In Part I of this series, I reviewed problems with headway reliability on the 29 Dufferin route. An issue commonly raised by operators is that there are times when schedules do not provide enough time for vehicles to make their journey, and this results in a variety of problems including irregular service.
In Part II, I turn to the actual time required for buses to make their journey on the route during the month of March 2012.
Updated March 20, 2013: In the comment thread, there was a question about whether different vehicles operating on this route showed any difference in travel times. I have added a section to the end of the article to address this. (The short answer is “no”.)
Original Article from March 10, 2013:
To understand what is going on, it is necessary to look at segments of the route so that local effects can be seen clearly, and so that time taken for layovers at terminals is considered separately from driving time.
Weather & Special Events
The weather in March 2012 was generally very mild with small amounts of snow on only a few days.
Weather_March_2012 (Source: Environment Canada)
March Break fell in “Week 3” of the month from the 12th to the 16th. There were no statutory holidays because Easter weekend fell in April.
Scheduled Running Times
The section of interest is for February-March 2012 under the columns:
- RTT: Round trip time allocated to driving. A one-way trip should use half of this value.
- Rcvry: So-called recovery time. In practice, the values allowed bear little relation to the time of day or route conditions. In fact, this is usually an amount added to the round trip so that it will be an even multiple of the headway. A fairly common adjustment the TTC makes is to convert “recovery” time to “driving” time in reaction to increased congestion on a route without actually lengthening the total time allocated to a trip, changing the headway or adding vehicles.
- Total: The sum of the driving time and the recovery time.
In reviewing actual times, I will treat the time between points near the terminals separately from the terminals themself in order to isolate fluctuations due to layovers, if any. On the Dufferin route, the points of reference are King Street (just north of Dufferin Loop), and Transit Road at Wilson (just south of Wilson Station). Driving time should cover this interval with enough left over for entering, serving and leaving the terminals.
Northbound from King to Wilson Station
These charts have the same layout as the headway charts in the previous article, but show travel time between points rather than headway at a point.
Running times on weekdays have a common pattern with values of 35-40 minutes through the AM peak and into late morning. The values are highest in the afternoon slightly before the PM peak itself. This suggests that congestion is actually worse before the onset of peak period traffic restrictions. Peaks on Fridays, and to a lesser extent Thursdays, tend to be higher than earlier in the week.
Running times fall back to lower values as the evening goes on, although again Fridays tend to stay a bit higher than other days.
The cloud of data for all weekdays shows the same overall pattern, but has a notable feature in the small clusters of unusually low values at around 9:00am and 7:00pm. These are buses hurrying back to Wilson Garage.
The PM peak data are more spread out with the “cloud” of points noticeably opening up from about 4:00pm to 6:00pm. There are also fewer data points here because the 29A Tycos short turns don’t reach Wilson are are not included. The values are spread over a range from 40-50 minutes with several reaching into the band up to 60. During this period, the scheduled trip time on the 29D (Wilson Station to Princes Gates) is 54 minutes one-way, and it is clearly impossible for most buses to achieve this when they will need at least 45 just for the section from King to Wilson & Transit Road.
On Saturdays, the peak comes earlier, at about 3:00pm, and it is higher than the PM peak travel time. On Sundays, the travel times rise to a fairly steady value by noon and drop off in the evening.
Saturday afternoon scheduled times allow 51 minutes, all in, for a one-way trip. Some of the trips actually observed, notably on March 24 and 31, took that long just to get from King to Transit Road. Even the lowest of the trend lines on March 17 lies at about 44 minutes, and this leaves only 7 minutes for the route south of King and north of Wilson including terminal time.
Normally I do not include all of the detail for the “uninteresting” parts of a route in these articles, but in this case I will give the whole set to show how there are parts of 29 Dufferin where running times are quite consistent across many periods of service. This is not a situation where the entire route is beset by unusual, unpredictable traffic events.
King to Bloor
Running times in this segment range from about 5 to 11 minutes, but stay within a narrow band that varies slightly and in a similar way over all weekdays. One source of delay is the Dufferin Mall, and this effect depends on shopping patterns. There is no visible AM peak, and only a small PM peak.
On Saturdays, times longer than on weekdays are seen in the afternoon, particularly late in the month. Sunday times are not unlike weekdays, but with lower values in the morning.
