Gen-set Failures

[seppburgh2]

 

Not that ago, in the pages of Trains, the Gen-set was praised as the next big thing for moving to a greener tomorrow.  Well, tomorrow is here, and the Gen-set has not lived up to the hype.  Interesting given the advances in computer control (have you driven a Ford lately) this concept isn't working.  Now, the future is the battery-power can the Experts here chime in on the reasons why the Gen-sets failed?

Just so folks have a visual of a Gen-set, and commentary on their weak point, I'll direct you over to RailPicutres.net:

https://www.railpictures.net/photo/759522/

 

 

[SD70Dude]

In general terms, multi genset locomotives are far more complex, and complexity does not mix well with the railroad environment.  

I understand that there would be a delay between when power was requested and when a shut down engine would actually fire up.  This quirk would make these units very unpopular with crews.

 

 

[creepycrank]

The front office likes them because they can show that the railroad is going "green". The purchasing agent likes them because its a new locomotive that somebody else is paying for. The shop forces  like them because of all the overtime they get for keeping them running. The only group that does't like them are those that have  to operate them.

If they have to operate them in tandum doesn't that defeat the purpose of having the power on each unit divided between 3 diesels ?

[Overmod]

Remember that the dream of adjusting power through 'just the right' combination of engines has been a will o'the wisp of designers -- wacky and good -- for many years.  One of the express premises of the Baldwin Essl locomotive was that it could (more or less seamlessly, given the electrical connections) start and stop its 6000 nominal hp of power, in 750hp increments, depending on trailing load.  This was 'not dissimilar' from the design premise of at least some of the gensets that only the necessary combustion power would be running at any given time, something probably of more interest to bureaucrats in places like SCAQMD or CARB, or bean-counting people responsible for fuel contracts, than actual railroaders trying to Run Trains.

Of course, in the late '30s, things like short-term pollution at start or restart, or mandatory long spoolup times for engine acceleration, would have been less significant.  One thing about well-designed gensets was that they could share cooling loops, so if running on one engine, the other two could be kept at reasonable running temperature in the cooling jackets and engine structure and be more easily and less 'pollutingly' fired up; likewise the oil pressure on all the engines could be maintained, effectively as with a pre-lubing system, if any one engine were running, making it more practical to start, stop, and throttle them as "desired".

The great problem being that, as with the earlier generations of battery switcher, it appears that no one who understood how the locomotives would actually be used was involved in approving the final design.

One of the early comments I made to the Carnegie-Mellon 'integrated GIS' plan for autonomous fuel management was that it promised to make genset operation far more 'intuitive' to railroaders, as it could reasonably predict many types of increased power requirement and start and throttle-up engines as needed with less silly fribbling.  Likewise, for flat switching there needed to be some 'advance' control, ideally of great simplicity, that when pulled would start the engines cycling on for maximal programmed acceleration and alert the engineer when 'ready' to be quickly loaded down.  To my knowledge this was never implemented on the current range of gensets.

There are also potential issues with the source and design of the actual engines and 'gensets' used in these locomotives.  To the extent they are sourced on OTR truck engines, with the vast wealth of horror literature regarding the smog equipment and expedient design (ahem, early-2000s Caterpillars?) in some generations of them, we could expect to see trouble x3 in a number of respects.

There have been a couple of other stabs at select-a-power in a single unit, the Cat PR43 being a particularly interesting example.  This instead of having a buncha little prime movers had one big one, a C175 sized for 'average' use, and a little dozer-size C18 to be fired up either when higher peak horsepower or lower 'maintaining' horsepower was desired.  If the cooling and oil systems were not interconnected, someone missed a stitch... particularly if an 'idle' or shut-down engine did not have hydrodynamic lubrication against road shock, something that might not be obvious to an OTR guy.  This was an interesting and, to me, pretty-well-thought-out design -- but it does not seem to have caught on other than where PR30s were already in use.

