CN: 4-8-4 vs 4-8-2

In the string on Boston & Maine's R1 4-8-2 (which considers the question whether B&M would have been better off with a 4-8-4), there is a bit of a digression about Canadian National's 4-8-4 and 4-8-2 classes. I expressed a bit of surprise that CN went for a 4-8-2 (the 6060-6079 "Bullet Nosed Betty") in 1944 AFTER a long period of buying 4-8-4 (their final 4-8-4, the 6200-6264 series, were built in 1942-1944). Other people -- notably Pneudyne -- set me right by pointing out reasons they might have done this: go read that discussion before finishing this one.

What I only recently noticed (in Drury's "Guide to North American Steam Locomotives") is that CN had, by that time, had a lot of experience comparing the two types. Their earlier 4-8-4 (6100-6189) had been built in 1927-1930, and their earlier 4-8-2 (6000-6058 (with two built for subsidiary GTW)) were built in 1923-1930. So evidently in 1929 (?) they ordered BOTH types, having six years of experience with 4-8-2 and two with 4-8-4. So: they -- with a fair bit of supporting evidence! -- concluded that both types had a role, mainly in different services.

---

CN seems to have thought of its 4-8-2 as passenger power, with the 4-8-4 primarily used on freight (though both types could, in a pinch, fill in on either sort of train). My current guess (my current effort to fill in details of what others have tried to explain to me is this):
---With similar cylinders and drivers and boiler pressure, either type can accelerate a train from a stop.
---At high speeds, a passenger train doesn't need as much power as a 4-8-4 can produce: so a 4-8-2 will do the job.
---At medium speeds (high speeds for freight!), freight trains need more power: so a 4-8-4 is better.
But of course, even if this is right, it's only a first approximation. After all, many railroads (looking at you, B&M; looking at you New York Central) were happy to use big 4-8-2 on freight!

I suspect that it was mostly a matter of matching the locomotive power curve to the haulage and speed requirements.

If you have access to the book “A practical Evaluation of Railroad Motive Power” by P.W. Kiefer, then the various charts in it provide a good illustration. For example, the NYC L-4 4-8-2 may be compared with the S-1 4-8-4. Where the haulage and speed requirements were within L-4 capability, then that was probably a more economical solution than an S-1. I suspect that CN may have done the analysis for some of its passenger haulage requirements with the finding that an updated 4-8-2 would indeed be a good match for the job. In a relatively small fleet, standardizing on a dual-service 4-8-4 might have been prudent, but in a large fleet, such as CN had, there was probably minimal utility penalty incurred by also fielding a lighter 4-8-2 design, meaning that its lower operating costs could be taken advantage of.

If one looks at eight-coupled road locomotives as a group, then the reasonably feasible wheel arrangements were 2-8-2, 2-8-4, 4-8-2 and 4-8-4. The number of carrying axles used was partly determined by the power, and thus the boiler size required. Speed was another determinant. For locomotives that habitually operated above say 50 mile/h, the four wheel pilot truck was preferable. For a given weight on drivers, possible boiler sizes were smallest for the 2-8-2, somewhat bigger for the 4-8-2, a little bigger again for the 2-8-4, and with another increment to the 4-8-4. Driver sizes varied according to desired top speed and where one wanted to place the peak of the power curve relative to the modal operating speed. But empirically it would seem that with the 2-8-2, there was little to be gained by going beyond around 63 inches. On the other hand, the 2-8-4 justified going to 69 inches or thereabouts, but not beyond because the limitations of the two-wheel pilot truck. The 4-8-2, not being speed constrained in this way, could go to around 73-74 inches where desired. The 4-8-4 could go a step further to around 80 inches for the fastest requirements, although it was fine at around 69 and 73-74 inches where those met the speed requirements.

It would be interesting to see the power curves for the CN late 4-8-2 and 4-8-4 designs. I suspect that they would be closer together than was the case for the NYC L-4 and S-1. But there might have been some convergence at the upper end of the speed range, if say the final 4-8-2 had been specially configured to have a shallower droop down from the peak.


Cheers,

I have extracted some of the charts from the Kiefer book as general illustrations, below.

The L-4 4-8-2 was probably reasonably representative of the late 4-8-2 as a species, perhaps even a little better than average when its weight and grate size are taken into account. It was more powerful than the J-3 4-6-3, but its key advantage was probably less its greater power than its higher adhesive weight that allowed reliable starting and acceleration under pretty much all ambient conditions, and without the need for a booster.

