santiago: a low-tech approach to fast exits from a subway station

So you're on a crowded subway train on Santiago's Line 4, the dark blue line on this map.  You're northbound, approaching the end of the line at Tobalaba station.  


Everyone on the crowded train will get off at once.  Most customers are changing to an intersecting line 1, which has  side platforms on the level above.  That, means you can't exit the platform at just any stairwell; each of the two stairwells goes to just one direction of the connecting line.

So customers tend to collide as they exit the train trying to get to the correct stairwell for their preferred direction, creating massive platform congestion that slows people's exit from the train.  this increases the dwell times of the train and thus reduces the possible frequency, which in turn only makes the trains even more crowded.

Massive infrastructure solutions were proposed.  My friend Juan Carlos Muñoz, a professor of engineering at the Catholic University of Chile, came up with a simpler solution (Spanish with English subtitles):


A gate blocking the platform halfway along it forces people to exit at the door nearest to them, which in turn teaches people to be in the correct part of the train for their preferred connection. People who try to exit the wrong exit are stopped at the staffed gate, and let through last only after the crowd has cleared. These people are irritated, and a few write to their elected officials, but most people just learn how it works, and work with it.  

UPDATED: Shouldn't people have figured out anyway what part of the train to be in to be close to their exit?  No, becuase in this case, there's an exit at the front end of the platform and another in the middle.  Juan-Carlos explains:

There is one set of stairs coinciding with the middle of the train. Let´s call them A.

Only 40% of the passengers in this train wants to take these stairs.

Thus if we were to assign every passenger a position inside the train we would put all these passengers at the back half of the train. Then the front half of the train would be full of passengers taking the stairs at the front end of the station (stairs B).

However, a great place inside the train to take stairs A is in the back of the front half of the train. Indeed every train used to have around 120 (out of a total of around 1500) such passengers taking such a strategic position. You can see them in the video! These are the passengers causing the problem, not only because they cause the counterflow but because they force some passengers wanting to take the B stairs to enter the back half of the train. The gate forces to act otherwise leaving some room for more B passengers into the front half. They can now exit the station much faster.

So this was a "tragedy of the commons" problem.  People optimizing for their own outcomes were in conflict with the most efficient way to get everyone out of the station before the next train arrived.  

Note how Juan-Carlos refers to the "greatest good." The implication is that we can't let a few people's anger get in the way of solving the problem in a cost-effective way.

12 Responses to santiago: a low-tech approach to fast exits from a subway station

  1. Jesse-the-k December 21, 2013 at 8:59 pm #

    The low-tech effectiveness is indeed gratifying. However, I want to challenge your frame that the problem and the solution is for especially inattentive people, who need to be forced to think through the consequences of their position on the train, and take responsibility for those consequences.
    It’s unhelpful at best (and hostile at worst) to claim that first-time riders, visitors, tourists, people with vision, mobility, reading, or cognitive impairments need to take more responsibility for non-apparent systems.
    Teaching/instructional material available outside the press of a subway car would be the first step.

  2. D December 21, 2013 at 10:00 pm #

    There was a similar problem and solution in Toronto, at Bloor station on the Yonge line. This is a major transfer point between the east-west crosstown subway line and the Yonge subway heading downtown. Yonge trains are already quite full when they reach Bloor, and then passengers from the Bloor-Danforth line have to try and jam themselves in.
    The problem was that the stairs and escalators between the two lines are situated at the far north end of the platform. As a result, passenger distribution along the platform was very uneven (passengers coming up the stairs to the Yonge line would tend to stay in the north third or half of the platform). Dwell times increased both because it takes longer to serve passengers when they are not using the doors evenly, but also because the rear two cars were packed like sardines. With headways in the order of 2-1/2 minutes or less, any excess dwell time has a major impact on reliability and train throughput.
    In 2009, the TTC discovered that they could relocate the transfer point away from the north end by erecting temporary barricades directing passengers about halfway down the platform before being permitted into the trackside half of the platform. This evened out platform loading and reduced dwell times. It also helps separate flows of boarding and alighting passengers. The barriers are in place during the AM peak period when platform congestion is worst. The only real drawback from a passenger perspective is that it adds a little more walking distance/time.
    More information and photos:
    Steve Munro’s web site:
    The Torontoist:

  3. Juan Carlos Munoz December 22, 2013 at 8:25 am #

    Thanks a lot for posting my video Jarrett. However I have one disagreement with your main argument. The door was needed because expert daily commuters would take a door in the train that was great for them but bad for the group of travelers. And this is why a communicational campaign inducing passengers to change their door badly failed not making any behavioral change at all. On the ears of those travelers the campaign would hear like: “we invite you to take a door that will take you longer to exit from the station once you reach Tobalaba”… So why would anyone adhere to such an invitation. I am thinking here on the passengers that wanted to exit through the stairs in the middle of the station taking the back doors of the front half of the train. These were the passengers that caused the counterflow and needed to be shifted to the back half of the train. They were daily extremely expert commuters so they would never do that if they were not forced! And by staying in the front half they caused some passengers that would benefit from being there to stand in the back half instead.
    The gate forced some of them to do it and that made everyone’s situation much better.
    The beauty was to intervene in a very decisive and delicate way, almost chirurgical so that the situation improved with a minimal change.
    We should bear in mind that in a congested system travelers are causing delays to other travelers that they do not perceive. In some cases if these travelers perceived those delays as if they were their owns, they would change their decisions. Thus, very often we have opportunities to improve the system if we force them to act differently or if we make them internalize the cost they cause.

