Classic trains photos from the Denver Public Library

[Jones1945]

More photos you might find interesting:

Night scene during record run, about 57 hours from Los Angeles to New York, 5 cars. Photographed: at Cheyenne, Wyo., October 23, 1934.

https://digital.denverlibrary.org/digital/collection/p15330coll22/id/67624/rec/119

 

Train #17, Super Chief; first run; 6 cars, 75 MPH. Photographed:  east of La Junta, Colo., May 13, 1936.

https://digital.denverlibrary.org/digital/collection/p15330coll22/id/62310/rec/170

 

Train #17, Super Chief; first run with new stainless steel cars; 8 cars, 50 MPH. Photographed:  east of Trinidad, Colo., May 19, 1937.

https://digital.denverlibrary.org/digital/collection/p15330coll22/id/62311/rec/171

 

Denver Zephry, after record non-stop run from Chicago; 1017 miles in 12 hours 12 minutes, averaging 83 MPH; 8 cars, standing, night scene. Photographed: in Denver, Colo., October 23, 1936.

https://digital.denverlibrary.org/digital/collection/p15330coll22/id/62699/rec/183

 

Train #11, The Kansas Cityan; 12 cars. Photographed: west of Monica, Ill., June 23, 1946.

https://digital.denverlibrary.org/digital/collection/p15330coll22/id/55074/rec/240

 

Train #17, The Super Chief; 8 cars, 30 MPH.  Photographed: eastside Raton Pass, Wootton, Colo., May 19, 1937.

https://digital.denverlibrary.org/digital/collection/p15330coll22/id/58610/rec/321

[Overmod]

BaltACD

Suspect they could do Denver to Chicago without adding fuel, but not Chicago to Denver.

I was thinking, when I read that, that the route from Chicago to Denver might be better typified as 1035 miles -- 1034 across and one straight up.  But I got the strong impression that the idea was to go either way without a fuel stop... 

I have no ideas how either the S4as or the two oil-fired S4bs were actually fueled (or not) on that route; it would be interesting to know either way.

[Jones1945]

CB&Q really cared about the image of their premium trains. Both Aeolus locomotives were built as backup locomotives for their streamlined diesel-hauled Zephyr passenger trains, but still, they had a full-width diaphragm on the back end of the streamlined tender. Many RRs didn't consider it a problem when their streamliners were powered by a non-streamlined engine/unit/power/locomotive.

[rcdrye]

There were two CB&Q streamlined Hudsons, 4000 and 4001. Both were named Aeolus.  4000 was a rebuild, 4001 was built by West Burlington Shops using a Baldwin boiler and frame.  Both were class S-4a.  Known as "Big Alice the Goon" after a character in the Popeye comic strip (at least I think it was Popeye).  They had a reputation of being easy to service, with reported turnaround times in Chicago of a couple of hours between arrival from Denver and departure for the Twin Cities.

[Jones1945]

Overmod

The Q's Hudsons are some of the most underrated steam locomotives.  Something truly eye-opening are the 'before and after' augment numbers for the S4b reported in the Timken ad in the '47 Cyc.

It should also be remembered that one of the engines achieved 112.5mph on test, which is pretty good anywhere.  To put that in perspective, that's about 10mph faster than the ATSF's 3460s, with taller drivers and more advanced construction, could manage, and right up there with the best ATSF 4-8-4s.

Interesting! The speed record of hitting 112.5mph was exactly what the MILW Atlantic No. 1 also achieved during a test run on May 15, 1935, pulling (only) 5 cars. I wonder how many cars the CB&Q Hudson were hauling during the test. I thought ATSF got the "best" Hudsons in terms of top speed, followed by MILW and NYCRR...

Close-up view of (CB&Q #4004) drivers and roller bearing side rods. Photographed: Denver, Colo., June 29, 1939.

https://digital.denverlibrary.org/digital/collection/p15330coll22/id/42355

Train #49, Empire Builder, locomotive "Aeolus." Photographed: St.  Paul, Minn., July 26, 1940.

https://digital.denverlibrary.org/digital/collection/p15330coll22/id/46104/rec/26

 

[BaltACD]

Suspect they could do Denver to Chicago without adding fuel, but not Chicago to Denver.

