On Autonomous Buses in Edinburgh

[danieldwilliam]

I am more interested in the impact of ground vehicle autonomy on public transport than is strictly speaking good for me, or, if you are reading this, you.

However, I've been mulling over the impact on Edinburgh's bus services of shifting to an autonomous fleet (which will happen any day now - they will be fusion powered).

Lothian buses are the main bus operator in Edinburgh with some other services being offered by another operator. Mostly those other services are routes coming in from the commuter belt not urban routes.

Lothian buses look to have about 60 routes.

Assume average frequency of every 20 minutes. 3 buses an hour. 4 hour round trip on each route is 12 buses per route. 720 buses operating each day. The Lothian buses web site says they have more than 700b buses so this seems like a reasonable set of assumptions in the round.

If you get a 1/3rd cost reduction you can increase the number of buses by 50%. If you assume an additional increase in revenue from ridership increase due to service frequency and additional routes you can add another 50% on top (maybe). So your 720 buses goes up by 150% times 150% to 1,620 (that seems pretty generous, but lets roll with it).

You could double the average frequency on all existing routes. Taking you from buses on average every 20 minutes to every 10 minutes. That's a nicer improvement than it sounds because the increase in service frequency disproportionately improves services on less frequent services. There are already some services at 10 minute frequency and more at 15 minute frequency. You do potentially start to have problems with congestion on the main routes in very centre of town. Lothian Road is already pretty close to being full of buses during rush hour. A model that fits these assumptions sees Lothian buses move from the majority of routes at 15 or 20 minute frequency to the majority at 10 minute frequency with a few at 5 minute frequency.

That is going to cost you another 720 buses - taking you to a total buses in use of 1,440 . That leaves you with about 180 buses spare to put on to new routes. With 6 buses per hour and a 4 hour round trip each new route would take 24 buses. So you get 7 maybe 8 new bus routes at a 10 minute frequency or 11 new routes at a 15 minute frequency.

Those buses could be running in theory 24 hours a day. Would you actually want to run a bus service with 5 minute frequency at 3am? Not sure you would but you certainly don't have a drop off in frequency in the evening and Sunday.

That is a pretty decent improvement in service but not quite enough to move to a fully metropolitan service which I define as one where buses are so frequent that you do not issue a timetable and travellers do not need to optimise their journey by picking alternative routes based on the exact time they are travelling. It's not far off though for small city like Edinburgh, especially if you consider that many journeys will have overlapping routes.

Comments

[armiphlage]

One thing to consider is that the size of a bus can be varied. Where I lived five years ago, my local buses were articulated, and held ~140 passengers (although I really think they squeezed in more at times). I then moved to a small town, where the biggest bus holds 20 passengers.

The articulated buses had frequent mechanical issues where the two bodies coupled, and were inefficient outside of peak hours when most seats were empty. The only advantage over several smaller vehicles was that the articulated bus only needed a single driver.

If you take driver payroll costs out of the equation, smaller vehicles are an attractive option. More maneuverable and able to fit into smaller gaps in traffic (improving routekeeping), easier to maintain, and less prone to cause "knuckles" in the asphalt at transit stops. Smaller vehicles lets you increase frequency without increasing overall capital expenses.

Of course, my region's experiment with autonomous vehicles ended after one of them decided to drive through a tree.

https://blog.metrolinx.com/2021/08/25/self-driving-electric-shuttle-heading-to-whitby-go-as-part-of-av-pilot-project/

[danieldwilliam]

The variation in bus size for sure is important. In Edinburgh we don't have articulated buses but we do have an expanding network of trams. The trams carry up to 250 people each. Those will end up bulk moving people along fairly straight and level corridors.

With a small, single decker passenger bus - say 30 people you could run routes in Edinburgh that you currently can't. Specifically what comes to mind is some routes up and down the Royal Mile and behind the Castle, some routes through the squares in the city centre, and some routes where a right turn is useful but too difficult in a large double decker bus.

If you are no longer optimising to reduce driver labour costs you can have much more choice about bus size and the routes those buses take and how those routes are structured. For example, rather than taking 100 people from the airport to the centre so that 50 of them can go to the north east and 50 to the south east of the city you can run two buses direct.

[andrewducker]

I had a quick look at the buses that go past me (24 and 41) - 2 hour roundtrip for the 24 (ERI to Granton) - 7 buses currently on that route. 2 hour roundtrip for the 41 (King's Buildings to Cammo) - 7 buses also currently on that route.

Furthest bus trip I can see is Seton Sands to Clerwood (the 26) - that's a 3:15 roundtrip. I count 21 buses currently on that route. So clearly a lot of variation!

Overall, I think your assumptions aren't bad. But I do think that actually if buses are every ten minutes then you don't need a timetable. The variation in departure times due to traffic is easily going to be greater than the mean waiting time.

The biggest difference, to me, will be on Sundays, when the 24 and 41 are both half-hourly, and the 41 doesn't start running until 9am (with the 24 being hourly until 9:30). Increasing the number of buses in those circumstances suddenly makes buses a feasible method of being somewhere on a Sunday morning! And reducing the need for cars means making things possible which currently aren't, so that'll be a huge help.

[danieldwilliam]

But I do think that actually if buses are every ten minutes then you don't need a timetable.

Not needing a timetable is one of the defining characteristics of a metropolitan service for me and for exactly the reason you suggest, the natural variability in progress will be often greater than the waiting time. Putting it another way, you would not look at your watch to see if a bus was due. I'm not sure if that point kicks in at 10 minute intervals, which is a 5 minute mean waiting time or at 8 minute or 5 minute frequency. I suspect that might turn on how often people have to change buses to complete their journey. If you generally don't change or only change once then a 5 minute mean wait might be much more acceptable than if you are changing 3 or 4 times.

[andrewducker]

One of the advantages of hub/spoke is that you tend to only change once for most journeys. Certainly I've never had to change twice for a journey in Edinburgh.

But then you have a lot of congestion around the changeover points.

[danieldwilliam]

There's clearly a strategic difference in point-to-point systems and hub-and-spoke (see Boeing vs Airbus). Advantages of hub-and-spoke, economies of scale. Advantages of point-to-point quicker and less faffy journeys.

The airlines have been testing the two strategies after a change in engine design meant smaller planes had the range and the safety margins to travel intercontinental distances and point-to-point seems to be winning.

What I think it not tested in Edinburgh buses is whether there are enough potential journeys by bus that currently do not happen using the current hub-and-spoke method but would happen if you had more point-to-point capacity.

I suspect that two or three circular routes would solve most of the desire for more point-to-point services and still only require one change - particularly in the world of the future where there are a few more bus routes running anyway.

The congestion is a problem. For example, on occasion I've had to wait 5 minutes at the bus stop at the bottom of Lothian Road to get on the 24 because a 10, 11, 15, 16 and 47 we all trying to load an unload at about the same time and getting caught in the traffic lights. Not a huge disaster but indicative of a road junction that is full.

[danieldwilliam]

I appear to have gotten the right bottom line but with at least one pretty wrong assumption. I've assumed bus routes twice as long as they actually are (thanks for checking). So either buses are on average more frequent than I assumed or fewer of the buses are in use at anyone time.

The weekend service is where you get the biggest improvement and I think that's what makes me bullish about ridership. If you could actually get around the city on Sunday on a bus as quickly as you can in your car (allowing for finding a parking space) then it becomes much more useful as service to a big chunk more people, particuarly if you are able to combine with with increased frequency on the suburban routes and more point to point journeys.