THE magazine of railroading

SEARCH TRAINSMAG.COM

Enter keywords or a search phrase below:

Linked valve gear on steam locomotive

1113 views
18 replies
1 rating 2 rating 3 rating 4 rating 5 rating
  • Member since
    May, 2004
  • 3,244 posts
Linked valve gear on steam locomotive
Posted by 7j43k on Wednesday, February 22, 2017 11:20 AM

First, apologies for what doesn't quite look like the right term.

But.

I have noticed over the years comments about how steam locomotives with two engines (articulateds and duplexes) could have one engine slip.  I wonder why the two engines didn't run off of essentially one valve gear--it seems that would make it impossible for one engine to slip.  I am thinking especially about the duplexes, as it seems the linkage between the engines would be easier.

Or, of course, I am wrong.

Comments?  Please.

 

Ed

 

Oh.  On consideration.  I DO know that the above locomotives had "linked" valve gear, in that there was one single control for the gear, usually a power reverse.  But I am talking about sort of only having ONE valve gear for TWO engines.

  • Member since
    February, 2005
  • 1,612 posts
Posted by timz on Wednesday, February 22, 2017 12:32 PM

If just one valve gear for the two sets of drivers, you'd have to connect the two sets of drivers too, with another pair of rods. So the locomotive couldn't be articulated.

  • Member since
    May, 2004
  • 3,244 posts
Posted by 7j43k on Wednesday, February 22, 2017 1:01 PM

timz

If just one valve gear for the two sets of drivers, you'd have to connect the two sets of drivers too, with another pair of rods.

 

 

I don't see why.

 

Ed

RME
  • Member since
    March, 2016
  • 1,243 posts
Posted by RME on Wednesday, February 22, 2017 7:13 PM

7j43k
I don't see why.

It's really simple.

Valve gear events have to be precise at parts of the valve stroke to within 1/64", repeatably and at both ends of the cylinder.  If an engine slips, even a little bit, or if its wheels wear to a different size, even a little bit, that single valve gear won't maintain anywhere near the correct precision, and of course there is no mechanism at all to restore the engine to synchrony with what the valve gear is doing.

The situation is really so bad that you can't use long Cardan shafts to drive two sets of Franklin type B from a single gearbox -- there is too much torque windup and whip in the shafts.  (This is a reason for the almost cartoonlike heaviness of the shaft frames provided for some engines with type B gear). 

Now, there's nothing that says you can't use one reverser to control the two sets of valve gear (or one control to do the comparable adjustment for a Franklin nightmare box - see PRR drawing 439426A, the general arrangement for type B-2, for an example).  This was in fact the arrangement anticipated for the NYC C1a duplex (one reverser running a linkage that would lift the gear on both engines equally) and in fact most of the time you'd want a cutoff control that did in fact adjust both gears nearly equally.  The problem is that at the most important times, it's even more desirable to have fine-adjustment capability for the reverse than to have fine adjustment on the throttle flow 'by engine'.  And that is true both in terms of quick correction and recovery, as in stopping a slip while maximizing starting power and acceleration, and in terms of being able to 'trim' one engine separate from the other.

Interestingly enough, even if you conjugate the engines with the 'original' type of Deem gear drive, there's enough difference to put serious shock and wear on the gear teeth and shaft splines.  And of course my version of that gear has a Ferguson or MR clutch, and allows the engines to be arbitrarily dephased even though there is detenting to bring the engines back into some semblance of phase 'in time'.

It would have been interesting to see how the valve drive for the ACE3000 would have been detail-designed, since Withuhn conjugation is nominally little different from eight-coupled rod connection in terms of angular precision.  I believe you'll find that lateral whip in the lightweight Timken rods is a greater source of inaccuracy than anything associated with the rod roller bearings or their spherically-ground sleeves and seats.

  • Member since
    May, 2004
  • 3,244 posts
Posted by 7j43k on Thursday, February 23, 2017 7:47 PM

RME,

Thank you.  I'm certainly not following everything you're saying, but I am getting a sense of it.

 

Ed

  • Member since
    March, 2013
  • 379 posts
Posted by Dr D on Friday, February 24, 2017 1:36 AM

7j43k

Ed,

Your initial question is a logical one.  Most modern designs of internal combustion engines operate multi cylinder valve systems through one valve drive mechanism.

Steam railroad locomotives developed in the 19th Century through the fourth decade of the Twentieth Century were potentially working towards this more modern valve gear concept in the development of "poppet valve" steam admission mechanisms of the 1930's.

-------------------------

Lets examine some basic engine concepts here.  The elemental piston ported Steam Engine consists of a (flywheel) or wheel, drive rod, piston and cylinder. 

The operation of this piston can be controlled in several ways. 

1 - the oscillating cylinder design in which the pivoting cylinder alternately charges and exhausts the working steam.  This is common in childrens toy live steam engines. 

2 - the piston valve design which has a secondary parallel admission valve to control steam exhaust and admission through the use of a basic valve gear mechanism.  This is your usual valve piston operated by, Stephenson, Walscharts, or Baker valve gear designs. 

