American steam locomotive efficiency- the effect of blastpipe size and superheat levels.

[GP40-2]

Burgard540

 

If you look at the chart, notice that the proportions of heating surface are nearly identical regardless of the type of locomotive.  Of the total heating surface: 63% is tubes & flues, 7-8% is firebox, and 29-30% is superheater.  Notice that the ALCO built UP locos had the highest proportion of firebox heating surface at 8.4% for the articulated's and 7.8% for the 4-8-4's. 

Just to add some additional data, the 2-8-8-4 EM-1, built by Baldwin in 1944-45 had its boiler comprised of 61.2% (4,540 sq.ft.) Indirect Heating Surface, 28.6% (2,118 sq.ft. Type E) Superheater Surface, and 10.2% (758 sq.ft.) Direct Heating Surface. Baldwin clearly changed the boiler design on the EM-1 from the earlier AC10 and M3/M4 designs by enlarging the size of the firebox area vs. the fire tube area.

[Burgard540]

It would be interesting ot compare these various designs differences based on actual performance.  A lot of data would be needed beyond most reports of drawbar horsepower, coal & consumption, and design parameters.  Things like flame, firebox, flue & tubes temperatures measured at various points; percentage of imcomplete combustion & weight of air required for combustion; stack gas temperature, weight of flow & composition analysis; steam chest pressures and superheat temperatures; continuous pressure readings throughout the cylinder cycles; etc.

Too bad a lot this testing would require a time machine...or generous funds and cooperation by someone like the UP steam program or similar group that operates a late design steam locomotive.

Given that only about 12% of the energy of the steam is converted to mechanical energy at the wheels, the focus should be to try to design a system where the total heat of the fuel fed into the firebox, the total heat produced by combustion, and the total heat absorbed by the boiler producing steam is as close to unity as possible. 

[Neil Robertson]

[selector]

Neil, if you are using Internet Explorer 9, and didn't just mistakenly forget to type a message above, you will have to look for the small grey 'compatability view' icon that looks like a torn square, meaning a sheet of paper.  It is just to the right of the top URL bar, along with a magnification glass icon, the curved 'refresh view' arrow, and the close-out X.  Just click on the torn sheet of paper, at which it will conform to the view intended for this server and then it will turn blue.

Welcome to the forum.

Crandell

[Neil Robertson]

Thanks, I'll try again. Giesel in his Anatomy of the Steam Engine (in German) mentions tests the LMS did in 1934 comparing different blastpipes on Jubilee 4-6-0s, which initially were not great steamers. In all cases steaming rate was 20000 lbs/hr. For each arrangement he gives the total power in the exhaust at the blastpipe, the pumping power (that used for draughting the boiler), the shock losses (caused when the steam impacts the flue gases in the smokebox), other losses (not totally wasted as it lifts steam out of driver's line of sight), the smokebox vacuum, pumping efficiency (ratio of pumping power to total power) and blastpipe pressure. For the standard single blastpipe the results were: 238 HP, 9.5 HP, 154 HP, 74.5 HP, 91 mm WS, 4.0% and 0.45 atu. For an improved single blastpipe: 196 HP, 11.6 HP, 126 HP, 58.4 HP, 103 mm WS, 5.9% and 0.36 atu. And for a double blastpipe: 143 HP, 13.1 HP, 94 HP, 35.9 HP, 118 mm WS, 9.1% and 0.26 atu. 

One sees a progressive fall in exhaust power, improvement in draughting and also implied a fall in cylinder backpressure, giving a further gain in useful power.

[Dreyfusshudson]

Dear Neil,

Thanks. I think the tests you describe are very important in that they had a crucial impact of UK locomotive design ever after. I have still to locate the full report, but what ex LMS senior engineer AJ Powell reported was that (surprisingly) there was no improvement in coal or water consumption for the Kylchap set up vs an extremely restrictive 4 ½” single blastpipe.  I am at a loss to explain this. He notes that the Kylchap device used was probably far too powerful for the quite small Jubilee boiler, and that there was a continuous shower of sparks from the chimney, which may explain the lack of reduction in coal.

I have done some calculations ( see below) on what the benefit of the Kylchap would be, assuming it was of the dimensions used by the LNER, which were similar to the French Compounds. There is, even at the relatively low steam rate of 20000lbs/hr, an improvement in IHP of about 8%, so you would expect an 8% reduction in water at the same power. ‘At the same power’ may be the root of the problem. Because the cylinder backpressure is lower, the steam consumption at a given cut off goes up by about 10%, so power goes up by 10%. So it may be that on these road trials, the Kylchap locomotive was producing more power, and this interfered with the economy assessment. Hopefully the report will throw light on this.  

	Kylchap					Standard
Steam rate, lbs/hr	Cut off, %	Blast pipe pressure psig	IHP	SSC, lbs/ihp-hr	vacuum, ins	Cut off, %	Blast pipe pressure psig	IHP	SSC, lbs/ihp-hr	vacuum, ins
13700	13.6	1.0	850	16.1	3.8	14.8	3.7	820	16.7	2.7
19100	19.6	2.0	1300	14.7	5.9	21.5	7.6	1210	15.8	6.0
23100	23.7	3.0	1590	14.6	7.2	26.4	12	1420	15.3	-
26400	26.8	4.0	1780	14.9	8.0	30	16.2	1530	17.2	-

 

The LMS later tried a Kylchap on a Duchess Class Pacific, and found no benefit. Here, I think the result is explicable. The Duchesses were fitted with double plain blastpipes with a similar free nozzle area to the Kylchap; no backpressure benefit expected then. A Duchess was tested on the Rugby test plant where it was concluded that the draughting was adequate, though not brilliant. Now what better draughting such as a Kylchap does is increase the specific evaporation rate (lbs steam/sqft/hr) possible. The Duchesses had grates of 50 sqft and in service were never required to steam at more than 30000lbs/hr (600lbs/sqft/hr) because they were hand fired. On test, they got 800lbs/sqft/hr; a Kylchap can deliver 1000lbs/sqft/hr- both these higher rates are quite beyond any fireman. So because only modest specific evaporation rates were possible in service, these could be satisfactorily delivered with a very ordinary plain double blastpipe.

So, in two sets of tests, the LMS concluded that Kylchaps weren’t worth it. The LMS mafia were in charge of BR locomotive design after nationalisation and guess what- no Kylchaps. BR standard designs were hobbled with very primitive exhausts. The mafia even tried to stop the ex LNER region from fitting Kylchaps to their Pacifics, where there was a clear need.

