Overmod If you have a better package for analyzing reciprocating steam locomotives...
OvermodCan you scan and post the corresponding Zeuner diagram
Percent of cutoff in full gear is DH divided by DE
To get DH you need the cosine of angle DOP
Angle DOP is angle DOY plus angle YOP
Cosine of angle DOY = (lap + lead) divided by half the valve travel
Cosine of angle YOP = lap divided by half the valve travel.
For the 6-4-4-6, angle DOY is 54.315 degrees, YOP is 60 degrees, and DH/DE is 0.706.
If you assume the piston and the valve are each moving with a sine-wave motion, the Zeuner diagram is the result (or the equivalent Reuleaux diagram) and the above is mathematically the same.
Timz
Thanks for these additional data; from another source (book by UK author Reed) I now have valve events as lap 1.875” (Cf RME), lead 0.32”, exhaust clearance 0.25” valve diameter 12 in. Unfortunately, this source does not give the exhaust dimensions, which are absolutely critical, and if he doesn't have them, it will be a long search!. It also says that the tube plate distance was 22’- very long indeed, which will I fear tend to reduce superheat (exhaust gases at exit below the superheated steam temperature). I suspect that at what I suppose were normal maximum steam rates of less than 60000lbs/hr superheat would already be compromised, because the specific combustion rate/sqft grate would be quite low, leading to low gas temperatures.
In my initial stab at the S1, as I wrote, I gave it the equivalent of a 9” diameter blastpipe, and assumed a discharge coefficient of 0.99 (i.e. a plain blastpipe), for no better reason than if the PRR wanted to exploit the full potential of the boiler, it would need something like this! The C&O H8, which had a similar grate size, had a blastpipe area equivalent to a 9.9” plain blastpipe, but its cross pieces would reduce to the discharge coefficient- maybe to 0.9 which would make it in effect equivalent to a 9.5” plain blastpipe, this allowing to produce the 8000+IHP needed for its reported peak DBHP of 7500; the Q2 had Kiesel star nozzles with an area equivalent to a plain blastpipe of 8.55”, which also allowed 8000IHP.
However, the first thing the exhaust system has to do is produce enough draught to produce steam, and it may well be that to do this, the S1 blastpipe would need to be less than 9”. As I wrote, this would have pretty dramatic consequences at high steam rates. The first line in the Table below shows the power estimate on the original assumption. The second shows what would happen at a more realistic 280 psi steam chest pressure, a blastpipe diameter equivalent to 8.5”, with a discharge coefficient of 0.9 (typical of PRR set ups of that era). There is a significant reduction in steam rate, power and efficiency. The third line shows that you can get back to the original steam rate by lengthening the cut off to 37.5%, but under these new assumptions power is 900HP lower at 7200. The third and fourth lines show the same exercise with an 8” blastpipe- now 1250 less HP.
So, knowing what the blastpipe dimensions were is absolutely crucial at these high steam rates- it wouldn’t matter that much at what I imagine would be the normal maximum steam rate of 60000lbs/hr. An additional complication is that the programme used for these calculations assumes that the blastpipe flow is the same as the cylinder flow. However, with a feedwater heater, around 10% of the steam gets recycled, meaning that back pressure will be significantly lower than the programme estimates. To get round this and calculate at the correct back pressure, you have to icheat and input a blastpipe diameter that is 0.5” larger than actual.
So, the third line in the Table gives a fair estimate of what the performance would be if the true blastpipe diameter was 8”.
Cut-off
Speed
St.cons
IHP
BackP
Effy
bp diam
discharge
SCP
%
mph
lb/hr
Psi
ins
coeff
psi
35
120
118182
8092
22.05
12.6
9
0.99
290
108363
6888
27.54
11.7
8.5
0.9
280
37.5
117988
7215
31.65
11.3
104401
6310
31.7
11.1
8
38.8
117777
6843
38.03
10.7
It seems that there is no data on the S1 in the Locomotive Cyclopedia, and if there is nothing on the exhaust system in RME, then the only hope is that someone has unearthed this for the PRRHS in Keystone. I do not have access to this.
In summary, I think we can estimate power reasonably accurately, but the figures I posted need updating with the new info, and crucially, the blastpipe dimensions if they can be located.
Thank you very much, Dreyfusshudson.
From my understanding, there was not even one article about PRR S1 #6100 on Keystone, not even one! This was one of the reasons why I joined this forum. I already gathered all the books and articles recommended by forum members on Classic Trains forum, except this one, recommended by Reed:
Redwards Some years ago Feltonhill had recommended the following article on the S1: "The S1's history was covered in a 7-page article by the late Charlie Meyer in the Jan 1992 (Vo.10, N0.1) issue of Milepost, a magazine published by Friends of the Railroad Museum of Pennsylvania. I believe they're still in existence, maybe out of Strasburg, and this is available as a back issue. It's well worth trying to get. It's probably the only detailed account written at this point." I managed to find a copy on eBay and as he states, it's the most detailed account I've seen on the S1. --Reed
Some years ago Feltonhill had recommended the following article on the S1:
"The S1's history was covered in a 7-page article by the late Charlie Meyer in the Jan 1992 (Vo.10, N0.1) issue of Milepost, a magazine published by Friends of the Railroad Museum of Pennsylvania. I believe they're still in existence, maybe out of Strasburg, and this is available as a back issue. It's well worth trying to get. It's probably the only detailed account written at this point."
I managed to find a copy on eBay and as he states, it's the most detailed account I've seen on the S1.
--Reed
On the other hand, in "HAGLEY DIGITAL ARCHIVES" ( https://digital.hagley.org/ ), you could find some non-copyrighted photos of S1 during its construction, showing its smokebox, boiler and firebox, I hope this will help!
Jones 3D Modeling Club https://www.youtube.com/Jones3DModelingClub
You can see the Kiesel star nozzle and blastpipe height in one of the pictures. I am presuming there are two of these (one for each engine) exhausting into the two stacks, as this was common practice in that general era. At minimum this gives a check on the cross-sectional area and effective entrainment coefficient that would be applicable, even if we don't have hard data on the flow streamlining of the upper front end.
I would confidently predict you will not find any superheater damper arrangements on this locomotive, and for any high mass flow through the front end the effective superheat on at least some of the elements would become dramatically higher than 'expected' -- that has certainly been the observation on some lesser locomotives, notably Greenbrier 614. Consider the subatmospheric TOF difference relative to effective gas-to-tube heat transfer for that 'extra' 2' or so of tube over 'correct' 40.6x ratio and tell me if you think heat transfer would result in measurably reduced gas temps in the front end at high speed.
I still cling to the 'entrainment' theory of complex nozzle geometry (including the practical effect of things like Goodfellow tips) vs. the theory that Jos. Koopmans has come to develop regarding multiple nozzles (the most efficient being multiple circular nozzles at appropriate angle and spacing to produce the effect of x long, thin conventionally-scaled chimneys). I don't think we will resolve this for Kiesel nozzles until multiphysics simulation is backed up with empirical observation ... with steam, in situ, not using some assumed approximation like blowing with straight air. I do think that we don't care about gas conditions at any particularly high 'economy' cutoff in terms of the present discussion, which implies operation at 'best power' at high cyclic, almost certainly within shouting distance of 40%, with corresponding mass flow and incident gas-plume kinetics (and relying on the very large boiler and its presumed high radiant heat uptake to source that mass flow ... there should be little difficulty assuring adequate superheat for it given the comparatively restricted expansion past significant initiation of nucleate condensation that would characterize very short cutoff operation. I am tempted to say we need to concentrate on the gas generation and physical coupling into steam generation (and then superheat) from the physical boiler, rather than doing theoretical calculations of ihp that inherently assume steam-chest pressure derating, even as slight as the ~30# mentioned.
Jones1945.
Thanks for the info on the PRR speed limits. I believe the NYC was also generally 80mph, with similar flexibility upwards if necessary?
Thanks also for the pictures of the S1, which confirm it had a double chimney with Kiesel star blastpipes. I did find another reference on the S1 here:
https://oldmachinepress.com/2018/07/05/pennsylvania-railroad-6-4-4-6-s1-locomotive/
It cites Reed, and I suspect leans heavily on him. It does not give blastpipe dimensions, but it does say that the chimney exits are 21” in diameter, 4.8 sqft in total. Now the Altoona report on the Q2 says that, with a similar set up, the chimney exits are 5.33 sqft, or 22” in diameter. The Q2 blastpipes had an area equivalent to 8.5” diameter, so I am thinking that the S1 blastpipes probably had somewhat more restricted area (no rigorous logic for this, just a hunch).
So, I have recomputed the power map for the S1 based on a blastpipe area equal to 8” diameter, allowing for the reduction in backpressure given by feedwater recycle in the manner described earlier. i.e. actually computing on the basis of 8.5"
This shows lines of power at different cut offs and speeds, and also the power developed at three different steam rates vs speed (the more horizontal lines). The lowest rate of 78000lbs/hr is 550lbs/sqft evaporation, the two higher rates 675 and 800lbs/sqft/hr.
What you can see ( or could if I could figure out how to insert the imaage!) from this is that at what is probably something like the maximum rate of steaming, 113000lbs/hr, you get about 7200ihp with this more restricted blastpipe configuration, which aligns with the quoted maximum power. At the low rate of steaming, well inside the boiler’s comfort zone, you are getting about 5500IHP. At what I suspect were service maximum needs of 3500-4000HP, you require but 15-20% cut off at speed- if the loco could be driven that way.
Now on the spreadsheet I am about to send you, I have analysed the running of a log with T1 5501+1060 US tons working from Crestline to Chicago, culled from the pages of the Journal of Made Up Data.
From Crestline to Fort Wayne, the T1 observes the 80mph speed limit, and slows to 70 mph restrictions across a number of diamonds and round the sharper bends. It gets to Fort Wayne in 102 minutes without exceeding 4000IHP, often less on easier stretches. The Broadway was allowed just over 2 hours, including the Fort Wayne stop.