3 - the use of geared "poppet valve" admission and exhaust through use of cams and driven valve train in the form of axle geared drive shafts or crank mechanism from the crosshead.  This would be similar to modern internal combustion engines in concept.

---------------------

Whatever the case the definition of the "basic engine" consists of one cylinder and piston and one drive rod to (flywheel) or wheel. 

In the case of the basic railroad locomotive there would be two such "steam engines" one on each side of the railroad locomotive each with its own easy to service primitive admission valve gear.

In the case of a more complex articulated Mallet type of railroad steam locomotive there would be four basic steam engines, one on each side for each pair of wheels.  Each would have its own valve gear mechanism.

---------------------

The introduction of the "poppet valve" steam admission design featured the use of automotive type camshaftes and valves with one or more intake and one or more exhaust valves.  These early designs copied railroad practice of the time and featured each an independent "poppet valve" system for each basic steam engine of the railroad locomotive.

Some systems featured a gearbox and drive shaft similar to automotive chasis practice - that was driven from the center locomotive drive wheel.  Some systems of railroad "poppet valve" gear dispensed with this and drove the "poppet valve" camshaft by a linkage attached to the piston crosshead of each cylinder.

------------

Because the right and left sides of the basic railroad steam locomotive were joined by a common set of (flywheels) wheels - I see no reason that they both could have been driven by one admission valve drive mechanism - aka drive wheel gearbox and drive shaft or combination rod from one crosshead. 

This was I believe the original point of your question.  

Interestingly such a common admission valve drive mechanism for the railroad steam locomotive was designed in the 1900's and did exist in three cylinder design locomotives.  In this case the center cylinder admission valve mechanism was eliminated and the valve was instead driven in "congugated" fashion from the two outer cylinders.  

This system was logical from a conceptual standpoint, but in actual practice often gave trouble as the "congugated" valve gear required a great deal of maintaince.  Such a congegated three cylinder railroad locomotive still exists in St. Louis, MO and at the Southern California Railroad Museum - the Union Pacific type 4-12-2.

-------------------

It seems likely that if a traditional piston operated steam locomotive were to continue to be designed today it would include the valve gear design you are asking about.

------------------- 

Its a good question.

- Doc  

RME
  • Member since
    March, 2016
  • 1,243 posts
Posted by RME on Friday, February 24, 2017 2:04 PM

[quote user="Dr D"]Some systems featured a gearbox and drive shaft similar to automotive chassis practice - that was driven from the center locomotive drive wheel. Some systems of railroad "poppet valve" gear dispensed with this and drove the "poppet valve" camshaft by a linkage attached to the piston crosshead of each cylinder.[quote]

The normal place you found the 'drive' was on the main driver, often because it was simple to adapt the type of return crank used to drive SHM (Walschaerts or Baker) so that it would turn a centered gearbox.  Driving from another coupled wheel would necessitate a special longer rod pin.

The technical name for shaft drive with either single or double (constant-velocity) U-joints is "Cardan shaft" -- it is amusing to watch people research who 'Cardan' was and find out the approximate date of the invention.

You should mention that the practical crosshead-drive or 'scissors' gear systems drive one side relative to the OPPOSITE crosshead of a quartered engine.  (See Woodard's patent and the application to Franklin type A, which was an oscillating-cam gear, for examples)

Because the right and left sides of the basic railroad steam locomotive were joined by a common set of (flywheels) wheels - I see no reason that they both could have been driven by one admission valve drive mechanism - aka drive wheel gearbox and drive shaft or combination rod from one crosshead.

In the case of crosshead drive, there are a number of good geometrical and physical reasons why drive from only one crosshead will not be reliable, and in fact why it may exhibit different precision on the two sides over a wide range of operation (which to me is a tremendous no-no, on the order of having throttle bodies on a twin-six IC engine not synchronized...)  This is not as true for gear-and-shaft drive, and in fact the first system of gear drive from the main I saw as a boy was the Reidinger system as illustrated in the Encyclopedia of World Railway Locomotives, which shows the shaft only on one side and a cross-shaft at the valves.  By extension this system would drive three or four cylinders with equal facility, and nearly optimal precision if the effective shaft whip between cams was kept low (which is not difficult with simple detail design, e.g. by making the shaft tubular and larger-diameter for the same mass of material, as with driver axles)

The problem according to Franklin was that if anything went wrong with this style of drive, the whole locomotive would instantly be crippled in a way that was difficult to deal with on the road.  Hence they arranged to have a gearbox on each side, and adjust the constraints and lost motion accordingly to make it work, for type C.  Type D uses short camshafts, not linked 'across the middle', so there is a separate drive arm, gearbox, and shaft run on either side.  This is strictly for power and rotational timing; the lateral shift is arranged under separate control (on type D that is by nonproportional air servos)

Interestingly such a common admission valve drive mechanism for the railroad steam locomotive was designed in the 1900's and did exist in three cylinder design locomotives. In this case the center cylinder admission valve mechanism was eliminated and the valve was instead driven in "conjugated" fashion from the two outer cylinders.

