Tuesday 23rd January 2024
Technology company Prospective and First Bus are seemingly very keen to promote and publicise the use of AI in schedules and timetable compilation.
Following the article in trade magazine Coach & Bus Week last month, the story appeared on the BBC News website last Thursday repeating the claim about “20% more punctuality during some peak periods” (note: no definition of “some” nor context for “20%” is given, rendering it a meaningless claim) and a benefit lauded as “AI’s greater processing power and ability to learn enables the company to alter timetables more often if needed ……. it can also make automatic adjustments on any given day if required by road congestion, which, the company says, helps to prevent bus services bunching up.”
Constant changes to timetables is are the fastest route to putting passengers off travelling and you don’t need expensive AI to prevent buses bunching up due to road congestion “on any given day”, you need a human being overseeing the vehicle tracking and controlling the headway.
Still, it all sounds impressive to those without close knowledge of how the bus industry works and I’m sure all this news coverage will be making top dogs at Prospective and First Bus feel good about what they’re doing.
I’d intended to follow up my recent blog about First Bus’s use of AI in timetable and schedule production, which led to one of the biggest responses from readers yet, so following last week’s BBC News coverage, I thought I’d carry out a more detailed examination based on my own travel experience of how these new, so called punctuality proof, timetables perform in practice.
At the suggestion of one of the commentators from that previous blog, I’ve carried out a detailed analysis of the new timetable set to begin this Sunday on First Berkshire’s route 8 which runs between Slough, Windsor, Egham, Staines and Heathrow Terminal 5.
The existing non AI timetable ends this Saturday. It’s been in operation since last August. It comprises pretty much a standard half hourly frequency operated by five buses with departures at regular intervals from both Slough and Heathrow Terminal 5. The first journey leaves Slough at 05:15 (Mondays to Fridays) followed by departures at 06:00 then 00 and 30 minutes past each hour until 15:30, 16:05, 16:35, 17:10, 17:45 and the last departure at 18:50.
From Heathrow Terminal 5 departures are at 06:27, 07:13, 07:51, 08:25, 08:57 then 27 and 57 minutes past each hour until 15:57 then 16:40, 17:13, 17:44, 18:15, 18:46 and 19:20.
Off peak journeys between 09:30 and 14:30 all have a standard 78 minutes running time in both directions with up to 10 minutes added to peak journeys.
The daytime timetable presents itself as an easy to remember half hourly service.
All that is swept away from next week when a new AI generated timetable is introduced whereby, unbelievably, every single one of the 52 journeys (26 in each direction) on Mondays to Fridays has a unique distribution of running time between the intermediate timing points. There are not even two consecutive journeys across the whole day with identical timings. The differences are mostly variations of just one or two minutes. Similar variations apply on Saturdays and Sundays.
You can see the detail in the spreadsheet I’ve compiled below. It shows the journey time between each timing point in minutes. For example, whereas the 09:36 and 10:06 departures from Slough (one of the few gaps between buses leaving that terminus that is exactly 30 minutes) both have 74 minutes running time to reach Heathrow, the distribution of that 74 minutes varies between six of the ten timing points by one minute making for uneven departures (ie not exactly 30 minutes) from those points.
Which, frankly, is stark raving bonkers.
It can’t possibly be the case traffic conditions vary from one journey to the next across six timing points that justifies a minute’s difference that is worth building into the timetable. All it does is present a timetable to the public which appears unduly complicated and off-putting. Rather than making the service more punctual and reliable to attract passengers as proponents of this latest tech derived development claim, it’ll increase exasperation and frustration at the complexity of it all and put people off.
The spreadsheet is littered with similar such unnecessary complexities which, when all put together, makes for an embarrassment of a timetable to present to the public.
However, that’s all a desk based analysis. In BusAndTrainUser Verify I like to actually test out the veracity of whatever is being claimed in the media by taking a look on the ground.
So I headed over to Slough yesterday to give the service a try out.
Now, obviously whereas the AI software is analysing hundreds of journeys (the BBC News story quotes to train its AI, Prospective “says it uses billions of data points”), I am only sampling one journey in each direction but it gives a snapshot of whether this whole exercise is another Kings New Clothes or whether there really is something in it.
Spoiler alert: if I was running First Bus I’d save the company £4.5 million by immediately cancelling the contract with Prospective. I reckon it’s no better, and possibly inferior, to what a seasoned human timetable planner, with intimate knowledge of his or her local patch, could achieve.
I caught the 11:30 departure from Slough yesterday morning. It’s scheduled to arrive Heathrow Terminal 5 at 12:48, giving 78 minutes running time. Next week the equivalent journey departs 11:33 with an arrival at 12:50 shaving a minute off the running time.
