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

142 replies
1 rating 2 rating 3 rating 4 rating 5 rating
  • Member since
    December 2007
  • From: Georgia USA SW of Atlanta
  • 10,105 posts
Posted by blue streak 1 on Wednesday, April 13, 2011 7:23 PM

J: I believe that N&W retired their big steram last Switchers were first to go in my area.. I think that a "Y" was the last to drop its fires under regular service. Anyone more info? 

  • Member since
    April 2001
  • From: Roanoke, VA
  • 1,882 posts
Posted by BigJim on Thursday, April 14, 2011 3:22 AM

blue streak 1

J: I believe that N&W retired their big steram last Switchers were first to go in my area.. I think that a "Y" was the last to drop its fires under regular service. Anyone more info? 

An S1 or S1a was the last to drop its fire. Sorry, don't have the exact engine # at my desk, however, it was written up in "Classic Trains" mag a year or two ago.


  • Member since
    June 2002
  • 17,359 posts
Posted by daveklepper on Thursday, April 14, 2011 4:31 AM

Just for the record,  N&W diesilization started with the Harrisberg Lline and its 4-8-0's.    Plus, of course, replacement of J's by run-through Southern Ry E-Units Monroe - Bristol.

  • Member since
    March 2016
  • From: Burbank IL (near Clearing)
  • 12,311 posts
Posted by CSSHEGEWISCH on Thursday, April 14, 2011 6:37 AM

Last operating steam on NKP was an 0-8-0 at Calumet Yard.

The daily commute is part of everyday life but I get two rides a day out of it. Paul
  • Member since
    April 2001
  • From: Roanoke, VA
  • 1,882 posts
Posted by BigJim on Thursday, April 14, 2011 9:17 AM


Just for the record,  N&W diesilization started with the Harrisberg Lline and its 4-8-0's....

No it did not!


  • Member since
    February 2005
  • From: Vancouver Island, BC
  • 22,546 posts
Posted by selector on Thursday, April 14, 2011 9:42 AM

Nor is it topical. 


  • Member since
    June 2008
  • 121 posts
Posted by Dreyfusshudson on Friday, April 15, 2011 4:27 AM

Great discussion, turning into a full time job to keep up!

I am in sympathy which most of what is written.

I do agree with CJ Hegewisch (12/04/11) that the perspective of operators was different to that of the gurus running testing stations. Our test stations were interested in cost savings from efficiency, because that is what they were designed to measure. Their work led to a few, trivial improvements to efficiency, all of which were ignored by the operators and not implemented. The only things they did that were accepted were to narrow the blastpipes of some designs, which made efficiency a bit worse, but improved the reliability of steaming. The basic philosophy (not well executed) of latter day UK designs was low maintenance.  Efficiency was pretty good, but draughting stone age, because of the predilections of those in charge of design. Clearly if any feature which improves efficiency adds to maintenance costs, that needs to be factored in. 

So, as Juniatha so eloquently puts it (13/04/11) efficiency is one thing, operating a railway another, and operational needs win out over efficiency if they are in conflict.

(Great memories of Hegewisch by the way; I remember well trying to walk from the station to State Line Crossing in a foot of snow, 13 degrees, early 1974, Train Watcher’s Guide to Chicago in hand. Crazy. Have a slide of an old CSSB unit there the following summer).

With respect to NM Coot 09/04/11, 10/4/11 and Juniatha 10/4/11 and 13/04/11 on the subject of Chapelon, he surely was one of the all time greats, and his designs demonstrated that. Juniatha’s stories of what he achieved are well known and oft repeated over here.

Secondly, it is in my experience nearly always the case that we often selectively cherry pick data to support our cases;  A number of our 100ton (110US ton) Pacifics have achieved 3000IHP, one approached 3500IHP. Did they get anywhere near this in daily service? No. Could this have been expected? No, certainly not with handfiring, only at great inefficiency with Mechanical stoking. So, you will find me less interested in extreme claims, more interested in what operational practice and needs actually were, for this will determine how much benefit could be achieved e.g. by better draughting.

Whilst it is true that going faster and heavier, which is what efficiency allows you to do ought to be a universal good, experience over here says that speed and train length were not benefits always avidly sought. Those operators again.  

Thirdly, with respect to his designs, the improvements obtained were simply by applying known best practice- three simple tenets, high superheat, low back pressure, high expansion ratio- nothing magic. (And, in the firing department, he was helped by Chauffeur Marty, who it seems was able to shovel approaching 10000lbs/hr of coal, for short periods at least. Over here, our guys seem to peak out at about 4500lbs/hr. Wimps. (I think I’m good for about 200lbs/hr)). High superheat was found in best US designs. Expansion ratio was sometimes poorer than it might have been because of the use of two cylinders only in many designs, which means that to generate high powers, particularly at low speeds you need to use longer, less efficient cut offs. However, I think it would be wrong to criticise this; I prefer to believe that the engineers had done their sums and in net terms, the simplicity of two cylinder designs far outweighed the benefits of going to more complex three and four cylinder arrangements, which could have operated more efficiently, except when they had no choice. So that leaves a draughting system with lower back pressure as the potential easy win.

As I said right at the beginning, it does seem to me that with 20:20 hindsight, it could well be that some US designs were suboptimal in draughting; this seems to me where Chapelon might have helped. I am interested in finding out if this is true, and whether under the operating conditions which were actually required, improvements would have made much difference. I suspect not, certainly not to a degree that would have saved the species for any length of time.  There is no doubt, as Juniatha writes that with later European practice, e.g Chapelon, you can break the mould of the single blastpipe, where you do generally trade efficiency for steaming, and get the highly desirable combination of better draught and reduced back pressure. This would appear to be a low maintenance efficiency improvement that could have brought about some economy, potentially an easy win. How much is what I would like to establish.

Over here the high ups in British Railways were so anti fancy foreign practices that even the Giesl was killed off. In fact, this particular decision was not too unreasonable, since it turned out that the class on which it was tested were not often steamed at rates where its benefits would have kicked in. So the real issue was that the class was not being used to their potential. The Giesl could have been applied to good effect on other designs.

One of our railways, whose CME was a good friend of Chapelon, introduced Kylchap to this country; after his death, and following nationalisation and the expiry of the patents, it was widely adopted on their passenger designs, but only in the face of strong opposition from the top. Even as late as 1960 there is a vitriolic letter from a high up slating the use of such equipment, even though it had patently transformed the classes concerned. So I think an additional point is that, over here at least, design was determined as much by personal opinions and rivalries as rational considerations.