Conjugation is not at all the thing he's talking about, and indeed if either of the side valve gears is at all unsynchronized relative to the other, the events on the center cylinder go to Hades (much as they do with either whip or lost motion in the setup).  All Gresley or three-cylinder Holcroft do is to geometrically define a 'trajectory' for the center valve spindle relative to the other two (it is much easier to comprehend this if you watch something like a YouTube animated diagram of different kinds of conjugated gear); the setup has no other relationship of any kind to the valve drive on either side but only the relative movements of the two valve-rod spindles.

I tremble to think of the problems that would occur if you tried to use reciprocating or shaft motion from two mechanically unconjugated (here the word refers to linking the two piston-and-siderod drives mechanically so they stay 'in step') engines to drive a conventional conjugated valve-lever arrangement.  It would be amusing, that's for sure ... and I know of no way, other than very stout and non-disengageable mechanical engine conjugation, that would even make the arrangement work during the short time its bearings and clearances were kept at tight tolerance.

An interesting 'alternative' was proposed by Bulleid for the sleeve valves on Hartland Point and the Leader class, which had a drive that moved the sleeves fore-and-aft and also rotated them (to help, supposedly, with a tendency for all the rings to seize when 'stopped' each time they reversed longitudinal direction).  The arrangement that 'walked' the sleeves was driven off a sprocket on one of the axles, similar to the somewhat abortive drive seen on the Merchant Navy locomotives but differing in that it neither reversed direction periodically nor had great changes in demanded force over each cycle.  It would not have been difficult to use such a power drive to turn a RC setup (or an OC setup through suitable mechanism) so that it can be timed with a precise method that does not need to exert mechanical power to move the valves.  That, for example, is one way that a valve gear using something like an 8192-position optical encoder for rotary reference and some form of PLC to derive valve positions might implement 'riding cutoff' on a reliable mechanical system that is 'in time' for less rigorous speeds or high-speed loads.

 

  • Member since
    May, 2004
  • 3,244 posts
Posted by 7j43k on Friday, February 24, 2017 2:59 PM

Thank you, gentlemen, for continueing discussion of my topic.

I think it might help and/or clarify things if I divulged my motive for bringing up the subject:

For most of my life I've casually tried to design MY "ultimate" steam locomotive.  In MY particular case, it comes down to a duplex drive with cylinders at each end of the rigid wheelbase.  It seemed (and seems) that if one could control the rotational timing of the drivers, one could arrange things so that the reciprocating masses of the main rods would ballance out.  Individual engine slip would also be eliminated.

My feeling is that it would be simple with current technology.  The valve control system would be essentially electronic.  Actual activation of the valves is my big sticking point with this system.  Well, perhaps not a "sticking point", because it can clearly be done.  But getting it small enough and reliable enough to work and fit on a steam locomotive is more the challenge.

 

Ed

RME
  • Member since
    March, 2016
  • 1,243 posts
Posted by RME on Saturday, February 25, 2017 12:45 AM

7j43k
For most of my life I've casually tried to design MY "ultimate" steam locomotive. In MY particular case, it comes down to a duplex drive with cylinders at each end of the rigid wheelbase. It seemed (and seems) that if one could control the rotational timing of the drivers, one could arrange things so that the reciprocating masses of the main rods would balance out. Individual engine slip would also be eliminated.

I assume you are familiar with the Withuhn conjugated duplex (first described in a 1972 issue of Trains Magazine, and then famously incorporated into the ACE3000 design) which balances all the first-order rod forces with a pair of appropriately-phased quartered internal rods (acting on cranks between the inmost set of duplex engine axles, #2 and #3 on an eight-coupled locomotive).  That does exactly what you are proposing, but the practical implementation of the thing with roller-bearing axleboxes in an appropriately strong frame structure is difficult.

Riley Deem's proposed method of conjugating the Q2 chassis (with gears instead of rods) and my variants of that approach (with clutches and detent phasing at 45 or 135 degrees instead of Withuhn's full balanced opposition) are a different approach.  Technically, both Withuhn's and Deem's approach have sufficient strength and precision that a single set of valve gear could be used for all four cylinders.  If you have not studied the discussion in the ACE3000 patent, I recommend it to your attention; I also encourage you to look at other ways of keeping the engines mechanically synchronized without interfering with suspension compliance and other mechanical 'desiderata'.

An alternative approach to balancing the reciprocating masses is in theory a bit more direct: that is the 'salmon rod' approach used on some French locomotives, perhaps most notably the infamous Algerian Garratts.  This uses a return-crank drive to rotate a quartered jackshaft, which then drives an OC mechanism that operates not poppet valves, but 'drop valves' which are basically poppet-valve sized but use circumferential ring sealing like piston valves.  The French decided that if you weighted the eccentric rods appropriately at their forward ends, the inertial forces of the reciprocating component of the main and of the eccentric rod could be made to oppose each other in balance, essentially in the same longitudinal plane, and thereby eliminate both yaw and surge components without resorting to duplexing or zero overbalance.

