Just How Large can Steam Locomotives Scale up?

[L-105]

I’m not sure were to ask this question, I’ve searched high and low but can’t seem to find anything that properly answers my point. Hopefully there may be someone here who can give a bit of insight, as having looked through previous threads, it looks like there are some users here with fairly in-depth knowledge about steam locomotives.

 

It appears that the design of powerful steam locomotives particularly benefited from having a larger loading gauge in which to place larger boilers and fire-grates for more steam producing capacity, especially since the maximum pressure that can be practically obtained with a fire tube boiler is ~350 psi (Ralph Johnson, The Steam Locomotive). Specifically, I am interested in just how large steam locomotives could potentially have been made, not being limited by a given track gauge, loading gauge, weight of rail / axle load, tunnel / bridge clearances etc. I am aware that locomotives like the UP Big Boy, C&O H-8 and NP Z-5 were just about the limit of what could be crammed into the allowed loading gauge, but I’m talking beyond standard gauge and its associated limitations.

 

For example I estimate that a 5.5ft track gauge, with a loading gauge allowance of roughly 19.25 ft height by 12.75ft width would have been sufficient to fit a boiler and firebox large enough to competently provide steam for a simple triplex articulated engine. A hypothetical 4-6-6-6-6, with 3 sets of 6 drivers all under the boiler, rather than have the rear set under the tender like the Erie and Virginian triplexes. Grate area would be 216.7 sq ft (with grate dimensions of ~ 130 inches wide by 240 inches long), firebox direct heating surface of ~1300 sq ft, maximum boiler diameter of 135 inches with 26 ft long tubes and flues, providing ~11200 sq ft of indirect heating surface making a combined evaporative heating surface of ~12500 sq ft. Superheating surface would be ~5750 sq ft. Boiler pressure of 275-300 psi, ~70 inch drivers and 6 cylinders measuring ~24.5 x 32 inches. Track would use ~ 190 lbs per yard rail. This is all very rough, but gives you an idea of what I’m talking about.

 

There is a couple of issues with this that I can think of. Firstly, the length of the tubes and flues. I’ve seen from Ralph Johnson’s The Steam Locomotive, that it is stated that ~21-22ft long tubes are the optimum length due to diminishing heating returns. However in Ralph Johnson's The Steam Locomotive, the “heat taken up by boiler tube length” data (p.89) seems to be obtained from a USRA Light Pacific, which has a 66.7 sq ft grate. I’m not sure how this optimum tube length would apply to boilers with substantially smaller or larger grates. I’d imagine that a boiler with only 10 sq ft grate would have shorter optimum tube length, whilst a boiler with a 200 sq ft grate would have a longer optimum length for example. Then there is the structural issue, as apparently longer tubes have a tendency to flex. I don’t know if utilizing some sort of internal support in the boiler could solve this, but you wouldn’t want it to impede water circulation.

 

Secondly, I don’t know just how large grates can get. How far can a mechanical stoker spread coal? Would using multiple stokers help ? Would oil burners work better for very large grates? I know of some ship boilers with large grates such as the Yarrow water tube boiler which had 493.5 sq ft grate, although I think these had access from both sides.

 

The only steam locomotive designs that I know of that exceeded anything built on standard gauge were some of the 3000mm gauge German Breitspurbahn proposals. The Breitspurbahn would have utilized 191 lb per yard rail and locomotives were typically 6.85m (22.5 ft) high and 6m (19.7ft) wide. One of the designs has a 24 wheel tender bracketed by two 8-8-8 engines with 3000mm (118.1 inch) drivers. Each engine has what I estimate to be a 250-300 sq ft grate and a 4000mm (157.5 inch) maximum boiler diameter. The length of the flues and tubes looks to be ~ 10000mm (32.8 ft).

 

[BaltACD]

L-105

I’m not sure were to ask this question, I’ve searched high and low but can’t seem to find anything that properly answers my point. Hopefully there may be someone here who can give a bit of insight, as having looked through previous threads, it looks like there are some users here with fairly in-depth knowledge about steam locomotives.

 

It appears that the design of powerful steam locomotives particularly benefited from having a larger loading gauge in which to place larger boilers and fire-grates for more steam producing capacity, especially since the maximum pressure that can be practically obtained with a fire tube boiler is ~350 psi (Ralph Johnson, The Steam Locomotive). Specifically, I am interested in just how large steam locomotives could potentially have been made, not being limited by a given track gauge, loading gauge, weight of rail / axle load, tunnel / bridge clearances etc. I am aware that locomotives like the UP Big Boy, C&O H-8 and NP Z-5 were just about the limit of what could be crammed into the allowed loading gauge, but I’m talking beyond standard gauge and its associated limitations. 

For example I estimate that a 5.5ft track gauge, with a loading gauge allowance of roughly 19.25 ft height by 12.75ft width would have been sufficient to fit a boiler and firebox large enough to competently provide steam for a simple triplex articulated engine. A hypothetical 4-6-6-6-6, with 3 sets of 6 drivers all under the boiler, rather than have the rear set under the tender like the Erie and Virginian triplexes. Grate area would be 216.7 sq ft (with grate dimensions of ~ 130 inches wide by 240 inches long), firebox direct heating surface of ~1300 sq ft, maximum boiler diameter of 135 inches with 26 ft long tubes and flues, providing ~11200 sq ft of indirect heating surface making a combined evaporative heating surface of ~12500 sq ft. Superheating surface would be ~5750 sq ft. Boiler pressure of 275-300 psi, ~70 inch drivers and 6 cylinders measuring ~24.5 x 32 inches. Track would use ~ 190 lbs per yard rail. This is all very rough, but gives you an idea of what I’m talking about. 

There is a couple of issues with this that I can think of. Firstly, the length of the tubes and flues. I’ve seen from Ralph Johnson’s The Steam Locomotive, that it is stated that ~21-22ft long tubes are the optimum length due to diminishing heating returns. However in Ralph Johnson's The Steam Locomotive, the “heat taken up by boiler tube length” data (p.89) seems to be obtained from a USRA Light Pacific, which has a 66.7 sq ft grate. I’m not sure how this optimum tube length would apply to boilers with substantially smaller or larger grates. I’d imagine that a boiler with only 10 sq ft grate would have shorter optimum tube length, whilst a boiler with a 200 sq ft grate would have a longer optimum length for example. Then there is the structural issue, as apparently longer tubes have a tendency to flex. I don’t know if utilizing some sort of internal support in the boiler could solve this, but you wouldn’t want it to impede water circulation.

