Boiler Design – issue 4765

[JasonB]

On JasonB Said:

I have actually posted the long form as well as the short as that gave details of what subjects the qualifications were in.

You yourself posted the link to it earlier in this thread so  not sure why there is a problem with me reposting the link or a screen shot of what is on your forum that you were happy to post the link to.

I would imagine that Neil’s main reason for not including it was simply one of space. You have what is in the short version included in the title photo and the rest would not really have added anything to the content of the article.

 

Before I go any further I detect a bit of a Sock puppet here. Are the two of you posting as one member?

I will also repeat that I am sorry for my minor Dyslexia most people would make allowance for it once aware.

Hi JasonB I can assure you that all the posts up to now on this forum have been by myself only. Les Smith has tried repeatedly to create an account but the system so far has not let him do so. I do hope that Les can resolve this issue and start responding.

I think that his posts will both be in a different style to mine and will add further value to our message. It could be of immense benefit to the model engineering community if people would start to engage with Les in a positive manner on our subject topic. Les has a lot to offer, if only people would listen to him.

I have no issue with you posting information from our forum here on this forum provided that the source is mentioned. Can this be agreed to be  a reciprocal agreement?

[SillyOldDuffer]

On SillyOldDuffer Said:

An technical article isn’t proven correct just because the author has engineering qualifications, and nor is it proven wrong because it disagrees with the readers past practical experiences: experience is unreliable.  Rather technical content stands or falls on it’s merits, not opinion.

I welcome this interesting article and it’s stimulated debate.   Touches on a problem, which is a small-c conservative inspection system that makes it difficult for model boiler makers to innovate in the UK.  Unfortunately, challenging the status quo always upsets people. Here, there’s a rift between modernisers and those who want live steam to be done in the time honoured ways they understand.  The rift might lead to an emotional exchange of schoolboy irrelevances as opposed to a quality discussion.

This is about engineering, not ego.   We’re supposed to be reviewing the article, not rubbishing the author or readers.   Polite review allows mistakes to be corrected.   And if the words and music have been misunderstood, the author is here to explain.  We all benefit from a civilised review, where the rules of debate are followed:

1. Simply announcing an article is wrong isn’t good enough.  Critics have to show why it’s wrong. Experience alone does not explain “why” because experience could be out-of-date, sub-optimal or irrelevant.
2. Don’t criticise authors, criticise what they say.
3. Concentrate on important points, not trivia.  Though it’s easy to get hung up on them, typos rarely matter!

Dave

 

Dave I am delighted to read your post above. It has given me a boost of confidence in the integrity of the forum.

I agree with everything you say and you put it so elegantly. Thank you.

I am sad that the model engineering community – as represented on the MEW forum at least – seem to be missing out on a golden opportunity to further everyone’s (including our) knowledge of model boiler design, construction and testing.

Les, (coauthor of YPM) and I through the knowledge we have and also have gained recently particularly through my multitude of calculations comparing the different methods (YPM, UTS, AMBSC) could contribute a great deal more to the model engineering community if we are allowed to do so.

It seems that this chance to open up the debate that could have led (amongst other things) to deciphering the AMBSC Code, has been squandered through short sightedness of some posters and by closing their minds to anything but the established status quo.

Instead, I fear that the YPM will just fade from people’s memories and they’ll carry on doing it the old ways.

I hope that is not the case and that we can continue to expand on the YPM as published and post more of our insights on the forum without attracting unwarranted criticism.

We are not infallible. Despite rigorous checking a mistake crept into our calculations that we immediately acknowledged ad corrected.

As Albert Einstein said “The man who never made a mistake never tried anything new”.

[Nicholas Farr]

On Nicholas Farr Said:

Hi Dave, my understanding of these replies is about deformation or not in the relative ME&W article. The point being that the yield point of the material is the deciding factor, but of course, that should be of the weakest material in the design. So the design is best if it’s use is in the lowest part of the elastic range in the weakest material. Once a material reaches the yield point, there will be permanent deformation, once that happens, there is no second elastic limit, and deformation will continue to the eventual breaking point, just as the Tensometer graph above shows. I think the science shows why things are wrong.

Regards Nick.

Hi Nick, you are, I believe, absolutely correct up until you say ” I think the science shows why things are wrong.”. I don’t understand that last sentence. What things are going wrong?

By the way, what you have described about the yield point and plastic deformation, to quote “Once a material reaches the yield point, there will be permanent deformation, once that happens, there is no second elastic limit, and deformation will continue to the eventual breaking point is so elegantly written and so clear it should be obvious to anyone reading it should understand what our Yield Point Method article is all about.

