Earlier this year, I published two analyses of the 29 Dufferin bus operation for March 2012.  Part I dealt with headway reliability, and Part II looked at running times.

In this article, I will review the details of operations on four sample days from the month to illustrate the fine-grained detail of service on some representative days.

Tuesday, March 13, 2012

March 2012 was very mild with little precipitation.  On March 13, temperatures ranged from 2.7° to 18.1°C and there was a small amount of rain (0.6mm).  This was the second weekday of the March school break when traffic is typically lighter than usual.  As weekdays go, these are fairly benign conditions.

• 29_20120313_Headways_NB
• 29_20120313_Headways_SB
• 29_20120313_Chart

The headway charts each have seven pages corresponding to seven points along the route starting at King Street in the south and ending at Wilson and Transit Road in the north.  The time of day is read across the chart, and each point represents a bus and the spacing between it and the vehicle ahead.

If every bus came at exactly the same spacing, all of the dots and the trend lines interpolated between them would be straight except where there is a change in the scheduled headway.  The trend lines do roughly track the scheduled headways, but the variation between vehicles can be seen in the sawtooth pattern of individual points.

Where a point is on the zero line, this means that two buses (sometimes more) passed the location “together” within the measurement accuracy of the vehicle tracking system.  A very low headway is always accompanied by high headways corresponding to the service gap on either side of the bunched service.

The TTC’s goal is for 65% of the service to lie within three minutes of the scheduled headway (in effect, three minutes either side of the trend line).  With a few exceptions this is the case northbound at King until about 2pm (1400 hours), but things get more ragged as the day wears on.  In the evening, at around 8pm (2000 hours), there are two gaps of about 20 minutes, and several of the data points thereafter lie outside of the three minute target band.

At Bloor Street northbound, the wide gaps in the evening have lessened thanks to short turns, but headways are still spread over a 12-minute range.  By the time we reach St. Clair northbound, the gaps between vehicles are getting longer starting around 3pm (1500) and clustering of vehicles is quite evident.  This continues all the way to the north end of the route.

The southbound headway charts need to be read from the last page backward to see the progression down Dufferin Street.  Service leaving Transit Road (just outside of Wilson Terminal) southbound has a fair amount of scatter around the trend line except for a brief period at midday.  By the pm peak, wide headways (commonly over 12′ with one spike to 24′) are common and service does not get back to a better behaved pattern until mid-evening.  Bunching is still evident until about 10pm (2200).

Working south along the line, the erratic headways of the pm peak settle down once we reach Eglinton and the Briar Hill short turn service has merged into the route.  Even so, the headways range over a 12-minute interval for much of the day and into the late evening.  The closer we get to the south end at King, the more pronounced the gaps become.  This is a typical pattern with poorly spaced service because trailing buses in short gaps catch up to their leaders and the gap behind them grows.

What is quite clear on these charts and others like them in this article is that no attempt is made to smooth out the service by breaking up packs of buses and forcing them to operate on something close to a regular headway.  In both directions, vehicles leave their terminals unevenly spaced, and it is no surprise that this condition simply reinforces itself along the route.

The service chart maps every vehicle on the line in space and time.  Each bus travels back and forth from north (top) to south (bottom) of the route, and its chart moves from left to right in time.  Any irregularity in service shows up with uneven spacing between vehicles, or bunching caused by events such as traffic accidents and congestion.  Where lines run horizontally, a vehicle is stationary for a time either at a traffic signal, a busy stop, or a layover point.

The first major gap shows up just after 7am southbound at Eglinton when five buses show up together at about 7:15am.  Two of these originated from the Briar Hill short turn, and the other three from Wilson.  They travel together southbound until one of them (the yellow line in the chart) short turns at College, has a short layover, and then returns north as a trip to Briar Hill.  Two of them go as far as Dufferin Loop (at Springhurst) (the black line and a faint green one).  The final two make a trip to the eastern entrance of the CNE, but then return to Dufferin Loop where they have a layover before returning northbound.

Although there is a bit of bunching, this is an example of a pack being sorted out to restore better spaced service on the return direction.

The remainder of the am peak is fairly uneventful although some bunching is evident.  We know from the headway charts that this not a problematic part of the day.

Similarly, the period from 10am to 1pm is fairly well behaved with only occasional bunching and no wide gaps.  It is worth noting here that vehicles get time for a reasonable layover at both the Wilson Station and Dufferin Loop terminals.  There appears to be only one short-turn southbound at Queen at about 11:50am (bright green line on the chart).

Coming into the afternoon, the character of the chart changes and we can see gaps developing.  Note that this is not accompanied by a change in the overall slope of the lines on the chart that would indicate congestion (i.e. slower operation) at specific troublesome parts of the route.  A gap leaves Wilson Station southbound at about 1:30pm and it gets progressively wider as a cluster of five buses makes its way south to Dufferin Loop.  Much of this cluster returns north, and it is only on their next southbound trip, after 3:00pm that somewhat better headways are restored.

