Steam speed records including City of Truro and Mallard

I don’t access the Forum for a day and all hell breaks loose but, at last, we are getting there. I’ve been questioning the chart drive for some time and Tom has picked up on what were really my unanswered questions. The vibes I’ve got are that ‘we’ have no idea of how the dynamometer works and that includes Simon. Until we do, no sensible analysis of the chart can happen. It is all guess work based on a little knowledge and a lot of ignorance.
I did suggest a while back the possibility that the take up spool was driven by the fifth wheel and hoped that it wasn’t. I’m beginning to wonder if this is actually the case.
We need to know exactly how the chart is driven and how the distance pen is driven. We also need to know the same about the time pen but this is almost certainly by some form of chronometer. We also really need to know what type of gear train is employed in driving whatever it does. Are they relatively crude open gears or machined precision gears running in an oil bath and is there any backlash elimination system.

 

Andy - I’ve explained why working from the roll itself isn’t possible. My copies were done by trained NRM staff who have provided a scale for the photographs, and the quality of the photographs is incredibly high.

Given my circumstances - in work and also doing a PhD - and the needs of the museum against the base cost of having access to the roll for an extended period of time, I am doing the best I can.

 

Steve, I’ll send you a PM in a day or so. I’m frankly a bit fatigued.

 

As some of us see it you have kept saying that your examination of the roll (or the Museum copy) is telling you different from what Andrews saw, but not telling us just what you are seeing that is different.

 

As Simon has pointed out, the roll does have one trace showing actual time elapsed and one showing actual distance travelled. Provided those are not seriously irregular, the exact speed of the paper, and whether it is constant or varying, doesn't matter.

 

It doesn't, it has one showing measured time and one showing measured distance.

 

I think that's overstating it a bit. The roll will have a lot to say that hasn't necessarily been published before without going to that level.

Yes, the gold standard would be a very high resolution flat bed scan of the roll. If you want to go down to the absolute limits of what the measurements tell you, establish the tolerances of the measurements, analyse to as many decimal places as possible that's what you need to have. With that, and a really talented graphics analyst/programmer you could go down to every bit of detail that's there, locate the centre of every ink line to a couple of decimals of millimetres, all that. Note that for this to be useful you would need to only that super high res image, but also some very high end talent on the analysis that is certainly way beyond my capabilities, I don't know about the rest of you.

Simon is telling us he has good quality photographs, presumably from a rostrum camera - at least that's how I would do it. That pretty much precludes super high end analysis because of lens distortion. I suppose sufficient talent could minimise that, but its still there. So that pretty much eliminates the last decimal or two that would be possible to achieve with the first method, and the error bars have got a lot wider. That's wider relative. The difference between +-0.1% and +- 0.001% is a gulf in terms of accuracy, but for most analysis its neither here nor there. And you have to make a few assumptions. The pen overrun ticks, for instance. With the high res flat bed scan you could locate those to a fraction of a millimetre, and discover how much variance there really is from the flow of the ink, bumps in the track etc. Without that the best you can do is make an assumption they're accurate, which of course they aren't, but I'd assume they're probably good to ( I dunno, I'm not expert at this stuff) lets say within quarter of a line width, which isn't too clever measuring individual ticks, but is just fine for an average over say ten.

Perhaps this is best illustrated by a graphic. I've taken the wave form enlargement from Andrews' iMechE paper, and assuming its a genuine scan of the roll. (It may not be of course, the rest of the graphic is clearly his own).

What I've done is to import this into a vector graphics programme and draw by eye, and as well as I could, three lines based on three different parts of that individual tick. They're all I think reasonable points to consider, the rise where it crosses the zero, the point of the tick and its centre. As you can see each measurement is different (the units are irrelevant), but they're all within a close range - they actually vary over about half a percent. Its also unsurprising that the measurement to the centre of the tick is closest to mean. Its the most difficult to judge, but probably the most useful. As this is just one sample and others will be better or worse, I suggest its not humanly possible to measure these ticks (if they are indeed genuine from the roll) individually to an accuracy of better than 1%. If you do 5 ticks at a time I reckon it goes to 0.2% and so on, so there's the power of even short distance averages.

All this means that short distance averages are well within the capabilities of what Simon has got, and although he can go down to measured speeds against individual ticks, even this crude analysis of one variable starts showing an error bar.

OK, so where does this get Simon. It means, I think, that trying to establish a point speed is futile, and I can almost hear Gresley's shade shouting in my ear "Of course it is, that's what I said all along". On the other hand it means that given sensible short distance rolling averages he can tell us a great deal about acceleration, power outputs, all that sort of thing analysing the run to an extent that certainly hasn't been published before. All that is valuable. I must admit I'll also repeat my suggestion for getting a GPS log of the route and seeing if we can get a good idea of the real gradients as opposed to the approximations used on the route chart.

 

It actually displays a lot more than that, those are just two of the wave forms that we have discussed.

 

Thank you Jim. Appreciate this erudite analysis.

