Rail Joints

[L&Y Robert]

I have been reading RAIB reports – they are full of interesting stuff. However, one of them (R242014-141113-ECML-Rail-Breaks.pdf) deals with broken rails, and documents several instances of rail breaks occurring at or around fish-plate joints. I have often wondered about this type of joint, used throughout our railway system. Their numbers have been much reduced in recent decades by the introduction of long welded rails, but they are still predominate on some lines and are common around points and crossings and so-on. The butt joint seeks to unify two lengths of rail which are effectively jointed by clamping two additional short members (the fish-plates) to the rail web, one on each side, using four (now six, I see) nuts and bolts. The rail ends thus clamped are themselves short cantilevers from the last point of support - the preceding sleeper. This arrangement is then repeatedly subjected to rolling loads of varying intensity but of short duration as the wheels of a train pass over, producing intermittent bending moments and shear forces in the rail ends. I see that the clamping action between rail head and rail foot of modern fish-plates will, to some extent, transfer the deflection of the roll-off rail to the roll-on rail, and then, as the wheel passes over, the reverse. Nevertheless there is - must be - some residual bending moment in the rail end, and pictures in the RAIB of rail-end failures support this. An added complication is that the rail web has to be drilled to admit the fish-plate bolts – right where the forces are greatest. One day at the Ffestiniog railway I saw that the butt joints of the original railway were made in a different, and in one respect to my mind superior way. The ends of both rails were arranged to fall centrally in an extended chair, secured to a wider sleeper, and held in place there by wooden keys. The wooden keys might be a bit doubtful, but the point here is that the rails are supported right to their very ends, so do not become cantilevers. Bolted fish-plates could be used in such an arrangement, but the rail would be fully supported at the joint. For this I envisage some kind of combined chair/base-plate/fish-plate fabrication and a wider sleeper – perhaps even a single longitudinal sleeper (well, it’d be a pair of course, the other rail as well – sort of ‘H’ configuration). Maybe it’s been tried, but if so I’ve never come across it or heard about it. What do we think?

 

[furnessvale]

Years ago, fully supported joints were the norm on main lines but fell out of favour.

Two main reasons have been given for this.

Maintenance of the sleepers and ballast, especially mechanised tamping, is compromised by the non standard sleeper spacing a supported joint causes.

There is evidence that, despite the support, there is more rail head end batter at a supported joint.

 

[Joseph_Locke]

It's a very old design, long out of favour on the main line as it is hard to make sure both rails stay close together, end on.

The UK doesn't use six-hole fishplates at common joints; those are glued and Huck (MGL) pinned together. The main issue with common joints is the need to allow them to slide to "lose" thermal expansion. The nearest 20th century equivalent is the "tight joint" which allows no sliding (and has no expansion gap) - these aren't lubricated and can be done up as tight as you like. They rarely fail.

Another major preventative is cold expansion of bolt holes (CBX) after drilling, which largely eliminates the conditions in which minor drilling marks can become stress cracks by compressing the metal around the bolt holes. This treatment is now mandatory for new fish bolt holes.

We also don't (and haven't for a while now) install anything that fits the description of long welded rail (e.g. 120' jointed or 180m with intermediate breathers); we install continuous welded rail, in nominally infinite lengths - no joints at all.

Lastly, another major cause of rail breaks, slag inclusions from rolling ingots, is on the decline as a) all rail post-1976 in concast and b) all pre-1976 is being removed from track as a matter of policy.

That just leaves Hydrogen bubbles, wheel burns, RCF, tamper damage ....

 

[Dr_Paul]

We also don't (and haven't for a while now) install anything that fits the description of long welded rail (e.g. 120' jointed or 180m with intermediate breathers); we install continuous welded rail, in nominally infinite lengths - no joints at all.

How does this cope with expansion on hot days?

 

[najaB]

How does this cope with expansion on hot days?

It would be more accurate to say how does it cope with contraction on cold days. It's pre-heated when it's installed.

 

[Railsigns]

using four (now six, I see) nuts and bolts.

Or two.

 

[Mugby]

Six hole plates are only for insulated joints. ordinary joints usually have four hole plates.

The picture above shows nine inch plates (two holes) which were once used extensively on lightly trafficked routes and some survived until relatively recent years.

 

[Emblematic]

It would be more accurate to say how does it cope with contraction on cold days. It's pre-heated when it's installed.

Pre-stressed rather than pre-heated. In the UK, it's tensioned to be nominally stress-free at 27 °C. Extremely hot weather can still cause problems.

 

[najaB]

Pre-stressed rather than pre-heated.

Thanks for the correction.

 

[furnessvale]

How does this cope with expansion on hot days?

Euston station is mounted on some very big rollers which the public can't see.

Seriously:

On a cold day the rail goes into tension.

On a hot day stresses build up within the rail.

Both of these are resisted by the heavy sleeper and ballast arrangements employed and the rails just stay where they are under stress.

Very occasionally, often when the ballast is not quite enough, the rail stresses will win and a buckle will occur.

 

[L&Y Robert]

Six hole plates are only for insulated joints. ordinary joints usually have four hole plates.

The picture above shows nine inch plates (two holes) which were once used extensively on lightly trafficked routes and some survived until relatively recent years.

Yes, I remember those when I was a kid. I thought at the time that "That was how you did it", observed on what is now called the Copy Pit route at Kitson Wood (just near our house). When I first saw four-hole fishplates somewhere I thought it was "overkill"! - but then I would be about seven at the time. I wouldn't say the line up from Tod was "lightly trafficked" though, coal trains of 20 wagons or so, hauled by, and always banked by big dirty engines (8F type).