Just shows what happens when a) you have dodgy data and b) no sound quantitative theoretical model. Politics wins.

[Neil Robertson]

Thanks for info, which I was unaware of. The blastpipe nozzle of 4.5 in would give a steam rate of below the critical value of about 93 kg/(cm2 h) for 20000 lb/h, but not by much.The minimum ssc you give for this single blastpipe (15.3 lb/ihp h) seems to indicate a steam temperature of at least 330 C (626 F) for 225 lb/in2 (following a steam table in Giesel's book Locomotive Athletes for a cylinder volume of 200 L), which would need to be corrected upwards to allow for the relatively small cylinders.

Giesel also gives the results of testing the West Country class pacifics with both the Lemaitre and Giesel exhausts. In going from the Lemaitre to the Giesel, blastpipe energy fell from 165 to 78 HP lowering pressure from 0.30 to 0.13 atu, increasing pumping work from 13.1 to 15 HP, which increased smokebox vacuum from 118 to 126 mm WS, lowering shock loss from 126 to 15 HP, and increasing other losses from 25.9 to 48 HP, which was enough to lift the exhaust clear of the engine. Pumping efficiency rose from 7.8 to 19.3%.

[Neil Robertson]

Coming to Kylchaps and A4s I have a question that your computer program might help to answer. Coster says in his book on the A4s that the development of the LNER boiler might, with the benefit of hindsight, have been developed somewhat differently. That the 41.25 sq ft was quite adequate given the Kylchap but that firebox volume might have been increased.

So have a combustion chamber 1 ft longer than in an A4 and tube length 1 ft shorter (about 17 ft) as in the V2, A1 and A2. So, compared with A4 there is less water in the boiler so for a given amount of heat produced in the firebox one might expect a higher combustion rate. Less water in boiler reduces the overall weight so boiler plates can be a little thicker and boiler pressure increased a little as Gresley planned. The possible problem is that with the rear tubeplate 1 ft further from the firegrate so also are the rear ends of the superheater tubes. And lower superheat would be very negative from the efficiency point of view, perhaps negating the point of this thought exercise. Does your program make any prediction for the superheat in such a boiler, i.e. as an A4 but with about a 2 ft long combustion chamber and a barrel as in an A1?

[Neil Robertson]

Sorry there is an error in my last post.

The second sentence of the second paragraph should read: So, compared with A4 there is less water in the boiler so for a given amount of heat produced in the firebox one might expect a higher evaporation rate. 

[Dreyfusshudson]

Dear Neil,

This responds to your three recent posts.

Firstly, with respect to the inlet steam temperature on the Jubilee, I had it rising from 298 to 326⁰C as steam rate increases in line with the test data on 45722. Steam chest pressure was assumed to be 215psig. I should say that, after ten years of using this fluid dynamics engine programme, it is in my view in a different league to anything available in steam days, providing of course, you know what you’re doing! Giesl’s approach is now superfluous.

On the Giesl results, I’m somewhat sceptical of Giesl’s claim. In UK test plant data, there are plenty of examples of misleading backpressures, due to problems with where the backpressure was measured. The programme mentioned does compute blastpipe pressure, in line with standard discharge theory. If you look at most Test results from Rugby, the agreement is very good between experiment and theory. There are a number of case where, for a given free nozzle area, measured blastpipe pressures are higher, but this is on locomotives where an orifice plate was fitted to the blastpipe. (An American idea, I believe). This reduces the discharge coefficient from 0.99 to about 0.90, so the effective area (working back to a common 0.99 coefficient) is about 10% less- so not as good as you might believe from a backpressure point of view.

There are then locomotives which show lower back pressure than they ought. Amongst those tested at Swindon, there are instances where the reported pressure is about half what it ought to be (Duke, DC, King DC, MN Lemaitre, to a lesser extent King Single chimney). Indeed the MN results at Swindon are about half of what Rugby reported (and the Rugby results are slightly lower than you would expect). At Rugby, the Duchess DC is about half of what it should be, and half of what the LMS reported. I have worked through the Duchess case carefully, and the problem is due to the fact that the pressure was measured at the base of the blastpipe, and there is very slight narrowing to the orifice, hence pressure increase at the tip. I suspect, but haven’t proved that this is the problem at Swindon.

Rugby did test a Giesl ejector vs a Standard orifice plate DC on a BR9. The free nozzle area for the Giesl was given as 30.2 sqins, though the reported back pressures are lower than expected, more consistent with 33.2sqins free nozzle area. I am assuming, but have not proved that this is also a measurement problem. Looking at the data you quote, the backpressure for the Lemaitre is consistent with a steam rate of about 25000lbs/hr. For the Giesl to be as good as Giesl claims, the free nozzle area would have to be over 40 sqins, equivalent to a 7 ¼ “ single pipe. I do not know what 34064 had, but I’d bet it wasn’t this. So, non -comparable measurements I think. I don’t think 34064 was indicated, so Giesl’s power calculations are on the basis of what would happen if the backpressure drop was as large as claimed. The Giesl on the BR9 didn’t dramatically change perceived performance. The reason, I think is simple. The steam rates required on Newport duties for BR9s were so low as for the reduction in backpressure achieved to not be significant. Giesl couldn’t believe how little and how lightly the BR9s were used.

It’s so long since I wrote in the thread, I’m not sure which boiler programme you are referring to. I have one of my own, rough and ready but works, but it cannot answer questions of the kind you ask. I have one from Professor Hall (He was chairman of the British Heat Transfer Group, so knew what he was doing!), which he abandoned uncompleted. He was concerned about some aspects of the theory; it is also obvious that it is a heat transfer model that does not take into account all factors involved in boiler efficiency, most crucially coal loss. I had rather ignored it, but recently discovered that it makes a quite remarkable (to me at least) prediction, namely that if there were no radiant heat transfer to the superheater, superheat would be very poor indeed. Obviously there is a radiant section at the beginning of the firetubes, and it is here where the majority of superheating occurs. What is critical than is the surface area of the superheater in this radiant zone, which will be greatly influenced, as you suggest by how near the superheater gets to the firebox. Chapelon consciously got his tubes as close as he could; because of the risk of burning the elements, most others preferred to keep them some distance away to reduce maintenance, but at the cost of superheat.

