Steam That Could Have Been

[Eddie Sand]

Sometime back around 1968-70, Trains featured an article, possibly two, about what might have been if a "stumble" or two had allowed one more cycle in the refinement of steam power. The ideas revolved around things like cross-connected "balanced" drives involving four cylinders -- two at each end of one set of drivers, and the end products depicted were a 4-10-4 built around the idea listed above and a 4-8-6 that represented a "souped up" Lima "mass production" Berkshire. Another feature I recall from about the same time argued that very little refinement in steam technology emerged after 1928; it didn't actually "rank' various classes of steam power so much as argue that the later products of the steam locomotive builder's art sometimes didn't pan out because they didn't fit into the service for which they were eventually assigned -- WM's 4-8-4's and P&LE's Berkshires (Alco's last steam order) as prominent examples. Conversely, the Decapods which were Lehigh and New England's last seam power drew high praise.

[Overmod]

Eddie Sand

Sometime back around 1968-70, Trains featured an article, possibly two, about what might have been if a "stumble" or two had allowed one more cycle in the refinement of steam power. The ideas revolved around things like cross-connected "balanced" drives involving four cylinders -- two at each end of one set of drivers, and the end products depicted were a 4-10-4 built around the idea listed above and a 4-8-6 that represented a "souped up" Lima "mass production" Berkshire.

I think you have got this a little sideways: the 'balanced' Withuhn drive requires an even number of drivers to work correctly (you need a Deem-style geared conjugation to do a 4-4-6-4 or similar) so it would be eight-coupled or 12-coupled (the latter almost certainly having so long a resulting effective rigid wheelbase as to make a modern Challenger a far better approach). 

The souped-up Berkshire would be a 2-8-6; there was no reason on any fast-freight engine to use a four-wheel lead truck.  The 4-8-6 was actually pitched by Lima (see Hirsimaki's book) but as a better 4-8-4 dual-service engine. 

It's easy with 20/20 hindsight to make claims about 'practical steam technology improvement ending in 1928' -- but in actual fact, quite a few of the real significant changes came well after that date, and no few cockamamie misadventures predated that.  (Amusingly enough, one of the great dead ends, the Central Machinery Support 2-12-6, was pitched to the International Railway Fuel Association in 1928...)

Look at Alco and its assumption that three-cylinder power was THE shape of the future as seen from the mid-Twenties, or anyone's experimentation with the Schmidt system or other approaches to very high working pressures stemming from power-station technology improvements after WWI.  Look at the massive trouble involved with the articulated trailing truck on the original Berkshires, a design guaranteed to keep you in the drag-speed range Or Else with an eight-coupled locomotive.

Meanwhile, consider the work done after 1928, for example by Eksergian, regarding high-speed cross-balancing (in better-designed driver centers) leading by the late '30s in the recasting of decidedly slow engines like the original T&P 2-10-4s as quite capable fast-freight power.  At around this time, Voyce Glaze at N&W was writing down what became the low-overbalance calculations and rod-configuration details that made the J locomotive capable of high running speeds.  It's not entirely fair, although certainly appropriate in its place, to note the disaster involved with experimentation with nickel-steel alloys for boilers in the '30s and '40s (followed by fairly heroic levels of replacement boilering activity in the '40s and '50s!), or to speculate (as in the Classic Trains article) that better balancing didn't make the NYC A2As 'good enough' road locomotives, or to claim as LeMassena did that the steam passages in the S2a Niagara were 'unintentionally' made poor-flowing.

L&NE is something of a dubious example because they were essentially the transportation arm of a mining business, and were otherwise known for "successfully" (by their standards) using things like Very Large 2-8-0s with Bedlam Auxiliary Locomotives on the tenders.  Did not keep that stuff running very long into the diesel era, either, IIRC...

Which brings up the potentially valid issue of what DID make locomotives 'fit into the service for which they were eventually assigned'.  K4s survived until quite late in the game on Long Branch commuter trains, not exactly the stuff for which 80"-drivered locomotives are best suited.  UP ran the Nines long after they had high-speed power better suited in almost all respects.  Milwaukee got rid of the F7s with almost uncanny speed, nearly as soon as they could -- I personally think this had to do with unavoidable and chronic main-pin breakages, the utter danger of such occurrences at high speed being fairly easy to imagine, but still, for such a 'successful' design to be abandoned so quickly...

