Crowding data in light of Coronavirus


in relation to the Crowding data available at,

I have several questions:

  • the most recent data set is for 2018 - does the data set represent all of 2018, or just November 2018?
  • when will the 2019 data set be published?
  • are there any plans to generate this data set on a more frequent basis, i.e. monthly or weekly?
  • for example, the recent Coronavirus has presumably made most of this data not as accurate; a monthly or weekly dataset would be much more reliable to use. are there any plans for developing something like that, perhaps also some sort of (near) real time crowding API, like the other existing APIs for e.g. Tube?


1 Like


I’ve been trying to work out how you might go about working what the right new figures would be. So far, I think that I can see that:

  • The space needed per person is the circle around them (πr2), so that’s 6π (12.6m2)
  • The “new” capacity of a train would be worked out from the floor space divided by 12.6.
  • So a Jubilee Line train (17.77m x 2.630m x 7 = 327m2 area) could support 26 passengers in total. This is one passenger sat in every 9 seats.
  • Escalators capacity would be reduced. At Angel the 61m escalator could carry only 30 passengers at a time (each).
  • Trains that have passenger operated doors (London Overground) will need to change to opening automatically to remove the need for a common touch point on the trains.
  • It might be possible to use the extra trains used to boost services during peak for longer hours to provide more trains to travel on as capacity is so reduced.
  • “Dwell Times” should be a thing of the past
  • The stations with only lifts which are closed will not be able to open again.
  • It might be necessary to alter the software on the entry gates to a LU/LO station to restrict the passenger flow. If the station can only deal with 30 passengers a minute then the gateline software should introduce delays to slow down entry to the station.
  • Passengers should stand in the middle of the escalators to help with horizontal spacing.

And my London Underground bête noire need sorting - ** All stations should be made KEEP-LEFT** to prevent people bumping into each other. For example Seven Sisters (one end), Highbury & Islington, North Greenwich are oddly “keep right”.

1 Like


Anna Leach and Pablo Gutiérrez have had a stab at some of this on TheGuardian.

It’s even harder to work out what the figures might be for an interchange station like Stratford, as there are many possible interchanges as well as entrance/exits.

This is an interesting train of thought, pun kind of intended.

I’m confused by the maths, particularly around the number of people who can fit in a Jubilee line train, and I wonder if you’ve made some assumptions that I haven’t or vice versa? My own maths abilities are absolutely appalling so it’s entirely possible I’ve missed something very obvious.

For example, if a Jubilee line train has 7 cars long and holds 26 passengers total, this means fewer than 4 people per car. Each car is slightly in excess of 2 metres wide and so I guess it’s touch and go whether people could stand on opposite sides of the car. With a person in each corner of each car as a minimums , this gives a capacity of 28 for the train but my first assumption there is that social distancing doesn’t need to be observed through the walls between the cars, and also that people can cram themselves into the corners enough to maintain the 2m gap.

If standing on opposite sides of the car works, and each car is around 17m long, I reckon you could get 8 people on each side of the car for a total of 16 per car, or around 124 per train (without factoring in cabs).

I think the difference comes down to how we’re spacing people out. I think the figure of 12.56m^2 per person double counts the space between people, and you end up with people placed 4m apart rather than 2? If you assume each has a 1m radius around them, and you place two 1m radius circles with their circumferences touching, the distance between the two centre points is 2m. If you then divide the 327m^2 train by 3.14m^2 you get almost 34 people per train with the caveat that dividing a rectangular space into circular areas wastes some space because circles don’t tesselate well. You can’t fit people into those spaces so assuming then that the effective area per person is 4m^2, i.e. a square with sides of 2m per person, you get 327/16 people per train which is about 20. This, of course, assumes that people don’t stand against the walls of the cars.

So if you then assume that people stand directly on either side of the cars and you place 4m^2 squares between each, you effectively have a car with 4.630m width, around 82m^2 area per car, 573m^2 area per train or 35 people per train. Assuming 20tph it’s around 180 passengers per hour difference.

Out of interest I then imagined a Jubilee Line train to be more like an escalator. If you’re dealing with straight lines of people like on an escalator, the train has 17.77*7 metres of length, and dividing by 2 gives 62ish people per train.

Have I got totally lost here? I’m genuinely intrigued by the figures, certainly provided some food for thought with the insomnia!



Of course, I’ve made an error in the calculation around them, it’s only 1 meter per person as you say which reduces the area to 3.14m2) which would make the Jubilee Line train carry four times more passengers, 104 passengers.

Another issue is that quite how you get these passengers to socially distance when waiting for a train as there is only room for 62 of the to wait on the platform.

here is that social distancing doesn’t need to be observed through the walls between the cars,

This could be a false assumption. There is no active guidance that say either way. Also quite a few trains are now walk-though (Met/Circle/H&C/District and London Overground).

Have I got totally lost here? I’m genuinely intrigued by the figures, certainly provided some food for thought with the insomnia!

Indeed. Your assumptions and calculations seems very reasonable to me. Of course the problem here is that I can’t see TfL surviving - or London too - with such measures in place indefinitely.