 

My feeling is that it would be simple with current technology. The valve control system would be essentially electronic. Actual activation of the valves is my big sticking point with this system. Well, perhaps not a "sticking point", because it can clearly be done. But getting it small enough and reliable enough to work and fit on a steam locomotive is more the challenge.

If you are using poppet valves, both the actuators and their positional encoders (either optical or LVDT) are available in the required size and duty cycle (the technology of one company's product having been evolved for the Apollo program in the '60s).  Either electropneumatic or electrohydraulic systems can accomplish the necessary time precision and physical resolution.  Different systems would be used to provide the drive power and 'fluidic amplification' on a steam locomotive, but again most of the needed capacity is found in the air-brake system and the workability of a Lewty booster system.

I would argue, however, that a purely electronic valve gear (like the electronic camshaft" approach on some compression-ignition engines) is NOT a good idea for large reciprocating locomotives with thin-section Timken rods.  What I recommend instead is a mechanical Reidinger-style gearbox and shaft drive to a camshaft, with the electronic actuators (however they are implemented) working on mechanisms in the followers.  That way, if there is a failure in the electronics or one of the actuator systems, the 'default' will be mechanically-timed operation, or some form of Wagner or Trofimov bypass.

One other little note:  Nobody really got over the problem of the inlet and exhaust piping to the rear cylinders of an opposed duplex.  The situation begins to approach the complexity, and the consequent weight consequences, of the arrangement on the Lima Alleghenies (Big Jim probably remembers the N&W comment, either from Pond or Glaze, about that) and it really isn't worth it compared with running a conjugating shaft with a pivoted center bearing through the transom between cylinders in a cast (or strong fabricated) underframe.  At some point it becomes preferable to go to a turbine arrangement rather than a complex and inherently-flawed reciprocating setup.  In my opinion, either a Bowes drive or a magnetorheological clutch gets around some of the difficulty with turbine drive, and proper active elements in the front-end design deal with most of the rest of it.

  • Member since
    May, 2004
  • 3,244 posts
Posted by 7j43k on Saturday, February 25, 2017 3:30 PM

Thanks for your insight and comments.  I will poke around in the areas you suggest.  

I do think your idea of not depending completely on non-mechanical valve timing worth considering.

I rather like having the rear set of cylinders placed to the rear of the drivers because of the shortening of the rigid wheelbase.  At first glance, the only difference between the steam pipe routing of this setup and a more conventional "mid-mount" cylinder is the length of the pipe run.

I will confess that I have certain "aesthetic considerations" in mind with this project.  And something that was a box with wheels under it would lose my interest, even if it were transcendentally efficient.  So much for rationality.

Ed

  • Member since
    July, 2004
  • 2,145 posts
Posted by Paul Milenkovic on Saturday, February 25, 2017 9:55 PM

So that was the setup behind the Cossart valve gear?  Did those "salmon rods" on 180-deg eccentric cranks really supply the reciprocating counterbalance -- why was this setup not more popular?

If GM "killed the electric car", what am I doing standing next to an EV-1, a half a block from the WSOR tracks?
RME
  • Member since
    March, 2016
  • 1,243 posts
Posted by RME on Saturday, February 25, 2017 11:06 PM

Paul Milenkovic
Did those "salmon rods" on 180-deg eccentric cranks really supply the reciprocating counterbalance -- why was this setup not more popular?

Part of the problem was that it only 'worked' right geometrically when the two rods were driving a rotating jackshaft, in the right 'phase' as you noted, not when they worked valve rods and combination levers as on a conventional SHM valve gear.  It probably didn't help that the setup was French, and that there wasn't an American industrial powerhouse like Baldwin (with Caprotti) or Franklin to popularize the idea.

But another part of the problem is that it didn't work as originally intended.  If you look at pictures of the setup, you will notice that the lightening holes in the big end of the 'salmon' get larger and larger, perhaps as recognition sets in that the balancing and inertial effects aren't working as expected.

Meanwhile, things like the Langer balancer for surge and the Berry Accelerator (look it up on Dr. Leonard's site) were never even tried in the field; they came just a bit too late, and everyone had stopped caring.

Personally, I liked the idea of the dwarf drop valves themselves a great deal, and I have to wonder whether they could be combined with Wagner 'servos' to produce valves with high precision and very low wear at high speed. 

RME
  • Member since
    March, 2016
  • 1,243 posts
Posted by RME on Saturday, February 25, 2017 11:20 PM

7j43k
I rather like having the rear set of cylinders placed to the rear of the drivers because of the shortening of the rigid wheelbase.

But it cramps the firebox, chamber and throat terrifically, and just look at the fun you have accessing the outer cylinder heads or trying to remove pistons to do rings, or machine and dress the bores...

