If you register, you can do a lot more. And become an active part of our growing community. You'll have access to hidden forums, and enjoy the ability of replying and starting conversations.

Saint Class 135 ish mph

Discussion in 'Steam Traction' started by Reading General, May 5, 2017.

  1. Courier

    Courier New Member

    Joined:
    Dec 1, 2010
    Messages:
    197
    Likes Received:
    117
    Thank you for your polite and civilised points.

    When I first went to the NRM to see the roll I thought it possible that I would find evidence for 126 or even higher - because there was a theory that at high speeds the paper wasn't travelling as fast as it should have been. I would have been happy to be the one to prove this. I realised I could check this theory by measuring the distance between the milepost marks - ie were they really 24" apart. Over one mile that might not be very accurate - but when you measure it from milepost 105 to milepost 89 - ie 32 feet of paper you can be much more precise.

    So what did I find?

    In general the paper was moving at a very precise 24" to the mile - the difference I found was less than one part in a thousand and is as likely to be caused by my measuring method as anything else.

    As mentioned before by looking at a long section of the roll it was possible to see that the paper was not moving at a constant 24"/mile. During every mile it sped up and slowed down. I suspect that the gears in the mechanism were very slightly eccentric.

    The high speed section of the roll was marked off in 5 second sections and annotated with the speeds calculated by the LNER. You can back-calculate the distance travelled for each 5 seconds, and then sum up the distances to compare with the mile marks. This shows that the speeds were overstated. Over three miles you can see a consistent error mile by mile. In D R Carling's accounts he says they used a scale to directly read off speeds from the 5 second distances, not a vernier. That scale was slightly out.

    You might have guessed from the following that I am an engineer - not a scientist.

    In a different way you can still extract knowledge from the data around City of Truro. You can't prove everything, but that is not the same as saying that you can't prove anyhing. I will post an example of that at some point.

    I don't expect any of my work to change the generally held opinions on Mallard - the plaques on the side will still say 126 mph. However I hope it will be of interest for the small minority who are interested in the details of old steam speed records.
     
    MellishR, Copper-capped and LMS2968 like this.
  2. Big Al

    Big Al Nat Pres stalwart Staff Member Moderator

    Joined:
    May 30, 2009
    Messages:
    21,065
    Likes Received:
    20,773
    Location:
    1016
    Heritage Railway Volunteer:
    No I do not currently volunteer
    This is truly bandit territory. On the 90 mph Bittern run from Newcastle to York in December 2013 in the dark where all we had was GPS equipment (and there was quite a lot of that on the train), you would be surprised at the variation in speed shouted out by folk from their different devices. At one point I did wonder whether my part of the train was travelling at a different speed from everyone else's. I know what the maximum speed was, where it happened and for how long it happened - I think! My point is that all this retrospective analysis is technically interesting but of little real value. I think we can agree that all recorders at the time will have been inclined to 'talk up' the maximum and that includes an A4 touching 126 that it did not sustain. The thread title is far less believable. The CoT speed on Wellington is slightly more believable but also arguably unlikely. And so the debate continues........
     
    S.A.C. Martin, gwalkeriow and LMS2968 like this.
  3. Jimc

    Jimc Part of the furniture

    Joined:
    Sep 8, 2005
    Messages:
    4,052
    Likes Received:
    4,665
    Occupation:
    Once computers, now part time writer I suppose.
    Location:
    SE England
    Heritage Railway Volunteer:
    No I do not currently volunteer
    I wouldn't be. GPS spot speeds are notoriously prone to artifacts. I have a GPS trace that shows my sailboat (max speed at best possibly 20mph) jumping sideways directly towards the wind at 50mph, and you'd be amazed at how big the waves are on my little lake, judging by how much the measured height above sea level varies. If you want to do serious speed recording with GPS then (excluding clever tricks like differential GPS and ground stations) then you need a recorded trace, and to chart it up, exclude anomalies and average over a reasonable distance with a decent number of data points.
     