[Overmod]

The Q's Hudsons are some of the most underrated steam locomotives.  Something truly eye-opening are the 'before and after' augment numbers for the S4b reported in the Timken ad in the '47 Cyc.

It should also be remembered that one of the engines achieved 112.5mph on test, which is pretty good anywhere.  To put that in perspective, that's about 10mph faster than the ATSF's 3460s, with taller drivers and more advanced construction, could manage, and right up there with the best ATSF 4-8-4s.

An interesting note is that the S4a class was supposed to be able to make Chicago-Denver, 1034 route miles, without recoaling.  This with 27 tons in the tender, which would appear to indicate that the NYC's much higher consumption for lower mileage (the best uncoaled-run figure I know being that for the C1a, with a Niagara boiler but lower water rate, requiring 64T in an extended pedestal tender to go 34-odd fewer miles with no engine change at Cleveland) was related to much higher acceleration and sustained speed.

Then there were the two oil-fired S4bs...

[M636C]

Overmod

 

 

M636C

Remember that ATSF couldn't run the "Super Chief" with even a string of 3750 class changing at division points.

 

 

There is no 3750 class; 3750 was last of a class of 4-8-2s, a fine passenger engine in its day but hopelessly incapable of Super Chief sustained speed.  The 3751 class 'as built' with 73" drivers was not much better; even ATSF referred to them as 'Heavy Mountains' when new and that was their design philosophy.  On the other hand, it was NOT long after introduction of the Super Chief that ATSF realized the importance of rebuilding these (as they started to do) in 1938 into true high-speed locomotives, fully competent as protection power in the same sense the 3765 (and 3776) classes could be.

I really should have remembered the class change number.

I remember stumbling across 3751 in San Bernadino one day some time before it returned to service....

On the other hand, it seems the CB&Q Hudsons were fairly capable of running the (presumably lighter and articulated) Denver Zephyr, but the CB&Q diesels were only good for 3000HP compared to 3600 for Amos and Andy and the E1 double units.

Peter

[Overmod]

M636C

Remember that ATSF couldn't run the "Super Chief" with even a string of 3750 class changing at division points.

There is no 3750 class; 3750 was last of a class of 4-8-2s, a fine passenger engine in its day but hopelessly incapable of Super Chief sustained speed.  The 3751 class 'as built' with 73" drivers was not much better; even ATSF referred to them as 'Heavy Mountains' when new and that was their design philosophy.  On the other hand, it was NOT long after introduction of the Super Chief that ATSF realized the importance of rebuilding these (as they started to do) in 1938 into true high-speed locomotives, fully competent as protection power in the same sense the 3765 (and 3776) classes could be.

[M636C]

Firstly, we should consider Otto Perry and his amazing photographic record of this period in particular. Clearly he was made aware of this problem with No 17 and went out to photograph it. These may be the only photographs of this event and they have been published before.

Remember that ATSF couldn't run the "Super Chief" with even a string of 3750 class changing at division points. It was good to have them in case of a failure of 1A and 1B but they couldn't meet the speed requirement to run the train regularly.

To expand on my comments in the Trains thread on EA51, notice that in these photos, apart from the truck shrouds being long gone, only 1A has a pilot, the other three pilots fitted when new having been removed. The additional radiators on the roof at the leading end of 1B are visible in these views.

Peter

[Overmod]

You have those pictures backward, and there's going to be more to the story.  

#17 is the westbound Super Chief.  It was brought west to Ja Junta, with only one of the 'one-spot twins' running... I'm trying to see if I can tell which from the smoke... using that 2-6-2 ... which indicates an engine failure on one of the diesel units; then the same train is shortened one car and merits first-stage 3751-class 4-8-4 power to snap it over the passes.  What this indicates to me is little if any diminution of the train's timekeeping in case of on-the-road engine trouble...

[Flintlock76]

Oh baby.  Oh bay-BEE!  Steam and smoke, just the way it's supposed to be!

What happened?  What the hell happened?  Is this a better world without those incredible machines?  I don't think so!  Well, as the song says:

"Don't it always seem to go, and you don't know what you've got 'till it's gone!"

[Jones1945]

Three more interesting photos from the archive:

Train #1, Denver Zephyr; 10 cars. Photographed: Chicago, Ill., August 11, 1939.