 

Secondly, I don’t know just how large grates can get. How far can a mechanical stoker spread coal? Would using multiple stokers help ? Would oil burners work better for very large grates? I know of some ship boilers with large grates such as the Yarrow water tube boiler which had 493.5 sq ft grate, although I think these had access from both sides. 

The only steam locomotive designs that I know of that exceeded anything built on standard gauge were some of the 3000mm gauge German Breitspurbahn proposals. The Breitspurbahn would have utilized 191 lb per yard rail and locomotives were typically 6.85m (22.5 ft) high and 6m (19.7ft) wide. One of the designs has a 24 wheel tender bracketed by two 8-8-8 engines with 3000mm (118.1 inch) drivers. Each engine has what I estimate to be a 250-300 sq ft grate and a 4000mm (157.5 inch) maximum boiler diameter. The length of the flues and tubes looks to be ~ 10000mm (32.8 ft).

All railroads are constrained by the gauge, clearance plate and load limts that the track structure is built to withstand.

If you want LARGE boilers and steam engines - go to sea.  Marine steam engines were much larger than railroads - just as marine diesels are several times larger than railroad diesels.  By the time railroads got the the 'Super Power' steam engines, the marine use of steam had moved onto steam turbines.

[54light15]

Balt- when was Super Power first used in locomotives? 

[selector]

The general consensus is that it commenced with the Berkshire class 2-8-4.

[Overmod]

L-105

It appears that the design of powerful steam locomotives particularly benefited from having a larger loading gauge in which to place larger boilers and fire-grates for more steam producing capacity, especially since the maximum pressure that can be practically obtained with a fire tube boiler is ~350 psi (Ralph Johnson, The Steam Locomotive).

The limit is not so much in the 'firetube' portion of the boiler as in the conventional staybolted-plate firebox construction (Johnson in part was referencing Baldwin 60000, which was able to use a nominally higher practical pressure for compound operation, with a watertube firebox, but was limited somewhat by practical considerations of riveted construction and mid-Twenties practical metallurgy.)  "Locomotive" boilers in the oilfields routinely ran at 500psi -- admittedly these were in stationary applications with reasonably continuous load and no issues with water supply, but here again a "better" firebox design (say, welded Jacobs-Shupert or waterwall) would allow somewhat higher convection-section pressure with proper material thickness and detail design -- the critical issue now being staying of the tube/fluesheets if the "boiler" proper is not fully populated with structural tubes.  (Most conventional locomotive boilers are not, of which more a bit later)

In practice a large modern locomotive would probably run either with careful Cunningham jet circulation in its firebox water space or explicit forced-circulation Lamont-style waterwall circuits feeding good cyclone/gravity steam separators.  These have the watertube-boiler's inherent resistance to overheating or bursting, but don't have the prompt consequences of tube failure that make the various Schmidt high-pressure systems so impractical on crewed locomotives.

... I am interested in just how large steam locomotives could potentially have been made, not being limited by a given track gauge, loading gauge, weight of rail / axle load, tunnel / bridge clearances etc.

There are several potential classes of answer to this question.  The immediate one has less to do with size than with economy: larger locomotives imply much larger unit horsepower, and even with reasonable Rankine-cycle heat recovery and recompression the limiting factor quickly becomes effective water rate.  In the 'old days' you could scoop stagnant water with pollen and leaves and soot in it and get away with boiler structure that only lasted a couple of years before full replacement.  In a modern world of deoxygenation and antifoams (see Porta Treatment or the earlier McMahon-Porta discussions) scooping becomes less and less practical, we pass into a world of almost NASCAR-like fast pumping transfer ... and a range between required water stops on even the best Eastern standard gauge that actively approached 120 miles on the latter classes of large PRR power (both duplex-reciprocating and mechanical turbine, about the most water-rate-efficient of steam designs).

Meanwhile the vertical separation of the boiler water (which has a kind of indeterminate level, like milk in a saucepan boiling as it is scalded) from the intake to the dry pipe is critical for higher power density (which of course is a product of mass flow).  You may remember Woodard experimenting with the outside dry pipe: this was an attempt to get the vertical separation nearly as much as the overall vertical loading gage would allow.  Russian engines, with ridiculous overhead clearances, often provide outsize tall domes and arrangements to provide great vertical 'steam space' net of separation, and one place 'double-stack clearances' become very useful in modern steam design is that you can, finally, run the boiler completely populated with flues and tubes. get rid of palmstays and eccentric loadings, and still have excellent passive separation at high mass flow even with a certain amount of foaming/priming due to water quality or treatment issues.

A related issue is maximum beneficial contact-patch weight transfer (for adhesion and running load).  There are limits on effective contact-patch size across all drivers that can start to become a kind of 'does Texas have the largest midgets' issue -- probably long before you get to 3m Breitspurbahn gauge.  Actual rail section weight has to be scheduled net of 'minimized' contact patch, too.  At some point you're going to go to intermediate bearing rails or even a full 'double-track train' structure at which point, again, the discussion starts to go places not of much relevance here.

The 'largest' reciprocating steam locomotive practical for a given loading gage will almost certainly be a Garratt, and arguably either a simple-articulated Garratt (of the approximate size of a pair of Alleghenies for fast service, or 2-8-8-4s for heavier) or Mallet-Garratt with IP injection (to keep the effective running water rate minimized).  This has no effective limit on boiler diameter up to the loading gage as there need be no accommodation for drivers and suspension under any part of the boiler, the water sidewalls in the radiant section can be as long as practical, and the ashpan arrangements on solid fuel can not only be easily taken down close to railhead height but also built to periodically empty and 'ashavey' to reasonably controlled storage.  Naturally you would use proportional air throttles for control and probably for trim between engines, so actual operation would be no more difficult than any modern steam locomotive.

There are limits, of course, on curve radius clearance (now on the 'inside' along the center of the boiler, instead of on the overhanging end(s) as on a typical large Mallet chassis) and this may mandate opening out center-to-center distance on curves.

To lessen that and other problems, you can get the same 'basic' effect with a mechanical turbine, where the underframes carrying the boiler are lower and can hence be carried close to a central deep firebox (as for example in the PRR V1 design and its early offshoots at N&W.  This can be built bidirectional with the general geometry of a Meyer locomotive, with active mass transfer to the bunker and cistern capacity on the chassis itself.  For best 'overhang' you could handle your weight distribution and geometry to put the chassis pivots on the effective 'quarter points' of the main frame of the locomotive, and articulate the chassis frames toward the ends if you want to make them longer.

