Regenerative braking: basic-level questions

[Allen Hazen]

Please forgive me. I never took Electrical Engineering 101: I am a complete ignoramus! I have a couple of curiosity-questions; I am sure there are people in the Railroad.net "community" who can answer them in a minute: I'll be happy with pretty short answers.

My impression is that regenerative braking in a pure DC system (so: locomotive or powered car with DC motors, taking current from a DC overhead or third rail) is reasonably straightforward: it was used by the Milwaukee on its Pacific Extension electrification, so we are talking about early 20th C technology.

What about AC? For definiteness, I'll specify two particular historical periods.

(1) I'm not 100% confident, but I think I remember that the E-44 rectifier locomotives built by GE for the PRR in the early 1960s had dynamic brakes (as on many diesels: current generated in traction motors for braking used to heat grid which are cooled by fans) and NOT regenerative. Am I right in thinking that regenerative braking for a locomotive of this sort (i.e. DC traction motors, current from an AC overhead) would need fairly sophisticated (microprocessor controlled?) inverters to use regenerative braking, and that this technology was not available when they were built? Is it a technological possibility even now?

(2) "Classic" PRR electrification used AC motors: voltage reduced by transformers on the locomotives/m.u. cars, but motors running on the same single-phase 25 hz kind of current carried by the overhead. As far as I know, the PRR did NOT use regenerative braking, but is it feasible for this sort of system? If the PRR had succeeded in extending its electrification from Harrisburgh to Pittsburgh (making this economically feasible would perhaps involve MAJOR changes to the political and economic history of the 1930s and 1940s, but for the moment I'm only interested in the technological question!), would it have been possible to use regenerative braking there with 1940 technology?

(Questions occasioned now by a "what-if" question on the PRR forum, but topic of more general interest: I think I've wondered about it for years.)

[Nasadowsk]

) I'm not 100% confident, but I think I remember that the E-44 rectifier locomotives built by GE for the PRR in the early 1960s had dynamic brakes (as on many diesels: current generated in traction motors for braking used to heat grid which are cooled by fans) and NOT regenerative. Am I right in thinking that regenerative braking for a locomotive of this sort (i.e. DC traction motors, current from an AC overhead) would need fairly sophisticated (microprocessor controlled?) inverters to use regenerative braking, and that this technology was not available when they were built? Is it a technological possibility even now?

Correct - they had dynamic, but NOT regeneration into the catenary. You'd need inverters to do it. Though actually, today it's handled by the AC to DC converter, which is bidirectional and an active device on most locomotives. I.E., they don't use banks of diodes anymore.

(2) "Classic" PRR electrification used AC motors: voltage reduced by transformers on the locomotives/m.u. cars, but motors running on the same single-phase 25 hz kind of current carried by the overhead. As far as I know, the PRR did NOT use regenerative braking, but is it feasible for this sort of system?

Not really. At least, I'm not aware of anyone anywhere that ever did it. So the answer is likely 'no'. Tri phase systems with synchronous motors COULD regenerate, but were so limited in application that they're really a footnote in history.

If the PRR had succeeded in extending its electrification from Harrisburgh to Pittsburgh (making this economically feasible would perhaps involve MAJOR changes to the political and economic history of the 1930s and 1940s, but for the moment I'm only interested in the technological question!), would it have been possible to use regenerative braking there with 1940 technology?

No (see above).

IMHO, the enabler for Pittsburgh would have been the practical rectifier locomotive and commercial frequency electrification. That would have allowed 60hz catenary, which was FAR cheaper than the 25hz system, plus more efficient to boot. The PRR lost a good 10% to 20% of their power consumption right at the rotaries.

In short, had the E-44's technology come along, say in 1930, the PRR would have gone out to Pittsburgh. They must have had some interest in further electrification after the war, because they purchased a few experimentals from Westinghouse to test it out, though not on commercial frequency.

[Allen Hazen]

Nasadowsk--
Thank you!
As for the post-war W'house rectifier experimentals... Note that PRR also, in the same general time-frame, got a few AC-motored electric locomotives from GE: perhaps they wanted to compare the two types as a guide to future plans, and wanted to compare the rectifier locomotives against modern, rather than 10-to-15 year old AC, technology.