From Fort Wayne onwards, the crew threw caution to the wind, and worked at 5000 IHP, (a still comfortable 60000lbs/hr) and left quite a few diamonds in worse shape than they were. The stretch from Larwill on is slightly downhill, and the 68.7 miles from Winona Lake to Valparaiso were covered in in 42.5 minutes, an average of 99 mph. After easy running in from Gary, they got to Englewood in 102 minutes vs the Broadway schedule of ca. 130mins.
The point of this fable is that a T1 was more than powerful enough to beat the Broadway schedule with a heavier train – hence the S1 was grossly overpowered!
Overmod:
Well the subjects of predicting superheat and draughting behaviour are worth extensive separate threads, but after much thought on these matters over the years, I have concluded that I don’t really have much to say.
My understanding is that the first principles CFD sums to predict superheat from a given set up are very hard if not impossible. What is true is that for shorter boilers, flue gas exit temperature exceeds superheat temperature, but for boilers longer than about 15’ in the UK, it is less. For the 21’ Q2, steam temperature is always higher, for the 19’ K4 this is also true at low rates, but at higher rates, steam temperature is higher. Whether a low exit temperature of the flue gas can actually cool the steam in the elements (as I was suggesting) I do not really know. Some over here assume this to be the case, and it certainly looks like longer boilers deliver marginally less superheat than you might expect. But you would need expertise in heat transfer, Reynolds Numbers etc. to demonstrate this was a fact not a mirage- beyond me.
I stopped stressing about this some while ago however, because
a) The effect, if it does exist and small and
b) The effect of a small change in superheat on efficiency is small and
c) Having spent years trying to work out how to design for best thermodynamic efficiency- something all designers aimed for and their Boards wanted- I concluded that small efficiency differentials were completely irrelevant to the operators ( the folk who have to make things work) - they had much higher priorities. This is why I think that whilst the Franklin T1 is somewhat less efficient than it ought to be because of leakage, this would not figure at all in the operators assessment of how good the T1s were.
d) I think I can predict from number of elements and combustion rate/sqft what the superheat range of a given boiler is likely to be, within a band within which the differences do not matter in practical terms. Superheater area is not relevant. What might matter, as Altoona showed is how close the elements are taken to the firebox. Chapelon exploited this, as did Gresley, but maintenance costs persuade Gresley to back off.
On draughting, I have come to a view that designing a good draughting system from first principles is likely an intractable problem- not only are the sums on front end behaviour very hard, but what we are interested in is combustion, not draughting, and there no good data on the state of firebeds, nor models of how they work. And, the state of the firebed is usually determined by that most unpredictable of things, human behaviour. not to mention variation in the qulaity of coal. I have been meaning to argue this out logically this year, but have never got round to it, because, if I am right, it is a most unhelpful and depressing conclusion!
I don’t really understand what you are driving at in the text beginning ‘I do think that we don't care about gas conditions…..’ I’m not arguing that there is somehow a problem with S1 superheat, only that because of its low combustion rate/sqft in actual operating circumstances, it might not be as good as it could be- but as noted above, I don’t think this is likely to matter much at all.
What I do believe is that Alco, Baldwin and Lima got it as right as it matters in terms of efficiency, even though their theoretical understanding was limited- there is no golden upland of superior performance beyond what they achieved for reciprocating engiens with firetube boilers; and, they set standards for robustness and mechanical reliability that it is hard to believe. How could a steam locomotive run 25000miles/month? Or run from Kansas City to Los Angeles, arrive in the morning and set off back the same evening? An amazing triumph of human endeavour. Unfortunately, there were better solutions!
Dreyfusshudson Jones1945. Thanks for the info on the PRR speed limits. I believe the NYC was also generally 80mph, with similar flexibility upwards if necessary? Thanks also for the pictures of the S1, which confirm it had a double chimney with Kiesel star blastpipes. I did find another reference on the S1 here: https://oldmachinepress.com/2018/07/05/pennsylvania-railroad-6-4-4-6-s1-locomotive/
My pleasure, Dreyfusshudson! I am glad to know that you willing to keep updating the simulation of S1 and other engines!
Regarding the speed limit of NYCRR, I think this post is worth reading on this topic: http://cs.trains.com/trn/f/741/t/116632.aspx please check it out if you like.
I read the reference link you posted not long ago, and I believe you noted that many, but not all info in that article can be found in S1's Wiki page which is having a major update since last year. Tons of references were added in the Wiki Page with the help of some familiar username (wink). There are some brief data of S1’s performance in Wiki page with reliable sources and references in the “Service History” which was never published on the web before. These data might help you to decide how to adjust the parameter of S1 if necessary.
By the way, your pic doesn’t load so I can’t read your new calculation result. You could send the pic to me so that I can post it here or you can upload your pic to free photo hosting website like “imgur” and post the link in the forum! Looking forward to your new discovery!
There was a very extensive discussion in Classis Trains forum recently about PRR’s experiential steam engines. Many forum members agree that T1 was powerful enough to handle the official schedule and PRR S1 was an “overkill”. I suspect that S1 was specially built for the 1939/1940 World Fair to represent America railroad industry thus it was built unnecessarily large and powerful. But during World War II, her power and her speed did help PRR a lot to handle the wartime traffic. After the war, there was a rapid decline of passenger trains service thus S1 or even T1 didn’t have many things to play with.
By the way, I would like to know that If your simulation could simulate or show wheel slip during heavy train starting or high-speed wheel slip? Thank you very much!
Jones 1945
I wouldn’t know where to begin in setting up a ‘slip’ facility. At the most basic level, setting and adhesion factor of 4 says you are betting the coefficient of friction at the wheelrim will be at least 0.25 if you were to start at full maximum NTE i.e. longest possible cut off, full regulator. A loco with an adhesion factor of 5 could get away with 0.20. What the actual friction coefficient is is highly variable of course, e.g. in terminal areas where standing locos are dripping oil and water onto the tracks. Not least for this reason, I suspect full NTE was used on starting only when you had too e.g. when starting a heavy freight on the ruling gradient. At speed, wheel rim tractive effort (WRTE) is nothing like NTE even at full power, so things didn’t really ought to slip, but they do. Fluctuations in WRTE around the piston cycle might have something to do with it. In this country, there is a mainline stretch of 4 miles at 1.33% gradient. There are a fair number of steam specials each year, and crews like to have a go and get to the top in record time. Unfortunately there is a slight curve near the bottom (105mph restricted for modern traction) which has a flange lubricator right where you are hoping to put down maximum WRTE. Guess what happens.
There are complex mechanical factors involved also which may play a part in the T1’s propensity to slip. Quite beyond me.
So it often seems that slipping is a random event with no clear indication of what the cause is.
The real issue I would see is that, given the random nature of events, there is absolutely no data I’m aware of against which you can check the general validity of your model.
The former British Swiss built Brown Boveri gas turbine locomotive was rebuilt with a single powered axle replacing one of the former idler axles (it had been A1A-A1A). This axle was powered from an AC electric locomotive and was specifically taken up to wheelslip while investigating different rail surface and adhesion conditions.
Peter
Perhaps the 'simplest' form of slip determination is that proposed for T1 5550 on its 'roller rig'; although the contact patch is somewhat artificial (as for historical test plant operations) it still technically understates baseline adhesion conditions. The rollers are specifically suspended to allow simulation of cross-level instabilities, non-flange-bearing frog traversal, and most if not all the other issues that destabilize a four-coupled duplex.
Kiefer famously conducted high-speed rotational balance testing on a greased track, and interpreting some of his stated results made measurements relative to different degrees of induced track motion, which would be secondarily admissible in an analysis of inducing factors for high-speed slip. Although of course rotational testing at 'high speed' involving slip leaves out a great many factors involving the transients likely to induce many of the problems.
Does ANYONE have current access to the AAR movies showing driver lifting in augment?
The superheat issue is very important in North American practice (relative to British practice where much more attention was paid to consistent 'automatic action' across a wide speed range). Even the data Mr. Pawson produces shows detected excessive superheat in one of the T1 tests. Ross Rowland IIRC is the inventor of the term 'crazy high' superheat as observed on some of his trips with locomotives equipped with superheat gauges. I very strongly suspect excessive superheat to be one of the major causes of the valve seizure of N&W 610 that abruptly ended the PRR high-speed testing.
While I do think that there is an increased benefit (in retarding nucleate condensation) to "excessive" superheat, there is little doubt that it adds little relative 'elastic power' to the steam in terms of producing extra torque. But it DOES add considerable heat to parts of the valve system right at the time lubrication is most compromised by even slight additional overheating, and therefore the maintenance costs of the excessive superheat are by far the things of greatest importance. This was one of my theoretical concerns with performance of an unconjugated T1a at 'achievable high speeds' relative to a Franklin-valved B-2 alternative with correctly cast and sprung spools. (Yes, something similar to what Wardale used for valve 'cooling' might be applied to the valve stem gland surrounds if tribology there suffered, as it well might have, from high superheat...)
As a small peripheral note: the "letter" in a Franklin system refers to the valve drive, not the valve configuration itself. That is why, for instance, the RC system tried on the T1 was called "B-2"; type B was the three-valve rotary cam system, B-2 was the special modification with bridges used for B gear to correctly actuate the four valves built for the type A installation (and sketched in Kirchhof's little corporate heraldic 'shield'), which would have been much more difficult to modify to the 'more economical' setup on, say, ATSF 3752.
Overmod and M636C
Thanks for your comments.
On slipping, there was and still is a tremendous amount of research going into the subject of adhesion, and it will be most interesting to see what can be learned about the specific issues 5550 might face. All I was trying to say was that I don’t know how one could, other than by introducing a few arbitrary criteria, build slipping into a steam locomotive performance simulator, since so much of what actually occurred was seemingly random.
