Max speed on UK network

[The Planner]

Performance allowance is part of the Timetable Planning Rules, as they are now.

Operator led though. Very rare that they are put in by NR.

 

[edwin_m]

Is GSMR a continuous feed or does it update via fixed eurobalises? I’d be surprised if the braking curve updates in real time.

Level 1 ETCS updates via balises. Level 2 updates continuously via GSM-R but still uses existing track circuits or axle counters to locate the trains. Level 3 will be as per level 2 except that the trains transmit their locations and confirm completeness via GSM-R. In all these cases the train is updated with movement authority (it is allowed to proceed as far as position X and/or to observe maximum speed Y at position Z) and calculates its safe speed according to its own braking curve and the details of the line ahead. More details at: 3-ERTMS-Levels.pdf

The ERTMS “levels” define different uses of ERTMS as a train control system, ranging from track to train communications (Level 1) to continuous communications between the train and the radio block centre (Level 2). Level 3, which is in specification development phase, will further increase ERTMS’ potential by introducing a “moving block” technology and reducing trackside equipment for train detection. Whilst it is commonly acknowledged that to date, ERTMS level 2 offers considerable benefits, the use of level 1 already brings significant advantages for the railways and allows for High Speed travel.

 

[Wilts Wanderer]

Level 1 ETCS updates via balises. Level 2 updates continuously via GSM-R but uses existing track circuits or axle counters to locate the trains. Level 3 will be as per level 2 except that the trains transmit their locations and confirm completeness via GSM-R.

Thanks for clarifying.

 

[Annetts key]

Is GSMR a continuous feed or does it update via fixed eurobalises? I’d be surprised if the braking curve updates in real time.

Edit - as I engage brain - or do the Eurobalises just describe the route characteristics to the train. GSMR being a radio system of course.

Depends on which ETCS is being used. As what edwin_m says.

ETCS level one and level two use Eurobalises to update the train with its exact position. Otherwise the on board computer uses less accurate systems.

 

[Kneedown]

Don’t blame them. Was such a buzz to make up time back then.

I remember mentioning to my DM what impressive bits of kit the 91's were when I went in to sign the route. He promptly rollocked me for riding on them, until I pointed out that he told me to learn the Grantham to Northgate bit, and what the bluddy heck else was I supposed to ride on!

 

[peteb]

I find it quite rare for 125 mph to be maintained over any long distance. Top Speed seems to hover around 122-124 mph.

Isn't it something to do with the accuracy of the speedometer so 125mph on the speedo is actually up to 5% less?

Certainly car Speedos are reknowned for their inaccuracies so much so that the police give you a 10% margin of error?

 

[waverley47]

Isn't it something to do with the accuracy of the speedometer so 125mph on the speedo is actually up to 5% less?

Certainly car Speedos are reknowned for their inaccuracies so much so that the police give you a 10% margin of error?

The average speedo on a train without a TMS, as in anything before computers so BR legacy units, sprinters, HSTs ect, will have a speedo based on mechanical components ultimately derived from how fast the wheels turn. Wheels, springs, gears ect all get worn out, so after a reset the speedo will gradually wander from its true value. Drivers will notice and eventually it will get reset, but at any given point you don't know how accurate it is, or its margin of error.

Certainly it wasn't unheard of under BR for speedos to be quite substantial in their wanderings, and there was quite a famous story of a driver pulling into Carlisle on an electric southbound train to be met by the signaller. The signaller asked about some excessive speed coming down Beattock, and the driver replied saying he had been 110mph dead on, and it must be the speedo which was wrong. He refused to carry on with an inaccurate speedo, and the loco was swapped. While waiting, the driver wandered over and had a chat with some enthusiasts asking about it, where he answered that of course he had been speeding, but that the inaccurate speedo would cover up all manner of sins.

In comparison, most modern trains use either electronics or doppler radar to check speeds, which has less of a tendency to wander, but still will have a margin of error.

Most electronic components you'd find in your average modern train cab have margins of error at less than 0.1%, which isn't very much. However if you've set a speed limiter, it might be that the mechanical components have a greater margin for error, and the train can only regulate it's speed to between 123-127 mph, so you'd err on the cautious side.

 

[CrispyUK]

Certainly car Speedos are reknowned for their inaccuracies so much so that the police give you a 10% margin of error?

Under UK law a car speedometer is permitted to over-indicate the true speed by quite a margin, but it must never indicate a speed Lower than that at which the vehicle is travelling.