Bloor to Lawrence
From Bloor to St. Clair, there is little variation in weekday times except for a short and small peak in the afternoon just before 4:00pm. This could be the effect of peak traffic buildup before rush hour parking restrictions cut in. As with the section south of Bloor, the longest running times are seen on Saturdays.
The section from St. Clair to Eglinton shows almost no variation at all through all periods and days.
Between Eglinton and Lawrence, the times are quite consistent, although there is a noticeably lower average value in the evenings.
Lawrence to Wilson
Weekday running times are fairly consistent across the day with a slight rise in the afternoon and a drop-off in the evening. The last two Fridays are noteworthy for their peak at a higher level than other days. This shows the effect of congestion near Yorkdale Mall which grew later in the month on weekends thanks to the balmy weather in late March.
Similarly on Saturdays, running times in this segment roughly double from 4-5 minutes in the mornings and evenings to over 10 minutes between 3:00 and 5:00pm.
Sunday shows a similar pattern, but a much smaller peak value.
Wilson to Transit Road
The short segment on Wilson from Dufferin to Transit road eastbound shows no variation in running times throughout the week.
The following chart shows the round trip times from Wilson & Transit Road through the terminal and back again. This includes driving and loading time plus any layover.
The trend lines behave a bit oddly on weekday mornings, and the reason can be seen in the all-weekdays data cloud on page 6. There is a peak in values after the morning rush hour with some quite high values. My suspicion is that these represent crew changes where a vehicle sat without an operator for an extended period.
The weekday values tend to be lowest during the PM peak with a trend line at about 7 minutes, and most values lying about 3 minutes either way of this. The values are more spread out and tend to be higher in the evening.
On Saturdays, the times sit at around 10 minutes mid morning and early evening, with a drop to about 6 minutes around 3:00 pm.
Sunday times are spread out especially in the early evening.
The observed values suggest that a round trip on the order of 7-8 minutes represents common operating practices, although a shorter value is physically possible, but likely with only minimal time for a station stop.
Southbound from Wilson Station to Dufferin Loop
Running times from Transit Road & Wilson south to King Street have a short peak for trips southbound between 8:00 and 9:00am. The band of values is about 10 minutes wide through the day until the onset of the PM peak at about 3:00pm, and this lasts until around 6:00pm with a quick fall-off into the evening. As with the northbound trips, the time required to travel the main part of the route leaves little in reserve for the terminal areas.
No one part of the route is responsible for the majority of the extra time (unlike, say, the Queen car where specific sections exhibit large swings in running times while others are almost constant through the day).
Wilson to Lawrence
Weekdays on this segment show little variation through the day with some trips lying well outside the main group of data points in the PM peak, but not much change in overall values for the peak period itself. Early morning and evening running times are a bit shorter and the values are more tightly clustered.
On Saturdays, the running time is higher than on weekdays, notably late in the month as we saw for northbound service. This is the effect of Yorkdale Mall. On Sundays, there is a slight increase in running time during the mid-afternoon, but not as pronounced as on Saturdays.
Lawrence to Eglinton
Between Lawrence and Eglinton, there is little variation in running times, although there are some odd clusters between 6:30 and 7:00am, and again in the early evening. I will examine these in detail in the third article in this series.
Weekends show little variation in running times.
Eglinton to Bloor
Running times from Eglinton to St. Clair, and from St. Clair to Bloor show little variation at all times. I have omitted the charts for them as there is nothing unusual to see.
Bloor to King
Between Bloor and King, weekday running times are quite consistent except for March 30 when service was interrupted during the PM peak. The largest times are seen on Saturdays.
King to Dufferin Loop and Princes’ Gates
There are two terminals at the south end of the Dufferin route. Most service turns at Dufferin Loop just outside the western entrance of the Exhibition Grounds and only a few blocks south of King Street. During some periods, part of the service runs through to a loop near the Princes’ Gates at the eastern entrance. These have been separated so that we can see the behaviour of each set of vehicles.
What is immediately noticeable about the service to Dufferin Loop is that the running times fluctuate widely, much more so than for driving time along other segments of the route. This shows that some drivers are taking substantial layovers at this location especially on weekends. The typical driving time on weekdays from King to Dufferin Loop and back is about 10 minutes during the daytime, somewhat less in the evening.