The Denver Zephyr pulling by a "non-streamlined" CB&Q Hudson.

https://digital.denverlibrary.org/digital/collection/p15330coll22/id/46106/rec/37

 Train #17, Super Chief; 7 cars, 60 MPH. Photographed:  west of La  Junta, Colo., January 6, 1937.

ATSF 3760 hauling the old, diesel-powered (ATSF #1A, #1B) heavyweight Super Chief

https://digital.denverlibrary.org/digital/collection/p15330coll22/id/44766/rec/4

 Train #17, The Super Chief; 8 cars, running late, diesel trouble.  Photographed:  east of La Junta, Colo., January 6, 1937.

AT&SF locomotive, engine number 1105 2-6-2 helping the heavyweight Super Chief.

https://digital.denverlibrary.org/digital/collection/p15330coll22/id/58529

 

[Overmod]

timz

Overmod says it would be "straightforward" to calculate the mean eff pressure in the 4-8-4's cylinder. (Hard to believe, but he did say that -- take a look.)

What I recall saying is that there is little point in calculating MEP over the stroke when it is the curve of effective pressure over the stroke that is the practical thing of interest -- as measured by piston thrust, not indicated pressure with its latency and compression errors, etc., but that's related to what I said about the point of measuring "MEP" to start with.  (I believe Tim referred somewhat disparagingly to formulae that use it, which I agree are oversimplistic.)

What else could the simulator have done wrong? If he assumed Davis-formula train resistance the acceleration would come out better than he said, assuming the GS3 was good for 2500+ dbhp at 75 mph. Only other possibility is wrong arithmetic, which sounds unlikely.

Absent seeing the actual modeling of the engine, which Jones1945 said he was trying to either obtain or recover, the discussion hinges on whether or not any careful physical modeling a la Hall or Flyte was even done.  I tend to agree it isn't a mistake in Davis assumptions or arithmetic, but again we'd have to see the actual model code to "know" anything.  Most of my points are that it's silly to assume that modeling the engine is impossible simply because gamers may have done it differently.   

...3/4 inch is a better guess.

I suspect the port length in the bushings, something that can be directly measured from drawings and which I suspect is longer than 3/4" regardless of width or percentage of circumference, is necessarily the thing that determines inlet duration -- the time perhaps being low because the valve is already moving at high speed when the steam edge begins to open the port, and of course is moving at high speed when the steam edge closes the port to steam.  Keep in mind that these are piston valves, not slide valves which minimize physical motion (for usually valid reasons) but try to make up by using Trick porting or massive parallel passages a la 'waffle grid' to get rapid opening flow and high mass flow at short cutoff, both of which are critical for achieving reciprocating-engine high speed.

Overmod seems to imagine someone could do it; whether they can or not, they're not going to.

I'm certainly not going to; the point is that the result can be 'close enough' to the real physics without going beyond things already well-established in more recent steam locomotive design.  

But no measurements?

You're the historian; surely you have copies of the T1-T4 test runs peaking at 152mph.  Those alone would be worked examples of how formula application was done; I don't see any reason why scaling laws won't reasonably approximate larger consists of the same equipment.  

Timz

Overmod

timz

The Davis formula says 2000 dbhp will pull an 18-car 1000-ton train at 75 mph on the level. After 1938, did anyone assemble such a train on a long stretch of level track, and try it?"

[The issue is, why would any result, whether replicating the 100mph test train and conditions or not, be different or 'more valid' if conducted in subsequent years than results from 1938?  Did something in the physics of reality change just before the War?]

The question is clear. What's the answer?

Again, you're the historian, but I'd strongly expect that the PRR T1 testing (880 tons at 100mph) would surely contain 'answers' completely relevant to the question you're actually asking, or a reasonable interpretation of it anyway.  

Naturally there have been dramatic improvements in vehicle-dynamics simulation since "1938" and it would not surprise me to find that someone, somewhere, has plugged in appropriate factors for an 18-car consist on assumed constant-resistance track and gotten results.  Again, though, the issue is 'can it be done' rather 'did someone have enough business case to produce results' (which then survived the collapse of any passenger business justifying that combination of speed and load) and, as you're fond of noting, 'we don't know' the latter... at least, I certainly don't, and I don't have the time to spend at the Hagley or other primary source or repository to try to find one.  The issue remains that it is not difficult to get meaningful characteristics for the consist, and I'd think this puts gross error squarely in the engine modeling until demonstrated otherwise.

