Gradient Mystery

[Andy873]

It seemed a long time before I came across a gradient profile of interest to me, it was by John Marshall who wrote many books covering the L&Y network, and his work is held in high regard. This profile is in a section of one of his books from the early 1970's covering one specific branch line. I've independently checked all his facts, figures, costs and dates etc and all are correct, with that, I took his gradient profile to be accurate.

A couple of years later I came across a book concerning signalling and the boxes on the branch written by Chris Littleworth. Sadly Chris is no longer with us, but in one of his books written in the early 2000's, he too has a gradient profile. Knowing Chris' work as being very precise / accurate I just looked at the graph and thought "yes, they are same" - My mistake was I didn't read those gradient numbers.

Finally, a few days ago I came across the gradient profiles for all of the L&Y network produced by the Mechanical Engineering Dept. of the L&Y (Horwich) in 1889. Looking at the same gradient profile of interest, this is different...

This morning I decided to compare all three and they are all different from each other.

John and Chris both show 18 gradient changes, the L&Y only 11?
Also, John and Chris mainly agree with each other on the eastern section, but differ greatly on the western section.
One portion is listed by John as 1 in 64, by Chris, 1 in 1270, and the L&Y as 1 in 800!

Who would you believe if you wanted to draw your own profile of this branch?
Could the L&Y one be a more general profile?

Out of the three, I suspect Chris' profile will be the most correct, he was writing about the signal boxes which we know had gradient profiles for their sections.

 

[Rescars]

I have no specialist knowledge of this, but I wonder why the Mechanical Engineering Dept produced a gradient profile, rather than the Civil Engineers. Perhaps the Mechanical Engineers needed something against which to assess the suitability of different classes of locomotive for working various loads over different routes. In this case, a slightly simplified interpretation of the civil engineering reality might have sufficed.

This is pure speculation on my part and I suspect it is merely coincidental, but 1889 was the year in which the Armagh accident occurred - and one of the contributory factors was the suitability of the locomotive provided to work the heavily loaded excursion train. Given how profoundly this accident impacted railway operating practices, is it possible in the aftermath that the L&Y undertook that an urgent review of locomotive capabilities for different routes? In which case a full set of gradient profiles for the mechanical engineers would have been helpful. I wonder if the L&Y board minutes have anything to say on the matter.

 

[Bevan Price]

There is another source of L&YR gradients in Eric Mason's "Lancashire & Yorkshire Railway in the 20th Century". Not all lines are included (one being "your" North Lancashire Loop), but some of these differ from those in John Marshalls' books - some show fewer detail changes in gradient.

So I see several possibilities:
1. They are "averaged" over longer distances.
2. There are differences between the gradients estimated during the line survey and the line as actually built.
3. In some areas, the gradients have changed, for example due to mining subsidence.

 

[Pigeon]

I reckon I've seen examples of all of those on odd occasions when I've found myself comparing gradient profiles from different sources for the same line. In particular on some of the "theoretical" profiles there are sections with several gradient changes in the space of much less than a mile, usually labelled in illegibly small writing, which in more "pragmatic" renderings - and quite likely on the line itself over time - get averaged out to a single value.

The railwaydata.co.uk site is supposed to have gradient profiles from modern surveys but they often suffer from other inaccuracies which seem to be down to nothing but carelessness. The site's own javascript to render the data into gradient plots doesn't work because of course it doesn't, but it's straightforward enough to replace it with something that does; it then turns out that there are two separate sets of data for each line - an approximate one with average gradients between points spaced at regular intervals, and an "accurate" one with the gradients between the points where they actually change (ie. as usual) - and they don't match up. The "accurate" data is speckled with random corruptions where the sign of the difference between two successive points is wrong in either the y or the x direction, or digits in a number have been replaced by adjacent digits on the keyboard, so the plotted gradient profile has crazy zigzags and impossible loops where it turns back on itself for a bit and then continues in the right shape but displaced horizontally and vertically from where it should be.

 

[Rescars]

Presumably modern traction and rolling stock are rather more tolerant of modest inaccuracies in gradient mapping than steam hauled (and banked) unfitted goods trains.

 

[Andy873]

I have no specialist knowledge of this, but I wonder why the Mechanical Engineering Dept produced a gradient profile, rather than the Civil Engineers. Perhaps the Mechanical Engineers needed something against which to assess the suitability of different classes of locomotive for working various loads over different routes. In this case, a slightly simplified interpretation of the civil engineering reality might have sufficed.

Well it could very well be the reason, I have no specialist knowledge either.

There is another source of L&YR gradients in Eric Mason's "Lancashire & Yorkshire Railway in the 20th Century". Not all lines are included (one being "your" North Lancashire Loop), but some of these differ from those in John Marshalls' books - some show fewer detail changes in gradient.

So I see several possibilities:
1. They are "averaged" over longer distances.
2. There are differences between the gradients estimated during the line survey and the line as actually built.
3. In some areas, the gradients have changed, for example due to mining subsidence.

All valid possibilities, although I can't think why you'd go from one section being 1 in 64 by one person to 1 in 1270 be someone else, that's a big difference in a gradient.

The railwaydata.co.uk site is supposed to have gradient profiles from modern surveys but they often suffer from other inaccuracies which seem to be down to nothing but carelessness. The site's own javascript to render the data into gradient plots doesn't work because of course it doesn't,

Talk about inaccuracies on that site, I messaged them that it was wrong about the section of the branch line covered by them, but never got a reply, nor was it changed!