So many thanks for that.

 

[SillyOldDuffer]

Hi MEinThailand, maybe science is the wrong word, but may statement was in reply to what SOD said, shown below, and with respect to the Tensometer graph that I’ve shown. I could have misunderstood what he ment with his statement though.

On SillyOldDuffer Said:

 

1. Simply announcing an article is wrong isn’t good enough.  Critics have to show why it’s wrong. Experience alone does not explain “why” because experience could be out-of-date, sub-optimal or irrelevant.

Dave

 

Regards Nick.

[JasonBModerator @jasonb]

I don’t know about not accepting it as I said

I don’t have a problem with the suggested method but with errors it is going to be hard to convince people to change. Even more so when the corrections are made and there is only a few thou difference in wall thickness between the two methods.

I also said early on

It would have been nice to see some actual test pieces and compare them to the theory.

 

Although the authors say at what pressure the barrel will start to deform permanently they do not say by how much, if only a thou or two is that going to be a problem or will it end up looking like a balloon? Paul, is there an allowable deformation in diameter in the Aussie code or must it be zero?

You have not replied with anything that quantifies the results of the two methods. For example, if a boiler built based on the old calculations was tested to 2 x WP and a similar boiler constructed to your method, would we see a 1 thou permenent deflection, a 10 thou or a 100thou. And related to that what would be an acceptable amount of permanent deformation. Those with experience here say you will get some permanent deformation, are they all wrong?

You ended your article with “Why not adopt it for future designs?”  For me personally, I would want something more compelling as the difference is so small even without all the other variables. If we were having lots of reports of boiler failures and your method could be proven (not  just on paper) to produce a significantly safer boiler then I may be more inclined to consider it.

But it is not me you need to convince it is the members of the various committees that write the codes that we have to work with. These are all people who have built many a boiler between them and tested countless others, a good many of them will also have direct experience with pressure vessels. I think one of the main members who worked on the Australian code was a consulting engineer for full size boilers.

[Michael GilliganParticipant @michaelgilligan61133]

I don’t know if this will help clarify things for anyone else, but I think I have recognised a couple of important features of the YP method which may expressed in Plain English rather than Maths:

1. The Yield Point is a “tipping point” at which one characteristic tips over [irredeemably] to another.
2. In the present context, it is an excellent indicator of “incipient” failure … which is the point at which failure is just about to occur, but hasn’t actually happened yet.

Combining these two concepts suggests to me that a sensitive detection of Yielding could provide a powerful tool in the design/development process [rather than specifically an alternative test method]

[ Feel free to ignore my wittering. ]

MichaelG.

 

[JasonBModerator @jasonb]

Quite agree with that Michael. I think I even said somewhere that both calculations could be done and the builder makes the remaining decisions from there.

But one needs to remember that the calculations shown are for the Minimum Thickness. Once all the other factors have been taken into account the thickness of the boiler is unlikely to be at that minimum let alone the few percent thinner than the Evans method currently gives. So the statement “Arguably a safer boiler would result” is very unlikely to happen in real terms.

For that to happen the calculated thickness would need to fall on an available material thicknesses and be taken straight from the pickle to 2XWP test. (I’m happy to accept that soldering could anneal the whole boiler)

Just about anything else will not see a boiler any safer than the current Evans method so that is why I can’t really see and have not been shown anything to convince me of the need to change, what’s the saying? If it aint broke…………

I’m happy to expand on why in practice a stronger boiler is unlikely to occur if anyone wants me to. (sorry Charles)

[Michael Gilligan]

On Michael Gilligan Said:

I don’t know if this will help clarify things for anyone else, but I think I have recognised a couple of important features of the YP method which may expressed in Plain English rather than Maths:

1. The Yield Point is a “tipping point” at which one characteristic tips over [irredeemably] to another.
2. In the present context, it is an excellent indicator of “incipient” failure … which is the point at which failure is just about to occur, but hasn’t actually happened yet.

Combining these two concepts suggests to me that a sensitive detection of Yielding could provide a powerful tool in the design/development process [rather than specifically an alternative test method]

[ Feel free to ignore my wittering. ]

MichaelG.

 

Michael, first, let me quote a classic textbook, Strength of Materials – JP Den Hartog:

“27. Plastic deformation. All the formulae discussed so far were derived on the basis of Hooke’s law and therefore are valid only in the elastic range, i.e., for stresses smaller than the yield stress. In practice, however, beams and other structures are sometimes subjected to loads that would give rise to stesses higher than the yield stress by the existing formulae, and such beams often do not fail but patiently keep on carrying their loads indefinitely.”