The original V1 was designed with two turbines on separate chassis using what was essentially a Q2 boiler with slightly higher efficiency, for 8000hp (water rate about 132 miles or so with the largest PRR coast-to-coast 16-wheel tender) and PRR publicists subsequently cheated that number higher without carefully considering what it meant to do so, first to 9000hp and then to unspecified moar.  

That water rate was the explicit reason the original V1 was 'not proceeded with', and while there was something of a conceptual revival with a slightly different transmission arrangement after 1947, you really didn't see alternatives to E and F units much after then.  There are credible technical answers to feeding additional fuel and water -- but most of them only make sense in a limited number of applications, like the N&W running 15,000-ton trains from Williamson over the Ohio River under very strict and careful supervision.  Otherwise you are better with more modular power, and that's a different discussion with different criteria from this one.

... I estimate that a 5.5ft track gauge, with a loading gauge allowance of roughly 19.25 ft height by 12.75ft width would have been sufficient to fit a boiler and firebox large enough to competently provide steam for a simple triplex articulated engine.

Except that a 'triplex' isn't the right answer, either 'all under the boiler' or using a motor tender (which has never, ever worked quite right in actual practice).  

The appealing 'alternate history' here would be if the Erie and other railroads had continued the 6' gauge through the development of heavier track and engines, as that's a workable picture for comparable loading-gage evolution (and of course high-wide vertical clearances if you want them).  Again you'd be better with a relatively shorter and fatter convection shell, now with more possibility of vertical steam separation, running the convection shell entirely full of stressed tubes like a HRSG, and using better folded-return gas path and additional economizing and exhaust-steam heat recovery.  You're limited here more by the (considerable!) mass of the feedwater than by its packaging in cisterns or auxiliary tanks.  

One very attractive thing that 'happens' on 6' is that you can practically use roller bearings combined with accessible inside cranks, so you can go either to 3-cylinder or 4-cylinder drive and get rid of most of the overbalance-related augment concerns -- with zero overbalance many of the issues with higher rotational speed of reciprocating locomotives become much less significant.  Technically this lets you optimize bore-stroke and wheel diameter in ways not possible on narrower gauge; you can also use larger-diameter journals and tunnel cranks for the larger bearings that go with heavier axle load.

Where 'auxiliary' engines actually went, less than a decade after the Henderson multiplexes, was of course the geared-engine solution implicit in the Franklin booster, Bethlehem auxiliary engine, and Sentinel engines.  There have been advances in this technology since the 1920s, perhaps the 'best' being some variant of a Lewty booster (a Canadian innovation like Cunningham circulation!) where an efficient multiple-expansion positive-displacement engine drives axles separately without the packaging and gearing problems in typical steam-only reversible boosters.  To the extent you might want to motor carrying or 'tender' axles, doing it asynchronously with relatively low (and standardizable) wheels is infinitely better than replicating the faults of a large-drivered 2-cylinder simple under grossly variable weight. 

Grate area would be 216.7 sq ft (with grate dimensions of ~ 130 inches wide by 240 inches long), firebox direct heating surface of ~1300 sq ft.

I trust you are not achieving these numbers just by 'scaling up' the dimensions of something like an Allegheny boiler.  You might, in fact, not require additional 'grate area' at all -- and might want to increase radiant surface only to allow better thermal-barrier coatings to give longer luminous-flame path before quench -- instead dropping the grate as low as possible, using primary and secondary air heating, and employing better control over exhaust-steam and combustion-gas heat exchange.  As you get to larger nominal firebox and chamber size a number of things begin to scale poorly for a 'Stephenson' style firebox: you still need to provide a reasonably large mass of luminous gas to equalize radiant uptake to keep thermal distortions minimized, and you need to adjust effective TOF to keep the uptake profile 'as expected' along the full length of the plume to the rear tubesheet.  This implies that some methods of sliding-pressure firing for part-load economy will be less than efficient, compared to (say) the use of a multicell firebox structure with parallel tables and jetting/flinging for mechanical stoking, or multiple burners and primary/secondary air provision.  At a certain point it begins to be technically feasible to use a small cyclone furnace instead of fixed or traveling grates, at which point the radiant uptake begins to take on powerplant-like proportions (remember that it scales as the fourth power of luminous temperature, but has trouble in quench) ... and your locomotive stops being a simple upscaled reciprocating machine and starts looking like a powerhouse on wheels.

... maximum boiler diameter of 135 inches with 26 ft long tubes and flues...

Doesn't work that way.  Efficient aspect ratio is independent of size (and, really, of practical gas speed) and the range is fixed at about 20-22'.  More actually loses you efficiency, even before you add the additional structural weight.  And this is net of using a Chapelon-style 'leaky plate' front-end feedwater-tempering chamber at the very front of the convection section (Joe Burgard calculated the dimension for the T1 boiler and it's about 3' and a fraction for maximum gain).

What you'll do instead is use more of the available space for heat recovery and more advanced feedwater-heat train management, probably including a non-corroding Franco-Crosti style combustion-gas economization down to reaonably low gastemp.  

Superheating surface would be ~5750 sq ft.

You can't size it arbitrarily like that.  'Superheat' beyond that that eliminates wall and a certain degree of nucleate condensation on 'economical' long expansion contributes little, and starts to have very bad effect on your tribology (rings rods, and glands) as well as contributing to differential heating issues.  To the extent a large engine requires effective superheating turndown, you might want to go directly to supplemental gas-side superheat or even separately-fired superheat as in marine applications.

Boiler pressure of 275-300 psi, ~70 inch drivers and 6 cylinders measuring ~24.5 x 32 inches.

With welded boiler construction and a waterwall box many of the historical constraints on 300psi are not there.  On the other hand the weight of the boiler and especially any internal circulators in the firebox space increase at the higher pressure, and of course initial fireup and recovery times are more protracted to reach the higher pressures.  Remember that according to Brashear the original ATSF 2-10-4 plan (from about 1923) involved nominal 325psi steam, and Baldwin's calculations indicated that neither the weight distribution nor the axle load would have been tenable.  Much of this also factors into ...

Track would use ~ 190 lbs per yard rail.