The increase in rigid wheelbase is usually minimal in any meaningful sense, as the forward engine has a long-travel lateral-motion device by necessity, and the other axles have permitted lateral.  Note the way the Q2 design 'necked' the cylinder saddle at the plane of the drivers so that the cylinder ends could overlap the wheels; the formal rigid wheelbase is, if I recall, actually shorter than that of an ATSF 5011-class 2-10-4, and there are a surprising number of 'lateral motion devices' on the five axles of one of those locomotives.  This design was done after considerable experience, first with observing the George Emerson 5600 prototype on B&O and then PRR's own Q1.

 

At first glance, the only difference between the steam pipe routing of this setup and a more conventional "mid-mount" cylinder is the length of the pipe run.

But remember that it's not just one 'pipe run'; it's four, and the exhaust from that rear engine has all the drivers' worth of distance to travel but has to be equalized with what's transiting the very short tracts in the front-engine cylinder block.  Much easier to 'siamese' the exhausts as on the T1, especially if you're conjugating the two engines with some sort of phase-preserving arrangement.

(Now, if you're using some version of Franco-Crosti 'economizing' preheat, and using slipstream drafting for a proportion of your high-speed draft as you should on a true high-speed locomotive, there are advantages to having your exhaust manifolding plena terminate at the center of the driver wheelbase.  But I suspect you're not quite that far along in the detail design yet...)

Please don't get me started on the ACE3000 as it was presented.  Aside from having more bugs than a New York apartment with peanut butter smeared on the walls, it was pathetically underpowered for all the complexity it required (but the wack fueling system apparently depended materially on the low average power).  We won't go into the expediency (or lack thereof) of requiring compound operation in American summer conditions and the like.  And the theory that 'making it look like an SDP40F' would somehow make railroad buying agents more amenable to it was not exactly rocket-engineering grade thinking.

  • Member since
    May, 2004
  • 3,244 posts
Posted by 7j43k on Sunday, February 26, 2017 11:36 AM

RME
But it cramps the firebox, chamber and throat terrifically, and just look at the fun you have accessing the outer cylinder heads or trying to remove pistons to do rings, or machine and dress the bores...

The increase in rigid wheelbase is usually minimal in any meaningful sense, as the forward engine has a long-travel lateral-motion device by necessity, and the other axles have permitted lateral.  Note the way the Q2 design 'necked' the cylinder saddle at the plane of the drivers so that the cylinder ends could overlap the wheels; the formal rigid wheelbase is, if I recall, actually shorter than that of an ATSF 5011-class 2-10-4, and there are a surprising number of 'lateral motion devices' on the five axles of one of those locomotives.  This design was done after considerable experience, first with observing the George Emerson 5600 prototype on B&O and then PRR's own Q1.

 

I surely would like to read the various reports generated about the Emerson and the Q1.  I wonder, for example, if there are ways to design around their complaints that weren't available to them.  

I am, by the way, envisioning six driving axles with 70" drivers.  And figuring on oil burning, which removes the bulkiness of ash handling.

 

 

 

But remember that it's not just one 'pipe run'; it's four, and the exhaust from that rear engine has all the drivers' worth of distance to travel but has to be equalized with what's transiting the very short tracts in the front-engine cylinder block.  Much easier to 'siamese' the exhausts as on the T1, especially if you're conjugating the two engines with some sort of phase-preserving arrangement.

 

 

Yes, four pipes.  But the Q2 had four pipes, just not as long.  And the extra length is not something that can be ignored.  I figure, at a minimum, there would have to be twin exhaust stacks, split front/rear.

 

 

Ed

RME
  • Member since
    March, 2016
  • 1,243 posts
Posted by RME on Sunday, February 26, 2017 12:24 PM

7j43k
Yes, four pipes. But the Q2 had four pipes, just not as long. And the extra length is not something that can be ignored.

The Q2 had all the cylinders at the front of the engines, which limits both the length and tuning problems.  Even then, there turned out to be severe problems that were, at the time, attributed to steam momentum.  You should read some of the test reports and determine the source and magnitude of that problem in your pipes.

The locomotive that bears inspection as to 'pipes' is the Q1, which ran one set of pipes very high and outboard, with some interesting curves -- note in particular the location of the diaphragm cover for the PRR-standard snifting valves on the rear cylinders.  There's a relatively high cost in developing all this plumbing compared to what is involved on the Q2 or T1 with all cylinders facing the rear (and heads easily accessible) where the admission steam and the exhaust plena have room and reasonably direct routing.

Six axles is already over the top for a rigid-frame engine; there's little point in not going to a good simple articulated with compliance control for the 'hinge', set up as Voyce Glaze and later Alco did with minimal vertical compliance in the lead engine so it equalizes like a rigid 12-coupled but hinges neatly in the middle.  Or, if you're a glutton for punishment, build an American Garratt...

 

I figure, at a minimum, there would have to be twin exhaust stacks, split front/rear.