[Nasadowsk]

I suspect it was a 'test the waters' move, but the EL-C came around shortly afterwards and probably settled the issue as far as anyone was concerned...

I'm not aware of any AC powered MUs built after WWII that weren't rectifier units - the Washboards were and all the MP-85s were.

[timz]

In the 1950s Rwy Age mentioned that rectifier electrics had been considered unusable because of their interference with the lineside pole line. The article on the NH MUs mentioned that they had a capacitor and resistor in series across their transformer windings for this reason. (No details, tho.)

Would other ignitron electrics need that too? Don't recall any mention of it in the articles on the NH EP-5 and VGN rectifiers, or the earlier PRR rectifiers.

[Nasadowsk]

I'm surprised a pure rectifier unit would throw off that much EMI. The Westinghouse experimental locomotives, yes - they fired at weird phase angles (this per the maintenance manual), so they could do interesting things. But AFAIK, the Jets and the E-33s were pure rectifiers.

Today, most AC electrics have pretty involved filter circuits, and I've even heard of software modifications to avoid screwing up lineside equipment.

I could check the book on the Pioneer IIIs, but I don't recall it having much of a filter on it's transformer. Maybe New Haven's lineside stuff was more sensitive?

[Pneudyne]

Regenerative braking was certainly used from quite early times on electric locomotives working from AC supply systems.

One way was to use motor-generator conversion and DC traction motors, and this technique was sometimes chosen because of the ease with which full-scale regeneration could be achieved with good power factors. American post-WWII examples were the GN W-1 and VGN EL-2B.

In Switzerland regeneration using AC low-frequency (16⅔ Hz) single-phase commutator motors was commonplace in SBB locomotives and EMUs, going back to the early days and used until the last such series of locomotives, the 10 000 hp Re6/6 class of the early 1970s. There were several ways in which this was done. Keeping the TMs generating at 16⅔ Hz and keeping the power factor reasonable were evidently the major challenges. Such regeneration was also applied to a small number of the early French (SNCF) 50 Hz AC locomotives that were fitted with AC commutator motors. SNCF also had a small fleet of trial 50 Hz locomotives that used rotary equipment (including what was tantamount to a rotary transformer) to provide variable frequency, variable voltage three-phase AC to feed squirrel cage TMs, and these had full-range regenerative braking.

The advent of the rectifier era initially made regeneration more difficult, but it was done during the mercury-arc period. SNCF (those guys again) had some early rectifier locomotives fitted with regenerative brakes. These had excitron rectifiers which might have been easier to arrange for inversion than ignitrons. SNCF tried pretty much all approaches to 50 Hz AC locomotives in the 1950s, including ignitron rectifiers, excitron rectifiers, multi-anode rectifiers, 50 Hz motors (using different design approaches), AC-DC motor-generators and AC-AC conversion as mentioned above.

Nevertheless, even when silicon rectifiers and then thyristors arrived for AC-DC locomotives, dynamic (rheostatic) braking became the norm and regenerative braking was rare, but not unknown. This changed with the arrival of power electronics and three-phase motors.

Conversely, regenerative braking was the norm on DC-DC electric locomotives, with dynamic brakes only rarely found. And when fitted, they were often of the streetcar type, self-excited, sometimes for safety reasons. Separately excited dynamic brakes on DC-DC locomotives seem to have been quite scarce.

Cheers,

[Pneudyne]

I had forgotten that I had wanted to add to this thread at the time that it was current. I remembered recently when I was looking at some braking curves.

Looking at the question:

(2) "Classic" PRR electrification used AC motors: voltage reduced by transformers on the locomotives/m.u. cars, but motors running on the same single-phase 25 hz kind of current carried by the overhead. As far as I know, the PRR did NOT use regenerative braking, but is it feasible for this sort of system? If the PRR had succeeded in extending its electrification from Harrisburgh to Pittsburgh (making this economically feasible would perhaps involve MAJOR changes to the political and economic history of the 1930s and 1940s, but for the moment I'm only interested in the technological question!), would it have been possible to use regenerative braking there with 1940 technology?

The answer is yes, provided of course that the power supply system could handle regeneration, about which more later on.

