There's a reason there are no SPADs on the Northern Line anymore. There aren't any signals to pass.
And Arnos Grove has a depot, so perhaps that's related.
There's a handy map of the whole system here.
'A' signals are Northern Line.
Hundreds of London Underground's automated trains were cancelled last year thanks to automated train operation system SNAFUs, blowing a hole in claims that replacing bolshy staff with computers is the best way to prevent delays. Service-busting failures included the Jubilee and Northern Lines' main vehicle control computers ( …
There's a reason there are no SPADs on the Northern Line anymore. There aren't any signals to pass.
And Arnos Grove has a depot, so perhaps that's related.
There's a handy map of the whole system here.
'A' signals are Northern Line.
depends what you want the map for. Harry Becks map is useless if you don't know where you want to go. For example. I'm a country bumpkin from the north of England, where trains are concerned they run every hour and mostly bypass your station so I don't really need to worry about them.
Recently I took a trip to London to see the marvels of piped gas and mains sewerage. We were stopping near to Tower bridge so choosing a starting station was easy - Tower but I wanted to go to the natural history museum - Kensington. But WHICH station was closest?
I ended up using a rather magnificent app called "ulmon London" that had offline mapping (googlemaps) and overlayed the tube maps. That made finding a local tube station a cinch. So sure, Harry Becks is great at locating your start and destination stations (and mapping changes inbetween) but is rubbish if you don't know your station to begin with.
It has historical notes, shows accurately the position of pointwork, sidings, stations, closed stations and where lines cross each other vertically. OK, maybe not "handy" as in small, but "handy" as in it it has more detail. There's also a nice scrolly, zoomy version on the same website. And I believe the guy's done the same for other Metro systems.
But indeed interesting to see how it's actually physically laid out in a true geographical sense, rather than from the traveller convenience viewpoint.
You can actually get a similar thing on Google Maps - if you enable the Public Transport layer, it shows you London tube lines
OK, I've had some replies about the Arnos Grove thing.
There's a higher than typical number of speed-control signals in the area, which can be tricky, and there's a gradient up to Oakwood (so the SB trains may be picking up speed from that).
There's a lot of conflicting movements possible in the area, so signals may remain set to danger for longer than anticipated.
Crew may be distracted by preparing for crew-relief at Arnos Grove.
PJ320 is a 10mph draw-up signal halfway along a platform which may not clear as anticipated due to e.g. a reversing service on the adjacent (converging) track.
And the inner home signals for eastbound and middle platforms have a dual role as semi-automatics or draw-up signals depending on the route that has been set.
NB Speed control and draw-up signals are peculiar to LU and were introduced to shorten the gaps between trains, particularly at platforms, by allowing a slow moving following train to approach closer (to the rear of a stopped in platform or just pulling away train) by means of a cascade of time-delayed signals - if you get to the next red before the timer runs out, you're going too fast and the tripcock will apply the emergency brake.
I've had more replies about the SPaDs. Seems like it's a feature of the layout of the tracks in that area.
"A904 if you miss the R904 and come out of the tunnel doing 30mph plus in the rain its a spad all day.."
"When I didn't have gray hair and used to be a trainer, I'd always tell trainees that if R904 was on start braking immediately. If they didn't..."
"One of the problems with A904 is that drivers don't react soon enough to R904 if it's on. They come down the hill from Southgate, see R904 with a yellow aspect, but don't immediately start braking, hence the number of SPaDs at that signal. "
" A hidden, recurring fault is often the most difficult to diagnose and cure – and may be indicative of a deeper problem."
Get in their first with a new line in Phsycology you could even name it after yourself.........
https://www.youtube.com/watch?v=jC_-r-J69qA
ATO systems aren't 100% perfect, I guess.
Well, what a surprise.
Then, as the author notes in the middle of the article, they're actually a damn sight better than what came before.
Then, for some inexplicable reason, he concludes that Mk 1 eyeballs are better after all.
I note there's no mention of the DLR, which has been happily trundling around using Seltrac and no Mk 1 eyeballs for decades.
