Stresses at intermediate points on pipe

[Mr.Pipe]

Hello Community,

Generally the practice in piping stress analysis is to model the system without intermediate node breaks. There could be moments and forces all along the length of pipe contributing to stresses. Due to this limitation of beam element software's, is there a chance that a highly stressed point is missed in the analysis as no node was created at that point. Any insights on this ?

Thank you !

 

[GBTorpenhow]

No because the program only allows forces moments etc. to be applied at nodes. The way a beam element is defined there is no possibility for 'surprises' along its length in between nodes.

You can satisfy yourself on this by comparing two models, one with and one without many extraneous nodes, you should get the exact same results.

The program's visualization in terms of the color gradient display for stress may differ, like when importing a piping model with extraneous/artifact nodes at tee welds and just being lazy efficient and leaving them in, you get the whole color gradient for the branch stress crammed into a real short element, if you delete the extraneous nodes at the tee welds you get the same stress at the branch but a longer element so the stress color gradient can often look visually different, but the stresses end up being the same.

 

[Mr.Pipe]

Thank you for the response. I will now bother you with a few more questions

1) Do we have any way to technically justify this using principles of beam theory instead of comparing the models ?

2) Some codes such as ASME B 31.1 specifies an SIF of up to 1.9 max for butt welds for pipes with nominal thickness t<0.237 inch or a welding misalignment of up to 0.0625 inch. I am wondering what happens with butt welds added on site(site welds) at an intermediate point to facilitate installation. With an SIF above 1 the stresses reported on these welds would be higher if it was modeled in the software. Is it an industry practice to record such welds on as-built drawings and perform a re-calculation to check these welds ? Some WPS or welding specification would allow a higher misalignemt that could actually trigger the condition to apply 1.9 SIF.

Have a great day !

 

[GBTorpenhow]

1) Non-mathematician hand waving explanation but the argument would be that beam theory doesn't allow for discontinuities in moment/shear except at applied point loads/moments or sharp changes in flexural rigidity or distributed load. The way the program functions these things that can cause discontinuities must occur at nodes. As any section between two nodes must be continuous it is always explicitly solvable, in fact it just requires a relatively simple (for a computer) 4x integration and the solution is just a nice continuous polynomial. Adding extraneous nodes where by the definition of extraneous nothing changes and no point loads/moments are applied doesn't and can't change the solution, it just shows you more points on the same result polynomial.

E-B with constant EI (x position, w is deflection, and q is distributed load):


2) I live in B31.3 land so can't comment on industry practice under B31.1. My completely uninformed reaction would be that yes if you become aware of a situation that would trigger a SIF greater than unity after the fact you would be obligated to go back into your stress model and add a node with the appropriate SIF. I also perhaps cynically assume that in practice this doesn't actually happen a lot.

 

[RVAmeche]

I'd agree - if you somehow knew the system was misaligned you need to include it in the stress analysis. However if it's a random mid point weld where the stress is maybe 15% allowables, your SIF of 1.9 makes it all of 28.5% allowable.

Beam theory has its limitations but by and large works. If you have specific situations that arise, you can go to FEA and analyze a specific tee or component that is having issues. But I think it's still impractical and wasteful to assume the whole system needs to be analyzed with FEA vs beam theory stress analysis or something because "its better".

 

[XL83NL]

I would add that is useful, and sometimes necessary, to add intermediate points. They may help in better pinpointing (and understanding) where and how loads or stresses may peak.
I've had situations where I initially made incorrect conclusions on stress models that didn't use (sufficient) intermediate points. Indeed, the highest stress will not change, but it's important to know where it occurs, and what remedies are best fit for purpose.
PS: don't think it matters but I'm familiar with AutoPIPE, perhaps CII has a different way of visualizing stress concentrations over a pipe.