Boiler Design – issue 4765

[DiogenesParticipant @diogenes]

re. WS – shouldn’t that be working pressure rather than test pressure? ..D x 110 rather than D x 220?

[MEinThailandParticipant @meinthailand]

Yield Point Method – AMBSC Code Uses it!

The Australian Miniature Boiler Safety Committee(AMBSC) Code Part 1 Copper boilers, on page 13 of Issue 8 – 2012 provides a formula for calculating stay pitches and diameters.

The formula uses the maximum allowable stress as that for annealed copper, 26,000 kPa (3,771 psi).

The code does not give a formula for the thickness of boiler shells but if it uses the value for annealed copper for stay design, wouldn’t it be incongruous if it used the UTS value of hard copper, plus a 6 to 10 times ‘safety factor’?

Perhaps someone with an inside knowledge of the AMBSC Code could clarify?

[Diogenes]

On Diogenes Said:

re. WS – shouldn’t that be working pressure rather than test pressure? ..D x 110 rather than D x 220?

No. The maximum stress a boiler will (normally) be under is during the hydrostatic test = 2 x Working Pressure.

[DiogenesParticipant @diogenes]

Apologies, I misunderstood the ‘W’..

[MEinThailand]

On MEinThailand Said:

The UTS Method Explained – Part 3 – Safety Factors

The factor of safety (FoS) in engineering, particularly in the design of steam boilers, is a measure of how much stronger a system is than it needs to be for its intended load. It is defined as the ratio of the ultimate strength (or failure stress) of a material to the allowable stress (or Working Stress – WS) applied in operation.

Mathematically, it is expressed as:

Factor of Safety = Ultimate Strength / Allowable Stress

Example 1
From the article “The Yield Pont Method ” by Les Smith and Alan Brown, published in Model Engineers & Workshop magazine in Volume 134, Issue 4765, June 2025 edition, page 26.

Taking a known published design for a G.W.R. boiler for a 5” gauge loco from Page 150 of The Model Steam Locomotive as an example, where:
Diameter, D = 4.75″
Working Pressure, WP = 110psi
Test Pressure = 2 x WP = 220psi
Shell Thickness, T = 0.092″

Using Barlow’s Formula. S = (P x D) / (T x 2)
WS (Stress) = (220 x 4.75) / (0.092 x 2)
WS = 5,679 psi

Ultimate Strength for annealed copper = Yield Point for annealed copper (to avoid permanent distortion) = 4,830 psi

Factor of Safety = Ultimate Strength / Allowable Stress
Factor of Safety = 4,830 / 5,679
Factor of Safety = 0.850

Yet the UTS Method proclaims a FOS of 8 (Average)

Clearly the UTS Method is both incorrect and misleading, because it does not allow for the Yield Point of annealed copper.

What you explain is how YOU back calculate a FOS in a given case against a given material property.  What you don’t explain is why passing yield or proof stress is deemed a failure.  The fact remains as proven by others on here and best part of a century of practice that either method results in a similar and safe design when used with the appropriate Factor of Uncertainty.  I use my term deliberately having explained how this mis-named FOS is actually used.

Martin

[Martin Johnson 1]

On Martin Johnson 1 Said:

What you explain is how YOU back calculate a FOS in a given case against a given material property.  What you don’t explain is why passing yield or proof stress is deemed a failure.

Which is exactly where we started 150+ posts ago.

[JasonBModerator @jasonb]

You don’t really need any qualifications to see that the FOS is just about cancelled out by the larger stress figure being used in the UTS equasion as the eight is above the line and the stress below which is why the two methods give a very similar result and why the UTS has served us well for years.

All you need to do is divide top and bottom by eight to get rid of the FOS and what does that leave at the bottom……………….5000 rather than 4830 of the YP method.

[duncan webster 1Participant @duncanwebster1]

If we believe MEinT’s figure for proof stress 4830 psi (please stop calling it yield), then boiler shells built to all the other criteria will all take on a permanent swelling at 2* pressure test, where they could be at 8000 psi (JasonB’s working stress of 4000 psi * 2). I’ve never seen one, and I’ve been involved in model engineering for 40+ years. It would be obvious, as the end near the tube plate is reinforced and so would not permanently deform. I’ve found another figure here which is for C12200, which claims to be equivalent to C106, which is what we in UK will use. It quotes 0.5% proof stress at 10,000psi. My figure was 0.2%, so I’d expect it to be a bit lower. As others say it wouldn’t matter if you did get some permanent set, but UK national standards are built around staying below proof stress.

