Another B31.3 Pipe Stress Analysis Question 4

[Robster1us]

I am a Mechanical Engineering PE in the state of Florida with mostly Industrial Ammonia Refrigeration experience. I have recently had the opportunity to branch out into some other aspects of the pressure piping design world, specifically process piping, and started looking around at various sources of information. I have B31.3 2006 and initially began my search based on the required flexibility analysis.

From other threads on the site, as well as some other sites on the Web, it became clear that simply having a pipe stress analysis package like Bently was a matter of "garbage in, garbage out", and unless one really understands what's going on, it's best to leave this to the professionals (meaning actual pipe stress engineers, not any old yahoo with a Bently or CAESAR II license).

However, you have to start somewhere, so I purchased copies of Rip Weaver's "Piper's Pocket Handbook" and both volumes of "Process Piping Design". They introduced me to my first pass/fail method of culling through piping arrangements for those that definitely have adequate flexibility and those that may need some analysis, an important thing when you could save thousands by honing in on only the items that need analysis.

That's it for the long preamble. My question is this: both books I mentioned have essentially the same exact text on flexibility and minimum leg lengths for the "L" method and its analogs. The chapters also mention guidelines for reaction forces/limits for various equipment types. I am having trouble interpreting these two portions of the text. It would seem to me that piping flexibility is a separate issue from reaction forces at equipment and/or anchor points. My interpretation of the "L" method in the books is that, should you come up with an answer within the criteria, the PIPING is adequately flexible and stresses are pretty-much garanteed gelow allowable from a thermal expansion point of view. However, what does this have to do, if anything, with acceptable reaction forces at anchor points and equipment connections? The text doesn't seem to address how to find reactions, and I must be missing something, but I don't know what it is. Can someone with more experience and familiarity with Mr. Weaver's work please shed some light on this? Am I even approaching this method correclty, i.e. can you provide a better expanation of what "adequate flexibility" means with respect to pipe stress added by the pipe flexing?
Incidentally, as a side question, when he talks about anchoring pipe, what type of support would this be? Is it permitted to weld the pipe itself to a support (I guess if the pipe is not penetrated, it might be OK, at least in Normal Fluid service in B31.3)?

Sorry for the long question, I wanted to give sufficient background for a targeted answer(s). For those who are curious, yes I do realize that the Refrigeration Piping Code, B31.5, requires adequate flexibility as well, but I myself have never performed one of these analyses, and they are uncommon in that industry unless dealing with especially low temperatures or suitability of certain, non-impact-tested materials for lower-temperature service.

 

[StevenHPerry]

Thank you for making an honest effort to learn about piping. I think you'll get some good answers. [I guess we're like the French Canadians. Just say "hello" in french and they'll be happy to talk in whatever other language you prefer.]

Another excellent book that I recommend is Peng & Peng's Pipe Stress Engineering from ASME presses.

High thermal stresses = lack of flexibility.

Equipment loading from a THERMAL perspective is directly related to pipe flexibility.

Imagine two anchors 100 feet apart north south and only 2 feet part east/west and in hot service.

Flexibility: If you pipe 50', elbow 2', elbow 50' the stress will be high. The system is very rigid. All the expansion has to go into that 2' perpendicular leg. Make the routing 25', elbow 10 feet, elbow 50 feet, elbow 8 feet, elbow 25'. Now you have 18' (10+8) of perpendicular leg to take the growth and you have a much better chance of passing thermal stress range.

For anchor loads just think of a cantilever beam, 2' long or 18' long. To take up 4" of growth, the free end of the beam has to move 4". Does it take more force to deflect the 2' long piece or the 18' long piece?

- Steve Perry

 

[Robster1us]

Steve,

Thanks for the very helpful reply. It certainly makes sense to put it in terms of a moment arm being longer to reduce a force on the fixed support, whether it is equipment or other. I would still love some clarity on what those reactions are or how to calculate them, as I can see many instances where this would be useful (i.e. pipe is anchored to roof stand that is anchored to bar joist. Am I going to distort joist when the system starts up?)

