Boiler Design – issue 4765

[Nicholas FarrParticipant @nicholasfarr14254]

Hi, all I can say is the the UTS is in the plastic range, and there is no elasticity in the elastic range. When I started work, high tensile bolts were not in general use, and when you unbolted things, the same bolts were used to bolt them up again. It soon became clear to me, that used bolts would stretch a little bit more every time they were done up again, and the stretching would accelerate, to the point it was clearly obvious, and the tightness of the bolt could not be made to the original, and it would eventually fail trying to even reach any sort of tightness at all. The UTS tells you the point that a material is most likely to fail. When I did my testing on the Tesometer, you could feel the stretching takin place, as the effort used to turn the handle was getting les as it was approaching the UTS.

Regards Nick.

[noel shelleyParticipant @noelshelley55608]

I take a dim view of being referred as a LUDDITE or the use of other derogatory terms ! May be this thread has run it’s course and should be closed ?  Noel.

[Paul KempParticipant @paulkemp46892]

Magic numbers, my thoughts;

We want to build a boiler.  We know it’s physical dimensions and the pressure we want to operate.  The absolute minimum theoretical shell thickness considering pressure alone will be that thickness that produces a circumferential stress in the cylinder that is equal to the UTS of the material we want to use (ignoring distortion as it has to distort before failure if it is a ductile material).  Using that thickness to build the boiler will result in certain failure when it exceeds operating pressure.  But wait, we are required for the first pressure test on the shell to pressurise it to twice working pressure, we can’t do that, it will fail, so actually to guarantee a pass we need to use a thickness of our material that will resist a pressure of 2xWP +1.  I suggest +1 because we want the absolute minimum thickness at which we can pass the 2xWP test with certainty.  We already have 2 magic numbers, my arbitrary +1 because I wanted to be certain it wouldn’t fail at 2WP and in fact the multiplyer of 2 for the test pressure, who decided 2 was the optimum number?  Whatever, we now have a minimum strength thickness for our boiler that will resist a 2xWP test.  It is already twice as strong as it needs to be based on material strength alone to be operated once at its design working pressure.

We now need to consider the other properties of the material we wish to use outside its basic UTS (or YP), it’s resistance to corrosion, creep, ductile properties, susceptibility to fatigue, behaviour at elevated temperature at the same time as considering how long we want it to last, how many pressure cycles will it go through in its design service life, what external stresses or other stress raisers is it subjected to? For each of these considerations we can assign a correction factor (magic number) or limiting factor through modelling or calculation to define the minimum operating thickness.  We don’t have the luxury of being able to ask Mr Greenly to explain how he arrived at his magic number of 8 but as it provides a result that is close to the alternative proposal I am assuming there was some science behind it, suggested by the results of the two methods being close to each other.

Personally I like magic numbers, I wouldn’t be happy with a plane, train, boat, bridge etc having a magic number of 1 for design strength and with systems that have a poor FMEA, ie single point of failure.

The sole advantage proposed for the YPM appears to me to be ensuring there is no distortion when the 2xWP test is applied and the material remains in its annealed state.  But as the results in terms of thickness correlate quite closely between the UTS and YPM is there a fundamental issue?

I can’t comment on the Aussie code, I don’t have it and thus haven’t read it.

On Lez’s point re RIDDOR, I am very familiar with it and Duncan is correct the definition of workplace is a very grey area and as various individuals have found to their cost the fact they were operating in a volunteer environment for no financial reward is no defence and Courts have ruled in favour of HSE in some instances even where the public were not present.  I mentioned RIDDOR as one example of why an incident would be reportable.  That is a separate discussion though and a distraction from the YPM debate.

As for holding the hobby back, I have no issue with modern manufacturing methods or using available technology to assist design.  The EU accept stainless steel for boilers, the UK should review this.  My criteria for change is it needs to bring clear and tangible benefit over what we had before.  With the small difference in results even ignoring Luker’s points I am on the fence.  If a notified body (that is after all who will approve our design) accepts this is the future direction then I will happily comply.

