Pipe Stress Analysis 4

[jrfroe]

In our office here, some of our mechanical engineers are looking at purchasing software for analyzing mechanical pipe runs for thermal expansion (along with all the other hydraulic and gravity forces on the pipe). They are looking at purchasing Caesar II by Intergraph mostly, but they are all fairly expensive and they supposedly require several days of training. As one of the structural engineers in the office, I'm being asked what I know about these types of programs, and I mentioned that we already have a finite elements program, Visual Analysis, that could analyze most pipe runs with just a little more time spent inputting the correct layout, materials, cross sections, etc. The only qualification I brought up was that I wasn't sure if I could handle analyzing the stresses in the elbows. I said I could give them the forces that the elbows need to safely resist and if the elbow manufacturer has some guidelines or if there are some industry guidelines for the strength of that fitting, we could get by with the FEA program we already have. But I'm being told that this type of analysis is extremely complicated and it needs a dedicated program to design these piping runs properly, but the way I see it, it doesn't seem too difficult to accurately model a pipe run in an FEA program like Visual Analysis with all the hydraulic and thermal loads and check the pipe for yielding, buckling, etc.

Has anyone here ever used an FEA program intended for conventional structural engineering for analyzing mechanical pipe stresses? Is there anything I may be overlooking about the analysis of mechanical piping runs?

 

[JStephen]

One issue I've run into in the past (actually using VA, by the way) was that finding the forces in the structure is about half the work, and evaluating those forces to see if they comply with a particular design code is the other half. If you are using a design code that is not incorporated into the software, it then becomes a major pain to do the work. So if you are designing to meet a pipe code, using output from a structural software, it could become a major headache just trying to evaluate the output.

Some other questions:
Does VA allow you to use the pipe grades needed?
Are the methods of handling wall-thickness underrun the same in both the VA software and the pipe codes?
How do you allow for internal pressure using VA for pipe? (Including longitudinal loads due to pressure and poisson's ratio effects due to hoop stress)
Do the piping programs allow for varying pressure/temperature along the way?
Is it ever necessary to import somebody else's pipe model, or to export your model for somebody else?
Do the pipe programs automatically allow for water hammer, thrust due to directional changes at ells, or other dynamic effects?
Are the engineers in question already familiar with the pipe programs and not familiar with VA?
Do the pipe programs offer prettier output?

 

[JedClampett]

In nuclear design, the pipe guys used to use a program named PIPSYS. I did a internet search and it is a pipe program, but I have no idea if it's current or it's been absorbed into something else.

 

[LittleInch]

The other issue is that I'm not sure what VA does, but support loads and allowance for movement at different types of supports is part of C2, but the key is that it does all the combined stress for different load combinations and multiple stresses, especially contents weight and internal pressure, which I think VA would struggle / just not do. It also does it for different codes which have different ways of calcualting the same thing with different acceptence limits by way of a simple setting at the input stage.

I don't know any elbow or tee supplier who would state the "strength" of their fitting as "it depends" on lots of things so any ifo he gave would turn out to be wrong in some combination of circumstances. Stress analysis in piping rarely fails due to buckling or yielding, but tends to "fail" at elbows, tees and supports.

There is other software about that does stress analysis, but C2 is definetly a key player.

Your main issue is as I see it is - if you used VA, designed it, built it and then the pipe fractured due to excess stress - you'd be out of a job and the firm would be sued because you used a tool that wasn't designed for the job, but you said it was.

As I say to other things like this, the proof is in doing it. Find a reasonably complex design with elbows, valves, tees, supports and anchors, guides etc which has been analysed using C2 with the output. Model it yourself in VA and then see if you can generate the same answers.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[jrfroe]

Does VA allow you to use the pipe grades needed? I believe they would have pipe grades for most steel pipe, other than stainless. If they don't, it is not too difficult to create a custom material.

Are the methods of handling wall-thickness underrun the same in both the VA software and the pipe codes? I'm not sure... something to consider, but I would imagine that it could be accounted for by either modifying the section or the material properties. In VA, pipe can be input as a standard parametric shape (with just outside diameter and wall thickness) instead of a database shape. If not using a database shape, I would have to do a bit more work on the design side instead of letting VA do the code checks for me, but as you brought up, steel pipe codes might be different than AISC

How do you allow for internal pressure using VA for pipe? (Including longitudinal loads due to pressure and poisson's ratio effects due to hoop stress) hoop stresses can't be applied in VA. Longitudinal loads (whether thermal or hydraulic thrust forces) can be applied fairly easily. The hoop stresses would have to be accounted for by hand, but I would suspect that these stresses would be fairly low for everything we design

Do the piping programs allow for varying pressure/temperature along the way? I would imagine it would. VA could do this as well, but more cumbersome.

Is it ever necessary to import somebody else's pipe model, or to export your model for somebody else? I don't see this being a problem in our office. We've survived this long without any pipe analysis software. Up to this point, as far as I understand, pipe runs have been checked only using old fashioned empirical methods. I've never been involved with the actual pipe design any more than occasionally designing pipe supports where needed, but it would be a structural engineer (not necessarily myself) that would be trained in the pipe analysis software if we purchased it.

