Steam speed records including City of Truro and Mallard

I think what's challenging is seeing the evolution of the ideas in real time, without the whole hypothesis to test or comment upon. As a layman, I'm finding this interesting even if it is well beyond my scientific/engineering depth; but do wonder where the questions take us.

Given that a PhD has to be original research, I would hate for a discussion like this to undermine the originality of what you propose to do - while also struggling to see the detailed analysis of that roll as more than a component of that wider thesis.

Your previous work has done sterling service in blowing through the fog of 70 years of partisan reportage; that is something that can be challenging to encounter when previously solid facts prove shaky.

 

I doubt if it is intended to be 'more than a component'. In my experience a lot of theses are good pieces of work woven together more or less successfully. What we don't know yet is what the big picture is into which this episode will fit. At this stage, less than 5% in, that probably doesn't matter too much. More important that the peers say 'yes that looks a good piece of work which takes us forward'. Academic journal publication is the gold standard test of that.

 

I don’t think anyone is suggesting reinventing it though…

 

I thought @sacmartin had dealt with this a couple of pages back - the LNER crew in the dynamometer car did what they’d been tasked to do with the data. There’s loads of other stuff they could have done, such as what is now being proposed, but they didn’t because they could do what they needed and move on.

The point isn’t about second guessing the LNER, it’s about building on their work in a way that no one else has done in the intervening period?

 

Are we at cross purposes? What designed in feature are you referring to? The apparent variation of speed above and below a smooth acceleration, as shown in Andrews' graph "Instantaneous Speed as Percentage of Average Speed over One Mile" is not noise. It is far too consistent for noise, over the whole 11 mile distance covered by the graph. I assume (but am open to correction) that the speeds in that graph are those calculated by the LNER from the distances covered in each 5 second period. Are you arguing that your calculations, based on the distances covered in each one second period, do not show that same systematic variation (although they would surely show considerably more noise)? Or are you arguing that what is shown for those 11 miles disappears for other sections of the run?
'

 

Just as an aside, having travelled at 94+mph behind Bittern in a Mk1 a few years back, I can't help wondering what it must have been like in the 30 year old dynamometer car with a clerestory roof behind Mallard at 125mph on non-continuous track. It must have been challenging to stabilise the measuring equipment and the pens so that the traces on the graph could be interpreted to a few thousands of an inch. Anyone know how they managed that?

 

I'd like to start by pointing out that I am a carpenter who knows he is out of his depth here. Some people contributing to this thread are frighteningly clever.
However... I started pondering the same question as Jim. The 5th/9th wheel looks to be about 4' diameter, so its going to make 420 (ish) revolutions per mile. Possibly it is sized to give a an easily manageable number of revolutions....360 0r 400, but I cannot find the actual diameter. Whatever that number of revolutions, there is some kind of gear train to ultimately end up with a mechanism that moves a roll of paper at 2' per mile- yes?
Andrews suggests that there is a 1 per mile sinewave.
Hermod says there is also a 4 per mile sinewave.
Tom says a sinewave indicates a circular motion.
I can't quite imagine where, in that gear train a 'one rev per mile' gear wheel would be. Its fairly specific. It was not difficult to make precision gear wheels for chronometers etc, so how could anyone allow that one very specific gear to be flawed? If 1 mile =24" , and that gear were to turn once ...24/pi is 7.64" which would be a very big cog in a precision mechanism.
But there is this:


One rev of a slightly fatter roll of paper would be 24". square drive, or 4 securing screws would be one per 1/4 mile.....
According to the discussion here:https://www.lner.info/forums/search.php?author_id=293&sr=posts, whence came the images, this roll is associated with the LNER.

 

IMG IMG does not show and link does not go nowhere
Please try again.
Sounds maddingly interesting.
My idea was something with uneven spacing of the cog holes in paper if it feeds like a cine movie film,but this will need to many coincidences.

 

With all this discussion of possible inaccuracies in the dynamometer car scrolls for mallard’s run, why is nobody questioning the accuracy of the recording equipment for the German record run?

 

Probably due to the title of this thread, Tom.

 

I’m amused because I have said repeatedly on this thread a few things that keep being ignored.

I want to know where the contradictory statements on the operation of the dynamometer come from.

The seconds on the graph are controlled by chronograph. The chronograph isn’t powered by the fifth wheel, as I understand it.