Rail profile might be different, but I suspect it would not 'scale' directly from 132 or 135# sections up to what a Breitspurbahn would use.  PRR 155# was probably a steel-wasting mistake in many respects.  With the lower augment and some care with vertical and lateral spiraling and elevation, you might not need much enhancement of the rails for the larger imposed loads.  Bridges and tie arrangements might be better places to look at constraints on the MOW side.  I'll be interested to see what MC contributes to this ... given the fundamental size and design constraints for the power so far.

...  the length of the tubes and flues. I’ve seen from Ralph Johnson’s The Steam Locomotive, that it is stated that ~21-22ft long tubes are the optimum length due to diminishing heating returns. However in Ralph Johnson's The Steam Locomotive, the “heat taken up by boiler tube length” data (p.89) seems to be obtained from a USRA Light Pacific, which has a 66.7 sq ft grate. I’m not sure how this optimum tube length would apply to boilers with substantially smaller or larger grates.

I can see how this might be confusing; there is a discussion of convection design and scaling in Ell's book, but it is largely masked by the large empirical constants he uses which, as you note, are relative to early-20s dimensions at best.  Most of the theorists in the '30s and '40s appear to have converged at right around 1:400 (I think the ACE3000 numbers from theory came out at something like 1/408 and a decimal) right across the board.  This is net of a certain amount of implied black magic from the Superheater Company about effective areal drag due to efficient vs. cheap elements in flues balancing flow in different regions of tubes -- part of the fun in the Master Mechanic front-end arrangement being how to balance flow from the firebox plume across tubes a long way off the effective plume axis.

I’d imagine that a boiler with only 10 sq ft grate would have shorter optimum tube length, whilst a boiler with a 200 sq ft grate would have a longer optimum length for example.

Lots more to it than that.  (And you have to iterate/recurse when you change something, to see how all the other operational effects change and start to affect each other.)

Smaller grate implies lower combustion-gas mass flow, but the kinetics and physics of fuel combustion don't scale with size.  So radiant plume length is more a function of gas speed (which is induced by the front end net of flow resistance in the tube arrangement) and of course with that fixed 'just' so the glow drops down near the rear tubesheet contact, flow speed in the various tubes and populated flues becomes relatively determinate... as will the lolog heat transfer down the tube length for the corresponding 'time of flight'.  

Note that any consideration of 'automatic action' (one of the important 'assumed' priorities of practical locomotive design) or of excessive tendency to superheat at high mass flow have not been discussed.  They remain critically important.

Then there is the structural issue, as apparently longer tubes have a tendency to flex.

 Difference of 2-4' over a span of 20+' is vanishingly important to begin with.  Do not forget that the tubes are effectively in tension -- they are the primary reason the tubesheets at front (where the prossering is done) and at the rear (where the structural full-pen external fillet welds are) don't deform at full pressure.  That plus the hydrodynamic drag on the tube cross-section presuming it starts to deflect make it unlikely to me that actual damage from this cause would be significant, even with lower-pressure tube wall thickness.  It would not be that difficult to put a light structure 'orthogonal' to effective circulation pattern in the convection section ... assuming you've mapped what that pattern actually does (which you would for the Cunningham pump locations) if it proves for some probably unaccountable reason that the tubes are in fact contacting and fretting at some points in service.  The various design considerations I leave as an exercise for the alert (and not MEGO'ed) reader.

... you wouldn’t want to impede water circulation.

I doubt that circulation patterns in the water mass need be adversely affected.  At one point in the '20s some people actually advocated a baffle ACROSS the tube nest as a shield against the 'wrong' sort of circulation ... even more astoundingly it didn't seem to have that much effect either pro or con.  Aside from ensuring that the greatest mass of 'downcoming' circulation feeds into your firebox-circulator inlets (this is FAR more important circulation than convection around the tubes!) you're more concerned with assuring a relative absence of spot overheating and perhaps Eisenhoffer/Leidenfrost DNB on some places in the tubes..

How far can a mechanical stoker spread coal?

Some designs have substantial 'throw' and can be adjusted for reasonable projection accuracy (this is a reason Porta liked that wacky Elvin design).  The key, though, is that both 'economy' and 'performance' are historical factors of great importance, and a proper stoker design has to be sensitive and economical enough to generate the demanded gas mass without requiring a PhD and constant foreground attention from a 'fireman' or equivalent.  

Multiple stokers would help, just as multiple burners in pulverized-coal firing are needed for larger boilers.  Here, assuring the correct gas mass for the evolving combustion plume is the thing you start with, and having the right shape and characteristics for the plume is the next thing to watch ... but if the arrangement costs more than the 'fuel savings' or needs tetchy adjustment or even a few times breaks down on the road -- there goes your big savings!

Would oil burners work better for very large grates?

I am not being snarky in saying this, but most modern oil burners don't have grates.  (There are dual-fuel designs that do, and some early 'burner' designs sprayed oil over the equivalent of hot lava rock for full effective carburetion and preheat) but most large oil-burning power involves generation of a combustion plume that *just* fills the combustion space in firebox and chamber during peak firing -- very similar to what a subbituminous-fired Big Boy does with the large proportion of levitated fines with which it is fired.   Here too one school of development uses multiple smaller burners, sometimes directional, to handle the turndown necessary in practical railroad firing, and facilitate lightoff/flameholding.

I know of some ship boilers with large grates such as the Yarrow water tube boiler which had 493.5 sq ft grate, although I think these had access from both sides.

It may be no coincidence that the most advanced locomotive built, in relative great secrecy, in Britain used a more or less purpose-built Yarrow, and was not enough of a success even to keep the watertube arrangement.  A boiler of this type would probably benefit from one or more (I'd use 3) differentially-drivable chain grates, as in the B&W boiler of ridiculously inadequate size used on the N&W TE-1, probably with some active control over the secondary air to adjust some of the volatilization and burnup rate.  This would be supplemented by some kind of jet or flinger stoker for more specific edge-to-edge control ... probably under at least semiautomatic observation and control.

The only steam locomotive designs that I know of that exceeded anything built on standard gauge were some of the 3000mm gauge German Breitspurbahn proposals. The Breitspurbahn would have utilized 191 lb per yard rail and locomotives were typically 6.85m (22.5 ft) high and 6m (19.7ft) wide. One of the designs has a 24 wheel tender bracketed by two 8-8-8 engines with 3000mm (118.1 inch) drivers. Each engine has what I estimate to be a 250-300 sq ft grate and a 4000mm (157.5 inch) maximum boiler diameter. The length of the flues and tubes looks to be ~ 10000mm (32.8 ft).