On this logic, if you make the firebox bigger, you could well reduce the radiant area in the firetubes which would not be good for superheat. Looking at US designs with large combustion chambers, even with A type superheaters, the level of superheat achieved was good but not top of class, although the PRR T1 with an E type superheater was excellent. It was short between the tubeplates, and I wonder if this meant the ends of the superheater were allowed nearer the firebox. The Chapelon 4-8-0 also had a very short boiler, and I wonder if, when tubes were short, designers, who were trying to maximise superheater areas, didn’t finish up with elements nearer the firebox.

The superheat measured on the V2 at Swindon was very good, also on the BR7 boiler of similar length, even better on the even shorter BR9.

So, I think the question you put is really interesting, but Bill’s boiler model is not good enough, nor is it validated against data, to allow conclusions at the level of detail your query requires.

On your corrected comment, I’m not sure why less water in the boiler would increase evaporation? Combustion is determined by amount of coal and draught, heat transfer is independent of boiler level surely? The amount of steam/hr/ gallon water would go up with a smaller boiler, but surely not the gross amount?

[pete1950]

 

Just in case the PRR test results for 8 compound engines tested in St Louis in 1904 come within your sphere of interest and are not in the 2 bound volumes, and if no-one has already mentioned this, here are details:

The test reports are not numbered, as with other test bulletin numbers but identified by title only, eg 'Tests of De Glehn Atlantic Type locomotive, PRR'.

The test results are in the 1905 publication 'The Pennsylvania Railroad System at the Louisiana Purchase Exposition'. The volume includes a detailed description of the Altoona test plant as erected in St Louis, together with chapters on calibration of instruments and methods of recording and conducting tests.

Cheap, print-to-order copies are available from ABEBOOKS. I was thrilled to pick up a 1905 copy in a PRR memorabilia fleamarket stall near Altoona before the advent of the web made such things so easy, albeit now at a price ($hundreds).

Regards,

Pete1950

[Dreyfusshudson]

Dear Pete,

Many thanks- I was unaware of the references you quote, and they are of great interest. Having completed the ‘American Project’ and written it up into an article for publication, I turned my attention to the feats of Chapelon’s Compounds, which in the 1930s were shown to be able to deliver more than twice the power at the drawbar of contemporary simples of similar size in this country (UK). There are two possible kinds of explanation of this; firstly the Compounds were markedly more efficient, so produced more power per unit of coal fired; secondly, the enhanced draughting allowed evaporation rates/sqft grate that were much higher than achievable with the simple draughting devices used in the UK at that time. This has been an enormous endeavour, wading through hundreds of pages of French text, and simulating the behaviour of Compounds using the computer programme.

Apart from possibly the Chapelon 4-8-0 (there are apparently conflicting sets of data in the same report), there is no evidence that French Compounds had any significant boiler or engine efficiency advantage over simples with the same superheat and exhaust pressure. The French Compounds were way ahead of most UK designs in these two respects in the 1930s, so will have had economy benefits (Explanation 1), maybe 20%, meaning overall, explanation 2 is the more important one, and if you work out the evaporation rates/sqft required for the highest claimed outputs, they could only be attained with superior draughting. However, if you then look at the coal rates required to achieve these steam rates, they are generally way beyond the amount it is reasonable to expect any fireman to achieve (most Compounds were hand fired). So, whilst the drawbar power claims are factually correct, they are simply demonstrations of potential, not of what might be expected in service, which, at similar efficiency would be no more than a simple could achieve if fitted with the same draughting and raised superheat. Many of the cylinder powers claimed are several hundred HP too high- Chapelon had a very unrealistic view of Locomotive Resistance- hence cylinder efficiencies also.

The Computer modelling of Compounds suggests that, as has often been stated, Compounding can provide up to 10% benefit in efficiency. This depends on exactly what you choose as the simple to compare with the Compound. If you fed live steam to both the HP and LP cylinders of the Compound i.e. operated it as a simple, this would have similar engine efficiency to the Compound, but you would have a brute with enormous tractive effort relative to adhesion weight, which would have to be worked in extremely short cut offs at speed- in practice an non-starter. It is against practicable simples that the benefits begin to show.

However, a number of factors can eat away at the 10%. Some of these issues Chapelon rectified, meaning older Compounds would not achieve 10%, and 10% also requires that you use the right combination of HP and LP cut offs.  Most importantly, my calculations suggest that unless the inlet steam temperature is above 750^oF (as on the 4-8-0, but not the other Compounds), there will be significant condensation on the walls of the LP cylinders, reducing efficiency; on simples, cylinder wall condensation ceases above about 630⁰F.

All of which says that when experiments were being done on Compounding in the early 1900s, it is quite possible that there would be little or no efficiency benefit for Compounding vs a well optimised simple. The Great Western Railway bought four De Glehn Compounds in this period and came exactly to this conclusion. That the de Glehn system was actually tested on the PRR plant is very exciting news, for one should be able to analyse the results to find out exactly what was going on in machines of this era, which by and large sealed the fate of Compounding. If the De Glehn reports do not form part of the numbered sequence of Altoona reports, then they are probably not at the library in London.   I have found a source in Google books, The Pennsylvania railroad system at the Louisiana purchase exposition: locomotive tests an ... but unfortunately this exceeds the 700 page limit, so will not reveal its contents. I may be able to source this in the UK.

 

[GP40-2]

Dreyfusshudson

 

 

...All of which says that when experiments were being done on Compounding in the early 1900s, it is quite possible that there would be little or no efficiency benefit for Compounding vs a well optimised simple.

 

This is an interesting statement. The B&O EL Class were originally built around 1916-1920 as compound articulated locomotives. In the late 1920's into the 1930's the B&O did a rebuild of the ELs in their own shops, converting them into simple articulated locomotives. Among the many changes, the boilers were redesigned to increase the draft and increase the superheat temperature. The test results not only showed a large increase in horsepower at speed, but an overall increase in efficiency. The final result was they took a locomotive that was designed as a slow speed drag locomotive, and ended up with a locomotive that was quite good at pulling manifest freight at moderate speeds, with a decided HP advantage over their S Class 2-10-2 (which were highly regarded freight haulers in their own right). The ELs ended up in service on the B&O right up into the mid 1950's before being scrapped.

[Neil Robertson]

Dear DreyfussHudson,

Thank you for your interesting reply. I agree that some historical data seem inconsistent or wrong, even from Swindon and Rugby. For an interesting but very demanding review of locomotive resistance, people might be interested in a piece 'Steam Locomotive Resistance' by John Knowles that has been posted on the web.

