Business Analytics Blog

Business Analytics Blog

Opinions expressed on this blog reflect the writer’s views and not the position of the Capgemini Group

Would standing on the left get you through Tube stations quicker? by Shivam Desai and Lukas Dobrovsky

Category : Figure It Out

An incredible 1.305 billion passengers use the London Underground on an annual basis. Not only does Transport for London (TfL) have the inconceivable job of moving these passengers quickly around the 402 km network, they need to get these passengers around the various stations as quickly as possible.

That is why TfL are trying an experiment at Holborn Station: they are making everyone stand on both sides of the escalator.

What is the issue?

So, how would this improve the customer journey? Well, it’s all about wasted space and queues.

On an escalator, commuters stand on the right and those who want to walk, on the left. Those who stand on the right-hand side tend to use every second step, whilst those who walk are likely to take up three steps per person, which wastes space.

Moreover, the majority of travellers prefer to stand, particularly in stations with long escalators, which further decreases the utilisation of the left hand side.

Green Park station on a Monday morning: 11th April 2016

So by making passengers stand on both sides the available space should be used more efficiently.

The model

The blogging team decided to build a simulation model to mimic the behaviour of commuters on an escalator on the London underground during the rush hour period on a Monday morning.

These types of models are built on computers to mimic real life activities. Even 3D models are possible! The method has many useful applications in the business environment such as improving passenger flows through airports and the sizing of a call centre.

The blogging team made some observations and timed themselves walking and standing on an escalator over a number of days at Green Park station to add realistic data inputs into the model. The first job was to try to replicate the as-is behaviour. Here’s what we found:

  • Walking up the escalator took 26 seconds
  • Standing on the escalator took 40 seconds
  • 40% of people prefer to walk (Guardian)
  • Fewer people can fit on the ‘walking side’ of the escalator – around 70% as many as those standing. This is because a ‘walker’ occupies more steps than a ‘stander’.

What we didn’t measure, which is what have simulated, is how long it took us to queue just to get on the escalator. It is at this stage where TfL believes the time will be saved. In the model, we assumed that trains arrive every 2 minutes on each of the two platforms and that that about 65 people get off each train, which leads to roughly 4,000 persons who use the escalator in one hour.

It is important to note that we built variance into the model to mimic reality as much as possible. For example people walk at different speeds. Some people face obstacles whilst walking up the escalator. For example, those who stand on the right sometimes have their large bags blocking people, or the annoying tourists that stand on the left.

The results

To investigate the impact of the new policy, we ran our simulation for two different scenarios.

  • Scenario 1 assumes the current situation in which 40% walk and 60% stand on the escalator
  • Scenario 2 assumes every person stands on the escalator

So the results show that, on average, commuters will take 59 seconds to get to the top of the escalator after the change.

For those who would be standing anyway; they would spend an average of 79 seconds less queuing for and travelling on the escalator.  This will benefit 60% of all passengers.

Figure 1: Average time to the top of the escalator

So what about the 40% who like to walk up the escalator, including our bloggers Lukas and Shiv? We would end up spending an extra 13 seconds in the system; which works out to be a 29% increase.

Another benefit is that average queue lengths would drop by enforcing the standing policy, as illustrated below. This means less time spent at the bottom of the escalator, and more people can flow through the station.

Figure 2: Average queue length

So, the majority of commuters would benefit from the proposed changes. However, the benefit comes at the expense of those in a rush.  There goes our plan of a 30 second lie-in.

Check out previous blogs on the London Underground or simulation that Shivam has previously contributed to.

The NHS missed their waiting time targets – an accident or an emergency? (Link)

It could be you, but it probably won’t be (Link)

The price of misery (Link)