The subject of ‘crazy high’ superheat does indeed take us back to where we started, namely poppet valves. I correspond regularly with the chap who was at one time ‘minder’ of our big preserved Caprotti Pacific, Duke of Gloucester, which operates at high superheat (about 7500F at a high working rate of 30000lbs/hr- as did many of our latter day designs). Not surprisingly, he’s one of its biggest fans! He thinks that the primary benefit of poppet valves is not their supposed or actual efficiency benefit, but that they do not suffer from the high carbonisation found in high superheat PV designs, hence are less likely to suffer from the valve problems you allude to. He has first hand steam age experience of hammering out seized up PVs when locomotives came in for shopping at the local Works.
More generally on superheat, the Table below shows some calculations on the effect of varying superheat on a British Pacific- no reason to doubt the same trends would be experienced everywhere, this is very consistent with 1912 Altoona tests on a K2.
inlet steam, degF
cut off %
Speed,
MPH
steam, lbs/hr
efficiency, %
eff relative to 650 deg F
403
21
70
27673
1341
10.4
75%
450
26465
1375
10.9
79%
500
25350
1434
11.6
84%
550
23989
1482
12.3
89%
600
22890
1534
13.1
95%
650
21933
1577
13.8
100%
700
21140
1611
14.3
104%
750
20499
1627
14.6
106%
800
19934
1646
14.9
108%
Going from saturated steam to 600oF gives a 33% improvement in efficiency, this due to both the favourable thermodynamic effects of higher temperature, and gradual elimination of cylinder wall condensation- as also shown by Altoona in the same 1912 tests- also because backpressure is falling a little. Beyond that, there is only the thermodynamic benefit to be had, and decreasing returns begin to set in, so one might question a priori the wisdom of aiming for very high superheat, never mind the practical consequences associated with it. As you say, increasing superheat does not have much effect on power developed above about 650oF- the principal effect is on steam consumption.
Thank you very much, Dreyfusshudson. In case you missed it that in your post on Nov 21, there is a picture you posted within your message but it is not displayed on my end. It was between these two paragraphs:
"I have recomputed the power map for the S1 based on a blastpipe area equal to 8” diameter, allowing for the reduction in backpressure given by feedwater recycle in the manner described earlier. i.e. actually computing on the basis of 8.5"
This shows lines of power at different cut offs and speeds, and also the power developed at three different steam rates vs speed (the more horizontal lines). The lowest rate of 78000lbs/hr is 550lbs/sqft evaporation, the two higher rates 675 and 800lbs/sqft/hr."
Would you mind posting it again? Thanks a lot!
Regarding simulation of locomotive wheel slipping, I understand that the method you are using is different from the train simulators I am using. This is a video showing wheel slipping in another train simulator "Railworks Train Simulator":
I wonder if there is any way to combine your method with 3D train simulators so that it could show the wheel slipping problem of some locomotives with FA lower than 4. Examples like PRR S1 and T1s; N&W J Class, Santa Fe "Northern" 2900s etc. As you may note that 3D computer simulation is widely used nowadays, there are some other examples of 3D vehicles dynamics simulation available to general public. They are not designed to simulate railway vehicles but the principle and some formula used should be the same. Here is another simulator you may find interesting:
Simulator like this can simulate vehicle dynamics including suspension effect without using a very powerful computer. Some of these simulators were developed by a small group of people. Thanks!
Jones 1945.
I sent you the missing diagram by e:mail on the morning of the 21st- if you have not received it, send me a message and I will try again.
I don’t have the skills, nor really the interest to integrate what I have done with modern simulation software, though have worked a bit with some providers to enhance their programmes.
The work I have done was driven by the desire to be able to compute the power developed in accurate, detailed running logs of steam locomotive performance, to find out what they were capable of in practice. I think we now know enough about the resistances involved in the UK to make a fairly decent stab at the powers being developed.
There are thousands of such logs for steam age running available in this country- timing of trains was and still is a big hobby, and recorders were generally meticulously careful. It turns out that each type, when pressed, (often on late or overloaded trains) performed within a characteristic band. Especially for larger types, this is often well below what they could achieve when pushed to the absolute limit- as happened e.g. on the T1 plant and Niagara road tests.
This band is, I think what the operators would need to understand when planning the operation of the railway, not the extreme stunt outputs.
I would love to do the same thing for US locos, but as far as I am aware, the data needed does not exist- I know of less than 50 US logs, some very sketchy, with the good ones being collected by foreigners who were interested in that kind of thing! And, as I indicated earlier, exact resistances in the US, particularly for different kinds of coaches are a grey area.
The second thing I have been trying to do is to see if I can reproduce the drawbar economy (lbs coal/ dhp-hr) measured on UK tests, assuming locos were of, and driven to test plant standards. Mixed results here. Sometimes locos are better than I estimate- I think because those doing the tests were trying to prove how good they were for publicity purposes, and adjusted the test protocol to deliver the desired result. More often, the test results show inferior efficiency to what the test plant would indicate. I think there are a number of very plausible reasons for this, so the more general conclusion is that test plant data will overstate the efficiency found in practice, and efficiency will surely deteriorate between shopping.
My point is that neither power developed, nor economy found in practice can be defined with pinpoint accuracy, so you cannot create a simulator which will give you highly accurate rendition of what a locomotive could do, for you are dealing with is a ball park, not a point. I do not know how the commercial simulators work, but it may well be that, even if the analyses are not as well rooted in test data as mine, they will nonetheless get you into the right ball park.
I am sure all manufacturers of modern railway equipment will have highly advanced simulators of how they perform. Unfortunately, there were no computers in the first half of the 20th Century.
Sorry for my late reply, Dreyfusshudson! My email inbox is flooded with Christmas advertainment from different railroad model shops, but I found your email arrived safely with the data you promised. I would like to express my gratitude for your generosity and your high-precision calculation of my favorite steam engine! I will reply to your email separately asap!
It is interesting to know that operating data for different kinds of steam locomotive was carefully and systematically recorded in detail for operational use in the UK. This explant that why one of the train simulations which is targeted to UK railfans; seldom can make a steam locomotive of the States. They probably can’t find some important data of post-war steam engine of the States and they don’t want to ruin their reputation by making inaccurate train simulation using too many estimative data.
I am glad to know that you expressed your willingness to simulate more US locomotives, but I can understand that the difficulties. PRR S1 #6100 was an extreme case that I can hardly find the complete monthly average mileage of her (merely 5.5 years in service); such record could be found in various sources which only shown the data from 1940 to 1942; Let alone a set of complete data of her technical specification. So, what you have done to simulate PRR S1’s performance is truly amazing. Duplex cases like PRR S1 or T1 needed not only accurate data but also rich experience in steam locomotive simulating.
There were many rumors or stories about the performance of PRR S1 as well as her younger sisters the T1s, so my original goal was to find out what they really can do in normal circumstance and their rate of acceleration. I am glad that I already got what I wanted from your simulation. If there is a way to render their wheel slip problem would make the whole thing even more perfect, but It seems to me that It would require another type of effort. Thank you very much!
Dreyfusshudson · However, as a number of contemporary observers had suggested, analysis of the test plant results with modern techniques says: o The increase in engine efficiency on the test plant over the Standard K4 was entirely due to higher steam chest pressure and wider exhaust, not the Franklin valves. o The higher steam flow at a given cut off allowed by the Franklin valves ought to have given a further significant improvement in efficiency by reducing the very long cut offs needed to work at high steam rates. This does not appear, and analysis of the data says that, as with other poppet valve set ups of that era, steam leakage was occurring which negated any benefit that the lower cut off gave. o The enhanced efficiency benefit of 5399 at very high speed was again not due to the poppet valves, but rather that 5399 was operating at much lower back pressure, thus lowering the over compression that occurred at high speeds on the original K4. o In any event, 70000lbs/hr is quite impracticable in service from a 69 sqft grate. o By inference, the superior performance of 5399 in road testing was almost entirely due to the fact that the greater steam flow at a given cut off given by poppet valves allowed 5399 to be steamed at higher rates than the standard K4. A similar result could have been achieved by raising the boiler pressure to 250 psi, and indeed Altoona testing of a 250psi K5 Pacific showed it had superior efficiency to all K4s tested, but changes to the boiler design on the K5 made it a temperamental beast. o It is not possible for 5399 to have averaged 100mph for 25 miles on end as claimed, nor does the time course data support this claim. · As with 5399, analysis of the results says that the engine was not as efficient as it ought to have been. Using test results on the Walschaerts Q2 and K4 as a comparator, it can be shown that the Franklin valves on the T1 were also leaking to a degree which significantly reduced overall engine efficiency, negating the benefit of shorter cut off, so contrary to even recent commentary, did not provide an efficiency benefit, even at full boiler pressure. This coupled with the fact that four valves and cylinders will inevitably leak more than two means that the T1 was likely no more efficient than a PV 4-8-4 working in longer cut off, and under typical service conditions, quite possibly worse. · The combination of two radically new technologies, Duplex drive and Franklin valves led to a number of operational issues with the 50 T1s purchased on the back of the 6110 results. Development work was done to eliminate these problems, including a newly designed rotary cam Franklin B valve, but the T1s were made redundant before they were even built by the purchase of diesels, so it was all irrelevant. The leakage problem was not recognised, and whether it was a tractable one is not known, nor whether Franklin B solved this problem. · Overall, one can say that the misinterpretation of results on 5399 meant the speed and economy logic for poppet valves on the T1 was quite spurious. Further, leakage of the poppet valves, which paralleled earlier UK experience with both oscillating and rotary cam valves, had not been solved at that time. · More fundamentally, one may speculate that the creation of 5399 was a consequence of the failure of the K5, which could have matched 5399. This failure has been linked to deficiencies in its boiler design, consequent on the firmly held belief that, since greater evaporative capacity was needed the evaporative surface area of the K4 boiler had to be increased. Modern analysis says this is not so; heat transfer does not limit evaporation and power.