  4. Steve

    Steve Resident of Nat Pres Friend

    Joined:
    Oct 7, 2006
    Messages:
    11,930
    Likes Received:
    10,088
    Occupation:
    Gentleman of leisure, nowadays
    Location:
    Near Leeds
    Heritage Railway Volunteer:
    Yes I am an active volunteer
    I have never really examined the dynamometer car or even seen the trace, copy or original. You say that you have carefully checked the distance of the trace but I am not sure what you have checked it against. To do that you have to move the vehicle a given distance and carefully check the trace against that . Similarly, you have to check the time base against an accurate timepiece. I'm not suggesting that this needs to be done now but there really should have been a calibration check before and after to confirm the accuracy of the measurements recorded. Was this done, I wonder?
     
    andrewshimmin likes this.
  5. Matt37401

    Matt37401 Nat Pres stalwart

    Joined:
    Jun 8, 2014
    Messages:
    15,328
    Likes Received:
    11,666
    Occupation:
    Nosy aren’t you?
    Location:
    Nowhere
    Heritage Railway Volunteer:
    No I do not currently volunteer
    I can't remember which one it was, but before the locomotive exchanges in 1948, but wasn't one the Dynomometer cars found to not to be caliberated correctly?
     
  6. LesterBrown

    LesterBrown Member

    Joined:
    Jan 20, 2009
    Messages:
    995
    Likes Received:
    761
    Location:
    Devon
    While the LNER and LMS companies claimed maximum speed records IIRC the GWR speed claims (Truro and Connaught, Fishguard Boat Trains, Cheltenham Flyer etc.) were all based on start to stop average speeds.
     
  7. Forestpines

    Forestpines Well-Known Member

    Joined:
    Jun 5, 2009
    Messages:
    1,681
    Likes Received:
    2,438
    Gender:
    Female
    Location:
    Somewhere in the UK
    Heritage Railway Volunteer:
    Yes I am an active volunteer
    As someone who works with interpreting GPS data as a large part of my day job: what you really need to do is record the device's DOP values and potentially also the satellite count as part of your data set, and use this to compute error bars.
     
    MellishR and S.A.C. Martin like this.
  8. Big Al

    Big Al Nat Pres stalwart Staff Member Moderator

    Joined:
    May 30, 2009
    Messages:
    21,065
    Likes Received:
    20,773
    Location:
    1016
    Heritage Railway Volunteer:
    No I do not currently volunteer
    Sounds sensible to me. On the Bittern run in the dark I noted speeds every 20 seconds to create the distance/time/speed log via Excel. You could easily see the odd reading that was anomalous. Roll back to 1938 (and earlier) and it's logical that with what was available at the time there may be a similar query. But I have to say that to begin to suggest that a dynamometer record (e.g. Mallard) might have the same level of (in)accuracy as a stop watch and a pair of eyes (City of Truro) is stretching it a bit.
     
  9. Spamcan81

    Spamcan81 Nat Pres stalwart

    Joined:
    Aug 25, 2007
    Messages:
    35,121
    Likes Received:
    20,773
    Occupation:
    Training moles
    Location:
    The back of beyond
    Heritage Railway Volunteer:
    Yes I am an active volunteer
    The LNER also had a start to stop average speed claim when the Coronation streamliner beat the Cheltenham Flyer's average IIRC.
     
    S.A.C. Martin and LesterBrown like this.
  10. Jamessquared

    Jamessquared Nat Pres stalwart

    Joined:
    Mar 8, 2008
    Messages:
    26,103
    Likes Received:
    57,430
    Location:
    LBSC 215
    Heritage Railway Volunteer:
    Yes I am an active volunteer
    If I read @Courier correctly, he is suggesting that the dynamometer roll was not moving at a completely linear speed through the recording apparatus, possibly due to s slight non-circularity in the driving mechanism. So over a long distance, it was running at exactly 24"/mile, but at times it was moving fractionally faster and at times fractionally slower. The impact of that would be that if, hypothetically, you had a vehicle that was running objectively at an exactly constant speed, but then plotted the apparent speed from the dynamometer, you would get a slightly wavy line either side of the true speed. That would give very slight peaks and troughs that would be an artefact of measurement rather than the "real" speed. If, serendipitously, one of those peaks happened to just be at the point that Mallard reached its actual objective peak speed, it would (a) mean the peak speed would be anomalously high and (b) there would be a more pronounced acceleration to (and deceleration from) that peak than was actually true.