I agree with Juniatha’s critique that what I am interested in doing is looking back with 20:20 hindsight; I am not trying to resurrect the species, which is dead.

On Juniatha’s original input, point by point

1.       My original query was exactly to ask what the backpressures experienced actually were. I am making no assumptions as to what they are. I agree that measurement is a problem. The model I use uses the standard equation for flow through a plain orifice, with the possibility of varying the discharge coefficient. This fits the measured values of blastpipe pressure with a discharge coefficient of 0.99 in nearly all UK tests. The exceptions are a) some higher values, when orifice plates, which I believe were a US invention were placed over the blastpipe nozzle. These reduce the discharge coefficient to 0.85-0.90. i.e. pressure is higher than it would be at the same orifice area without these devices. (Incidentally, although our blastpipe diameters were specified to the nearest 1/8”, and adjustments of that magnitude made, in service there could be deposits of up to ¾” of carbon below the lip- so clearly these locomotives would be very well draughted! b) lower values. These occurred in three cases. In two cases this was due to the pressures being measured at the base of the blastpipe stand, and there being very slight taper in the blastpipe. Allowing for this, the predictions are spot on. The other exception was the Lemaitre set up, where pressure measured was a bit below the theory. I have not investigated the reasons for this.

2.       I was specifically vague about the Niagara blastpipe pressure at high outputs, because there are lots of unknown factors. Precisely what it is, is part of data that I am looking for, and precisely what I think it should be will fall out when I have other design details. My programme gets the pressure for the Q2 more or less spot on.  It says the T1 had a free nozzle area equivalent to ca 8” diameter- this must be a matter of record, so this will say how good the programme is for this type. The T1 was 29 psi at 100000lbs/hr. The Niagara had a diameter of 7.5”, so by my estimation it would be higher, and the relationship is not linear. And the higher the backpressure, the more steam you need to generate a given power, so at a given target power you lose twice. So I don’t think my approaching 40 psi is too far wrong and I’ll wager it’s not too far out. We’ll see. The programme says at 52500lbs/hr the Niagara nozzle was behaving like 7.2”. This could be a measurement problem, a design feature, a build up issue, or a small inadequacy in the programme.

3.       The H8 quote is from a very respected British Engineering Journal, quoting in great detail what I suppose are primary sources; some of the original graphs are included. Blastpipe pressures are given for both front and back engines, the former being somewhat lower than the latter. The 13psi is a fair average of back and front on the highest rate test, where the evaporation is quoted as 100800lbs/hr, the tank feed 91300lbs/hr. I understand the question of where the measurement was made is important (See above), and this may lead to misleading values.

4.       Firstly I should say that in UK testing, the difference between exhaust port pressure and blastpipe pressure was negligible, especially considering the difficulties in measuring both temperatures and pressures of fast moving steam. On your comments about the Gresley set up, you make assertions, but provide no facts. The facts I know of are these a) There are plenty of UK tests on engines with similar lap to the original A1 that show no deficiency in thermodynamic efficiency. b) tests on the original short lap A1 showed the steam chest pressure was typically about 110psi. So, to develop any kind of power, engines of this size had to be worked in 50+% cut off at speed whereas the later designs with longer travel valves were worked at full pressure in ca 30% cut off. This makes a big difference to economy.  c) one of the high preists responsible for perpetrating the original efficiency story recanted once he’d figured out what the computer programme was doing and saying d) At around the time the valve gear design was changed the superheat was also increased, then boiler pressure, which is why the A3 and A4 were more efficient still. I agree that the performance of the original Castle it was compared to with was far from inspirational. The claimed 2.83 lbs coal/dbhp for that design is quite unrealistic, as later tests showed. I think I’ve figured out how the 2.83lbs claim was made; a little not unreasonable leger de main was involved to generate a great publicity claim. Like all claims, once you’ve made it, it’s very hard to deny it. (The reason the Castle got to Plymouth faster was basically that the own line driver ran through speed restrictions over the limit. The A1 was actually developing more power going uphill, but because it weighed 30 tons more it didn’t go much faster). I’m not claiming that high MEPS generated in long cut off are not a bad thing- of course they are; but as noted above they are a consequence of sticking with 2 cylinder designs. My point’s a bit more subtle, but I’ll spare everyone that.

5.       I agree I’m not taking pumping efficiency into account; I’m rather assuming that everyone understands that with more sophisticated exhausts you can, as noted above get a highly desirable reduction in back pressure whilst actually improving pumping, so, reducing back pressure looks like a bit of a no brainer now.

6.       More than happy to look at the differences between the German types if you have the relevant dimensions.

7.       Well, I suppose that’s another thing I’d like to understand; exactly how hard and for how long were US designs worked?  I could ‘build’ the line from say North Platte- Cheyenne- Laramie, and if people have passenger and freight schedules, and tonnage ranges, I could work out how much power was actually needed on this long uphill stretch.  Experience in this country says that whilst the draughting usually allowed 800-900lbs/sqft/hr evaporation, 1050lbs/sqft/hr on occasion, much over 600-700lbs/sqft/hr was rare. Maybe it was different in the US. That’s what I’d like to find out. But even the 20th Century only averaged 60 mph, and that on a Water Level Route.

  • Member since
    April 2003
  • 305,206 posts
Posted by Anonymous on Saturday, April 16, 2011 3:45 PM

"I’d like to understand; exactly how hard and for how long were US designs worked?"


Have a look here...

  • Member since
    September 2010
  • 17 posts
Posted by NM_Coot on Saturday, April 16, 2011 5:28 PM

Volumes have been written on this!  North American freight service, with a few exceptions, is NOT the place to look for continuous performance capabilities.  Here are a few passenger examples.  ATSF  4-8-4 classes worked 1791 miles, Kansas City to Los Angeles, 12 crews, Raton, Glorieta, continental, Cajon passes.  Average engine worked 18 to 20 thousand miles per month.  (2900 class 4-8-4 on freight service via Belen cutoff, only averaged 8000 to 9000). NYC 4-8-4 S-1b, in field tests, consistently produced 6600ihp at speeds of 75 to 85 mph and were capable of accelerating a 1000 ton train to 75 mph in 19,400 feet.  They averaged over 20,000 miles per month.  Kiefer was never able to optimze them because of dieselization taking place and bumping them from passenger service.  SP GS3 regularly handled 925 ton trains up compensated 1.0% grades at sustained speeds in excess of 55 mph, which I believe is equivalent to about 5500ihp.  NYC J-3a Hudsons worked the 20th Century, usually about 1000 tons, from Harmon to Chicago, 925 miles, in 16 hours with 7 intermediate stops.  In practice the J-3a boiler, in line with most modern boilers, could produce more steam than the engine could consume.  What made the J-3a so much more powerful and efficient was the care taken with the dimensions of live and exhaust steam passages and the high degree of superheat (about 750 degree F).  J-3a's on average worked about 12,000 miles per month.