[timz]

Here's what started this discussion:

 

"The [simulated GS-3] engine would slowly reach 80mph or above but it took another 15 mins running on level track. If I use the original e-specs, the engine would go even slower"

 

15 minutes from 68 to 80 mph, he said, with 16 cars. Actual time would be maybe half that, so the question is, where did the simulation go wrong? What wrong assumption did he use?

 

First guess: wrong power for the engine. Since then, this argument has been about how much the simulator can/does know about a GS-3's power. Overmod seems to imagine the simulator could know a lot, with the right help. Overmod says it would be "straightforward" to calculate the mean eff pressure in the 4-8-4's cylinder. (Hard to believe, but he did say that -- take a look.)

 

What else could the simulator have done wrong? If he assumed Davis-formula train resistance the acceleration would come out better than he said, assuming the GS3 was good for 2500+ dbhp at 75 mph. Only other possibility is wrong arithmetic, which sounds unlikely.

 

Overmod

timz

What fantasy. The valve opens -- 1/60 of a second later, it's open to its maximum (offhand guess, less than half an inch)

Even a cursory knowledge of long-lap long-travel valve design would give you a better 'offhand guess'

You're right -- 3/4 inch is a better guess.

 

timz

How does the temperature and pressure in the steam chest decrease during that time? What's the temperature and pressure in the cylinder when the valve closes? What's the pressure in the cylinder after the exhaust valve closes, and just before the intake valve opens? What's the pressure in the cylinder after the exhaust valve closes, and just before the intake valve opens?

No one knows, so no one can calculate the GS-3's power at 75 mph. Overmod seems to imagine someone could do it; whether they can or not, they're not going to.

 

Overmod

timz

Aside from the 1938 AAR tests, who in the US has measured passenger-train resistance in the last hundred years?

I myself have seen probably hundreds of calculations done...

But no measurements?

Overmod

timz

The Davis formula says 2000 dbhp will pull an 18-car 1000-ton train at 75 mph on the level. After 1938, did anyone assemble such a train on a long stretch of level track, and try it?

So you're saying what, that up to 1938 engineers were lying wholesale about... [etc]

The question is clear. What's the answer?

[Overmod]

timz

It wasn't simple in 1940 -- what has changed?

Well,aside from an enormous improvement in OTS measurement technology, computational resources and the algorithms to run on them, and the development of whole fields like CFD, and then of nondeterministic and 'fuzzy' versions of the techniques -- about 80 years of general engineering evolution.  Much of which, of course, wasn't applied to steam, but people like Bill Hall certainly didn't let that deter them.  I am quite sure by this point that you have no practical idea at all even about Wardale's FDCs.

 

What fantasy. The valve opens -- 1/60 of a second later, it's open to its maximum (offhand guess, less than half an inch)...

i see more abject ignorance here.  Even a cursory knowledge of long-lap long-travel valve design would give you a better 'offhand guess' and a study of things like trapezoidal porting would hint at the magnitudes of porting, unporting, and shrouding in valve-gear design.

...and it closes 1/60 second after that.

The issue, as you really ought to know, is related to the actual admission mass flow.  A couple of the amusing problems with various approaches to poppet valves, particularly the 'advantage' in British Caprotti of having only effective 3 to 5% cutoff in either inlet or exhaust, are related to misinterpreting this.

What matters is primarily the mass of admitted steam, and its movement and quality change from the moment of unporting to the moment of cutoff.  Note that a couple of theories regarding compression affect the actual kinetics -- I was taught, for example, that 'ideal' compression should give roughly equivalent inlet-tract pressure (nominally in steam, with no air or significant gas content) right up through the ports net of any heat-transfer losses, so the effective steam flow at the valve opening does not wiredraw or show pressure latency.  In modern large engines this requires much more careful reversible compression control than on anything designed in 1940 -- there may be practical concern in doing this effectively, just as with Okadees or equivalent in 'lossy' compression limiting, but I think this stuff might as well be on the Moon given your current knowledge of what it needs to be used for.