I’d noticed the bus was significantly late heading into Slough on its previous journey and in the event instead of arriving at the scheduled 11:15, pulled up to the stop at 11:33.
Six passengers boarded and we departed at 11:34.
I’ve shown the journey progress in the table below.
We encountered two sets of temporary traffic lights. One, in Old Windsor, delayed us by one minute and the other four-way set in Englefield Green delayed us by six minutes.
The route crosses two level crossings – one in Egham and the other in Slough. The former was open as we passed through but we got held for two minutes at the latter.
We stopped at 19 bus stops with 37 passengers carried.
As you can see our end to end journey time was 76 minutes – two minutes less than the non AI timetable applying this week and one minute less than next week’s AI version.
None of our departures at the timing points along the route were in accordance with either the non AI or the AI timetable, so that objective wasn’t achieved.
It leads me to conclude there’s absolutely no point nor benefit from the AI generated timetable. We encountered nine minutes worth of delays and met the overall existing schedule. Next week those temporary traffic lights might be in a completely different location so no AI magic dust will counteract that.
The return journey from Terminal 5 was even more interesting.
As you can see from the spreadsheet earlier, AI has given most Slough bound journeys a very generous extra dollop of running time compared to this week’s timetable, in particular between the last two timing points – Chalvey Windsor Road McDonalds, on the southern outskirts of Slough, and the terminus in Slough town centre.
Whereas this week’s 12:57 departure from Terminal 5 has just four minutes, from next week, the same timed departure at 12:57 has a whopping 12 minutes – three times as much. That’s mainly why the overall journey time has increased from 78 minutes to 94 minutes.
This week that journey is timed into Slough at 14:15. Next week it’s 14:31.
It’s as if First Bus are worried they’ll be liable to Delay Repay and are using the technique favoured by many train companies to add padding to the end of the journey.
Google maps tells me walking time between these points is 15 minutes and I can usually shave a couple of minutes off that even at a moderate pace. First Bus’s £4.5 million AI timetable construction is allowing 12 minutes by bus (and 13 minutes on some journeys).
We left Terminal 5 a minute down at 12:58 but the driver obviously knew what he was doing. By the first stop at Stanwell he paused for a minute to kill time and even so we arrived into Staines bus station two minutes early at 13:10. From next week AI reckons we won’t arrive until 13:19 – that’s ten minutes for the driver to lose and then gets three minutes pause time in the bus station leaving at 13:22. There’s almost time for passengers to nip into the adjacent shopping centre for a quick purchase.
We left the bus station at 13:15 (this week’s time) reaching Pooley Green shops at 13:21 where we paused for another minute (not being due there until 13:24, and next week not until 13:28).
We’d passed over Staines level crossing but this time got caught at Egham’s including a four minute wait for two trains to pass.
Despite this, by Egham Church Road we were spot on time at 13:32 on this week’s timetable – next week it’s 13:40.
Then we met the four way temporary traffic lights at Englefield Green which cost us eight minutes including two to sort out a ticket problem with a passenger wanting a return but the driver had issued a single.
Fortunately we sailed through the next set of temporary traffic lights and arrived at Windsor Hospital at 13:59, four minutes down on this week’s 13:55, but four minutes up on next week’s 14:03.
We’d clawed a minute back as we arrived at the Parish Church at 14:02 instead of 13:59 (next week 14:07) and another two minutes by McDonalds, passing there at 14:12 instead of 14:11 (next week 14:19).
And then, as described earlier, it’s all fantasy timetabling from next week for the final run into Slough. We picked up another minute compared to this week’s schedule doing the journey in three minutes arriving spot on time at 14:15. Next week we’re not due in until 14:31 and are given four times the timing we achieved, ie 12 minutes.
We’d stopped at 17 bus stops and carried 32 passengers.
I can only conclude from yesterday’s journey experience First Bus have been duped with all this AI nonsense. Despite every single journey having its timings altered from next week, my experience shows there’s nothing wrong with the current timetable during the off-peak. It pretty much worked bearing in mind the vagaries of two level rossings and two sets of roadworks.
But with the new timetable passengers and drivers will suffer a worse travel experience – in the Slough bound journey example, a slack journey time of over 20 minutes in a theoretical 94 minute journey – as well as an uneven headway, non-sensical, unable-to-remember, timetable.
Maybe instead of wasting time and huge amounts of money on useless technology like this it might be better to give some attention to vehicle presentation and how the company’s brand is portrayed to passengers.