Not sure if this what you wanted.  In any case it is just a sample.


  • Member since
    May 2008
  • 4,612 posts
Posted by Hamltnblue on Saturday, April 16, 2011 6:42 PM

Whenever I see a thread like this I usually do a quick google for the challenger freight video.

It puts it all into perspective.  The fact that you have so much black smoke shows that there is a lot of unburned fuel/inefficiencies, but she is a beaut.

Springfield PA

  • Member since
    June 2008
  • 121 posts
Posted by Dreyfusshudson on Monday, April 18, 2011 7:35 AM

Thanks for these comments; with respect to the video of UP3985 posted by Hamltnblue, then if you want to know why I find this topic compelling that explains everything perfectly. It has to be high up in my top ten of steam videos on You Tube. Unfortunately those of us with over active left hand sides of the brain try to explore this magic, and try and reduce everything to numbers!

Thanks to Lars Loco for the link to the previous thread, which contains some helpful stuff on feedwater heaters, a topic I’m going to need to get my brain round (they never caught on in the UK, we settled for exhaust steam injectors, which recycled roughly 6% of the heat in the exhaust steam), since this determines how much of the cylinder feed goes up the blastpipe, a very important correction especially at high steam rates. Also, the data on what the Big Boys put out at 14 mph (ca 4200IHP) is the kind of stuff I’m after. Lots of other interesting stuff in the thread too, don’t want to divert onto that!

With respect to NM_Coot’s data, that’s helpful, but not quite enough. Let me tell you what I’d ideally like, what I’d happily settle for, and why I think the point is important.

Between you, you have helped focus my original query into something like ‘How much would Chapelon style exhaust technology have improved the efficiency of latter day US steam locomotives?’ My original Porta quote implies he thought the answer was ‘a lot’. I think the answer is more likely ‘a bit, nothing game changing’. This depends on the rates at which locomotives were habitually worked, not what they could do on the occasions when they were worked flat out for short periods, for it is overall rate of working that determines overall efficiency.

Let me illustrate with some calculations I did some years back on the Milwaukee Road F7s. (Don’t take numbers too literally, my thinking has moved on quite a bit since then, but they are good enough to make the general point). In Jim Scribbins book ‘The Hiawatha Story’ are what I’d really like- detailed logs of the working of the heaviest/fastest US expresses. (My strong suspicion always was, as NM_Coot writes that freight is not the place to look for sustained high power; there are of course mountain grade where locos will have had to be worked flat out at the maximum end of their tonnage range, but a) because the speeds will be low, the cut off will be long, the cylinder efficiency down, and hence the maximum power will be some way below what the locomotive is capable of, as the Big Boy example shows) b) this is probably not typical of what they had to do over the whole trip: the short stretch from Cheyenne to Laramie is different to Laramie to Ogden or North Platte-Cheyenne. I don’t know what US freight speeds on reasonably flat terrain were, but am assuming it was probably at best 50 mph(?)), which will not require too much power - about 4000IHP with 4500 tons).

I have analysed the runs in the book, from Milwaukee to La Crosse. There is a northbound run with 100 plus 9 cars than got to Milwaukee in 73 minutes, 2 minutes inside schedule, having lost 3 minutes as far as Pacific Junction, possibly due to adverse operating conditions.  Accelerating from Mayfair, #100 produced about 4000hp for a couple of minutes to get up to 90mph by Glenview. Thereafter, 2500-3500 ihp sufficed to cover the 56.4 miles on to Oakwood in 36 ½ minutes, maximum about 100mph.  Coming south there is another run with 100, from Milwaukee to Chicago with 9 cars in about 68 minutes, net of a slowing at West Lake Forest, maximum claimed speed 110mph, well within the schedule and I suspect at the top end of the range of daily working. A full analysis is shown below; times were clearly noted at every milepost. Note that IHP rarely exceeded 3500 and was often less. The very detailed time course provided suggests there was a brief maximum of 107 mph just before Gurnee. The two claimed miles at 110mph are quite impossible- accelerating from 103-110mph over a mile is quite beyond an F7- an artefact of the timing process used!

Engine 100

Tare/gross wt

430/460 tons

Loco wt 395 tons


Total wt 855 tons






Distance/cut off

Est speed

Log speed

Log time

Av speed

Av ihp











Washington St









Kinnickinnic Draw




































Tower A68



































































































West Lake Forest




















54.25 brake







Tower A20




































Morton Grove


















Forest glen




































Tower A5









Pacific Jn


and drift







Western Avenue


















 These and the other logs I have analysed suggest that a continuous output of 3500IHP from an F7 was more than enough to keep time with the standard load. Now there is also a report from the French Baron Vuillet that in 1943, an F7 took 16 cars from Milwaukee to Chicago in 63 minutes, averaging almost exactly 100mph from Oakwood to Edgebrook. This would require about 4200IHP sustained over the whole distance- perhaps a truer measure of what an F7 could do, though, I suspect, rarely required to.  It is claimed that an F7 averaged about 120mph with 9 cars on test, and this would also require this level of power.

Now from some source or other, trustworthiness unknown, I have a blastpipe diameter of 7” for the F7. Using this, and making an intelligent guess on F7 superheat, I reckon that 3500IHP would require about 50000lbs/hr to the cylinders, and if all this went up the blastpipe (which it didn’t), this would give a backpressure of about 10psi, probably in truth nearer 8psi. 

However, suppose what was actually required on a daily basis was more like 4500IHP. The rapidly rising backpressure means that efficiency would decrease and steam to the cylinders required would balloon to ca 72000lbs/hr, possible from a 96.5 sqft grate with Indiana coal, but seriously high blastpipe pressure. M. Chapelon could then claim he could deliver 10+% in coal savings, which would surely have got the MILW’s attention.