The PRR would have had two workable choices in locomotive technology. One would have been to employ motor-generator (M-G) locomotives over the mountain section, following for example the GN precedent. The other would have been to regenerate directly from the single-phase traction motors, following for example long-established Swiss practice, which had evolved several approaches. These were applicable to 50 Hz as well as 16⅔ Hz, so there should have been no problem with 25 Hz systems.

I suspect that PRR would have opted for the second, although M-G locomotives might have had a role as helpers/pushers. M-G locomotive capacity was limited by that of the motor-generator set(s), so unless these were overspecified, as it were, then they did not have the same short-term reserve capacity as locomotives with single-phase motors. Illustration of the latter is provided by the PRR E2b case. Its continuous rating was 625 hp per motor at 26.5 mile/h, but it could accelerate trains up to 33 mile/h at 25% adhesion, corresponding to 1300 hp per motor, and haul tonnage trains at 41.5 mile/h over ruling gradients at 41.5 mile/h at 16% adhesion, corresponding to 1050 hp per axle.

Here is the regenerative braking curve for the VGN EL2B M-G electric locomotive:

VGN EL2B Regenerative Braking Curve.tif

Except for the truncated peak, it is of much the same form as a typical dynamic brake curve (non-extended range) for a diesel-electric locomotive with DC traction motors. The parabolic curve to the left and the hyperbolic curve to the right represent the motor limitations, at maximum field current for the parabola and maximum armature current for the hyperbola.

Those limiting elements for the motors are fleshed out in this curve for a diesel-electric locomotive, which is representative, albeit of quite modest size.

Diesel-Electric DB Curves.tif

The truncated peak in the EL2B case would, I imagine, represent the motor-generator set power absorption limitation when in regenerative mode.


I have also found a set of curves for the single-phase regenerative case:

AC Regenerative Braking Curves from Dover 4th, p.207.tif

These are differently presented, but they show braking effort is retained down to a standstill. This would stem in part from the availability of multiple transformer tappings, allowing progressive stepup of the diminishing voltage from the regenerating motors as speed dropped. I’d guess that only very short-term braking at low speeds could be tolerated without risking commutator damage though. Still, as an example, the SBB (Swiss) Ae6/6 locomotive AC regenerative brake was specified to be able to continuously hold back the same train weight, when running down the Gotthard, that it could haul up, and to have enough reserve capacity to brake the same train from full speed to nearly a standstill. I imagine that that kind of capability, appropriately scaled and ruggedized, would have appealed to the PRR.


Whether the power supply could handle regeneration would depend upon its nature. If say a 25 Hz single-phase system were supplied by frequency converter stations (using motor-alternators or perhaps rotary transformers) from a large 60 Hz grid, then there would not likely be any problems. The converters would work “in reverse” and the grid would be able to receive power. But if it were supplied from captive generating stations, then regeneration would be difficult, unless there was always a high probability of a balancing load from locomotives moving uphill. I guess that supply from a remnant industrial 25 Hz network would also have been a possibility, although that would have invoked concerns about single-phase loads on three-phase systems, such as were raised in the early days of industrial frequency electrification, but turned out to be mostly a non-event. Whether a 25 Hz industrial network could absorb regenerated power might vary case-by-case.

In hindsight one wonders whether GE did any work on regeneration during its development of the PRR E2b design. The relative ease with which regeneration, with high overload capacity, could be applied to single-phase motors could have been a selling point for that technology as compared with early mercury-arc rectifier locomotives. Whilst the latter could be, and sometimes were configured for regeneration, the relative rarity of such installations suggests that the “hassle” coefficient was of rather large magnitude.

Or perhaps the PRR saw dynamic braking rather than regenerative braking as best meeting its needs. In Switzerland, whereas the SBB had embraced regenerative braking, the BLS – with a much smaller network - had chosen dynamic braking. So there was not a universal panacea. A consequence of this divergence was that whereas the BLS adopted silicon rectifier locomotives for its Re4/4 in 1964, this being the next generation after its Ae4/4, the SBB stayed with single-phase motors for its next two generations, its Re4/4 II in 1964 and its Re6/6 in the early 1970s. (Although I have a vague notion that signals and telecoms interference/immunity was also a consideration in the SBB case.) Of course, in respect of the SBB case, 16⅔ Hz was a near-ideal frequency for single-phase motors, whereas 25 Hz was viewed as a compromise, chosen not because it was a good number for the purpose, but because it was there.