In my opinion, a number of the design considerations used in the Budd RDCs (or the Essl locomotive) are (or ought to be) used on genset engines.  Having the engine installation designed for easy maintenance is one; having the whole genset 'modular' for easy replacement without keeping the locomotive itself out of service is another.  I of course would argue for using full SCR (with a little additional DEF to be paid for as for a fuel surcharge as an "environmental accommodation cost") for 100% of the NO reduction, thereby getting rid of the entire EGR boondoggle and perhaps the need for any sort of DPF kludge as well.  

A proper battery/supercap hybrid gets rid of much of the 'predictive' issue with the genset instantaneous power accommodation, provided that someone who actually understands this type of power transmission has done the detail design.  I am watching with great interest to see how the people in Fullerton manage to pull this off.

[Overmod]

This deserves its own discussion:

creepycrank

If they have to operate them in tandem doesn't that defeat the purpose of having the power on each unit divided between 3 diesels?

Not necessarily.  All that means is that 'peak power requirement' -- probably only representing a few minutes' or even seconds of actual engine-hourmeter time -- exceeds what three gensets can put out conveniently.  (I used to think that four or more engines run a bit less than peak nominal output produced more efficient power than three run at the equivalent of advertised WOT, but that is apparently no longer the case with modern engines, if indeed it actually ever was).

There can be little doubt that, if a single genset locomotive is more flexible than a large single medium-speed engine, then a pair of them is far more so if operated in the equivalent of MU.  I would expect it to be very rare that a given turn required the full power of two gensets substantially "all the time" rather than for particular parts of the profile or operating requirements ... with the lower combustion emissions, and perhaps 'wear and tear' on rotating machinery, etc., being lower at all other times the peak power wasn't needed -- although there might be sense in having a single efficient larger locomotive providing 'baseline' power from idle to Run 8 and the genset providing incremental power as with a more extensive version of the supplemental engine in the PR43.

I find I don't remember if the gensets can MU with conventional power.

[oltmannd]

Minor fuel savings at what cost?  The power generation of a turbocharged engine is nearly linear from notch to notch, once you account for the "idle burden".  So, the fuel savings is really the idle fuel - 2 to 3 gallons an hour.   That's a lot of complexity to pay for such a small benefit.  

If I was still in Equipment Engineering at the time these beasts came around, I'd have voted "NO!"  

 

(Also, wasn't a fan of the AC6000s and SD90s when they first were offered.  Too much new stuff in one box.  Was a fan of SD80MACs though...)

[CSSHEGEWISCH]

Perhaps we could use some input from Peter as to how the variable horsepower NR class is working out in Australia.

[Shadow the Cats owner]

If a genset is ever going to work in the railroad environment 3 then 3 systems must be combined on the entire set.  First off is the cooling system so that the entire power package is kept warm.  Second is the oiling system so there's always pressure in the non running engines.  Lastly reprogramming the software on the engines so that whenever the engine that was last started reaches 50 percent load the next one starts up and comes online automatically instead of waiting till 100 percent.  

[rdamon]

Sounds like a good use for the Cadillac 4/6/8 engine :)

[Overmod]

Shadow the Cats owner

Lastly reprogramming the software on the engines so that whenever the engine that was last started reaches 50 percent load the next one starts up and comes online automatically instead of waiting till 100 percent.  

That's a valid idea but doesn't do some of the things really needed in practice.

There is one other sneaky system I didn't mention, but that those of you with Mercedes and BMWs will perhaps ruefully appreciate: the intake-tract charge air has to be fully pressurized (and heated) at the time any of the engines is cranked.  That means that careful control over the running engine's boost, perhaps with careful use of heated brake air, needs to be used to provide each sequential additional engine with the right conditions before its own turbo has fully spooled up.  (In new practice you might use one of the electric turbo setups, or additional electrical spooling of an exhaust-driven turbocharger, to accomplish this -- another of the good reasons to incorporate hybrid-style energy storage in a locomotive consist...)

The loading isn't fully progressive in the sense of opening notches on a typical 'time-delay' GE throttle.  Now, if it were, your kind of proportioning would work at least as well ... but each engine will still have the same ramp-up to full power necessary for pollution control, which may be in the 20 to 30-second range.  If this is sequentially taken up, the response may be ridiculously glacial ... with the risk that as soon as engines have been fired up and enter their acceleration and then load-down, the locomotive may no longer need the horsepower, and the corresponding deceleration to save pollution then has to be implemented...