And if he wants to see a really brave implementation of ETCS and ATO, look at what NR are planning for on the Thameslink core in 2018.
Quote
And if he wants to see a really brave implementation of ETCS and ATO, look at what NR are planning for on the Thameslink core in 2018.
I think the NUR will have more than a few words to say about that given the present problems Sarf of the River. The last southern train from my local station goes at 16:28 this week.
If NR think they can get 24 tph (trains per hour) in each direction in the peaks through the Thameslink core then can I have some of what they are smoking please?
If NR think they can get 24 tph (trains per hour) in each direction in the peaks through the Thameslink core then can I have some of what they are smoking please?
And even more, temporarily, when "recovering from perturbation".
Which is why I descibed it as "really brave". Yes, mad might be a better description. Especially given how many separate routes need to be merged together coming into the core (and even more now they're proposing to add in Rainham via Dartford, because Windmill Bridge can't handle the traffic...)
> If NR think they can get 24 tph (trains per hour) in each direction in the peaks through the Thameslink core then can I have some of what they are smoking please?
Wondered the same thing myself - I take the Thameslink, FSM help me, and dwell times at rush hour through the core stations are easily in excess of a minute, frequently two, and this that doesn't leave much room for getting the train on and off the platform. So I'm left incredulous as to how it's meant to happen even with new rolling stock that's easier to get in and out of.
There's talk of a new signal type called POSA to be used in the core however to achieve the supposed peak of 30tph, and I don't think there's to be any through trains past blackfriars from the south-west loop.
http://www.railengineer.uk/2013/11/13/signalling-the-thameslink-programme/
Edit: grr stupid non-clickable links
DLR was designed from the outset for automatic trains, with the luxury of not opening lines until all the work was done and thoroughly tested.
All the other lines are conversions from manually-controlled systems with work generally done under enormous time pressure in a few hours window each night in old tunnels full of god-knows-what and testing also has to be crammed into that window.
One of the main reasons given for not putting in extra tunnels is cost, but the cost of only having one line in each direction with no resilience and no ability to take a line out of service for proper heavy maintenance (some tunnels are still running on century old rail which should have been replaced years ago, and as such are subject to low speed limits) far outweighed the cost of boring new ones a long time ago. Doing it this way would allow safe 24 hour services too.
Crossrail _could_ have addressed this, but oh no, the same blinkered mentality applies. 10-20 years after opening when heavy work is required it's going to have the same problems as the current lines do.
A brave politician and TFL head would take the bull by the horns and announce major new programs but this is unlikely to ever happen.
A brave politician and TFL head would take the bull by the horns and announce major new programs but this is unlikely to ever happen.
Or just sit down and be quiet and be thankful for what you have and distribute the money around the rest of the country where it will make a heck of a lot more difference:
IPPR report from 2013 (PDF)
See tables 1.5, 1.6 and figures 1.3 and 1.4 for example.
Some of the analysis is disputed though, and even the Western Mail is a bit more restrained:
Transport spending per person in London is twice the level in Wales
Note that this is in relation to an earlier version of the IPPR report, from 2011.
M.
So you're saying that the system that can put overpaid underworked union dinosaurs out of a job, which is maintained by other overpaid underworked union dinosaurs, isn't fit for purpose according to the data produced by yet more overpaid underworked union dinosaurs?
Colour me surprised.
To talk sensibly about safety critical infrastructure control you need to distinguish between failing and failing safe. Otherwise you will overinvest in wailing about why everything is so broken.
Of course you need to reduce fail safe events, but start from the premise that FAIL SAFE GOOD.
Example: signal failures are vanishingly rare. Signal fail safes are all too common. All that means is that one of the two or three redundant signalling comms signals doesn't agree with the others, and since the system has no way of judging which one is right, it goes red and waits for a human to come along and take a look. In the meanwhile, everyone stays more or less alive.
Alive good. Dead bad.
