Girder crown stays – are they overstressed?

[Paul HorthParticipant @paulhorth66944]

Many boiler designs use girder-type crown stays for the firebox crown (rather than stays connecting the crown to the boiler shell). There must be hundreds of engines in steam with such crown stays. I have made a 2 inch Durham Traction engine, designed by John Haining, which uses this type of crown stay.

I was going through some old copies of M.E when I read an article (Jan 1999) by Ross Forsyth, from Australia, where he showed that a typical copper girder crownstay would have stress which greatly exceeded the limit set in the Australian model boiler design code (AMBSC). Consequently, many boiler designs now used in the UK would not be accepted in Australia.

 

Their maximum stess allowed for copper is 3770 psi for any location in the boiler. When I checked the stress in the crown stays on my traction engne, I found that it came to 16100 psi, more than four times higher. This compares to the ultimate stress limit for annealed copper of 25000 psi, as advised by Martin Evans.

 

I would be interested in the opinions of others on this subject. Has this argument been settled over the last ten years, or should I be concerned about the high stress in the girder stays in my boiler? The same concern, if valid, would apply to many others.

 

Mr Forsyth says that the AMBSC does not prohibit girder stays, but it is clear that to meet the AMBSC stress limit, such stays would have to be really massive, so in effect they would be not practical, and stays to the outer wrapper would have to be used.

 

Please let me hear your comments…..

Paul

 

 

Edited By Paul Horth on 03/12/2011 21:19:08

[Paul HorthParticipant @paulhorth66944]

a query about the Australian boiler code

[JasonBModerator @jasonb]

I suppost the question is has anyone actually seen a girder stayed boiler fail, can’t say I’ve seem mention of one letting go.

 

I was chatting to an Australian who was over on business recently and he was not allowed to use girder stays on his 3″ Fowler Traction engine and there were a number of other details that had to be changed, this is what he ad to do.

 

J

[Jeff DaymanParticipant @jeffdayman43397]

I’m not familiar with the DNY traction engine drawings, but I recall that John Haining was partial to steel boilers in his models generally. 16000 psi working stress would be OK in a steel structure. Is it possible the girder design was intended for a steel boiler structure?

 

I agree 16000 psi stress in any part of a copper boiler is pretty high. It may not break or bend but the safety factor will be on the light side. If in doubt I would suggest you beef it up or fit some stays, maybe.

 

JD

Edited By Jeff Dayman on 04/12/2011 00:28:00

[JasonBModerator @jasonb]

Haining has both steel and copper boiler designs for the D&NY. But you could be right as the “bridge stays” are almost identical in profile and neither extend to the top of the barrel.

 

On the otherhand his Fowler boiler on has far more bridge staying and two are girders that extend to the top of the barrel.

 

J

[Geoff RogersParticipant @geoffrogers81118]

During my time with Lloyd’s Register I approved the design of many steel boilers produced for larger scale traction engines and locos by John Rex and Bell Boilers. For this design approval I applied relevant sections of BS 2790 (Shell Boilers) and in this code girder stay stresses were limited to 1.5 times that allowed for the barrel. This is standard pressure vessel and boiler practise where the item in question is loaded in bending – such as girder stays not attached to the outer wrapper.

 

These calculations apply to steel and copper boilers – the trick is knowing what the allowable stress should be. For steel this can be found from various sources, but copper is more difficult. The allowable stress depends on the design temperature ( ie desired steam pressure) and published data for copper tends to run out at 150C (from PD5500), which is equivalent to about 60psi.

 

Geoff Rogers

[Paul HorthParticipant @paulhorth66944]

Thanks for the replies, I look forward to more comments.

 

I have now written a spreadsheet to calculate the stress for different crown stay designs, which makes it easy to check a few other cases. Some of the recent designs by Martin Evans for copper boilers come out as follows:

 

Roedeer 7.25 gauge 7600 psi

Harringay 5 gauge 5270

Tennant 5 gauge 14000

Greene Queen 7.25 gauge 14780

Some of these stresses are pretty high, it seems to me.

If i can figure out how to attach a spreadsheet, I will send it in, so you can try this check for yourselves. I no longer have the earlier John Haining boiler designs, apart from the Durham, so I haven’t checked those.

Neville Evans has used rod stays in his recent boiler designs.

The design stress quoted by the UK Copper Board for annealed copper tube at 170 C is 26 MPa, or 3770 psi. This is the same as the limiting figure used in the AMBSC.

 

There are a large number of Maritn Evans boilers and John Haining Durham traction engines currently in steam. As Jason says, there have been no boiler failures…yet. I would like to think that this evidence of large numbers operating at higher stress without a prroblem, means that the higher stress is in fact safe. Do others agree??