Keep in mind that in some ways this was more of a prestige railway than a carefully-engineered solution, and it appears to me to have had too many architects and shipwrights in the mix.  The very obvious thing this would have gotten would be a German naval-style forced-circulation boiler arrangement, not anything familiar to the contemporary German locomotive-builders (who appear to have been stymied even in making pulverized-coal firing work on the fastest express locomotives in Europe).  A look at the putative MAN-powered diesels may be a bit more gainful although no less grandiose; one would also have no difficulty implementing an Essl-style modular design (up to a huge number of wheels!) now presumably with a slave economy and outsized national pride and arrogance behind development rather than Depression-era railroad purchasing beancounters.

[Overmod]

54light15

Balt- when was Super Power first used in locomotives?

Not 'all' of it appeared precisely at the same time.  Some aspects date back to Mikado 8000; some others didn't really appear until the N&W A (and perhaps arguably the Seaboard R engines) improved on the Baldwin 2-6-6-2 idea for an explicit high-speed simple articulated (which itself was early-Thirties).  And a couple of real whoppers in the original Super-Power needed 'extensive rework' or replacement before the real promise of, say, 2-8-4s in high-speed service could be achieved.

[54light15]

From Wikipedia, Super Power in locomotives first appeared in the 1920s. The first turbine powered ship was Charles Parsons "Turbinia" which was launched in 1894. It still exists in The Discovery Museum in Newcastle, U.K. 

[L-105]

Thanks for that Overmod, absolutely fantastic insight. Incredibly helpful.

[blue streak 1]

Does long boiler tubes become a limiting factor for track grades.  Can imagine one of these monsters going down hill uncovering boiler sheet ?  How did the SP cab forwards handle the water level  problem going up hill calling for max tractive effort ?

[Overmod]

blue streak 1

Does long boiler tubes become a limiting factor for track grades.

The 'tubes and flues' are not really the limiting factor as there is an absolute limit on their length no matter how large the diameter of the shell.  Just as the firebox only occupies the 'lower' portion of the outer wrapper, the tubes and flues only occupy the lower part of the cylindrical shell -- that is part of why the superheater-element drop from header to tube entry is so pronounced.

Part of the reason for the 'taper' in a wagontop boiler is to accommodate the "tilted" water level in a boiler going up- and downhill.  Some railroads went so far as to put a fixed plate on the backhead indicating the 'correct' water level to maintain on steep (~2%) grade; careful balancing of the feedwater-heater pump and the injector would be used to keep the water 'safe' but no higher ... careful balancing.  

[/quote]Can imagine one of these monsters going down hill uncovering boiler sheet?[/quote]That was often where the problem emerged for the careless -- the engine would come up to the top of the grade, under high draft with the fire piled up ... and over the top the water runs forward before the fireman can get enough pumped/injected in.  With oil firing you can cut back with only thermal cycling a concern, but with coal you need the experience to get the heat you need but not more heat than you can use downhill... 

How did the SP cab forwards handle the water level  problem going up hill calling for max tractive effort ?

This is one of the 'problems' with the turning-the-boiler-around  designs: the raked crown that keeps the firebox water level 'higher' going uphill now works at cross purpose.  I don't have detailed firing instructions for those locomotives but I suspect they are available, either through SPHTS or a place like CSRM.  What I suspect is that water was kept higher than 'optimal' for resistance to carryover, and that along with the tuning of firing would come very careful incremental water management.

[L-105]

Overmod

Doesn't work that way. Efficient aspect ratio is independent of size (and, really, of practical gas speed) and the range is fixed at about 20-22'. More actually loses you efficiency, even before you add the additional structural weight.

I suspected this as much, but wasn't fully sure until your response. Whats interesting about this is that as a steam locomotive increases in size, it would increasingly deviate from the "traditional" shape that most people would associate with a tender engine, in which the the combined length of the firebox, combustion chamber, barrel and smokebox arrangement more or less matches the wheelbase of the engine. As sizes increase, I'd imagine the engine wheelbase to increase out of proportion to the total length of the whole boiler arrangement, leading to more and more "free space" on top the locomotive frames not taken up by boiler.

Overmod

At some point you're going to go to intermediate bearing rails or even a full 'double-track train' structure at which point, again, the discussion starts to go places not of much relevance here.

Perhaps such intermediate bearing rails would be a pair of rails placed just inside the usual outer rails, with a set of wheels directly behind the outside wheels, somewhat similar to a dual rear wheel arrangement on a truck.

 

[Overmod]

L-105

Whats interesting about this is that as a steam locomotive increases in size, it would increasingly deviate from the "traditional" shape that most people would associate with a tender engine, in which the the combined length of the firebox, combustion chamber, barrel and smokebox arrangement more or less matches the wheelbase of the engine.

Actually, very quickly upon gauge increase, you start to get into packaging that doesn't use a Stephensonian layout at all; some of the smaller package-boiler watertube arrangements begin to fit, with ample room for the auxiliary equipment to run them effectively.  The same is true for the kind of chain-grate high-pressure boiler on the N&W TE-1, which was tested years ago to be scaled to 6000hp within standard AAR clearance but couldn't be.  

Likewise at some point you can use modular boilers with return.  Donlee discovered (back in the 20th Century!) that if you use steam injection to reduce NOx generation from hotter combustion, a long enough pass would let you recuperate the heat of condensation from the water of combustion -- this was an interesting savings.  Multiple boilers with reasonable turndown could be provided in what would otherwise be 'excess free space' on the chassis somewhere, and it becomes very easy to produce the vertically long centrifugal steam separators for a forced-circulation boiler of adequate steam-generation capability.

As sizes increase, I'd imagine the engine wheelbase to increase out of proportion to the total length of the whole boiler arrangement, leading to more and more "free space" on top the locomotive frames not taken up by boiler.

It certainly does that on a double-Garratt, but to an extent you'll put some of the Rankine-cycle equipment on those frames, and the 'transfer mass' keeps the adhesive weight where it 'needs' to be (see the methods used on large Garratts to address loss of adhesion as fuel and water are consumed)

Perhaps ... intermediate bearing rails would be a pair of rails placed just inside the usual outer rails, with a set of wheels directly behind the outside wheels, somewhat similar to a dual rear wheel arrangement on a truck.

That would solve comparatively little except the spot pressure problems between the wheel and rail contact patches (which become excruciating when a comparatively small increase above current HAL loading is imposed on many forms of rail steel).