Cheers,

[Allen Hazen]

Pneudyne--
Thank you for those two very informative letters! (I have only JUST seen them, having been away from home and away from my computer for a stretch.) It'll take some re-readings and a lot of thinking before I have absorbed them, so I won't try to reply of comment now.

Except…
Re: "In hindsight one wonders whether GE did any work on regeneration during its development of the PRR E2b design."
--- Particularly since two of the E2 units were built to demonstrate on the Great Northern, and went there first before being added to the PRR fleet. … I recall, however, being told or reading that GE had originally wanted to build locomotives with DC motors (so: either motor-generator or rectifier) for the Pennsylvania, so competing directly with W'house's rectifier units, but that the PRR wanted to compare straight AC locomotives with AC/DC and insisted. So maybe the E2 was something built because a big customer insisted, and not something GE really had its heart in.

[Pneudyne]

Well, GE certainly seemed to like the M-G concept, so I’d expect that it would have tried to sell it to the PRR in the immediate post-WWII period. In the late 1960s, it offered the Milwaukee a DC-DC MG electric (based upon the U36C) as a lower cost alternative to a conventional DC electric. And its SL50E switcher (from the 1980s, I think) was of the DC-DCMG type.

Still, the apparent attention to detail in the design of the PRR E2b, particularly its motors and their low-speed control aspects, suggests that GE did expend quite a bit of effort on it, even if it had to be pushed by the PRR so to do.

(1) I'm not 100% confident, but I think I remember that the E-44 rectifier locomotives built by GE for the PRR in the early 1960s had dynamic brakes (as on many diesels: current generated in traction motors for braking used to heat grid which are cooled by fans) and NOT regenerative. Am I right in thinking that regenerative braking for a locomotive of this sort (i.e. DC traction motors, current from an AC overhead) would need fairly sophisticated (microprocessor controlled?) inverters to use regenerative braking, and that this technology was not available when they were built? Is it a technological possibility even now?(

Yes, the PRR E44 had dynamic braking, as did the VGN E33 before it.

As mentioned above, regenerative braking was possible with early rectifier locomotives, but was not much used, in part I think because of the complicated switching required to operate rectifiers as inverters. SNCF France did apply regenerative braking to some of its early rectifier locomotives in the 1950s, for which it used excitron mercury arc rectifiers rather than the ignitron type. I imagine that the fact that the excitron was permanently ignited (once switched on) rather than cyclically ignited like the ignitron made it a better choice for the inverting function. The excitron was probably developed in the USA (in the 1940s), but customizing it for traction applications took place in Europe and Japan. (It was also used for some non-regenerative locomotives.)

I happen to have the “French Railways Techniques” 1967 catalogue, which throws up some interesting points. By that time, SNCF’s non-regenerative AC locomotives had been built with silicon rectifiers for several years, and it was on the cusp of moving to thyristors. But it was still using excitrons for its AC regenerative locomotives, and anticipated that it would take some time to adapt thyristor technology for this purpose. That said, JNR Japan had thyristor-controlled regenerative AC-DC locomotives in service from 1968.

Cheers,

[Pneudyne]

Here is a set of curves for an AC-AC unit with DC-excited rheostatic (dynamic) braking.

AC Rheostatic Braking Curves from Dover 4th, p.187.tif

As might be expected, these show zero braking effort at zero speed, unlike the AC-AC regenerative case. One would expect the PRR E2b curves to have been broadly similar.

There were Swiss examples where AC rather than DC excitation was used for rheostatic braking, although I haven’t found any curves. In this case the braking effort at zero speed may have been non-zero.

In contrast to the AC-AC cases, here is an example of a conventional DC locomotive with regenerative braking.

