Applicability of the M.W. Kellogg Methods

[Robster1us]

After reading some opinions in some of the threads in this forum, I would like to get some specifics, if possilbe, on the applicability of the M.W. Kellogg methods. Just in case anyone didn't know, brand new paperback copies of the second addition are available for around $35 on Amazon.

As background, I am just starting to get into the field of pipe stress analysis, and have been recommended the Kellog Book, as well as Peng and Peng's, and Rip Weavers Books, all of which I have purchased.

I have delved into the Kellogg book more deeply than Peng as it appears to have something you can sink your teeth into and find some results (their analytical method), not just explanations of the theories and considerations in pipe stress. While I believe understanding these considerations is the most important part of driving toward accurate results, I don't see how someone could pick up Peng's book, read it, and do a flexibility analysis or really even a sustained loads analysis on even a simple system. There just doesn't seem to be a method there.

Contrastingly, the Kellogg book has step-by-step approaches that (with particularly tedious study of the details, at least on my part) will get you a number, and they even tell you how accurate that number may or may not be. Also, figuring out exactly how they got there and duplicating their numbers is an excellent way to get to understand the concepts, far better than just reading a paragraph about it.

I've seen a number of forum members say that Kellogg is outdated and not of much practical use beyond it's theoretical information and code history, particularly John Breen. What I'm trying to get a feeling for is, in the opinion of the forum members, does this imply that in light of computer technology and finite element analysis, the Kellogg methods are basically worthless? If so, is it then a logical extension of that that if I don't have an AutoPipe or Caesar II license, I might as well not try to predict stresses because there is no good way to do so?

From what I am gathering from the text, it seems very applicable to today's codes, so long as you use material properties, flexibility and stress intensification factors, etc. from up-to-date codes and not necessarily those in the book.

I have enldess questions on this subject, but I'll refrain from asking any more until I get some opinions. Your responses are welcomed, the more detailed, the better.

 

[ColonelSanders83]

The M.W. Kellog book was developed in the age before computers.

You are correct in you assesment of the two tomes, however you must understand that the purpose of the books is different.

The Kellog book was developed and intended to be used by those doing hand calculations, as such many of the methods and layouts are simplified. This will normally drive a conservative design with posibbly more supports and ancors than neassacary.

Peng and Peng's book is geard towards use with modern pipe stress analysis softwares commonly used in industry. If you aren't performing analysis using software I could readily see how it would not seem very usefull.

You can use the Kellog methods to predict the pipe stresses. If you are doing so by hand this will drive you to locate more anhcors in the system until you have something which resemble one of the Kellog method layouts. These stresses will be accurate for the layout in question. You should also keep in mind that the code rules were developed for use with methods like those in the Kellog book (or Piping Stress Calculation Simplified by S.W. Spielvogal).

In my opinion you can design the piping using the rules in the Kellog book, and you can get a code complient system by doing so, however it is very inefiecient to do so. Depending on how much piping you need to design I would reccomend either getting the software or have a specialist contractor perform the work.

Just my two cents worth.



A question properly stated is a problem half solved.

Always remember, free advice is worth exactly what you pay for it!

http://www.ap-dynamics.ab.ca/

 

[Robster1us]

Thanks Colonel for that excellent reply. I understand that, for large systems, the Kellogg method would be cumbersome, although looking at the people I know who have licenses for pipe stress analysis software, I'm beginning to understand more and more that they don't know that they're doing. I hope that understanding the Kellogg method will at least allow me to check simple systems that seem marginal at a glance, and if the time should come where I have the software, at least understand the inputs better for a more accurate result.

I would still like other peoples opinions on the above, but I have an additional question on the method if you are familiar with it. If I'm analyzing thermal stress for a cold system rather than a hot system, it seems as though I need to set the cold temperature as the "ambient" temperature and the ambient temperature as the "hot" temperature. Then at the end of the analysis, one would simply reverse the signs of the reactions.
Because of coordinate sign conventions with the direction of expansion of a "free" end, it would seem to me that trying to use a negative expansion factor ("e", ft/ft) would garble up the analysis and give you incorrect answers. Is my assessment accurate in your experience? Or would I just use a negative "e", follow the sign convention of the movement of the free end, and come out correct? I plan to check the out myself as well, but if you have any experience with pitfalls with the method in this regard, I would love to hear about it.