The distance pen is, and distance on the graph is shown by square waves with specific forms for each quarter, half, three quarter and full mile.

The wave form for the seconds drawn gets wider as the speed of the train goes up.

The wave form for the distance drawn gets smaller as the speed of the train goes up (in other words, the inverse of seconds).

People keep saying to me that the length of the roll going though the paper equates to a set length of roll per mile.

But that cannot be the case. We can evidence why by looking at the roll itself.

Measuring seconds against the known distance of a quarter mile, you get expected answers for instantaneous speed where roughly you know where similar calculations were done by the LNER team. E.g. if I calculate the quarter mile speed at what is likely milepost 91.5, by using the known variable of a quarter mile distance, and measure the seconds taken to achieve that quarter mile, I get 122.75mph. The LNER team measured 123mph around that similar point.

So we have different answers but they are “in the same ballpark” so to speak. Doing this for other known speeds produces similar results, albeit I think there will be some understandable variation, because I am trying to be consistent in what I am measuring, and the LNER team only did half and full mile calcs in order to prove the 125mph sustained, in 1938.

So I think there has been some misunderstanding about how the dynamometer car operates. Part of this I think lies in the fact that the LNER operating instructions are not preserved in full, as I understand it. I need to go back to the NRM and confirm this.

Which is why I have been sceptical (and remain so) of Dr Andrew’s work, because fundamentally I know how many plotted points on his graph he has, and my work is projected to cover roughly double the number of plotted points.

 

OK, now I am really confused.

On the one hand, you are saying that "it cannot be the case that [] there is a set length of roll per mile".

Andrews says "this showed the paper was moving at an average rate of 24 inches per mile with an error of less than five parts per ten thousand".

You can't both be right.

Broadly, there are three possibilities I can see. One is that you drive the paper from the wheel in connection with the track. That gives you a fixed length of paper per unit distance travelled, but the paper gets faster and slower in unison with the train. That is what Andrews suggests, to a very high degree of precision over quite a long length of chart (24 feet) and track (12 miles).

The second option is you drive the paper with some kind of clockwork mechanism, such that you get a fixed length of paper per unit time. The paper speed past the pens stays constant.

(There is a third option, which is the paper is driven according to neither, but the obvious response to that is why?)

You have remarked that the dyno roll has marks given at 1/4, 1/2, 3/4 and full miles, which presumably are driven off the "fifth" (ninth) wheel. So there is a very easy check you can do to distinguish. Measure the distance on the chart over a selection of quarter miles and over a selection of (say) 10 second intervals at different points on the chart. Ideally choose a couple of different regions where there is a significant difference in speed - say somewhere adjacent to the 120 mph point, but also somewhere earlier in the trace where the loco was travelling at less than 100 mph.

No need to do any calculations, or anything else - just measure how long in inches on the chart a 1/4 mile section is in the low speed and the high sped regime; and how long (in inches) 10 seconds is in the low speed and high speed regime.

If the chart is being driven at a constant length of chart per unit length of track, then the two 1/4 mile measurements will be the same number of inches on the chart, but the 10 second time interval will be longer (more inches) in the high speed region. By contrast, if the chart is being driven at constant length of chart per unit time, then the two 10 second time intervals will be the same number of inches on the chart, but the 1/4 mile will be shorter (fewer inches of chart) in the high speed region.

There's no need to calculate speeds (beyond broadly ensuring you are in quite distinct "high speed" and "low speed" regions, well separated). You don't even need to be precise to the thousandth of an inch if you choose regions in which there is a significant speed differential (say somewhere at 120+ mph and somewhere at below 100mph).

Tom

 

Exactly!

 

Well yes ... But Andrews is categorical about what he says about the chart being driven at a constant length of chart per mile. It is also clear in his blog post that (a) he had access to the chart and (b) he had a set of callipers to make actual measurements.

It is one thing to believe that he has misinterpreted the data, i.e. you disagree with his analysis. But you are saying that his data are wrong - so I think the onus is on you to disprove his statement that "this showed the paper was moving at an average rate of 24 inches per mile with an error of less than five parts per ten thousand".