In a very real sense, many of the designs that survived were significant particularly because they were NOT supersophisticated, with boiler alloys that would rust away in months, or equipped with fancy auxiliaries that railroads couldn't keep maintained, or developing high power at great size.  What this says most prominently to me is that there was a fundamental ... I won't call it an error, because it made sense at the time, particularly in the WWII era ... call it a poor life choice, made by most American railroads in applying high technology in and after the Depression years.  Very little of the modern improvements were made to smaller power, and when they were (as on the Frisco locomotives like 1351/1352) a great deal of the improvements were whiz-bang gimcrackery rather than fundamental reliability enhancement.  I suspect, economics being what it was (especially with regard to tax law) the outcome wouldn't have been otherwise... especially with dieselization posing a particularly attractive alternative for what most smaller power did.  But most big power generally didn't 'pan out' because diesels were perceived as a much better solution for what they did -- witness the T1, and the WM 4-6-6-4s, as two pointed examples we've discussed fairly recently -- and no matter how good or how sophisticated they were, they were just too big and complicated to do smaller jobs effectively.  (Tuplin noted that a Niagara was quite capable of doing a cost-effective imitation of the 2-8-0, but that didn't make it sensible to operate one in 2-8-0-size service even between required shopping or service intervals...)

[Paul Milenkovic]

Duplex/Deem drive are maybe in the gimcrackery category.  Yes, badly designed, badly operated, or badly maintained steam locomotives could bend rails with their dynamic augment "hammer blow" and wreck large sections of track.  But the sense I get (from the cough, H. F. Brown paper, cough) is that you could balance two-cylinder locomotives to not be hard on track.  The most recent thinking is to underbalance them, that is balance against the rotating mass but not the reciprocating mass, and to have yet-another-Franklin-gadget in the form of an articulated connection with the tender take up the back-and-forth surging that results.

But what else was a gimcrack gadget?  The feedwater heater?  The superheater?  The Valve Pilot?  Two of those gadgets made substantial contributions to water economy and in turn fuel economy, and by increasing those economies, you also increase the HP limit of the locomotive.  The Valve Pilot allowed doing more with less locomotive and also gave important diagnostics.

[daveklepper]

And modern drivers, and cast engine beds, and lightweight rods and roller bearings

[Overmod]

Paul Milenkovic

Duplex/Deem drive are maybe in the gimcrackery category.  Yes, badly designed, badly operated, or badly maintained steam locomotives could bend rails with their dynamic augment "hammer blow" and wreck large sections of track.  But the sense I get (from the cough, H. F. Brown paper, cough) is that you could balance two-cylinder locomotives to not be hard on track.

The conjugated drives are only significant if you are running high cyclic rpm and have a need either for 'diesel-grade' levels of track maintenance or high-speed alignment.

You are correct that for most American needs a two-cylinder engine with lightweight rods produced 'acceptable' augment, but one thing that was coming up at the end of steam (and as noted was probably a deciding factor in the relatively early scrapping of the Milwaukee F7s) was main pin failures.  Other aspects connected with large two-cylinder power involved valve gear and main rod deflection (leading to buckling distortion or failure) and separation or breakage of piston rods (they are hollow in good lightweight practice). The incidence of these problems goes up geometrically with increased cyclic, and problems that are vanishingly slight at, say, 60 mph are clearly emergent at 80.  Dividing the drive is already a significant improvement on, say, a 6000 hp locomotive that is intended to operate at 100 mph (see the NYC C1a design, for example) and while the conjugation appears to be gimcrackery much of the time, it is intended in particular to arrest high-speed slips (exacerbated by the unloading of short wheelbases by track imperfections and shocks) which can propagate to high speed when the locomotive has precise valve gear such as Franklin poppets.

The most recent thinking is to underbalance them, that is balance against the rotating mass but not the reciprocating mass, and to have yet-another-Franklin-gadget in the form of an articulated connection with the tender take up the back-and-forth surging that results.

