Analysis of Route 39 Finch East: Part I — Introduction

When I started the analysis of operations on various routes, I requested data for a number of suburban lines as well as the downtown streetcar routes.  Up until now, I have only published material on 29 Dufferin, but it’s worthwhile getting some of the other data out in view before it is completely stale.

The data in this series come from December 2006, but the conditions on Finch have not changed too much in the interim.  The service is not as complete a mess as we have seen on some of the major streetcar routes, but it is far from perfect.  It’s worthwhile looking at how a frequent bus route operates in conditions that should be generally favourable to transit service.

As I have done before, I will start with Christmas Day 2006 because this shows the basics including the route’s behaviour when there is no excessive passenger demand or traffic congestion.  For contrast, I will also present a weekday Friday, December 1, 2006.

Christmas Day, 2006

December 25 Service Chart
December 25 Headways Westbound
December 25 Headways Eastbound

For those who are unfamiliar with these analyses, I refer you to the first series on the King Car.  Briefly, there are three types of chart here:

  • The Service Chart is based on a standard method of analysing railway operations dating to the late 19th century.  The horizontal axis is time, and the vertical axis is location.  Each line plots the movement of a vehicle back and forth on the route.  The slope of the line is roughly proportional to the vehicle’s speed (see notes below).  The spacing of the lines indicates the relative position of vehicles.  Both directions of operation appear on the same chart, but once you are used to the format, it is relatively easy to concentrate on only one direction’s information.  This layout makes it easy to see bunching of vehicles that persists beyond a terminus and continues into future trips.
  • The Headway charts are, in effect, horizontal slices through the Service charts at specific locations.  Each page of these charts gives a different location on the route and shows the spacing of vehicles at that location.
  • The Link Times charts (not included for December 25 as they are only of interest in spotting traffic congestion) show the times individual vehicles take to get from one point to another on a route.  Where these times are unchanging over the course of the day, vehicle progress is unimpeded by traffic congestion or excessive dwell times at busy stops.

All of this is prone to errors introduced in the data by the TTC’s vehicle monitoring system, CIS.  Two problems affect this route:

  • Near terminals, resolution of vehicle location is imprecise.  Some vehicles may not register their arrival until they are about to leave, while others may appear to leave well before they actually do.  On the large loop at the east end of 39 Finch, the exact location of the vehicle may not be accurately reported.  To correct for this, the entire loop at Neilson is treated as a single location so that all buses are “there” regardless of exactly where within the loop CIS says they are.   Headways at the terminals may not be correctly reported, but near the terminals, they are accurate.
  • CIS uses a combination of odometer readings and electronic signposts to locate a vehicle.  When a bus passes a signpost, the system corrects the vehicle location to match, and this can make a bus seem to move backward in space if the odometer is reading high.  The data analysis routines I have written filter out most of this, but can cause vehicles to appear immobile for the time between the calculated arrival at a signpost and the time it actually reports this.

The Service Chart shows what we would expect to see for a holiday day.  The service is fairly well-behaved running back and forth from Finch Station to Neilson Road.  Even so, some variation in speed and in vehicle spacing is evident.

When we look at the headways, we see a pattern familiar from other routes.  Page 9 of the eastbound chart shows the data at Willowdale Ave (between Yonge and Bayview).  This is far enough away from Finch Station that terminal effects do not muddy the data.  Although the headway is nominally six minutes, the actual headway lies in a wide band particularly in the morning and evening (when the headway widens to nine minutes).  The range of headways considerably exceed the TTC’s own target of a 3-minute band, and many riders will encounter headways much wider than the advertised schedule.

This pattern continues at various points eastward along the route.  Note that data shown for the terminal, Neilson Road, reflect the time vehicles arrived in the general area of th terminal, not exactly at Finch and Neilson itself.

Westbound headways starting at Tapscott (page 2) show the same pattern as we see at the west end of the line.  Headways lie across a band of values exceeding the +/- 3 minute TTC standard.