    I think the bottom line is that in both cases (CoT and Mallard), the figures that have entered the popular imagination (102.3mph; and 126mph) have simply been quoted to a greater number of significant figures than is justified by the limits on the measurement technique chosen. 99% of the time that doesn't really matter; however, in those two specific cases, it matters, because in each case a significant landmark was being claimed that could only stand up if the highly accurate figure is accepted. For example, given the likely errors in measurement, maybe CoT did 100mph +/- 2.5; and Mallard did 125mph +/- 1mph. Those errors would encompass the claimed speeds, but would also encompass speeds that lost them the claim to fame: in the case of CoT the lower limit would mean it didn't do 100mph; in the case of Mallard, the lower limit would mean it wasn't the fastest ever steam locomotive. So the issue of errors in measurement becomes much more fraught than in most other cases.

    The one caution of course would be that an analysis of error bars only makes sense to prove or disprove the claim of, say, Mallard, if you also do an equivalent study on the German locomotive. I don't know the circumstances of that latter run, but if you are going to make a claim that Mallard maybe didn't actually achieve a speed above 124.5mph, it is a bit meaningless if the 124.5mph claim for the German locomotive contains a similar level of optimism as the 126mph claim for Mallard.

    Tom
     
    MellishR, andrewshimmin, Jimc and 5 others like this.
  11. Big Al

    Big Al Nat Pres stalwart Staff Member Moderator

    Joined:
    May 30, 2009
    Messages:
    21,065
    Likes Received:
    20,773
    Location:
    1016
    Heritage Railway Volunteer:
    No I do not currently volunteer
    Yes, I get that and it's possible that anything mechanical may always have that margin of error but you have to take on trust that for a key speed measurement the equipment was checked and calibrated beforehand. This was the LNER after all. Actually, in the case of Mallard, 125 would have secured the higher maximum although I can see that 126 put clear water, as they say, between the UK and Germany. In the instances of GW speed claims, I remain very sceptical for engineering capability and human reasons, with the greatest of respect to Rous-Marten. And we haven't even started to discuss the fact that mileposts (as now shown by GPS positioning data) are not all (and therefore were not all) at precisely the correct location.
     
    S.A.C. Martin likes this.
  12. Spamcan81

    Spamcan81 Nat Pres stalwart

    Joined:
    Aug 25, 2007
    Messages:
    35,121
    Likes Received:
    20,773
    Occupation:
    Training moles
    Location:
    The back of beyond
    Heritage Railway Volunteer:
    Yes I am an active volunteer
    Gresley only claimed 125 for Mallard. It was his successors who re-examined the data and went for the 126. Like you, I accept that there is a margin of error but I refuse to accept that the margin for a calibrated dynamometer car is equal to or greater than the human eye and stopwatch, especially given the known inaccuracies in milepost spacing.
     
  13. Copper-capped

    Copper-capped Part of the furniture

    Joined:
    Apr 19, 2017
    Messages:
    2,563
    Likes Received:
    3,316
    Gender:
    Male
    Location:
    Stanthorpe, QLD, Australia
    Heritage Railway Volunteer:
    No I do not currently volunteer
    Fact: Every engineering job to come out of Swindon was more exacting - chains included. ;)
     
  14. sir gilbert claughton

    sir gilbert claughton Well-Known Member

    Joined:
    Jan 17, 2017
    Messages:
    1,061
    Likes Received:
    511
    Gender:
    Male
    Occupation:
    retired
    Location:
    east sussex
    Heritage Railway Volunteer:
    No I do not currently volunteer

    yep , I realise that , but a loco travelling forward with the gear just in reverse , and the regulator cracked open , but not enough to drive the wheels , would give the effect of a counter pressure loco - or would it not ?
    whether anyone could do this in an emergency is open to debate ,but it is probable somebody has tried it in a non emergency situation to see what happened
     