About the author

Shivam Desai
Shivam Desai
Shivam is a Senior Consultant within the Customer Experience & Analytics team at Capgemini Consulting. He has over ten years of experience in analytics. His experience covers a wide range of clients across different sectors and he specialises in both Operational Research and Data Science, enabling businesses to make data-driven decisions to give them a competitive edge in the market place.
20 Comments Leave a comment
I use the left as part of my fitness drive...I'll burn less calories and get frustrated. :) Nice blog guys. CP
Thanks Colin and I think a lot of people will be frustrated. It will be interesting to see the results.
I think this analysis suffers from two important flaws: 1/ The 40% stat for would-be walkers is not as it seems as reported in the Guardian. The way it is reported makes it seem that it is a observed data point and a contemporary one. But it is not. It is taken from a 2002 study commissioned by TFL from LSE. It out of date by over a decade and is not taken from observed data but a focus group type survey. You will find the study on line. 2/ Because of the misleading 40% stat, your modelling misses out an important third scenario: people asked to WALK on BOTH sides of one or more escalators. Those who could not walk (or did not want to walk) could use the third escalator as normal. What would your stats on though put be if people walked on both sides? My guess is that you would get both greater efficiency in terms of through put (TFL happy), time saved (commuters happy) and better public health (commuters, TFL and NHS happy). I also think that in the age of the Fitbit and the cash-rich-time-poor consumer the message "everyone walk" is a whole lot easier to sell than "everyone stand still".
Thanks for the suggestions. The actual percentage varies according to different studies and is dependent on the height of the escalator (the higher the escalator, the more likely people are to stand). The majority of people like to stand, hence the queues. TFL are not trialling a walk on both sides escalator policy. Going through Green Park this week, it seemed a lot more than 60% of people were standing. The queues were much better longer than usual. The walking only policy would only work if you could get the majority of people wanting to walk which just isn’t the case.
I am a writer for Inside Science News Service at the American Institute for Physics and we'd love to do a story on the tube experiment. Could you email me a time when we can Skype Monday or Tuesday. I'm in Maryland, the east coast US.
That sounds great, please email me at and we can set something up.
If two lines of people are getting on the escalator every other stem and both lines are sitting still, then the number of people getting off the escalator is one person/step time. If the people on the left decide to move, it doesn't effect how fast people get on. Both columns still climb on every other step. People in line waiting for the escalator will not notice whether the people of the left move or not - just so long as they climb on every other step. So the guy on the left climbs on at the same time as the guy on the right. during the next step-time, no one gets on but the guy on the left moves ahead one/half step. The second step time passes the guy on the left takes an additional half step and two guys board - one on the left and one on the right. the guy on the right sees one empty step between him and the guy in front of him (and guy on his left. The guy on the left sees 2 empty steps in front a a guy on his right. They continue to board at the same rate and therefore depart at the same rate. If the guys on the left walk faster, it gives even more empty steps on the left, but the boarding and exiting rate must remain the same.
sdesai's picture
This is not what happens in reality, you just need to go and see what happens on the London Underground on a daily basis. A key point here is not even between walkers and people who stand.
The fact that walkers occupy more steps than standers doesn't seem to be relevant considering the throughput of walkers is higher than standers. I just read a NY Times article which cited that as the reason for the improved results. The main reason, though, seems to be simply because there are a lot more standers than walkers.
sdesai's picture
There are many dynamics at play here. Agreed, that more standers than walkers is key to the problem. That ratio varies escalator to escalator. Capacity is also very important which is dependent on the steps occupied by a single person.
Do the standing people and walking people have different queues? If the time taken to get to the entry point of the escalator is the same, and the difference between times spent on the escalator is 14 seconds (26 vs 40), why is the overall time spent in the system about 90 seconds more for the standing people? Something doesn't seem right here. Variance should not have that big an effect.
sdesai's picture
There no official separate queue but that is how it works out. Standers are in the system for longer as there are more of them, so the queue for them to board the escalator is longer. That is where most of the time is spent for them.
The title should be "Would standing on the left get you through Tube stations quicker DURING PEAK HOURS?". If the line to take the escalator is short, standing on it would take longer with no benefit. And since we are asking people to change their behavior, asking them to not leave steps between each others would also save time.
sdesai's picture
It be interesting to see how the latter point could be enforced :)
Why is there no mention of the possibility of the "collapse" of the escalator if there are enough people standing still, most especially, if there is a power failure? Should such an event occur, all those people standing still would SLIDE DOWN to the bottom crushing those who have just stepped on. The new Q subway stations in NYC have LONG escalators (and elevators) with no matching STAIRWELL, should there ever be a power failure!
sdesai's picture
The analysis is based on a typical day. Factors such a power failure can be brought in if required.
How about everybody walking?
sdesai's picture
We might publish a follow up blog this week and include that.
I suggest that another flaw is the assumption that the only metric to optimise is the time taken. I find walking more satisfying, and hope it keeps me fitter. Throughput might also matter. I also think you should state how the various strategies perform depending on the volume of traffic: it could be reasonable to maximise throughput by all standing at especially busy times.
sdesai's picture
Throughput was is observed in the simulation model but given this is a blog not all outputs have been shown. TFL also found throughput did increase in their pilot. We only looked at rush hour when this is going to be most beneficial.

Leave a comment

Your email address will not be published. Required fields are marked *.