A while ago I was asked to check an above ground pipeline model. The pipeline was a 16 NPS High Pressure Steam line (320°C, 11240kPa), several kilometers long with 1000mm of growth into the expansion loops. The expansion loops were typical with the lateral legs at 45° to save material.

The problem I had was that the analyst had put large gaps at the guides near the loops to reduce stresses and structural loads. Typically we specify 6 mm gaps to prevent binding in the guides, but these ones had 50 mm gaps.

A model only simuluates what happens in the first cycle, so the calculated guide loads are 25% lower than with a no gap model. But what happens after the next cycle? The pipe isn't going to return to it's original position is it?

I was able back then to check 3 cycles in sequence using AutoPIPE because you can sequence the load cases in the one of the advanced dialogues. I haven't figured out how to do this in Caesar. I recall after 3 cycles the calculated guide loads were just as high as with no gaps and I want to be able to prove this critical fact with Caesar as I no longer use AutoPIPE.

I've seen too many failures in above ground pipelines as a result of structural deflection and binding in guides. I need to be able to prove my theory to convince future analysts from continuing to use this naive approach to reducing structural loads. I am developing a paper/article/presentation on this subject so the more evidence I can get the better.

Any guidance or references you know of would be much appreciated.

Interesting Paul.

Do you recall if it was just friction that kept the pipe from fully coming back in successive cycles, or was it due to elastic shakedown / yielding in the loop in the first few cycles?

I never was a fan of fine-tuning gaps like that either, if for no other reason there's the chance the field could miss the special gap requirement and just put in a standard guide (or directional anchor, or whatever). I would prefer to use a standard gap and guide the next support back to control the loads, but then your lateral loop displacement increases also.

Hi Dave!

It was just the friction, as it was just an AutoPIPE model I was checking. The non-linear effects of yielding are not simulated. In reality you may be right, there could be non-material behavious excaserbating the issue.

I agree Dave, tweaking gaps to tweak loads is dubious at best, after standing in the middle of an expansion loop during start-up I realized how idealized beam element models are. It's an estimating tool to make sure we have enough flexibility and control to conservatively estimate stresses and structural loads.

What I am really hoping to find out is if there is a way to simulate the behaviour of the pipe in sequences of cycles. Ie., the T2 load case starts with non-linear restraint status of load case T1, and so on.

My "hypothesis" is that after a few cycles the pipe will nearly be sitting next to the guides nullifying the benefit of the large gaps.

I inspected last year several km of steam piping with higher temperature and pressure than your pipes. Sizes 30" to over 40". Lines are some 30 years old. Most of them have 50 mm gaps in guides. I did not find damaged guides or steel. Lines were in correct position excluding cases where designer (or erection contractor) forgot guides. One line was bad. Line had "walked" nearly one m. Guides missing and it was design error.

I assume the old design standard was to have large gaps. For me the system worked well. Guides prevented excessive movements but nowhere were locked. According to CAESAR II a lot of the supports are lifting. Not in reality. Calculation and real life tend to be different.

I did not try to calculate multipple cases. With time we have to do NDT and replicas as the lines are in lower creep range.

Paul, what you're describing is load stepping, which CAESAR II doesn't do. To do load stepping correctly you would also have to have non-linear materials (with plasticity, strain-hardening, and more). I'm not sure if you can predict how much shake down occurs in each cycle without all of this.

this is not to my knowledge done by any beam element analysis program but is found in some 3D FEA packages.

That's true John. I have a general FEA program that has the non-linear materials and contact with friction so I am going to give that a try. Stay tuned, as I find spare time to create the model. When I get some some theoretical answers I post here. Might be easier to build something with copper pipes in my garage though and measure the forces

I would think your garage is the best bet. I have a case where specialist FEA company has spent about 6 months trying to calculate couple ducts. Results are still questionable. OK they have couple of my compensators in the system to make their life difficult.

As you know FEA is not easy. Interesting to hear your results.