Friday, December 1, 2006

December 1 Service Chart
December 1 Headways Westbound
December 1 Headways Eastbound
December 1 Link Times Westbound
December 1 Link Times Eastbound

Weekday service in December 2006 was somewhat different than today:

  • AM peak service consisted of four branches on 6’00” headways:  local services from Finch Stn. to Seneca College and to McCowan, plus express services to McCowan and to Neilson.
  • Midday, all service ran to Neilson on a 3’00” headway.
  • PM peak service had four branches:  locals to McCowan and to Birchmount each on a 7’00” headway, and expresses to McCowan and to Neilson on an 11’00” headway.  This meant that service to Neilson was much worse in the PM peak than at any other time of the day.
  • Evening service was every 4’00” to Neilson dropping to 8’00” in the late evening.

Today (June 2008 schedules) the service is:

  • AM peak on the same four branches, but each on a 5’15” headway.
  • Midday all service runs to Neilson, but half as express, on a combined 2’30” headway.
  • PM peak service has only three branches:  the McCowan and Birchmount locals each are  on 6’30” headways, while the Neilson express runs every 6’00”.

I will refrain from pointing out what a great candidate this route would be for some “higher order” transit service as that’s not the point of this thread.

The Service Chart shows the very dense service on this route, and if you look closely, you can see the expresses passing the locals during the peak period.  The service spacing is not ideal at the end of the AM peak with two gaps opening up westbound to Finch Station between 9 and 10 am.  However, there are so many buses on Finch that even a “wide” gap is comparable to a typical day on the Queen car.

By the afternoon peak, gaps start to appear again as well as some delays eastbound between Highway 404 and Victoria Park.  As we will see in a later post, this does not happen every day.  The weather on December 1 was heavy rain, and there may have been some overflow congestion from the highway.

Particularly noticeable is the widely spaced service to Neilson Road.  Although this is scheduled at a 12-minute headway, the actual service is much more scattered as we will see in the headway charts in more detail.

Things settle down in the evening, but service spacing is still irregular.

The headway charts show the variations much more dramatically.  Note that the first and last pages, corresponding to the terminals, have to be taken with a grain of salt because CIS does not reliably report departure times.  Therefore, the charts for Tapscott and Willowdale (near Neilson and Finch Station respectively) are more reliable.

For December 1, the westbound headways from Neilson are appallingly bad in the PM peak, and we will see this pattern through the month.  Although the service is supposed to run every 12 minutes, many buses are on headways of 18 minutes or more, and there is a gap of over half an hour.

At McCowan, the short turn service has merged in, but there are still gaps over 10 minutes throughout the day all the way over to Finch Station.

The return service eastbound is a little better.  At Willowdale, most of the headways lie within the band of 0 to 6 minutes and therefore fit the TTC’s “on time” standard.  However, as we continue east on the route, just as with the streetcar lines downtown, the short gaps get shorter while the wide ones get longer and gaps of 10 minutes are common.  Beyond McCowan, the service to Neilson reminds me of the Queen car at Long Branch with pairs of buses and wide gaps.

I believe that what we are seeing here is the result of applying a laissez-faire attitude to a route that has lots of service on paper.  There is no need to manage it because for most of the route, another bus will be along any moment.  Service at the outer end is left to fend for itself.  As we will see in the whole-month reviews, this situation is common to weekday operations, not the result of a single day’s foul-ups.

I should mention that I cannot explicitly identify the express trips from the CIS data.  During periods when the Neilsons all run express, one can assume that these are express trips.  However, it is impossible to definitively identify the McCowan locals and expresses.

The Link Time charts are interesting because they are so boring.

Again a caveat about terminals:  Because the terminal departures are not always correctly reported, the link time the next time point I have chosen may vary.  This typically shows up as spikes well above the prevailing level caused by some of the layover time being included in the departing trip time, and in wide variations in link times approaching a terminal such as Finch Station.  (It would be nice to quantify congestion problems there, but the CIS data is not at a level of detail and accuracy allowing me to do this.)

Westbound, the link times show little of the variation we saw on Queen Street, and what we do see fits into the classical peak periods rather than the off-peak.  For example, there is a slight rise in times from Seneca College to Leslie in the PM peak, but only by one or two minutes over the all-day values.

Eastbound, the link times don’t vary much except from Seneca College to Victoria Park in the PM peak.  This corresponds to the congestion we saw in the Service Chart earlier.  Other than this, the link times have little variation.

This is a good example of a route with very frequent service and little traffic congestion compared with ohter parts of the system we have looked at in this series.  Nonetheless, service does bunch, and the one part of the route with scheduled wide headways suffers from the randomness of the frequent service projected into a branch where this causes wide variations.  The lesson here is that special care is needed for routes where a comparatively infrequent branch is part of a much more frequent whole.