· However, as a number of contemporary observers had suggested, analysis of the test plant results with modern techniques says:
o The increase in engine efficiency on the test plant over the Standard K4 was entirely due to higher steam chest pressure and wider exhaust, not the Franklin valves.
o The higher steam flow at a given cut off allowed by the Franklin valves ought to have given a further significant improvement in efficiency by reducing the very long cut offs needed to work at high steam rates. This does not appear, and analysis of the data says that, as with other poppet valve set ups of that era, steam leakage was occurring which negated any benefit that the lower cut off gave.
o The enhanced efficiency benefit of 5399 at very high speed was again not due to the poppet valves, but rather that 5399 was operating at much lower back pressure, thus lowering the over compression that occurred at high speeds on the original K4.
o In any event, 70000lbs/hr is quite impracticable in service from a 69 sqft grate.
o By inference, the superior performance of 5399 in road testing was almost entirely due to the fact that the greater steam flow at a given cut off given by poppet valves allowed 5399 to be steamed at higher rates than the standard K4. A similar result could have been achieved by raising the boiler pressure to 250 psi, and indeed Altoona testing of a 250psi K5 Pacific showed it had superior efficiency to all K4s tested, but changes to the boiler design on the K5 made it a temperamental beast.
o It is not possible for 5399 to have averaged 100mph for 25 miles on end as claimed, nor does the time course data support this claim.
· As with 5399, analysis of the results says that the engine was not as efficient as it ought to have been. Using test results on the Walschaerts Q2 and K4 as a comparator, it can be shown that the Franklin valves on the T1 were also leaking to a degree which significantly reduced overall engine efficiency, negating the benefit of shorter cut off, so contrary to even recent commentary, did not provide an efficiency benefit, even at full boiler pressure. This coupled with the fact that four valves and cylinders will inevitably leak more than two means that the T1 was likely no more efficient than a PV 4-8-4 working in longer cut off, and under typical service conditions, quite possibly worse.
· The combination of two radically new technologies, Duplex drive and Franklin valves led to a number of operational issues with the 50 T1s purchased on the back of the 6110 results. Development work was done to eliminate these problems, including a newly designed rotary cam Franklin B valve, but the T1s were made redundant before they were even built by the purchase of diesels, so it was all irrelevant. The leakage problem was not recognised, and whether it was a tractable one is not known, nor whether Franklin B solved this problem.
· Overall, one can say that the misinterpretation of results on 5399 meant the speed and economy logic for poppet valves on the T1 was quite spurious. Further, leakage of the poppet valves, which paralleled earlier UK experience with both oscillating and rotary cam valves, had not been solved at that time.
· More fundamentally, one may speculate that the creation of 5399 was a consequence of the failure of the K5, which could have matched 5399. This failure has been linked to deficiencies in its boiler design, consequent on the firmly held belief that, since greater evaporative capacity was needed the evaporative surface area of the K4 boiler had to be increased. Modern analysis says this is not so; heat transfer does not limit evaporation and power.
Before we switched the focus to PRR S1 a few days ago, I note some meaningful discussions about PRR T1s and the Franklin poppet valve gear. I learned a lot from it. After rereading Dreyfusshudson's first post, I think I still have some questions about PRR #5399, T1 and Franklin's poppet valve gear.
Base on Dreyfusshudson's discoveries, does that mean if PRR #5399 never equipped the Franklin type A poppet valve gear, her enhanced performance would have been the same after she was rebuilt by Lima? Is it true that the poppet valve didn't contribute even 1% of her improved performance? If the answer is positive and the management of PRR noted this solid fact (but they didn't) before 1944, I think it is reasonable to assume that PRR wouldn't insist to use Franklin poppet gears on the 50 production T1s.
Another question is that if those 200 sets of Franklin type A poppet valve gears equipped on the 50 production T1s didn't have the steam leakage problem or it could be fixed easily, was it possible that T1's performance would have been better than a T1 using Walschaerts valve gear (like T1a)? Or the situation was the same as PRR #5399? (or even worse due to the problems of the poppet gear itself increased maintenance cost, and worsen the performance due to steam leakage)
In short, is it correct to say that the Franklin Type A poppet valve gear equipped on PRR K4s #5399 and 52 PRR T1s contributed absolutely nothing in terms of performance enhancement?
I really like the design of S1 and T1. I am glad to see that using Dreyfusshudson's simulator, S1's performance is very satisfactory in my book (Yes I know she was too large in reality), but T1s performance is unexpectedly dissatisfying; it is still a powerful and fast engine but not as good as I imagined. If the post-war decline of passenger service didn't occur in America, PRR would have needed to improve the design of T1 or simply jumped on the diesel train.
An advertisement of Franklin Type B rotary cam poppet valve gear from 1948:
As it turned out, PRR did jump on the diesel train, the first E7 A/B/A set was delivered while the T1's were still being delivered.
Jones45
I don’t think the PRR was looking for a superior K4, they were after something remarkable- the S1 had already given a hint of their ambition. The Franklin K4 was really just a test bed for a technology that was widely believed at that time to offer the chance of a step change in performance.
I think I made it clear that, in practice 5399 was more powerful than the conventional K4, because its Franklin valves allowed it to be steamed at a higher rate at speed. I also showed that the test plant version of 5399 could have been matched by a Walschaerts K4 with a wider blastpipe and higher steam chest pressure, if the comparison being made is power developed at equal steam rate (50000lbs/hr, report Table 2). However, the poppet valve version would, even allowing for higher leakage, be developing this power in shorter cut offs than the Walschaerts version, so a crew might well judge the Franklin version to be more powerful, even if it was using somewhat more steam.
I do think that if they’d fitted a 250 psi boiler to the K4 and tweaked it a bit they would have had something superior to both. Unfortunately the execution in the K5 didn't succeed, even though I think it was fundamentally a good idea.
As I showed, misinterpretation of the 5399 results led to its adoption for the T1. In the context of the K4, Franklin did offer benefit because the Walschaerts versions struggle to use the full potential of the grate at speed, and the higher steam flow of poppet valves help. This argument does not apply to the T1. Indeed, I would argue that the steam flow at a given cut off given by the combination of four cylinders and Franklin was excessive on the T1, hence the need to test at 170 psi boiler pressure at Altoona.
I do not think that the T1 performance is in any way disappointing against real needs- it had far more power potential than was ever required on the PRR. As I argued earlier, I do not think the fact that the Franklin version was not quite as efficient as it might have been had any bearing on its reputation at all. Would a T1a have been less trouble?- almost certainly. Any less powerful in practice? No. Of course against the extraordinarily high ambition of 6000HP it is short of the mark, but this was never needed.
In fact, at the risk of setting a hare running, I do wonder if there was ever a need for a four axle design purely for passengers in the east. The extra driver helps you start faster and climb mountains, run even heavier or faster trains if that’s the operating plan, or be a fast freight hauler as well. But as far as I know, the PRR only had 12 miles of serious mountain to deal with on its principal routes, and with 1000 ton trains and 80mph speed limits, a 4-6-4 can deliver the goods or rather, passengers perfectly well everywhere else. I’ve just ‘driven’ a CBQ S4+1000 tons up the west slopes to Gallitzin and it does just fine. With 1400 tons a J3 could get from Crestline to Fort Wayne in about 1’ 55”. So, maybe a fourth axle was only really needed for an anticipated future that never happened?
Was the high ambition set for the T1 ever realistic? My view is that as soon as the Sante Fe was running the Chief with diesels in 1936, the answer was ‘No’, ultimately for reasons nothing to do with diesels vs steam.
On a separate point, there was some discussion earlier about simulating slipping. Whilst looking for something else, I came across a Timken report on high speed slipping, which may be of interest-Railway Mechanical Engineer, April 1939, p132. They concluded Timken technology was terrific.
Thank you very much for your detailed response, Dreyfusshudson. After your thorough explanation, the whole picture is much clearer for me now. Due to my shaky English, I always missed some important points.
I agree with you that the 6000hp requirement was set too high for post-war traffic load, even though the PRR did set the bar lower: from S1's "100mph at 1200 tons" to T1's "100mph at 880 tons". I believe the reason behind it was that not long after the order of two T1 prototypes (6110 and 6111) was placed by the PRR in 1940, the United States Congress declared war on the Empire of Japan in response to that country's surprise attack on Pearl Harbor on December 7, 1941. The whole progress of the construction and the development of T1; as well as other duplexes like PRR Q1 and Q2 were during World War II.
When both freight and passenger traffic reached the historical high during the war. The "heroic" wartime traffic probably affected the management of PRR to not adjust the bar they set for their new steam engine projects since they didn't know how long the war would have lasted. The PRR needed engines that were capable of towing longer consist which was mixed with heavyweight and lightweight cars which could easily exceed 1000 tons at higher speed. They also needed freight engines like Q2 which was capable of higher speed and heavier loads than the J1s. The order of 50 production T1 was placed in Mar 1945, the month when more than 8,000 American troops crossed the Rhine River in Nazi Germany.
Once the war is over in 1945. US railroads' passenger service encountered a "Pendulum Effect"; declined rapidly within two years. The 6000hp duplexes had no scope for the exercise of their ability, they were like some big fishes in a small pond. PRR S1 and Q1 were withdrawn from service by 1947, other duplexes lasted merely 5 to 6 years longer.
Regarding the performance of T1, we have a long, thorough and interesting discussion on page 8 of the post I mentioned before ( http://cs.trains.com/ctr/f/3/t/271182.aspx?page=8 ). please feel free to take a look if you like. This is an extract from the "PRR Chronology 1944" :
Dec. 5, 1944 - PRR begins one month of tests with borrowed N&W Class J 4-8-4 No. 610 in freight and passenger service on Fort Wayne Division; tests made at request of VP-Western Region James M. Symes, who is not impressed by performance of T1's and Q2's; makes 2 freight and 12 passenger runs at speeds up to 94 MPH; less power than a T1 at speeds over 42.5 MPH but better acceleration. (Hirsimaki)
By the way, is the issue of Railway Mechanical Engineer, April 1939 available on web? Thank you very much!
FYI:
Full report: https://www.nber.org/chapters/c5625.pdf
Jones 1945:
Thanks for you further comments. I do have some reports on the comparative testing of the N&W J and PRR T1. RME is not avaialble on line as far as I am aware.