The key is, where possible, to predict where the incremental horsepower would be needed -- or provide a different engine control that is proportionally pulled up to tell the engines to go to a particular "horsepower" ASAP, with an indication when they have approached that.  I suspect that a crew 'kicking cars' would rapidly come to know the lags involved, and be able to predict when to run the engines up and down independent of the physical 'throttle' to accelerate and decelerate the locomotive.  

This is one of those things, like having a second or 'trim' throttle on an unconjugated duplex-drive steam locomotive, that seems obvious if you actually want to run a multiple-motored locomotive efficiently, especially within the potentially arbitrary desired bounds set by pollution-control agencies and the like.

[mvlandsw]

 

"I find I don't remember if the gensets can MU with conventional power".

 

 

The CSX one I used to use could MU with conventional power. ( I don't remember who built it). We always had to take another unit along on a 20 mile round trip because the Genset was so unreliable, mainly because of computer problems.

Probably more fuel was burned hauling it back and forth to the shop than it ever saved in operation.

Mark Vinski 

[Big Cat]

i think that Avon had two of them for use in the puller (Hump Bowl) operation and that the town of Avon provided some 'green' funding.  I was told that they didn't play well together so each one was always paired with a 6-axle.  Never heard that they left the yard.

[Shadow the Cats owner]

On modern OTR truck engines which are what genset locomotives use for powerplants.  These engines have no direct connection between the throttle and the injection system its all done by electronic signals.  So if say the engineer puts the engine in notch 3 the second engine of the 3 starts then at 5 the last one fires up for use.  It is all a matter of programming.  Heck the shop here can change a 450hp engine into a 550 one with a couple of key strokes.  Why it's all in the programming.  

[Overmod]

Shadow the Cats owner

These engines have no direct connection between the throttle and the injection system its all done by electronic signals.

I was not implying you would not program FADECs to coordinate the engines, only that more than a simple ramp at 50% would often be better considering the unavoidable control lags.

One likely difference is that OTR trucks necessarily have random throttle control; to have them 'load' the way, say, a GE FDL in a dash-9 does would be suicidal.  But for each engine type there will be an ideal map for 'rate of speed change' (both for acceleration and deceleration) that provides least pollution; in my opinion there is also benefit in transiently unloading the 'generator' as the Diesel engine is physically increasing its rotational speed.  Both these will benefit from anticipatory loading as you indicate, but I suspect in many cases you'd have many more starts and stops of engines over the course of a day -- this is less troublesome if you have tied the lube pressure together or use the Cat-style prelubing that uses the starter motor, and have preheated the engine with some combination of coolant circulation and oil heating, of course, but I suspect many railroads would still want to minimize the actual number of starts vs. equalizing out the number of engine-hours per prime mover even so.  That is not difficult to integrate into FADEC programming, just that it is more involved than just starting new engines at 50% load as a general rule.

To me the important thing is predicting when you need the additional power and accomplishing the necessary start(s) and engine runup(s) with minimum pollution in the time before the crew wants the engine to pull or physically accelerate a load.

It is interesting to consider how a battery-enabled hybrid control balances using the rate of discharge and charge of the battery vs. acceleration and regenerative braking.  There has been considerable work over the years in how hybrid automobiles 'built to a price' with the smallest possible battery capacity actively manage this -- and to me it's clear here, too, that prediction of loading greatly enhances the choice of use of the battery vs. combustion-engine progressive acceleration/deceleration  or sequential startup.

[Paul Milenkovic]

I recall taking the public tour of the Queen Mary berthed in Long Beach.  We were taken into the turbine gallery of the engineering space of the ship, and we were played either a recording or a dramatization of the communication between Bridge and Engineering on avoiding a collision.

A stuffy upper-class English voice demands, "We need steam!", presumably to have enough to manuever around the hazard, and the reply comes with an undecipherable Scot's accent to which Star Trek actor James Doohan never came close.  I would have laughed out loud were it not for not being rude in front of strangers.