Because of this and to allow for tyre wear, etc. speedos are designed to over read slightly.

 

[O L Leigh]

Certainly it wasn't unheard of under BR for speedos to be quite substantial in their wanderings, and there was quite a famous story of a driver pulling into Carlisle on an electric southbound train to be met by the signaller. The signaller asked about some excessive speed coming down Beattock, and the driver replied saying he had been 110mph dead on, and it must be the speedo which was wrong. He refused to carry on with an inaccurate speedo, and the loco was swapped. While waiting, the driver wandered over and had a chat with some enthusiasts asking about it, where he answered that of course he had been speeding, but that the inaccurate speedo would cover up all manner of sins.

The problem you had to face, however, was when you had to deal with cab with a genuinely inaccurate speedo.

I took a Cl317 down the Lea Valley some years ago and recall that it was terribly slow. It seemed to take an age to get up to linespeed. I braked at the usual point at Cheshunt only for the platform to rush up at me uncomfortably fast. Some judicious use of the biggest anchor got the thing stopped, but when I had the same issue at Broxbourne I realised that summat was amiss. After that I drove the train according to routine (this much power until here, coast until here, and so on) and I found that the train behaved as it ought, I could use my usual braking points comfortably and I was still keeping to time. It also allowed me to work out that the speedo was under-reading by about 10% which, at the normal speed along the Lea Valley of 80mph, meant I was actually doing around 88mph. It's no wonder I was having trouble getting the thing to stop.

 

[Bald Rick]

The problem you had to face, however, was when you had to deal with cab with a genuinely inaccurate speedo.

I took a Cl317 down the Lea Valley some years ago and recall that it was terribly slow. It seemed to take an age to get up to linespeed. I braked at the usual point at Cheshunt only for the platform to rush up at me uncomfortably fast. Some judicious use of the biggest anchor got the thing stopped, but when I had the same issue at Broxbourne I realised that summat was amiss. After that I drove the train according to routine (this much power until here, coast until here, and so on) and I found that the train behaved as it ought, I could use my usual braking points comfortably and I was still keeping to time. It also allowed me to work out that the speedo was under-reading by about 10% which, at the normal speed along the Lea Valley of 80mph, meant I was actually doing around 88mph. It's no wonder I was having trouble getting the thing to stop.

Indeed. Not an unusual turn of events. Heard a rumour that a 91 / Mk IV set has been running around recently at around 140mph with similar issues...

 

[D365]

Why?! That’s really silly. Is it just because the system can’t judge actual rail conditions so errs on the safe side?

Well, yes. ETCS is not designed to replace any element of the driver's function. The computer contains limited information about the unit/locomotive and does not have enough sensor capability to read rail condition. Any braking input by the computer is very much a last resort, when the computer believes the driver is coming in too fast. Repeated ETCS interventions will likely result in a tea-without-biscuits for the driver concerned.

 

[notadriver]

My observations are that speedos are quite accurate on modern units - as they are on tachograph fitted vehicles. There might be a 1 or 2 mph difference. In cars a 2 mph varience at 50 mph could turn into a 10 mph one at 125 mph. It’s critical that doesn’t happen on a train.

I remember departing Tiverton parkway with gps on. It wasn’t until we reached 90 mph that we were overtaking most but not all cars.

 

[dm1]

As I understand ETCS braking curves (which is an extraordinarily complex topic), there are multiple braking curves for different situations, train types and boundaries.

From what I've heard (from fairly indirect sources), when the ETCS specifications were being standardised, there were fairly heated discussions about how exactly to set the braking curves and what margins of error to allow for. Various countries had wildly different legacy standards that were all combined to give ETCS, but under the main consideration that they should be "safer" than the systems that came before, at least on paper. There is still some debate about this - if I remember correctly, there is a parameter that can be set that influences the braking curve, but only once per country, rather than for every signal. Ideally you could adjust the braking curve parameters based on the type of hazard you are protecting (i.e. more conservative for situations where an overrun is critical, less conservative where an overrun is annoying but unlikely to be dangerous, or taking factors such as gradients into account)

There have been issues with ETCS failing to deliver the promised capacity boost because of these braking curves, particularly when going towards a red signal or buffer stops, when compared to the legacy signalling systems that came before.

At least part of the problem comes from the way the signalling behaves - the driver is not shown the braking curve itself, but rather only a continuously changing target speed. If this target speed is exceeded, or the system thinks it will be exceeded at a later point, then an emergency brake application is triggered (rather than a service brake application until the curve is being followed again), meaning drivers are forced to drive even more defensively in order to ensure this does not happen.