I thought it might be of general interest to add part of the visit report of RC Bond to several Canadian and US railroads in 1947. Bond was about to become #2 in the British Railways design hierarchy, and he and some others were sent to explore North American practice before the design of BR Standards began. On poppet valves,his report reads: (I have emboldened things that I don't recall having seen from US sourcesand added some comments of my own in italics)
(b) Poppet Valve Gear.
Bearing in mind the large power output at high speeds demanded .in normal service from many American locomotives, Poppet valve gears have been used to a surprisingly small extent. (So he was a believer in the (myth of) suitablility of poppet valves for high speed) Early experiments were not encouraging and I found there is still no great enthusiasm for this type of gear.
The performance of the oscillating cam gear on the first two Pennaylvania T.1 class locomotives was unsatisfactory and Maintenance costs have been heavy. Although recommendations-were made to fit rotary cam gear when further engines were built it was decided on commercial grounds that the oscillating cam gear should again be used. Continuous trouble with broken valves has been the outcome of this decision.
The valves were originally steel drop forgings but they have been replaced, without much improvement, with centrifugal castings. A number of engines is now being fitted with modified cams which will reduce the impact of the Valves on their seats by 50% in some cases and by 75% in others, in an attempt to avoid valve breakage. These modifications, if successful, will be applied to all engines of the class notwithstanding that the main object for which this type of gear was fitted, viz., attainment of maximum power output at very high speeds, will be considerably curtailed thereby. ( i.e. not surprisingly, reliability was the absolute priority)
The rotary cam gear fitted to one of the K4s 4-6-2 .engines has given better results but with it, too, there has-been some trouble with broken valves. Comparative tests carried out last year with this engine and two others of the same class with piston valves showed that little benefit from the point of view of power output can be expected from Poppet valve gear in comparison with the best design of piston valve, engines at speeds below 60 m.p.h. (I am aware of no record of these comparative tests, nor the exact design of Franklin B K4 3847, nor what 'best design' was) Based on their experience to date, the views of the Pennsylvania Railroad may be summed up as follows:
.(1) Under no circumstances will any more oscillating cam gears be fitted. (Interesting that in 1947 the PRR was still contemplating improvements to its steam fleet, even though diesels were arriving in force)
Poppet gears are justified only when maximum power output at very high speed is required. (as discussed earlier, the BR Hierarchy were still believers in poppet valves, hence their attempts to make Caprotti a reliable design and then implement it). They should certainly not be fitted to freight engines, particularly multi-cylinder locomotives, where there is no difficulty in providing piston valves of generous proportions in relation to cylinder-volume.(The T1 report makes celar that the reason that K4 5473 was tested in 1945 was because it had been fitted with 15" instead of 12" piston valves. It also had a single pass sine wave superheater, which had shown very significant benefits to Franklin valve 5399 in 1939. As such it should have been a significant improvement on the original K4. I have also just spotted that the likely reason that K4 5341 was tested at Altoona in 1937 was that it too was fitted with the sine wave superheater, and, it would appear, a direct comparison made with the normal A type. My sources here say that no report was ever issued on the 5341 tests, but this would be an extremely valuable set of results to get hold of for students of PRR design development. I am not sure how widely the single pass superheater was fitted to other K4s)
(3) The reduced coal and water consumption is not sufficient to balance the considerably higher maintenance costs involved with any type of Poppet valve gear, of which they have so far had experience.
The New York-Central, who have fitted rotary cam Poppet valve gear to the last of their 'Niagara' class 4-8-4 locomotives, are not yet ready to express a final opinion on its merits. Dynamometer car tests have shown-that the Coal consumption of this engine is very little less than that of identical piston valve locomotives.
Dreyfusshudson,
It might be worthy of note, that no self respecting automotive engineer would consider the use of "spool valves" or "piston valves" of any significance in modern automotive engine design. The "poppet valve" with all of its cam and valve seat problems was entirely superior in its rate of operations and unobstructed gas flow usage and its modern technology also thoroughly worked out.
It would seem the late design advanced cam Type B Franklin Valve gear was the "way of the future" in spite of the obvious limits revealed in the usage by the Pennsylvania Railroad of the 1940's of of the Franklin Type A.
Then there is the remarkable US Military development of this Type B Franklin system into the 1950s - with one surviving example - which among other things allowed soldiers with little practical knowledge of steam locomotive valve gear operation to effectively operate in a variety of railroad situations which they were professionally untrained for.
I would expect you are aware that the Franklin Type B system was also rebuilt into one Pennsylvania Railroad T1 4-4-4-4 Duplex locomotives no 5500 and was reported to have been the best of the type. I believe the Type B was also built into several new Chesapeake and Ohio Railroad late model L-2a 4-6-4 Hudson locomotives. I have never seen reports of this "poppet valve Hudson" usage but expect it might have outperformed the C&O 600 series "Greenbrier" 4-8-4's.
I believe Paul Kiefer the steam locomotive designeer for New York Central Railroad considered the "rotary cam poppet valve" design to have produced the hightest performing steam locomotive in that railroad's history - its Franklin Type A S2a Niagara 4-8-4 No. 5500 - and was preparing to offer a substancial engineering study regarding this. Although he never completed his "poppet valve" study his book A Practical Evaluation of Railroad Motive Power remains an impressive read concerning the steam vs. diesel controversy of the time.
It seems the Central finished with steam development and Kiefer still felt that the steam locomotive as designed at that time had an unrealized competitive potential against the diesel locomotive of the period.
I am excited about the steam locomotive "rotary cam poppet valve" design study! Which will likely continue the controversy in the future regarding the operation Pennsylvania Duplex T1 - No. 5550 the modern recreation of this historic locomotive type - to be equipped with a new generation of Franklin Type B valvegear!
If only the famed New York Central "Niagara" 4-8-4 No. 5500 could now be also recreated with its "rotary cam poppet valve" design - well that would sure settle some unanswered questions about "duplex" and the "poppet valve" - which are actually two separate subjects which Pennsy managed to tangle together. The Pennsy sure took on a great deal of work by developing the 8 axle 8 drivered locomotive in both "poppet valve" and "duplex" formats all at once!
I guess if the NYC "Niagara" No 5500 or No. 6000 were ever recreated there would be a controversy over the practical or the visionary formats for a "poppet valve" or "spool valve" version of this famous and historic engine. Possibly you would then see a realized 6000 horsepower in Paul Kiefer's fabulous design - supposedly its why the first engine was numbered 6000.
Dr. D
I have some comments, for what they may be worth:
Dr DIt might be worthy of note, that no self respecting automotive engineer would consider the use of "sleeve valves" or "piston valves" of any significance in modern automotive engine design.
This is in a very different context, with very different physics (and one might note that where mechanical concerns dictated use of sleeve valves, they were remarkably successful in contemporary modern airplane-engine practice. The thing to remember (as, for example, by comparison with something like the Aspin valve) is that both the design and operation of 'poppet' valves in modern automotive practice has little if anything to do with the designs at all successful in steam practice.
The "poppet valve" with all of its cam and valve seat problems was entirely superior in its rate of operations and unobstructed gas flow usage and its modern technology also thoroughly worked out.
But what Mr. Pawson is repeatedly saying is that at least some of the purported 'advantages' of poppet valves in reciprocating locomotives are overrated, and perhaps actually deleterious, in comparison to what could be achieved with proper (cf. Wardale for example) piston-valve practice. Much of what would be needed to refute his claims necessarily involves empirical operation -- either very well simulated or actual -- and it will be some while before that is likely done to the necessary 'rigor' (it certainly will be in the 5550 project).
Yes ... but kinda, sorta, not really when you look at it carefully.
There are two parts to the PRR implementation of type A: the layout of the OC gear, and the layout of the physical valves and their operation. The RC conversion only affected the former; the whole point of varietal type B-2 being that the RC gear was bridged and otherwise modified to actuate eight valves per cylinder in the 'legacy' configuration ... something that I recall optimizing high-speed flow, but NOT steam economy as in the three-valve type B that Vernon Smith optimized so well.
Whether the type B variable cams would have held up well in high-speed service is an open question; more so when you consider contemporary PRR maintenance and service as applied to the T1 fleet. My guess is that you'd see fairly rapid (and probably expen$ive-to-fix) preferential wear of channels in the cams, and perhaps surface damage to parts of the spherical roller followers, especially if the valve springs had to be made variable with effectively stiffer spring rate close to seating (which I suspect was the gist of what Franklin did to the springs in the latter '40s). It may be highly notable that no actual locomotive was ever ordered with shifting-cam "type C" (as patented), which is why I always use that term in quotes.
Then there is the remarkable US Military development of this Type B Franklin system into the 1950s - with one surviving example - which among other things allowed soldiers with little practical knowledge of steam locomotive valve gear operation to effectively operate in a variety of railroad operations.
This has been formally termed 'type D' for a variety of reasons, one of which is the somewhat-interesting set of design changes made by Vulcan et al. in making the refit kits. This is not at all the same thing as poppets used for high-speed efficiency OR poppets used to reduce the water rate and fuel burn. This setup intentionally induces wire-drawing for a large portion of operation in order to approximate the effect of cutoff "physically" (by limiting effective steam flow per stroke!) in order to get reasonable automatic steam economy without a dedicated proportional mechanical cutoff-control system. While this would fairly obviously help with some kinds of slipping, and the one-lever control system certainly had its elegance, I wouldn't consider this anything of interest to American railroad use of large power, and I doubt anything it could bring to switching practice would counterbalance the effective advantages of diesel power there.
[quote]I would expect you are aware that the Franklin Type B system was also rebuilt into one Pennsylvania Railroad T1 4-4-4-4 Duplex locomotives no 5500 and was reported to have been the best of the type.[quote]
As noted, this was B-2, different from any other practical type B installation of which I'm aware. Some of the type B versions have incredibly overbuilt external shaft arrangements, the C&O engines you mention being examples. Note that none of the type B arrangements seem to have survived very long, even though showing nominal advantages in operation ... they just weren't enough to beat the competition. Even the first-generation competition, most of the time...
I believe Paul Keifer the steam locomotive designer for New York Central Railroad...
Once and for all, the man's name is Kiefer. Kiefer, Kiefer, KIEFER.