  15. LMS2968

    LMS2968 Part of the furniture

    Joined:
    Sep 1, 2006
    Messages:
    2,987
    Likes Received:
    5,084
    Gender:
    Male
    Occupation:
    Lecturer retired: Archivist of Stanier Mogul Fund
    Location:
    Wigan
    Heritage Railway Volunteer:
    Yes I am an active volunteer
    Yes, it would act like that, and yes, it was done in emergency. I have a photo of two 8Fs in the Middle East following a head-on collision. Was is definitely in back gear.
     
  16. Jimc

    Jimc Part of the furniture

    Joined:
    Sep 8, 2005
    Messages:
    4,052
    Likes Received:
    4,665
    Occupation:
    Once computers, now part time writer I suppose.
    Location:
    SE England
    Heritage Railway Volunteer:
    No I do not currently volunteer
    That's not really the best way to do it. Mr Pines can correct me if needed, but GPS isn't measuring speed directly, its measuring position, and the instrument generates a speed measurement from the positions. These may be averaged, depending on the instrument, but you still don't want to work from them. Ideally what you do is set the instrument to log all its positions and times, and then process the data later. When you check the logs and plot the positions, you'll almost certainly find that, instead of having a nice straight line of plots exactly on how ever many feet one is sitting above track level, the plots are scattered left/right/above/below where they should be. This matters because the distance between a plot thats 15m to the left of the track and the next one that's 10m to the right of the track and 5m above is rather greater than the distance between where the instrument actually was when the measurement was taken. Almost every plot will be inaccurate, but some more so than others. Given the log you can average the plots over a large enough distance that the errors are insignificant, but if you just work from the individual speed plots then a large chunk of the measurement error still remains and the measured distance is liable to exceed the real distance travelled, so the speed recorded will be on the high side.

    To give you an idea here's a partial trace (only one I have handy) from a racing boat doing 3 laps of a course. Its more difficult because it doesn't travel in a straight line anyway, but you can see how much the trace is jiggling about through measurement errors. The error is going to be less with something travelling 10 times faster so the points are ten times further apart, but the point remains. Every form of measurement has errors, and there are always error bars.. That's why most speed record bodies won't accept a new speed record unless there's a certain margin above the old one - 1% with aviation I believe.

    new-1.jpg

    A smartphone will record a GPS log given the right software, and further software can be used to convert the log to graphs or charts like the one above.
     
    MellishR and Forestpines like this.
  17. Forestpines

    Forestpines Well-Known Member

    Joined:
    Jun 5, 2009
    Messages:
    1,681
    Likes Received:
    2,438
    Gender:
    Female
    Location:
    Somewhere in the UK
    Heritage Railway Volunteer:
    Yes I am an active volunteer
    This is all very true - position is the primary thing being measured and speed is derived from it. Well, if you want to be completely pedantic, time is the only thing GPS devices measure - it's the difference between the time signals that is used to geometrically compute position, but there you go.

    The big caveat if you deal with speed directly is that you don't necessarily know how different devices compute it. Some might use additional data such as an accelerometry sensor, so that jumps in speed that are incompatible with the measured acceleration can be smoothed out. Some might be sampling position at a higher frequency than they output it. Some might take the HDOP (horizontal dilution of precision) figure into account in speed smoothing. In applications such as rail and road they could potentially be using some degree of location-snapping, which of course requires an accurate map. In short: unless you have a friendly manufacturer, you don't really know how the speed is being computed.
     