Almost twelve months on, two of the five buses out on route 8 yesterday were still sporting red branding from two withdrawn Southampton routes in February 2023. Two others were wearing First Berkshire blue livery and the fifth was in yellow BeeLine livery.
Does nobody at First Bus care about confusing passengers with nonsense timetables and a sloppy, inappropriate image on the road?
Roger French
Blogging timetable: 06:00 TThS.
Comments are welcome but please keep them relevant to the blog topic, avoid personal insults and add your name (or an identifier). Thank you.
BusAndTrainUser 
I know I was in a minority in the last set of comments by saying I didn’t think it was a problem on an urban service with a 10-15 minute frequency as most people would go straight to the journey planner for the next arrival. This, however, is utter madness. In terms of presentation to the customer its horrible and as Roger says its impossible to believe that the endless +/- 1 minute fiddling with the running times will have any serious impact on the actual reliability. I’m not even going to try and guess what’s causing the 12 minute running time in Slough, that seems like the underlying data is screwed up in some way.
LikeLike
A bus that spends considerable amounts of time sat around because it’s early (making public transport even less competitive), with confusing inaccurate branding for another town tens of miles away, and the high possibility of missing the bus because you can no longer “leave at :xx in order to get the :30 service” with a new timetable that’s about as unfriendly as it’s possible to get.
But it’s an AI timetable and that’s the trendy technology now so it’s got to be good. Ugh!
LikeLike
This isn’t even AI, artificial intelligence. All this is doing is capturing actual running time data from the ticket machine and coming up with an average running time for each journey. There is nothing intelligent about that. And anyone who travels on our roads knows that every day their journey will be different for a huge number of reasons, including the weather. And of course in many areas the difference between school/college holidays and school/college days is significant as well. This is yet another attempt to reduce costs by centralising scheduling and getting rid of regionally based people who know their patch by more senior managers who have no idea how to run a bus company.
As a competitor of First I’m all in favour. The more bonkers ideas the better for the rest of us.
LikeLiked by 2 people
AI though can collect a vast amount of data and schedule more realistic AVERAGE running times. What it cannot do is take into account unplanned roadworks
Clockface timetables may look pretty but in most cases they are not realistic and the buses do not keep them
LikeLike
Traffic is generally unpredictable from one day to the next, from one hour to the next. Buses won’t keep exactly to clockface timetables, or to timetables generated by rolling dice and flipping coins like this one – so the best bet is to approximate the times as best you can in as user friendly way as you can. Clockface timetables are easy for passengers to use, and everybody recognises that buses won’t arrive to the exact second. AI generated garbage won’t increase reliability one iota and it throws the user experience in the bin.
LikeLiked by 1 person
Anonymous at 7.21 is correct, because of numerous things, roadworks being only one of them, buses cannot keep to clockface timetables. But neither can this AI produced version, for exactly the same reason.
LikeLike
Another downside to this waste of money is that it will encourage a mindset amongst the local and district councils that AI will solve unreliability problems, therefore less need for bus priority measures.
LikeLiked by 1 person
It would be great if this new scheduling did help operators improve reliability. Excellent outcome. And if that means less need for bus priority that’s also good. Except bus priority also helps with reducing journey time which is a good. I wonder how many local authorities really are reporting accurate bus reliability stats to the DfT each year? I suggest not.
LikeLike
This is a bus route I use quite often and the new system sounds even worse than I thought. While the random nature of traffic delays means buses will rarely if ever turn up at the exact odd times quoted, I hadn’t realised about the extra running time built into Slough bound journeys. I wonder if that is the AI trying to take account of the frequent but, in reality, unpredictable delays which affect buses going west out of Staines (particularly) when the M25 southbound jams up south of J13 and a lot of traffic heads south down the A320, clogging the road past Staines Bridge. On one recent occasion my bus left Staines bus station on time but took 20 minutes to get across Staines Bridge. But we were only 10 minutes late on arrival in Slough even with a wait at the second level crossing. So the current timetable with sufficient turnround time at Slough, coped fine.
LikeLiked by 1 person
As a regular traveller on that route it is a nightmare, using live tracking via the app is the only way to know where a bus is and even with this yesterday morning it was 15 minutes later than indicated towards Heathrow. Journey times towards Windsor are now so slow it’s better to head into Staines from Pooley Green and catch the train. Don’t see how AI is making any difference to passengers here
LikeLike
As a note on the quality of the fleet it is awful. I think the previous buses used are now on the Bright Edinburgh Airport service. The ex Southampton fleet have no luggage space which is terrible for a Heathrow Route.
With regard to timings many local people use it to connect to the Elizabeth Line at T5 which is half hourly to Shenfield and cheaper than the adjacent SWR line, also it is not as prone to strikes however the random bus times mean it is almost impossible to guarantee any kind of connection.