Regarding the West Country, a steam rate of 25000 lb/h seems typical of everyday use. Looking at curves of blastpipe pressure against steam flow rate for various degrees of superheat given by Giesel, for a pressure of 0.13 atu, flow rate is around 40 kg/(cm2 h). For 25000 lb/h this needs an area of 283 cm2 or 44 in2 (7.5 in diameter single blastpipe). This seems too high given that the Kylchap in the A4 was 39.23 in2. The only example I have found so far for a Giesel is for a Class 78. This was a medium power fast tank engine. When new in the 1930s, Giesel says blastpipe area was 128 cm2 but by the early 1950s some had only 95 cm2. Perhaps they had steaming problems similar to some British types postwar. From 1956 they received Giesel ejectors and higher superheat (through throttling gas flow through the large tubes while easing it in the small tubes). Giesel gives the ejector area as 232 cm2. This corresponds to a single blastpipe of  6.75 in diameter. Nominal steaming rate was 12200 kg/h or 26900 lb/h. The 7.5 in blastpipe would only look consistent if Giesel thought the nominal steaming rate of a West Country was 32800 lb/h with an all-out maximum of at least 36000 lb/h.

Regarding his time in the US he says he rode on a NYC 4-6-4 during tests in the early 1930s. The engine was one of the 1927/30 batch. With coal was of 7000 calories he says and normal evaporation rate was 60000 lb/h, maximum 70000 lb/h. This corresponded to an evaporation rate per sq m of heating surface of 77 kg/(m2 h), 90kg/(m2 h) maximum. This he says was far over the prewar German limit of 57 kg/(m2 h) and even their 1950s 'high performance boilers', which were 75 kg/(m2 h). Blastpipe pressures were a high 1.2 resp. 1.6 atu. He says the exhaust sounded like gunfire. He comments that the gas free area was relatively good. Further, in connection with the high specific and especially absolute boiler tube resistance, together with the self-cleaning smokebox, and the high performance, the overall design was good. Certainly better than the German.  

  

[Paul Milenkovic]

What is the compound supposed to do?

Is one thing to get a greater expansion ratio of the steam within the limits of a particular type of valve gear (i.e. Walschaerts)?  That is, if you had a simple with the right valve gear without the "wired drawing" (throttling) limitation at short cutoffs, it could achieve higher expansion ratios?

Is another thing thermal, that by expanding in separate stages you keep the HP stage hot, the LP stage not quite as hot, and reduced thermal losses?

Is a third, possible one of cylinder friction, that cylinder optimized for a certain inlet pressure you have a certain amount of friction that there is a diminishing return for short cutoff and high expansion ratio?  That a LP cylinder can have lower friction than a HP cylinder of the same volume on account of the lower steam pressure to seal against.

Working against any efficiency gains of compounds appears to be indicator diagrams I have seen.  It seems there is considerable white space on the indicator diagram separating the HP indicator loop from the LP indicator loop, suggesting there are losses in the HP exhaust, LP intake and the receiver in between to contend with.  Does this make sense, that compounds incur losses in transfering steam from the HP to the LP cylinders?

[Dreyfusshudson]

Thanks- most interesting- more evidence that early 20th Centrury Compounds weren't fulfilling their potential. I did read Warden's book on the NW Ys in the unfulfilled hope of getting some test and technical performance data on them. He clearly was a big fan of the Ys, but never talks about efficiency benefits, and himself wonders if it would not have been better for the NW to have built more versatile As instead of the last Ys.

[Dreyfusshudson]

Dear Paul,

 

The $64K of question, of course! I drafted my thoughts on French Compounding into an article for a general audience, deliberately skating over some of the hard questions. I circulated to local experts who all came back and said that the fundamental questions needed tackling head on, so a complete rewrite needed. I now have four pages that attempt to answer your question, enough to send even the most dedicated reader to sleep. So I have  had a go at trying to simplify it in response- still over a page- so see what you think. (I'll get to your specifics at the end).

Why Compounding?

 

There are two ideas to consider.

 

Firstly, improve expansion ratio, the volume of steam in the cylinder when exhaust ports open divided by volume of steam in cylinder at cut off, including clearance volume.  The higher this is the better, providing the expansion is isentropic, as it is reasonable to assume in a simple engine.  If you calculate expansion ratio for a compound, even flat out in 70% cut off in both cylinders, and allowing for 25% CV in the HP cylinders, this is in excess of 3, whereas for simples, the range varies from about 1.5 to 2.5 as cut off reduced from 50% to 20%.  There is however a problem, not always recognised in steam age literature, namely that the expansion from HP inlet to LP exhaust in a compound is not isentropic, because of what happens between HP exhaust and LP cut off, so the above ratios are misleading, and the expansion ratio benefit is not actually known from these simple sums.

 

So the second, underlying idea is to maximise isentropic efficiency, in other words reduce irreversible losses associated with increases in entropy.  The maximum amount of work possible from a heat engine-the isentropic (constant entropy) ideal is, for a given inlet temperature and pressure and exhaust pressure, fixed by the second law of thermodynamics.  There are in fact four major sources of irreversible losses:

 

• Incomplete expansion; the ideal is for steam to be completely expanded down to exhaust pressure, which also ideally should be as low as possible- atmospheric or even below. In practice, the cylinder pressure is always above exhaust pressure when the exhaust ports open, so this is one source of irreversible loss. These losses are greater at lower speeds than higher speeds.
• Adiabatic flow through the ports with an associated pressure drop. When this occurs, the entropy of the steam increases, hence the isentropic ideal is not met. This happens a) at inlet, when compression pressure has not reached steam chest pressure and b) during inlet, when the flow of the steam through the ports is not fast enough to maintain cylinder pressure at steam chest pressure, as always occurs in practice at higher speed (wiredrawing).
• Losses due to condensation on cylinder surfaces.
• Losses due to leakage from and within the steam circuit e.g. past piston rings.

 

So how do Compounds fare on each of these criteria?