Paul.

[lezsmithParticipant @lezsmith]

Hi Luker,
With reference to your post, the YP method presented, clearly states that it was for use with “Copper” therefore your point about not using it for other materials is clearly as stated in the proposal.

You go on to say “Design a shell to the minimum thickness” is NOT what was stated in the proposal, it clearly states the calculated value is the minimum thickness required to resist permanent deformation during the 2x working pressure testing, the proposal goes on to state an appropriate safety value MUST be added to account for operational conditions the boiler will encounter, rather than an arbitrary 8x used in the current UTS method.

You state “Practically, this example brings the YP design method much closer to failure than the UT method. And that’s not what we want…”
How can this be true if the points in previous posts above and in the examples given, the minimum copper thickness is generally thicker?

As you said, I also don’t want to get into a debate here on the correct application of numerical methods, therefore I will refer this question to my old professor of Engineering at MIT, hopefully some undergraduate can take this on as research for his/her degree.

 

 

[lezsmithParticipant @lezsmith]

Hi Paul,

First off, thank you for a very good post, no insults just interesting commentary 🙂

Taking your last comment first, fair point on RIDOR, it clearly states workplace, but as you point out “how to define a workplace is debatable”, if I employ a gardener to cut my grass is my personal home now a workplace?

However, an interesting fact was raised “where do I find the RIDOR reports” just doing some research I’m trying to find all the reported “Model Boiler failures” but I’m not able to find any, clearly their have been some, at least 2 as mentioned in the above posts, so where are the records?

I’m not going to comment on the YP vs UTS anymore, we put the presentation and methodology together not as you must abide by this, it was meant to be here is an alternative that does NOT use magic numbers and follows other standards such as the AMBSC that have used the Yield point of annealed copper as the reference rather than the UTS. Our hope is it would provide an alternative by using well defined values that can be referenced in easily obtained literature.

If you are ok with your boiler deforming during the 2xWP test no problem as the builder and perhaps the designer that is your prerogative, just because we provided another tool for the workshop it does not mean you have to use it, but at least it is available and like most tools covers a specific function (defines a minimum copper thickness that will not deform under the 2xWP test) the reality is 99% of boilers will use copper that is significantly thicker than this minimum value, and so they should, they need to increase the thickness by adding a safety margin based on the environmental situation the boiler will be used, then the normal next step is to find what thickness is commercially available, this could be slightly thinner than the calculated value, however with the YP method it is very easy to work backwards to find the calculated safety value for the thinner copper, allowing the builder/designer to consider if the decrease in the safety value is ok or not.

 

 

[MEinThailandParticipant @meinthailand]

Lezsmith and all, just to clarify the following statement in your post:-

“an appropriate safety value MUST be added to account for operational conditions the boiler will encounter, rather than an arbitrary 8x used in the current UTS method.”

I just want to reiterate and emphasise once again that the so-called safety factors in the UTS method are not Safety Factors at all.

They can best be considered as Conversion Factors, used to convert the strength properties of the material used in the UTS calculation (hard drawn copper) into the actual properties of the material when a boiler has been silver soldered, ie annealed copper.

To refer to the UTS ‘safety factors’ as actual Safety Factors is misleading and could lead copper boiler builders into thinking that they have a large safety factor (e.g. 8) in their boiler when in fact they do not, due to the reduction in strength of copper during the silver soldering process.

 

 

 

 

[MEinThailand]

On MEinThailand Said:

Lezsmith and all, just to clarify the following statement in your post:-

“an appropriate safety value MUST be added to account for operational conditions the boiler will encounter, rather than an arbitrary 8x used in the current UTS method.”

I just want to reiterate and emphasise once again that the so-called safety factors in the UTS method are not Safety Factors at all.

They can best be considered as Conversion Factors, used to convert the strength properties of the material used in the UTS calculation (hard drawn copper) into the actual properties of the material when a boiler has been silver soldered, ie annealed copper.