There is little room to route the 'exhaust stack' at the rear, even if you use a full Franco-Crosti exchanger and air preheater (which would work well for oil firing, as its exit is right at the throat where the burners would be, but will only have concurrent flow).  You would not want to route it under the firepan, and it can't be placed as it was on the Triplexes because you (sensibly!) have no engine under the tender.  Your combined stack, nozzling, and steam flows in the rear will have to occupy a long, relatively thin duct going around the circumference of the chamber just ahead of the firebox legs; if you do that, I'd see if a Giesl ejector, perhaps with limited 'fanout', would work.  The historical examples like the British 9F used only one stack, probably a cost issue but also concerned with visibility; there's no objective reason you couldn't use a pair if you have the free dollars^3 to make it go.

There is then the issue of how you split the combustion gas to optimize draft contribution from the front cylinders, or make better use, quickly, of the exhaust steam from the front engine at low back pressure.  This might be a suitable application for Holcroft-Anderson recompression, as the forward engine will be slightly biased for lower power, but the size of the condensers is enormous given the necessary cylinder power the locomotive will be developing.  Likely you'd try to run all the exhaust-steam auxiliaries off the forward exhaust, and even then there will be limits above which you'll have to throw away both mass and heat.  Only about 1/6 of it (max.) could be used to heat feedwater at any point before the injector, as you're probably already aware.

Vision becomes a critical issue here: the rear stacks make forward vision about as good as operating an F unit backward, and any Holcroft-Anderson condensers in the front would, for anything near the ground, do the same.  So you will have to use some sort of periscope or CCTV system for working forward vision ... this is an operating advantage anyway, and among other things makes night-vision operation much more optimal, but here it's almost a necessity.

  • Member since
    May, 2004
  • 3,244 posts
Posted by 7j43k on Sunday, February 26, 2017 1:39 PM

RME

 

The Q2 had all the cylinders at the front of the engines, which limits both the length and tuning problems.  Even then, there turned out to be severe problems that were, at the time, attributed to steam momentum.  You should read some of the test reports and determine the source and magnitude of that problem in your pipes.

The locomotive that bears inspection as to 'pipes' is the Q1, which ran one set of pipes very high and outboard, with some interesting curves -- note in particular the location of the diaphragm cover for the PRR-standard snifting valves on the rear cylinders.  There's a relatively high cost in developing all this plumbing compared to what is involved on the Q2 or T1 with all cylinders facing the rear (and heads easily accessible) where the admission steam and the exhaust plena have room and reasonably direct routing.

Six axles is already over the top for a rigid-frame engine; there's little point in not going to a good simple articulated with compliance control for the 'hinge', set up as Voyce Glaze and later Alco did with minimal vertical compliance in the lead engine so it equalizes like a rigid 12-coupled but hinges neatly in the middle.  Or, if you're a glutton for punishment, build an American Garratt...

I surely would like to read these reports.  Can you suggest a source?

I certainly do agree that the pipes to rear-set cylinders would be longer.  And that needs to be dealt with.  Further complexity of the run, for me, is sort of up in the air.  I can see that it is possible that it would be more complex.  But I am not sure it necessarily is.

Well, six axles is a lot, for sure.  But there's one of my faves:  the 88 4-12-2's that worked for the UP for about 35 years.  I figure my rigid wheelbase would be about the same.  I am increasing driver diameter by 3", but I am eliminating the 18" between the first and second axle on the UP.

I want to stay away from articulation, because then the individual engines would eliminate the possibility of reciprocating ballance.

 

 

 
I figure, at a minimum, there would have to be twin exhaust stacks, split front/rear.

 

There is little room to route the 'exhaust stack' at the rear, even if you use a full Franco-Crosti exchanger and air preheater (which would work well for oil firing, as its exit is right at the throat where the burners would be, but will only have concurrent flow).  You would not want to route it under the firepan, and it can't be placed as it was on the Triplexes because you (sensibly!) have no engine under the tender.  Your combined stack, nozzling, and steam flows in the rear will have to occupy a long, relatively thin duct going around the circumference of the chamber just ahead of the firebox legs; if you do that, I'd see if a Giesl ejector, perhaps with limited 'fanout', would work.  The historical examples like the British 9F used only one stack, probably a cost issue but also concerned with visibility; there's no objective reason you couldn't use a pair if you have the free dollars^3 to make it go.

There is then the issue of how you split the combustion gas to optimize draft contribution from the front cylinders, or make better use, quickly, of the exhaust steam from the front engine at low back pressure.  This might be a suitable application for Holcroft-Anderson recompression, as the forward engine will be slightly biased for lower power, but the size of the condensers is enormous given the necessary cylinder power the locomotive will be developing.  Likely you'd try to run all the exhaust-steam auxiliaries off the forward exhaust, and even then there will be limits above which you'll have to throw away both mass and heat.  Only about 1/6 of it (max.) could be used to heat feedwater at any point before the injector, as you're probably already aware.

Vision becomes a critical issue here: the rear stacks make forward vision about as good as operating an F unit backward, and any Holcroft-Anderson condensers in the front would, for anything near the ground, do the same.  So you will have to use some sort of periscope or CCTV system for working forward vision ... this is an operating advantage anyway, and among other things makes night-vision operation much more optimal, but here it's almost a necessity.