 

Paul

 

[Sub MandrelParticipant @submandrel]

I have never made a loco boiler, but:

 

It has been commented that girder stays mean a lot of extra copper over the top of the firebox and they reduce free circulation in the area where most heat transfer should be taking place.

 

As for the stresses, the freshly annealed copper boiler will change shape under test pressure/first steaming and this will share the stress between crown, wrapper and stays, so that if any are overstressed they stretch until another component takes the load or the boiler fails.

 

Presumably this is why so many copper boiler designs made without detailed stress. calculations defy the Cassandras who predict mass-destruction.

 

Neil

 

[61962Participant @61962]

Paul,

 

I’m not very familiar with the Australian codes so you might enlighten me on the 3770 psi maximum allowable stress you have quoted. Is this the working stress or the stress at test pressure?

 

Anealed copper has a very low yield stress of 6000 to 7000psi. In normal elastic design it is considered that a structure has failed when yield is reached, therefore in a copper boiler under an initial cold hydraulic test pressure of twice working, good design would keep test stresses under the yield stress. Because the strength of copper reduces with increasing temperature it is prudent to design for working stresses less than half yield. There was (and may still be) a British Standard, BS2871, for copper pipes used in heating systems which tabulates copper pipe sizes and thicknesses for 100psi and 250psi. These when analysed show a working stress of around 2450psi on average, and I think that is a good guide to working stress for the design of model copper boilers. If you go back to the model engineering designers who actually quoted design stresses you will find Martin Evans at 2500psi (25000psi strength with factor of safety of 8 and 20% reduction for temperature, not the right route but a good enough answer), KN Harris at 3125psi, Keith Wilson at 3500psi and John Haining the same as Evans but with a 30% temperature correction giving 2187psi.

 

On your girder stays, if your calculations are correct, then there is no doubt they will bend permanently under load with a stress at that level. In fact the boiler would fail the initial test. The rod stay solution shown is now quite common, and is simply in line with full size practice. Note the cross stays to prevent the outer wrapper moving out, which would allow the arch of the wrapper to flatten and the firebox crown to drop. Good practice all round. Many professional builders have been using this techinique for years as it is easier and more economical than plate or girder stays and it is better for ensuring the joints have been properly made.

 

Girder stays are fine when adequately designed and this should not be difficult to achieve in relatively short fireboxes. The usual designs have a L or T section girder with no top flange for ease of manufacture. As the web is also the compression member of the beam it becomes subject to longitudinal buckling and therefore needs to be supported transversly. Not many designs address this and it is a difficult design calculation. Most engineers would avoid beams with this configuration, using Ts and Ls in the inverted position so as to put compression in the flange rather than the web. Can’t easily do this in boilers

 

Eddie

 

 

 

 

 

 

[JasonBModerator @jasonb]

“On your girder stays, if your calculations are correct, then there is no doubt they will bend permanently under load with a stress at that level. In fact the boiler would fail the initial test.”

 

So how come there are plenty of D&NY traction engines running around with this design of boiler? Can’t have failed their initial test.

 

J

[Paul HorthParticipant @paulhorth66944]

Eddie,

 

Thanks for your reply – but Jason asks the right question, which is really what I was asking. You must be wrong.

As well as the numerous boilers to this design owned by others, I have two, one made by me and the other made professionally. Neither shows any bending in the firebox crown whatsoever, despite being tested to 2 x working pressure several times and each boiler having been run in steam over several seasons.

Therefore, either my calculations of stress are wrong, or your conclusion is wrong and the higher stress can in fact be tolerated and accepted? This is what I am asking.

 

I am not from Australia, all I know of their code is what I have read in Ross Forsyth’s article. I undestand the limit of 3770 psi to refer to working pressure not test pressure ( that would be really low). Martin Evans may quote a design stress of 3125 psi but in fact his designs would have a much higher stress in the girder stays. This was Mr Forsyth’s point and I have now confirmed that. Neither has John Haining stuck to your quoted figure of 2187 psi, the Durham design exceeds that by a factor of about nearly 8!

Unless my calculations and Mr Forsyth’s are not correct.

 

I repeat, I am not talking about a new design which I can modify after taking good advice, but about an existing boiler already completed and in service, similar to many others which are likewise tested and proven. Likewise for many of Martin Evans’ loco designs, there must be hundreds of these in service.

 

Paul

 

 

 

[61962Participant @61962]

Paul, Jason,

 

Materials science and basic engineering conclude that at 16000psi, five or six times the accepted safe working stress in copper, there is more than enough yield to fail the boiler. Given evidence that numerous boilers work perfectly well built as the drawing, is the logical conclusion that materials science and engineering are wrong or that there is some doubt over the calculation of the stress? Maybe Paul can show us his method.