Not entirely true. Each life has a value. Each commuter minute wasted has a value. At some point holding up (or being fined) for a delay exceeds the lives lost number and then we let people die. This is usually done at the inception of a project, but also applies to upgrades and ongoing maintenance.
This post has been deleted by its author
Simply you can have more trains on the same tracks. Which in a country where building railway lines is near impossible (because there is always someone's house in the way) is pretty much essential. Cross rail for example will have to be ATO in the tunnels just in order as to achieve 20 trains per hour, though for reasons of us implementing such things at a geological pace,it will revert to conventional signalling above ground.
That's why, sooner or later, the number of system requirements, especially those related to Built In Test, self-monitoring, and logging of all data, will grow and grow and grow. The system spec and proof of compliance matrix will each become the size of a telephone book. Then a new Program Manager will take over, realize it's all safety critical, and force DO-178 Level A onto the process. You'll need an entire dedicated track network to execute full scale test programs. Everything will become ultra-difficult, expensive, but still making progress.
Then, quite unexpectedly, the accounting folks will try to implement SAP. The entire network will be out of service for a year.
* Perhaps a reader knows why Arnos Grove is so prone to SPADs? 29 Category A SPADs occurred there in the 12 months from March 2015, the highest concentration in any one Tube location over the year. Particularly problematic were signals A902, A904, PJ320 and PJ350, with each having been SPADded four or five times.
Well I've never been there, so my suggestion is uninformed, unfair, malicious and probably wrong. Could it be because it's a hole, and nothing (not even the machine) wants to stop there?
Or ley lines...
"Could it be because it's a hole"
A bit harsh but that could be the reason for a local nickname "Anus Grove"!
The station is about 2 miles from my humble abode which is nearer to my eponymous station on the Northern Line. AG not only has wacking great sidings where tubes go to sleep at night, but it also has a strange central platform which can operate in both Northbound and Southbound modes and doors that can open on either/both sides.
It confuses the hell out of me and I'm just a passenger so what it does to train staff is anyone's guess.
On the tube, or are you just pissed with the RMT over the Southern thing? Because the way I see it
(1) effective management doesn't antagonise the staff to the point of dispute
(2) driver only operation is not as safe as two-person operation (where both of the roles carry a safety responsibility)
(3) drivers operated doors are not as safe during train dispatch as guard operated doors are (and the RAIB agree - "[...] trains operated by Southern [...] are equipped with external CCTV cameras which cover the area along the sides of the train, and provide images which are displayed in the cab. These images are suppressed once the train has begun to move. It is important that the driver is not distracted from observing signals and the line ahead during departure from stations." The images for most of these systems are not recorded and many have a lag of 0.5-1.5 seconds - presumably due to compression required to send them over the air to the monitors in the cab. Some images are displayed on screens on the platform, which can be affected by dirt and sunlight and which pass out of the driver's line of sight within moments.)
(4) anyone who says "well, the tube has been DOO for years" has missed the fact that stations are categorised as A or B type depending on the amount of curvature / amount of platform that can be seen from the cab, and class 'B' stations require a member of staff on the platform to assist with train dispatch - though the ultimate responsibility for starting the train is that of the driver. Even class 'A' stations have SATS (Station Assistant, Train Services) - they hold the baton up to signal that they think it's safe to depart. In the end, the tube is quite a controlled environment compared to the mainline.
I believe that many of the complaints revolve around the fact that most other train services (ignoring intercity routes) don't have a guard on the train. Therefore why does Southern require this where others don't?
Yes, having an additional person on the train ought to improve safety to a certain degree. Whether or not this makes any real difference is up to a lot of arguments and will naturally depend on the trains and stations involved.
"most other train services [...] don't have a guard on the train"
Really? Strange definition of "most". Look at Section 3.29 of this 2014 consultation document which says "For example, trains across the region are currently operated with both a driver and guard. However, on almost a third of train services in Great Britain, including many recently electrified routes, everything is controlled by the driver in what is known as 'driver only operation' (DOO), with no need for a conductor or guard to operate the doors, or for train despatch. This means any other staff on-board the train are able to focus on customer service and revenue protection."