If we adopt MEinT’s figure, all we achieve is more expensive boilers, not more safe. I very much hope that subsequent instalments of this saga will remain in the unused pile.

 

[JasonBModerator @jasonb]

I think you will be Ok there Duncan, chatting with Neil even he and Diane were unsure of publishing it.

[JasonBModerator @jasonb]

It should also be bourne in mind that this is a calculation for teh MINIMUM thickness and there are a lot of other factors that will go into arriving at the final tube thickness, to name a few.

– use Factor – will the boiler be a structiural part in which case you may add 20-25%

– Joint Factor – will there be a butt joint or tig welded copper joint, piercings such as bushes and domes add 20%+

– Construction considerations – will you need thicker tube where the barrel meets the firebox, will the barrel be unwrapped to form the firebox sides 10-20%

– nearest Available – Always go up in size to the next available which is becoming less as 13g has all but gone

All of these are going to increase the minimum calculated using the UTS method by more than the 3.5% difference between UTS and YS method so what’s the point in changing?

[duncan webster 1Participant @duncanwebster1]

You dont need to increase the thickness for domes, bushes etc, you just add local compensation, all detailed in BS and Aus code. For structures, which I’m more familiar with, a properly performed full penetration weld is reckoned to be as strong as parent material,  but Tubal Cain states a reduction to 80% for but straps in silver soldered boilers

[JasonBModerator @jasonb]

Ah, I was thinking of Haining but that was under steel boilers where 1 is used for an unpierced seamless shell and adjusted for holes and joints. Copper he only mentions joints, 0.8 (80%) for soldered or welded and down to 0.5 (50%) for rivited but unlikely to see that used now.

Would “localised compensation” be a thickening plate or similar?

 

[duncan webster 1Participant @duncanwebster1]

If its a small bush, the bush itself is usually enough (but needs checking) for a dome hole it can be a doubling plate around the hole. Best get a copy of the code.

Traction engines  are another thing altogether,  working to a lower acceptable stress for thd bits that ard loaded by the engine parts would be one way.

[Martin KyteParticipant @martinkyte99762]

Just as a point of interest (as I happen to be reading a book on the subject) the navy design office was operating on a safety factor of between 1.7 and 1.85 for it’s A class and later Submarine pressure hulls dependent upon steel grade. I guess they were doing a great deal more calculations than the average steam boiler design but given that an average crew was around 60 blokes the consequences of failure seem rather more extreme than one of our models ‘letting go’
I’m not in the least suggesting we don’t take pressure vessels seriously but the low failure rates seem to indicate we are well positioned n the cautious side.

Keep doing the calcs.

regards Martin

 

[duncan webster 1]

On duncan webster 1 Said:

If we believe MEinT’s figure for proof stress 4830 psi (please stop calling it yield), then boiler shells built to all the other criteria will all take on a permanent swelling at 2* pressure test, where they could be at 8000 psi (JasonB’s working stress of 4000 psi * 2). I’ve never seen one, and I’ve been involved in model engineering for 40+ years. It would be obvious, as the end near the tube plate is reinforced and so would not permanently deform. I’ve found another figure here which is for C12200, which claims to be equivalent to C106, which is what we in UK will use. It quotes 0.5% proof stress at 10,000psi. My figure was 0.2%, so I’d expect it to be a bit lower. As others say it wouldn’t matter if you did get some permanent set, but UK national standards are built around staying below proof stress.

If we adopt MEinT’s figure, all we achieve is more expensive boilers, not more safe. I very much hope that subsequent instalments of this saga will remain in the unused pile.

 

Duncan, I do apologise for not responding until now but I have been on a trip to Hong Kong for my son’s engagement party and whilst there it was a delight to travel on the old trams, a British legacy from 1904.

It’s one of the few tram systems in the world that has continuously operated electric double-decker trams since its inception.

Against a back-drop of modern high-rise skyscrapers it reminded me of the Luddite movement that began in 1811 in Nottingham who organized raids to destroy the machines and sometimes factories themselves in resistance to industrialization and technological change.

Much like the Luddites who, sadly, seem to dominate the model engineering community today – at least on this forum.

Don’t get me wrong, Duncan, I’m all for preserving our wonderful industrial revolution by making working models from the period – I’ve built a half-sized Little Samson Traction Engine traction engine here in Thailand just for example – but I used CAD to make drawings, not a drawing board, Tee square and set square, I used Excel to do calculations not a slide rule, although in my teens those were only the implements available.

So if we are to base our modelling techniques on those available in the days when our prototypes were made, we should throw away our DROs, throw away our digital callipers and go back to dials and verniers.