LikeLike
I’ve noticed that fleet presentation and image across First’s southern division has taken a downward dive since a particular senior manager left last year to pursue other business interests, just when it seemed to be improving. It seems there’s a tendency now to focus on what’s considered to be “trendy” rather than the actual basics.
These images from Slough sadly remind me of the worst of deregulation time in the late 1980’s when an independent operator would buy whatever it could at a bargain price and put it into service in “as acquired” condition .
I’ve just re-read the excellent book on Badgerline written by Martin Curtis and Mike Walker, two of it’s former directors . Having grown up on the southern edge of the Badgerline patch, it shows how much local strength and goodwill was destroyed by First and although that was decades ago, it seems the lesson still hasn’t been learned.
Keith Bryant
LikeLike
I’ll have to correct anon 08:55. The Bright Bus service in Edinburgh uses former First Borders vehicles that have recently had luggage racks fitted. I struggle to recall the previous vehicles (Volvo 7900 and Citaros) having much in the way of luggage racks either.
The lack of attention to detail (i.e. out of area branding etc) is all too apparent in First after a period when things did seem to improve. A consequence on cutting the management overhead when they restructured?
Sadly, the move to AI designed timetables (and yes, the other poster has pointed out that it isn’t AI… It’s simply automated timetable construction based on data point averages rather than learning) is just another way in reducing back office overheads. However, you can set certain parameters that at least maintain clockface departures from the main timing points. It’s a mess.
BW2
LikeLike
For many years It’s been the practice in London to schedule extra running times when conditions justify it – for example between 0700 and 1000 on weekdays , between 1500 and 1900 on weekdays, and generally on Fridays. That is acceptable from a scheduling point of view because the routes mostly don’t have clockface timetables and most are “turn up and go” for pasengers.
But we all know that it’s an operational nightmare, with long intervals alternating with bunching because of the random effect of incidents which slow the overall progress – eg bad weather, roadworks, bad parking, police activity, traffic light misalignment, vehicle problems, slow crew changes, disruptive passengers. These are the factors which challenge any scheduler, whether AI or busman, and they lead to excessive regulation (eg short turning to avoid shifts overrunning) and excessive waits en route (especially annoying at the last few stops before the terminus).
Outside London there is less timing variation over the day, and in practice this means that bus drivers confidently set off 5 or 10 minutes late in the offpeak because they know the slack timing allows them to catch up the schedule. However at peak times the schedules are too tight and late arrival is the rule not the exception. (I see that the Stagecoach MK1 leaving Bedford for Luton at 06:45 on Mondays has an extra 15 minutes allowed, and that the next departure which used to be at 07:45 and was changed to 08:05 at the last timetable revision is to revert to 07:45 shortly.) But on busy urban corridors such as Banbury Road into Oxford, or the A6 from Walkden into Manchester, traffic is inevitably slow and the schedules don’t cope well.
Ian McNeil
LikeLike
In the more rural areas morning and afternoon times and even the route varies to fit around school hours in fact even trying to get a taxi at these times is a nightmare as they are all doing school runs
LikeLike
Variable running time has been used in London going way back many decades and is also widespread among competent urban operators elsewhere.
Doing what you suggest, scheduling for the peak running time, would be operationally completely unworkable because the inevitable early running or holding back at the start of the route would cause congestion at stands, as well as an erratic service due to each driver interpreting differently what time they need to depart at the start of the journey.
Steve
LikeLike
Having extended running times in the peak with non-clockface timings during the peak period & the reduced running times off-peak with clockface timings (& often even more reduced in evenings & early mornings when traffic is lower) is standard practice in the provinces and pretty much has been for the 20 years I have been a scheduler. It is normal to see timetables with lower frequencies & irregular departures during the peaks where running times are extended to reflect the extra road traffic, but due to cost constraints there are few extra resources so you can’t just drop a couple of extra buses in to maintain frequency. Being a bit irregular in the peak isn’t ideal but partly this is viewed as workable as more peak travellers will have a fixed travel time around their work shifts so will catch the same bus every day so will have regular routine whereas leisure passengers will be more random when they leave home so a regular headway is easier to aim at.
LikeLiked by 1 person
In the interests of balance . . . if the route is high-frequency, I kind of get what they’re trying to achieve. In TfL-land, operators are penalised if the route doesn’t meet quite stringent headway targets; if certain journeys are always affected by traffic or other delays, then flexing the run time will help in meeting such targets, and Joe Passenger won’t really notice (every 9-11 minutes is sufficiently accurate in such cases).