To refer to the UTS ‘safety factors’ as actual Safety Factors is misleading and could lead copper boiler builders into thinking that they have a large safety factor (e.g. 8) in their boiler when in fact they do not, due to the reduction in strength of copper during the silver soldering process.

 

 

 

 

Ref my own post I would like to choose better words.

I said:-

They can best be considered as Conversion Factors, used to convert the strength properties of the material used in the UTS calculation (hard drawn copper) into the actual properties of the material when a boiler has been silver soldered, ie annealed copper.

A better choice of words would be:-

They can best be considered as Compensating Factors, used to compensate for the fact that the strength properties of the material used in the UTS calculation (hard drawn copper) is much reduced to the properties of the material when a boiler has been silver soldered, ie annealed copper. (Approximately by 1/6 depending upon the source of information for the properties of the materials.)

[MEinThailand]

On MEinThailand Said:

On MEinThailand Said:

Lezsmith and all, just to clarify the following statement in your post:-

“an appropriate safety value MUST be added to account for operational conditions the boiler will encounter, rather than an arbitrary 8x used in the current UTS method.”

I just want to reiterate and emphasise once again that the so-called safety factors in the UTS method are not Safety Factors at all.

…

To refer to the UTS ‘safety factors’ as actual Safety Factors is misleading and could lead copper boiler builders into thinking that they have a large safety factor (e.g. 8) in their boiler when in fact they do not, due to the reduction in strength of copper during the silver soldering process.

 

 

 

 

…A better choice of words would be:-

They can best be considered as Compensating Factors, used to compensate for the fact that the strength properties of the material used in the UTS calculation (hard drawn copper) is much reduced to the properties of the material when a boiler has been silver soldered, ie annealed copper. (Approximately by 1/6 depending upon the source of information for the properties of the materials.)

I fear inventing ‘Compensation factor’ as an alternative to ‘Safety Factor’ only muddies the water!

Though there are variations Safety Factor is long established in engineering,  see Wikipedia’s ‘Factor of Safety‘  In my view, this whole debate is about understanding Safety Factor.

Simply put, the underlying YP and UTS formula were derived scientifically. At root, the burst pressure of a cylinder depends on the outside diameter, how thick the wall is, and the strength of the material.   Easy enough on paper, unfortunately the real world is complicated!  Especially “strength of material”.

Material strength can be expressed imperfectly with Yield Point or Ultimate Tensile Strength.  Both can be measured and used to give reasonable answers in calculations,  but material strength is best explained with a graph.   More!  The formula assume a perfect cylinder and ignore all the external factors, which matter.   For example, material strength varies with temperature, construction matters (joints, end-caps and penetrations),  and so do the operating conditions.   Metal fatigue!

In practice, the underlying formula is used to derive the minimum thickness of a pressurised cylinder,  and then an arbitrary safety factor is applied.  How big is the safety factor?   My machine design book says:  ‘To compensate for the fact that not all the assumptions involved in a rational analysis are true, the machine designer must choose a safety factor which in his judgement will permit satisfactory machine operation.’    

Safety Factors are usually a multiplier decided in light of practical experience, often radically increasing strength above formula!   SF’s are often embedded in codes of practice, likely updated after a catastrophic collapse.

I pointed out in post #800663 that the Safety Factors in the YP and UTS formula dominate the answer.  In my view the difference between YP and UTS is academic in this application because both are heavily weighted by the Safety Factor and produce similar results.  Not worth arguing about.

Unfortunately, the size of boiler Safety Factors are lost in the mists of time!  But they are based on practical experience – boilers built to them giving a good service life without going bang.   Practical experience increases the Safety Factor to cover unknowns like material differences.  The strength of Copper depends on which alloy is used, how it was worked during the build process, and then on how hard, hot, and often the boiler is cycled.   Safety Factors are not an exact science.

Worth challenging current standards because they’re excessively high compared with a good design built by a skilled worker using well-chosen material.   But they cover weak designs, built by semi-skilled workers from scrap!