  4. Joe Busman December 23, 2013 at 9:07 am #

    In America Munoz would be fired, because in America, his solution deprives the agency of using more money to build massive infrastructure which is usually federal money here. You’d never find a simple solution like that in America. The solution that costs the most is almost always the better solution.

  5. Dave December 23, 2013 at 10:39 am #

    Actually Joe, since the federal approval process for new infrastructure takes years if not decades, the transit provider would still need an intermediate solution as it waits for the massive infrastructure solution to work its way through FTA.
    I think the more likely reason this wouldn’t work in America is the cowardice of politicians and the availability of weapons here. I see all those people nicely listening to the woman guarding the gate and think: in America, at best, the pissed off people would complain to the authorities who would end the project immediately… and at worse, a single pissed off person would simply flash his/her concealed weapon once, and nobody would ever want to staff that gate again.

  6. Simon Vallée December 23, 2013 at 1:07 pm #

    Ah, yes. The bane of free-market economics: cases where a rational agent’s optimal course of action is suboptimal for the group as a whole because of externalities. Also known as the “prisoner’s dilemma” or Braess’s paradox. I also call it a “cheater’s game”: a game where any individual is better off by cheating the rules, but where everyone is better off if everyone follows the rules, so the collectivity has a rational reason to enforce the rules and the individuals have a rational reason to work around the rules.
    The more authoritarian models of municipal governments in Latin America do help a lot when comes the time to implement such rules. Doing something like that in the US or Canada would effectively be fraught with peril, because of the main problem with local democracy: most people are indifferent and don’t care, but the only ones who actually care a lot are those opposed to the decision, so the only voices democratic representatives hear are people who keep yelling “NO!NO!NO!”.
    It also explains the success of NIMBYism: NIMBYs tend to be a minority, a vocal minority, mind you, but a minority still. The problem is that, most of the time, the majority is indifferent, those who “support” a project are more likely to be “yeah, it would be nice if it came to be”, whereas those opposed go “if that project goes through, I will die in agony!!!” and to campaign against it like their immortal soul was at risk. So if you end up in a referendum, you’ll get 75+% NO votes in general… but with a 20% participation rate.

  7. Wai Yip Tung December 23, 2013 at 1:49 pm #

    “we can’t let a few people’s anger get in the way of solving the problem in a cost-effective way”. Kudos to Santiago for making it happen. 5% drop in travel time for the whole line is huge! All this done for minimal cost with no construction. Transit managers and political leaders should all learn from this and make the right call.

  8. Jarrett December 26, 2013 at 1:54 pm #

    David Vartanoff submitted this comment:
    While the solution applied seems to have worked, it points up previous design decisions which caused the problem. The video shows that the car fleet apparently is of isolated married pairs so that if a rider boards at a given point in the train, s/he cannot walk through to a better spot. (Why I dislike the NYC/Chicago prohibition of walking through the train which was not the case several decades ago when I lived in those places.)
    We also should note that the trains in the video all had significant wasted space with both curved fronts and too many cabs per trainset. A more New York like fleet with cabs only at front and rear of a 4 or 5 car train would carry more passengers in the same platform length.
    Though you write ” Over time, of course, everyone but the occasional rider or tourist figures it out.”, in my observation, many riders do not seem to figure it out as for instance BART riders who are still seated 4 rows from the door as it opens for their stop. The overcrowding on the insufficient stairways mirrors the Yellow/Red Line cluster#%&* on DC Metro where the platforms are poorly aligned leading to crowding. Imagine how the Red/Yellow transfer would be different if the Yellow had side platforms–there would be only half as many riders using each set of stairs/escalators.

    End of David Vartanoff comment

  9. anonymouse December 27, 2013 at 6:01 pm #

    Of course the orignal problem with the whole design is that the circulation area on the platform isn’t quite big enough, and the stairs themselves block some of it and form a chokepoint, and the door is a clever workaround that keeps the flow of passengers smooth. Which all just goes to show the importance of modelling things like passenger circulation, because poor passenger circulation can lead to longer dwell times, and dwell time is ultimately one of the things that determines total line capacity.
    As for BART riders, their habits are just completely baffling to me. They actually line up in lines to wait for the trains, standing directly in front of where the doors of the train will be and stretching back across the platform, which is a pretty inefficient use of platform space given that the trains only have two doors per side of 75-foot car. And BART seems to not really be doing anything to manage dwell times despite dwell time at Embarcadero being the main constraint on the number of trains they can put through the Transbay Tube.

  10. happy wheels February 19, 2018 at 12:15 am #

    Thanks for your post! It’s been a long time since I read a good article and such a meaning! I hope you will continue to write articles like these for hobbyists!

  11. Jimmy May 17, 2019 at 10:58 am #

    It is such a beautiful post. Your article is very useful for me. You point out many points that no one follow and just ignore them. This really a big problem that we face in our life. I appreciate your work. Roads are made for the help of people that they can easily reach their destination but they take it non serious and don’t follow the rules. That is very bad so follow the rules for you and for other people safety.

  12. carry January 8, 2020 at 5:42 am #

    It can be most broadly defined as the entities, both material and immaterial, created by the application of mental and physical effort in order to achieve some value. Printer Repair helps you to solve your all printer issues quickly and at affordable company london