The problem comes with low-frequency routes. In Roger’s example on Route 8, the new timetable is plainly daft, and I can only guess at where the timings have come from. As has been mentioned earlier, this isn’t AI as such . . . as the original piece in Route One says, millions of data sets (a posh name for times at bus stops!) have been used to aggregate average running times.
This isn’t new tech either; it’s been available in London as a by-product of Countdown for around 10 years now . . . it still needs analysis by the compiler, but a skilled compiler can average run times out in a day’s work for a sizeable route, and use those to compile the schedule.
I’ve previously mentioned Uno’s Route 602 . . . this is a particularly daft new timetable (on Mondays-Fridays only), and is found here: https://passenger-line-assets.s3.eu-west-1.amazonaws.com/intalink/service/fe056966-43bd-4a90-afb4-2ed27b4b8b8e-timetable-20240107-7c4267b9.pdf (hopefully the link will work; if not go to the Intalink site and ask for a 602 pdf).
Departures from Watford are at:1022; 1052; 1126; 1201; 1226; 1259; 1328; 1359 . . . departures from Hatfield are similarly random. However, the running times are exactly the same during the daytime on every journey! There is no flexing of run time at all . . . I couldn’t understand why this was so . . . until I checked terminal allowances at each end . . . exactly 15 minutes stand time between every trip !! So the daytime timetable is derived from the stand time after the AM peak trips.
I’m sorry, but this is nothing to do with “AI”, but the inability of the compiler to set a parameter correctly in the program used (minimum stand time all day at 15 minutes at each terminal) . . . and more to the point, to actually check their work afterwards; a cursery glance would’ve shown what was wrong. If I’d turned in a timetable such as this to my manager, I would’ve expected a sharp comment, and been told to “do it again, and do it properly this time”.
There is no substitute for (a) experience and (b) a second glance.
Sorry this is so long, but if I was dead, I’d be spinning in my grave!!
LikeLiked by 1 person
The timetabled running time on our local high-frequency route in Cardiff varies between 41 minutes (early morning/late night) and 55 minutes (peaks) with a “standard” time of about 48 minutes. But even this is a counsel of perfection, especially around afternoon school time and 5-6pm.
Andrew Kleissner
LikeLike
I think the 8 would be almost impossible to timetable reliably. Lots of comments have already mentioned some of the factors but:
Staines town centre traffic is incredibly unpredictable both in bound from the A30 and in both directions over Staines bridge. If there’s a major incident on the M25 it can often cause gridlock in Staines as traffic tries to divert.
There are 2 level crossings with a minimum 8 trains per hour. These have attracted news coverage due to increased down time of up to 15 minutes following resignalling. Ironically one of these was the much quicker automatic barrier type until an 8 driver caused a train crash by stopping on the crossing!
Then as Roger experienced there always seem to be temporary lights somewhere. End result is high variability in running time that’s hardly ever in the same place. I think the only way to approach this is to allow some recovery time at Staines and Windsor and then allow generous layovers at either end so at least you’ll get a right time departure at the start of the route.
Surfblue
LikeLike
Definitely agree on the branding point. If this is a route that First is drawing attention to then it should be ashamed that the route 8 bus says 6 on the side, let alone Netley and Hamble. Not sure how a passenger is supposed to know which bits are right and which to ignore. Transforming travel at its best. I expect they are waiting to be repainted but if it was bee network and not bee line then they wouldnt stand for it
LikeLike
Something I’ve been saying for a while is that rather than AI (more like ML) generated timetables, the real opportunity here is AI being applied to routes.
Anyone who has an Android smartphone (unless you turn the feature off) has their travels recorded. This data can be anonymised and made available to build a picture of where people travel from and to, and at what time. Using that to plan service provision would be much more transformational than tweaking times by a few minutes here and there, but still not providing what passengers, and more importantly potential passengers, would actually benefit from.
I’m sure there will be some who are paranoid about invasion of privacy, but there are plenty of safeguards in place to stop individuals being identified and this is one of many areas where the big data collected by phone companies and OS providers could actually bring significant benefits.
LikeLike
The driver distraction trying to keep to the correct times must be phenomenal.
LikeLike
The excess time in the last timing point will see no waiting and effectively just extra layover to cater for the various delays that could happen as being mentioned – but actually adding time in to give the customer an indication of when they might arrive, so if a customer arrives early, they are not unhappy… Isnt this what airlines do?
LikeLike
I’m trying to decide if this is good or bad in terms of anyone wanting to make a connection. Ones that would previously be viable may now seem impossible, while what looks like a tight connection on paper will actually be viable most of the time.