Throughout the afternoon, pairs and trios of buses leave Wilson Station southbound.  Almost none is short turned, and where one does turn back at Queen (pink line, about 2:45pm), the bus immediately becomes part of a pair that grows to a quintet of buses headed north.

After 3:00pm, the effects of congestion are evident for northbound service as the lines tilt over and vehicles take longer to traverse their route.   One factor is the problem of overcrowding which can affect vehicle performance and stop service times.  Buses running in gaps can take longer then evenly-spaced counterparts over the same segment of a route.

A gap that appears to be congestion-induced begins northbound from King and widens to 24 minutes by the time we reach Lawrence.  At roughly the same time and location, a pack of three buses is travelling south from Wilson Station.

From 4:00pm to 7:00pm, gaps and bunching are common, and there is severe congestion south of Queen Street.  Service to the CNE Eastern Gate is erratic with many of these trips short-turning at Dufferin Loop to save time.

By the early evening, the congestion problems are gone, but bunched operation is the norm with multiple vehicles leaving from both terminals.  Short turns at various locations attempt to put buses back on time, and some of these moves even fill gaps to even out the service.  However, the pattern continues up to 10:00pm (2200).  Things are somewhat more reliable in the late evening, but an example of, shall we say, unprofessional operation is evident with a pair of vehicles (turquoise and dark blue) that leave Wilson southbound at about 11:00pm and run as a pair for two round trips until “dark blue” goes out of service at 1:10am.

That was a more-or-less “ideal” day for operating conditions.

Friday, March 23, 2012

March 23 was another day with tolerably good weather.  Temperatures ranged from 11.9° to 17.4°C, and there was 0.8mm of rain.  This was the beginning of a fair weekend, and spring traffic was definitely in evidence.

• 29_20120323_Headways_NB
• 29_20120323_Headways_SB
• 29_20120323_Chart

The overall form of the northbound headway charts is similar to what we saw on March 13, but the range of values is worse especially in the evening.  Afternoon peak effects appear a bit sooner, and headways are more extended toward the north end of the line.

Late in the pm peak, there is a huge gap (over 30 minutes) southbound from Wilson Station southbound.  This is part of a large gap that echoes over the route through the peak as we will see in the detailed chart.

As on March 13, many buses are leaving their terminals close together and this produces the inevitable pattern of gap-bunch-gap-bunch. throughout the day.

On the service chart, the differences in the character of service show up in many ways:

• During the late morning and midday, there is slightly more short turning, and pairs/triplets of buses are already forming.
• At certain times, congestion both ways approaching St. Clair is evident, but the cause is unclear.  Nothing in the TTC service alerts for March 23 mentions the Dufferin route.
• By the early afternoon, packs of four or more buses are evident.  Although there is some short turning to break these up, they reform just as quickly.
• After 3:00pm (1500), congestion northbound around Yorkdale Plaza is evident.  (This sort of thing showed up in an analysis I did of 29 Dufferin on Boxing Day several years ago.)  This effect disappears after about 6:00pm.
• After 4:00pm (1600), a wide gap travels north from Bloor and opens out to over half an hour by the time it hits Wilson Station.  Some short turns come into the southbound echo of this gap, but it persists to the south end and …
• After a quintet (!!!) of buses leaves Bloor northbound at about 5:45 pm, there is a gap of nearly half an hour that travels north to Wilson.
• The early evening is a shambles with a parade of seven buses reaching Dufferin Loop southbound around 7:40pm in a gap of about 20 minutes.
• Despite some short turning, the service continues with bunches and gaps through the evening.  Even the late evening service isn’t well-behaved, but there are fewer buses on the street (and lines on the charts).  By this time, nothing is short turning, but there is also no attempt to space out the service and provide an even headway.

That’s an example of a really bad Friday, although the weather was not a factor here.  The problems may arise from greater demand and congestion, or from a different (or laissez-faire) management approach, or from the schedule which would have been the same for the two days.  To what extent each contributed to the wide difference in service character I cannot say definitively, but the difference between a light traffic and heavy traffic day are certainly evident.

Now we will turn to the weekend service.

Saturday, March 17, 2012

From the chart of running times northbound from King to Transit Road published in a previous article, we know that there was considerable difference between March 17 (the weekend at the end of March break) and March 24.  On the 24th, running times were much longer, and this is reflected in a very different service pattern.

Temperatures on March 17 ranged from 8.6° to 11.5°C, and there was no precipitation.

• 29_20120317_Headways_NB
• 29_20120317_Headways_SB
• 29_20120317_Chart

The northbound headways show a similar overall pattern to what we saw for weekdays.  The spread of headway values exceeds the TTC’s target three-minute band throughout the day, and this becomes noticeably worse in the evening.  Many vehicles leave the south end of the line close together, and the corresponding gaps in service travel north getting wider as they go.