This diagram should help:

On the left, in pink we see the layers of molecules in the crystal structure stretched apart elastically. A the top, in red, some yielding has taken place. What happens is that irregularities in the matrix, called discontinuities, start to move, permanently altering the shape of the crystal. When the load is removed, as shown in brown, the elastic behaviour returns the layers into closer contact, but the plastic strain remains.

Subsequent loading and release will go up and down the dark brown line without any further plastic strain, unless the maximum stress at the top of the brown line is exceeded, in which case some more resistant discontinuities will be moved and further plastic strain will take place, creating another line parallel to, and to the right of, the brown one.

An excellent book on how materials behave is designed for non-technical readers and contains very little maths: The New Science of Strong Materials or Why You Don’t Fall Through the Floor – JE Gordon – Penguin.

[Charles LamontParticipant @charleslamont71117]

To continue a little further, plastic strain is not necessarily a problem. Every time you take a new nut and tighten it firmly there will be localised plastic strain in parts of the thread, and probably on the mating faces of the nut and the adjacent part. Failure is nowhere near incipient. Up to a point, it can be just a case of the parts settling into their work and getting to know each other.

[Martin Johnson 1Participant @martinjohnson1]

It seems to me this whole debate is much ado about nothing.  The only thing that Smith and Brown actually propose is using the material’s yield point as the limiting stress without (as far as I can see) any stated Factor of Safety on top of that.  The Barlow formula (not a name I know it by, but no bother) stands regardless.

We haven’t escaped the much stated problem that determining yield point for annealed copper is a tricky business as demonstrated by the spread of figures presented by S&B and in this forum.  This is a physical problem in that we can’t measure the yield stress without yielding the material and if we yield the material, then the yield stress will change due to work hardening.  That is why the yield value for annealed copper is a slippery customer that defies catching.

It seems to me that the underlying assumption that “any permanent deformation in a copper boiler shell during a 2WP hydraulic test is not allowable” is fundamentally flawed.  It is a constraint that does not in fact exist.  As I have previously stated, an FEM analysis by Wim Merks published in ME around 2020 or 21 showed that firebox corners and stays will see plastic deformation in a well proven design.  As has been stated by others in this thread, how is anybody going to know exactly what diameter of shell you started with when you have already done your own 2WP hydraulic before getting a boiler tester involved?  You will be presenting it in a semi work hardened state by that point.  Finally, most telling of all, point to a boiler failure that has not involved poor workmanship in jointing.

Next we have Jason’s and S&B’s figures which show the actual difference in result between the Yield point and UTS based stress calculations is small.  Could this be that the UTS criterion is used with basic FOS of 4 (as far as I know from copper boiler calculations) and looking at S&B’s graph on page 27, annealed copper has a UTS around 30000 psi and a YP of 7500 psi – coincidentally a factor of 4.  Or predictably depending on your point of view.

The “YP method” seems to have an unstated assumption that a cold hydraulic test will always be the worst case.  Depending on working temperature and the associated temperature de-rate in material properties and the number of times hydraulic required by the test authority, this Ain’t necessarily so, as the song goes.

However, the real problem in my view is that the “YP Method” is not a method for designing boilers.  It says nothing about allowances for stress in the seam of a shell, it says nothing about concentration factors for penetrations like domes, clack or safety valves.  It also is not applicable to the tricky bit where shell meets firebox or smokebox tubeplate, is not applicable to the flat surfaces of inner or outer firebox or flues and tubes.  For something that covers all those tricky sums, you need a design code.  To attempt to mix the YP method for the shell with the remainder designed to your chosen code is a short cut to getting your sums thrown back at you with “Do it again” scribbled in red.  (Memories of school days?).

So we come round to Duncan Webster’s view of a couple of days ago that sticking to an established design code such as the Aussie one is the only way to sensibly proceed.  I wish to be associated with the remarks of that gentleman.

Smith and Brown’s article is interesting, but it is not a design method.

Martin

p.s. Just to join in the willy waving as we now seem to be indulging in the sport: B. Tech (1st class) in Mechanical engineering, retired as a C. Eng, F.I. Mech. E.  Model engineer since 1969, designed an 18″ shell steel boiler to BS 2790 and obtained approval from a Notified Body for same for my last project, helped with translating the FEM article by Wim Merks.

[Charles Lamont]

On Charles Lamont Said:

On Michael Gilligan Said:

I don’t know if this will help clarify things for anyone else, but I think I have recognised a couple of important features of the YP method which may expressed in Plain English rather than Maths:

1. The Yield Point is a “tipping point” at which one characteristic tips over [irredeemably] to another.
2. In the present context, it is an excellent indicator of “incipient” failure … which is the point at which failure is just about to occur, but hasn’t actually happened yet.