My underlying suspicion is this: everyone likes to quote what a steam class could do; it is reasonable looking at the data already to hand that if they operated at ‘could do’ level then improved front end design would lead to significantly more efficient locomotives; however if you look at ‘daily needed to do’ as the basis, most of the case against US front ends will disappear.

To understand this for other designs needs similar information on what they were actually required to do on a daily basis.

Now, whilst we had tens if not hundreds of folk timing trains all over our country with great precision from about 1900 on, so any train that moved was logged on some occasion, I’m not sure that detailed logs of the kind for 100 given above are common in the US. It would be great if they existed for the heavy high speed workings on the NYC, PRR, and Santa Fe, say, or any other railroad for that matter. If they don’t, just knowing typical loads, and schedules between the stops on say, the steam hauled 20th Century, Broadway and whatever the fastest steam hauled Santa Fe service was would suffice. The data would need to be for fast, straight stretches where it may be reasonably assumed there were no speed restrictions for curvature nor frequent speed restrictions for other causes, and what the speed limit, if any was. Routes that fit the bill would I think be for example Buffalo and Crestline to Chicago, Garden City to La Junta.

In all this, I am making the assumption that the second benefit of Chapelon style exhausts, namely providing more air to the fire, hence more reliable steaming was not really an issue in US practice, the need to operate reliably in harsh environments having already been given priority over other considerations. (It was in this country when people tried to draught increasingly large boilers with single blastpipes, in an environment where the distance between the top of the blastpipe and the top of the chimney was severely restricted by the loading gauge).

Thanks again for all your inputs.

  • Member since
    August 2010
  • From: Henrico, VA
  • 8,955 posts
Posted by Firelock76 on Tuesday, April 19, 2011 7:12 PM

Ok folks, I think we're getting to the point here where we can't see the forest for the trees.  All this technical data is starting to make my eyes glaze over and my head swim!  I'm not an engineer, but I am an amateur historian, so let me give you all a little history lesson about American railroads, especially to our brothers and sisters "across the pond".

As opposed to railroads on the European continent which were state run, American railroads were strictly a private enterprise.  No national treasury to back those boys up in the 19th Century when things got rollin'.  American railroad builders had to raise the capital themselves.  AND those investors were looking for a return on their investment as soon as possible.  Ever hear the old saying  "the best is the enemy of the good"?   In the United States that meant a good roadbed, a good locomotive, good rolling stock, was a lot better than someting perfect in the future.  The roads had to start making money NOW.   Improvements could always be made in the future when the money started coming in.  Certainly mistakes were made, often with tragic consequences,  but it's too easy to critisise from the comfort of an armchair when you're not the one on the spot with stockholders looking over your shoulder.

So, American locomotives had to be built tough and durable.  If some efficiency was lost, well,  that's something they were willing to live with.  Remember, if you build a perfect locomotive you need a perfect road bed to run it on, and in the old days that was usually lacking here in the U.S.  By the way, those tough American engines were also a pretty attractive puchase for those building railroads in what we'd call  "Third World"  countrys, but I'm digressing.

Now if the locomotive builders got it wrong at times, well who's to blame them.  No computer assist back in the old days.  All the boys at Baldwin, ALCO, or Lima had to go on was their own knowledge of physics, geometry, metallurgy, and prior experience, and if they had a slide rule to help them they were lucky.  Sure they got some things wrong,  but boy when they got it right!   Niagaras,  Mohawks,  Big Boys,  Berkshires,   the Norfolk and Westerns  mighty Class A's,  J's, and Y's!   Even the 4-4-0  "American"  type was a masterpiece in it's own right.   By the way, even the present day diesel builders get it wrong from time to time.

So, remember the American philosphy was "keep the tonnage moving!"   Get 'em in, get 'em serviced, get 'em out again.  If the engine can't take it you get rid of it and get another.   Keep them trains rollin'!

Mind you, I'm not attempting to denigrate any other nations locomotives.  British, German, French, all were good machines for the evironments they operated in.  All were gems in their own right.  Hey, some were so beautiful they LOOKED like pieces of jewelry!  Would they have been sucessful here in the U.S?  Well, I don't know.  I DO know that American designers were aware of what was going on  with European steam, but there wasn't much they were willing to borrow.  Superheating, the Belpaire boiler, the Walschaerts valve gear, as far as I know that was as far as they went.

So how hard was American steam run?  Well, as the song says "I'll run her 'till she leaves the rail, 'cause I'm eight hours late with the southbound mail!"

Gotta go, my eyes are glazing over again!   I guess it beats writers cramp!


  • Member since
    September 2010
  • 17 posts
Posted by NM_Coot on Tuesday, April 19, 2011 9:50 PM

Some brief comments if I may.

1.  Water treatment is far more important to good boiler operation (and maintenance) than feedwater heating.

2.  Chaplelon was far more concerned with the entire steam circuit than he was about the exhaust configuration per se.  The exhaust point is but one of many critical points in the steam circuit.  He was as concerned about the drafting interaction with combustion as he was about back-pressure.

3.  I don't quite understand the pokes at Chapelon.  He was a modest engineer who achieved remarkable results.  In fact, I will go so far as to say that if you do not know and understand Chapelon's work you do not really understand all the important elements and inter-relationships of modern Stephenson locomotive design.  His book which summarizes his fundamental work was published (in French) in 1937 and revised and expanded in 1952.  A English translation was published in 2000.  The contents are as valid today as they were in 1937.

4.  Stephenson steam locomotive design is a constant trade-off of factors.  Once you get the exhaust about right, continued tweaking isn't going to gain much in day to day service.  For example, soft exhaust often led to problems with smoke and water vapor obscuring forward vision.  Farrington, in The Santa Fe's Big Three, reprints many Santa Fe test reports.  The 3460 class developed 4350ihp at 65 mph and 3600dhp at 50 mph.  Because of drifting smoke, the ATSF test engineers gradually reduced the nozzle openings from 3 3/4" to 3 1/2" and the stack diameter from 26" to 24" without any appreciable change in performance.  The smoke problem was particularly problematic when working lightly.  There are also some interesting relationships between firebox draft, smoke box draft and oil atomizer pressure.  How about that folks!  Anyone interested in seeing how a railroad evaluated locomotives in service should read this book.