 

• On the first, the whole idea of Compounding is to increase expansion, but as noted under the first idea above, this may not be as good as one might think.
• On the second, there are a number of conflicting factors, and it’s not possible to say a priori; indeed this points to the root of the theoretical problem in steam days. This is in essence a thermodynamic analysis, which is great for reversible processes, but once irreversibility is involved, quantitative predictions are not possible. So, thermodynamics doesn’t help; what you need is a fluid mechanics approach, coupled with thermodynamics, such as in the programme I have mentioned, but fluid dynamics was not well developed enough in steam days for them to apply it.
• The problem of condensation was recognised in steam days, but heat transfer was not applied in a quantitative sense, so its occurrence was poorly understood. Chapelon believed there would be condensation in Compound HP cylinders- quite wrong at normal superheat levels. We have enough understanding now to scope out the problem, and as I wrote, it seems that if inlet steam is above ca 630^oF in simples, you’re OK, but in Compounds you need to get above 750^oF to avoid LP condensation. At low superheat or with saturated engines, you have more cylinder surface in Compounds, and it could well be that you net out worse (haven’t attempted these sums yet).
• Since leakage within HP cylinders finishes up in the LP cylinders of a Compound, one only really needs to consider LP leakage, where MEP will be lower (probably good) but cylinder diameter greater (probably bad). Again, not possible to say a priori which wins, but Compounds may have an inherent advantage here.

What’s important to recognise in this is that if you take a simple engine working in short cut off, and high superheat, i.e. no condensation, at realistic backpressures you can in principle get isentropic efficiencies of ca 85%- the cylinders are doing a pretty good job of getting the most that is possible from a given inlet temperature and pressure. The ‘in principle’ refers to the fact that, from UK test data, there is usually at least 5% leakage to take into account, which means that isentropic efficiencies in the low 80s were in practice the best achieved.

 

So, if a good simple can get 85% isentropic efficiency, there’s only 15% to go for, whether through compounding or valve gear design. The Chapelon 4-8-0s did, by my as yet provisional calculations achieve isentropic efficiencies in the low 90s, 10% better than actual simples. However this figure only applies if a) the superheat is high enough into eliminate LP condensation, and in most other cases it wasn’t and b) the right combination of LP and HP cut offs is used, which depends on both the design of the engine and the actions of the crew and c) there is no pressure drop between HP exhaust and LP inlet.

 

The design constraints on a steam locomotive are very considerable, which means, I think, that it is difficult to get the few % improvement in thermodynamic efficiency that is possible, which is why most abandoned the attempts to do so. Compounds were successful in a marine environment. The Titanic had triple expansion Compound engines rated at 46000HP. However, there are no hills or stations between Southampton Water and the Hudson, nor could the operating department add extra coaches for peak periods, nor was adhesion an issue, nor were there constraints on cylinder size- a much simpler environment.

With respect to you specifics;

a) It's clearly an objective in any engine to reduce wiredrawing, hence attempts to improve valve size etc. Whether, because Compounds were working at high HP cut offs against a higher backpressure, hence with higher pressure at cut off there was a net benefit is hard to say, but bear in mind if there's only 15% to go at, and a lot of this is due to incomplete expansion, it's maybe not as big a deal as people thought. Also bear in mind that you will have a second set of wiredrawing losses in the LP cylinders.

b) Not sure what you mean by 'reduced thermal losses'- do you mean e.g. convective and radiative losses? These seem to be small

c) Cylinder friction losses are not that high. I have worked with a friend to produce a good quantitative model of Locomotive resistance. A key element of this is to estimate 'machine friction' losses from first principles, which includes of course losses associated with cylinder friction. The friend's estimate comes out as rising from 3HP at 20mph to 14 HP at 90mph for a large UK 4 cylinder Pacific. Even if this is out by a factor of two, it's not such a big deal.

d) On the earlier French Compounds, the LP inlet pressure was below the HP exhaust pressure, so there was an unhelpful pressure drop of the kind you indicate (see above). Chapelon's development work elimiated this- so it is possible with the right valving. 

 

[Paul Milenkovic]

I think we are pretty much in agreement that steam locomotive 2-expansion piston compounds can have losses between the HP exhaust to the receiver and the LP intake from the receiver.  As I mentioned earlier, this shows up in compound indicator diagrams where there is the HP pressure-volume loop, the LP pressure-volume loop, and this big gap between the two loops representing what you call entropy loss (i.e. lowering in pressure without doing any mechanical work).

With respect to the thermal losses I had suggested, one of the knocks on steam-locomotive piston engines is that you have alternating hot inlet steam and relatively cooler exhaust steam flowing in and out of the the ends of the pistons, and such things as the Skinner Uniflow engines, used on the Lake Michigan ferry boats but never in railroad applications are supposed to reduce thermal losses from this effect.  Separate into HP and LP cylinders, in the words of a gimmick sandwich package from a fast-food restaurant, are supposed to be better at "keeping the hot side hot, the cold side cold."  What I am thinking of are conductive losses, of heating a chunk of metal in the cylinder with inlet steam and then cooling that same chunk of metal with exhaust steam.

With respect to conductive losses, why is condensation, especially in the LP cylinder so bad?  I can understand you don't want too much fluid in the piston that you burst something with hydro-lock, but why the worry about saturated steam?  The only thing I can think of is that if steam is condensing on the cylinder walls, you are transfering enormous amounts of heat as losses to those walls whereas if the steam stays as steam, you get much less heat transfer to the walls?

With respect to high superheat temperature, I don't think that the upper limit on superheat should be lubrication of the cylinder.  You probably don't want high superheat during "hard working" -- 85% cutoff -- but with respect to high superheat under 15% cutoff where you are getting a lot of expansion, the high piston temperatures are only intermittent, much as a Diesel engine that they know to keep oiled in the face of very high intermittent temperatures.  On the other hand, oil is typically injected into the steam, which has the effect of exposing the oil to full superheat temperature even though the steam cools down upon expansion in the cylinders.  Perhaps one would have to inject lube into the cylinders more directly.

The real limit on superheat temperature is the superheater tubes.  Otto and Diesel cycles can have very high cycle temperatures because the applied heat is intermittent, and there is also water or air cooling on the outside of the cylinders and cylinder heads.  Gas turbines cannot have quite as high turbine inlet temps, but again, there is a hot side and a not-hot side of the turbine blade, and various cooling schemes can be applied.

With a steam superheater, the material has to withstand the full superheat temperature on both sides of the tube, which I see as a limit to steam cycle temps, even if you go to exotic "superalloys" for the tubes as you do for turbine blades.

In light of this, the most efficient steam cycle, within the limits of boiler pressure brought on by scale formation in a cycle that is not using highly distilled water, and within the limits of not using a condenser as you don't have access to river water in a moving steam engine and you do in a stationary power plant, that steam cycle should use a high degree of reheat between HP and LP stages, yeah, yeah, the reheat tubes and headers also add to pressure drop.

