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.

Sir Nigel Gresley - The L.N.E.R.’s First C.M.E.

Discussion in 'Steam Traction' started by S.A.C. Martin, Dec 3, 2021.

  1. 30567

    30567 Part of the furniture Friend

    Joined:
    Aug 7, 2012
    Messages:
    5,612
    Likes Received:
    3,512
    https://rchs.org.uk/wp-content/uploads/2021/03/FINAL-Wilson-LNER_2.pdf

    Yes, see page 43 of this paper. I think the finance scheme was effectively an interest on loan subsidy.
     
    S.A.C. Martin likes this.
  2. huochemi

    huochemi Part of the furniture

    Joined:
    May 6, 2008
    Messages:
    2,751
    Likes Received:
    1,393
    Gender:
    Male
    Location:
    UK
    Heritage Railway Volunteer:
    No I do not currently volunteer
    Much of the detail is in the Railways (Agreement) Act 1935. It is fairly complicated but in broad terms, the railways paid interest on loans from the Railway Finance Corporation Ltd (a government vehicle) at the rate at which the RFC raised debt from the market (which debt was guaranteed by the Treasury), which was 2.5% p.a., issued at 97. The railway companies were more or less obliged to borrow the monies they signed up for subject to a few get-outs like failure to get relevant approvals etc. I suspect the railways were not necessarily overjoyed about the arrangement, nor would they necessarily have had sufficient "shovel ready" schemes, but felt they had to go along with the goverment's wishes, and they could justify it by selecting schemes which they were probably minded to do anyway based on an analysis of the projected return (although they could hardly admit as much in public comments). As much as the reduced interest rate (v. the terms they would get on their own credit - probably at least double the cost), it was a convenient source of funds as I suspect for most if not all of them, there would have been a lukewarm response to a debenture issue by any of the Big Four at that time The LMS incidentally allocated part of their share of the loan to the purchase of 369 new steam locos.
     
    S.A.C. Martin, ragl and 30567 like this.
  3. 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
    A final successful lecture of the year for the RCTS last night and now all eyes on the editing copy of the book.

    I will make sure to update the thread when we have a release date.

    I'd like to thank everyone for participating in the debate, which has been for the most part helpful, friendly, and given us all a lot to mull on and consider.

    For my part, I have come out of the research and writing period again with a renewed respect and admiration for Gresley, whose skills and talents have been given fresh meaning to me.

    Until then, this is a last call for anyone who would like to provide feedback ahead of the book's publication - please get in touch if you'd like to read it. As always, you will be thanked accordingly in the special thanks section of the book.
     
  4. bluetrain

    bluetrain Well-Known Member

    Joined:
    Mar 3, 2019
    Messages:
    1,326
    Likes Received:
    1,460
    Gender:
    Male
    Location:
    Wiltshire
    Heritage Railway Volunteer:
    Yes I am an active volunteer
    That is an interesting point. I believe that the LNER concentrated loco building at Doncaster & Darlington with limited further building at Gorton, but discontinued building at Stratford and Cowlairs once outstanding Pre-Grouping orders were complete. Stratford did, however, carry out much rebuilding during the LNER period. Many ex-GER locos acquired Gresley-type boilers and round-top fireboxes, while the B12s and Clauds were given more extensive rebuilds with larger boilers and modified front-ends.

    Gresley placed fresh orders for a number of locos of GC and GE design. David Jenkinson ("BR Carriages of 20th Century" - Vol2 P14) wrote of similar action on coaches. The GE Section in the 1920s could not at first accept the new LNER standard 61ft 6in corridor stock (Reasons are unclear - possibly throw-over on curves against tight clearances?). Gresley's interim solution was to place an order on York carriage works to provide the GE section with NE-design 53ft 6in carriages, pending the production of shortened versions of the LNER standard designs. Jenkinson speculates that Gresley may have had a political motive here - to help get the ex-NER York establishment (the LNER's largest carriage works) "on-side" with the new regime.
     
  5. 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
    So yesterday I began the task of examining Mallard's dynamometer roll. It is a truly humbling thing, seeing the roll itself for the first time. I am very grateful to the copy service team at the NRM for reproducing it for me.