To my eye anyway, the charts seem to show that if you wanted, and could justify on fleet utilization grounds, a fast passenger locomotive for low-grade routes, but you also wanted a boosterless design with more reliable adhesion than a 4-6-2 or 4-6-4 then a 4-8-2 was a logical and probably ideal choice. There would have been no need to go to a 4-8-4 unless the locomotive was also required for freight service, in which case the extra mid-range power would probably be useful.

Of course, in steeply graded territory, the extra power of the 4-8-4 would be worthwhile, particularly in the mid-range.

Also apparent from the charts is that for the NYC, the S-1 4-8-4 was really overkill – perhaps developed more as a prestige project than to meet a real need. (The design improvements that allowed higher monthly mileages, etc, could equally have been applied to a 4-8-2.) It did offer better acceleration, but that may have been a byproduct rather than a specific design objective design. In the event, NYC found that 4000 hp, two-unit diesels could do the heavy passenger job reliably in terms of point-to-point speeds.

If nothing else, the charts show that in terms of peak power, steam locomotives had to be significantly overpowered in order to have enough elsewhere along the speed curve. Nonetheless, recording very high peak power outputs seemed to be an obsession with some railroads, and even more so amongst observers. In that vein, kudos to the UP who didn’t seem to care much about such matters. Rather it designed its late steam locomotives to undertake, and to do well, their particular haulage tasks, with their power curves being a consequence, not a cause.

The 4-8-2 seems to have suffered somewhat, at least in the observer community, from not being a “superpower” locomotive, disqualified from that club by its lack of a four-wheel trailing truck. Yet it was the wheel arrangement of choice for some roads that had previously dabbled with the 2-8-4, including the B&M and the IC. And whilst the MoPac rebuilt some 4-8-2s in kind, its 2-8-4s were rebuilt as 4-8-4s.



Cheers,

And one more attachment.

I think that you're onto something. Allen. Maybe Northerns were overkill in many places. The New Haven was a road which took the Mountain to an extreme, by having three classes. They first got USRA, as well as post war clones, then the R@, and finally the three cylinder R#. As for the boiler capacity point brought up about the design not being Super Power, apparently they were able to keep up enough steam capacity to feed the three cylinders, as well as maintain the high speeds needed to keep up their time slots. This was very important on the Shoreline amongst the numerous passenger trains on the important New York to Boston route.

Based on the same information, I can see why my employer, the D&H went with Northerns. They were used in both passenger and freight on a hilly and curvy line. I see why something so powerful was needed to jump out of the numerous 25-30mph restrictions.

The D&H K-62 was probably a good example of a mid-sized 4-8-4. Although derived from the established Rock Island 4-8-4, it had some major as well as detailed changes, to the extent that one wonders how it got past the WPB stipulation that only existing designs be used. One could say that it was probably “half new”.

Compared to the RI original, it had a longer driving wheelbase, to accommodate 75” drivers, and a larger firebox, with 96 ft² grate area as compared with the 88 ft² of the original. Perhaps Alco has already done the redesign work, maybe in anticipation of an RI order, so was able to claim it that it could be treated as an existing design because it would not require significant extra drawing office hours.

The detail changes included, as might be expected from the D&H, a higher steam pressure (285 lbf/in² as compared with 250 lbf/in²), traded off against a smaller cylinder diameter (24½” as compared with 26”). It had the “full works” Alco lever-principle running gear, with graduated lateral resistance, and with lateral motion devices (LMDs) on the first three driving axles. In that aspect it followed the UP FEF-2; it had the same weight on drivers, 270 000 lb, as well. In detail generally it was intended to be homologous with the D&H J-95 4-6-6-4, which was an improved but dimensionally similar version of the UP original (of course with higher pressure and small diameter cylinders).

The D&H 4-8-4 design was then used for subsequent batches built under WPB aegis for the RI and the Milwaukee. The RI version reverted to the original 26” cylinder diameter, with 270 lbf/in² pressure and 74” drivers. It had an LMD on the leading driver only. The Milwaukee variant had 250 lbf/in² pressure and 260 000 lb on 74” drivers. I understand that it had more than one LMD, but beyond that the details are unknown.