Please indulge me a little further if you will. As you explained it, the basic concept is very clear. However, it raises some other questions. I imagine that an unrestrained pipe has zero (or very low, since I guess there is no truly unrestrained piping when considering friction, connections, etc.) stress due to thermal expansion. Therefore, an adequately flexible system would not introduce additional thermal stresses that, in simple terms, would push the pipe stress anywhere in the pipe above allowable. Building on that premise, depending on the other factors (pressure, static loads), how do we know that a little bit is not too much? If the static and pressure loads put the stress right on the edge of allowable, then no matter how flexible the system is, thermal stresses will put you over the top, right?
Where I'm going with this is that it sounds like whatever you do, designing an adequately-flexible system according to simple rules, placing supports in the right spot, etc. in no way gaurantees your stresses will be below allowable. In other words, if you're not sure you have plenty of additional wall thickness, flexible or no, your theoretical stress may be too high. Is that a fair assessment? If so, then what is reasonable? If I follow that logic, a flexibility analysis, while necessary, is not anywhere near sufficient, if you get my meaning, and without gettign too crazy, once complete with the pass/fail method of flexibility analysis, you're still at a loss as to how much is too much thermal stress without knowing your other stresses. (I say "without getting too crazy" because I understand that for low pressures (300 psig, say), most of the time sch 40 pipe has several times the necessary wall thickness so you're probably good). Can you speak to the practical side of this?
If anyone can give guidance on calculating reactions as well, I would appreciate it. Please feel free to respond to any and all of my questions because I am the type who benefits from a broad variety of perspectives and knowledge.

P.S., Steve, I just bought Peng and Peng's book, as well as the MW Kellogg Design of Piping Systems. Pretty reasonable on Amazon. Thanks again. I look forward to perusing both when they arrive next week.

 

[BigInch]

Wow. A good attitude to learning pipe stress isn't so common these days.

The first 10 rules of pipe stress,

Rule #1

Adding a restraint increases stress and forces, but with the advantage of reducing movements. Therefore never add a restraint, if you can deal with the expected movement. A free pipe will expand with temperature without creating stress. A restrained pipe will not expand, but create great stress.

You get a lot more flexibility with beam bending deflection, rather than axial elongation or contraction. Always try to provide as much beam bending as possible. Unfortuanately providing beams for bending is contrary to the ideal pipeline practice, which is making straight lines from point A to point B. Straight line pipe develops high axial forces that are difficult to restrain. If you throw in a few 90s, axial forces reduce considerably, but its true bending stress goes up. The ideal practice is to balance the two, using a pipes cross_sectional area to full advantage for axial stress and its section modulus to full advantage to carry bending stress such that the combined stresses are less than allowables. Then getting the beam deflections to occur in areas that are not critical to movement; ie far away from equipment. Increasing bending stress, reduces those powerful axial forces.

Unfortunately attaching pipe to equipment always seems to be a requirement. Some of that equipment acts like anchors, because they are extremely rigid in relation to long skinney pipe. As a consequence the equipment does not allow movement at the connection and therefore must take the loads caused from completely restraining the pipe.

Some lighter equipment, especially pumps and compressors, are very sensitive to movement, because they easily distort under relatively light loads resulting in shaft misalignments and extreme bearing wear. They don't provide much of an anchor thereby allowing relatively large pipe movements, so it is extremely critical to have very flexible pipe configurations in the vicinity of these types of equipment, so the pump will move the pipe by providing only a slight anchor load, rather than the pipe moving the pump straight into the motor.

Rules 2 to 10,

Read Rule #1 nine more times.

While most anchors and guides and equipment flanges are typically considered to "full anchors", they can have some degree of flexibility, if the equipment is relatively weak, as a pump vs. a fat heat exchanger, so a "full anchor" at a pump will not be nearly so full at a pump and you need to recognize a full anchor at a heat exchanger may provide 20,000 lbs worth of force, while the same full anchor of the same piping configuration provided by a pump flange may only provide 150 lbs, because actually the pump flange will deflect, twist and distort the pump, so its important to recognize the difference of the degree of anchor that whatever is on the other side of the pipe flange can actually provide and actually model that equipment as much as you need to to get a good representation of how rigid that anchor point really is. The software today will let you do that. It wasn't so easy back in the old days where you had a choice of anchor or no anchor.

Some codes allow welding supports to pipe, some don't. Usually its not a good idea and it should be avoided when possible. Pipeline codes do not allow welding directly to pipe. We like flexibility and not welding to pipe is a good way to get some. For example, we would prefer to use a butt stop up of a pipe with a full encirclement wear plate pushing against a chunk of concrete, rather than welding on a wear plate, then welding the plate to the support. Butt directions are chosen to direct movements away from sensitive equipment.