    (Please don't ask me to post it publicly, I am restricted by the terms of the license).

    There's a very interesting blog on the iMechE site here: https://imechearchive.wordpress.com/2020/07/03/guest-blog-a-matter-of-time-and-space/ which gives an interpretation of the Mallard dynamometer roll, from David Andrews of the iMeche. In short, he doesn't believe Mallard attained 126, but does believe it attained 125mph and also demonstrates that it sustained at least 124mph for around 3/4 of a mile.

    Key to his analysis is this section here:

    When I started looking at the roll myself, I had a few things in mind.
    • The dynamometer car records one second intervals
    • It also measures the quarter mile automatically
    • Milepost positions are not recorded automatically but added later
    • The LNER speeds are taken as an average over five second intervals
    The actual roll looks something like this, which is David Andrew's graphical interpretation of the roll:

    The LNER assistants never actually wrote 126mph on the roll. 125mph is written, but there is a mark showing where they believed 126mph was achieved momentarily.

    So my thoughts are:
    • The LNER records don't use the automatic quarter miles on the chart, they used five second intervals to measure an average speed over that time frame.
    • Knowing that the dynamometer car automatically records quarter mile stretches, we can use these datum points instead of that the LNER assistants and David Andrews have used.
    • Using the speed/distance/time equation, we can then use our trusty excel spreadsheet to plot a curve showing the speed at each quarter mile recorded (using a formula to convert seconds to hrs:mns:secs to get mph)
    So in short, my approach would show every quarter mile attained and the speed at each quarter mile, instead of (as per everyone else) the speed at each milepost and then for a brief period, each quarter mile and five second interval over the 126mph claim.

    The +/- for the error in recording on the paper was taken as a 1/4mph either way, so for example if you had 126.25mph you would say it was a range of 126mph to 126.5mph.

    Does that seem like a logical approach?

    If it does, please find below my initial interpretation of the speed curve, to be peer reviewed.

    upload_2023-5-26_8-24-50.png

    Note that the bottom axis only represents the number of points on the graph, and not the distance between them (I am trying to work out how to represent that properly over the next few days).

    So my speed curve resembles David Andrews' curve quite closely in the shape and didn't have the noise in the data that he encountered by re-measuring the LNER's approach.

    My curve also presents a higher top speed than 126.2mph and shows that Mallard likely sustained at least 126mph for around 2/3rds of a mile, unlike the conventional wisdom which suggests that it was for a very short period.

    My suspicion has always been that the Dynamometer roll showed us a different story since I learned of how it actually records the data. When you remember that we are dealing with an analogue device in the hands of very clever men, but they are working without the advantages we have of instantaneous calculations via electronic devices, you can see how re-examining the data could give a different result.

    Anyway, it has convinced me that the 126mph claim should stay put, for certain. The dynamometer roll does show this was achieved.

    Whether you think it was achieved for a short period of time, or a longer one, is dependent on how you interpret the data contained on the roll. For my part, I have outlined my approach and if you think that's logical and acceptable then it does show the achievement claimed.

    My next bit of work, to do after I have finished my work for today, is try and plot the LNER's curve against mine on the same graph to see how similar they are to one another.

    I agree with David Andrews that the real achievement is in fact maintaining over 124mph for a sustained period, though he puts it at half a mile and I have something closer to a mile.

    Bear in mind we are using different datum points from the same graph, and my time measurements between the quarter miles do not equate to the same milepost approach the LNER assistants used.

    I thought this might be of interest!
     
    Spinner, ragl, RLinkinS and 5 others like this.
  6. maddog

    maddog New Member

    Joined:
    Apr 7, 2011
    Messages:
    192
    Likes Received:
    89
    This is beyond my head at the moment, but have you been in contact with David Andrews of the iMeche? He seems like most important person to converse with on this.

    does this mean your method avoids the error he highlights?
     
  7. 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, this has crossed my mind and once I am happier with my graphs I will drop him a line through the iMechE.