... considered the "Rotary Cam Poppet valve" design to have produced the hightest performing steam locomotive in that railroad's history - its Franklin Type A S2a Niagara 4-8-4 No. 5500 - and was preparing to offer substancial engineering studies regarding this. His book A Practical Evaluation of Railroad Motive Power is an impressive read.
You will note already that part of what you said is gibberish. Type A, as Kiefer used it, specifically involves OC gear, and you will note that the 'forthcoming' studies promised in the railroad motive-power report were quietly never published... and the supposedly more 'economical' Niagara with them retired long before most of the T1s. Reminds me of Khan's last sentence.
Kiefer I believe felt that the steam locomotive as designed at that time had an unrealized competitive potential against the diesel locomotive of the period.
And he was probably right. However, we don't find him tilting at windmills to continue steam experimentation even months longer than the NYC decision to stress 'dieseliners' in the only service where Niagaras were truly competitive with EMD passenger power.
If only the famed New York Central "Niagara" 4-8-4 could now be also recreated with rotary cam poppet valve gear - well that would sure settle some unanswered questions about Duplex and the Rotary Valve design - which are actually two separate subjects. ... I guess if the NYC "Niagara" were ever recreated there would really be a fight over the practical or the visionary formats for a poppet valve or spool valve version of this famous and historic engine. Possibly you would have seen a realized 6000 horsepower in Paul Kiefer's fabulous design.
I am tempted to say that it wouldn't be hard to see a realized 6000hp ... perhaps not just IHP ... with a properly modernized piston-valve setup, especially if first cost and maintenance expense were of relatively little object (as they would be on a well-cared-for excursion engine). I for one would want to see a Niagara design with two independently-driven piston valves (for different control of admission and exhaust) to get practical comparisons of that particular aspect of "type C" practice. You could easily build this with a 'section' through the cylinder block that would allow retrofitting of other valve arrangements as desired, especially now that laser keyhole welding is becoming a mainstream and costed-down technology...
Overmod,
Paul Kiefer makes the following comments to our discussion - p17 Railroad Motive Power, Reciprocating-Type Steam Locomotive,
"Another recent and significant reciprocating type steam locomotive for use on American Railroads is the four-cylinder, simple-expansion, non-articulated type for the movement of passenger and freight trains at relatively high speeds in territory where the character of the traffic and the topography of the railroad have imposed partricularly heavy demands on the motive power, involving in some cases the use of helper locomotives over grades.
The first unit of this type was built by the Pennsylvania Railroad in 1939 and exhibited at the World's Fair in New York. It was of the 6-4-4-6 wheel arrangement with 84-in. drivers, designed for heavy main-line passenger service. This was followed by 27 freight and 52 passenger locomotives, the latter having 4-4-4-4 wheel arrangement with 80-in. drivers, designated as Railroad Class T-1.
The Class T-1 four-cylinder passenger locomotive is comparable in capacity with the two-cylinder "Niagara" type Class S-1 of New York Central, described above. Some of the principal characteristics of both types are shown in Table 1.
The four-cylinder type has an advantage over the conventional two-cylinder design of equal capacity through reduction of thepiston thrust and the load on the main crankpins and connecting rods where the conventional rod drive is usedon the latter, together with a reduction in weight of revolving and reciprocating parts per cylinder with correspondingly lighter counterbalance, but introduces the complication of maintaining two additional complete sets of cylinders and machinery, besides necessitating increased over-all length and wheel-base with the attendant complications in handling at terminals and the disadvantages of increased total weight and cost.
A number of reciprocating locomotives have recently been built with poppet valves, notiably the 52 Class T-1 of The Pennsylvania Railroad, which received the arrangement known as the Franklin System of steam distribution. As these engines were of completely new design, a direct comparison of the performance characteristics with the same design equipped with conventional piston valves and gear could not be made.
An earlier application of poppet valves was made to one of the P.R.R. Class K-4 "Pacific:" type, of which there are a large number, and it is understood the results obtained led to the use of this type valve gear on the T-1 class.
When the New York Centeral "Niagara" engines were ordered, it was decided to equip the last one of the lot with Frankling poppet-valve arrangement having four intake valves of 6 1/2 -in. diameter and six exhaust valves of 6-in. diameter per cylinder. With this single exception, the engine is identical with the others, all of which have large-size piston valves actuated by modern-design valve gear.
Thus, a means for making an exact comparison of capacity, acceleration and performance for these two kinds of steam distribution was made available, and comprehensive road tests with a dynamometer car are now being conducted with the poppet-valve engine, which, with the tests already made of the piston valve equipped S-1, will give this exact information.
No predictions of the results of these tests are at present being attempted, but when worked up and analyzed later during the present year, they should provide a basis for definite conclusions."
-----------------
Paul Kiefer also notes the following test information,
- Chart A: Power Characteristic Curves Passenger Locomotives Reciprocating Steam-Electric-Diesel Electric
Rated horsepower - NYC "Niagara" - S1 - 6600 Indicated Horsepower
5000 Drawbar Horsepower
NYC "Hudson" - J3 - 4700 Indicated Horsepower
3900 Drawbar Horsepower
Chart A - Chart "A" shows the progressive increase in drawbar capacity for reciprocating steam built during the last two decades , from Class K-5 type 4-6-2 to the S-1 "Niagara" 4-8-4. These curves, from which the detailed cut-off information has been omitted to simplify the composite chart, were developed during road tests using the dynamometer car and made under the direct supervision of qualified Equipment Engineering Department personnel, using suitable length trains and a second engine back of the test car for close regulation of trailing load resistance so as to insure the elimination of practically all acceleration effects during separate runs made to produce the required increments of the complete curve as here illustrated.
With approximately rated boiler pressure and the locomotive, as a whole, in resonably good condition, there is a cut-off for each incidental speed at which maximum power output is made available from a given design, regardless of whether the capacity of the steam generating plant or the engine cylinders is the limiting factor. Any longer cut-off not only reduces the power output at that speed but also is wasteful of steam and fuel. At a shorter cut-off, the power output becomes less and it follows that there is a corresponding reduction in steam consumption.
By developing each increment of the pull speed curve, the succession of cut-off points throughout the range of operating speed to produce maximum power is definitely established and consequently each portion of any such curve can be reproduced at any time in regular train service by using the proper cut-off in relation to speed. This type of operation is utilized for all System reciprocating steam road locomotives, freight and passenger, and is based on the principle of cut-off selection which can be applied directly and conveniently through the guidance of a device known as the "Valve Pilot" which by means of a duplex gauge indicates the incidental speed and position of the cut-off at all times. Thus, full available capacity may be currently be produced while accelerating to running speed and, thereafter, the cut-off may be shortened consistent with load, profile and speed conditions.
As illustrated by the chart, there has been a gradual increase during the twenty-year period from 2500 dbhp at speed of 45 mph obtained with a K-5 class to 5050 dbhp for the S-1 (NYC Niagara), with 275 psi, an increase of 102 percent. The curve for the S-1 (Niagara) with 290 psi, derived from the test with 275 psi, shows a maximum of 5300 dbhp and increase of 113 per cent of the K-5.
--------------------------
Here Paul Kiefer indicates that the NYC 4-8-4 "Niagara" S-1 piston valve locomotive with a boiler pressure increase from 275 psi to 290 psi showed an drawbar horsepower of - 5300! Indicated horsepower was near 6600!
------------------------
The horsepower for the New York Central S-1a poppet valve "Niagara" No. 5500 was apparently tested and never revealed by Kiefer.
The horsepower for the K-5 "Pacific" 4-6-2 was 2500 dbhp.
The horsepower for the New York Central J-3 "Super Hudson" 4-6-4 was 3800 dbhp and 4700 ihp.
The horsepower for the New York Central L-4 "Duel Service Mohawk" 4-8-2 was 5400 ihp.
----------------
Keep in mind I posted a link to a readable copy of this report ... it may be very familiar to you as it was scanned from the copy at University of Michigan.
https://catalog.hathitrust.org/Record/001615135
Everyone reading this thread should look at this carefully (perhaps several times) and take notes.
Will someone with a copy of Know Thy Niagaras advise if it contains detail pictures of the valve layout for the poppet-valve Niagara? I'm not sure how the 'three' exhaust valves per end would have been arranged or driven; presumably there was one extra centered valve between the two paired inlets.
I have in my possession an interesting book on railroad engineering titled The Steam Locomotive by Ralph P. Johnson. This particular copy is signed in the front cover - "Fred S. Broadbent with the compliments of Ralph P. Johnson, October 1944".
It is a notable volume on steam locomotive engineering, second edition, Simons - Bordman Publishing,New York, 1942, and 1944. More notable is that Fred S. Broadbent worked for American Locomotive Sales Corporation - ALCO - and Ralph P. Johnson M.E. was Chief Engineer for BALDWIN Locomotive Works and one of the primary builders of steam railway locomotives in America and also one of the engineers behind the Pennsylvania Railroad's purchase of the first two experimental T-1 4-4-4-4 Duplex locomotives.
Among other engineering concepts Ralph Johnson was a leading proponent of "Rotary Cam Poppet Valvegear" and as the story goes also insisting on the railroads use of a locomotive booster engines for the T-1 Duplex locomotives - advice the Pennsy sadly failed to heed - as it turned out the second prototype engine No. 6111 was fortunate indeed to be so equiped.
The date of the signing of Ralph's book on locomotive engineering was "October 1944" during the height of World War II and was four months after the D-Day landing of British and American forces in Normandy at which time the railroads of America were major players in the success of the "Arsenal of Democracy" that brought about Allied Victory.
In engineering the Pennsylvania Railroad T-1 locomotives Ralph Johnson gives the following explanation of the design ideas behind the use of Franklin Rotary Cam Poppet Valve gear,
Poppet Valves - With the growth of high speed and high pressure service the shortcomings of the piston valve with the conventional Walscharts and Baker gears are becoming increasingly important. These gears are reciprocating in action and the steam distribution is by means of a single valve which controls both inlet and exhaust, thus mechanically joining two separate and distinct phases to the disadvantage of both. The poppet valve offers a means of separating these two phases of steam distribution.