Given ETCS resets can take a horrendous amount of time, they are something that drivers are at pains to avoid.

 

[507020]

Most electronic components you'd find in your average modern train cab have margins of error at less than 0.01%, which isn't very much. However if you've set a speed limiter, it might be that the mechanical components have a greater margin for error, and the train can only regulate it's speed to between 123-127 mph, so you'd err on the cautious side.

I’ve sometimes tried to crudely measure speed using a GPS speedometer app and the fastest I’ve supposedly experienced is 131mph on a Pendolino north of Crewe, but I’ve found it’s not particularly accurate when running under the wires.

 

[waverley47]

I’ve sometimes tried to crudely measure speed using a GPS speedometer app and the fastest I’ve supposedly experienced is 131mph on a Pendolino north of Crewe, but I’ve found it’s not particularly accurate when running under the wires.

Exactly. While doppler radar pointed down at the track will give you a very accurate speed, it's only as accurate as the stuff it's connected to.

The most prominent example of this is in the Olympics. Whereas sprinting is timed to the hundredth of a second, swimming is only measured to the tenth as the building tolerances for the length of swimming pools mean the margin for error would be to great.

Too much stuff in the way, or movement of the track, or tyre ware all contribute to inaccuracies, hence the pendolinos and most modern units having their speed limiters set to ease off the power at slightly below 125mph.

Unless you can guarantee every component has the same margin for error, you have to accept the highest individual margin as the baseline; in the case of GPS, that margin is the heights and positions of satellites which do vary somewhat from their origins.

 

[notadriver]

I don’t believe there is any automatic easing off of power nearing 125 mph. It’s down to the driver. However speedos can over read so an indicated 125 may beonly 123 mph.

 

[hexagon789]

I don’t believe there is any automatic easing off of power nearing 125 mph. It’s down to the driver. However speedos can over read so an indicated 125 may beonly 123 mph.

Pendolino speedsets do 'ease off' before the set speed is attained there's plenty of evidence about that demonstrates this, so do 90s and 91s. In fact I'm sure it's a pretty standard feature so that -

1. The desired speed isn't exceeded by still being under full power as the set speed is reached
2. When the desired speed is attained the power isn't immediately cut from full to zilch to prevent an overspeed and which particularly on a loco-hauled train could cause a bit of a jolt as evidenced by the frequent jolts on 225 sets in the TSOE when the driver cuts power on the 91and the coaches close in or the loco closes in on the coaches.

 

[notadriver]

Pendolino speedsets do 'ease off' before the set speed is attained there's plenty of evidence about that demonstrates this, so do 90s and 91s. In fact I'm sure it's a pretty standard feature so that -

1. The desired speed isn't exceeded by still being under full power as the set speed is reached
2. When the desired speed is attained the power isn't immediately cut from full to zilch to prevent an overspeed and which particularly on a loco-hauled train could cause a bit of a jolt as evidenced by the frequent jolts on 225 sets in the TSOE when the driver cuts power on the 91and the coaches close in or the loco closes in on the coaches.

The poster was referring to a speed limiter and not a driver selectable speed set device. They are different things.

 

[waverley47]

The poster was referring to a speed limiter and not a driver selectable speed set device. They are different things.

A speed set will limit speed. Given that as far as I am aware, no train in this country is blessed with a speed limiter, you could have made the connection.

 

[XAM2175]

There is still some debate about this - if I remember correctly, there is a parameter that can be set that influences the braking curve, but only once per country, rather than for every signal. Ideally you could adjust the braking curve parameters based on the type of hazard you are protecting (i.e. more conservative for situations where an overrun is critical, less conservative where an overrun is annoying but unlikely to be dangerous, or taking factors such as gradients into account)

I have a vague recollection of hearing that there's a (perhaps theoretical) capability for the braking curve to be varied depending on the train's capabilities. A basic implementation of it certainly exists in the current German train protection systems PZB and LZB - the train is placed into one of three "types" based on its braking performance, and that type then informs the maximum permissible speed (which can be lower than the line speed) and the braking curve.

 

[notadriver]

A speed set will limit speed. Given that as far as I am aware, no train in this country is blessed with a speed limiter, you could have made the connection.

I don’t believe that is correct. Some do, some don’t. A speed set device or cruise control is a luxury not all trains have too. The class 397 is notable in not having one.