That, I’m sure you’ll admit is a retrograde step. As is adherence to the so-called UTS method of calculating boiler shell thickness. The UTS method is outdated, illogical and just plain wrong. Anyone who supports it it is either stupid, uneducated or has a ‘political’ agenda, and I’m including biased moderators on this forum in this category.

Now let’s get to answering your technical points.

Quote: “If we believe MEinT’s figure for proof stress 4830 psi (please stop calling it yield)”

You don’t have to ‘believe MEinT’s figure for proof stress 4830 psi’, In our article we presented a range of figures for the yield point of annealed copper and gave the urls of the sources. We chose 4830 as being representative of published figures. The boiler designer can chose whatever figure he/she believes is appropriate.

Quote: “It quotes 0.5% proof stress at 10,000psi. My figure was 0.2%, so I’d expect it to be a bit lower.”

What is it then Duncan? 0.2% or 0.5%? Seems you can’t make your mind up. The FACT is that annealed copper dose not have a precisely defined ‘yield point’, as does, say steel. The value varies greatly (as stated in our article) which is why we chose to call it the ‘yield point’, because the yield point is a readily understood term but cannot be precisely defined for annealed copper.

Indeed when you say “(please stop calling it yield)” all the website references that we gave in our article use that exact phrase ‘yield point’. If you have an issue with this go and complain to the websites not to me.

Quote: “but UK national standards are built around staying below proof stress.”

Precisely. And this answers the stupid question being thrown around on this forum about ‘why is is it necessary to avoid distortion’.

Quote: “I very much hope that subsequent instalments of this saga will remain in the unused pile.”

Luddite. Go back to your slide-rule.

By the way, much to your chagrin and that of a certain ‘moderator’ (= agitator) I’m now back and ready to answer any further technical questions you may have about the correct way to assess model copper boiler shell thicknesses.

Quote: “I’ve been involved in model engineering for 40+ years”

So what?

Quote: “involved in model engineering..”

‘Involved?’ What does that mean? Simply reading a magazine means you are ‘involved’.

In contrast I have been making models for well over 60 years. I built an LBSC 3 1/2″ gauge Juliet chassis when I was 16 years old.

I’ve been a Professional engineer for 44 years. (Certificate below).

 

I’ve moved with the times – you seem stuck in the past.

I’ve got a drawer full of certificates covering for example steam boilers, pressure vessels (welding procedures ARAMCO – Saudi Arabia), I have a Collier’s Mechanical Engineer’s Certificate, and was a member of the Chartered Institution of Building Services Engineers (MCIBSE) (as well as MIMechE) until I retired. I’ve designed, specified and installed steam boiler installations up to Megawatt capacity. Think I don’t know a thing or two?

PS NO ONE has addresses my previous comment that the AMBSCE code (Australian Miniature Boiler Safety Committee Code pt. 1. Copper boilers, issue 8, 2012) uses the Yield Point of annealed copper and not the the Ultimate Tensile Strength of un-annealed copper modified by arbitrary factors (called ‘safety factors which clearly they are not) in their Tables.

Please explain why you would accept the modern AMBSCE code based on the Yield Point Method, but still cling on to the irresponsible UTS method.

[lezsmithParticipant @lezsmith]

Sorry it has taken me a while to respond, issues creating a new login, thanks Neil for helping me resolve it.

Wow, some strong comments but exchanging emails with Neil we thought the article would generate a lot of discussion 🙂

OK, just for clarification on some of the earlier comments, the goal of the article was to point out that using the tensile strength of copper to figure out the minimum shell thickness was incorrect, during manufacture the copper would be annealed additionally the use of magic numbers “safety factors” should be well understood and applied with the knowledge of how they affect the calculation.

Therefore, Alan and I proposed the yield point method, “this does NOT specify the copper thickness for a boiler, it simply calculates what the minimum thickness of copper should be”.

Additionally, the YPM does NOT include additional safety values that should be added to account for boiler usage e.g. the dynamic loads and stresses the boiler will be subject to during its usage. Given the multitude of uses we felt it better that the boiler designer adds an appropriate safety factor based on the design criteria.

Clearly after calculating the minimum thickness is to find the nearest commercially available copper plate that is equal to or greater than the minimum calculated thickness. Quite often that will be thicker than the minimum calculated thickness.

Note: the YPM just the same as UTS methods do not calculate the amount of deformation, all materials yes even ceramics deform under pressure, however the difference between YPM and UTS is the deformation under the YPM will remain in the elastic region, (non-permanent deformation).

For those who say who cares, also consider all deformation will cause some level of work hardening, however deformation beyond the elastic region will also cause thinning. I have done extensive FEA modelling and it appears the work hardening and thinning tend to cancel each other out resulting in a weak spot that will slowly grow over time.