 

On the subject of the various working stresses used by boiler designers, the figures quoted are taken from articles and books by them, and not back calculated from their drawings. My own investigation into how these figures were derived resulted in the premiss that the maximum stress should be based on the prevention of yield at initial test and I am happy that there is general agreement between the designers quoted, the British Standard and my own thinking on the allowable working stress.

 

I would point out that the factor of safety method is flawed in that the copper only acts elastically up to the onset of yield. After that basic Hooks Law is not valid as the material is significantly changing section during the plastic phase. Saying that no designer in industry would want to show a factor of safety as high as 8 or 10, as that simply means he has considerable doubt over the validity of his design.

 

Regards

 

Eddie.

 

[Paul HorthParticipant @paulhorth66944]

Eddie,

 

Yes I agree with you that either my calculations are wrong, or the prediction of the behaviour from the assumptions used in structural design and materials science are not as good as we think. I would dearly like to be wrong about the stress, so I can feel a little more relaxed when running the engine. But if I am wrong then so is Mr Ross Forsyth, because my calculation follows his and gives results in agreement with his , and he is a structural engineer and speaks for the Australian code.

I will gladly send to you my calculation spreadsheet but I have not found a way to send in a spreadsheet to this forum, if you wish i can email it to you directly if I have your address.

Basically, Ross Forsyth considers the girder stay to be a beam simply suported at the ends which carries the load on the firebox crown as a uniformly distributed load. Two beams share the total load, equal to the pressure divided by the area of the firebox crown.

I have calculated the section modulus of the girder according to published equations for a tee section beam.My calculation yielded the same result for the section modulus of Caledonia and Roedeer stays (Martin Evans), as from Mr. Forsyth in his article.

 

I suggest that perhaps the girder stays do not act as simply supported beams, that the evidence shows that the beam and firebox assembly is somehow stronger than the simple model can predict.

 

I am reassured by your observation that the structural model predicts collapse, but that has not happened, not just in one lucky case but across many examples. if it has not happened yet then this model is probably not accurate and I am probably not going to face a boiler failure in the future..

 

 

regards

Paul

[Paul HorthParticipant @paulhorth66944]

ERROR ALERT…

 

“equal to the pressure divided by the area of the firebox crown.”

 

That obviously should have been MULTIPLIED.

My apologies!
Paul

 

 

 

[JasonBModerator @jasonb]

I would have thought that as the girder is soldered along the top of the firebox it is more than just a simple beam, the top of the firebox will become in effect a web in tension and takes some loading from steam pushing down on it. If the two were able to slide against oneanother then the girder would just be a beam with even load.

 

Again the load is not really even as the front & back of the firebox will have a staying effect.

 

Just my thoughts.J

[Geoff RogersParticipant @geoffrogers81118]

I would agree with Jason and add that there is no such thing as a simple beam. The load capability of any beam depends to a large extent on the nature of the end restrains (or degree of fixing). Hand calculations are limited to ‘simple’ ie free to rotate/deflect or ‘fixed’ ie rotation/deflection prevented and quite different results can be obtained for a given beam and loading case.

 

For those with the access, finite element analysis will probably reveal all, but for us others we are limited to such books as Roark’s Formulas for Stress and Strain. A good source for beams on elastic foundations – which is what a copper boiler girder stay is.

 

as Jason says, just my thoughts

Geoff

[Paul HorthParticipant @paulhorth66944]

Thank you, Jason and Geoff. These answers help my understanding.

I readily admit my lack of knowledge in structural engineering, and I started this thread in the hope that others could offer an explanation of this seeming contradiction between the calculated stress in a simple beam, and actual experience.

The firebox crown is arched, not flat, and this must confer some strength (resistance to bending) which is in addition to the girders and acts to relieve them from some of the load. I can see that the whole assembly of girders, cross bracing and arched crown, rigidly joined together and rigidly joined to the flanged end plates (these are not knife edges) acts as a kind of box which is stronger than two loose beams with empty space beneath. However, calculation of such a structure would be completely beyond me, I have not learned the basics sufficiently to use Roark..

 

We may have to conclude that the simple analysis of Mr Forsyth and the AMBSC is indeed too conservative, something I am not qualified to judge except by the evidence of my non-collapsed boiler.

 

Paul

[61962Participant @61962]

Paul,

 

As a civil engineer of more years experience than I care to remember, it is not the first time that I’ve encountered people drawing the wrong conclusions. I’m reminded of the wheel tapper who condemned the wheels in several trains before he realised his hammer was cracked!