The principal difficulties are the beams constituting the axles, and the relative wheelset rotation as the axle becomes very long.  Retaining a shallow coned tread is of some value, and if you remember 'the case for the double-track train' there are advantages to having four rails in reasonable geometry, with trucks on adjacent 'pairs' linked transversely with loadable bolsters.  (This also lets you, in theory, continue to run narrower-gauge trains on the roadbed of something like an advanced RRollway.)  The idea is to spread the 'intermediate' rails across the nominal outer gauge, not space them close to give an effect like a broader contact patch.

[Paul Milenkovic]

"Double-track trains"

The GATX RRollway rail-borne automobile ferry proposal had been depicted in the press as riding on double-track rails.  Part of this may have been the model train exhibited to promote the concept, which if my memory serves correctly, rode on double-tracked HO gauge.

I think this was a shortcut taken by a commercial model builders to avoid having to hand-lay track.  The concept was always ultra widegauge two-rail.  My dad and a GATX colleague are credited with a patent of letting the wheels rotate independently the the manner of what Talgo does, with the patent describing a parallelogram linkage joining the wheels spanning the wide track gauge, and to use rollers gripping one of the rails on one side for positive-force steering rather than relying on cone-tapered wheels connected by a solid axle.

Those steering side wheels raise the question about what to do about track switches?  The answer is that RRollway would require a kind of moving frog switch, somewhat akin to the switches used with the Tru Scale milled roadbed for HO scale model trains.  You would also need a moving-rail replacement for conventional points.  Considerably more complicated than a conventional switch, but a moving frog, I have seen, was used in France for high-speed lines. 

[Paul Milenkovic]

Wardale in "The Red Devil" writes about a Double Garratt proposal for the Argentinian 2-foot narrow gauge coal hauling line.

 

 

One of the problems with ultra-large steam locomotives is that you don't have the modular flexibility as with MU connection of diesel "units."  Yeah, yeah, when EMD's single-engine units were limited to 1500 HP, the MU capability was more to make up for the limited horsepower, and with 6000 HP out of a 2-cylinder 4-8-4 (OK, it was the rare Northern having even that much indicated HP, but the 1500 HP out of an F unit involving 16 cylinders and electric drive didn't wasn't HP at the drawbar, either).

My question/concern is using steam, hydraulic fluid (the Lewty booster) or anything other than electricity to convey power to multiple trucks (bogies).  The advantages of electric drive led to the C&O turbines and Jawn Henry on the N&W, which had problems with coal dust flashing over the traction motors it was said.

So how does a "steam traction motor" work out, with the booster engine, the front engine on a Mallet-style articulated, along with the drive system on Bulleid's Leader and Turf Burner locomotives?  With the 2-cylinder (or even 3-cylinder or 4-cylinder arrangements driving crank axles, all placed near the smoke box), you have short, direct steam passages for admission and exhaust.

As Wardale points on in his "would have been" if his project would have been to enhance the G/MAM Garratt class instead of the 4-8-4 25 class, a Garratt introduces serious pressure drops along the long steam pipes needed.  He claims that this could have been surmounted by "drilling holes in the boiler cradle" and running more pipes in parallel, but I don't know if he had analyzed this more than in the most superficial sense.

As to the Lewty booster, isn't this a hydrostatic drive akin to a garden tractor or a light-utility farm tractor?  For all of the trouble and expense of electric drive, at least the wires don't leak and drip hydraulic fluid.

Haven't all of the large-scale diesel hydraulic locomotives been "hydrokinetic" torque converter drive, which only works of the drive is self-contained with power transmitted to trucks using Cardan (u-joint) shafts in the fashion of a geared steam locomotive?  The knock I have heard on hydro-static (positive with variable-displacement swashplate driven pumps, positive displacement (piston-type) motors) is the friction losses knocking down overall efficiency.

Maybe with a booster, you don't care that much for something only used for starting a train, that the booster "cuts out" to "cut your losses" once underway?  The thing with a booster, however, is that you end up using it for the length of ruling grades, which cuts into efficiency.

And one more question.  How was the Lewty booster supposed to disengage?  Was the motor more of a gear or a vane pump arrangement to which you could just shut of the oil supply and let it spin?  The clutchless gear engagement of the Franklin steam booster was problematic.

[Overmod]

I'll get back to RRollway in another post, as that idea is still dear to my heart and I think at some point it should be tried.

Paul Milenkovic

Wardale in "The Red Devil" writes about a Double Garratt proposal for the Argentinian 2-foot narrow gauge coal hauling line.

That is just the opposite of the situation here.  You physically can't get meaningful TE for 'South African consist density' on 2' gauge with only two engines that even with radiating axles would go around the curves.  So a double-Garratt (as in the original Beyer-Peacock drawn proposals) makes eminent sense there.  Even if what you get is a reversion to the original Mallet chassis, with only four axles to the whole combination locomotive...  

One of the problems with ultra-large steam locomotives is that you don't have the modular flexibility as with MU connection of diesel "units."

There is no need to make excuses for modularity; as I keep discussing with Tom Blasingame and a few others, there is a minimum horsepower for any conventional modern steam locomotive -- essentially set by the need to provide and package sufficient Rankine-cycle auxiliaries and low-efficiency expanders to make the capital and maintenance 'worth' the putative fuel saving -- and for one with an externally-fired boiler that works out to somewhere larger than a couple of 4400hp units.

Some of the Besler and Doble powerplants inherently 'scale' appropriately between small and large power, as of course do some of the Sentinel locomotives with exotic boilers.  The Oxford Catalysts system inherently scales in a remarkably small and light form factor while retaining astounding effective turndown ... what a pity there are 'certain inherent issues' with its practical adoption!

The good news is that from a variety of approaches we converge on two things: (1) that the optimal size for six AC-driven axles is right around 4500hp, and (2) that all components of the locomotive should self-diagnose, and be capable of easy replacement on a reasonably FRU basis 'before' failure.  This applies right up to wide-gauge engines using multiple parallel multiple-pass 'package' boilers to run modular genset expanders, which is the thing that any reasonably wide-gauge heavy operation would likely use.  

The alternative is a light cyclone with SCR aftertreatment, and proper ash recovery via slag handling.  The minimum size for this is most strongly affected by the dual issue of limited practical turndown combined with continuous sinking of the output steam power -- you benefit, for example, from continuous catenary by dumping electricity into it much if not most of the time.  This tends to make railfans who are not also EPRI members look up with that 'wait, what?' expression, but I suspect you of all people will quickly understand how the trick is put together.