Do the pipe programs automatically allow for water hammer, thrust due to directional changes at ells, or other dynamic effects? I'm not sure about dynamic effects like water hammer, but I would imagine they'd account for thrust. Again something to consider.

Are the engineers in question already familiar with the pipe programs and not familiar with VA? No one in our office has used any of the pipe programs. Only the structural engineers in our office have been using VA

Do the pipe programs offer prettier output? Most likely... question is if prettier output is worth a 5-figure dollar amount

Thanks for the questions... that was exactly what I was looking for. I guess now, I'd be looking at using VA with some outside calculations being performed to account for hoop stresses and dynamic forces and possibly some additional code checks.

 

[jrfroe]

Thanks for the quality advice, LittleInch. VA would be able to account for support loads and allow movement at the supports if desired. I have very few concerns about getting accurate support reactions (as a structural its my bread and butter). Most of my concerns are regarding the strength of the pipe and fittings... As JStephen brought up, the piping codes which I'm not all that familiar with come into play. I would have to work closely with our mechanical engineers who are familiar with the pipe codes.

Just curious, how do the tees and elbows generally fail? Obviously it depends on the type of connection, but if its a threaded connection, does the pipe rupture at the reduced section where the threads start, or is there some other failure mechanism?

 

[paddingtongreen]

Dedicated programs take account of decreased/increased flexibility and stress changes of the elbows depending on whether they are being straightened or bent further.

Michael.
"Science adjusts its views based on what's observed. Faith is the denial of observation so that belief can be preserved." ~ Tim Minchin

 

[LittleInch]

Most of my experience is with welded pipe where elbows and tees fail on combined stress usually because of bending stress. Tess usually fail in the crotch or occasionally sheer stress in small connecting pipes attached to larger pipes.

If the pipes you deal with are low pressure, low temperature pipe where hoop stress is low you might get away with it but think of this the other way and how you would feel if a pipe stress analyst decided he could use C2 to design structural steel using columns, aka pipes.....

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[JoshPlumSE]

When I first started out (almost 20 years ago) as an EIT at a large PetroChem engineering company, I took a pipe stress "cross -training" class. Surprisingly, I still have all of the class notes... I might just have a bit of a hoarding problem with technical references!!

As I flip through the notes I notice the following items that could be potential issues with using a typical structural engineering program for pipe stress:

1) Modeling of expansion joints
2) Including flexibility of nozzles in the analysis
3) Including friction loads in the analysis
4) Accounting for Gaps and guides and such
5) Thermal bowing (thermal gradient for partially filled pipes)
6) Anything associated with fluid flow (slug loads, water and steam hammer, et cetera)
7) Material non-linearity or capacity issues associated with high temperatures
8) Variable support spring hangers

Caveat: Some of these items are based off of information from excerpts from a Caesar II User's Guide that was included with the class. That's the program our Pipe Stress group usually used. Therefore, the list I compiled may be biased towards the sorts of things that Caesar advertises rather than what is typically done in pipe stress. I certainly can't tell you what is typical... as that cross-training class was the most exposure that I ever got to the pipe-stress world.

 

[jrfroe]

More good advice. I've heard enough to see that VA might be inadequate in many cases... exactly what I was looking for a recommendation on when posting this thread. Thanks everyone!

 

[TLHS]

Personally, I wouldn't touch this with a ten foot pole.

Firstly, the actual modeling will be a pain in the butt. Pipe models generally require things like supports with variable gap before they engage, can require interesting types of springs for spring supports, friction supports, and all kinds of fun things. These are items that may not be available in structural packages and even if they are will make it difficult to actually make the model converge to some sort of solution.

Then if you actually look at interpreting the results you get into all sorts of exciting things. A lot of piping codes combine stresses in ways I would consider odd. So you'll be allowed to do things like reach a certain percentage of allowable with ring stresses and another percentage of axial load, but then maybe double allowable due to localized stresses in places. Then you'll have different load cases where you might be allowed different allowables for various reasons. With some spreadsheets you can deal with some of this, but you'll have to hack around this stuff, which will leave you open to all sorts of mistakes, plus your software might not even have the capabilities to report some of the items. If you use beam elements, for example, how are you going to model internal pressures or get ring stresses? You're likely going to end up doing some parts of your work by hand still. You'll also have things like stress intensification factors at various fabricated pieces that you can't build in to your model.

The piping codes also aren't written in the most user-friendly ways, so I really wouldn't suggest doing this yourself unless you're going to take a fair amount of training. Instead of having one engineer working on this, it's going to be a structural guy and a mechanical guy overlapping on labour that one person really should be doing.

If you're doing this on rare occasions, you can likely detail your designs so they can be checked by a combination of hand work and something like you're proposing. If this is going to be regular work or you're doing anything reasonably complicated, then the current state of practice really seems to require specialized piping software.

This is also the sort of thing where if this is a reasonable part of your work, the savings you're getting in software licencing probably doesn't justify the dicking around you'll have to do on every project to make a non-piping specific program do the work. If it isn't a reasonable part of your work, could it become a larger part if you become more efficient?

 

[jsha811]

Try Bentley AutoPipe

Moment curvature analysis,reinforced concrete column design and prestressed concrete pile design at

http://www.structurexpress.com