DC Regenerative from Dover p.202.tif

As can be seen, in this case regeneration will brake the locomotive down to a fixed speed dependent upon the motor grouping in use. If the train speed attempts to drop below this, then without intervention or switching, the motors revert to motoring and restore the speed. This is simply a property of shunt machines, wherein below their nominal speed they are motors, and above it generators. Given that there was a minimum speed below which regeneration was not available, it was not unknown for regenerative conventional DC locomotives to also be fitted with rheostatic braking to cover the gap. By way of an example, the British Rail class 76 DC electric was retrofitted with rheostatic braking to supplement its regenerative braking. In the electronic age, blending of regenerative, rheostatic (dynamic) foundation brakes became readily possible. I don’t think I have ever come across regenerative brake blending from the electro-mechanical era. Blending of dynamic and air braking was done in the UK in the 1960s (86 and 50 classes), but I imagine that it would have been done in the USA before then.

Cheers,

[Pneudyne]

To complete the picture, here are two sets of curves for DC locomotives with rheostatic (dynamic) braking.

DC Rheostatic Separately Excited.tif

DC Rheostatic Self-Excited.tif

The first is a (fairly rare, I think) locomotive with a separately excited rheostatic brake. Each successive braking notch represents an increase in excitation level, the motors being loaded on to fixed resistances, in this case portions of the (force-ventilated) starting resistances. Although presented differently, the curves can be correlated with the form of the diesel-electric DB curves.

The second is for a locomotive with self-excited rheostatic braking of the streetcar type. Each successive braking notch represents a step reduction in resistance in the circuit, and with manual control, the braking controller needs to be steadily advanced to maintain the braking effort as the speed drops. Sometimes such locomotives have an interlock which inhibits electric braking above a maximum speed, beyond which motor armature currents would be excessive even with maximum resistance in circuit.

It might be said that the extended-range form of diesel-electric locomotive dynamic brake borrowed somewhat from self-excited rheostatic brake practice.

DB Extended Range.jpg

As may be seen from the above set of curves, the range extension derives from progressively reducing the resistance in the motor armature circuits at low speeds, this usually being done automatically by current relays.

By the way, the down-step at the high-speed end is not unusual. Conventional DC traction motors require a certain minimum field current-to armature current ratio for satisfactory commutation. If that cannot be maintained at high braking levels at the high speed end, where less field current is required to produce a given armature current, then limiting the amount of DB and hence the maximum armature current over that portion of the speed range becomes desirable.

Cheers,

[Pneudyne]

Although the primary subject here is electric regenerative braking, the original questions also referred to dynamic braking. And in practice it is difficult to discuss one but not the other. In respect of dynamic braking, perhaps I should be remiss not to include this:

Hydrodynamic Brake Curves.jpg

which is a hydrodynamic braking curve for a diesel-hydraulic locomotive. At the low-speed end the curve is a parabola, determined by the torque absorption capability of the braking coupling(s) at full fill. At the high speed end the curve is a (constant-power) hyperbola, determined by the power absorption capability of the hydraulic fluid cooling system. Coupling fill is progressively (and automatically) reduced as speed increases.

Cheers,

[Pneudyne]

Regarding the PRR and regeneration, the fact that the PRR bought and operated the ex-GN Y-1 class M-G locomotives indicates that its power supply system, or at least partof it, was suitable for regenerative locomotives. But the regenerative braking systems of yore were probably best employed for holding train speeds down long grades rather than decelerating them on the flat, so the utility to the PRR was probably limited. I imagine that the transition from power to braking involved something of a delay. Not so with more modern systems, though, Apparently the AC-DC (sepex) locomotives here in NZ can go very quickly from power to regen braking, much more quickly than (conventional) diesels can go from power to dynamic braking.

Cheers,

[Allen Hazen]

Re:
"the fact that the PRR bought and operated the ex-GN Y-1 class M-G locomotives indicates that its power supply system, or at least partof it, was suitable for regenerative locomotives"

Well, IF the PRR actually used the regenerative braking feature of the Y-1. The Y-1 were cheap (PRR was competing against scrap metal dealers for them when GN sold them!), and something of a stop-gap: PRR never expected them to last more than a few years. They were used in fairly low-speed service on low-grad lines. This is just a GUESS, but my GUESS is that PRR used them without using any non-air brakes.

(I will post a query about this to the Railroad.net Pennsylvania Railroad forum: somebody may actually KNOW.)