I am thinking that not only do you want the exhaust to be above condensing, you want the exhaust to approach the boiler temperature.  That way you could use exhaust steam to not only boost the feedwater temp up to atmospheric boiling of 212F, you could boost the feedwater temp up towards the boiler water temperature of around 400 F?  One would have to work all the details of the thermodynamics, but I believe highly-optimized stationary powerplants use reheat between compound stages along with transfer of heat from LP exhaust to boost the feedwater to a high temperature (the feedwater heater needs to apply heat to the feedwater after injection to boiler pressure for this to work).  And if automagically you could use this excess exhaust heat to pre-heat the combustion air, so much the better.

Before you think I am crazy, the British 5AT project (they have a Web site) has some kind of metal partition and use of the front part of the boiler at the smoke-box end as a kind of preheat-the-injected-water-to-boiler-temperature setup, but they are using the leftover heat from combustion in the flues just prior to the smokebox.  I am proposing using exhaust steam heat for this purpose by applying reheat in a compound to boost exhaust steam temps, and I am thinking the power plants already do this.

But yeah, yeah, in the steam locomotive application, these schemes to wring out more efficiency involved more hardware (I am not giving up on compound expansion and I am adding a reheater in addition to the superheater along with a high-pressure feedwater heater), and in the end, it was the maintenance costs not the fuel costs that were the dominant economic effect.

[pete1950]

Thank you for your reply.  I must correct an error in my post though - only 6 compounds.

The 8 tested were:

1. simple PRR 2-8-0 type H6a

2. simple LS&MSR Brooks 2-8-0 class B-1

3. cross compound MCR Alco 2-8-0 class W

4. tandem compound AT&SF Baldwin 2-10-2 class 900

5. DeGlehn PRR SA de CM essentially an exact duplicate of those supplied to the Northern Rlwy of France

6. Vauclain balanced compound AT&SF Baldwin 4-4-2 class 507

7.balanced compound w/ superheat Royal Prussian Rlwy Admin  HMAG 4-4-2 class S8

8. balanced compound NYC&HRR Alco 4-4-2 class I1

Pete

 

[Neil Robertson]

One of the better early compounds was the Bavarian 3/6 built between 1908 to 1931. Even in the 1950s they were preferred by many crews to the larger 01s and 03s on the hilly curved route from Munich to Lindau. Coal consumption for a round trip was about 2 tonnes less and oil consumption was similar despite the newer engines having only two cylinders. Their better balancing allowed them to start more easily than a two cylinder engine and acceleration up to 62 mph was better. Above this their machine resistance and corresponding wear rose rapidly. Maximum speed permitted was 74 mph (coupled wheel diameter was 1870 mm). Compared with the 3 cylinder 01 and 03 10s they were easier to lubricate (all had bar frames). The workshop personnel preferred the simpler 01s and 03s but overhaul costs were of the same order because damage to frames was unusual in an S3/6 on account of their smaller and better balanced piston forces.

Minimum coal consumption was 5.8 kg/ IPS, an excellent figure corresponding to about 795 C steam temperature in a simple expansion engine. With tube length limited to 17.25 ft evaporation rates up to at least 70 kg/(m2 h) were attained (2300 IPS), well over the official 1770 to 1830 IPS.

There was a small series with 2000 mm coupled wheels, which were used also outside Bavaria, including on the Rheingold in the 1920s and 1930s.

Despite their modest adhesive weight of 52.7 tonnes they handled trains of over 600 t. 18 478 is again operational.

They were a well balanced type with strong frames and a good boiler, as well as the good balance of their 4 cylinder layout.These other factors make it is hard to say how much their being compounds contributed to their success.

 

[erikem]

Paul Milenkovic

The real limit on superheat temperature is the superheater tubes.  Otto and Diesel cycles can have very high cycle temperatures because the applied heat is intermittent, and there is also water or air cooling on the outside of the cylinders and cylinder heads.  Gas turbines cannot have quite as high turbine inlet temps, but again, there is a hot side and a not-hot side of the turbine blade, and various cooling schemes can be applied.

With a steam superheater, the material has to withstand the full superheat temperature on both sides of the tube, which I see as a limit to steam cycle temps, even if you go to exotic "superalloys" for the tubes as you do for turbine blades.

Paul,

My understanding is that the amount of superheat is limited by water disassociating into OH- and H+ at much higher rates than at room temperature. Both ions can be very corrosive and make life miserable for any metal in their presence.

- Erik

[Dreyfusshudson]

Dear Neil,

Thanks for this info from Giesl. HIs conclusion about typical steam rates with NYC Hudsons conflicts with my own analyses, which suggest that no more than 3500IHP was required to gain significant amount s of time of the fastest schedule, the 20th Century with about 1000 (US) tons, no more than 50000lbs/hr cylinder flow, say 45000lbs/hr blastpipe flow with 10% feedwater recycle. I seem to have found some data that the J1s blastpipe as having increased from 6.75" to 7 3/8" over time, but with 6 sqins lost due to  1/2" bars. At the lower area, you would indeed get blastpipe pressures around 1 atmosphere, at that kind of steam flow.

You will find me sceptical to the point of tedium on most claims from the steam age (maximum speeds, powers, efficiencies etc) especially where there is a personal or company interest involved. Many don't stand up to detailed scrutiny, or are data massaged in such a way as to present the interest group in the best possible light. (Nothing new here, then). So, I feel like a bit of a curmudgeon, but getting to the bottom of matters is what interests me.

I was interested in your comments about maximum German evaporation rates. I haven't looked at German practice much, but I did do some analyses of running at the end of steam from Bremen to Osnabruck, and ws surprised to find that the running of a heavy 600tonne train which gained 10 minutes on the schedule was done with little more than 2000IHP, this from an oil fired Pacific ( no fireman limit) with  a 43 sqft grate, no more than about 600lbs/sqft/hr depending on boiler feed system Given this effort was significantly in excess of what was normally needed, I formed the impression that German boilers were not steamed hard.

I corresponded extensively with John Knowles during my researches into Locomotive resistance, and the basic approach we adopt is the same, seeing overall locomotive resistance as the sum of its resistance as a leading vehicle, given by the Davis equation, and the resistance of the machinery between the pistons and the wheel rims (MR). This, I later discovered is the Baldwin/RP Johnson approach, much more realsitic then the German Strahl equation. We were actually quite close in our estimates. Where we differ is in the value of MR; he uses an empirical approach which gives values somewhat higher than given by the first principles anaylsis done by a mutual friend mentioned in another post. I think these first principles figures are in much better agreement with UK test plant data for MR, and overall Locomotive resistance as measured in road tests- but we are talking relatively minor differences on a topic where there are very widely divergent published views, that do not really affect the outcomes mentioned here. Interestingly, the value of MR proposed by Johnson is I think way too high. It is based on Altoona testing, but these figures can be shown to be erroneous, a minor blot on an otherwise priceless set of data.  