How does the temperature and pressure in the steam chest decrease during that time?

Well, ideally, as little as possible.  Look at Chapelon's designs for 'jacketed'  primary flow, not only to preserve a large effective 'reservoir' at the valves but reduce wall losses.  Again you're measuring the wrong thing, for the wrong reasons, and calling the mistake 'meaningful'.

What's the temperature and pressure in the cylinder when the valve closes?

Hard to measure,meaningfully; what matters is the subsequent integral of measured piston thrust -- which arguably can be measured quite sensibly with pressure transduction between piston rod and cross head, reasonably isolated from rapid temperature effects on strain gages.  You modulate the valves to optimize that.  Who cares about the effect of instantaneous thermal transients other than in initial design and proportioning, e.g. to preclude gas cutting or reversible throttling losses in unavoidably pressure-cycled tracting?

A steam engine is a pressure engine, not a 'heat' engine, with respect to its horsepower production.  While entropy and similar concerns are important in assessing its economy, they are little direct connected with its output performance.  It can be fun to dance around this with semantics, and generations of engineers have done just that, but peculiarly, ever so much of the time the big performance savings don't seem to mirror the theoretical practice...  

What's the pressure in the cylinder after the exhaust valve closes, and just before the intake valve opens?

Would you explain what this statement is supposed to mean in the context of a piston-valve reciprocating locomotive?  

Aside from the 1938 AAR tests, who in the US has measured passenger-train resistance in the last hundred years?

I myself have seen probably hundreds of calculations done in the '70s alone, substantially following the Davis formula or only trivial refinements (or simplifications) to it.  The design of the Metroliner test fairings alone involved them, let alone the production 'versions'.  I don't think numerical simulations in things like ADAMS or VAMPIRE have completely supplanted the criteria in the Davis formula even to the extent it is not used in the algorithms for such programs 'at present' -- and I think you would be hard-pressed to argue differently given your evident disdain for the ability of simulations to model 'the phenomena' better than Davis purports to do.

The Davis formula says 2000 dbhp will pull an 18-car 1000-ton train at 75 mph on the level. After 1938, did anyone assemble such a train on a long stretch of level track, and try it?

So you're saying what, that up to 1938 engineers were lying wholesale about understanding the magnitude of train resistance and that we have to conduct full-scale tests to prove the methods?  That all the subsequent NACA/NASA research into scaling laws for resistance analysis is bogus, too, on your say-so?  Count me as from Missouri on those assertions, at least for now.

Like I said, we can't figure out where his simulation went wrong without knowing what train resistance he's assuming. Think we'll ever find out? 

Well, there is this little thing called 'contact the simulation developer and ask him what his assumptions were, and how he modeled things.  Yes, I think if you did that, you would know.  

[timz]

Overmod

it would be simple, straightforward, and not particularly expensive to DETERMINE it

It wasn't simple in 1940 -- what has changed?

Overmod

there are many people in that field who could model the required CFD for the superheated-steam path in a GS4 at high speed with comparative ease

What fantasy. The valve opens -- 1/60 of a second later, it's open to its maximum (offhand guess, less than half an inch) and it closes 1/60 second after that. How does the temperature and pressure in the steam chest decrease during that time? What's the temperature and pressure in the cylinder when the valve closes? What's the pressure in the cylinder after the exhaust valve closes, and just before the intake valve opens?

Overmod

I have never heard of [the Davis formula] deviating from 'conditional' physical results by anywhere near that percentage.

Where would you have heard of it? Aside from the 1938 AAR tests, who in the US has measured passenger-train resistance in the last hundred years?

The Davis formula says 2000 dbhp will pull an 18-car 1000-ton train at 75 mph on the level. After 1938, did anyone assemble such a train on a long stretch of level track, and try it?

Like I said, we can't figure out where his simulation went wrong without knowing what train resistance he's assuming. Think we'll ever find out? 

[Overmod]

timz

 

Overmod

... I could tell you this.

Indeed you could, if you knew it, which you don't.

You mistake my point, which is that it would be simple, straightforward, and not particularly expensive to DETERMINE it ... were there actual interest in precise numbers for things like EP variation in the stroke (far more important than 'MEP' in a steam rather than a gas engine, BTW.)

And no one is going to learn it.