    I believe so.
     
    maddog likes this.
  8. Miff

    Miff Part of the furniture Friend

    Joined:
    Jun 17, 2008
    Messages:
    2,857
    Likes Received:
    2,793
    Interesting. Makes me wonder since the dynamometer automatically measured 1/4 mile intervals why did the LNER (and Mr. Andrews) rely on the, manually recorded, milepost observations - or why did they record them at all? If it was intended as a calibration check how would LNER’ve compared it at the time? In any event the three methods of analysis seem to show very similar results, all agreeing 125mph or higher, suggesting LNER’s methods & conclusions were fairly robust for that time.
     
    Last edited: May 27, 2023
    S.A.C. Martin likes this.
  9. 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
    It’s not quite as simple that - my apologies, I must clarify that where the speed record is concerned the approach was 5 second intervals taken from the roll with the mileposts used as reference to when the speed of 126 mph was actually attained. These are average speeds over that interval.

    The advantage of using the dynamometer car automatically recorded quarter mile posts is that you can calculate the speed at that point since you know the distance covered and the time from the last quarter mile.

    I think the approach taken between the LMS, DRG and LNER was consistent with one another which explains the use of the average speed over 5 second intervals.

    You get similarly shaped speed curves for mine and David Andrews’ approach which confirms the robustness of the original rolls data to some extent. The difference in top speed is down to the position you take on the roll.

    You could argue, legitimately, that my approach could be invalidated by the use of scans and digital measuring tools, and that’s perfectly fine too. For my own peace of mind, I feel the evidence suggests strongly that 126 mph was maintained and not just touched. It doesn’t change the record books or the plaque on the loco.
     
    Miff likes this.
  10. 30567

    30567 Part of the furniture Friend

    Joined:
    Aug 7, 2012
    Messages:
    5,612
    Likes Received:
    3,512
    With an asymptotic curve like that it doesn't make much difference, but it is perhaps worth noting that your intervals are effectively 7.5 seconds. This would matter with a sharper peak.

    I haven't got my head around the sentence beginning 'note that.....'. Why isn't the horizontal axis simply distance with the observations in quarter miles? I must be missing something there.

    Obviously the size of the assumed data error term is crucial when using words like 'for certain' and 'strongly'. What is the rationale for it being 1/4mph rather than 1/2 mph?
     
    S.A.C. Martin likes this.
  11. 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
    That’s not true, the intervals vary. That’s the point. The dynamometer car records the distance automatically and this is set against one second intervals on the graph paper. Therefore the interval between each quarter mile post goes down as acceleration goes up. The interval is set by distance, not time.

    As I said earlier I am still playing around with how best to show it. As the distance is set but the time isn’t it doesn’t show all of the data.

    I took the same approach as the LNER which was to apply 1/4 mph +/- to their results which presents a range.
     
  12. 30567

    30567 Part of the furniture Friend

    Joined:
    Aug 7, 2012
    Messages:
    5,612
    Likes Received:
    3,512
    Thanks Simon, yes agreed on the first point. I should have said 7.5 secs at the peak speed which is the speed of interest.
     
    S.A.C. Martin likes this.
  13. Jamessquared

    Jamessquared Nat Pres stalwart

    Joined:
    Mar 8, 2008
    Messages:
    26,105
    Likes Received:
    57,438
    Location:
    LBSC 215
    Heritage Railway Volunteer:
    Yes I am an active volunteer
    I think the point is that Simon has plotted equal time intervals along the axis, not equal distances. (i.e. each point is 5 seconds). Since the speed varies, the distance represented by each point varies. (Nothing wrong with that, but I think we are accustomed to thinking that such charts would have distance along the axis).

    For plotting purposes, I suspect the answer to Simon's conundrum is to set a datum, and then work out how many yards the loco travelled in each 5s increment, and therefore where it is relative to the datum. Your data set would then, instead of looking something like:

    1 - 120
    2 - 121.5
    3 - 122.3
    etc etc

    would instead look something like

    0 - 120
    347 - 121.5
    712 - 122.3
    etc etc

    (I made this number sup, rather than calculate, just to show the point, i.e. the x axis is going up in uneven increments).