This means that 2/3 of the services are not operated solely by the driver. Having just one person responsible for the train is ridiculous. Just resetting a passenger emergency alarm (aka a passcom) can take five minutes of walking down the train - if the driver is doing that instead of driving the train... and then there's the communications equipment to the line controller which is located in the cab - so when the driver is away from that, there's no communication with the controller (they can take a portable radio, but most trains do NOT have an onboard radio system, so messages sent by the line controller can't be guaranteed to be relayed along the whole length of the train)
I'm of the opinion they should replace the rubber buffers on doors with interlinkings blades that slice things off that get wedge in closing doors. Fuck em.
Don't see many people blocking the doors and Hayes anymore. Ealing Broadway on the other hand, yes lets all try and get into the front two carriages that are always already pretty full leaving paddington and ignore the other 3 or 4 carriages. I remember when they managed to break the doors on two trains with their amazing intellect. God I hate ealing broadway afternoon commuters.
Interlinking blades!
That should be just a start to improve passenger discipline. I suggest large rotating knives on the train sides (inspired by Boudicea Chariots) to keep passengers from getting too close to the platform edges. I accept that you would have to fit drainage channels in the platform pits of busy stations to sluice away the blood but give it 20 to 30 years and the indigenous rodents that live in all stations will evolve a taste for human remains and act as nature's clean-up crew.
You could then use said plump rodents to feed the LUL staff and make further operating cost cuts by eliminating staff feeding arrangements!
"most other train services [...] don't have a guard on the train"
For what it's worth, in comparison, here in Germany, where there haven't been ticket barriers for many, many years, all trains except suburban ones have guards, who do signal from the platform that the train can leave and keep one door open for the inevitable last minute arrivals. Making sure the train doesn't leave with someone half-out the door is probably more difficult than driving the damn thing!
On the fully-automated skyrail at Düsseldorf airport, people get in and out of the (tiny) trains using different sides, which are never open simultaneously. I guess this is because people with luggage trying to compete for door space is probably a perfect definition of chaos!
Getting rid of ticket barriers and stupid, stop-based pricing would also be a great idea. Otherwise there's not a lot you can do with much of the UK network without spending vast sums of money on new track, signalling and roll stock.
"Getting rid of ticket barriers and stupid, stop-based pricing would also be a great idea."
Indeed. Anything which encourages people out of their cars and onto public transport inside zone 4 would be a good thing.
Then again, whilst a travelcard is nearly 20 squid from my station (I'm outside Zone6), it gets me to Waterloo/Victoria in 45 minutes at worst (vs 2-3 hours by car) and parking alone runs to about £12/day, so it's economic. If they could solve the issue of "no services home past 11:30pm" I'd be much happier though. (Night busses don't go to many places in zone 6, let alone beyond and it's a £40 taxi fare from the last night bus stop to home)
WRT "spending vast sums of money": This is exactly what's needed, along with using eminent domain more often. The UK rail network has suffered decades of underinvestment(*) compared to our european cousins and it shows. The money spent on Trident would go a long way towards a decent rail network (as would starting work on HS2 simultaneously at Birmingham and Manchester, building towards each other and south from B'ham. By the time the line reached the Chilterns they'd have finally decided on final routing and it could be completed quickly, vs taking 20 years from the time routes are decided.)
(*) Under private companies, BR and Railtrack alike.
Southern claimed that 961 trains where cancelled due to the strike on Monday http://www.bbc.co.uk/news/uk-england-36987040 and is running 341 less due to its emergency time table http://www.standard.co.uk/news/transport/southern-rail-crisishit-operator-confirms-it-will-cut-341-trains-in-new-timetable-a3288026.html
The idea that automatic trains are bad or unreliable isn't the correct conclusion here.
It's that *this* system is (possibly) unreliable. Although 750 is less than two per day on (one of?) the worlds busiest and most complex underground railways, so I'm curious if this actually constitutes a "big number" compared with other failures and other networks.