Presumably the D&H had convinced the WPB that it needed a 4-8-4 rather than a 4-8-2. And that the available existing designs were either too big and heavy, with above 68 000 lb axle loading, and those that met this constraint were not quite powerful or fast enough. Hence the “half new” design. If the 4-8-2 option were in the mix, then of the existing 4-8-2 designs that might have fitted in axle loading terms, the B&M R-1 was probably a candidate. Also perhaps the 1927 MoPac design. This had basic boiler dimensions very close to those of the original RI 4-8-4, except for an 84 ft² rather than 88 ft² grate area, and I suspect was the origin of the RI boiler. It was bigger, I think, than the USRA Heavy 4-8-2 boiler. Earlier, the MoPac had developed a version of the USRA Light 4-8-2 with 73” drivers, and its running gear and cylinder dimensions, but with 250 lbf/in² boiler pressure, were then used for the 1927 “heavy” design.

Re the WPB, one could say that although it accepted what was a half-new design, it then achieved some measure of standardization by applying it across three roads, allowing the usual minor customization. The RI was probably happy to have this improved design, but it may well have been imposed upon the Milwaukee.

Curiously though, in the same time period, the WPB allowed the LV to acquire from Alco a slightly modernized version of its 1931 T-2 class 4-8-4, retaining the original key dimensions. On its face, that was quite reasonable. But a closer look at those dimensions shows that the T-2 was simply a clone of the 1929 RI design, with slightly higher (255 rather than 250 lbf/in²) boiler pressure. Thus the on the one hand the WPB was allowing more-or-less simultaneous production of the original and established RI 4-8-4 design and on the other hand it permitted the development of a “half-new” derivative.


Cheers,

W.r.t. the original Rock Island 4-8-4, you say "It was bigger, I think, than the USRA Heavy 4-8-2 boiler."
Yes.
The USRA heavy 4-8-2 only had 76 (and some fraction) square feet of grate, but even allowing for the difference in firebox size between a Mountain and a Northern, its boiler was smaller than that of the Rock Island engine: maximum outside diameter of 96 inches versus 98 on the Rock Island's Northern.(*)
--
The USRA Heavies were not BIG locomotives by the standards of the late 1920s and later: the aim was to have locomotives that could be used on much of the national railroad network, with a target weight of 60,000 pounds per driving axle: the Heavy 482 came in at 243,000 pound on the drivers.
--
(*) To my delight, I find that my "reference library" has descriptions of both types. I can provide more dimensions of each if people are interested.

Hi Allen:

Yes, by itself, that reference to the USRA heavy 4-8-2 does look somewhat out of context.

To expand somewhat, I was thinking in terms of the Robert Le Massena essay “The USRA Heavy 4-8-2 and its Illustrious Ascendants” (1). The claimed “ascendants” included various 4-8-2 and 4-8-4 designs right through to the WWII WPB era D&H, CRIP, LV group. And for example, the original RI 4-8-4 was described as having the USRA boiler.

That essay was painted with a very broad brush, though, perhaps overlooking some of the dimensional changes involved. BMOD went from 96” for the USRA 4-8-2 to 98” for the MP heavy 4-8-2, staying the same for the RI 4-8-4. But the DLW 4-8-4 had a 95” BMOD, which would appear to have been a non-trivial change, but presumably logical in the context of what its designers wanted to achieve.

Nonetheless, LeMassena presumably had access to information about how the various designs were worked out, including their starting points. A simple dimensional analysis, such as can be done from available information today, misses such factors.

That said, the USRA heavy 4-8-2 was amongst the largest of its era, although subsequently heavier examples were built, as shown in this 1926 table:

The LeMassena ascension analysis does show that there was not a quantum gap between the 4-8-2 and 4-8-4 types, something relevant to this thread more generally. It was more the case that locomotive size and power output was a continuous function overlaid on an axle count as a step function. Both the MP 4-8-2 and the RI 4-8-4 had a 98” BMOD; the respective grate areas were 84 and 88 ft², not that far apart. The grate area steps went from 76 ft² in the USRA 4-8-2 to 84 ft² in the MP 4-8-2 to 88 ft² in the RI 4-8-4 to 96 ft² in the D&H 4-8-4. It just happened that the axle count step occurred between the 84 and 88 ft² numbers, I think not far enough apart to have allowed an infusion of some superpower magic.

(1) The subject LeMassena essay was in RL&HS Newsletter, 2003 Spring.