Hope that gave you some ideas about the difference between "anchor-anchors" and "equipment-anchors". "Anchor anchors" from supports, or specially designed concrete anchor blocks are considered to be much more rigid than the anchors provided by equipment flanges.

I always tell people wanting to really understand pipe stress to get a length of stiff copper wire and keep it on your desk. Bend it into a small-scale shape of your pipe configuration. You'd be surprized how pushing the ends together and pulling them apart and watching how it bends can give you a good understanding of how axial force and bending deflection can work together to reduce combined stress. In fact, I've heard rumors that is how they used to do pipe stress back in the really, really old days.

"We have a leadership style that is too directive and doesn't listen sufficiently well. The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.liv

 

[C2it]

Totally agree with the comments above, top marks for aiming to get into stress the right way. Peng's book, excellent. You will find Kellogg a bit long in the tooth (pub. 1955) but still worthwhile.

Not wishing to push their product, good though it is, Coade who own Caesar II hold very good seminars over 4 or 5 days, which give an excellent grounding to stress in petrochem generally. their web site is

http://www.coade.com.

 

[SNORGY]

Further to the above, you will find that almost all stress "problems" - the ones that are more challenging to solve - involve unacceptable forces on connected equipment even though the pipe itself is not dangerously stressed.

This challenge will be exacerbated by people (typically Clients) who will tell you "...The layout is what it is. There is no room to make the changes you want to make. Your calculation must be wrong. Stress engineers are too conservative. We have done this before a hundred times and nothing has blown up. Come up with something that looks like this...".

Agree on Peng and Kellogg.

Regards,

SNORGY.

 

[Robster1us]

Thank you all for your answers and encouragement. I can see this is going to be a lifelong process, but hopefully it won't take that long get some useful concepts under my belt. I'm feeling good that I'm already on the path, with good guidance on where to go next. It sounds like (some of) the details will be revealed in the volumes I've purchased, so I'll start reading when they get here.
I was hoping someone could address one of my more fundamental questions a little more directly. I appreciate the responses on reaction forces, and can see that it's just going to take reading and understanding to ge the hang of those.
I would love to get a more thorough answer to the question I pose above, basically, what good is a pass/fail method for determining which pipes need full stress analysis if you have no information that your pipe wall is conservative with respect to your other stresses (static loads, pressure, and wind/seismic if you have to get into that)? Is Rip Weaver giving us anything of value with his published method? If so, exactly what is it (and what are its limitations)?
Thank you all again.

 

[prex]

I would add a few more practical considerations.
Consider first of all that for most low pressure low temperature piping it's the conformance to previous successful installations that will suffice to justify them.
Concerning the reactions, the only possible method with a simplified approach is the cantilever beam model suggested by StevenHPerry. You have an expansion in the long leg and this is taken in bending by the short leg (and of course this only works for simple 'L' configurations). Elbow flexibility however may add a lot to it, and it should not be difficult to account for it, but I don't know of an accepted method to do so. Also, piping handbooks should have tables to calculate the reaction forces for different typical configurations (I have the Crocker & King, and it does).

If the static and pressure loads put the stress right on the edge of allowable, then no matter how flexible the system is, thermal stresses will put you over the top, right?

Well, that's not really right. Thermal stresses are a very different beast from the stresses due to mechanical loading. The proof is that increasing the thickness will generally increase the thermal stress and possibly also increase the total stress.
Thermal and mechanical stresses should be kept separate from an allowable stress perspective, and indeed, if I recall correctly, B31.3 requires a separate check, where the allowable for expansion is reduced as a function of the actual mechanical stress (but is not zero when the mechanical stress equals the allowable).
The limit on expansion stresses is not given by the usual a fraction of yield criterion, you can go well beyond yield without necessarily harming the pipe strength. Understanding this part of the story is fundamental before going on with any real world flexibility analysis on an expansion critical piece of piping.

prex

http://www.xcalcs.com

: Online engineering calculations

http://www.megamag.it

: Magnetic brakes and launchers for fun rides

http://www.levitans.com

: Air bearing pads

 

[BigInch]

Almost. Just take the simplest case of a straight fully end-restrained pipe.