That may be true.  I'm certainly more interested in modeling and testing something with actual high-speed potential, and my active interests do not, as Jones1945's do, include simulation modeling or rendering.

But you seem smugly sure that no one in that community can acquire the not-exactly-high-tech skills needed to recover reasonable data via proper modeling.  Perhaps you're right, but there are many people in that field who could model the required CFD for the superheated-steam path in a GS4 at high speed with comparative ease, and the necessary computer and software support is cheaper by the month.

We all agree the Davis formula is a "reasonable approximation" -- within 50%, we hope.

I have never heard of its deviating from 'conditional' physical results by anywhere near that percentage.  On the other hand, if you mean the real-world behavior with wind and weather, I'll grant you it's a crapshoot compared to easily-acquired continuous data acquired at the drawbar of a regular Daylight train -- which might be approximated by tinkering with historic equipment but I'll cheerfully grant you cannot be known even within meaningful limits.

If so, the simulation's wrong calculation is likely due to some other error, like wrong power for the engine.

It might be interesting just to crunch the historical numbers with a couple of proportional corrections within the other limitations of the simulator's 'physics engine' (I would suspect that most of a 'wrong result' would be here, but presumably using better approaches could fix that).

That historical 'good design' could colossally miss the boat is well known.  Good enough modeling can show you why an ATSF 3460'soerformance falls off a cliff between around 100 and 105mph, despite it being otherwise apparently capable of much higher speed; the really interesting thing a proper simulation might reveal is the shortcomings of the C&NW E-4 which not only couldn't reach 100mph with the (measured resistance) AAR test train but was something of a child film star in bouncing-driver home movies.

 

[/quote]

[timz]

Overmod

I could tell you this.

Indeed you could, if you knew it, which you don't. And no one is going to learn it.

We all agree the Davis formula is a "reasonable approximation" -- within 50%, we hope. So is that what his simulation is using? If so, the simulation's wrong calculation is likely due to some other error, like wrong power for the engine.

[Overmod]

timz

No one knows the MEP in the GS-4's cylinders, of course.

A few conversations with Doyle and Ray and a couple of hundred bucks on eBay and I could tell you this.  Not that involved to ballpark it with multiphysics either since only like .007" of the wall thickness actually thermally cycles at GS4 'high speed' so the losses vs. extraction are known along with the modeled steam flow through the valves each cycle.  We assume about 40% cutoff at full throttle opening as that is consistent within about 2-3% for a wide range of historical high-speed designs.

In the literature, contribution of machine friction in a modern engine is only something like 4% variance across tested prototypes, some of which were better and some of course worse than this reasonably-evolved Lima design, and this is a minor factor in determining service speed.

 

Are you suggesting that the Davis formula applied to a Daylight consist is somehow impossible or irrelevant in giving a reasonable approximation of train resistance?

[timz]

I naturally judge their data to be wrong, since there's no reason to hope it's right. The right data doesn't exist, and never did, far as we know.

If you imagine it's right, spell it out. A GS4 weighing how many tons, pulling how many cars, with how many axles, weighing how many tons, will accelerate from 50 to 60, to 70, to 80, to 90 mph in how many seconds, on the level?

While you're at it, give the assumed dbhp vs speed for the GS4, and the assumed train resistance.

[Jones1945]

timz

You say the simulator said a GS-4 took 15 minutes to accelerate 16 cars from 68 to 80 mph.

Nope, I was talking about GS-3 in the "Trainz simulator", not GS-4.

timz

You happily assume the simulator makers must have good data. No reason to hope for that, even if the good data existed, which it doesn't. Their wrong simulation of the GS-4 acceleration shows what their data is worth.

I didn't even post the acceleration data of the GS-4 in "Train Simulator 2020", how would you judge the accuracy of their data? 

[timz]

Overmod

simplistic modeling of ihp from a formula like PLAN...

For those who don't know what he means by PLAN -- it just means the obvious: if you know the mean effective pressure in the cylinder, you can calculate the indicated horsepower. That's just the definition of indicated horsepower.

No one knows the MEP in the GS-4's cylinders, of course.

[timz]

Jones1945

many data of the steam engine like the 4449 is not hard to obtain from various sources

Sure -- plenty of data. Just not the data you need to calculate a GS-4's acceleration.