 

[BigInch]

Depends on the system. It may not be so well defined. If you had a pipe between two 100% full, linear, all directional anchors, what you're saying might be generally true, however the same pipe contracting between two stops would, of course never see any restraining forces at all, so your assumptions could be invalid for any directional type of anchor. Lift-off conditions would not follow your assumption either. A vertical pipe turning onto a piperack may lift off the first support when expanding leaving a long span to analyze, or actually might pull down on the support when contracting. You may also have to look a little farther than the anchoring device itself too, for instance, to what actually the anchor is attached to. A steel pipe attached an anchor at each end, with the anchors themselves attached to a parallel steel beam or column, may not experience any load, if the temperature of the pipe and the beam happened to be nearly equal. A pipeline expanding against an anchor which itself was attached to a "retaining wall" would have a maximum anchoring resistance equal to the retaining wall's active pressure resistance, similar to what a bulldozer experiences when pushing against and slightly lifting a soil embankment, but if pulling away, the pipe might actually accquire an additional load equal to the force resulting from the "fall-back" of the soil against the wall as the wall moves slightly away from the embanking soil, that being the "passive pressure" of the soil on the retaining wall.

"The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward X-CEO BP
"Being GREEN isn't easy." Kermit

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

http://virtualpipeline.spaces.live.com

 

[Robster1us]

Sorry, BigInch, I'm not sure which question you're responding to, the general applicability of the Kellog method or its applicability to pipes cooling down rather than heating up. If it's the latter, are you saying it's not necessarily OK to use the method as if the cold temperature were ambient and then change the signs? Sorry, I'm just not clear on what you're responding to.

 

[BigInch]

#2

"The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward X-CEO BP
"Being GREEN isn't easy." Kermit

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

http://virtualpipeline.spaces.live.com

 

[Robster1us]

OK, fair enough. If I look at B31.5, it appears to be based on the same principles as B31.3, and the Kellogg method, and actually talks about expansion stress in the exact same way. The only clue it gives that it may actually mean pipe contraction is that it gives negative e values in Table 519.3.2. The fact that it is a code for refrigeration piping makes me wonder why they're talking about expansion rather than contraction, but I'm sure we'll never figure that one out.
The negative "e" values suggest to me that you would simply perform the Kellogg analysis as normal, adding a negative sign in for the contraction you anticiapte, per the table. Of course, no support is ever completely rigid, but if your inputs were chosen carefully, it seems reasonable that you could calculate reactions using the method, and at least have a conservative idea of what additional force (aside from weight) would be pulling down on your support of the vertical riser going to horizontal run pipe of your example. That way, the roof stand it was sitting on could be beefed up if necessary, and the structural engineer could get some information letting him know that his joist girder might need to be stronger at that pont to resist the contraction. Is that not fair to say?
In light of the above, doesn't the Kellogg method still work for reduced-temperature pipe contraction? It just seems that you would in fact use the negative "e", and the sign conventions would take care of the rest. My earlier suggestion of making the cold temperature the "ambient" and the ambient the "hot" would not be valid, but you can still use the negative "e" value and get a good answer. Am I still not getting it?

 

[BigInch]

As I thought I already explained, there is nothing wrong with changing the sign convention, as long as you don't have any of the conditions that would cause it not to work. I thought you were asking about cases where your assumpitons about sign or temperature reversals would not work, so I told you some to watch out for.

"The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward X-CEO BP
"Being GREEN isn't easy." Kermit

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

http://virtualpipeline.spaces.live.com

 

[JLSeagull]

Ask yourself this. "Does KBR use the M.W.Kellogg Method?"

 

[BigInch]

I did, but I couldn't answer.

"The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward X-CEO BP
"Being GREEN isn't easy." Kermit

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

http://virtualpipeline.spaces.live.com

 

[Robster1us]

I'm not sure I follow the gist of the comment. However, again, for non-critical service, at low temperatures (below 400°F), and relatively low pressures (and probably some higher pressures too), it seems unreasonable to dismiss the Kellogg method as useless just because a multi-billion dollar global company can afford to buy the software. The more I dig into the method, the more I am understanding about stress modes and their combinations in piping systems. Anyone can own the software, but I'm becoming more and more convinced that there are people out there (who I know personally) who do own it and have no clue if their answers are right or wrong.
I appreciate all opinions, but I am making a sincere effort to understand the limitations. It's not constructive or reasonable to blanketly say that every situation requires a computer analysis, as has been said by Rip Weaver in his books. It seems a little like saying you can't do unit conversions without MathCAD because the table in the front of your textbook only carries things out to 3 decimal places, so you couldn't possibly get an accurate answer by hand.
That said, if I ever become versed and successful enough at pipe stress analysis to afford FEA software to do it, I know I'll have a good basis for determining whether or not the answers are reasonable. Thanks for your comments.

