustained stress case. This will give you have a conservative, but not too unrealistic, estimate of the occasional stress when the support is failing.

Hope this helps....

 

[uaepiping]

Curtis has given you the information which you have asked for.
" could it predict the amount the pipe would drop due to the loss of the hanger?"
Model it with spring hanger in place. Check the displacement of the point (of interest).
Then model without spring hanger and check the displacement at the same point.
Interpret the result with your experience!

 

[lewist57]

Thanks guys (and gals?). Obviously I don't do this for a living. Here is a restatement of the issue:

A pipe hanger supporting a pipe segment failed, allowing the pipe segment to drop about a foot. The pipe near the hanger did not break. However, a 90 deg elbow at the end of the line (about 14 total linear feet of pipe away from the failed hanger) which came off the main distribution pipe fractured. Our theory is the elbow fractured due to the forces imposed upon it by the sudden drop of the pipe segment. Although in my mind Caesar, CAEPIPE, AutoPIPE et al are design tools for *new* designs, we are trying to figure out if a piping stress analysis program can be used to reverse engineer the failure of the elbow, and show that was the cause of the failure, and not something else. It is probably more simple than I am making it, but I have never used a piping stress analysis program for this purpose.

 

[curtis2004]

Lewist57,

You can analyze existing systems the same manner as "new" systems. Caesar doesn't know (or doesn't care) if the system being analyzed is in a construction stage or existing pipe in 100 years old plant.

One foot displacement seems too big. What pipe size we are talking about? What are spans to next supports, type of supports and fluid inside the pipe.
I think main issue here could be supports were not adequately designed to carry all loading conditions at first place.

Is it someone saw piping moved 1 foot during operations and broke the pipe or it is after the fact when it have broken they measured that the pipe sag 1 foot.

It is not necessarily pipe broke right after hanger was gone. In many instances pipe hangers brake due to water/steam hammer, vibration, earthquake, wind, snow, ice and other loading conditions. It remain undetected or neglected by operations and pipe become to subject to shaking or vibration which over time (many month or even years) brakes the pipe. You can see that by investigating pipe cracks, because fatigue cracks are different then ultimate, corrosion, or brittle fracture.

OK. What is now? A pipe is broke and have to be fixed. What are steps need to be taken?

1. Examine cracks, nearby supports (they could be broken too) and find out what was the main contributor to the pipe failure.
2. Analyze whole pipe line from one end to the other for piping stress for current layout and supports. Consider all operational, upset, occasional conditions.
3. Design new supports and/or layout based on #1 and 2.

There is so much you can do. That's what should be done in those circumstances, IMHO.

Regards,
Curtis

 

[dik]

Did the elbow fail due to torsional loading?

Dik

 

[RoboCop16]

Lewist57,

I understand your question to be whether you can use Caesar II to calculate what the stress the failed elbow would have seen when the support failed. The answer is YES, and I have already outlined to you two alternative ways for you to do it.

It is not clear to me what more you are looking for. We can't do the stress analysis for you. If you wish to pursue this line of enquiry, you will need to put a stress engineering on the job, i.e. someone who understand the principles of pipe stress analysis, including what is meant by sustained, displacement and occasional stress and knows how to set up the load cases. If you decide to go down the time history analysis route, the analyst will of cause also need to know how to set that up.

Good luck...

 

[curtis2004]

"...However, a 90 deg elbow at the end of the line (about 14 total linear feet of pipe away from the failed hanger) which came off the main distribution pipe fractured...

I think elbow fracture is nothing to do with broken hanger. Indeed it seems to me that both broken hanger and fractured elbow is the result of main distribution pipe displacement.
Please provide hand made isometric sketch of location with some notes what is around.

 

[RoboCop16]

Curtis2004,

Your points are all valid. But should we not concentrate on answering the question that is asked in this thread? Until proven otherwise, we must assume that the person asking knows what he is after.

Lewist57 is in this instance asking himself whether the sudden stress from a hanger failing could have caused sufficient stress in the elbow for it to fail. To me that seems sensible part of the investigation. Nobody is claiming that this was indeed what happened or that this is the entire investigation.

 

[curtis2004]

Terp16,

I do not think it is good idea to model broken hanger even if you could somehow model it. It is not necessarily give you accurate or meaningful results. Why not?

1. Even if one of hangers are broken, this will not necessarily will brake the pipe. If you follow code requirements keeping recommended distances between supports it will re-distribute loads to neighboring supports and location will sag bit more.

2. Dynamic analysis is mainly used for analysing dynamic loads like wind/wave, seismic, water/steam hammer, two-phase flow, PSV relief. Restrain report will tell you if any of supports might be gone. I think you analyze temperature, pressure, movements, weight, and other loads to see if pipe, supports, equipment loads are within allowable range or not.

3. Even if you can model "breaking rod hanger" it doesn't mean you can get meaningful and reliable results. What do you want see by modeling this? Let's assume you will get "higher" stress values at location of interest. So what? We all know that rod hanger should have not broken in the first place. We have to find out why it brake. Something tells me that when we find out why the hanger broke, we will find out the cause of pipe fracture, IMHO.