    MellishR likes this.
  18. Jimc

    Jimc Part of the furniture

    Joined:
    Sep 8, 2005
    Messages:
    4,052
    Likes Received:
    4,665
    Occupation:
    Once computers, now part time writer I suppose.
    Location:
    SE England
    Heritage Railway Volunteer:
    No I do not currently volunteer
    You really ought to take the actual measured speed as the centre of your +-. If we are being really pedantic the measured speed of COT over the last 1/4 mile before the brakes came on was (to one decimal place) 102.3 +- 1.2, and the measured speed over the last 1/2 mile 100mph +-0.6. If, as Rous Marten did, you accept the possibility of a one click (1/5 second) error then the two measurements are 102.3 +1 3.5 and 100mph +- 1.8. To my mind that means that a record speed of in excess of 98mph is as well authenticated as any rail speed record from that era is likely to be. If the train had been braked 9 seconds later and RM had recorded the next quarter mile at 9sec quarter mile then the 1/2 mile comes to 101.1 +-0.6, and we'd have been saved all that controversy!

    It probably needs reiterating that I don't believe that 98+mph down a 1:80 bank is anything like as great an achievement as 99+mph down a 1:200 bank.

    Of course the +- are far smaller for the later Dyno Car measurements, but they still exist.

    By contrast the Lady of Legend signal box timing of 135 timed to nearest 30 secs is +- around 35mph at the least,and we don't even have enough data to construct error bars for the footplate timings, but its hard to imagine it could be less than 120 mph +- 10mph. Its a fun tale, but as Collett said, there's nothing there that constitutes a record.

    @Courier , back in 2012 ( http://www.national-preservation.com/threads/city-of-truro-maximum-speed.29445/ ) you mentioned you'd written up a detailed analysis for a magazine, but didn't feel it appropriate to make it available whilst the original was in print. Are you in a position to make a copy available?
     
    Last edited: Oct 18, 2017
  19. S.A.C. Martin

    S.A.C. Martin Part of the furniture

    Joined:
    Aug 31, 2010
    Messages:
    5,591
    Likes Received:
    9,325
    Gender:
    Male
    Occupation:
    Asset Engineer (Signalling), MNLPS Treasurer
    Location:
    London
    Heritage Railway Volunteer:
    Yes I am an active volunteer
    Yes. The LNER dynamometer car was regularly calibrated. To the extent that Gresley ordered a new process for calibration when the results of one particular experimental investigation did not match his expectations. That was the point - the car was specialised equipment and there were select individuals who were called on to look after it, calibrate it, test it and run it.

    Hence my extreme skepticism that the LNER dynamometer car was in some way woefully inaccurate. We're taking about plus or minus potentially 1mph, not 10 or 20mph. If Scotsman attained only 99mph then it will still have done so with more accuracy and scientific practice than CoT or the 135mph Saint.

    If Mallard only reached 125mph in reality, then the dynamometer for what it was, is still by far a more accurate instrument than human sight, mileposts and a stopwatch.

    What I really object to in all of this is that there is somehow an equal footing between how the LNER and GWR recorded their speed "records". There isn't. The LNER's method was clearly by far more scientific and accurate.

    Calling their records into doubt casts a poor light on a group of very dedicated and well trained, intelligent individuals who ran the testing for the LNER - as opposed one time keeper with a stop watch.

    I hope my position can be seen clearly here.

    Yes - there were three dynamometer cars to be used. LMS, GWR and LNER cars. All three cars were found to require calibration before the tests. This is not unusual and not actually concerning.

    Unlike the 1948 exchange trials, the three dynamometer cars might go several weeks - sometimes even months - without testing. The 1948 exchanges kept the cars and their staff continuously in employment throughout the period which was out of the ordinary. Hence, as Cecil J Allen explains in "The Locomotive Exchanges", regular calibration was necessary.

    So the idea that the dynamometer car for the LNER was in any way wildly inaccurate is fairly preposterous. If it were, it would have been reported as such in the 1948 trials. As it was, it was the most used of all of the cars and no mention has been made of any dissatisfaction with it (which you would have expected, if such discrepancies existed).
     
    Last edited: Oct 18, 2017
  20. Spamcan81

    Spamcan81 Nat Pres stalwart

    Joined:
    Aug 25, 2007
    Messages:
    35,121
    Likes Received:
    20,773
    Occupation:
    Training moles
    Location:
    The back of beyond
    Heritage Railway Volunteer:
    Yes I am an active volunteer
    Where's the "round of applause" emoji when you need it?
     

Share This Page