5.  As best I can research, the Milwaukee F-7 class is a development of the A class.  These two classes handled low tare weight trains at high speed over a relatively flat line.  ALCO essentially designed both.  The 84" drivers were used to keep piston speeds within reasonable limits.  ALCO did pay attention to the steam circuit.  The relatively low superheat is unusual.   the A class could do about 3000-3300ihp and I would venture that the F-7 probably could do about 3500-3700 ihp (primarily because of rather low superheat).  In the example given, a total weight of 855 tons with an engine weight of 395 tons really means that the train was there to provide braking effort.  Both classes did what they were designed to do but of the two the A class is really the big step forward in performance.

6.  NYC J-3a's regularly handled trains of 800 to 1000 tons tare and regularly achieved schedule service speeds of 80-90 mph.  In my opinion a J-3a could handle the F-7 service and I doubt the opposite is true.  The only question might be the 79" drivers on the J-3a but I think the proven lively performance of the J-3a would compensate.  The ATSF 3460 class Hudsons had 84" drivers and could handle trains of 800 to 1000 tare at 90 mph when schedule compliance was required.

7.  Ther are numerous examples of excellent US design practice.  So the point isn't making "the case against US front ends."  The point was simply that when small locomotives significantly out-perform very much larger ones, the designers first step in making the larger ones perform better is to look at what makes the small ones perform so well.


  • Member since
    August 2003
  • From: Northern VA
  • 480 posts
Posted by feltonhill on Wednesday, April 20, 2011 8:02 AM

MEGO also!!  Help me out here.  I'm as good a number cruncher as the next person but I need some assistance.

The NYC J3a is credited with a superheat of 750 degrees.  I'll bet this is total steam temperature, not superheat.  The J3a's operated at 265 psi, which would be a saturated temperature of about 410 deg F.  For a total steam temp of 750 deg F, the superheat would have to be about 340 deg, which is a very high figure.  Where did it come from?  I'm not doubting it, just want to know the source.

I've re-read most of the thread and would also like to know the source of the MILW F7's "low" superheat?  Anyone have the figure?  I've never read anything that indicated the F7 was deficient in that area.  But stuff happens in any design.  Maybe I haven't read enough.

NMCoot - I believe you're in good company with the water treatment comment.  Wasn't Porta a proponent of this idea?




  • Member since
    February 2005
  • From: Vancouver Island, BC
  • 22,546 posts
Posted by selector on Wednesday, April 20, 2011 10:48 AM

Add flue sanding, keeping any necessary and installed smokebox netting clear, petitcoat pipes mounted correctly....they were complex machines that needed skilled and experienced, not to mention attentive, operators.


  • Member since
    September 2010
  • 17 posts
Posted by NM_Coot on Wednesday, April 20, 2011 11:29 PM

Here are some numbers with sources.

Loco Profile 2, Table II, text.  Loco Profile 26, text.  The Santa Fe's Big Three, text, tables, figures.  I've compared three contemporary 4-6-4's.  I put in some numbers that may be of use to others.  That's all!  This is not a mine's bigger, better, whatever, effort.


Railroad                                                NYC                 CMStPP          ATSF

Class                                                    J-3a                  F-7                  3460

builder                                                  ALCO                ALCO              Baldwin

year                                                     1937                  1938                1937

boiler pressure (psi)                               275*                  300                  300                  

grate area (ft^2)                                      82                     96.5                 99

total evap surface (ft^2)                           4187                 4166                 4760

superheat surface (ft^2)                          1745                 1695                 2080

superheat temp (deg F)                          750                   ***                    775

superheat/total evap (areas,%)                41.8                  40.6                  43.7

total evap/grate area (ratio)                      51.1                  43.3                  48.1


wheel dia (in)                                         79                     84                     84

valve diameter (in)                                  14                     12                     13**

piston diameter (in)                                22.5                  23.5                  23.5

max travel (in)                                        8.5                    7.5                    7

exhaust nozzle dia (equiv in)                   7.25                  7                      7.5

choke/stack dia (in)                               21.5                  18.5                  24

max hourly evap rate (lb)                        93,000##           ?                      89370#

max inp (hp@mph)                                4900@75          ?                      4350@65

max dbhp (hp@mph)                             3935@65          ?                      3600@50



* Later 265 psi.  Designed for 300psi. 

** Open throttle indicator cards show drop of about 30 psi, boiler to valve chest.

# 1st District, Arizona Division, heavy grade at 45 mph, 3997 ihp, steam consumption was 68700 lb/hr.

## Test, Selkirk.

*** Class A superheat was 680-700 deg F.  F-7 no reference value.

? No reference value in my library.

I think the numbers support a conclusion that the F-7 superheat was low typical.  The author of Loco Profile 26, Brian Reed, says, with regard to the A class: "This was not a very high degree of superheat."  I think the numbers above would tend to indicate a similar condition for the F-7.  The users of this data are completely free to draw a different conclusion!


Regards and happy contemplations



  • Member since
    August 2003
  • From: Northern VA
  • 480 posts
Posted by feltonhill on Thursday, April 21, 2011 9:26 AM

NM_Coot - Thanks for the figures!  I have two of the three sources: NYC 4-6-4's (Reed) and Santa Fe's Big 3 (Farrington).  I assume that Profiles 26 is the Milw F7 and/or A?  Don't have that

Farrington's book is the real gem here.  He apparently had the clout or $$ to put what he wanted into his books.  As you already know, and others should be aware, he frequently puts in copies of actual test reports.  He doesn't paraphrase or editorialize.  He lets the railroad's test department do the talking.  It's great stuff.

On the other hand, Reed tends to include a lot of numbers so that his analyses look technically rigorous,  but  he doesn't always use consistent data in his comparisons.  Some of his other articles where I got into the details, he got a bit fast-and-loose at times.

This will keep me off the streets for a little while.  Thanks again!

  • Member since
    June 2008
  • 121 posts
Posted by Dreyfusshudson on Saturday, April 23, 2011 8:00 AM

Thanks for more top rate inputs.