Conventional gears change the cut-off by changing the stroke of the valve and restricting the port opening. The poppet valve does not change the port opening for any ordinary cut-off, but keeps the valves wide open, varying only the time interval during which they remain open. The exhaust, being entirely independent of the inlet, and fully open at all cut-offs, provides a practically constant release and reduces back pressure. This characteristic makes the poppet valve particularly advantageous in high speed, short cut-off service. Thus, at 25 percent cut-off, where the piston valves release at between 50 and 60 percent of the stroke, it is possible with the poppet valves to carry the expansion to almost 90 per cent. Similarly, while the piston valve closes at anywhere up to 30 per cent of the stroke, it is possible with the poppet valve to reduce this to 15 per cent.
The conventional reciprocating valve is always in motion and, on one side or the other, is constantly engaged in restricting the port opening. Poppet valves, on the contrary, are fixed in their open or closed positions, with quick transition from one to the other. With this faster action, wire drawing is greatly reduced with compression ends of the stroke. The cylinder indicated horsepower is further increased by the smaller clearance spaces and reduced backpressure.
With a reciprocating piston valve, decreasing the cut-off increases the compression, which gives high efficiency for normal working conditions only. With increased cut-off the compression becomes insufficient and with decreased cut-off it becomes too high, resulting in a part of the power being worthlessly used in the cylinder itself. Therefore a poppet valve gear, wherin the cut-off can be independent of the compression, due to separate admission and exhaust valves, which in themselves provide a certain uniflow and enable reduction in cylinder clearances, is favorable to the more efficient utilization of the steam in the cylinders. With poppet valves, cut-off's as short as 5 per cent have been used successfully.
The poppet valve can handle higher pressure-temperature steam with less lubrication difficulties than the piston valve, as the use of lifting instead of sliding valves decreases the danger of "stuck" valves caused by carbonization of the lubricating oil. It also overcomes the disadvantages of the small areas for the entering and exhausting steam, even where the double-ported type of piston valve is made use of. Steam passages for poppet valves should be if anything, more liberal than for piston valves; so that full advantage of the increased port areas obtainable with poppet valves can be realized.
With boiler pressures as low as 225 pounds, poppet valves are advantageous in that at high speeds the distortion of valve events due to the "whip" in the piston valve gear is eliminated, owing to the absence of inertia effects in the valve motion.
Another advantage of the poppet valve is the greatly reduced power necessary to operate the mechanism. The Walschaerts or Baker gears require from 50 to 60 horsepower for their operation, while tests indicate the poppet valves require as low as 3 horsepower.
--------------------
At the time Ralph Johnson was signing this copy of The Steam Locomotive to Fred Broadbent my father-in-law was fighting Nazi's in Metz, France with General Patton. My own dad was in the Pacific with General MacArthur invading the Philippine Islands. It must have been a joy to go to work at Baldwin Locomotive Works and build steam locomotives for a living! To dream dreams of the great Pennsy Duplex locomotives - the speeding giants of the rails!
---------------------
Amusingly enough, my copy of Johnson's book came directly from Simmons-Boardman as the 1983(!) edition, in paperback, considerably cheaper than BS collector prices. And guess what? it is newly reprinted in the 2002 (!!) edition, and you can order it directly from here.
You can read the 1944 ('second') edition online free here.
Remarkable to me is that the book has little discussion of duplexing, either theory or detail design. You will also find relatively little discussion of more complex issues of balancing, which was still in active evolution in the early '40s when the first edition came out. I thought when I first read it, and still think today, that an edition revised with some of the interesting developments in steam-locomotive technology in the years since then would be significant -- the question being whether some of the controversial issues, like the issues Mr. Pawson raises about poppet valve practice, the application of jacket heat, tradeoff of dead space vs. excessive compression, and the Lima 'dogma' of long compression, can be treated fully objectively in an engineering text.
Steam locomotive design. Its an interesting engineering problem or should I say a historical one. On one hand not a lot of work is being done with steam Locomotive building as a contemporary subject - having been mostly concluded in the 1950's.
The steam railroad technology is OLD pertaining to an "industrial age" past and yet with modern engineering we get to evaluate what was done with the 20/20 hindsight and with 70 years further future science.
The fact that the T1 reproduction is ongoing does give some opportunity to eliminate some mistakes and make a few more! And how can one duplicate the experienced armies of capable locomotive builders - shopmen, foundrymen, machinists, master craftsmen and workers in steel?
At least when a surviving engine exists it represents the "extant evidence" of the engineering of the past - one that a post mortum can truely effectively evaluate.
But a new construction - no matter how well carried out - is in fact NEW!
With regards to ships there is a tradition - "no matter how many times the SS JANE is rebuilt it is infact the SS JANE - be there not a single timber left." That's the legal definition. A copy of the SS JANE is in fact the SS JANE II - a different ship.
And yet the chance to see a running Pennsy Duplex live and breathe again - its a project that cannot be turned easily away from!
Thanks Overmod and Dr D for these additional highly informed comments, and most especially for the link to Kiefer’s treatise, which I will read with great interest.
I may have said it before, but I am not an engineer, though am quite happy to accept the premise that, given modern materials and engineering design techniques, it will be quite possible to create a poppet valve suitable for use in reciprocating steam engines which will suffer neither from the mechanical failures that plague existing designs (even our Caprotti Pacific has experienced valve gear failures in service in the recent past), and eliminating the leakage losses from which most earlier designs seemed to have suffered. Our Caprotti Pacific seems to have got over these latter, when in top class condition at least, though analysis of the actual test plant and road test data on it say it was not, contrary to what was claimed at the time, any more efficient than a good PV design working at that superheat and backpressure. Even the falsely claimed advantage was in truth quite small, hardly revolutionary.
The question then is whether such a valve would have any step change advantages over PVs. My view is that from the point of view of efficiency at least, the answer has to be ‘No’. I say this on the basis of a large amount of work I did figuring out why the test results on most French Compounds say they were less efficient at the drawbar than simples (Yes, really). The reality is that the simple reciprocating engine does an excellent job of extracting the most of the work it is possible to extract from steam of a given inlet temperature and pressure and exhaust pressure. At the temperatures and pressures allowed by firetube boilers, once adequate superheat has eliminated condensation (heat transfer) losses, the second Law allows but little more. The poor efficiency of any steam operated device operating under these conditions has little to do with their design, everything to do with the immutable requirements of the Second Law. Inabilty to extract 100% of what the second law allows is due to entropy increases during irreversible expansion. These are difficult to eliminate, particularly during admission, and indeed the fact that poppet valves tend to have higher clearance space than PVs works against their efficiency to a small degree. Even the benefit of double expansion compounding can at best give less than 10% uplift.
If you are trying to make a silk purse out of a sow’s ear, as with the K4, then some meaningful advantage may be possible by fitting poppet valves. However, I would wager, (though it’s obviously difficult to prove directly) that a good PV design could always match (near enough) an ideal poppet valve on efficiency. The poppet valve will show apparent benefits in short cut offs by pulling harder, but that simply means the boiler is being steamed harder, which a PV can match by lengthening the cut off slightly.
Johnson’s arguments as to why poppet valves might be better at high speed are perfectly rational, but when you crunch the numbers it doesn’t add up to much. The key thing for high speed is not to have too restricted a clearance space to avoid over compression, and though, PRR designs apart, it’s difficult to get values for this quantity for US designs, my understanding is that values were usually quite generous. I would further say that given power/weight ratio considerations, the maximum speed you could ever schedule with steam on level track is 100mph. The PV MILW As and F7s were the only types ever to achieve this, and even then hauling lightweight trains (I often wonder how much they suffered from this incessant hammering).
Overall, I think that Kiefer’s belief (giant of a man though he was) that there was significant upward space for steam locomotive efficiency hence power is quite wrong. You can take most actual designs and improve them by up to 10%, but the reason this prize was not gained is often due to the fact that the cost and complexity involved in achieving it were not worth the effort. And 10% would never make steam competitive with other forms of traction, though it might have seemed able to tip the balance in the late 1940s. To get more you need to go to water tube boilers with very high pressures, and Compounding or turbines, but even here the benefits, if you are exhausting to atmospheric pressure are not really transformational. But even in 1947, long time steam professionals were still trying to prove this is not so.
I have a friend who has done first principles estimates of the several factors contributing to the HP losses between pistons and wheelrims (‘machine friction’). His estimates agree well with UK test plant data. Applied to US designs, this approach says valve HP losses would be about 25HP for a Niagara, 45 HP for a PV T1 at 90mph, so the benefit for poppets may not be as high as Johnson hoped.
It may be that a properly engineered poppet valve would be more desirable from a maintenance and reliability point of view than a PV- I cannot say. If so in an oil hence diesel free world, Franklin’s development efforts might well have paid off handsomely.
- Message deleted by me for posting in the wrong thread. Thank you. -
Well, the thread seems to have run out of steam on the original topic, so firstly thanks to all who have contributed. I have a couple of specific questions and thoughts to conclude:
1. To be more explicit about my main original query, I was looking for information about the state of the K4s in the 1939 AAR tests, a number of developments having been made since 1737 was tested in 1914. As far as I know, there is no book describing the detail of the progression of technical developments on the K4 and when and how widely they were applied? The T1 report implies 5341 was tested with a sine wave superheater in 1937, as applied to 5399. How common were these devices? Also, there is a paper by Vincent (RME 1938, p5) that describes the testing of a K4 with various exhaust systems, including the ‘star’ design, this also on an M1. Which system was actually adopted and when? As to what benefits these changes would bring, in the Pennsylvania State Archives in Harrisburg, there is a large number of Penn Central files, which include those of the PRR Engineer of Tests. https://www.phmc.pa.gov/Archives/Research-Online/Pages/PRR-Group-286.aspx I have an index of his 1295 items (they tested everything!), housed in a 71 cuft container (!). In addition to the reports I already have, of particular interest would be:
a. Item 26: front end arrangement locomotives 6707 +K4s, 1932- I suppose these are the tests described by Vincent.
b. Item 1068: K4s superheater units, single pass (sine wave), 1938-45.
