A bus is not a train.

Very often, I get suggestions for routes that cover a certain segment of the NUS campus. We all know that the stretch between Kent Ridge MRT and Central Library is chronically overcrowded, while the bus empties off after that. So people have proposed a short trip that simply goes back and forth between these two places.

That’s all good, I say. Sounds like a great idea. If it were possible, it would solve all the problems.

But where will those buses park? And where will they turn around?

These issues are often neglected by those who are unaware of the difficulties faced by transport planners. I don’t expect people to understand immediately the constraints that go into planning a bus route anywhere, but the more I receive suggestions that sound very reasonable to a layman but entirely impossible to anyone actually aware of the limitations present, the more I want to write them all down properly so I don’t have to repeatedly explain them.

I believe in this regard, people have been misled by how the MRT works.

With a driverless train, especially in a technologically advanced system like Singapore, it can just proceed to the end of the line, head off to a siding (side track), go to the other direction’s track and reverse back the way it came. And do that ten times over.

A bus, however, is something entirely different. A bus differs from a train in two main ways, one of which is covered here and one of which will be covered by another article.

Planning any bus route, whether in NUS or otherwise, is constrained by the availability and location of end points and turnaround points. Or as we call them, layovers and looping points.


First of all, a bus has a driver. This means that after the route is done, this driver needs to take a break which is called a layover.

The words “terminal” and “layover” are probably very familiar to anyone who’s taken a long haul plane trip. But they have very different meanings in bus terminology.

Whenever any of our drivers finish their trip, they wait at the start/end point of the route for a few minutes, giving them time to take a toilet break or have some water, following which they must tap their card on a machine (on schedule!) to register the start of a new trip.

A1 and A2, for example, will thus pause at Kent Ridge Terminal to allow the drivers to log their trips after every round, instead of going in continuous circles. Similarly, D1 has a layover at Carpark 11 between trips while D2 and K have a layover at PGP.

E might look like an exception, since the buses will go in circles. However, there is still a start and end point. Look at the E route: it starts and ends at UTown. What actually happens is the trip has its first stop occur there, goes for a couple rounds and then terminates (goes off service) after it reaches UTown for the last time.

Thus, a bus route cannot go on forever. You could, I guess, have the bus itself travel in a loop forever, by quickly swapping drivers whenever it reaches the termination/layover point.

That’s called jump-bus, but that doesn’t remove the need for a terminal where the drivers can formally end their route. Anyway, to keep to schedule, the bus has to park somewhere if it finishes its trip too early: this is done at the layover.

A circular route also doesn’t solve the need for a location along the route where the bus can break off for refuelling: a role which is, again, played by the terminal since it is a natural break in the route. If an A1 wants to refuel, it can do so after dropping everyone off at Kent Ridge Terminal; if the route circled forever, the driver would have to awkwardly order everyone off the bus like what is done at NTU. Better to systemise it, by having an end of the route even if it is only on paper (this is done with feeder bus services that end with a T).

Also, in NUS, our demand patterns dictate that it is better to always have trips depart properly from a place at a certain timing. By “breaking” A1 and A2 at Kent Ridge Terminal, we can control exactly when the bus starts from there, instead of it always depending on traffic conditions.

The implication is that in the NUS system, all our bus routes need to incorporate a bus terminal as part of their route, so that drivers have a layover point.

The above image shows the three “terminals” in NUS Kent Ridge Campus that serve as layover points: Kent Ridge Bus Terminal (starting point of A1 and A2), Carpark 11 (starting point of D1 and E) and PGP (starting point of D2 and K). The fourth “terminal” is the Oei Tiong Ham Building in Bukit Timah Campus (starting point of BTC and L).


Secondly, unlike a train which is double-ended and can be used facing either direction, a bus obviously has only one driver’s cabin and doors on one side. This means that when it reaches the end of a route, if it wants to double back and head the way it came (instead of heading to a termination point), it will have to turn itself around.

Our buses are twelve metres long. They cannot just use any U-turn they see to turn around. Neither can they do a 3-point turn.

Buses can only U-turn at specific wide areas, such as roundabouts, or a carpark/driveway big enough to allow a bus to enter and then exit it.

Thus, bus route planning is limited by the constraint of available looping points and how near they are to the first/last stop (to avoid wasting fuel and time).

For the NUS network, our looping points will mainly consist of roundabouts in the Kent Ridge Campus. The advantage of a roundabout is that you can use it as a looping point while coming from pretty much anywhere.

Thankfully, we have quite a few of them, but notice their positioning.

The six existing looping points in KRC are at UTown, the small roundabout in front of UTown/Museum/Raffles Hall, the roundabout near Yusof Ishak House, COM2 (Carpark 13) and two small roundabouts at FoS and NUH.

Service B formerly used COM2 as a looping point, as it would travel from AS5 into COM2 and then back out towards Ventus. It was later split into B1 and B2, which bypasssed COM2.

UT-CLB or UT-CCE used to loop at a very wide U-turn point outside Central Library. The wide U-turn still exists, but buses can no longer turn from CLB to IT because the CLB stop is now after the U-turn.

Service C formerly used the FoS roundabout as a looping point to immediately turn back to LT27 after it dropped passengers off at S17.

Services D1 and D2 use UTown as a looping point, since they enter and exit UTown via a link bridge. When it was first introduced, D looped at the FoS roundabout, continuing to do so after it was renumbered D1 and D2. This only stopped when D was actually split into two different routes in 2013, but that’s a story for another post.


Putting together the two constraints, we can see that bus route planning is not so simple after all.

A bus company will want to minimise what we call dead mileage — distance travelled by buses with no passengers (Off Service). A bus will experience dead mileage when travelling from the parking area to the first stop of the route, or from the daytime parking area to the depot. More dead mileage translates to more fuel wasted and time wasted.

Now if the route is a circle, it needs to have a stop along the circle that is considered the terminal for layovers.

If the route is a linear route between two points, the ends have to be either terminals or looping points. And there has to be a layover point somewhere, either way.

And to minimise dead mileage, the layover points and looping points need to be as near to the first or last stop of the route as possible.

Now we simultaneously look at the layover and looping points that we have in NUS.

Notice that all our three layover points are in the southern half of the campus, while there is a somewhat better spread in terms of looping points. (Note also that layover points themselves can and do function as looping points.)

It is thus immediately apparent that a bus route cannot run exclusively in the northern half of campus (eg MRT to CLB), despite that section being the most overcrowded.

Previously, in 2017, C would start from KRBT and head to Computer Centre, thus serving the northern half of the campus until FoS. However, as seen in this poster from the past, OCA explains why route C cannot loop at Kent Ridge MRT. After that stop, there is no looping point available until PGP!

If we do implement such a service, it will have to travel between two of the looping points, and after a certain number of trips it will have to terminate somewhere so that the bus driver can have his break. Such an arrangement was done historically with the direct services, but the direct services proved inefficient so we did not revisit them.

As mentioned, however, it is still carried out on A1E and A2E. So perhaps there is scope to look at introducing it. Probably not on full-day services, though.

Alternatively, we could have a service run through the northern half from CLB to MRT, but layover at KRBT and PGP. This, however, would again not be desirable because of dead mileage between KRBT and CLB, and between MRT to PGP.

In 2021, we did introduce a service that served the northern half of campus: Service K. This service did layovers at PGP, and looped around the entire Engineering area via CLB, SDE, Kent Vale and Museum. However, the long journey time meant that it had to be operated at a 40-minute interval.

At this point it should now be clear why solutions to connectivity are never straightforward.

3 thoughts on “ISB1101: A bus is not a train

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s