With respect to Firelock 76’s (20.4.11) comments, this is an almost perfect articulation of my belief, namely that, like any successful species, the North American steam locomotive was well adapted to its environment, (operating, social, political, economic and so forth). (It is by the way true that in this country as in the US, locomotive efficiency was relatively low in the operators priority list, getting the train over the road near the top; in the heyday of steam our railways were privately operated also, fighting to stay alive from the 1920s on as road transport took off, fighting for the country’s existence during the war years, so most of the other considerations about US railroads also applied). It has been suggested by Porta that efficiency improvements to US designs might have been possible by reducing backpressure. I am looking to establish if this was indeed the case; my view is that, given the relatively low priority of efficiency, any improvement would have to be very significant for it to have had a worthwhile impact. I do think that it is likely that the answer to this question will turn on how hard locomotives (let’s restrict it to passenger locomotives) were actually worked so I will come back to the point about the southbound mail later.


With respect to NM_Coot 20.4.11, we are I think in violent agreement on all points. I know the gentleman who translated Chapelon’s book into English, and he makes sure I don’t deviate from the straight and narrow, though of course there are factors that with 20:20 hindsight, which is what high speed computing can give you, aren’t quite as important as Chapelon thought. On point 4, you speak of ‘Nozzles’ How many did the 3460 class have? I will come back to NYC performance.


With respect to feltonhill 20.4.11, yes, I’m absolutely certain that the 750F for the NYC Hudson is inlet steam temperature, not superheat level. Given that I believe that superheat is the dominant factor in determining efficiency, it would be good if one could estimate what would be delivered from published dimensions. I have a book with a most unpromising title ‘The Great Book of Trains’ written by two top rank British writers, Hollingsworth and Cook, which is remarkable in that it makes a great general interest coffee table book, (super line drawings), and, as far as I can tell also provides reliable technical details on hundreds of steam locomotives from around the world, and exceptionally knowledgeable write ups. At $10 second hand on Amazon, it’s an absolute snip! I’m telling you this because last week, I spent 2 hours extracting all the technical data on the US designs featured, to try to address this point. I was surprised how different some of the key dimensions were, particularly in quoted superheater areas. (Sadly, my computer ate the spreadsheet, so need to start over!). However, I believe that superheater area itself is not the sole determinant of superheat achieved. Test results over here showed that superheat always increased with the number of flues; however they also showed that if the distance between the tube plates was more than about 18’, superheat begins to go down, possibly because the flue gases are cooler than the steam in the last few feet of the flues, so the steam gets cooled. Obviously, the longer the flues, the higher the superheater area at a given number of flues, but this does not guarantee high superheat. We found best practice was to have not much more than 17’ between the tubeplates. The class with the largest superheater area over here did not generate the highest superheat, in part I believe because it was 19+’ between the tubeplates. From the very limited US data I have, this also seems to apply, in that the T1 (18’ between plates, 1430 sqft superheater) delivered higher inlet steam temperatures (750+F) than the Niagara, (20’ between plates, 2060 sqft superheater, which gave ca 680F at 52500lbs/hr). This is a point I would like to explore further, hoping the texts I am getting will provide details on this important dimension. But 750F for a Hudson does not seem unreasonable to me.


With respect to NM_Coot, 21.04.11, thanks, this is just the kind of data I’m looking for. With respect to the 3460 class, the 44 sqins nozzle are must mean they had four 3.75” nozzles? Depending on the steam rates normally used, going down to 40 sqins wouldn’t be too tragic (See Below).


Now for some more analyses, which I’ll try to keep simple. Hollingsworth and Cook give a quote for a Hudson hauling a 940 tonne 20th Century, which made it from Toledo to Elkhart (133 miles) in 112.5minutes net, and on to Englewood (93.9 miles) in 79.5 mins, both about 71mph start to stop. I am assuming this is an example of very good running. Since it’s basically downhill form Elkhart to Chicago, and I’m not sure if speed restrictions applied beyond Gary, I’ve ‘built’ the line from Toledo to Elkhart. What I found was that, using standard resistance equations, and assuming a relatively modest start, I could get to Elkhart in 111.5 minutes without exceeding 3000IHP. On the gentle upgrades to Wauseon, speed rose to about 71mph (32mins). In the gentle dip down to Stryker, speed rose to 81 mph (43 ¼ mins), before falling to 67 mph up to Corunna. Finally, on the descent to Elkhart speed rose to 90mph. covering the 44 miles on to Dunlap in 32 minutes, and averaging 84mph over the 38 miles from Kendalville. If a more vigorous start had been made, power required later would have been less. Now my computer programme says that the Hudson would need about 42000lbs/hr to the cylinders to deliver 3000 IHP, and even if the Hudson blastpipe diameter were only 6.5”, this would not lead to excessive back pressure. So, adding this to my analysis of the MILW F7, and extrapolating wildly from just two data points, this says that the actual working rates of these Hudsons in normal service would not lead to efficiency losses due to restricted blastpipe dimensions.  Now I’m sure an NYC Hudson could sustain much more than 3000IHP. The question is, whether there is a run on the proverbial 8 hour late southbound mail that demonstrates this.


The second set of analyses was to look at the acceleration tests with a Niagara east from Utica with 1005-1875 short tons. The line is pretty flat, perhaps a very slight downgrade to the canal bridge. Now these represent the most astonishing running I have ever come across. I only have the bare details of loads, times, and distance to reaching 75mph, but the times reported can only have been achieved by working the Niagara in full gear right from the off, with power reaching about 6000IHP at about 45mph. From then on, cut off was gradually eased to maintain just over 6000IHP until 75mph was reached. I’m almost speechless! The sight must have been unbelievable, blastpipe pressures phenomenal. Now I recall David P Morgan reflecting on how every steam fan in the early 1950s knew that beyond about 20-30 mph, a Niagara could outperform a contemporary 6000 HP ABA E unit, and this is what these tests surely demonstrate. The point Morgan was really making however was that, even though this was true, it was irrelevant. I believe also, that even though this performance was surely possible, it was rarely needed in practice, except on the 8 hour late mail. If you put a Niagara on 1200 tonnes east from Toledo, accelerate more gently than the above tests, and work at up to 6000IHP on to Elkhart, then, having self imposed a 90mph limit, I get to Elkhart in 96mins, running at 90 mph up hill and down dale from Swanton on. Maximum power is about 5700IHP, less on the downgrades. Did a Niagara ever do this in normal service? I doubt it, but would love to be proved wrong. So, I suspect even the Niagaras were actually worked at steam rates which did not involve excessive blastpipe pressures. All data to the contrary very welcome.


Heading off to Catalan Spain for 10 days from tomorrow, to a computer free zone, so will be off the case for a while.