Also of interest:
c. Item 168: performance tests of M1 locomotives, 1926.
d. Item 398: Investigation of (Q1) loco 6130, 1943.
Has anyone gained access to these?
Of great interest also would be the full AAR report on the 1000 ton passenger tests with PRR, CNW and UP locos, summarised in RME 1939, p175. This famously shows that coach resistance was apparently different on the three railways. One can infer the kind of methodology used to deduce coach resistance from the K4 5399 test report, also some NYC tests when they worked J3 5408 up to the high 90s with a similar 1000 ton test train. They were doing the best with the analysis tools they had, but I have a hunch that if there is sufficient data to do a computer analysis, a different conclusion might be reached.
If anyone has access to any of these, please let me know.
2. It may seem I was unkind to the PRR K4 design. I was looking at them from the rather narrow perspective of power and efficiency- obviously a top concern for designers- and the K4 was surely off the pace in the 1930s. However, a class of 425 that worked for 40 years can’t be all bad - in fact, quite the contrary - it must have had an awful lot going for it! I would guess it was a rugged and reliable friend that was more than powerful enough for most of the duties assigned to it, the premium expresses apart. Bring on 1361.
As an act of penance, I analysed the one K4 log I have, of 5432 (maybe 5422) running 700 tons of Broadway over the 141 miles from Englewood to Fort Wayne in 118 minutes, average 71.6 mph, a splendid effort, including it would seem slowing to 55-60mph near Indiana Harbour, Gary and Etna Green, plus some caution over the water pans before Hamlet. Averaging over 70 mph with a train more than twice the weight of the loco is Hall of Fame stuff for steam. 5432 was saddled with an enormous ‘coast to coast’ tender, which in effect added another coach to the load. It was timed from the footplate by Baron Vuillet. Power was generally around 2400IHP which was sufficient to maintain speed in the mid 70s up the gentle grades beyond Valparaiso, with speed rising to 90mph on the downgrade beyond Larwill. The most remarkable feature of the run was however the acceleration to 71mph past South Chicago, which would require all the K4’s TE when starting, and all its boiler output once moving. One wonders if there wasn’t a Hudson on the adjacent track providing a bit of encouragement. This is not the kind of treatment the fire might expect at beginning of a long hard run, so maybe one should add ‘suffers abuse gladly’ to the K4’s list of virtues.
Vuillet says that up to Warsaw the loco was not worked at full throttle, so the cut offs would be significantly longer than 30-35% Altoona says is needed to produce this kind of power. The coal and water consumptions Vuillet quotes are horrendous if true-79 (UK) gallons/mile, 135lb coal/mile, about 60% higher than Altoona said should be achievable. These figures are consistent with the loco being worked in up to 60% cut off with around 110psi in the steam chest. It suggests that ‘Economy? Who cares? Let’s get over the road on time’ was the crew’s motto.
3. I shared my findings on the PRR poppet valves hoping in a way that there would be some comeback. Like all steam fans, I would like to believe that the PRR was right, and that there are technologies such as poppets which could advance the performance of latter day steam designs. The feedback has left me even more certain that this is not the case. Underlying all this, there is a good theoretical argument that I mentioned earlier that poppets cannot deliver much if anything in terms of efficiency, and it might help to spell this out.
The Second Law says that steam entropy cannot decrease between engine inlet and outlet. In practice, it will increase. Table 1 shows how much power you would get at a steam rate of 25000lbs/hr with a perfect engine where entropy is the same at inlet and outlet, at a (very optimistic) back pressure of 3 psig.
Condition
Inlet temperature, deg F
Inlet pressure, psig
Exhaust pressure, psig
Maximum thermal efficiency %
A
240
3
2155
17.0%
B
2246
17.3%
C
2343
17.7%
D
2446
18.2%
E
2554
18.6%
F
2668
19.1%
H
10
2109
16.0%
I
170
2129
16.1%
G
2788
20.0%
J
-14.6
5087
38.5%
As superheat increases (A to F), maximum possible efficiency increases. If back pressure rises, or inlet pressure drops (H and I) maximum efficiency falls. At the limiting temperatures and pressures from a firetube boiler (G), maximum possible efficiency is but 20%. If only you could exhaust to vacuum, this could be doubled (J). But with maximum possible values normally in the high teens, you are already in poor shape from an efficiency perspective with any steam powered device exhausting to atmospheric pressure.
The question then is how good a reciprocating engine is at delivering this maximum value. Table 2 shows some calculations for a common or garden Walschaerts engine with 1.625” lap and 0.25” lead. The first set of calculations makes an allowance for some leakage, and condensation when it occurs (below 650 degF). Actual efficiency is lower than the ideal in Table 1, but if you calculate the % of the maximum possible (IE, isentropic efficiency) then once condensation losses are fixed, IE values are just short of the 80% mark. This is typical for all test plant data. There is an effect of cut off in these calculations (generally shorter cut off= better IE), but in the range 20-25% cut off the effect is small. Note there is no simple correlation between IE and actual efficiency.
Britannia with leakage and condensation
Britannia with no leakage or condensation
Efficiency
IE
21.1
1597
12.6%
74.1%
22
1722
13.3%
76.7%
22.9
1833
13.9%
78.2%
24.8
2002
15.1%
85.4%
23.7
1927
14.3%
78.8%
25.6
2094
15.5%
85.6%
24.4
2009
14.6%
78.7%
26.2
2165
15.8%
84.8%
25.1
2081
14.9%
78.0%
26.9
2245
84.1%
24.9
1681
12.7%
79.7%
1861
14.1%
88.2%
32
1737
13.1%
81.6%
34
1831
13.8%
86.0%
20.4
2137
15.3%
76.6%
2366
16.9%
84.9%
The second set of calculations shows what would happen if a perfect leak free Walschaerts engine and boiler could be created. (steam losses from the boiler are conventionally included in engine efficiency estimates). Values for this set up now increase to the mid 80s. In other words, there is only 15% upwards headspace for any technological advance beyond simple Walschaerts at equal leakage.
The question then is where the increases in entropy that cause this 15% loss occur. It is possible to calculate how much entropy increase occurs at different parts of the engine cycle from indicator diagrams, and the answer is basically threefold:
a) Filling the clearance space -generally the smallest effect at speed.
b) During ‘wiredrawing’ as cylinder pressure falls below steam chest during admission. Pressure does not fall much in long cut offs at low speed, so here the effect is small, but at high speed, a large pressure drop occurs and this is a significant problem.
c) Losses when steam has not expanded to exhaust pressure when the exhaust ports open – the main effect at long cut offs, when cylinder pressure is relatively high at exhaust port opening, less so at high speed, when cylinder pressure is lower at this point, though still the largest effect.
A further possibility occurs if compression loops form, but this will be rare with clearance space around the 10% mark.
So if you want to increase efficiency at speed, you firstly want to tackle losses during inlet. But even with poppet valves there is still a pressure drop to cut off, so they do not really address this problem. The shorter cut off needed by poppet valves will lead to somewhat greater expansion hence lower pressure at exhaust opening, so a some benefit here, and when applied to the sub optimal K4 design, this is a quite significant effect at high steam rates. If release is delayed, this will further increase expansion, but pressure is falling relatively slowly at this stage, so it is not a large effect. However, at cut offs above 20-25%, poppet cut offs at a given steam rate are similar to a 1930s state of the art very long Lap design as e.g. on the ATSF 4-8-4s, so little potential benefit over PVs here either in terms of efficiency or increased power. If poppet events are organised such that exhaust closure is delayed, then this will lead to lower compression pressure, which could be counterproductive because there will then be more losses filling the clearance space. And, Clearance space tends to be larger for poppet valves anyway, to their disadvantage.
My calculations comparing a Walschaerts and low leakage Franklin T1 at equal steam rate (60000lbs/hr) at 75mph say the Walschaerts will be about 3% more efficient than Franklin, despite the lower cut off with Franklin. This effect is however entirely due to the fact that I have told the computer that the clearance space on the Franklin is 3.5% larger than Walschaerts (as on the T1), and computers do what you tell them! But what we can conclude is that in most circumstances the efficiency differences between a low leakage poppet and 1930s state of the art PV will, if anything, be small. And, in any case, as the run of 5432 shows, what matters in practice is not what the test plant or theory says, but what crews do with the tools at their disposal.
So, not only is the potential for efficiency improvement as a matter of principle quite small, it is also true that poppet valves are not likely to deliver much if any efficiency benefit over state of the art PVs at speed, even if their leakage problems can be solved. They are to my mind a classic illustration of how new technology sometimes raises false hopes of a breakthrough, driven by the belief that ‘we must and can do better’, and how results are then interpreted to support those beliefs and wishes. R&D people (not to mention steam fans) are by nature technology optimists. As noted earlier, you need full scale compounding to give yourself a chance (though often not in actuality) of gaining roughly half the 15%.
All a bit distressing for we steam fans, but that’s the way it is.
To conclude, if I may repeat an earlier point, this all has no bearing on the new build T1 in my view. Its modified Franklin valves should work just fine. They may not be as efficient as one might hope, but no one, least of all me, will care. Let’s see it roll!
Dreyfuss:
You stated:As superheat increases (A to F), maximum possible efficiency increases. If back pressure rises, or inlet pressure drops (H and I) maximum efficiency falls. At the limiting temperatures and pressures from a firetube boiler (G), maximum possible efficiency is but 20%. If only you could exhaust to vacuum, this could be doubled (J). But with maximum possible values normally in the high teens, you are already in poor shape from an efficiency perspective with any steam powered device exhausting to atmospheric pressure.
I have ofen wondered what happens at higher altitudes . So say at 8500 feet altitude standard atmosphere is 1/2 of that at sea level then what? For that matter what if any is there any difference from high pressure to low pressure systems and could that have comproised comparative test values?
Our community is FREE to join. To participate you must either login or register for an account.