 

[Ploughman]

We also don't (and haven't for a while now) install anything that fits the description of long welded rail (e.g. 120' jointed or 180m with intermediate breathers); we install continuous welded rail, in nominally infinite lengths - no joints at all.


That just leaves Hydrogen bubbles, wheel burns, RCF, tamper damage ....

Within the last 10 years I have installed 120fts on NWR tracks.
Have the rules changed within the last 10 years?

 

[Trog]

Within the last 10 years I have installed 120fts on NWR tracks.
Have the rules changed within the last 10 years?

Yes, nothing longer than 60'-0" for jointed for about five years.

 

[Dr_Paul]

Euston station is mounted on some very big rollers which the public can't see.

Seriously:

On a cold day the rail goes into tension.

On a hot day stresses build up within the rail.

Both of these are resisted by the heavy sleeper and ballast arrangements employed and the rails just stay where they are under stress.

Very occasionally, often when the ballast is not quite enough, the rail stresses will win and a buckle will occur.

Thanks for that. I take it that's why jointed track would have (at normal temperature) a gap of half an inch or so at each joint for expansion, but welded track needs only a sliding joint every few miles with a couple of inches play in it.

 

[Railsigns]

I take it that's why jointed track would have (at normal temperature) a gap of half an inch or so at each joint for expansion, but welded track needs only a sliding joint every few miles with a couple of inches play in it.

It is not necessary to install adjustment switches at regular intervals along continuously welded rail. They are only fitted as required at specific locations.

This may be of interest:

https://www.youtube.com/watch?v=KgWsJEgIDcA

 

[Emblematic]

Thanks for that. I take it that's why jointed track would have (at normal temperature) a gap of half an inch or so at each joint for expansion, but welded track needs only a sliding joint every few miles with a couple of inches play in it.

The long lengths of CWR don't have sliding joints, they are very firmly anchored and the forces involved are considerable. Sliding joints and shorter lengths are used where it's desired to keep the rails stress-free and not transmit any forces to the formation - for example on bridges and viaducts, you don't want the rail trying to push the decks off their bearings.
Your 10mm gaps on jointed track would need to grow to around a metre to keep a 2km length of CWR stress-free
--- old post above --- --- new post below ---
Or just watch the video in the previous post, which says it all

 

[furnessvale]

Thanks for that. I take it that's why jointed track would have (at normal temperature) a gap of half an inch or so at each joint for expansion, but welded track needs only a sliding joint every few miles with a couple of inches play in it.

No. You misunderstand what is happening.

It is perfectly possible to connect London to Glasgow with continuous welded rail with no expansion joints. The stresses in the rail, either tension or compression are contained without any expansion joints.

The only time you will see expansion joints is at the end of a length of continuous rail where the stresses cannot be resisted. That length could be 200miles.

In practice, a major station, for example Preston, could have expansion joints at each end so that all the complex pointwork through the station was installed without the need for strengthening but it certainly not essential.

 

[Joseph_Locke]

In practice, a major station, for example Preston, could have expansion joints at each end so that all the complex pointwork through the station was installed without the need for strengthening but it certainly not essential.

You may have to go jointed because of curvature, restrictions on turnout radii and "complex" S&C such as slips, and the spacing between S&C units, but current policy is CWR first choice.

It's worth noting that adjacent to an adjustment switch only the last 80 - 100m of the CWR will show any kind of thermal movement.

 

[furnessvale]

You may have to go jointed because of curvature, restrictions on turnout radii and "complex" S&C such as slips, and the spacing between S&C units, but current policy is CWR first choice.

It's worth noting that adjacent to an adjustment switch only the last 80 - 100m of the CWR will show any kind of thermal movement.

Indeed. I was trying to keep it simple just to dispel the notion that all CWR moves and needs expansion joints every so often.

 

[L&Y Robert]

No. You misunderstand what is happening.

It is perfectly possible to connect London to Glasgow with continuous welded rail with no expansion joints. The stresses in the rail, either tension or compression are contained without any expansion joints.

The stress producing a kind of induced STRAIN.
One remembers Young's modulus!
--- old post above --- --- new post below ---

The stress producing a kind of induced STRAIN.
One remembers Young's modulus!

And 'slenderness ratio' as well:-

https://youtu.be/_LoXgN1QWZM

 

[furnessvale]

The stress producing a kind of induced STRAIN.
One remembers Young's modulus!
--- old post above --- --- new post below ---


And 'slenderness ratio' as well:-

https://youtu.be/_LoXgN1QWZM

I forget most of my physics and I am probably wrong, but I chose the word stress carefully.

To me, strain indicates a movement whereas the whole point of restraining CWR was to stop movement resulting in a stress contained within the steel, but no strain occurring.

Probably completely wrong but I have no doubt I will be corrected.

 

[Ships]

Stress is force/area, strain is extension/original length. Stress and strain are related by youngs modulus E=stress/strain

 

[Deepgreen]

By contrast to the talk of CWR, I noticed the attached at Finsbury Park the other day - platform 1 is at least partially laid with 45 foot flat-bottomed rails - pretty unusual in my experience these days.

 

[furnessvale]

Stress is force/area, strain is extension/original length. Stress and strain are related by youngs modulus E=stress/strain

So I am correct. The rail does not move so all the forces on the rail caused by temperature are contained within the steel as stresses.

 

[Ploughman]

2 points in response to the film clip about maintaining Adjustment switches.

Have the standards changed? Because when I was instructed in stressing up to Lvl 4 the setting of Adjustment switches was governed by the switch overlap.
Dimensions for the gap opening and the overlap were provided but the overlap took precedence.
In the clip it purely relies on the gap opening.

The second point which may also relate to the other thread the NWR Linear Scrapyard.
It indicates that you should tidy up after yourself.