Increasing the thickness of a straight pipe does not increase axial thermal stress; that stays the same, axial thermal stress is always = [α] * [Δ]T * E.

But increasing wall thickness will increase thermal FORCE, as that is
= x-sectional area * Thermal Stress
= x-sectional area * [α] * [Δ]T * E.



"We have a leadership style that is too directive and doesn't listen sufficiently well. The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.liv

 

[pipesnpumps]

\quote{"what good is a pass/fail method for determining which pipes need full stress analysis if you have no information that your pipe wall is conservative with respect to your other stresses (static loads, pressure, and wind/seismic if you have to get into that)? "}

I can related to be a newbie maybe more than these career types who forget more about pipe stress than I'll ever know..

First thing, if you didn't already understand it, is the bending stress is /independent of wall thickness/.. That is a difficult concept for a beginner to understand, and based on your statement above, I just wanted to highlight that and make sure you were aware.

I take it from your earlier post that you are seeking a threshold for "is there a quick check test for when I should computer analyze for pipe reaction forces?".. You won't get a solid answer on that, but I will advise that if you keep the BENDING stress low relative to the TYPE of anchor point, then it will be fine.. Use 6 ksi max for steel nozzles, and 1.5 ksi max for cast iron nozzles and pumps. For steel and concrete anchors the weak point will likely be the anchor detail used. Use 12 ksi max for process industry type welded anchors. You should always model rotating equipment unless it is very flexible by simple observation. I use an old guided cantilever chart to rough figure the bending stress.. Add up the total amount of pipe in the plane perpendicular to the thermal growth direction, and calculate (look-up on a nomograph) the bending stress.

As for checking the wall thickness to see if it is adequate for other stresses such as pressure, you can do a quick check for wall thickness for pressure using the formula in B31.3. Don't forget mill tolerance and corrosion allowance. The amount of allowable stress allocated by the B31.3 committee for pressure has nothing to do with the cycling bending stress.. The code treats them differently, and they are NOT additive.. If you want to see this demonstrated, then open a model and severely reduce the wall thickness in one your pipes that passed bending stress.. It will still pass the code check, but if you look at the operating condition output the max stress will be too high.. The lesson is design for pressure first, and flexibility second. Pipe stress programs calculate bending stress.

To design for static loads, just go with a minimum span chart and your sustained stresses will be minimal. For seismic or wind put in guides as appropriate. For rough estimating, you can apply a horizontal force at the midspan between the guides and estimate the bending stress (or forces) due to the wind/seismic to decide where to put the guides.

my 2 cents anyway.

 

[StevenHPerry]

Pipesnpumps,

If bending stress is independent of wall thickness, how do you calculate I?

Stress = M*c/I

SP

- Steve Perry

 

[pipesnpumps]

Steve, you are of course correct I should have said "allowable bending stress" is independent of wall thickness.

The main point being extreme newbies assume you can just make the wall thicker and therefore lower the stress and fix the problem.. hardy har har.

for example, for a guided canti-lever,
Se = 6ERd / l^2

R=OD of pipe
l=length of leg absorbing
d = displacement
Se = expansion stress range

 

[Robster1us]

Thank you all again for the replies. I have now received both volumes, although I started on the Kellog book because I got it first and am making my way through it. It's quite interesting to see, as implied by references in the text, how the codes evolved since the 50s.
I am now keying in on an important point that I read above but it didn't quite click in pipenpumps's reply. I will use B31.3 as a reference point.
If I understand correctly, the allowable stress from Table A1 for a material is used for determining a minimum wall thickness (really max stress) due to static loads and pressure. The ALLOWABLE STRESS DISPLACEMENT RANGE is used in a completely separate flexibility analysis, and is at LEAST f(1.25Sc + 0.25Sh) from equation 1a. If Longitudinal stress is not at max allowable (at hot temperature), then this can actually be higher using equation 1b. The two analyses have nothing to do with eachother.
Is that correct?