In the case of Finch East, the issue is moot to the degree that the 2008 schedule no longer has the planned infrequent service to Neilson during the PM peak.

In the next installment, I will turn to full month charts of headways and link times.

8 thoughts on “Analysis of Route 39 Finch East: Part I — Introduction

  1. Why can they not have automatic headway control? Something like they have on the subway only different. For example, it would detect if the next bus is more then say 5 minutes behind where it is supposed to be (IE if the bus behind if you is supposed to be 7 minutes behind you, then if it is 12 minutes behind you, it will alert you) and tell you to hold at the next set of lights. Hopefully if drivers knew how far they really were from other buses they’d be able to take the proper self-corrective actions. For every two drivers who don’t care, one might, and that hopefully will see some of the worst gaps filled.

    Steve: This is possible in theory, and seems to be part of the design for an updated CIS that will be based on GPS sensing, not the old signposts. The mock-up displays for transit control show the space between vehicles in minutes, and if it can be done for Control, it can be transmitted to teh vehicles. Today, they are told where they are relative to schedule, but that’s almost meaningless in terms of headway management.

    Like

  2. The one thing that I wonder is how the revised express service (more closely spaced stops) compares to its former self in terms of travel time, and how much time is really saved in the off-peak times. With Finch being so frequent, why does all the service run to Neilson all day? I have been there, and the ridership is not there, at least not to the point of justifying 3-5 minute frequencies all week.

    Like

  3. Steve, as a Seneca College student who uses the 39 Finch East I can tell you that the Finch bus is frequent some days and others it is not. That being said, they usually bunch up with 4 buses showing up at once, one behind the other at Seneca College at any time during the day. It is not uncommon to be waiting 40 – 60 seconds before having a bus show up.

    I find that from Birchmount (where I get on) to Don mills in the AM the traffic is horrible and roadway jammed with traffic. This is going westbound. It usually lets up at Don mills. In the afternoon traffic is not that bad going east or westbound. It keeps on moving and the I can easily make it Finch station in less that 8 minutes via an express bus from Seneca College.

    One thing I would like to point out is that no matter what time of day it is, there is always at least 3 Finch buses waiting to pull into the platform at Finch station. How’s that for bunching up.

    Steve: Your comment about westbound to Don Mills is intriguing because the data for December 2006 show no such effect. Is this a recent phenomenon? Has there been some change that would cause traffic to back up on Finch this year that was not there in late 2006?

    Like

  4. IMHO, Steve, the buses need to be tackled first. Something on rubber wheels that can move around a slow bus and has headways short enough on the best of routes that even if 2 buses play leapfrog the whole way to the terminal, that you still have a 5-minute-or-less headway, at least on paper.

    IF you can get the buses to properly space their headways, THEN getting streetcars to do so is no problem. IMHO this must be done first because it is so much easier to do.

    Here’s an example. A “gap bus” sitting at York University Common can easily rush into service to fill gaps on many busy routes (the 196 Rocket, 35 Jane, 60 Steeles West, 36 Finch West etc etc), whereas a “gap streetcar” sitting at the Earlscourt loop is rather tied down in location.

    IF we can make the busing headway problem disappear and IF we can finally manage to get the buses to run on time, then logically it follows we should be able to do the same with streetcars with relative ease. Given how much more flexibility buses have, IMHO, we must start here first.

    Steve: It’s not quite so simple as you make out. When you see the all-month charts coming in the next post, you will see that Finch East has very little time-based congestion as compared with many other routes, especially the streetcars, and the buses seem to have little problem making up their running times. The problem is that on a frequent service where +/- 3 minutes is comparable to the headway, nobody bothers with vehicle spacing on those occasions when it is needed.

    Like

  5. Nick J Boragina wrote: “Why can they not have automatic headway control? Something like they have on the subway only different. For example, it would detect if the next bus is more then say 5 minutes behind where it is supposed to be (IE if the bus behind if you is supposed to be 7 minutes behind you, then if it is 12 minutes behind you, it will alert you) and tell you to hold at the next set of lights. Hopefully if drivers knew how far they really were from other buses they’d be able to take the proper self-corrective actions. For every two drivers who don’t care, one might, and that hopefully will see some of the worst gaps filled.”