    That would then make the graph with the same data, but with the x axis as distance rather than time.

    Playing devil's advocate, if two people analyse the same source data and get similar results, it doesn't validate the source data: it is basically inevitable because they are looking at the same data! I'm not questioning whether the data are robust or not, but if you want to make a claim about the LNER dynamometer car data being robust, you can only do that by comparing with an entirely independent data source. You can't infer anything about data quality by performing two different analyses on the same data.

    Tom
     
    Kje7812, S.A.C. Martin and Miff like this.
  14. 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'm not saying it is the case here but 1/4 mile posts can be notoriously inaccurate. The only value in recording them on the chart is to give a broad reference as to where the train is. At the end of the day you don't just want to be able to say that the loco reached 126 mph 'somewhere'.
     
    Hirn and Miff like this.
  15. MellishR

    MellishR Resident of Nat Pres Friend

    Joined:
    Apr 16, 2009
    Messages:
    8,069
    Likes Received:
    5,165
    Being pedantic, in some instances you can infer that the data quality is poor, for example if a train log were to show implausible acceleration. But I agree that multiple analyses of the same set of data cannot confim that it is good.
     
    S.A.C. Martin and Jamessquared like this.
  16. 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
    Hi Steve - the roll records quarter mile travelled, not the same as mileposts.
     
    Hirn likes this.
  17. 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 IMHO a very good piece of analysis by David Andrews. I particularly like the way he's picked up the cyclical error in the paper speed underlying all the other factors. I wonder if one could look at the mechanism - if it survives in the same form - and try and identify what components might be responsible. A really good statistician and engineer (ie not me) might be able to establish even more about inaccuracies in the mechanics.

    Also important are his questions: What is the error band? How capable was the measurement system...? If we apply those questions to, for example, Collett's notorious exploit with Lady of Lyons, we get the answers "Don't know, but very wide" and "Not very", so, like Collett, we don't claim it to be a record. If we look at Rous Marten's records of City of Truro, then, excluding those who hold the opinion that Rous Marten completely stuffed up/faked his measurements, we get about a 2mph error band for most of the timing, but about 4mph for the last one.

    Simon, I'm a bit nervous about your approach. Andrews had demonstrated, at least to my satisfaction, that we are seeing at least 1 - 1.5% noise. We have a pretty good idea how fast that train could be accelerated (not very) so any lumps and bumps in the trace that exceed that sort of acceleration *must* be noise. Averaging is of course a crude way of eliminating noise, but it does have the disadvantage that if you pick a region with the right random errors you will get a falsely high (or low) average. Andrews has demonstrated a 1 mile cyclical error. The implication of this is that any average over a distance of less than a mile is capable of being artificially high or artificially low. So if one goes hunting for the best 1/4 or even 1/2 mile the result will inevitably be an artificially high value. So I think the first thing one has to do with any analysis is to make Mr Andrew's correction for the mechanism error. I don't believe looking at averages, even over a different distance, is helpful unless you do that correction first.

    Jim C
     
  18. 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
    Hi Jim. I’m not taking averages.

    I think I’d better put together all of my data and do a proper explanation later this weekend.
     
  19. 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
    As I understand it, David Andrews has demonstrated that the recording paper was not moving at a constant speed, but was slowing down and speeding up slightly at one mile intervals. His chart demonstrates this, and suggests that the first two quarters of each mile are recorded too slow, and the third and 4th quarters too high, with the 4th quarter probably a bit higher than the third. Given 1/4 miles in the 7.2 second region, that also means that the LNERs analysis of 5 second averages would tend to a mix of slower and faster times in varying degrees, but not showing such a regular pattern as the 1/4 mile times.
     
  20. Hermod

    Hermod Member

    Joined:
    May 6, 2017
    Messages:
    985
    Likes Received:
    283
    Gender:
    Male
    Location:
    Klitmoeller,Denmark
    Heritage Railway Volunteer:
    No I do not currently volunteer
    Is mr David Andrews somehow related to the gentleman who wrote as Courier here around 2017?
    Searching here for Mallard and author Courier gives some very interesting pages.
     

Share This Page