Southbound headways from Wilson Station are better behaved especially through the morning and early afternoon.  They spread out somewhat from 3:00 to 6:00pm, tighten up a bit, but then become very scattered through the evening.  The pattern at Wilson deteriorates further south on the route with more bunches and wider gaps evident at Eglinton, and worsening by the time we reach Bloor.

On the service chart, we can watch the situation unfold.  The first page, before 7:00am, shows quite consistent running times and reasonable vehicle spacing.  Between 7:00 and 10:00am, some bunching begins to appear, but also the lines show that pauses at stops are now becoming more common, and some of these are fairly long, notably at Bloor.  For a few trips, one bus (the purple line appearing southbound between Lawrence and Eglinton at 7:00am) seems to have trouble staying on time, and its follower (black), although properly spaced relative to other service, manages to catch up.  At 9:00am, “purple” is short-turned at Dufferin Loop and gets back into the correct spacing.

Between 10:00 and 1:00pm, the same pattern continues, but with even more time spent at stops.  (Generally speaking, delay at stops will appear as short horizontal steps in all of the lines at the same vertical positions.  Delay due to congestion will be spread out and appear as a change on the slope of the lines.)  A gap southbound from Wilson at about 11:15 appears to be due to a bus spending a long time at the terminal (possibly a missed crew change).  This vehicle reappears in its proper place filling the gap southbound at St. Clair around 11:40.

By the early afternoon, bunching is becoming more common and some congestion effects are starting to appear around Bloor and near Yorkdale.  These get worse in the mid-afternoon, and bunching is quite evident.  Very little short turning occurs, and buses run back and forth with some gaps echoing across the route.  This continues into the early evening.

After 8:00pm, the headways widen, but nothing is done to correct the irregular spacing of buses.  They continue to travel in pairs and trios sometimes leaving terminals as a pack (northbound from CNE East just before 8:30pm, and again just after 8:40pm.  Bunching continues through into the late evening producing the wide swings in headway values shown on the earlier charts.

Saturday, March 24, 2012

Temperatures on March 24 ranged from 6.5° to 12.5°C, and there was 0.4mm of rain.

• 29_20120324_Headways_NB
• 29_20120324_Headways_SB
• 29_20120324_Chart

Northbound headways show the same pattern on the 24th as on the 17th, but with wider swings in the values, and an earlier onset of service well beyond the six-minute target band.  One notable feature is a sudden recovery to short headways between 6:00 and 8:00pm, but otherwise many trips are at headways much wider than the advertised service.

Southbound headways, notably in the morning, are not quite as regular as they were on the 17th and wide values of 12 minutes or more show up at around 3:00pm.  The evening headways are erratic as on the 17th with gaps over 15 minutes showing up regularly.

On the service chart, the day begins roughly as it did on the 17th although there is somewhat more bunching in the morning.  A gap opens up northbound from the CNE at 8:20am, and this echoes back from Wilson with a parade of five buses southbound at about 9:15am.  The late morning continues in the same vein with some bunching, and an increasing amount of time spent at stops.  One vehicle sits, presumably out of service, northbound at Davenport for a bit over half an hour (10:55 to 11:30am), but it is placed back in position with a Briar Hill short turn at 11:45am.

The familiar Yorkdale congestion begins to appear northbound at about 12:30pm, but this persists until about 6:30pm and is more severe than on the 17th.  Congestion between Queen and King both ways appears after 3:00pm, and this persists until about 5:00pm.

By 1:30pm, the service has become marshalled into groups of three or more buses, and these run back and forth in packs with almost no short turns or service spacing right through to the early evening.  Short turns do become more evident after 4:00pm, but comparative order is not restored until about 6:00pm.  This takes us into the hour of relatively well-behaved service until the point where many vehicles return to the garage.

From 7:00pm onward, and especially after 8:00pm, bunching is common, there are wide gaps in service, and there appears to be no effort to sort this all out.  Neither traffic congestion nor lengthy stop service times are evident.

Sunday, March 25, 2012

Temperatures on March 25 ranged from 4.3° to 20.0°C, and there was no precipitation.

• 29_20120325_Headways_NB
• 29_20120325_Headways_SB
• 29_20120325_Chart

Service on Sunday the 25th is less affected by congestion and stop service times than on the 24th, but more than makes up for this with some breathtakingly irregular spacing of vehicles.

Headways northbound from King are erratic right from the morning with some buses running at twice the normal headway or more at 9:00am.  Bunching continues through the daytime with pairs of vehicles running northbound right through to 6:00pm.  The evening brings headways ranging from under 6′ to over 30′ at King, although short turns reduce this range slightly at Bloor.  However, the wide gaps become progressively wider exceeding half hourly by the time the service reaches Wilson Terminal.