 

 

When I said twin stacks, I was thinking about the classic setup on the UP articulateds.  Both stacks would be in the "standard" location.

Your last paragraph does, indirectly, bring up the possibility of cab-forward operation.  It's a thought.

 

 

Ed

RME
  • Member since
    March, 2016
  • 1,243 posts
Posted by RME on Sunday, February 26, 2017 3:16 PM

7j43k
I surely would like to read these reports. Can you suggest a source?

They are available from a number of sources, I think including hathitrust.  Copies, and at least one analysis of the results, are in the T1 Trust file repository.  I think Joe Burgard posts here, and he can recapitulate some of the points he thought most important.

 

Further complexity of the run, for me, is sort of up in the air. I can see that it is possible that it would be more complex. But I am not sure it necessarily is.

Here is a picture from Bill's Pennsy Photos that shows some of the problem:

Note all the 'stuff' that the rear outside supply pipes have to navigate to get to position, and the fabricated bends to get piping with good flow smoothness that can be physically removed and installed for maintenance. 

The B&O locomotive did all this differently:

Here the valve cylinders are kept inboard, like early locomotives with piston valves and eccentric-driven valve-gear.  I do not have (but would like to see) a drawing that shows the pipe routing, and a description of how the rear valves were inspected and serviced when necessary.

An issue mentioned as significant with the forward-facing piston rods was that they were comparatively easy to 'nick' with bouncing ballast or coat with grit.  This is relatively easily fixed by giving them vented 'boots' like those on shock absorbers of 'mud trucks' with extreme lift kits.  There is also some accelerated wear to crosshead surfaces that would need to be addressed with good modern materials.

 

I figure my rigid wheelbase would be about the same. I am increasing driver diameter by 3", but I am eliminating the 18" between the first and second axle on the UP.

The problem isn't so much matching the wheelbase on the Nines as it is dealing with the pronounced overhang and polar moment of inertia with a design that has a larger and much heavier firebox and chamber structure.  Ideally there is a geometric formula for positioning the two axles and pivot point of a Bissel trailing truck (note sp.) so that it doesn't exert unbalanced lateral flange force when following curves relative to the rigid driver wheelbase.  In practice the axle locations may not be, for weight-distribution reasons, where this formula puts them (and Timken for example designed for this by providing lateral motion for the first trailing-truck axle via hardened steel rollers and wear plates).  I would be concerned about how you effectively control the long lever arm and hence fairly extreme overhang at the rear of the locomotive chassis in particular.

 

I want to stay away from articulation, because then the individual engines would eliminate the possibility of reciprocating balance.

In a sense, the engines of any Mallet type are 'duplex' in being physically separate.  What I think you mean is that the hinged engine can't be given effectively low overbalance because of relatively uncontrolled yaw -- that was certainly true of older articulateds that hinged in both horizontal and vertical planes and did not have compliance on the swing of the forward engine.  It is less true of the post-'30s designs, and of course the tendency with any long, heavy modern North American engine is to restrain yaw and its couples with inertia; the Mallet chassis itself being relatively insensitive to induced yaw from low or zero overbalance in the rear engine, the engineering comes to apply to the hinged engine.

Here the fun starts if you want to perform Deem conjugation - of course this can be done right across the hinged joint, if necessary by using a double pin and yoke arrangement similar to a Cardan-shaft universal end or clevis.  Then you use the methods Glaze used to stabilize the forward end of the J class, both with stiffer and more positive lateral on the pivot of the lead truck and with spring and dashpot arrangements to damp fast accelerations in front-engine swing without resisting necessary swing on curves.  It is not difficult even with '40s technologies to make this adaptive to a significant degree, or actually make it servo-assisted. 

I figure, at a minimum, there would have to be twin exhaust stacks, split front/rear [...] When I said twin stacks, I was thinking about the classic setup on the UP articulateds.  Both stacks would be in the "standard" location.

Note that many of the 800s, including 844, have two-stack arrangements for better draft (note the arrangements for the nozzles relative to the exhaust passages in the cylinder saddle).  There is no direct requirement for this; N&W for instance successfully provided a kind of 'rosepetal' nozzle in which the openings for exhaust steam from one engine alternated with the other's.  Then you would just use the appropriate larger stack diameter and proportions needed for actual draft-producing mass flow.

Of course I'm not going to tell you NOT to use a twin-stack arrangement, just to proportion things so they work at least cost and most optimal back pressure.  I don't really know of any 'side-by-side' arrangements that actually worked well (one British locomotive used triple thin stacks, but that was for making the engine harder to follow during air raids) but you might be able to keep the smokebox length controlled by using parallel Giesl ejectors and some internal baffling to keep the flow patterns good.

 

Your last paragraph does, indirectly, bring up the possibility of cab-forward operation. It's a thought.