One question though, I’ve been told by my local operator that buses can run up to 5 minutes early at intermediate stops – is this law or just their policy? If it’s law then, unless there are no intermediate stops between two timing points, then the bus can’t actually arrive significantly early at the final timing point without exceeding this.
LikeLike
The Traffic Commissioners before Covid and my retirement used to rule that no bus should be more than one minute early or five minutes late at timing points. Hence the paranoia by some operators to increase journey times especially at peaks, but also for some differentiate times between school days and school holidays.
LikeLike
I’m puzzled how First Group can justify spending £4.5m on this. I presume they know something about running buses without it. Unless this is being funded by someone else this looks like the cost goes straight to their bottom line.
Further the current timetable gives off-peak arrivals in Slough at xx15/45 with departures at xx00/30 giving a 15 minute turnaround. In next weeks timetable the journey you took is scheduled to arrive in Slough at 14:31 with the next departure at 14:34. I presume this gives a highly risky three minute turnaround or requires an extra vehicle to be scheduled.
On a different matter I hope you’ve noticed that DDRT problems have made the national media. This appeared in the Telegraph on Sunday:
https://www.telegraph.co.uk/news/2024/01/20/somerset-row-council-axes-bus-routes-villagers-stranded/
LikeLike
Yes … But, if millions of ££ are being thrown at this, why can’t First etc. throw similar amounts at the real problems which are causing buses to be late or unreliable? e.g. –
Level Crossings: as far as I gather, the criteria for spending money to eliminate these is safety related, and bus issues are not considered. Surely money could help here, plus CEOs taking up the issue with their politician friends?
Roadworks: the bus industry as a whole, plus the other sections of the ‘road lobby’ (most) who would support the idea, should press for better pre-publicity (to enable car/lorry drivers to divert), extra buses to keep the timetable the same as far as possible, and extra information for bus passengers at stops, on web-sites etc. – all to be paid for by a levy on whichever utility is digging up the road.
School-time congestion: perhaps the bus industry could offer a deal to schools – you introduce a staggering of school bours to spread travel over say three hours in the morning and three hours in the afternoon. We (the bus companies) will put on buses every 15 minutes over those times.
Time spent selling tickets: lots of options (e.g. copy what tram operators do) – this will reduce staff costs proportionately, and increase income due to a faster more reliable service. Part of the extra profit can go to paying drivers more – so less problems with getting staff!
Ever hopeful,
Rick Townend
LikeLike
I’d love to see the equivalent of road pricing for anyone who wants to do roadworks. If you insist on digging up a main road in the middle of the rush hour that’s up to you, but you pay a lot of money to the local authority to be able to do so.
It would incentivise concentrating workforce on one job at a time and working longer hours to complete it quicker. We seem to have 20 sets of road works on at the same time lasting several weeks, rather than a handful at the same time completed in a few days with no sign of anyone working most of the time.
A main route near us is disrupted for months so they can install some new water mains. I remember decades ago reading of microtunneling technology which would mean far less need to dig trenches and cause major disruption – no one seems to have taken it on, I suspect because they are only interested in minimising cost to them and don’t care about knock-on costs that others incur.
LikeLike
I remember the New Roads & Street Works Act from I think 1991, which laid down that proposals to undertake such works must be approved by the local authority, that works must be completed in a reasonable time (and the surface must be reinstated to at least the previous standard). There was a suggestion that the next step would be to charge the undertakers for disruption so as to stimulate them to get a move on – but I don’t recall whether that was enacted.
I wonder how much Network Rail may be coughing up for the disruption under the railway bridge in Oxford!
Ian McNeil
LikeLike
I was an engineer for Thames Water and built several new water mains in urban areas. I did indeed use micro-tunneling techniques to avoid excavations in roads. That was in the early 1980s! The problem is mainly that work to replace existing valves or to install new connections can only be done by digging down from the surface. I do agree that there should be time-related charges for obstruction of roadspace, with higher rates where bus routes are disrupted
LikeLike
From what I understand, the AI looks at average running times, including those at times when there is a driver changeover (so those specific trips have longer times)
So what happens when the duties get recut? The “problem” just moves.
LikeLike
From my dealings with First, I would suggest that they’re a dim lot with poor IT support, and that that is the reason for grasping at the “AI” panacea. I spent too much time in email conversation with them last year, trying to get them to see that timetables and route maps had disappeared from some areas’ websites, and to understand that use of the 26-hour [sic] clock is not helpful to would-be passengers.
LikeLike
Happy to be corrected on where the previous bus fleet went. We did have a fleet prior that had a largish luggage rack towards the front which was useful.