 

Conservative design treats each member of the structure in isolation. It’s a safe solution. Economic design treats the structure holistically but requires a much cleverer designer to ensure that all the possibilities are covered and that stress paths are not inadvertently doubled up or missed out. I tend to err on the side of safety. So I’ll stick with 2450 psi working stress. Finite element and limit state design might save a bit of material, but when it comes to boilers I like ’em heavy!

 

Regards

 

Eddie

 

 

[Paul HorthParticipant @paulhorth66944]

And yet, my boiler still stands, despite your prediction from conservative structural analysis that it should have collapsed.

How so?

My calculations stand ready for your checking.

Consider another example. An aircraft is a pressure vessel, with 1 bar internal pressure and close to zero external pressure. If the aircraft fuselage was designed according to industrial pressure vessel design codes, it would be far too heavy to fly. So it is designed differently, and yet the skies are not full of exploding aircraft. This experience is good enough for thousands of people to trust their lives to them every day. They don’t even get hydrostatically tested!

Conclusion: we should be ready to accept that the evidence of reality has priority over predictions from a simplified analysis. When he introduced girder stays, Martin Evans carried out a test at 3 x working pressure on a boiler he designed,, to prove their validity, whereas the structural analysis would have led to the mistaken conclusion of failure.

 

Paul

[Anonymous]

Paul,

 

While your example of an aircraft as a pressure vessel is valid, I think some of your numbers are a little awry. Commercial aircraft normally operate at an internal cabin pressure equivalent to about 8000 feet. From the ICAO standard atmosphere this is 753mb. It is rare for commercial aircraft to operate above 40,000 feet, where the equivalent pressure is 187mb. So the maximum pressure differential is 566mb.

 

I don’t know about modern airliners, but the original Comet fuselage was subjected to considerable hydrostatic pressure testing before certification. Unfortunately it didn’t find the metal fatigue failure that later led to two fatal accidents. I suspect that metal fatigue is a more serious problem for pressurised aircraft than straight over-pressure.

 

Here’s a link to a survivable decompression:

 

http://en.wikipedia.org/wiki/Aloha_Airlines_Flight_243

 

Regards,

 

Andrew

[Tony MartyrParticipant @tonymartyr14488]

‘All models are wrong but some are usefull’

I admire the confidence with which the quoted stress calculation had been done – I have not so much confidence and agree with Paul’s statement:

I suggest that perhaps the girder stays do not act as simply supported beams, that the evidence shows that the beam and firebox assembly is somehow stronger than the simple model can predict.

My boiler has inverted T beams tied together centrally by a transverse link and silver-soldered to the curving roof of the firebox that it overlaps at both ends. Looking at that whole structure under compression defies simple calculation. My boiler shows no signs of distortion and I have never heard, in recent time, of a beamed firebox collapsing so long as the water levels are held correctly.

I am still working (part-time) in an industry where the empirically established best practice of many years is now sometimes being challenged by some PhD student who has modeled a mechanical system on a computer and claims we should be dead.

Tony

[MICHAEL WILLIAMSParticipant @michaelwilliams41215]

To : Paul Horth ,

 

I can explain almost all of what is happening as regards stress levels in firebox/stay combinations and reasons for some of the anamolous results that simple calculations give .

 

I have already put some introductory explanations in a private message to you .

 

There is no point in trying to discuss any matter of a technical nature direct on this forum but if you would like to contact me direct after reading message I will give you every assistance .

 

Michael Williams .

 

[Sub MandrelParticipant @submandrel]

I recall Jim Ewins who was able to condemn a (succesful) Martin Evans design before Martin had even published a GA drawing.

 

He was much given to long and complex calculations which never seemed to be published for scrutiny..

 

Now Michael, I’m not suggesting you are a scoundrel, but I’m sure that if you did share your calculations there would be soime light generated, as well as heat.

 

Neil

[GoCreateParticipant @gocreate]

Paul, if it interests you I can look to do some fea analysis, I will need a copy of your boiler drg. or sketches etc.

 

FEA does have limitations and results have to be viewed with such limitations in mind. Allot depends opon the users knowlede of the application and skill in preparing the simulation however, it will give a clearer picture. The analysis will be linear, that is it assumes all strain is elastic no account being takes of re-distribution of stress in areas of high stress concentration etc. This can distort results a little.

 

All I am saying is, it’s not the be all and end all but can improve understanding of the application, in that context it can help identitify how and where to make improvements.

 

If you want to take this further send me a PM.

 

Nigel

[Geoff RogersParticipant @geoffrogers81118]

Nigel, Paul

I would be most interested in a FEA of this area, let me know by PM if I can be of assistance

Geoff

[Author]

Posts