I am a little confused as to why longitudinal stress is what's focused on in the code for the stresses due to sustained loads. The hoop stress is always twice the longitudinal from the pressure component for a closed cylinder. What I think the reason is for ignoring hoop stress once a suitable pressure-containing thickness is selected is that the only contributing factor to hoop stress (in any significant way) is pressure. So long as the wall thickness can support the pressure, everything after that (sustained loads, moments, etc.) causes an increase in longitudinal stress, making it the necessary focus of analysis for sustained loading allowable stress (as opposed to the hoop stress).
Is that correct? If so, why does the code specifically refer to longitudinal stress when maybe it ought to more appropriately refer to the principle stresses? Don't bending moments introduce shear in some arrangements, making the principle stresses higher than longitudinal and circumferential stresses?

I seem to be replacing every question with two more, but your help is greatly appreciated. I can tell that there is no way to be of any use on this subject without a good deal of reading and the coincedent struggle to understand.

Thanks as always for your responses.

 

[BigInch]

I think you're getting the right feeling. Bending and axial loads don't affect hoop stress, except perhaps by the addition of very minor secondary stresses, whereas bending and axial loads are primary stresses in the longitudinal direction.

Believe it or not, the codes generally like to keep things as simple as possible, hence they focus on limiting each individual component rather than principle stresses, although they will accept a combined stress analysis using Tresca (again prefered for simplicity), or Von Mises, if you chose that combined stress route.

"We have a leadership style that is too directive and doesn't listen sufficiently well. The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.liv

 

[BigInch]

And of course, temperature expansion in the radial direction is considered entirely unrestrained, so there's zero hoop stress from ideal thermal cases in thin wall pipe, just free expansion.

"We have a leadership style that is too directive and doesn't listen sufficiently well. The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.liv

 

[pipesnpumps]

Robster1us, sounds like you are getting it. To prove another point to yourself calculate the allowable bending stress and compare it to the yield stress..

The code allows the pipe to yield due to the thermal expansion. The code thermal expansion stress range is based on 7,000 heat up and cool down cycles. Remember the fatigue life curve from mechanics of materials? That is basically what is going on. Because thermal expansion is "self relieving" stress it is ok to let the pipe yield.. Once the pipe yields the stress "goes away" so to speak...So it doesn't continue to yield the pipe. This is not so for gravity or wind. The good thing about gravity on a simple horizontal pipe run, though, is the pipe will take on a cantenary curve, sagging until it reaches an equilibrium.

It is very important to understand these basic concepts. For example, how do you model wave forces on a pipe? As occasional loads? Ask yourself how many waves will be hitting the pipe..

 

[ZippyDDoodah]

It's interesting that the piping codes recognize and consider material nonlinear behavior in the determination of stress allowables, yet this same material yielding is almost always ignored when calculating piping loads on flanges, equipment, etc.. and also not considered in deflections or dynamic analysis.

 

[BigInch]

What have you got if you consider yielding with deflections, flanges, equipment and dynamic analysis. I mean I'd just hate to get hit by a wildly, swinging, oscillating, vibrating pipe span jumping around with 3 ft amplitudes. Yielding pump flanges don't do much for alignment. Calculating allowable deflections based on yield stress would make drainage a whole 'nother problem. In a few self-limiting cases it makes sense, but those cases are not the norm.

"We have a leadership style that is too directive and doesn't listen sufficiently well. The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.liv

 

[ZippyDDoodah]

BigInch, I'm not sure I follow your point, especially the 3ft deflections you refer to. Could you please elaborate?

You seem to be minimizing the importance of considering material yielding in the analysis. From a practical standpoint, most (all?) pipe stress programs are incapable of considering it so it's not something that is typically discussed. Yielding would typically have a relatively small effect on strain, but it could redistribute loads in ways that cannot be anticipated. I could see how it might also have a significant effect on the modal dynamic analysis.

Although material nonlinear behavior is an effect that is typically ignored, maybe it shouldn't be, especially with nonlinear pipe supports. Do you have first hand experience running side-by-side comparisons? The codes consider it in their stress allowables which tells you it's an important consideration... but it's ignored in other aspects of the analysis. That seems inconsistent and incorrect, even if it's the "norm" on how things are typically done. How big of an impact it would have on a typical piping analysis would be an interesting study. That's all I'm suggesting.

 

[BigInch]

Not at all. Arn't you confusing nonlinear material yield behavior with general nonlinear behavior. Nonlinear behavior of pipes lifting off of pipe supports (load to deflection with changing support conditions) or not does not necessarily have anything to do with plastic yield nonlinear behavior (stress to strain)

"We have a leadership style that is too directive and doesn't listen sufficiently well. The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.liv