 

[Robster1us]

BigInch, thanks for helping me understand what you were saying. You always provide well-thought-out responses, and I just wish you were accessible as a mentor on this subject. You should consider pipe stress tutoring after school to make some beer money.

 

[BigInch]

Robster, your observations of many pipe stress program users, note I didn't say pipe stress engineers, are consistant with mine (and I'll add that other types of programs are not excluded). What's even more unbelievable is many haven't even read the program's manual, never mind actually read a book on the subject of pipe stress. Read the MWK book, work out as many examples yourself as you can and you'll go a long way in the pipe stress business.

Thanks for the nice compliment, but I think you'll do even better just working out example problems yourself in XL. In pipeline work, stressing out an entire X*E2 miles of line is an exercise for idiots, as most problems can be easily isolated and worked out with XL in an hour or two. If you know when to use the big programs and what makes them tick, I promise that you'll get tired of doing stress analysis at "KB*" before "KB*" gets tired of you.

"The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward X-CEO BP
"Being GREEN isn't easy." Kermit

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

http://virtualpipeline.spaces.live.com

 

[joabjim]

Robster,
I have to calculate some anchor forces due to piping thermal expansion, for a simple piping system, this is a one time job for us. I am struggling with Kelloggs analytic method, I have the book, and I would like to see how you arranged the linear simultaneous equations using the shape coefficients. In the book(example, Page 121)they obviate this step and only show the final solution, it is not clear to me how to set them up, before I start solving the equations. Could You help on this?.
I am using Mathcad to solve these linear equations, and this helps a lot to speed the calculations.
Thanks.

 

[Robster1us]

joabjim,

That, I can certainly help you with. You're right, the book doesn't do a great job of explaining this, and I haven't tried to solve it the way they use (which was all they had 55 years ago). However, the easy way is this:

1. Calculate your shape coefficients, per the method.
2. Set up the equations in excel, with the "constants" of opposite sign as you list them on Form A. This will basically form an array (7X6), if you're just doing a simple system with two anchor points. This means that you have a 6X6 matrix of cells made up of the equations the method gives you (say in Form D-3 of sample Calc. 5.5) with a 6X1 vertical matrix made up of the coefficients, 0,0,0 for the moments and the -EIdX/144 coefficients for the forces.
3. Set all of the elements of the array below the principal diagonal equal to their mirror elements (i.e. make the cell in the 6,1 position of the array equal to the element in the 1,6 cell.
4.Use Excel's matrix inversion capbility to provide a solution matrix. Select a group of (6) cells (I do it vertically) and then in the formula bar, type:

=MMULT(MINVERSE(select your 6X6 matrix),(select your 6X1 matrix))

then hit {ctrl}{shift}{enter} simulataneously. Don't just hit {enter}, or it won't work. You will end up with a 6X1 matrix with the Mx, My, Mz, Fx, Fy, Fz solution values (in that order). These can then be plugged into Form F-1 and the method followed from there.

I found it frustrating that they don't solve all of the columns on Form F-1, just the one with the maximum stress in it or at most, each end. While you can solve each one and confirm they have picked the right one, the book implies to me there is some way to find which one to solve by inspection alone, which if you have little experience, I just don't see.

Anyway, I urge you to set up some examples in excel, confirm you are getting the correct answers, and then analyze your system after you have made all the mistakes gettign to their answers.

Hope that helps.

 

[KernOily]

Just a dumb question from someone entering the discussion late. Allow me to take a different tack on this.

Why are you trying at all to understand Kellogg, Peng, et al? Those methods are obsolete. No one uses them anymore, at least no one that is trying to get the job out the door on budget and on schedule.

Here is what I have told beginners learning and doing pipe stress.

1) Read and understand the Code book. Period. Learn how the allowables are calculated and know how the stress and displacement ranges are calculated.

2) Know your fundamentals down cold, specifically: Mohr's circle, the generalized 3D state of stress and strain, and statics. This is where most beginners get in trouble, in my experience.

3) Read the Rodabaugh papers.

4) Read the pertinent WRC papers.

5) Know the basic failure theories upon which the codes are based, e.g. Tresca and von Mises.

I think your effort to understand the Peng and Kellogg etc. methods is laudable but it should perhaps be left to a hobby effort. It will not help you understand what is going on inside the pipe, it will not help you understand why B31.3 does what it does, and it won't help you get your work done faster. And those three things are what is needed to, as I said, get the job out the door.

Perhaps I misunderstood your question...