Updated November 2, 2018: ERRATA: Capacity charts in the original version of this article omitted 514 Cherry cars from February to June 2018. This has been corrected.
Updated October 11, 2018: Charts have been added at the end of the article giving more detail about the effect of TIFF on operation of the King Street service.
September 2018 brought a major change on King Street with the presence of the Toronto film festival, TIFF, and diversion of service around the festival district. The service design was the same as in 2017, but last year the King Street Pilot and associated traffic restrictions downtown had not begun.
Other service changes in September included:
- The return to the standard 504 King routing from Broadview Station following completion of track construction on Broadview.
- Reinstatement of 514 Cherry following a split 504 operation during the construction period. (The split operation will become standard on Sunday, October 7.)
- The 503 Kingston Road car resumed tripper operations to York Street, but this lasted only to mid-September when the route was extended again west to Spadina (Charlotte Loop) to accommodate construction on Wellington Street.
Peak Travel Times
PM peak travel times continued the pattern seen over the past year, but the TIFF period produced major disruptions because of service diversions. Note that in the chart below, travel times across the pilot area from Jarvis to Bathurst include the time spent on diversion all day on Thursday and Friday, September 6 and 7. Diversions also occurred at some times in the week of September 10-14, but these trips are not included below.
- Westbound via north on York, west on Queen, south on Spadina
- Eastbound via north on Spadina, east on Queen, south on Church
The effect of TIFF diversions was worse in 2018 than 2017 with the 85th percentile of travel times on Thursday September 6 hitting 54 minutes. The chart below expands the first three weeks of September and includes four percentile lines rather than the two used in the chart above. Note that the four lines stay close together indicating there was little spread between the best and worst case values.
- Thursday/Friday September 6/7: The diversion via Queen more than doubled the travel time between the bounds of the pilot.
- Monday September 10: No diversion or temporary service blockage at TIFF affected the period from 5-6 pm.
- Tuesday September 11: Service was blocked at TIFF during the 5-6 pm hour:
- some cars were held producing higher travel times at the 85th and 100th percentiles;
- some cars diverted (not included below);
- some cars ran through unimpeded producing a 25th percentile similar to “normal” days.
- Wednesday September 12: Emergency sewer repairs west of Bathurst required a diversion via Queen for much of the day. No cars operated through King and Bathurst and so there were no trips on King between Jarvis and Bathurst to measure.
- Thursday/Friday September 13/14: Travel times returned to close to values seen before and after the TIFF period.
Travel times eastbound were also affected by TIFF and the diversions, and the effect was comparable to westbound data.
I will return to the effects of TIFF on all day travel times later in this article.
Here are the full sets of charts:
Eastbound at Jameson
With the return to the service design from Spring 2018, capacity during the peak hour inbound at Jameson fell back to comparable levels, although the proportion of CLRVs continues to decline compared with earlier in 2018.
Eastbound at Bathurst
At Bathurst, the 514 Cherry service adds to the line’s capacity and the service comes in striking distance of 3,000 per hour based on service design crowding levels. (Actual crush levels are higher, but service design values are used for these charts.)
[Updated November 2, 2018 with a revised chart.]
Westbound at Yonge
The PM peak hour westbound capacity at Yonge shows two effects in September 2018:
- For the early part of the month, the absence of 503 trippers heading west to Spadina drops the CLRV portion of the capacity at Yonge almost to zero as other service was provided mainly by Flexitys.
- On the first two days of TIFF, the service was badly strung out by added running time for the diversions, and capacity was considerably below the usual values.
[Updated November 2, 2018 with a revised chart.]
Westbound at Bathurst
At Bathurst, the capacities do not include the 503 Kingston Road trippers in the latter part of September because these turn back from Spadina, and only the 504/514 services continue westward. There is a double notch in the capacity chart where there is the effect of TIFF primarily on September 6/7/11 and the total lack of service at Bathurst on September 12.
[Updated November 2, 2018 with a revised chart.]
Westbound at Jameson
At Jameson, only the 504 King service is present.
The capacity at Jameson drops substantially during the first part of the TIFF period because less service reaches this part of the route. This shows how the effect of TIFF can stretch well beyond the festival district and affect riders well beyond that area.
