A Naive take on the Impact of Recent MRT Disruptions

by Yin Shanyang of Swarm

20/12/2011

Over the past week, Singapore’s subway system, the Mass Rapid Transport system ( MRT ), faced some of its worst disruptions in its 24 years of operation. The worst of which was a disruption of North-bound trains during after work peak hours, which lasted for about 5 hours. This disruption purportedly affected about 127,000 commuters.

Curious to place the impact of the disruption onto scale, I collected some data and created a few maps to visualise the scope of the impact.

Public Transport on a Normal Day

Let’s look at how Singapore operates on a regular day as a starting point for comparison.

Knowing that Google Transit contains public transport data for Singapore, I created the following map based on scheduled travel times at 7pm on 15/12/2011. The map shows the travel times required get to various locations in Singapore if one started her journey from City Hall Station.

Travel times across Singapore, starting from City Hall Station.

Overlaying the visualisation with a map of Singapore ( thanks Gothere.sg ), we can see that islands of faster commute coincides with the rail lines and major road systems. ( e.g. Bukit Timah Road. )

Visualisation overlaid on map. Notice how islands form along main transportation arteries.

Public Transport after the Disruption

Travel times after the disruption.

This is the same visualisation with the disruptions in place. Alternate routes are taken into consideration. As expected, it is mainly areas in the north and north-east that are affected.

For a clearer picture, I have isolated the areas affected and scaled them according the additional travel time incurred. General delays ranged between 5 and 30 minutes.

Areas affected, with additional travel times ranging from 5 – 30 minutes.

Best Case Scenario ( or being Naive )

Based on data from Google Transit and comparing alternate routes, the results I got in delays were a far cry from on-the-ground anecdotal evidence. 5 – 30 minutes of delays seem too optimistic. But to approach this, I have to lay forth my assumptions.

Assumptions
We start our journey from City Hall Station, which I think makes a good ‘average’ starting point for most communters’ journey home. Based on this approach, and available data from Google, I am assuming that all commuters have perfect knowledge of alternate routes home.

Other Factors, Unaccounted
Imperfect knowledge of alternate routes, as stated above. Disruptions or delays to other routes due to this disruption.

And hence I describe this approach as a naive take on the situation, though, I think it is a starting point to understanding the impact of such a disruption. Lest we can start to understand that even a partial breakdown in our transportation infrastructure can lead to a wide scope of impact.


The Approach Forward
To fully consider the impact of the disruptions, data collected from EZLink cards can be used to analyse the full impact of the delays on the ground, by comparing the average travel time between destinations against regular travel times. And from there, we can interpolate and infer the delays caused by, the lack of knowledge of alternate routes, the delays faced by extended waiting times, and other delays like traffic conditions.

And to take it a step further, the same data, collected in real-time, cross referenced with traffic conditions, historical commuter data and combined public transport schedules could also mean for a more adaptive system in handling unexpected incidents. LTA is partnering IBM after all.





The two of us at Swarm are looking at HDB’s resale data set. If you’re interested on our findings, follow me at @yinshanyang for updates.