 

[craigybagel]

A speed set will limit speed. Given that as far as I am aware, no train in this country is blessed with a speed limiter, you could have made the connection.

The electric members of the Siemens Desiro family have limiters, although they're not set by the driver - AFAIK you can leave one of those in full power and it won't exceed 100, or 110 for a 350.

Incidentally, I'm sure I heard somewhere that Avanti drivers don't tend to use the speed set on 390s these days but I'm open to correction on that matter.

 

[dm1]

I have a vague recollection of hearing that there's a (perhaps theoretical) capability for the braking curve to be varied depending on the train's capabilities. A basic implementation of it certainly exists in the current German train protection systems PZB and LZB - the train is placed into one of three "types" based on its braking performance, and that type then informs the maximum permissible speed (which can be lower than the line speed) and the braking curve.

That capability is definitely there. If I recall correctly, there is a number that is calculated for every train, given as a percentage (generally over 100% for passenger trains), which represents the braking performance of the train and is used to determine the braking curve. However, it generally doesn't include regenerative braking performance, so for trains without or with small braking resistors, it only considers pneumatic and magnetic brakes - the logic being that in an emergency it's possible that the OLE will not be able to take up the braking current.

Otherwise no every train would be limited in speed and braking curve to the performance of the worst-braked freight train on the network.

This is a concept that has been common in Europe for a long time, long before ETCS, in order to take into account the varying braking performance of different trains whilst ensuring they can slow down appropriately within the overlaps on a given line.

 

[edwin_m]

I have a vague recollection of hearing that there's a (perhaps theoretical) capability for the braking curve to be varied depending on the train's capabilities. A basic implementation of it certainly exists in the current German train protection systems PZB and LZB - the train is placed into one of three "types" based on its braking performance, and that type then informs the maximum permissible speed (which can be lower than the line speed) and the braking curve.

That capability is definitely there. If I recall correctly, there is a number that is calculated for every train, given as a percentage (generally over 100% for passenger trains), which represents the braking performance of the train and is used to determine the braking curve. However, it generally doesn't include regenerative braking performance, so for trains without or with small braking resistors, it only considers pneumatic and magnetic brakes - the logic being that in an emergency it's possible that the OLE will not be able to take up the braking current.

Otherwise no every train would be limited in speed and braking curve to the performance of the worst-braked freight train on the network.

This is a concept that has been common in Europe for a long time, long before ETCS, in order to take into account the varying braking performance of different trains whilst ensuring they can slow down appropriately within the overlaps on a given line.

3-ERTMS-Levels.pdf

Receiving the movement authority through Eurobalises, the ETCS onboard equipment automatically calculates the maximum speed of the train and the next braking point if needed, taking into account the train braking characteristics and the track description data.

The quote is for Level 1, unfortunately there's no equivalent quote for Level 2 and 3 but the higher levels incorporate all the lower level functionality. I assume national policy on adhesion levels is taken into account in this calculation.

With very few exceptions, trains have friction brakes on all the wheels but regenerative braking only works on motored axles, so the rate of deceleration on regenerative braking will never be more than with friction braking. Regenerative braking can't be relied on in an emergency and the two systems can't be combined to increase the total braking available, because applying too much brake on the same wheel would simply cause it to slide (they can be "blended" to improve the efficiency of the brake within the same maximum deceleration brake). Therefore no driver or ATP system should be allowing a speed faster than that which allows the train to comply with its movement authority by the use of friction brake alone.

 

[dm1]

3-ERTMS-Levels.pdf

The quote is for Level 1, unfortunately there's no equivalent quote for Level 2 and 3 but the higher levels incorporate all the lower level functionality. I assume national policy on adhesion levels is taken into account in this calculation.

With very few exceptions, trains have friction brakes on all the wheels but regenerative braking only works on motored axles, so the rate of deceleration on regenerative braking will never be more than with friction braking. Regenerative braking can't be relied on in an emergency and the two systems can't be combined to increase the total braking available, because applying too much brake on the same wheel would simply cause it to slide (they can be "blended" to improve the efficiency of the brake within the same maximum deceleration brake). Therefore no driver or ATP system should be allowing a speed faster than that which allows the train to comply with its movement authority by the use of friction brake alone.

That is true. But another major factor is that pneumatic/friction brakes take some time to activate (a few seconds at least - for a loco with coaches up to 30s), whereas electric brakes (in particular if electric braking independent from the grid into a braking resistor is possible) can be activated almost instantly, as can track brakes. For emergency braking curves, that could make a significant difference, if it were possible to take electric braking into account, which in most cases it is not.