My question/concern is using steam, hydraulic fluid (the Lewty booster) or anything other than electricity to convey power to multiple trucks (bogies).  The advantages of electric drive led to the C&O turbines and Jawn Henry on the N&W, which had problems with coal dust flashing over the traction motors it was said.

So how does a "steam traction motor" work out, with the booster engine, the front engine on a Mallet-style articulated, along with the drive system on Bulleid's Leader and Turf Burner locomotives?  With the 2-cylinder (or even 3-cylinder or 4-cylinder arrangements driving crank axles, all placed near the smoke box), you have short, direct steam passages for admission and exhaust.

As Wardale points on in his "would have been" if his project would have been to enhance the G/MAM Garratt class instead of the 4-8-4 25 class, a Garratt introduces serious pressure drops along the long steam pipes needed.

This is the 21st Century, and many of the complaints about the longer steam-line length are relatively spurious in a modern context.

Yes, the steam lines are nominally longer.  Nanoinsulate them and this poses little real issue; trace-heat them and there's less concern still.  If you are really itchy about it, put Wagner throttles far outboard, for example where Porta had them on the ACE3000, and treat the whole run up to there as a recirculated version of the dry-pipe/superheater arrangement on a reciprocating locomotive.  Easy to put sensing in there that automates what you'd do with a 'reversible booster' to start it: run mass flow adequate to clear any spot condensation out of the pipes before opening the cylinders to steam (or the cylinder cocks to subsequent condensate).  Etc. etc. etc.

Likewise any difficulties with long-term sealing of spherical steam-line joints have long been 'settled science' -- I suspect even though we have forgotten much of the knowledge of these in the railroad industry since the 1950s.

There are some consequences that (imho, at least) indicate some kind of dashpot regulation of throttle opening or reverser change, but this is trivial on any modern kind of power reverse, and likely advantageous for a variety of reasons on any modern reciprocating locomotive whether Garratt or 'otherwise'.  (Could be modeled in the PID or wheaver for the antislip system, too, but that's another discussion...)

 

He claims that this could have been surmounted by "drilling holes in the boiler cradle" and running more pipes in parallel, but I don't know if he had analyzed this more than in the most superficial sense.

As I recall the issue at the time was potential wiredrawing at high mass flow.  Read this in parallel with the idea of multiple 15" valves for 'normal' cylinder volume.

I don't share some of his concern with the importance of 'avoiding' wiredrawing -- there are a couple of contexts where it's flat-out useful, as in Franklin type D as built, or in the way Mallets avoid slipping.  But I'm not going to pretend it isn't either valid or desirable to 'streamline' effective mass flow to and from expander cylinders if it is cost-effective (or operationally valuable) to do so.

As to the Lewty booster, isn't this a hydrostatic drive akin to a garden tractor or a light-utility farm tractor?  For all of the trouble and expense of electric drive, at least the wires don't leak and drip hydraulic fluid.

It might bear remembering that, if I recall correctly, Lewty was an engineering 'amateur' in the right sense of the word -- wasn't he an ophthalmologist? -- and so some of the detail design of what he originally proposed circa 1970 was, how should I put this delicately, not fully optimized in its detail design.  Porta, bless his heart, wasn't known to let excessive concern for real-world railroad problems influence his positive thinking in many contexts -- and the "Lewty booster" as espoused by him has certain characteristics in accord with that.

Remember that the Lewty booster derives from an age when 'electronic' controls of anything on a steam locomotive were exotic science fiction, and the system to provide them would involve a whole bunch of time, cost, and introduced points of failure.  Using a frame-mounted triple-expansion reciprocating engine gave positive-displacement steam mass-flow economy at low-speed and part-load operations, while hydraulics involved comparatively simple and robust accumulators, reservoirs, valves etc. for 'fluid power' that normal shop crews could inspect, test, and learn to maintain easily.  The motor is on the frame so all the steam joints can be short and 'perfected'; even very large-bore hoses are a 'known thing' in industrial contexts (and the art of making them bulletproof and weather-resistant in harsh and negligent environments fairly advanced) so implementing a good combination of 'hardlines'/passages and some expedient hoses and shock/misalignment elastic couplings or hoses made reasonable sense.  (And yes, in a pinch, you could run the system on water if you 'had to' -- not something an electrical system does)

Haven't all of the large-scale diesel hydraulic locomotives been "hydrokinetic" torque converter drive...

Yes, and for good reason.  Even if you use the turbines only as 'fluid flywheels' (i.e. no pump torque-multiplication effect) the practical energy density you can run through them is enormous, and of course at near-synchronous speed the heating effect can be quite small (and the driveline can be easily locked direct as in RDCs without necessarily conveying any form of shock back to the attached prime mover).  If you want high speed out of a drive there are very few systems that are better, both for power and for low overall weight (a Bowes drive may be one) and that's a big part of why Mekydro transmissions and the like were so popular in Europe (and here for lightweight trains).  They were an instrumental part of making high-speed engines a practical alternative here in the late Fifties, attractive enough to get Alco in the game for a while with ordinary 251 power.

But if you adapt one of these to the specific context of a Lewty booster, which uses one dedicated source of shp and then bundles it around to various places, the advantages of a hydrokinetic 'transmission' -- in the usual fluid-power sense -- become much less.  If you do the thought experiment of separating the pump from the turbine, for example, the operating economy and transmitted power goes dramatically down and the weight and cost up -- even before the effect of flow restrictions changing flow energy into first pressure and then temperature.  And of course most forms of accumulator are less 'available' to turbines that are often at asynchronous speed from the pump, or are running close to 'locked rotor' under load.  There are also problems with using particularly high pressure in hydrokinetic systems that have more than glacial response times to control inputs -- and most of them wouldn't.

The issues with Cardan-shaft drive are largely solved; if you have equipment using body-mounted motors, the arrangement to driven wheels will already be present as appropriate, and likely a wide range of OTS or relatively easily-adaptable parts.

The knock I have heard on hydrostatic (positive with variable-displacement swashplate driven pumps, positive displacement (piston-type) motors) is the friction losses knocking down overall efficiency.

Friction is only the beginning of the problem.  The great putative advantage of hydrostatics is that they can use very high pressure to overcome the mass-flow issues.  But delivered hp is just as much a mass-flow consideration as steam demand is, and for (say) 1500hp per axle even very high pressure still requires considerable mass to flow.  Where the usual sorts of flow speed (and consequences of higher flow speed) will make themselves increasingly troublesome.