Combining these two concepts suggests to me that a sensitive detection of Yielding could provide a powerful tool in the design/development process [rather than specifically an alternative test method]

[ Feel free to ignore my wittering. ]

MichaelG.

 

Michael, first, let me quote a classic textbook, Strength of Materials – JP Den Hartog:

“27. Plastic deformation. All the formulae discussed so far were derived on the basis of Hooke’s law and therefore are valid only in the elastic range, i.e., for stresses smaller than the yield stress. In practice, however, beams and other structures are sometimes subjected to loads that would give rise to stesses higher than the yield stress by the existing formulae, and such beams often do not fail but patiently keep on carrying their loads indefinitely

[…]

 

.

Thank you for the quotation, Charles … and for your expert elaboration thereof

I would not presume to argue, but I must mention that we are perhaps at crossed purposes.

I was suggesting that actual, detailed physical measurement of what is really happening might be a useful design/development tool

… the truth of what you have quoted about the existing formulae [my emboldening] can be accepted as a ‘given’ .. it was ever thus!

MichaelG.

.

P.S. __ I can already feel the prickles as the usual suspects prepare to attack me with their comments about Angels on pin-heads !

[Martin Johnson 1Participant @martinjohnson1]

In my post of 13:56, I should also have mentioned that S&B’s method takes no account of axial stresses- that boiler is probably going to have ends which impose a stress in the axial direction.  That axial stress should be combined with the hoop stress and THEN  compared with your chosen yardstick of breakability.

Charles, I was typing while you posted but I agree.

Michael, I explain why you can’t measure yield stress in a highly work hardening material – the measuring interferes with the measured property.

Martin

[JasonBModerator @jasonb]

Yes that Factor of 4 that you mention Martin is really what makes the two so very similar.

Writing out the two formulas side by side with the Evans one on the left and the proposed on on the right.

You can then divide top and bottom of the Evans by 4, rearrange the order and the only difference is one is divided by 10,000 and the other by 9660. Which is a constant 3.5% higher number for the minimum thickness before other allowances.

As has been said in the article and on here that number at the bottom can vary so play with that and the percentage could just a seasily go the other way.

Interestingly as the two authors are working on small traction engines I just looked at Hainings formula as he is probably our most prolific TE designer. If you use his lowest stress value and an average of 8 for the FOS (6-10 suggested) the Minnie would be 0.040″minimum before other allowances so the suggested method would certainly not result in a safer boiler in that case as that returned a thickness of 0.026″.

[Martin Johnson 1]

On Martin Johnson 1 Said:

In my post of 13:56, I should also have mentioned that S&B’s method takes no account of axial stresses- that boiler is probably going to have ends which impose a stress in the axial direction.  That axial stress should be combined with the hoop stress and THEN  compared with your chosen yardstick of breakability.

Charles, I was typing while you posted but I agree.

Michael, I explain why you can’t measure yield stress in a highly work hardening material – the measuring interferes with the measured property.

Martin

It was a long time ago I learned this stuff about thin shells, and it is from memory so correct me if wrong:

• Assuming no stays or tubes, the axial stress is half the hoop stress
• Both stresses are tensile, and they act at right angles
• Radial stress through the thickness is not significant
• so the maximum stress is the hoop stress

[duncan webster 1Participant @duncanwebster1]

With a material like copper which doesn’t have a kink in its stress strain curve. you use the 0.2% proof stress, which is that stress required to cause a permanent 0.2% strain. You can only test the sample once, as it has then work hardened. Actually I think all ductile metals are treated the same, my little book which I got from United Steel in 1974ish, certainly quotes 0.2% proof.

If you use the Tresca theory  for combined stress to cause yield, for two tensile stresses at right angles you ignore the smaller one. See Roark, Formulas for stress and Strain, 4th Edition, Table 2, case 4. I know, it’s not intuitive.

I’ve looked into loss of strength with temperature. According to the Copper Development Association the proof stress at 204C is 90% of that at 20C, and we don’t go anywhere near 204C, for 7 bar (~105psi) it’s 170C. The cold *2 pressure test is the most stringent.

The Australian code is referenced back to their national standard AS 1210-1997-SAA Unfired Pressure Vessels. I based my thinking on BS5500, Fusion Welded Unfired Pressure Vessels and BS2790, Design and Construction of Shell Boilers of Welded Construction, although I haven’t actually seen a copy of the latter, only the excerpts in Alan Haig’s excellent book on full size boiler design. BS2573 part 2 Rules For The Design of Cranes, Mechanisms, is also based on proof stress, as I believe is BS 5950, Structural Use of Steel in Buildings, but I’m less familiar with that, I used to design machines.