LikeLike
Catching up with some points raised above:
Traffic Commissioners Window: 95% of bus services must operate within a 1 minute early 5 minutes late window. Currently, this window is measured at a specific timing point, and is witnessed in real time by a VOSA monitor, and is compared with the registered timetable. It should be noted that timing points must be less than 15 minutes apart, unless the route is routed as being “non-stop” between two timing points.
{From the Stagecoach Devon Public Enquiry, published 18/11/22: All bus operators who run local services in England outside London are now required to publish their timetable, fare and location data to a central system (BODS) operated by the Department for Transport. DVSA has access to the data as a means of monitoring bus operator service reliability}.
That was around 15 months ago . . . at that time, it was noted that some data was not present, and therefore remote monitoring was unreliable. I suspect that operators have now been told that remote monitoring WILL take place, and incorrect or missing data will not be an acceptable excuse.
In TfL-land, a low frequency route must operate within a 2.5 minutes early 5 minutes late window, but this is measured electronically along the route at timetabled timing points; these are specified by TfL as part of the tender specification. On some routes, it is possible to have timing points only 1 minute apart in the early morning or late night periods, where the daytime run time might be 7-8 minutes. That can be the size of the variation on some routes!!
Building in additional running time at (live) changeover points can be easily done . . . simply allow an extra (say) 2 minutes at the relief point between (say) 0900 and 2100, or whenever driver changeovers are scheduled. If they’re not required, then the bus waits as necessary. This will require the compiler to use a modicum of common-sense.
Staggering school start / finish times . . . yeah, good luck with that one!! I ran a network of school routes that weren’t quite profitable. I spoke to two schools about staggering start/finish times . . . if they could stand flexing to 0810 and 0910 instead of starting at 0840, then I could run two routes with one bus and driver. I told them that, if they didn’t at least consider the option, I would pull both routes.
Guess what . . . I pulled both routes. They both agreed to not even talk to each other about any change. Neither school had a green transport plan in place. I understand that schools have many calls on their time, and this might seem of small import, but even so . . .
Sorry about the verbiage again, but having the correct information is important.
LikeLike
Regarding early running, I had complained to Stagecoach about a service departing an intermediate stop early and was told that as it was ‘only’ 5 minutes early this was deemed acceptable. It had actually left the previous timing point, only 5 or 6 stops along the route, 2 minutes late which shows how much padding there was. I clocked it as 6-7 minutes late but just got the ‘we’ve referred it to the depot’ fob-off at that point.
(As an aside, it was made more frustrating by the fact the driver saw me waving at him as I was trying to cross the road but still drove past the stop despite being considerably early)
All that said, where traffic is variable and congestion common I can appreciate how difficult it can be to timetable a service for all eventualities.
LikeLike
How do the drivers remember the running times unique to each journey? I suspect they will just drive and hope for the best. Also how will the Traffic Commissioners monitor compliance?
Peter Brown
LikeLike
Drivers’ dutycards will detail every timing point on the route served . . . basically they’re an extract from the timetable.
Back in the day, drivers were given start and finish times of each journey . . . and were issued with a timetable booklet, plus amendment leaflets as required . . . we actually managed very well!!
LikeLiked by 1 person
Technology could tell them
LikeLike
Technology does tell them
LikeLike
Still fascinated by the “new, exciting AI” twelve minute running time from Chalvey (McDonalds) to Slough (Queensmere) ….they are actually in sight of each other! Who checked this? Someone in King Street Aberdeen? Almost making some TfL routes look fast……And are First Berkshire Drivers going to be disciplined if arriving by the Queensmere terminus possibly eight-ten minutes early?
At least the old mantra chanted at us, “It’s all on the internet” when daring to ask for a paper timetable will, by default, become reality. Clearly bus companies falling for this King’s New Clothes nonsense, are presumably going to constantly tinker with intermediate timings or worse, have different timetables every day of the week, so Lord help those who don’t have a smart phone.
Terence Uden
LikeLike
Wonder if a StreetLite will fit through the McDonald’s drive-thru? That should kill a bit of time!
Keith Briant
LikeLike
I’d love to have 4.5 mil to spend on my bus company!
LikeLike
The enthusiasm of FirstGroup for, so called, AI designed timetables seems to reflect that senior management appointees, nowadays, generally have no bus industry background and operational experience. They probably never make use of the services they provide, and have little understanding of what passengers look for in a bus service. Small wonder that they fall for the sales pitch of companies like Prospective.
Locally based schedulers with an understanding of their services are, sadly, almost a thing of the past. Coupled with the rationalisation of local managers, and Managing Directors being charged with running several businesses, and as a result being too remote from the markets they serve the whole operational scenario is heading in the wrong direction. The inappropriate focus on A1 as a tool for producing timetables is just another nail in the coffin of the bus industry, in my view.