As to the argument that it has worked for years why change, again doing the research it appears that there is no data to corroborate that statement, for commercial boilers any issue has to be reported by law to the health and safety authorities, however for model boilers no such law exists, talking to a club boiler inspector I was told boiler certificates are checked before all club events, however there is no requirement to report a boiler failure, if a model boiler has an accident, design or user error, it is never reported and no national record is maintained, even club records only show a valid boiler registration was seen at the time of the club event, if a model shows up with a new boiler no documentation is recorded as to what happened to the old boiler, or a boiler is just never seen again.

Therefore, all who say it has worked for years is quoting hearsay rather than quoting empirical data.

Additionally the YPM does NOT include additional safety values that should be added to account for boiler usage.

[Paul KempParticipant @paulkemp46892]

Not sure about others but I am struggling to understand the argument.  Trying to summarise the facts;

The YPM is a tool proposed to calculate the minimum thickness of a circular boiler shell subject to pressure only and no other external mechanical force or stress raisers due to penetrations (domes etc).

The YPM relies for proper accuracy on a defined “yield point” for the copper being used and yet everyone including the authors of the paper seem to accept that an accurate yield point for copper is unlikely to be available, hence they have taken an intermediate value (perhaps mean or average?) value from a spread of figures.

If the actual yield point of the material being used to construct the boiler is lower than the value used for the calculation, the accuracy is compromised.

Using UTS the first point above still applies, however as the UTS is a more reliable figure that can be verified if desired by tensile test of the actual material used to construct the boiler the accuracy of the calculation will be improved.

On the debate regarding distortion on initial test of a new boiler in the annealed state the UK code allows this under 7.1a “Owing to the ductility of annealed copper in a newly constructed boiler some minor distortion/bulging may take place and this should be allowed for”. Once that distortion has taken place the material will have work hardened and increased YP and UTS.  Assuming no higher pressure is subsequently applied then no further distortion (or thinning) should take place.

Surely the simplest way to validate the approach is for the authors to submit the paper to one or more notified bodies for acceptance that this is a method they will accept to validate and approve a design?  If they are satisfied then we can all go back to our sheds!

I guess the bit I am struggling with the most is no-one is going to construct a boiler to this theoretical minimum thickness whichever method is used due to available material thickness and the other considerations mentioned above, so what problem are we solving?

To Lez’s point re reporting what is stated is not quite true.  If a boiler failure results in significant injury then there would be a requirement under RIDDOR to report, additionally the insurer must be notified of a potential claim even if there is no expressed intent to claim at the time of the incident.  The boiler testing scheme does have the facility to report failed tests (but not failures in service).  I think all an event organiser is interested in is what is the condition of the boiler currently fitted, he won’t care what happened to a previous boiler so why would he record it?  His duty of care is to ensure the boiler being operated has been inspected and certified.

Paul.

[duncan webster 1Participant @duncanwebster1]

I think a significant boiler failure (in steam) would count as a dangerous occurrence and therefore be reportable. I mean a joint giving way big time or a shell bursting, not just a bit of a leak or a gauge glass breaking. However, I dont think it has ever happened in the model engineering world. Unless someone knows different.

[Nigel Graham 2Participant @nigelgraham2]

I do know different but of only one incident, and many years ago (1960s I think).

.

I recall my uncle telling me of one miniature locomotive boiler bursting, on a club track where he was a guest with his own locomotive. This though, was decades ago, before any particularly formal boiler-testing system let alone the PER-based, MELG-devised scheme. Investigation including laboratory analysis revealed the boiler had been over-strained by excessive hydraulic test pressure, leading to the longitudinal seam eventually breaking. Fortunately no-one was hurt but something narrowly missed the driver’s head. The locomotive was wrecked by the barrel partially unrolling.

I have not heard of any other such instances but obviously it could have very serious consequences.

….

Much more recently (1980s or 90s) I was once involved in failing a boiler brand-new – built faithfully to an appallingly bad design. It had virtually no stays and its inner firebox had collapsed under a preliminary hydraulic test, its very upset builder assured us, to working pressure. The builder, I should add, had recently retired from a career as a professional coppersmith; but was unfamiliar with miniature locomotive design and had taken the boiler drawing on trust.

We did not see the rest of the plans set, but the boiler drawing was poorly-printed, curiously anonymous and from an unknown source.

…

These are so exceptional surely, as a whole we must be doing something right without arguing over high-grade mathematics, who is qualified to design what, and over-stringent codes?

 

[Author]

Posts