 

The polished roll command the paper movement under pens and is synchronized by gears to fifth wheel.Be they eccentric or not.
The take up roll is driven very slip able from the polished command roll by a crossed sewing machine round belt.One of the sheaves can alter effective diameter and thus paper take-up tension can be kept under or on belt slip but not over.A very clever design methinks

 

Being, I freely admit, something of the geek/nerd (I once described myself in a business meeting as the alpha geek) I couldn't resist going a bit further, and measuring the half wave in the same way. (Remember, though, that we do not know this is a genuine extract from the roll, Andrews may have reconstructed it). Now the first thing you may notice is that the crest of the top tick is displaced to the right. That's the sort of thing that you get with ink on paper, and why you have to be careful.


OK, now I put those numbers in a spreadsheet.



You can see how that distorted crest distorts the measurements and we now have a much bigger variation. But even without that it appears that the first half of the wave is more than 50% of the total, so this upwave is shorter than this downwave. So, (still assuming that this is a genuine extract which it may not be) we've established what could be called an inaccuracy in the mechanism. But an inaccuracy that is quite irrelevant for the main use of the machine and even for any reasonable analysis work with it. But it does nicely highlight the point that no machine is perfect and we need to be very aware of the limitations of what we are dealing with.

(added later) Not to mention, of course, my own limitations. I probably ought to wipe all those lines off my graphic and draw them again from scratch, get a new set of measurements, and establish the amount of error there is in my measurements!

 

It’s always been my contention that the likelihood of the dynamometer car being “inaccurate” or the mechanism damaged or worn in some way is unlikely. The Dyno car was very important to the LNER over three decades and was given regular calibration and had regular work.

The difference in what I am saying versus Mr Andrews is the nuance between assuming the mechanism is not working in the manner intended versus the mechanism working as intended but going beyond its design’s limits.

It ultimately doesn’t change the price of fish in terms of whether anyone but it does change the story being told.

 

Whilst superficially using GPS to assess a potentially more accurate gradient profile has merit in
practice latter day deep ballasting means the results are unlikely to reflect the position in 1938. As
with so much of this debate we are left concluding the information available in 1938 is the information.

Picking upon some other posts. David Andrews worked on the original role ( see I.Mech Eng and SLS
articles ), he utilised a set of caliphers and he has/had the benefit of seven years as an archaeologist.

With regard to locomotive performance, evap rates etc there are IMHO many other examples of
A4 high output performances that merit as much, if not more attention. The only unique sector is
from 110 mph upwards, ( and then we have indications of the cylinders beating the boiler )

I think the only area of investigation that may offer (big if ) is the proposed detailed examination
of quarter mile sections, but with a loco working near to its limits ( RFO and 40% ) the
variation from half mile calculations may be minimal.

For what my opinions are worth, I think Mr Martin could profitably reallocate his precious
time to an analysis of the three pre WWII LNER streamlined services, Financial, operational challenges,
effect on other services, reliability, image forming etc and long term benefits etc.

This will, some will be glad to hear, my last offering to this debate.

Michael Rowe

 

Who says I am not already doing that?

In any event, it’s up to me what I choose to spend my time on.

That does not mean he or his equipment or approach is infallible, any more than my approach and the equipment/software I am using is.

The last few days of this thread has put me in mind of some of the early days of the Edward Thompson thread. “Why are you doing that?” “You’re wrong, OS Nock said this” “What a fruitless exercise” “You haven’t got the experience to say that”

It would be nice if a few members of the forum could step back, take a deep breath, and maybe consider (in their rush to defend Mr Andrews’ work at all costs) that the bottom line I’ve given is that I’m taking a different approach, I disagree with his analysis, and will respectively work on my approach and release my results when I can.

Some have argued that I would get greater scrutiny and more robust responses from those assessing my PhD. That might be true, but they would have had the benefit of the full work and results before giving their robust responses.

I like to think I have tried to be fair in my responses, have given my views, shared a lot more of my work than I need to do, frankly, and to be honest, I feel like it has become (intentionally or otherwise) a huge pile on to my work and approach without fully understanding what I have at my disposal and taking without question Mr Andrews’ work.

That’s the bit I find most heinous. Yes, there’s been past studies of the dynamometer roll, but I know for a fact I am taking a different approach to an analysis. I have discussed this with my supervisor and other colleagues and they’ve agreed with my sample findings thus far. So there is academic merit in it.

No one has to agree with me, and that’s life. But in sharing my approach and work, I did think I was providing something of interest that added value. It is a shame to get to Sunday of this week and feel absolutely exhausted in discussing the work.

 

I entirely agree. Such an analysis plays to his strengths rather more than trying to extract data from an analogue record that cannot contain anything else worthwhile other than that already established.

 

That really is a most ridiculous statement, and couldn’t be further from the truth.

 

But that's why it's time to draw a line IMHO Simon. You have chosen to share an idea, a line of study, with the community here. Not surprisingly there are some clever, knowledgeable people who have engaged. At the very least you now know some of the lines of cross-examination you will have to face. Away you go to your desk and write your paper referring back to the wheat which you have found here among the chaff. Onwards and upwards.

 

Was it also UK made 1904?

 

My meaning should be plain, but as you seem to misunderstand it, let me be clearer. There is no more useful data to be extracted from the dynamometer record, but further analysis of what has already been extracted, might further illuminate A4 performance.

 

Some of it can be seen here
https://www.lner.info/forums/download/file.php?id=16468&mode=view
The vertical worm most likely come from a gear driven by a worm on fifth wheel.
one of the handles is most likely a clutch
Other side
https://www.lner.info/forums/download/file.php?id=16467&mode=view