 

[Dreyfusshudson]

Dear Paul,

Thanks for these comments and ideas.

With respect to fluctuations in cylinder temperatures and the reason for the negative impact of condensation on LP efficiency, let me give you my understanding of the situation.

There’s no problem with saturated steam being formed in a cylinder during expansion, as you suggest. However, I’m talking about condensation not of steam in the cylinder, but of the steam onto cylinder walls, something which does impact efficiency. If the cylinder wall temperature is below the saturation temperature of the inlet steam, then a film of water forms very rapidly on the cylinder surface at admission. Since water is a poor conductor, it very quickly reaches a limiting thickness - up to 0.1mm. (Thickness depends on P^1/2, and the cylinder temperature). Now during expansion, the steam in the cylinder falls below the temperature of the cylinder wall and its film, and the water film re-evaporates, and disappears up the exhaust without having done any useful work- a kind of leakage. The amount lost also depends on the surface are of the cylinder; on smallish UK cylinders, the amount can reach 1000lbs/cylinder/hr, not  a large amount, but sufficient to take the edge of high efficiencies. I’m maybe making the LP losses through condensation more dramatic than they likely are, but they could eat away several % of any expansion ratio benefit compounding brings.

Question then is, what is, and what determines cylinder wall temperatures. These questions were addressed at Altoona in 1912; obviously the temperature lies between inlet and exhaust temperatures, and depends on the temperature of the inlet steam, and cut off- the greater the exposure to live steam, the higher the temperature. Knitting these data together with a theoretical model which calculates the size of the film under various operating conditions allows you estimate how much steam is wasted by this process. It is this kind of analysis that says that, in the HP cylinders of Compounds, where the exit temperature is always higher than the saturation temperature of the inlet steam, there will be no wall condensation. In the LP cylinders, where wall temperatures will be lower, there will be, despite the lower saturation temperature of the inlet steam.

This condensation process is the main source of heat transfer losses in cylinders. The Altoona data do not mention temperature fluctuation during the piston cycle, and Professor Hall, who developed the quantitative cylinder condensation model mentioned above built an apparatus to measure cylinder wall temperatures over the cycle. What he showed was that the block as a whole reached an equilibrium temperature, and if no condensation occurred, as with high superheat, there was no fluctuation in the temperature of the wall. When condensation did occur, there was a fluctuation in the very top of the surface, as heat flowed in from the condensation, then flowed out as the water film re-evaporated.

http://5at.co.uk/uploads/Bill%20Hall%20software%20and%20papers/The%20Effect%20of%20Superheat%20on%20Cylinder%20Condensation.pdf

 So, clearly, the condensation process apart, there is little heat transfer to and from the cylinder walls during the piston cycle, once the equilibrium temperature of the cylinder block has been reached. Chapelon feared heat losses along these lines, but his fears were I think unfounded.  

As to what would happen to cylinder wall temperatures if you went to very high superheat, obviously there’s no data, but like you I suspect any problems are surmountable. However, there is another effect worth considering which is that as you increase superheat, so the temperature of the exhaust goes up, and with it the back pressure at a given exhaust flow. With conventional exhaust systems, this means that once inlet steam gets up to 800⁰F, you are already getting into diminishing returns, and by 1200^oF the benefits disappear and things get worse, according to my calculations . I suppose you would be able to optimise the exhaust to deal with this, haven’t thought about this though.

With respect to your comments on lessons from modern power plants, perhaps the biggest lesson is that if you want to use steam efficiently in a heat engine, turbines are the way to go? Come back Jawn Henry? Seems to me you would only go down this road if there were only coal left to burn, and the economics of electrification didn’t add up.

[Dreyfusshudson]

Dear Pete,

Thanks for the clarification- a treasure trove that I need to get my hands on. Particularly interested in the De Glehn, as it ties two continents' data  together.

[Dreyfusshudson]

Dear Neil,

Thanks for the efficiency data on the 3/6s- at about 12.8lbs steam/ihp-hr, that's up with the very best simples. ( Best measured in the UK was about 13.2; the Caprotti Pacific Duke of Gloucester was claimed to be better, but if you look at the raw data, this is another example of the data being stretched to prove a point; I think it was no better than 13.2. The Kylchap A4, having similar superheat but better exhaust pressure was possibly in the high 12s). From what little I know of the 3/6s they had a large superheater area for a 17.25' tubes, so superheat was probably very good.

The comment about low adhesion weight is very interesting. When analysing logs of French Compounds, I was very struck by a run of a P-O 4-6-2 on the Bordeaux line, when the normal 235 ton Sud Express was made up to 600 tons for the fastest stretch, 1 hour for 70 miles with a 75 mph speed limit. This was for the benefit of A I Lipetz visting from Alco. The key to keeping the schedule was getting this huge load started quickly, which they did, with no more than 54 tons of adhesion weight, far faster than any Pacific in the UK would have done I think; these slithered even with 67 tons adhesion weight. The smoothness of the Compound drive has often been commented on, not sufficient reason to persuade the accountants it would seem.

[Neil Robertson]

Dear Dreyfuss Hudson,

Thank you for information about efficiencies of British pacifics and your thoughts about machine resistance.

I too was struck in rereading old Locomotive Profile No. 13 on Nord Pacifics by Brian Reed about the high loadings allowed compound engines. On Calais to Paris boat trains Nos 3.1252-90 hauled 600/675 tonne trains despite having only 56.8 tonne adhesive weights. Nominal factor of adhesion was a very low 3.33. I thought then that this was because this was a fairly flat route, but in the case of the S3/6 they were intended for the hilly countryside of Bavaria. The Norfolk &  Western also was a believer in low factors of adhesion. Perhaps an advantage in adhesion was one reason for the great success of the Class Y6.

Giesel, without giving any examples of particular locomotives, also says that compounds could start and accelerate better than simples. The reason he gives is that, although they worked at a shorter cut-off overall, the cut-off in the individual cylinders was longer than for simples. This gave them a more uniform torque, which translated into less likehood of slipping. I don't know anything about modern traction, but it may be here too that there is more uniform torque than with steam and lower factors of adhesion.