    If I am currently running on time and the bus in front of me is also running on time, do we hold both buses to even out headways? The bus behind could be delayed because it has had to deploy a ramp (or a lift in the case of 39) to embark a passenger using a wheelchair or a scooter. As an operator, I have seen some passengers load and place themselves into the accessible seating position as quick as a mobile passenger, but I have also had instances where a mobility-challenged passenger has taken upwards of 5 minutes to load and place themselves in position.

    Are we to delay the entire line to even out headways in this circumstance? As more routes become accessible and more mobility-challenged passengers start riding on a regular basis, dwell-time delays at stops will become more common. Your suggestion could have a ripple effect on the entire line if the scheduling were to be headway based as opposed to terminal departure and timing point based.

    I offer my opinion strictly from an operator’s viewpoint, but I am not defending those operators who engage in soaking or deliberate slow operation. I am stating that there are legitimate reasons why some vehicles are not maintaining headways at any given time on a route.

    Steve: The issue for me is far more on routes that suffer general delays through traffic congestion from weather or special events. It may be preferable to operate the line on a wider than scheduled headway, but this requires changes in crewing practices.

    Even on the current basis, a dispatching system should be able to distinguish between one bus that’s off schedule and a situation where many buses are running late. How long we will have to wait for CIS to have such an intelligent management strategy built into it I don’t know.

    Like

  6. Richard White’s comment about “bunching” is very interesting. I operate on several north-south connecting routes on a regular basis: Don Mills, Victoria Park, Warden, Birchmount. I have noticed this as well. Finch buses do tend to travel in “packs”. I have arrived at a red llight at Finch to observe a pack of several buses depart from the stop. By the time the light cycles, another pack arrives.

    I have ridden the Finch East on several occasions as well. Speaking with the operators indicates that the headways are very close and that it is not always possible to avoid the “pack” due to traffic conditions and passenger loading/unloading conditions.

    One other observation: once the express buses exit express territory and go into local territory at Brimley Road, all bets are off as I have seen packs of buses all vying for space at the stop to each let off one lone passenger.

    Steve: A related problem on suburban roads is the long cycle time for traffic lights. This tends to marshall buses into packs which, once formed, are hard to break. In my analyses, the frustrating effects are those where a pack arrives at a terminal and then returns intact on the reverse trip.

    Like

  7. Thanks once again for these analyses, Steve. This one is of particular interest to me as I take the 39 to work everyday from Finch stn to the far east.

    One general question/comment on your methodology here: had you considered adding the 139 route to this analysis? Looking at their behaviour might help explain what’s going on in this route sometimes.

    For example, in theory, those “appallingly bad” headways in the P.M. peak express service E of McCowan should be mitigated by the 139 service (itself on about a 12 minutes headway then I think) which in theory would make a combined service of six minutes or so — except for the fact that the 39 and 139 buses (both then and now) tend to run together in pairs.

    “I should mention that I cannot explicitly identify the express trips from the CIS data. During periods when the Neilsons all run express, one can assume that these are express trips. However, it is impossible to definitively identify the McCowan locals and expresses.”

    Sadly, the data won’t reveal this information, but I would love to know how many scheduled Neilson express buses got turned back at McCowan — long a favoured technique on this route.

    I don’t recall how much they were doing it in December 2006, but there were also sporadic experiments with short turning buses at Middlefield via Tiffield.

    Steve: I only have data for the 39, not for the 139, and so cannot build a chart of combined services. The headways east of McCowan suggest that short-turning is happening out there as you will see when I publish the all-month charts.

    As for the 139 mitigating the impact, the problem with this route is that anyone who starts west of Don Mills can’t use that service, but would have to transfer enroute. A lot of people tend to wait for a bus or streetcar that says it’s going to their destination rather than taking the first thing that comes along and changing enroute.

    Like

  8. Steve, this is not a recent phenomenon. I travel this route all the time. It is usually in the AM rush that the road is busy like this. I think it has something to with the 404 Nearby, and all the cars looking to get onto it. It might also have something to do with the amount of traffic lights on that stretch of road as well. There is a ginormous amount of them, There is one probably every 200 metres or so ( they are relativly new some of them), with the road as heavily travelled as it is, this could cause a bit of a back up. All I know is steve, during the AM rush, avoid finch ave east.

    Like

Comments are closed.