  • Member since
    February 2005
  • From: Vancouver Island, BC
  • 22,546 posts
Posted by selector on Saturday, April 23, 2011 9:29 AM

Just a FYI, and a bit startling, but lars loco has asked that his account be deleted.  He reports that he can no longer devote the time to the forum that he was used to.  I think it safe to say that he will be missed.  I wish him well.


  • Member since
    December 2005
  • From: Cardiff, CA
  • 2,930 posts
Posted by erikem on Saturday, April 23, 2011 12:13 PM


I enjoyed reading Lars Loco's posts as well.

- Erik

  • Member since
    August 2010
  • From: Henrico, VA
  • 8,955 posts
Posted by Firelock76 on Saturday, April 23, 2011 5:54 PM

Concerning the New York Central's high speed runs to and from Chicago, there's a great story I'd like to pass on.

Both the 20th Century Limited and it's arch-rival, the Pennsy's Broad Way Limited, used to depart Chicagos Englewood stationat the same time on adjacent tracks.  This always (well almost always)  led to a race between the two, to the absolute delight of the passengers.  Mind you, OFFICIALLY the railroads said they weren't racing,  and OFFICIALLY the train crews were forbidden to race.

One day, a NYC official was aboard the Century, when he felt an abnormal acceleration.  He looked out the window and saw the Broad Way running "neck and neck" with his own train.  As the Broad Way sped up, the Century did as well.  He saw a conductor passing down the aisle and called out  "Conductor!"  "Yes sir?" replied the conductor.  "Step this way, sir."  and the man came over.  "Are we racing the Broadway?"   "AHHHH, ERRRR,  OHHHHH,"  stammered the conductor.  "I see,"  said the official.  "Get word to the engineer.  If he looses,  he's  FIRED!"

  • Member since
    June 2002
  • 17,359 posts
Posted by daveklepper on Sunday, May 1, 2011 7:53 AM

Now I want to know:   If the Century was headed by a Niagra, was the Broadway by a T-1, or two K-4's?    If it was a J3A vs single K4 contest, well the winner would be quite obvious.

  • Member since
    August 2010
  • From: Henrico, VA
  • 8,955 posts
Posted by Firelock76 on Sunday, May 1, 2011 11:55 AM

You know, Dave, I have no idea!  The original teller of the tale didn't specify what was pulling the trains or when the incident occurred.  We just have to accept is as a fun bit of railroad lore and let it go at that.

  • Member since
    June 2002
  • 17,359 posts
Posted by daveklepper on Monday, May 2, 2011 1:40 PM

Anyway, I am glad I witnessed this race.   From Moutain View or Tower View on the rear of the Broadway (never dreaming that some day I would be a small shareholder in the owner of Mountain View, Pullman Classics Limited).   As I recall we lost and the Century won.   We had head end cars added on the Broadway, while the NYC (both still all-Pullman) began with the baggage dorm and had a shorter consist.   Both trains were pulled by EMD E-units at the time.   But I had traveled to Chicago on the Century, returning via the Brodway, change to Clocker at N. Phila, MU at Trenton, Princeton Jc. and back, and then mu to NYC.

  • Member since
    August 2010
  • From: Henrico, VA
  • 8,955 posts
Posted by Firelock76 on Monday, May 2, 2011 6:23 PM

To daveklepper, that must have been a fun race to witness.  I envy you!

  • Member since
    June 2002
  • 17,359 posts
Posted by daveklepper on Tuesday, May 3, 2011 4:58 AM

We (the Broadway) got the highball a few seconds before the Century, and then it took about  ten minutes for the Century to pass us.   He started while our obs was about even with the sleeper ahead of the dining car, which if I remember correctly was the two-unit type.   Our train was all red (E-units excepted, of course) and his was all grey. All equpment matched at the time.   I think the year was 1959.

YOu could arrange for a porter to have your pants pressed on the Century at the time.  I availed myself of this service and tipped appropriately.   This was not noted on the Broadway.  But the dining car food was excellent on both trains, and the interiors and general upkeep was very good at the time.

I contiued to use rail between the east and Chicago up to 1995, and I can also recall the worst Penn Central period.   Also one trip on the Erie-Lackawanna Lake Cities  --- with no complaint.

  • Member since
    June 2008
  • 121 posts
Posted by Dreyfusshudson on Friday, May 6, 2011 10:07 AM

Always the optimist, I went to Englewood 10th May 1974 to check out the ghost of the Broadway /Century race. What I got was a picture of my conveyance there (the 17.40 to Valparaiso) with Tuscan red PRR coaches, hauled by two ex NYC E8s 4039 and 4042, which I guess might have hauled the Century?  It all seemed very sad at the time, but on reflection, as good as you could hope for. I was the only person to alight at Englewood. From there, I think I must have walked under the tracks through this dangerous part of town to take the Dan Ryan El, for my next shot is at Pullman Junction, EL RS3 1042 hauling a short freight, on the third car of which was a small steam locomotive, possibly narrow gauge. Any ideas what this was?

On my vacation, I took the trouble to read the stuff I have already got more carefully. (Note to self: must do this more often). I have the executive summary of the NYC Niagara report, which has a big clue to one of my main queries. It talks in detail about the acceleration trials I mentioned earlier, and it was pretty much as I supposed- the locomotives were worked at full capacity up to 75mph, exceeding 6000IHP at about 60 mph, higher thereafter. However, there is another table which gives the game away on my basic query, for it gives the requirements to deliver 4000IHP required for ‘above average passenger train operation’.  In other words, Niagaras could deliver 6000+IHP, but in practice were not required to deliver more than 4000 IHP, evaporating no more than about 650lbs/sqft/hr. This is remarkable concurrence with operating practice in this country- where it was of course possible, but rarely if ever required to steam above 650-750lbs/sqft/hr (this with 12500-13500Bthu/lb coal, as on the NYC)- economics of boiler maintenance, not to mention fuel economy probably has a role here.

To illustrate what 4000IHP can do, the Table below shows some simulations of running from Toledo to Elkhart (all weights long tons). The run of streamlined Hudson 5450 is a simulation of an actual run that got to Elkhart in 112.5 mins net: this simulation uses no more than 3000IHP at any point. The first run of Niagara 6023 has the same load, but working at no more than 4000IHP; it gets to Elkhart 10 minutes faster. Alternatively, the last simulation shows that 6023 could get to Elkhart in the same time as 5450 at 4000IHP with an additional 300 tons (say four coaches).  (All these are ‘rough cuts’; there are some aspects of the resistance equations I’m using that I’m not happy with, to be fixed shortly).