To bring this (very interesting) discussion slightly back on-topic, I do wonder why magnetic track brakes are still so rare on the GB rail network. NR seems to be averse to them, but from my observations, many of the issues regarding leaf-fall could be largely solved, or at least significantly mitigated were they to be used. With additional, reliable brake force perhaps there are even locations where line speed could be increased.

 

[edwin_m]

That is true. But another major factor is that pneumatic/friction brakes take some time to activate (a few seconds at least - for a loco with coaches up to 30s), whereas electric brakes (in particular if electric braking independent from the grid into a braking resistor is possible) can be activated almost instantly, as can track brakes. For emergency braking curves, that could make a significant difference, if it were possible to take electric braking into account, which in most cases it is not.

To bring this (very interesting) discussion slightly back on-topic, I do wonder why magnetic track brakes are still so rare on the GB rail network. NR seems to be averse to them, but from my observations, many of the issues regarding leaf-fall could be largely solved, or at least significantly mitigated were they to be used. With additional, reliable brake force perhaps there are even locations where line speed could be increased.

I've often wondered about magnetic brakes - they could also be very helpful for rural lines with open crossings, where perhaps the train could approach faster and still be able to stop if the crossing wasn't clear. But they are quite an expense, probably create issues with EMC and axle load, and are too harsh for routine use. Trams need them because hazard situations are much more common on the street, and it's good to have a much more severe brake to use if something unexpected happens such as a car shooting out from a side road.

 

[Annetts key]

U.K. signalling has taken the braking curve of trains into consideration for many, many years.

Lots of variables affect a trains breaking curve, including available breaking force on the locomotive (or other powered vehicle/powered axles), the available breaking force on the carriages/cars/wagons (in other words, the train brakes), the maximum speed of the train, and the weight of the train. And yes, the time it takes for the brakes to become effective will also be taken into account.

The breaking curve is obviously also affected by other factors, such as inclines and the rail surface conditions.

With an ERTMS/ETCS or ATP system, the driver has to ensure that the system has the correct train configuration before the train enters traffic.

The placement of fixed signalling infrastructure also has to take a trains breaking curve into account. Such as TPWS track equipment. But this has to be a compromise between the different train types that are permitted. Because there are rather large differences between the braking curves of the various different trains that are permitted on the U.K. network.

The position of most colour light signals in track circuit block systems is determined by line speed and the worse case braking curve of the types of train permitted on that line. In some cases, this is the reason for differential speed limits. With freight having a lower limit, and disc braked trains having a higher limit.

And yes, because overshooting can in some places result in a potential high risk of collision, the systems are conservative. So yes, drivers have to adapt their driving style.

For example, now that drivers have learned about how TPWS works in practice, the number of “false activations” reported has fallen substantially compared to when it was introduced.

 

[edwin_m]

U.K. signalling has taken the braking curve of trains into consideration for many, many years.

Lots of variables affect a trains breaking curve, including available breaking force on the locomotive (or other powered vehicle/powered axles), the available breaking force on the carriages/cars/wagons (in other words, the train brakes), the maximum speed of the train, and the weight of the train. And yes, the time it takes for the brakes to become effective will also be taken into account.

The breaking curve is obviously also affected by other factors, such as inclines and the rail surface conditions.

With an ERTMS/ETCS or ATP system, the driver has to ensure that the system has the correct train configuration before the train enters traffic.

The placement of fixed signalling infrastructure also has to take a trains breaking curve into account. Such as TPWS track equipment. But this has to be a compromise between the different train types that are permitted. Because there are rather large differences between the braking curves of the various different trains that are permitted on the U.K. network.

The position of most colour light signals in track circuit block systems is determined by line speed and the worse case braking curve of the types of train permitted on that line. In some cases, this is the reason for differential speed limits. With freight having a lower limit, and disc braked trains having a higher limit.

And yes, because overshooting can in some places result in a potential high risk of collision, the systems are conservative. So yes, drivers have to adapt their driving style.

For example, now that drivers have learned about how TPWS works in practice, the number of “false activations” reported has fallen substantially compared to when it was introduced.

Signalling has to be designed around a braking curve, but with exceptions such as the differential speeds you mention, the signal spacing has to be designed for the worst case braking curve of all the trains using a particular section of line. ETCS gives the train information on the track ahead, and the on-train system works out the safe speed based on its own specific braking curve, so in theory trains with better brakes can follow others more closely.