Graeme Varley
LikeLiked by 2 people
Absolutely spot on. Thank you Graeme.
17A
LikeLike
Possibly not seeing things correctly. So, the new generation, say someone goes to work most days shortly after 10am. Never seen a timetable, but push a button on a smartphone, and the next bus is 10:27. However, one day they decide to go to work later, and instead spend an hour ordering up some avocados for an evening meal and surf some NetFlix trailers. Now the next bus is 11:34. Person is pleased, because all the work-from-home folk get bored around 11am, and head off out for a coffee. This makes the roads congested. This recognition that the bus needs an extra seven minutes means, in theory, less time waiting at the bus stop. It’s the new way…. (maybe)
LikeLike
I wonder it that Staines Bus Station time is for a Driver changeover ? Cannot explain the Slough one though unless the new timetable has dropped in the next departure time for the previous arrival time (I have had access databases have pointers going to the wrong underlying data point in the past)
LikeLike
I work in this space and use this type of technology, and thank goodness for its availability. However, relying on data when you have no knowledge of the real world operating experience (eg route surveys) is a slippery path to poor service delivery. It may also be an indictment of the shortage of seasoned professionals and the mistaken belief there’s a cheaper technological fix to hiring and retaining staff.
LikeLike
Using AI to help compile workable timetables makes sense. The industry still at present relies on using archaic manual processes which give poor results
LikeLike
“Using AI to HELP”….yes, fine, but Anyone who has an understanding of bus operation and the day-to-day issues of unpredictable traffic, knows an Operator can NEVER achieve a perfect timetable. So let AI do the donkey work, then produce as near as possible regular timetable the public can understand, which does require local knowledge and a human eye for the finishing touches. Clearly not a priority for First.
Terence Uden
LikeLike
I wonder what the young lad on that destination display in the photo thinks of all this?
(“Windsor Boy’s School”)
Roger G, Oxford
LikeLike
Are they using AI in conjunction with the Real Time Bus information. That knows or at least it should where every bus on a route is and where it is relative to the timetable. You could even write a program to automatically regulate the service. Where this sort of thing can fail is unplanned roadworks and RTC’s that usually needs manual intervention
LikeLike
In London, a byproduct of Countdown and I-Bus is a screen in the cab that advises drivers of their position with reference to the bus in front and behind.
In theory, drivers should self-regulate. Controllers can also advise drivers to wait to even out the service.
In practice, human nature being what it is, some drivers will just keep going!!
LikeLike
We have this nonsense to look forward to on route 3 to/from Uxbridge; I’ll be checking how it affects timekeeping with interest. As we have yet to see even the current timetable displayed at stops (with TfL, First Berks and Hillingdon Borough all claiming it’s not their responsibility), the mostly elderly and generally non-computer literate users will though just continue turning up at random and hoping not to wait too long.
Andrew S
LikeLike
An update to my comment on Route 602 . . . a new timetable has been registered to commence (under short notice) on 4 February 2024. There’s no updated timetable on Intalink; on bustimes, or on the Uno sites. One wonders if this next timetable will be more sensible . . .
LikeLike
I was going to watch football at Eton a long time ago using Route 8. The driver fell asleep at the wheel waiting for the level crossing gates to go up on the B388. The gates went up, traffic in front of us moved off yet we remained stationary! I was sitting front nearside. It was my duty to rouse the driver from his slumbers!
LikeLike
York Pullman Bus Company SOLD to First Bus
LikeLike
A possible subject for a future Verify – there’s a youtube video ‘The London bus that won’t stop’ (the no 18)
Very best – Rick Townend
LikeLike
Link for the Route 18 BBC feed
LikeLike
Two points of interest regarding customer satifaction and being “on time”
First in its latest Annual Report to Shareholders have changed the Key Performance Indicator (KPI) measure of performance for buses from Customer Satisfaction to percentage of timetabled mileages completed (I presume regardless of how late they are). For trains the KPI is percentage of Journeys “on time” at final detination. (ref Page 36 of First’s Annual Report 2023).
Whilst traffic commisioners has a “on time” metric of 1 minute early to 5 minutes late the DfT defines “on time” as 1 minute early to 5 minutes and 59 seconds. DfT staement copied here “Non-frequent services (fewer than 6 buses per hour) are measured by whether the bus departs within its
‘on-time’ window of 1 minute 0 seconds early to 5 minutes 59 seconds late. Buses that fail to run should be treated as ‘late’ and not ignored in the calculations.
LikeLike