Southbound, the situation is similar.  Buses leave Wilson southbound on an irregular headway, often in groups, right from 9:00am.  The evening service is badly bunched and gaps over 15′ are common even with the benefit of some short turns at Briar Hill to fill the widest gaps arriving southbound from Wilson.  Gaps over 20′ are common at Bloor, and when these are compounded with short turns, gaps to the south end of the line are common and large with a peak of 42′ at about 9:00pm.

The service chart shows some of the gory details:

• Bunching southbound from Wilson at 9:00am is clearly caused by vehicles entering service off time such that they tend to run together with a vehicle already in service.
• There is no congestion through the morning and midday period, but the effect of stop dwell times is evident.  Some bunching does occur, but this gets sorted out on return trips.
• A gap opens up after 12:30pm northbound from King with one vehicle (pink line) taking considerably longer than others nearby to make the northbound trip.  By the time this reaches Wilson, four buses are running in a pack.  Three of them return southbound together.
• As the afternoon continues, the pattern of bus platoons we saw on Saturday the 24th begins to appear notably southbound at about 3:10pm.  A group of four buses leaves Dufferin Loop northbound at about 4:30pm, and they remain together as close friends for one-and-a-half trips.
• By the evening, echoing what we saw on the headway charts, the service has become very badly spaced out.  Some short turns are completely ineffective in filling gaps.
  • Northbound north of Eglinton at 7:45pm, a bus (blue) that has been running as part of a pair is short turned and returns south together with a through bus (pale green).
  • A trio of buses leaves Bloor northbound at 9:00pm, and two of them short turn, but come back south accompanied by a through bus from Wilson.  That bus short turns at Bloor (salmon), but returns north together with a through bus (pink) from Dufferin Loop.
• Erratic headways and bunching continue right through to the end of service.

General Observations

The days I selected for presentation here show how widely the conditions in which service operates can vary from one day to the next even though the same schedule may be used (all weekdays, all Saturdays, etc).

From running time analyses in the previous article, we know that there are times when buses cannot make their trips within the allowed schedule.  Should this be fixed by scheduling to worst-case conditions, or to something less with an expectation of moderate numbers of short turns, or by moving completely to a headway-based scheme with running times adjusted on the fly to suit daily conditions?  This is not a simple question, but the TTC must grapple with it.

On “good” days, the headway actually operated would be better than on “bad” days, but at least the goal on both occasions would be the same – keeping service evenly spaced.  While on paper, this might produce “worse” service than the advertised schedule, we know that the common experience of riders bears little resemblance to it.  “Frequent service” really means “frequent gaps”.

This begs the question of whether it is practical to manage to a schedule when there are such divergent conditions on a route.  Managing to a headway may be more challenging (and would definitely require changes in contract language and crew design), but it could result in a target that might actually be achievable.

Of course, managing to a headway includes one important component: management.  Clearly there are periods on the Dufferin bus where little is done to organize the service and break up platoons to provide better service to riders.  Groups of buses do not just magically appear, but are created by uneven service from the terminals.  The condition worsens through the compounding problem of gaps and overloading.

A Personal Note

On occasion, I get complaints that I am being too hard on the poor old TTC, that I am too unreasonable in expecting better service especially on “difficult” routes such as Dufferin.  To this I make two replies.

First, the patterns of irregular service shown here are pervasive through all of the data I have looked at from the TTC, although the severity varies from route to route.

Problems with vehicle spacing commonly thought to be the preserve of streetcars (which cannot pass each other) show up routinely on bus routes.  The lack of management of short turns – ensuring that these vehicles actually fill gaps rather than simply creating new pairs travelling in the opposite direction – shows up everywhere.

Some operators play games and run hot, right behind their leader, for hours on end.  Some operators dawdle operating in a manner intended to force a short turn and siesta.  But these are exceptions, and they should be easy to spot from routine monitoring.

Second, the problems with erratic service show up on routes even at times and on days when no external “force of nature” bears down on the luckless transit system.  The TTC is too quick to blame outside factors without understanding the reality on the street.  To their credit, the TTC’s recently published route performance statistics show that they now see just how bad and how pervasive the problem is, but we have taken decades to reach this point.  We cannot afford to wait decades to fix the problem especially if we hope to make transit an attractive alternative to driving.

There are locations where the City must make hard decisions about how road space is used.  Should parking be allowed in periods when it triggers severe traffic congestion?  Should deliveries on major streets be restricted to specific hours when there is capacity to give over road space to handle them?  Should traffic signals more aggressively favour transit vehicles?

We are asking taxpayers to devote billions to new transit construction, but somewhere along the way, we need to make the Dufferin bus and other routes like it provide reliable service.