This was 'recent thinking' circa about 1945, and a bit muddled in this telling.  We call this 'zero overbalance', and it is made possible largely by the great stiffness, wheelbase length, and polar moment of large American road power.  The 'Franklin-gadget' is not an "articulated connection", it is the Franklin radial buffer, and its main purpose here is to couple the surge in the locomotive effectively to the substantial mass of the tender.*  (A subsidiary purpose is to damp the surge excursions going to the train, which can be a source of discomfort when tightlock passenger couplers are in use -- see the fun that was produced when ATSF first put its DL-109 consist on a passenger train!)   Note that by about 40mph the frequency of the surge is such that the actual motion of the locomotive is comparatively slight -- within the range of excursion of the buffer -- and above that speed the effect of surge is of little moment (pun intended) in terms of train handling. 

Surge is actually more of a problem on big unconjugated duplexes, as the PRR test plant results clearly showed -- this was especially bad when the engines were intentionally made to 'break phase' (by changing the stroke or the driver diameter of one engine) , as they would then come in and out of phase, with great surge (enough to make some of the test crew a bit nervous!) at the range of best synchronization.  Here you would need something like Langer's patent 2432907, which is otherwise clearly in the gimcrackery range, or Thearle's earlier 'take' on the idea for Alco (1828627).  But you'd have to do some 'overbalance' here, too, as the component of piston thrust going into yaw is not compensated by the device, and this is precisely the thing that causes the problem in the Q2 (evidenced by its only manifesting under about 35 mph indicated, which wouldn't be the case for dynamic-augment inertial concerns...)

The gimcrackery I was referring to was various sorts of device like the Bailey Foam-Meter, giving an idea about priming in the boiler, or the Elesco Steam Dryer (which, redesigned to have steam-separator geometry, might have been more effectual without an external drain) -- or designs of feedwater heater or ESE full of little proprietary parts that needed to be expensively sourced from the manufacturer when, not if, they broke or wore out.  It isn't so much that they don't do something 'useful' -- it's that they have a high cost, and non-graceful degrade or failure, for what they provide.

The Valve Pilot was hell to keep in running order, and depended on an empirical cam to match wheel rpm with expected reverser setting.  It also depended on good contact of its friction wheel with the driver rim, which is one thing with the engine neatly standing still and quite another at high speed, especially in the range where the driver may be cycling vertically at, say, 8 Hz at 480 rpm.  I suspect much of the 'value'  railroads actually derived from the thing came from pulling the tapes -- not to compare them against grade and speed, but to detect infractions of overspeed or slow orders.  (I admit that part of this is a bit of sour grapes -- there was a perfectly good system for doing automatic proportional cutoff control - none of this matching-needles-to-what-the-analog-cam-thinks business - in the early Twenties, and THAT did not catch on... perhaps <dons tinfoil hat> because it was closed-loop to backpressure and didn't have a tape that included road speed... </tinfoil>

A superheater is essential, if for no other reason than keeping nucleate condensation low over the full range of efficient cutoff.  A better question is where it's gimcrackery, and where it isn't, to use superheated steam for auxiliaries.  (My own opinion, FWIW, is to have an alternator on the Lewty booster engine, generating AC (pick the frequency that gives best OTS machinery choice!) which then runs all the rotating pumps and machinery on the locomotive, instead of having lots of inefficient little steam turbines.  I do not consider this 'gimcrackery' although I'm sure there are many who would...)

I'm on record repeatedly as confirming the need for feedwater heating, in several stages, as well as air preheat, some FGR, and (where cost-effective or justifiable) economizing return a la Franco-Crosti.  But I'm also firmly in line with the idea that thermodynamic efficiencies have to be cost-justified, and not make the locomotive less usable in actual real-world railroad service. 

Ross Rowland, for one, made a decision not to put feedwater heating (or Snyder preheaters, or Cunningham circulators) on 614, even in the 614T days.  He will, I hope, comment here on why he chose not to do so, and that you will recognize the results as enlightening...   ;-}

__________________________________________________________

*Amusingly enough, I got the geometry of the device wrong when I first started designing locomotives, thinking that it worked a bit like a ball-and-socket joint, rather than two opposed radii making line contact.  I then got into making buffer surfaces that were essentially spherical, with lots of twiddly compliance mechanisms... Franklin got the answer right, as it did with another 'novelty' that I consider essential on modern power, the self-adjusting wedge.