I reckon I've seen examples of all of those on odd occasions when I've found myself comparing gradient profiles from different sources for the same line.

So these kinds of differences are not unusual then?

The only thing I can say is that all these three profiles show roughly the same graph where the line goes up and down, that at least is some sort of constant.

Maybe the best I can do is draw the graph, split it up into gradient change section and state Mr X says this is a 1 in 64, Mr. Y says a 1 in 1270?

It's very disheartening when you think you've covered something such as this only to find out no one can agree, perhaps this is why my L&Y society book about the line doesn't have a gradients profile? which I was very surprised to find out it didn't.

 

[splashoutradio]

The railwaydata.co.uk site is supposed to have gradient profiles from modern surveys but they often suffer from other inaccuracies which seem to be down to nothing but carelessness. The site's own javascript to render the data into gradient plots doesn't work because of course it doesn't, but it's straightforward enough to replace it with something that does; it then turns out that there are two separate sets of data for each line - an approximate one with average gradients between points spaced at regular intervals, and an "accurate" one with the gradients between the points where they actually change (ie. as usual) - and they don't match up. The "accurate" data is speckled with random corruptions where the sign of the difference between two successive points is wrong in either the y or the x direction, or digits in a number have been replaced by adjacent digits on the keyboard, so the plotted gradient profile has crazy zigzags and impossible loops where it turns back on itself for a bit and then continues in the right shape but displaced horizontally and vertically from where it should be.

The data used for gradient profiles is sourced from the same dataset that NR use to compile gradient profiles for signalling purposes, released under FOI requests. It's not a perfect set of data by any means, but at the time of creation, it's what was available unfortunately. I'd be interested to know how you believe this is from modern surveys though, as I don't believe it is? There's definitely only one set of gradient data used on the site, so not sure where the two 'accurate' vs 'approximate' datasets per line has come from.

I'd be interested to know about the straight forward fix you mention, It'd be gratefully received, so it can be incorporated into future updates and I can provide a free website that is a little less "nothing but carelessness".

I messaged them that it was wrong about the section of the branch line covered by them, but never got a reply, nor was it changed!

I can only apologise for this. Spam, and for some reason CVs, tend to flood the inbox, so it may have got lost amongst them. I'll have a sift through for your email.

If the raw NR gradient profile data for a line is ever of use to anyone, let me know - I'd be happy to pop it in an excel file and upload it here

 

[Andy873]

I can only apologise for this. Spam, and for some reason CVs, tend to flood the inbox, so it may have got lost amongst them. I'll have a sift through for your email.

Sorry if I sounded a bit harsh, it wasn't my intension. The email I sent was I think over two years ago so it's probably not there now, but thanks anyway.

If the raw NR gradient profile data for a line is ever of use to anyone, let me know - I'd be happy to pop it in an excel file and upload it here

I can do something better for you, I spent this morning putting my data into a spreadsheet for your Railway Date site (Attached), you very welcome to use it on the site. I've measured and re-measured the distances many times and I think these (give or take a couple of feet) measurements are as close as anyone can get.

One thing your site did show was the original names of two consecutive bridges, Shuttleworth's No. 1 and Shuttleworth's No. 2 named after Lord Kay Shuttleworth of Gawthorpe Hall, Padiham!

Regarding gradients for the L&Y network, the L&Y Society has two general profiles for the entire line and all the other lines, they can be found in their Virtual Museum section of their website.

Attached is what I believe the Railway Data site should show for the PRG line section.

 

[splashoutradio]

Sorry if I sounded a bit harsh, it wasn't my intension. The email I sent was I think over two years ago so it's probably not there now, but thanks anyway.

I can do something better for you, I spent this morning putting my data into a spreadsheet for your Railway Date site (Attached), you very welcome to use it on the site. I've measured and re-measured the distances many times and I think these (give or take a couple of feet) measurements are as close as anyone can get.

One thing your site did show was the original names of two consecutive bridges, Shuttleworth's No. 1 and Shuttleworth's No. 2 named after Lord Kay Shuttleworth of Gawthorpe Hall, Padiham!

Regarding gradients for the L&Y network, the L&Y Society has two general profiles for the entire line and all the other lines, they can be found in their Virtual Museum section of their website.

Attached is what I believe the Railway Data site should show for the PRG line section.

No need to apologise. I think I'd be a little ruffled if I went through the trouble of compiling and sending updated data to not receive an acknowledgement.

Thank you so much, that's incredibly generous! I'll update the backend data this week and let you know when it's up. Thanks again

 

[Andy873]

Thank you so much, that's incredibly generous! I'll update the backend data this week and let you know when it's up. Thanks again

You're more than welcome!

Some more details for you:

PRG name - Padiham to Rose Grove West Junction.
At Rose Grove it connected to the East Lancs line under reference FRH5.
From the Padiham end, the rest of the branch was code NLC.

Gradients:
Rose Grove West onto branch - Level, approximately 5 chains.
Then 1 in 40 for 1.25 miles (down hill).
A 1 in 90 approximately 10 chains.
Level through station approximately 20 chains.
Then a 1 in 173 down hill.

Total length, 1 mile 76 chains.

Bridge Inspection dates:
The dates you have for the inspection of most of the bridges (2007) look good to me.
This would have been the last inspection of the line before it was lifted.

Hope that helps.
Andy.