As a counterpoint, the Singapore underground system is fully automated, and from what I understand is very reliable.
One of the problems with the UK is we tend to look at a failing system (education being the classic) and assume that all the problems are inherent to the system, rather than down to a poor implementation, and so we chuck it all away and bring in something new. Because we all know that the best way to fix something is to completely change it every few years...
Can help you put that number in perspective. Central Line plans to puts 79 trains into service every day. They make an average of 11 trips per day. They stop at around 40 stations per trip.
So, for the Central Line alone (the oldest and least reliable ATO currently on London Undergound) the number of times a train stops in ATO per day is somewhere in the low 30k area.
And that's just Central Line.
The Central central system very rarely falls over - it has happened, but it is pretty rare. Most ATO failures are as a result of the onboard ATO or ATP controllers locking up - or not talking to each other. Most of the time the delay is the time it takes to trip the ATP MCB, trip the ATO MCB, reset the ATO MCB and reset the ATP MCB. In that order. Getting it wrong makes for a longer delay.
The main problem with ATO in wet weather is that the braking rate is too much for the low rail adhesion. So the wheel slide protection cuts in (like ABS on a car) and stops the train in the shortest possible distance. Which tends to be longer than the dry stopping distance. So the train overruns the platform (a SPAD in ATO if it passes a red signal which is a pain and a 7 mins delay and pisses everyone off; if a green signal the train can continue to the next station which only pisses off the people who wanted to get off).
To avoid this, Central Line normally derates back to mk1 eyeball, manual control and "defensive driving".
"The main problem with ATO in wet weather is that the braking rate is too much for the low rail adhesion."
In other words, ATO is fine. The implementation is a failure.
Design for braking distances must cater to "worst case conditions", not "optimal".
Someone needs taking out the back and re-educating with a brick to the side of the head.
A railway is relatively self-contained, sure there are occasional junctions/points and the like but basically trains can go forwards or stop, and maybe occasionally reverse direction. It's rare to find people or other obstacles blocking the lines.
Not so a car on a road.
Not really: there are systemic differences because the cars are autonomous and don't have the same kind of hop-on, hop-off operation. But a lot of the data gathered from these kind of services, especially failing safe, has gone into the systems developed for cars.
Personally, I think humans are incredibly not suited to driving cars in traffic: we are so easily distracted, so roll on fully autonomous vehicles.
So 750 odd incidents of the automated system failing safe (how many of these were resolved by turning the train on and off again? I'm not jesting) which is exactly what percentage of total service hours for the line.
Versus number of sick days taken / strike action as a percentage of total service hours. Versus number of failures of the two systems not fail safes. Which is 440 odd life threatening incidents where a train went passed a signal at danger when humans were in charge where as while the automated trains had 10... Gee which makes me feel safer...
So the automated system definitely comes out better then human drivers when it comes to putting passenger lives at risk.
And also likely comes out better in train availability due to driver out of action incidents.
Most by trip and reset of the ATO / ATP system (see earlier reply). For length of delay, also see earlier reply
Not Fail Safe is EXTREMELY rare - even an alleged wrong side failure causes real ructions in the Signalling Team.
NOTE : A train running a red light and stopping is not Not Fail Safe. It is the system working as it should in a failure (human or machine) situation. A train running a red light and Not Stopping is definitely Not Fail Safe. There is always* a section that is unoccupied between trains. So you get section occupied by train, empty section, section occupied by back of next train, section occupied by front of next train.
* the only exception is when there is a dead train and one is drawn up behind it to push it out. Then the driver has detrain the passengers beforehand, and approach under direction at extreme caution speed, applying The Rule for passing a signal at danger.
I've always thought it's one of those obvious but smart things when they have to do a removal of something from the track at a station, and they stop a train at the tunnel mouth with just the driver's door on the platform - the driver gets out, locks the train & hands the key to the person going onto the track; so there is no way that train is moving until the person is off the track. So obvious, yet so elegant.