I actually can't take any credit for that.  ATSF of course is famous for its 6-4-4-4 oil-burning duplex design, which was to have been cab-forward, and the German 05 003 in particular was built cab-forward with coal firing (admittedly pulverized).  A concern I have is that operating a twelve-coupled engine stern-first puts an awful amount of mass on an awfully long lever arm out ahead of the lead and second driver pairs, where it might not be controlled very well by a pin-guided truck of the usual cab-forward style.

  • Member since
    May, 2004
  • 3,244 posts
Posted by 7j43k on Sunday, February 26, 2017 5:43 PM

Looking at that excellent photo of the Q1:  A lot of that "stuff" wouldn't be there in my design.  I think.  The valve gear hangers would go.  Or, at least, be much smaller.  That big air tank is a wonder.  It looks like they snaked the steam pipe around it.  I think I would have run the pipe straight and placed the air tank differently.  I also see what looks like an air-cooler.  Which can also be placed about anywhere.

I am thinking of having a shallow firebox.  Say, for example, similar to GN's N-3 2-8-8-0's.  There, the firebox is entirely above the rear two drivers.  Using 70" drivers will help out there.  If I do that, then the big-ass overhang off the back can be minimized.  This being a high-speed locomotive, some sort of trailing truck will HAVE to be there; but it won't be designed around being under the firebox.

I do wonder if the exhaust return from the rear cylinders could come forward in a single pipe.

 

Ed

RME
  • Member since
    March, 2016
  • 1,243 posts
Posted by RME on Monday, February 27, 2017 4:10 AM

7j43k
A lot of that "stuff" wouldn't be there in my design.

Remember that the 'stuff' is where it is at least partially for weight distribution (which in those days was tediously calculated inch by inch, both longitudinally and transversely).  It is relatively easy to forget this when doing 'crayonista' revisions of physical systems on a locomotive design.  Don't you be one of those!

I am thinking of having a shallow firebox.  Say, for example, similar to GN's N-3 2-8-8-0's.  There, the firebox is entirely above the rear two drivers.

Much better illustrative example: Jabelmann's late 4-6-6-4s, or the "Big Boys" -- for that matter, just about any Challenger design (which of necessity put the firebox forward end there).

However, it's very well established that deep firebox legs and a low grate are tremendously more effective than even a large area of grate 'up high' - this being a principle of the Q2, the Allegheny, and almost any of the Garratts.  With oil firing you have less difficulty in arranging effective combustion space and flame path, so a long fairly shallow firebox space could be made to work, but the radiant uptake area is going to be more limited, and the area of potential flame impingement, cooling, and consequent sooting may be higher.

Remember that whether or not you're using a Chapelon/Porta 'sectional boiler' the tube length is only going to be about 19 - 20' max, so at least some of the firebox tinkering could be extended even further forward, with a longer reflex path provided with a long refractory arch with circulating arch tubes substituting for some of the nominal combustion-chamber area, and careful control over the diffusion of combustion plume from the firebox exit to the rear tubeplate.  That would also reduce some of the big-ass overhang further.

This might be a place to experiment with defined waterwall forced circulation, as in a Lamont boiler, rather than just multiple circulators in welded-staybolted legs as in a Cunningham circulator.  You might be able to package multiple centrifugal separators ahead of the firebox space to get the necessary quick stripping and defined separation for the 6x or so circulating mass flow involved.

If I remember correctly (and I may be confusing things) the East Germans tested a Lamont boiler on a locomotive in the '50s, obviously with hindsight having problems that kept it out of practical further development.  I believe there is some discussion of this in European circles, in European languages, but I haven't actually found or tried to read any. 

This being a high-speed locomotive, some sort of trailing truck will HAVE to be there; but it won't be designed around being under the firebox.

Well, it will still be 'under the firebox', just not directly under the grate and having to contain the ashpan.  The principle is to pivot the trailing truck as reasonably far forward as you can, and put the transverse restoring-force means as far to the rear, arranged 'radially' like the guides in a Cartazzi axle, as you can.  That maximizes control of 'steering' the rear of the chassis dramatically, and simplifies what the tender or its lead truck then has to do in that respect on a modern design

 

I do wonder if the exhaust return from the rear cylinders could come forward in a single pipe.

It can, and if I remember correctly, on the Emerson 4-4-4-4 it did.  There may be flow difficulties involved with this at high speed unless you have very careful shape to the ducting into that single pipe at the 'cylinder' end -- remember that there will be alternating pulses from each side into the pipe from a quartered 2-cylinder engine, and the pulse frequency in the pipe will be 4x driver rps.

It may pay to look carefully at some of the special vaning and pipe flow shaping that was used on these duplex engines, particularly the provisions above the valves and the point where the steam-supply manifolds diverge to the engines on the T1.

Join our Community!

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

Trains free email newsletter
NEWS » PHOTOS » VIDEOS » HOT TOPICS & MORE
GET OUR WEEKLY NEWSLETTER DELIVERED TO YOUR INBOX
Connect with us
ON FACEBOOK AND TWITTER

Search the Community