Even if the whole tube system was upgraded to Moving Block signalling, driver-less trains and then double the number of trains (which would need to double the number of train stabling) and operated 24/7 the capacity drop here would still be unable to cope with the demand seen in 2019.

Then there are three more issues.

  • You can’t enforce the 2 meters thing without on-train CCTV and there are several lines without this;

  • The seat layout on the Bakerloo (and Met) lines isn’t the inward-facing “Overground” layout but the old groups of four the face/back the direction of travel.

  • Lift-only stations have been closed because of social distancing, but how do disabled passengers travel with the lifts out of bounds?

OK, I’ve done an actual diagram to illustrate the problem in the real-world. Here’s the Jubliee line train again, and when you try to cram in the 1 meter bubbles (for 2 meters seperation) you actually only end up with 15 people per car.

1 Like


You seem to be allowing passengers 1 metre separation from the carriage walls. Shifting the centres half a metre a more might make quite a lot of difference. In particular all four corners of the carriage might be usable. On older trains I suppose passengers on opposite sides of the glass (?) partitions (one sitting, one standing) might be OK too.

In practice proper separation doesn’t happen now so it won’t happen when more are back at work. It’s a nice thought experiment though.


Following the topic of the original post, rather than the thread the developed, TfL has real-time crowding data, based on anonymised WiFi signals. They can literally count the smartphones moving around the network that have WiFi switched on, which is almost the same thing as counting the people - certainly close enough for crowding purposes.

They have said they intend to make this data available. No idea when.

I assume they’re rather busy right now, but if there were a way to get it, I’ve been working on an app that’s designed to help people avoid crowding. The 2018 data is obviously useless at the moment, but if the real-time data were available we could fairly easily integrate it so anyone could see how busy the stations were.


Fair point, that makes the capacity here 30 which is more workable than 15. I’m not sure if you could 100% be sure that people wouldn’t need to come closer when boarding or alighting.

I’ve done the “fitting circles” for each line, as this will give a figure for the “new capacity” values. This time, I’m hoping these are correct!

1 Like

I think you need to allow for the width of the passengers as well, so the circle diameter needs to be about 60cm bigger. Those guys in the corners are looking pretty squashed.

When I was commuting I was in a “bubble” so the diagrams seem very apt. I am still trying to decide which of Brian’s patterns would look best for retiling our bathroom though.

I was under the impression that the 2 meter thing was from the centre of the person, rather than between people. etc

You would have to be very thin indeed for your centroid to be right in the corner!


Of course, this is because the “2 meter thing” is a product of observations from the 1918 Flu Pandemic, rather than a modelling of actual people.

It’s interesting that it’s only “6 feet” in the USA (183 cm) and only 1.5 meters in the Netherlands.

In practice, the area needed is less than that because – to take an extreme case just for explanation – if you had a square enclosure that was about 2.5 metres on each side, you could legally put one person in each corner. So, surprisingly each person would be occupying only 1.6 sqm. Though as I said, that’s an extreme case.

Conveniently, that means we could have one row of passengers down each side of the train and if we temporarily forget about the doors, a 17m carriage would alllow 9 people down each side. But a carriage with no doors wouldn’t be a lot of use, so unless you have an exact seat layout my guess would be 6 per side, 12 per carriage, 84 per train. That’s potentially rather a lot more than the 26 that you calculated. But with a proper carriage plan we could work out an exact number. And maybe squeeze in a few more if some of the doors could be locked out of service to allow a few additional passengers standing.


Thanks for that, I’ve already made corrections to those points above with these images …

Mathmatically, it’s down to Euclidean tilings by convex regular polygons

It is interesting there has been more work done on this:

That’s a much greater capacity and it’s down to the great foresight of TFL’s ancestors settling on a tunnel bore that seems to have been chosen precisely to enable the optimum width for correct social distancing. A larger or smaller bore size would have been much harder to fit with optimum efficiency.

It just begs the question that if TFL’s ancestors had such great foresight, could they not have warned us that covid was coming and given us a hundred years to develop a vaccine?

There is one further piece of advice that TFL needs to adopt to maximise capacity:

"If you are travelling at peak periods with another person that lives at the same address as you, please ensure that the other family member sits on your lap, to avoid disrupting the Optimum Social Distancing layout for other passengers"


@markw Thank you for the reply.

Does anyone else know if there is any update regarding this? Having some sort of semi real time data (or at least a latest “snapshot” of the data from any week of this or past month) in light of the current special service situation would allow developing various services to help passengers use public transport faster and safer when they require to use it.

1 Like


Also, it might be useful for everyone for some data around the idea that “commuters may have to queue outside stations at busy times”

“Due to the national requirement of maintaining 2 metre social distancing, passengers travelling at peak times (between 5.45am and 8.15am and between 4pm and 5.30pm on weekdays) might be expected to queue to enter stations and experience delays to their journeys.”

This an interesting idea, because if the number of people waiting on the platform is managed better, then the HUGE dwell times for the services could be reduced - which would mean the trains could run to the actual timetable!

1 Like