For service to the Princes’ Gates I have included only the consolidated weekday “cloud” given the gap in the hours of service and the limited number of data points for each day (both of these tend to produce meaningless trend lines). On the schedules, the difference between travel times for “Dufferin” and “Princes’ Gates” trips is 10-12 minutes, but the actual values for weekdays exceed this considerably. The values lie over a band about 10 minutes wide and with values in the range from 15-25 minutes for the PM peak. (AM peak values start off higher because, early in the day, buses are operating with peak period running times, but off-peak traffic.) On weekends, the trend lines lie at about 20 minutes for much of the day.
Both of these suggest that some buses are allocated (or take) considerable layovers at this end of the line.
Round Trip Times from King to Wilson Station
To get an overall view of the time required for buses to operate the 29 Dufferin route, we can look at round trips from just north of Dufferin Loop to Wilson Station. This includes most of the route, but excludes the south end with its separate terminals and widely varying layover times.
Note that the scale on these charts is different at 120 minutes rather than 60. The horizontal lines remain at 10 minute intervals.
Trips longer than 120 minutes which typically involve a long layover at Wilson Terminal have been omitted. What remains shows the typical time required for a bus to drive north from King to Wilson, serve the terminal, and return south to King.
The pattern is consistent for weekdays with a wider range of values coming into the PM peak, especially on Fridays. Round trip times quickly drop off from 90 minutes in the daytime to lower values through the evening dropping eventually to near 60 minutes.
On Saturdays the trendline lies higher from about 10:00am to 5:00pm especially on the last two days of the month. Sundays sit at a consistent value of 80-90 minutes through the afternoon with shorter trips in the morning and evening.
Another way to look at the same data is to consolidate it further to follow overall patterns and statistical behaviour.
There are three charts in this set showing data for weekdays, Saturdays and Sundays subdivided by half-hourly intervals. Each chart shows five lines:
- The mean value (dark blue) — the average value of all trips in the half hour period.
- The maximum and minimum values found within the half hour period on any day. Note that if one day in a month had unusually high traffic congestion, this could push the maximum value up substantially (see 8:00pm weekdays). However, a sustained level of congestion over many days would be needed to pull the mean up as well.
- The mean plus or minus one standard deviation in the values. The standard deviation measures the degree of dispersion of data, and if it is small this means that most values are tightly clustered around the mean. If it is wide, then the data have a greater range of values. Provided that the data behave “normally” (in the statistical sense), about 2/3 of the data points will lie within one standard deviation of the mean.
The following table compares scheduled running times with average values.
In cases where there is only scheduled service to Princes’ Gates, a calculated time for Dufferin Loop is shown. The difference between the scheduled time (Dufferin Loop to Wilson Station) and the observed round trip averages (King to Wilson Station) are also shown. In some cases, this time is insufficient to cover the King to Dufferin Loop trips (exclusive of layover) for the average trip times, let alone for trips that exceed the average.
A basic scheduling problem is the need for running times to reflect the majority of usual demands without being so generous that buses are idling along the route or at terminals. On a route like Dufferin with widely varying conditions it is possible for “average” values to be insufficient under common conditions such as shopping congestion and other activities specific to certain days, weather conditions, or short periods other than the expected AM and PM peaks.
In the next article, I will turn to details of specific times and days to review the behaviour of the route “on the street” with time-distance charts showing the movement of vehicles including bunching and short turns.
Updated March 20
The Effect of Vehicle Type on Running Time
Three different types of vehicles operate on 29 Dufferin: Orion V diesels, Orion “next generation” diesels, and Orion “next generation” hybrids. Given the hilly nature of the route, the amount of stop-start traffic and the heavy passengers loads, one reader asked whether there is a difference in vehicle performance.
To address this, I broke out the running time data for trips northbound from King to Wilson & Transit Road by vehicle type, and calculated the means at half-hourly intervals. The following charts show:
- All weekdays combined on one chart
- Each day, Monday through Sunday, on its own chart
The mean travel time is roughly the same for all three vehicle types (note that Orion Vs are little used on weekends and have been omitted). Weekday times are highest in the afternoon peak with Fridays being the highest of all. Saturdays look not unlike weekdays, but the peak period starts earlier and lasts longer.
Periods when the mean falls to zero indicate that there was no vehicle of the specified type that made the full trip from King to Wilson during the interval on the days in question. This can be caused by short turns north of King or south of Wilson, or by the luck of vehicle assignments to trips leaving within the period.