You say the simulator said a GS-4 took 15 minutes to accelerate 16 cars from 68 to 80 mph. The actual time would be half that, or less. (FWIW, NY Central's 6023 accelerated 15 cars, 1005 tons from 0 to 75 mph in just over 5 minutes, on a downgrade averaging 0.02%.)

You happily assume the simulator makers must have good data. No reason to hope for that, even if the good data existed, which it doesn't. Their wrong simulation of the GS-4 acceleration shows what their data is worth.

[Jones1945]

I think we discussed that before in another train simulation thread thoroughly. The developer of Train Simulator 2020 did a lot of research before they sell a 3D train model at a price that is high enough to buy a completely stand-alone new video game, and yes, many data of the steam engine like the 4449 is not hard to obtain from various sources, including books and from the engine's owner of course. 

Regarding the "infuriating" example of S1, the 3D model was created by a fan 10 years ago, I only reskinned it and created an e-specs for the engine to run. Since I *still don't have enough data to correctly calculate and adjust the suspension of the S1 in Trainz's physic system, I never take the rocking side to side reaction seriously because I know it is unrealistic from day one. I keep the video online as a reference to show the engine's potential, even though she runs like a roller coaster at a very high speed in the sim. I can manipulate the figures so that the locomotive wouldn't rocking like that but what's the point? That is cheating! Hell no, sir.

If you guys interested in vehicle simulators, try SnowRunner, new car simulator:

[Overmod]

timz

How could the simulator developers know what power a GS-4 had at any speed?

That's pretty damn obvious.  Even simplistic modeling of ihp from a formula like PLAN and some common-sense approximation of machine friction and resistances from the Davis formula will get you a useful simulation; I don't think the multiphysics to model steam flow through the valves and ports (which would very likely be the principal 'choke' on GS4 cyclic just as it is for the 3460s) would be that more conceptually difficult to set up in CFD with the 'usual' assumptions.

Now I think you are on much firmer ground questioning the 'results' from a typical train-simulation version of a locomotive that hasn't been modeled with appropriate physics.  One infuriating (to me) example was a simulated version (I think provided by Jones1945) of the PRR S1 being run at 'highest balancing speed' where the only pretext of dynamic concerns with guiding or suspension appeared to be to make the rendering of the locomotive rock around from side to side more visibly with speed.  That stuff fits Tim Z's criticism far more squarely.

I do have to say that if we needed detail data to model a GS4 we could relatively easily obtain it from 4449, which is an easy source for pointcloud or other inspection for things like detail port and passage or burner design... and it wouldn't be difficult to obtain results from a modern approximation of a dynamometer car just about any time 4449 is fired up to run.

[timz]

How could the simulator developers know what power a GS-4 had at any speed?

You seem to imagine they had some source of info. It's unlikely that any source exists, and if it does, it's unlikely they saw it.

[Jones1945]

Overmod

rcdrye

133 km/h is about right.

Even I could get it faster than 83mph.

I won't doubt it, because I can't see why they cannot run faster than 80mph by looking at their specs. But there are some strange phenomenons I want to share. I created a new e-specs exclusively for the GS-3 (not 4) in Trainz (a train simulator) to see how fast the engine can go, the engine's acceleration dramatically slow down after she reached 68mph with 16 lightweight cars behind her ( https://www.youtube.com/watch?v=JthOGAy6ahU ). The engine would slowly reach 80mph or above but it took another 15 mins running on level track. If I use the original e-specs, the engine would go even slower.

Two different 3D artists created the GS-4 for Trainz, both engines could hardly reach 80mph pulling a standard Daylight consist. GS-4 can also be found on another train simulator, the TS2020, she also gave me the impression that she wasn't designed for high-speed operation. Many think TS2020 is much more accurate than Trainz for steam locomotives, UP's FEF can easily hit 90mph or above in TS2020, the Daylight engine couldn't. I can't tell why when looking at the specs on the paper (Both engines in the simulator were developed by the sim's developer but not by fans). It would be interesting to know how SP would have upgraded these engines if the tracks between LA and SF were improved to allow a much faster schedule.

[Overmod]

rcdrye

133 km/h is about right.

Even I could get it faster than 83mph.