Thanks,

Curtis

 

[uaepiping]

Dear Lewis,
"One foot drop". Was it an impact loading which caused failure of the elbow? Also dropped on what? Did it hit some other support/structure? Or is it a sag in the middle at the hanger location?

I believe the supports in the piping system are not as per recommended pipe support spans!

 

[lewist57]

OK, guys, I am not asking anyone to do this analysis for me, I am asking the following:

1) If a clevis pipe hanger has a nut become loose, and then it drops the pipe about 1 foot, does not the pipe drop event cause a sudden load that could overstress the elbow at the end of the line and cause it to fracture? The fact is that a clevis hanger in the middle of the line dropped the line, and the 90 deg el which was within 12" of the main distribution pipe did indeed fracture. The metallurgical analysis showed the elbow failure was strictly due to mechanical failure, and not due to previous damage or degradation of the elbow. Now before we start the stress analysis, I am just trying to figure out if and how to include the sudden force on the line due to 30 lbs falling 12" would be modeled, or if it is valid to model the force of the pipe drop.
2) The line is 1-1/2" copper, soldered, chilled water, supported every 6' or less with clevis hangers. It was never modeled for stress, but when one models it as installed prior to the event, the elbow has virtually no stress. When you model it with the one hanger missing, the elbow is stressed less than 40% of allowable, but not stressed enough to cause it to fail. HOWEVER, it did fail after the pipe dropped. My *assumption* is that if one were to include the impact force generated by the pipe dropping in the middle that the calculated elbow stress level would be near or above the allowable limit. I am also concerned that the program is modeling the elbow as a simple 90 deg bend in the pipe, as it appears that the program is not set up to enter soldered copper fittings.
3) Although the pipe drop/elbow fracture event did occur, and a review indicates that there is no other credible cause for the pipe fracture other than the pipe drop, my boss wants to "prove" the pipe drop broke the elbow via a stress analysis. Which to me is like doing a finite element analysis of a car after it hit a tree to prove it hit a tree. But he's the boss, and I will follow through. Once again, I have not used a stress analysis program to do this type of diagnostics.

 

[curtis2004]

Why are you ignoring the effect of your main distribution line which is close to your fractured elbow?
Do you know how much this main distribution line moved due to temperature variation?
You have to model whole system (including main distribution line) in order to see if piping was stressed out in the first place.
Can you draw a sketch of you piping?

Curtis

 

[RoboCop16]

Lewist57,

In answer to you questions:

1a) '...does not the pipe drop event cause a sudden load that could overstress the elbow at the end of the line and cause it to fracture?' Yes. It is not very likely but cannot be ruled out with the information we have.

1b) '...I am just trying to figure out if and how to include the sudden force on the line due to 30 lbs falling 12" would be modeled, or if it is valid to model the force of the pipe drop'. I have already outlined for you how the modelling can be done. But - with all respect - the problem seems to be that you are not a stress engineer and do not have the background required to take the advice given to you.

It strikes me by the way, that another and even simpler method would be to take the vertical upwards support force from the normal operating case (let us call it F), remove the support and model the force F at the same location as the support, but going vertically down. The calculated sustained stress will be equal to the occasional stress when pipe 'bottoms out' in its downward drop after the support failed. I just made a test of this in a simple model and with a deviation of less than 0.5 % it it gave the same result as the static method I mentioned earlier. This is not surprising as the principle behind both methods are the same, i.e. that when the support is taken away, the pipe will drop the same distance past its neutral, unsupported position as it was previously held up from that position by the support.

2) '...I am also concerned that the program is modeling the elbow as a simple 90 deg bend in the pipe, as it appears that the program is not set up to enter soldered copper fittings'. I doubt that anyone has developed SIF's specifically for copper fittings so you need to take the standard code SIF's for elbows/bends - which is what Caesar II will do if given the correct input. If the elbow has another wall thickness than the straight pipe, you will need to model that.

3) 'Which to me is like doing a finite element analysis of a car after it hit a tree to prove it hit a tree'. Unless the failed support was carrying a heavy line load (e.g. a large valve) it is surprising that a single support failing will cause rupture of the pipe, in particular in a material as ductile as copper. I thinks it is wise of your boss not to accept that explanation without validation.

 

[lewist57]

To close things out, I finally got the model input, and it clearly shows the fractured elbow was indeed way over the allowable stress limit. And there was no need to impose a 'pipe drop' force or whatever to reach this conclusion. Thanks for everyone's input on this one.

 

[RoboCop16]

Hmmmm, for at calculation to demonstrate rupture for a one-of applied load, the calculated stress would have to be well above the ultimate tensile strength for the material, even with the SIF's set to 1. So if that is what you have, I say congratulations, you have solved the mystery.

I write 'well above' because Caesar II calculates stresses assuming elastic behaviour which means that as soon as the material starts to yield, Caesar II will grossly overestimate the actual stress in the material.

 

[MJCronin]

I always get nervous when a boss decides that someone should do a calculation to "prove" something one way or another.

I think that this is definitive proof of an MBA without an engineering background.....

MJCronin
Sr. Process Engineer

 

[DubMac]

Yup, thats how we got Global Warming........