Cheers,

Thank you for your reply, putting at USRA "Heavy" 4-8-2 into a broader context! You (and Le Massena) may be right in pointing to the USRA engine a a possible ancestor of several later types.
I don't know enough about the "culture" of locomotive design offices: maybe, at this late date, no one is in a position to know! My impression is that by the 1920s, American steam locomotive design was a fairly "mature" field. So I guess my default assumption has been that when, say, a new 4-8-2 design was needed, the designers would think "O.k., we know the relevant ratios (from the work of Cole and others), so let's see what we can do to meet the desire performance spec within the given weight specs." But of course, they might also have thought "O.k., the USRA Heavy is a good basic design: let's see if we can stretch, beef up, and generally tweak it to meet the requirements."
--
And thank you very much for the table from RME!!! It will repay poring over carefully... I will post comments if I think of any useful ones.
--
Umm. One, I suspect, significant proviso to my characterization of steam locomotive design as a mature field (as, to borrow Thomas Kuhn's terminology, "normal science" with an accepted "paradigm"). Some decades ago there was a fascinating article in "Trains," possibly by Le Massena, on the origins of "Super Power." Some time before the famous A-1 Berkshire prototype, Woodard, at Lima, designed a prototype of an improved heavy 2-8-2. Its overall dimensions were unremarkable, but he made one major change from the "paradigm": higher degree of superheat (and so more flues and fewer tubes in the boiler -- so probably a larger free gas area). The result was much higher performance! Enough of an improvement that the New York Central ordered a large number of copies as their H-10 class.
... So maybe I should do some arithmetic with the figures in the table to see whether and to what degree this idea was incorporated in the 4-8-2 of the 1920s.

That Trains article you are thinking of might be: “Why the H-10 was a Landmark Locomotive”, by Alan L. Bates, in the 1975 June issue. Let me know if you want a copy, and also of the LeMassena USRA 4-8-2 article.


Cheers,

Sounds like the right period for the article I remember. I've got a poorly organized bunch of old indues of "Trains": I'll look to see if I have that one (and if I don't, will write to you to take you upon your offer -- thanks in advance!).
--
Looking at the table of 4-8-2 from the 1926 "RME"...
First comment: the B&O T-class looks like a bit of an outlier: way heavier (and, at least as calculated, more powerful) than the rest. Backstory: it was built with the boiler from a 2-10-2: 2-10-2 types tended to be bigger and have bigger boilers than 4-8-2, so the size may just have been a consequence of this choice. History: Two T-class were built. They had reasonably long careers, and were regularly assigned to pull heavy passenger trains up steep grades, but B&O didn't get any more 4-8-2 for several years. One COULD try to read something into this, but given the general history of B&O motive power acquisitions in the late steam era -- they seem have been impecunious, had a penchant for semi-experimental locomotives built or re-built in their own shops, and were early converts to the advantages of diesels -- I won't try.

So, with reference to the "RME" comparative table and the idea that something of a revolution in American steam locomotive design came with the (Lima designed) New York Central H-10 Mikado and its greater superheater area, I said "So maybe I should do some arithmetic with the figures in the table to see whether and to what degree this idea was incorporated in the 4-8-2 of the 1920s."
Showing I hadn't studied the table yet!
"RME" did the arithmetic (for at least a first, crude, comparison) for us: the bottom line (except for notes) of the table tells us what the superheating surface of the boilers was as a percentage of the evaporative heating surface. The USRA Light and Heavy designs come out at 23.44% and 23.26% ... which (with one exception: the Santa Fe 3700 has 22.80%) are the lowest of the lot. The others, with one exception, range from 23.53% up to 28.54. (Suggesting that the idea that a bit more superheating was good had gotten around by the 1920s.) The exception: the New York Central's L-2 Mohawk, at ... 44.60%! (The footnote notes that the New York Central's design was apparently the only one to use the Type E superheater: my impression is that locomotives with Type E superheaters tend to have larger total superheating surfaces than those with Type A, and a higher proportion of "flues" to "tubes".) Suggesting that the New York Central, which had tested the H-10 against its own H-7 Mikado and gone on to order a large number of H-10(*), was convinced! (And -- Type E -- thought that the increased cross-sectional area for exhaust gas to get through the boiler was as important as the added superheat itself.) This is the sort of thinking that, 20 years later, produced the Niagara.
--
(*) The New York Central ended up with 301 H-1 Mikado and 300 L-2 Mohawk.