 

 

[Dreyfusshudson]

Dear Neil,

The Nord Paris- Calais route is generally very well engineered. Normally gradients are 0.3-0.5%, (1/333- 1/200), steep enough to require high sustained power outputs at speed, which the 35 sqft grate Nord Pacifics certainly achieved. However, leaving Calais ‘cold’ there is about 7 miles of 0.8% (1/125) to Caffiers, before dropping back to sea level at Boulogne, then a few miles of 0.7% out of Boulogne. Serious adhesion tests in a maritime environment, I would think. With a simple, even slogging away at 40 mph up this grade, you would be back in something like 35% cut off, so the more even torque from the longer Compound cut offs might come into play.

If you look at UK practice, by my estimates it was unusual for locomotives to use much more than about 16000-18000lbs TE when starting , 50-60% NTE only, even with the heaviest passenger trains of around 650 (long) tons gross (loco +train) weight. In the US, gross weights were easily double this, but since you can only increase TE in proportion to driver load for a given number of axles  (Max UK load about 22.5 long tons,  US 30-32 tons), a much higher proportion of starting TE would probably be needed on starting- not analysed enough logs to know for certain. Hence the use of boosters. This may explain why US designers were so hot on the ‘rule of 4’ for the ratio of driver load to NTE. Hence the limitation of the maximum cut off on the ATSF 4-8-4s, which surely could have used a bit more TE on their biggest challenge, climbing Cajon Pass. Maybe the rule of 4 could be relaxed in a less demanding environment?

So, if high power was not used on starting, why did UK Pacifics, often slip? The text book says full gear and part regulator on starting, which should give the same torque as a Compound? Maybe other factors, e.g. slow release of brakes, uneven track, sharp turnouts, leaking lubricant contributed. At speed drivers would often work in 25-30% cut off, part regulator, again maybe to get a more even ride, but not what the text books say. In fact the loss of efficiency due to lengthen cut off between 20% and 30% is pretty small, so this was not really an issue, except with the Bulleid Pacifics, which, especially with their original 280 psi boilers produced so much power relative to the schedule requirements that they were often worked in 40% cut off, ca100 psi, which is very inefficient, hence the class’s coal eating reputation.

Lots of mays, mights and maybes in this, which is a shame, because the subject of adhesion is clearly crucial to operation, but something it’s difficult to get hard data on.

Back to the subject of blastpipes and superheat. I thought this clip might be of interest. The MR main line from Bristol to Birmingham has a stretch of 2 miles at 2.6%, thanks to the intransigence of a local landowner in Victorian times, an operational nightmare to this day, banking imperative in steam days.  This spectacular recreation last weekend involved a Pacific with 3^rd Division superheat, primitive blastpipe- the least efficient UK Pacific by far- listen to the racket/music it makes- banked by two 0-6-0s with 4^th Division superheat (none), basically an 1890s design.  (from about 1’40” on) Which just goes to show that efficiency isn’t what gets people’s emotions.

http://www.youtube.com/watch?v=7nCakwix6pQ&feature=youtu.be

 

 

[NM_Coot]

Santa Fe test results showed that the 3765 class with limited cutoff could maintain full boiler pressure at all speeds and cutoffs.  Test runs with 3766 showed 5450 MIH at 55 to 70 mph and over 5000 IH from 35 to 90 mph.  Given the Santa Fe penchant for long runs at high reliability, this high HP over a very wide speed range makes for excellent performance with heavy, fast services where maximum output is required over wide speed ranges.  The concept of limited cutoff was recommended for implementation on the 3460, 3776, and 5001 classes.  In passing, the test of 3766, as equipped with roller bearings on all axles, gave a mechanical efficiency of about 90% up to 30 MPH and then dropped to about 82% at 60 MPH and 62% at 90 MPH.

[Dreyfusshudson]

Many thanks- I agree that at speed the limitation on maximum cut off is of no consequence, and at 0-20mph is a good idea if you need to put down maximum NTE on starting, as was the case. The only situation where it might have helped a bit is climbing to e.g. to Cajon at 20-30 mph. At this speed, even in full gear the engine can't deliver more than 70000lbs TE, so the 'rule of 4' is not broken, and you could get an extra 500HP, albeit at a pretty astronomical steam consumption.   

I read ATSF 2926 is being restored in Albuquerque. To see it running north to La Junta would be well worth the price of an air ticket, hopefully it will run before old Santa Fe route is closed, as many seem to think it will be.

[Neil Robertson]

Dear Dreyfuss Hudson,

Thank you for the recording of Princess Elizabeth on the Lickey Incline, I enjoyed this the more so because I saw it earlier that day at Bristol Temple Meads. I agree the sound is indicative of a constipated exhaust. The outside steam pipes struck me as small in relation to those on Castles and Kings. And the mechanical layout is somewhat similar that of the Thompson Pacifics with a related tendancy to loose cylinders and leaking exhaust steam pipes. They appeared to be an obvious candidate for at least a double blastpipe and if they had belonged to any other Region would probably have received one (the LMS tests of 1934 already showed the advantage).

British Pacifics appear to have had an advantage in percentage of weight on drivers compared with both Continental pacifics and American 4-8-4s. But the US engines had the advantage of bar frames and roller bearings.

The Germans, in the post-war (Witte) era belatedly adopted combustion chambers, reduced the grate areas somewhat, and moved in some cases to roller bearings. Giesel says chimney diameters became a bit less extreme (in the Class 50 the chimney was 590 mm diameter under Wagner and 520 mm after Witte's retuning). The reason for the reduced grate was to reduce radiative losses (German expresses usually had more stops than British ones). The roller bearings, sometimes also on the rods, were not a complete success. Several locomotives, including one or both the Class 10 Pacifics, were condemned because of broken rod roller bearings. Axle roller bearings sometimes overheated (especially on 03 10s), with the need for immediate further lubrication and delay to the train concerned. The mixed type of feedwater heater also was not a success, nor the American type superheater regulator (although the Class 10s also had an auxiliary regulator in the normal position in the dome that avoided the kind of disaster that befell Blue Peter at Durham about twenty years ago). Performance seems to have been a bit below the British level, although the mileages achieved were higher especially in the mid to late 1950s. There was a relative shortage of pacifics.The lifetime mileages of the Pacifics in the German steam engine museum is upper 60000 to low 70000 miles per annum for simples and 55000 miles per annum for the 18.6, although these include ususally a zero to low mileage at the end of the War and for several years afterwards.