Tare wt




gross wt




Engine wt




total wt





est speed

est time

est speed

est time

inst speed

est time


































































































































































Looking at these outcomes, 70-80 mph start to stop averages with 14-17 cars, I would be surprised if more than this was achieved on a daily basis from a Niagara, or indeed any US 4-8-4, except perhaps when climbing steep grades.  In this case, with exhaust rates not much above 50000lbs/hr (allowing for feedwater heater recycle), then with blastpipe diameters of 7.5+”, there really is no ground for Porta’s implied critique that more power/less fuel could have been achieved in the US with better designed front ends. Unfortunately the British interlibrary loans service is in chaos, so the books on the ATSF and N&W 4-8-4s that might provide counter evidence still haven’t arrived.

Thinking more generally about US designs, there was a practical limitation of about 30 long tons driver axle loading, say 210000lbs on 3 axles, 280000lbs on 4. Now there was also a desire to keep the adhesion factor (ratio of axle load: tractive effort) not less than 4, so TEs of around 50000lbs for 3 axles, 70000lbs for four axles. Now if you say that at speed, you don’t want cut off to exceed 30%, for efficiency's sake, then the question is, how much HP do engines with 50000 and 70000lbs TE develop at 30% cut off at ca 80mph? Now obviously this depends a bit on some design factors, but at reasonably good superheat, the answers are ca 3900IHP and 4200IHP. (Note that IHP does not increase in proportion to TE at speed). So at 4000IHP a 4-8-4 would be working in an economical regime. At the risk of getting caught in partisan cross fire, I would also observe that going to four cylinders, as per the T1 does allow slightly shorter cut off working to get the same steam flow and power, but that the difference in efficiency due to a few percent in cut off at 30% or less is pretty small.

  • Member since
    August 2010
  • From: Henrico, VA
  • 8,955 posts
Posted by Firelock76 on Friday, May 6, 2011 5:46 PM

Welcome back!  How was Spain?

  • Member since
    June 2008
  • 121 posts
Posted by Dreyfusshudson on Tuesday, May 10, 2011 5:02 AM

Thanks, Spain was great. The only thing that distresses me is that about 6 miles from where we live, we have a brand new TGV station, two island platforms long enough for double length TGVs, two arrow straight 200 mph centre road fast tracks separating them. At present, it is the terminus for two daily round trips to Paris with a single TGV, plus two connecting services to Barcelona. In addition there are four freights per week. The whole project to Barcelona is about 4-5 years behind schedule. I think I'm paying for this somehow, probably most other folk too.

Kratville's book on the UP had arrived. In the absence of solid data yet from Farrington, I checked out the run of the Super Chief on 31st April 1953, when a 2900 had to take over  a consist swollen with business cars to 850 long tons due to diesel failure at Kansas City, working a 39 3/4 hr schedule. On the long uphill 2500' climb from Newton KS to La Junta CO, the 2900 covered the 355 miles in 4'50", just inside schedule, including a 10 minute servicing break at Dodge City, working around 4000IHP, and cruising pretty consistently at 80mph. Maximum was about 85mph in the slight dip before a 30 mph service slack through Hutchison.

  • Member since
    March 2010
  • From: Fort Hood, TX
  • 28 posts
Posted by Burgard540 on Saturday, May 14, 2011 12:22 PM


I just saw this thread (haven't been on here for a while, in Iraq) and the program you're using sounds interesting.  Can you share it?  I've been doing research on the UP 4-8-8-4 Big Boys, and it would be interesting to see the results.  The Big Boys, Challengers, and FEF's shared many design characteristics and by extension many of the blueprints are shared as well.  Kratvilles books regarding each class are good, but often reference each other.  I bit the bullet and bought all three.  I'd be happy to send you the tables of data and analysis I've done based on the data from Kratville's Big Boy book that I have made in Excel. 

Each class of the locomotives utilized a double stack design (except the first FEF class, which were later modified to double stacks), with two multiple jet exhaust plates that had 4 nozzles.  It seams that UP made similar conclusions regarding nozzle design as Porta's Lempor drafting design.  The inside bore diameter of the tips were (Reference UP Dwg 443 CA 33736, 16 JAN 1951):

FEF-1: 2-13/16" (Total Area 49.70 sq in); FEF-2-3: 3" (Total Area 61.36 sq in)

4664-3-5, Challengers: 3" For oil (Total Area 56.55 sq in), 3-1/8" For coal, (Total Area 56.55 sq in)

4884-1-2, Big Boys: 3" (Total Exhaust Area 56.55 sq in)

The April 1943 tests showed that from Ogden to Echo the per hour running time the average water evaporated by the boiler was 83,200 lbs/hr, condensate return (via exhaust injector) 6830 lbs/hr, coal fired 19,100 lbs/hr, developed 5300 drawbar horsepower, speed  29 mph, and drawbar pull of 68,800 lbs.  Only 4.36 lbs of water were evaporated per lb of coal, and the coal was fired at a rate of 127 lbs per sq ft of grate per hour. 

From Echo to Evanston the per hour running time the average water evaporated by the boiler was 80,600 lbs/hr, condensate return (via exhaust injector) 6450 lbs/hr, coal fired 17,900 lbs/hr, developed 4560 drawbar horsepower, speed of 16 mph, and drawbar pull of 103,500 lbs.  Only 4.52 lbs of water were evaporated per lb of coal, and the coal was fired at a rate of 120 lbs per sq ft of grate per hour. 

Average boiler pressure of 294 psig, 96% full throttle, & reverser 26 notches ahead of center (what that means in terms of cut-off I don't know). This was pulling a 71 car, 3883 ton train.   Regarding total steam temperature, Kratville mentions on pages 13 & 18 of "The Mighty 800" that the Type E superheater would supply steam at 750 degrees or better to the cylinders igniting leaking lubricants! 

I look forward to the discussion,


"If a nation expects to be ignorant and free, it expects what never was and never will be." Thomas Jefferson

Join our Community!

Our community is FREE to join. To participate you must either login or register for an account.

Search the Community

Newsletter Sign-Up

By signing up you may also receive occasional reader surveys and special offers from Trains magazine.Please view our privacy policy