Along the lines of a "trim throttle", the simple solution would be a big red button marked "We need steam!" for "kicking" cars or dragging a long cut.  The locomotive engineer could press it multiple times to get all of the gensets to power up.  After some time with the throttle at a reduced setting, the gensets could power down?

Don't some of the hybrid cars have a "sport" and an "eco" setting?

[Overmod]

Paul Milenkovic

Don't some of the hybrid cars have a "sport" and an "eco" setting?

The hybrids I've driven use the term "power mode".  And I get better mileage in power mode than I do with the little leaves flowering on the dashboard or whatever -- just as I do in cars with the 'eco' button, like Hyundais, by leaving it turned off, and in the '91 Jaguar XJ6 in 'sport mode'...

The problem with any mode button is that you and I both know it would be left depressed all the time --- or there would be some damn edict about discipline if you left it depressed, or the computer would automatically deactivate it the next time the vigilance control alerted.  Bureaucrats have ways to act when they think you're "circumventing the air pollution control" or whatever.

What I had in mind is like a ring you pull up, with a light that says it's been enabled and another light that says it's ready.  Probably some kind of preset for speed or power, depending on what you want the additional 'drive capability' for.   Think of it as a pollution-friendly hand throttle...

The predictive power increases can show up automagically, too, just as in the Carnegie-Mellon 'predictive cruise control'.  

If the RPS commuter 'accelerator' does not have many of these features, someone is asleep at the switch, so to speak...

Interesting idea with the multiple presses to get more engines in the spool.  I can see a display that works like that for heated seats -- one, two, and three bars of LED showing how many engines will cycle on following something like STCO's demand-following.  Just toggle a bit more to change that number as desired... again as with the heated seats.  And it could default to off as heated rear windows do, with an adjustable timer.

[Paul Milenkovic]

Now all of this presupposes that when you press the button or pull the ring enough times, you get all of those gensets powered up for a heavy pull.

From what people are saying about the gensets, that may not always be the case.

[SD70Dude]

I think the genset concept has turned out to be something like the EOT brake assist feature.  A good idea in theory, but it turns out to have problems that aren't worth the time and effort to fix in the face of better alternatives, like AutoStop and smaller, turbocharged engines replacing all those old worn out roots-blown EMDs. 

If you really want to save fuel and cut down on noise from yard engines, I think a modernized Green Goat would be a good idea to explore, especially considering how far battery technology has come in the last 15 years.

BTW, it doesn't really matter what type of defeat devices and interlocks you install to try and keep crews from messing with the units, we WILL find a way to get around them.  And if we don't, productivity will drop because the unit won't work properly, and every little defect we can find will get reported, as we will look for any excuse to get rid of the POS and get another unit that actually does the job. 

[Overmod]

SD70Dude

BTW, it doesn't really matter what type of defeat devices and interlocks you install to try and keep crews from messing with the units, we WILL find a way to get around them. 

Believe me... we know.  There is a long and classic art of misdirection in 'doing what the government wants' while providing back-door methods to get the actual job done.  Note how this subtly differs from the pennies-on-the-governor cheating the manufacturer: you can go to jail for tampering with "emissions" -- defined as anything the government can halfway establish had something to do with political perception of pollution.  And there is a fun thing in the United States called 'strict scrutiny' that you can get invoked on you if you're caught doing something the government deprecates...

The key is to make the convenient acceleration fully "green" according to sensible criteria, and then make operations relevant to the capability.  I would argue that had that been done with the Green Goat, 999, and other unsuccessful things, we'd have had a very different practical outcome from battery (or energy-storage) assisted power, particularly now that flat switching has become a hot and frequently-used thing again.  

What I'm worried about is that a new generation of 'SPV-2000 designers' will apply automotive design and algorithms to railroad service as though it scaled.  And you guys will have to get it in the neck, possibly for decades of misery, when it is relatively simple (at least in my opinion) to get many things very right.

I learned very early about the lesson of the Miller Train Control: if you're selling practical ATC, you don't sell it to bureaucrats, or publicity-hound safety departments, or penalizing middle management looking for more brownie point opportunities.  You sell it to the people it makes safe, and enables better.  And if it doesn't actually make them safe ... the thing needs redesign until it does.