I suggest YP versus UTS isn’t the most important factor.    Both formula only apply to the cylinder and assume it has perfect ends, which is unreal.  Actual boilers have a smoke-box, tubes, holes punched in the shell for drainage, safety valves & the regulator, and a firebox full of stays.  MEInThailand recognises the problem, noting silvering soldering can reduce strength by up to 1/6.    Multiplying by a SF fixes that, but the SF is arbitrary…

My gut feel is that FEM offers a better way of challenging the status quo. The whole boiler can be modelled, not just the cylinder, the software calculates stresses throughout the whole structure, and shows where they concentrate.   Not a trivial job though.

I welcome articles like this and the resulting discussion.

Dave

 

 

[JasonBModerator @jasonb]

 the strength properties of the material used in the UTS calculation (hard drawn copper) is much reduced to the properties of the material when a boiler has been silver soldered, ie annealed copper. (Approximately by 1/6 depending upon the source of information for the properties of the materials.)

 

MEInThailand recognises the problem, noting silvering soldering can reduce strength by up to 1/6.

 

Be careful using “by 1/6″ as 24,000 reduced by (1/6 * 24,000) =20,000

When it is reduced to 1/6th the original value 24,000/6 = 4000

 

 

[Nigel Graham 2Participant @nigelgraham2]

Lezsmith –

Some posts back but re your:

…reported “Model Boiler failures” but I’m not able to find any, clearly their have been some, at least 2 as mentioned in the above posts, so where are the records?

If that includes the one I had described a while previously, that was decades before any such reporting was devised; and indeed before the test system we now use.

Possibly the other was also before the RIDDOR rule, but only two serious failures in service, although still two too many, shows just how thankfully rare they are; making me wonder the point of this entire thread. We all want to build and operate safe engines but let’s not give ourselves opening for making the hobby no longer feasible.

 

I can offer a third failure in service, and its lessons.

It was not catastrophic, by sheer luck. The locomotive, a doubled-up (so 7-1/4″ g.) version of LBSC’s Juliet, struggled to draw its loaded train into the station, with the driver reporting water in the visible part of the ashpan.

Assuming a tube leak we took the engine out of service and examined it. A leak had developed in the inside firebox wall, and quietly extinguished the fire. The boiler was steel and it proved possible to grind the damage out and patch-weld it, and after very careful, thorough testing return the locomotive to service for just two remaining events. Then it was withdrawn for a new boiler, now copper, by a professional boiler-maker although at cost as he was retiring and was also a club member!

The lessons, after I cut the old boiler up for full examination:

– Design fault: Lower gauge-glass bush too low so possibly the crown-plate occasionally ran dry. (No fusible plug.)

– Operational faults not recognised for years, leading to serious scale and corrosion. The water-legs were nearly choked by calcite and iron-oxide, sufficient to distort the plates. The locomotive had run for years before water-treatment became normal in model-engineering circles, and ours is a hard water area. Although the boiler was usually blown down at the end of running, a single blow-down valve is insufficient for proper cleaning.

– Deep crown-plate corrosion, perhaps liked to the times when mimium water indication might actually have meant a dry plate. (I did not think to measure this properly in the post-mortem.)  Some sectioned pits on the water side had floors ~0.03″ thick. Fire side moderately eroded.

– The rest of the boiler’s internal surface was rusty but evenly and not deeply, nor particularly scaled.

– Materials source and approvals: not relevant! The boiler had been assembled by a professional welder outside the club, although he did not understand locomotive boilers and had no reason to query the gauge-glass bush height. The materials were probably industrial off-cuts not knowingly certified for anything, let alone pressure-vessels; but no-one worried about that in the 1970s. The barrel was of standard steel pipe, the plates likely ordinary hot-rolled mild-steel. I think the tubes were of copper. This shows paper does not protect anything physically: you can quote all the numbers and codes you like but oxygen still attacks iron, and calcium-carbonate still precipitates from solution in water.

[Author]

Posts