Not Fail Safe is, IIRC, called Fail Dangerous.
As opposed to electronic door locks, which are classed as Fail Safe and Fail Secure, presumably failing safe for a door lock means the door can still be opened. For some reason an idiot thought it would be good to have the only door into our server room fail secure, so that infrastructure was protected in the event of supply loss. As I was delivering a whole pile of switches to there, I noticed the peculiar arrangement of lock, pointed it out to the building super who insisted it was fine until I got his lackey to trip the circuit breaker for the door's power and locked him inside the room. As there was no manual release on the inside, they couldn't really use the room properly for 4 months whilst they arranged to have a different lock (with a break-glass release) fitted. Managed to fit the switches, but it meant we had to have at least two people present at any one time and the door was very firmly wedged open with a breeze block. No they didn't bother replacing the glass after the first time it was needed, the power failed so often.
>>> So the automated system definitely comes out better then human drivers when it comes to putting passenger lives at risk.
The greatest risk by far of fatal injury to passengers related to train movement is when someone finds themselves on the roadbed in front of an oncoming train. I've had this happen to me six times in my short career as a train operator. I never hit anyone, but I came damn close a couple of times. I hate to think what might have happened had I not been there to recognize the problem in time to stop my train.
The last fatal accident involving a SPAD in New York was in 1991, which killed 5 people. I think the accident rate from being hit by a train in New York is about once every three days. This doesn't include the many near misses where tragedy was prevented by alert train operators.
I'm very confused at the assertion a train driver's "systems" are always operational (except when they're not - on strike, off sick, doing "training", "union activities" etc.).
By definition a SPAD incident on a manual train means the train driver's systems were not operational. And the statistics indicates the train drivers failed more often than the automatic systems.
Very strange conclusion from this article.
My old Ford Cortina got nicked from there many years ago. Recovered in a nearby housing estate where someone had 'borrowed' it to get home on a rainy evening.
Even better, the two Superbrain chassis' that I had in the boot (completely non-working, do not pass go, do not collect boot prompt) were also gone. Which saved me the hassle of driving them to the tip.
No damage to the car - you could open the locks with a penknife and the borrower obviously knew that.
LUL average 460 ish trains in service at any one time (higher in peak, lower off peak). Number of trips up and down the line vary by line, but if you assume somewhere in the 5-6000 area won't be far out. Multiply up by, say 300 or so to allow for this, that, planned closures, weekends, Xmas day etc and so on and you get somewhere in the 1.7M range ...
Stop the presses, systems can fail. Safety critical systems can have a tendency to fail more often, but in a safe manner.
The Central line was one of the first computer based signalling systems, and doesn't have quite the same levels of redundancy as more modern ones. The onboard controllers are also suffering from 20 years of living on trains.
Seltrac is well known in the industry for catastrophic failures when a VCC goes down. Having said that, it seems to work well enough on the automated DLR and countless other driverless metros worldwide.
Discussing SPADs in the same context as technical failures is kind of misleading. Most SPADs are a result of human error, and are generally regarded as a bad thing. So this is an argument for more automation, not less.
After the dot-com bust I spent 6 years working as a train operator for New York City Transit--the subway system. We have both one person and two person operation here, and I've worked both.
There are two problems with one person operation as I see it. First, it makes for slower passenger loading at stations. Our conductors (guards) work from the middle of the train and have a much better view of the platform and can better control door closures to make sure folks get completely on and off the train without blocking a door and delaying the train. The second problem is that it creates a distraction for the operator by having to switch from conductor mode to train driver mode at each station.This adds additional dwell time at each station.
These problems are particular to our implementation of one person operation. I often visit London and think they have done a much better job of it. In particular, London provides plenty of passenger control people on the platforms to help with loading. We have hardly any here.
We have one line in CBTC operation here. It is one of our shortest lines, with little interaction with other lines. The project ran way over budget and schedule and is far from trouble-free. Station overruns are common, as are mysterious system glitches. I often wonder if the enormous investment could have been better spent on other pressing needs.