It's a very easy measurement to make. Look at how long 1/4 mile is in the "90mph region" and the "120mph" region. If he is right, both 1/4 miles will be 6 inches of chart. But if, for example, the chart is moving at a constant time-rate, then 6 inches in the 90mph region will be 4.5 inches in the 120mph - you don't need a vernier gauge to measure that!

Tom

 

No, I’m not questioning his data. I am questioning his analysis of it.

I am happy with his plotted points. I am not happy with his analysis of it.

My point has always been that he’s not looking at all of the data on the roll.

You’ve posed three options there.

The one that is recorded by most academics on this, which is obviously the intention, is option 1.

What actually happens is option 3, derived from the intention of 1.

The chronograph gives a constant rate of time recorded in seconds and as the paper speeds up through the mechanism, due to the increasing speed of the train, the square waves get progressively longer. This makes sense: the driven paper is going faster through the mechanism.

In practice, you would expect then that the machine is calibrated to give that constant speed of roll paper and that it would give each quarter mile the same length, right?

That’s not what I have found. In fact in reality the distance between each quarter mile on the graph paper is going down as the speeds increase and increasing as speeds decrease.

This is a constant throughout the roll.

I think it is likely that the dynamometer car was obviously designed to run the paper at a constant rate. But in reality, the speeds being obtained mean that this doesn’t happen in practice. Bear in mind that this is the fastest the dynamometer car has ever been run, and normally it would not be expected to exceed speeds of 75mph on the mainline. It hasn’t been built for these speeds in the first place.

It actually doesn’t matter, though.

Provided your chronograph is indeed recording time consistently in one second intervals, and the driven wheel driving the paper is indeed recording each quarter mile as it should do - i.e actually recording the quarter mile mark on the graph at every actual quarter mile travelled - then you still have time recorded consistently against the distance actually recorded.

This is tested as I have said above, by checking various calculations to see if the new calcs match up to the existing ones. If they do, then you know that the approach works.

This is why when you look at the LNER team’s calculations on the area for the high speed record, you realise why they took 5 second intervals and calculated from there. They already knew that the mechanism was likely to change the length of the distance between the quarter mile marks on the graph, and they adapted how they did their calculations to compensate for this. If they did not know this in advance, they would not have adapted their calculations over the speed run section in this way.

They only wanted the average speed over those intervals to prove 125mph - all that was required was beating the German team. So their efforts and how they did it was tailored to this goal.

If they had had more time to analyse this - I think the discussions outlined in many secondary sources indicate there was a degree of urgency in setting up the run and in announcing the result - I think they would have realised that they could have plotted the entire run through the recorded quarter mile calcs and probably would have come up with the same result - 125mph - but moreover, as my research is indicating, 126mph - more easily.

But we’re not talking about huge changes in the lengths between the quarter miles recorded at the start of the graph and at the high speed ends of the graph - which is why I am confident in my interpretation and why I’m not convinced in the argument that there was an issue with the dynamometer car equipment. Rather, its limitations were known by the on board team who aimed for a set distance on the roll but knew that realistically, there would be some variation. But it doesn’t change the price of fish, as they say.

 

But the title of the thread does not limit it to CoT and 4468.

 

I would be interested in seeing the German dyno roll, if it still exists, as I think they must have been dealing with very similar issues when attempting their run. I cannot imagine that their method of recording time and distance would have been too different, if at all.

I think, given the variations between the lower speed and high speed variations in the LNER dyno roll are relatively small, and can be taken into account when calculating the average or instantaneous speeds, we can say with a degree of certainty that Mallard achieved 125mph and was faster than the German 01 loco. 126mph is down to various interpretations of that part of the dyno roll and for me, I am confident 126mph was achieved.

But there are issues with every interpretation of whether it did it or not! Accepting that the different approaches produce ever so slightly different results (in some cases, 1/2 to 1/4mph speed), you could argue reasonably that it doesn’t matter. Everyone agrees 125mph was achieved.

But again, I’m not as interested in that as I am the early parts of the run and that’s the bit I am focusing on, because that’s the bit everyone else has mostly ignored in the clamour for 126mph or not.

 

I know too little about the German run. But my understanding is that it carried passengers. Was it subject to a dynamometer reading?


Sent from my iPhone using Tapatalk

 

Your clarifications usually confuse me about what is your understanding of what is happening. This one suggests to me that your understanding is capricious. Can you please just answer Tom's questions as put?