There are much worse problems with regular hydrostatics, most notably their shock intolerance.  If you thought popping circuit breakers with improper transition was fun, try slack run-in effects to vane motors with any practical kind of overpressure mass venting -- it makes the Nathan fibbing about the number of fusible plugs needed for 'safety' look pale.  And internal damage to a positive-displacement motor is seldom going to result in 'freewheeling' of the deflicted machine afterward ... whether or not you can design the hydraulic circuits to let combinations of motors be cut out 'on the fly' or with easy field changes.  There are also a number of potential control issues at this combination of pressure and flow that will suggest themselves urgently to you.  Note how I avoided the fun involved with repeated overpressure to the hoses and joints...

Maybe with a booster, you don't care that much for something only used for starting a train, that the booster "cuts out" to "cut your losses" once underway?  The thing with a booster, however, is that you end up using it for the length of ruling grades, which cuts into efficiency.

One point of the Lewty booster was that it promised much of the increased thermal efficiency of a multiple-compound engine run near its 'best design range' with flexible output.  Not only would it provide effective boost faster than an equivalent Franklin 'high-speed' booster (which, remember, had truly pathetic cutoff arrangements) it would provide useful shp at better efficiency than the same mass flow of steam to simple reciprocating-engine cylinders if you were at higher speed but wanted to economize on water rate or whatever.

Add, or subtract, to this if you are also using the Lewty system as I propose: to run not only a booster but most of the auxiliaries on the locomotive, including via 220V 60 or 400Hz electricity to replace turbine wackiness with OTS sealed motors and conduit.  You would not have hydraulics all over the locomotive to little purpose-designed machines for air compression, high and low temperature feedwater or heat-balance tapping valves, etc.

And one more question.  How was the Lewty booster supposed to disengage?  Was the motor more of a gear or a vane pump arrangement to which you could just shut of the oil supply and let it spin?  The clutchless gear engagement of the Franklin steam booster was problematic.

The simplest version of that involves nothing more than a heavy sprag clutch.  Remember that a 'reversible' booster didn't mean that it boosted the train in reverse, just that you could run the motor and drive backward a bit if it jammed.  A proper booster would be bidirectional, of course, but most applications looked on it a bit like the center turbine on Titanic, only "useful" in starting trains forward with locomotives optimized to run front-end first.

A sprag clutch is inherently overrunning at any speed, has low to no engagement shock to speak of, and runs with hydrodynamic lubrication at almost any time it is not transmitting power.  In a sense you don't care if there is pressure oil applied to your hydraulic motor, as there will be very little additional resistance to 'powering' an idling axle when overrunning, even if its speed were very close to what the final drive of a geared vane motor's might be.  To me it followed that having some sort of reversible overrunning-clutch arrangement of suitable robustness would be easy to arrange if you in fact did want to be able to set back any train you could pull.

Very likely you will be running the actual steam motor at a speed and load corresponding to 'other purposes' so some means of spilling hydraulic pressure to the booster would be valuable.  As noted, with the overrunning clutch the speed with which that's accomplished is not critical; simple spill valves to a recirculating passage with accumulator/reservoir and some anti-shock pressurization would probably do the job as well as necessary...

[L-105]

Overmod

The principal difficulties are the beams constituting the axles, and the relative wheelset rotation as the axle becomes very long. Retaining a shallow coned tread is of some value, and if you remember 'the case for the double-track train' there are advantages to having four rails in reasonable geometry, with trucks on adjacent 'pairs' linked transversely with loadable bolsters. (This also lets you, in theory, continue to run narrower-gauge trains on the roadbed of something like an advanced RRollway.) The idea is to spread the 'intermediate' rails across the nominal outer gauge, not space them close to give an effect like a broader contact patch.

It does seem that using multiple tracks is the optimum solution and perhaps the only realistically viable and effective solution when it comes to supporting potentially huge broad gauge locomotives and rolling stock. The ability to still run conventional "single track trains" is a particularly good point as it allows more scalability of operations depending on demand.

Even a double track standard gauge arrangement would allow for extremely generous proportions. To a give a rough estimate, in "double track mode" the effective gauge would be something on the order of 19' with a maximum loading gauge width of ~25' (depending the loading gauge width of single track trains and track spacing). Perhaps any railroad intending to operate primarily double track trains may wish to comprise it of two narrower gauge tracks (although not so narrow to the point were it constrains the usefulness of single track locomotives). To give an example, a double track arrangemnt comprised of two 3.5' gauge tracks, with each individual track having a loading gauge on the order of 8' width x 12.25' height. Spacing between the tracks (from the centerline of each track) would be 10.5'. For double track trains this would give an effective gauge of 14' and a loading gauge width of 18.5' (so that the total width is same as two adjacent single track trains)  Loading gauge height I'm not so sure, maybe something in the region of 25-30'.

I'm not sure what such a double track steam locomotive would look like, the closest thing I can approximate is something like a fatter, taller C&O M-1 / N&W TE-1 / PRR V1 with mulitiple floors in certain parts. Also I'm uncertain as to how the boiler(s) would be arranged, maybe two M-Type water tube boilers (each with seperately fired superheater) arranged in tandem? Im struggling to find any information with regards to double track trains, the only thing that comes to mind are the famous 80cm Dora and Gustav railroad artillery pieces.

    

 

[Deggesty]

The mention of using a double track reminded me of a TV program in the seventies or eighties; it featured a luxurious (complete with a swimming pool on board) passenger train that ran out of Grand Central on a double track. I do not remember any episode, but I do remember that near the start of one. there was an announcement that the Silver Meteor was arriving in the station. 

Now, back to the topic of the thread.

[Paul Milenkovic]

 

Deggesty

The mention of using a double track reminded me of a TV program in the seventies or eighties; it featured a luxurious (complete with a swimming pool on board) passenger train that ran out of Grand Central on a double track. I do not remember any episode, but I do remember that near the start of one. there was an announcement that the Silver Meteor was arriving in the station. 

Now, back to the topic of the thread.

 

Supertrain!

https://en.wikipedia.org/wiki/Supertrain

 

From these pictures, it is apparent NBC and the Fred Silverman production company didn't understand that the loading gauge could be larger than the track gauge.

http://nbc_supertrain.tripod.com/

It did not appear to run on a double track, merely an ultra-widegauge dual-rail.