I’ve said somewhere in previous posts that I wouldn’t get upset about minor localised permanent distortion in the 2* test, but it’s a brave man who flies against the philosophy of national standards.

I’m not going to get involved in willy waving, it’s a bit crass in a forum like this

[Martin Johnson 1Participant @martinjohnson1]

When I was designing shafts in pursuit of money we used the Von Mises failure criterion which required the torsional and bending stresses to be combined basically by summing the squares and rooting the result (from distant memory).  I accept that the choice of appropriate failure criterion in this case could be quite a complex subject.  Anyway my point stands that it was not addressed in the original article.

I accept Duncan’s numbers in regard to temperature de rating, but again not addressed or stated in the original article.

Any reference to showing off was firmly tongue in cheek.  Too many folk on here appear to have had their sense of humour surgically removed.

Martin

[Martin Johnson 1]

On Martin Johnson 1 Said:

[…] This is a physical problem in that we can’t measure the yield stress without yielding the material and if we yield the material, then the yield stress will change due to work hardening.  That is why the yield value for annealed copper is a slippery customer that defies catching. […]

However, the real problem in my view is that the “YP Method” is not a method for designing boilers.  […]

I have quoted two of your statements there, Martin … NOT because I want to argue BUT because I would welcome further clarification.

I cannot understand why observing the first Yield Point need be difficult … nor can I understand why that knowledge should not be a useful benchmark in the design process.

I fully accept that lots of other stuff happens … but this is [surely?] the start point of whatever cascade might follow.

Obviously one could not merely  plug that information into existing equations, but [to this innocent] that first yield ‘feels’ like a significant event.

MichaelG.

[Paul KempParticipant @paulkemp46892]

Michael,

It’s over 40 years since I used to machine material samples and run them through the tensometer in the tool room for the met lab.  Steels tend to show a very clear yield point on the stress strain curve produced so it is relatively easy and reliable to identify the yield stress.  Copper does not show such a clearly defined point on the curve.  Therefore it is more common to use 0.2% or 0.5% proof stress as an artificial yield point.  This is done by taking the stress strain curve and drawing a line parallel to the clear elastic part of the curve at either 0.2% or 0.5% of the total extension at failure and then reading off the stress.  This takes out the guesswork of identifying on the curve the exact point where elastic deformation becomes plastic.  Proof stress is an artificial yield stress as it includes a very small amount of plastic deformation.

As UTS for copper is pretty easy to determine in a test and Yield point is harder to accurately define I am not sure the YPM is a major improvement.  That said the calculations in the article lead to shell thickness within a few percent of the UTS method then it may be a valid option to determine shell thickness.

I do agree with Martin though that the YPM is not a boiler design method, it is one tool to determine one of the criteria (shell thickness).  There is a lot more to designing a boiler than determining the hoop stress and choosing a shell thickness.

Paul.

[Michael GilliganParticipant @michaelgilligan61133]

Thanks for that, Paul

Clear, concise, and duly noted.

MichaelG.

[Martin Johnson 1Participant @martinjohnson1]

Michael,

I think Paul has done an excellent job of answering for me.  I was just trying to put a mechanism to it that would aid understanding.  The only thing I would add is that you are looking for a 0.2% deviation from a straight line while gazing through a fog of experimental error and the value you are looking for is the intersection of 2 near parallel lines.  So, do you still think you will get a precise figure for YP?

If you look at the graph in S&B’s article, you will see the UTS section of the graph covers quite wide range so picking off the maximum value is a breeze compared to interpreting YP.  Could that be why it has been used for design all these years?

Martin

[duncan webster 1Participant @duncanwebster1]

Why wouldnt you just repeatedly apply and release increasing load until the relaxed length is 0.2% longer than you started with

[Paul KempParticipant @paulkemp46892]

Because each time you apply a stress cycle even below the yield point with copper there will be a degree of work hardening so each time you re-apply load you are starting with a material with slightly different properties to the fully annealed sample you started with.  Look up age hardening of copper.  It will naturally harden with temperature (below the annealing temp), low stress cycles below the yield point, a combination of both or over an extended time.

The change in hardness / UTS might not be very much over a small number of cycles but  it will change and imagine the poor bloke doing the testing – oops I got 0.3% extension can I have another sample please?

To get reliable accuracy of the YPM I would say you need an accurate yield point rather than an approximation?

Paul.

[Author]

Posts