The full sets of charts covering four hourly periods at each of the locations shown above are linked here:
[Updated November 2, 2018 with revised charts for Bathurst and Yonge locations.]
The Effects of TIFF
Some of the TIFF effects are shown in charts above. This section dives into more detail.
There is no question that TIFF has a severe effect on both King and Queen streets and their transit services for the two weekday closures on opening weekend, and a lesser effect mainly on King on the Saturday/Sunday. The TTC operates a service using schedules wholly unsuited for the actual conditions, and riders suffer from the combined effects of congestion, diversions and considerably poorer service than they would normally receive. This is a known event, and the TTC should plan appropriately. That said, extra costs of providing good TTC service should be borne by the City as part of its TIFF support. Ideally, the street shutdown would be confined to the weekend when the effects were less, but actually forcing this change will take political will the current administration is unable or unwilling to muster.
During the first week of September, Thursday and Friday saw the service divert via Queen. The difference in travel times from Jarvis to Bathurst is obvious in the chart below where the stable running times without the diversion are replaced by much longer travel times on the 6th and 7th. There is a peak period effect with the diversion that is not seen on other days, and this peak extends well beyond what is typically thought of as the “peak hour”.
(There is one rogue data point just after 1 pm on Tuesday where a travel time of about 5 minutes is shown. This is caused by a GPS error in the data that fools my analysis routines. The more outrageous of these are filtered out.)
In week 2, two effects show up in the chart:
- On Tuesday, September 11, there were two periods when activity at TIFF held streetcar service. These appear as clusters of red dots well above the trend line which is itself pulled upward by these data points.
- On Wednesday, September 11, there is no data for the period of the diversion around sewer repairs west of Bathurst. This causes the trend line to dip in the middle of the day.
On weekends, the difference in travel times over the diversion routing is clear for both Saturday and Sunday/Holiday data.
Eastbound data are similar. Both sets are included in the pdfs below.
Headways at the outer ends of King saw some effect during the first TIFF week when diversions operated, although this was mainly on the west end of the line. This implies that most of the short turning to correct for extended running times took place westbound rather than eastbound. (I will leave a discussion of the number of very low and very high headways and the issue of gapping and bunching to another post, but the problem is quite obvious in the headway data here.)
Queen Street was affected by the complete closure of King Street. Here are the westbound travel times from Yonge to Bathurst for the first week of the month. The effect on the two weekdays when King was closed is quite clear.
In week 2, there is no major change, although the two sets of long times on Tuesday correspond to the periods when King was closed for TIFF activity (red dots below, compare to the King data for same period).
Weekends saw additional travel time on the first Saturday, but only a slight increase in times on the first Sunday. Note that the trend line for September 29 goes a bit wonky late in the day due to the onset of Nuit Blanche.
The eastbound operations show comparable changes on TIFF days. The full chart sets are linked below.
Service to Humber Loop was less frequent on the first two TIFF days because the congestion triggered more short-turning. Note that although the trend line for September 6 only tops out at about 10 minutes, there are many examples of headways at 15 minutes or more on that date. At Neville Loop, there was only a modest change in headway behaviour compared to the usual level of service irregularity at that location. (Note that for the east end of Queen, headways are measured at Silver Birch to avoid confusion about actual arrival and departure of cars due to on-street queuing at Neville.)
Updated October 11, 2018
This section contains additional charts showing the effect of TIFF on King Street operations at a more granular detail.
The next two charts show the average travel times and standard deviation values by hour broken down by week for westbound and eastbound trips.
- Week 1 included two days of TIFF diversions, and this pushed values higher than for the other three weeks.
- The smaller increases later in the day in Week 2 were caused by temporary TIFF diversions on two days of that week. Otherwise, operations on King through the Festival District were not substantially affected.
Another way of looking at the data is to break out the quartile values. The chart below breaks the data into four groups by quartile. The bottom of each column is the lowest value observed, and the top is the highest value. The boxes in the middle cover the second and third quartiles (25% to 75%) of the measured travel times, and the bar across each box is the 50% (median) line.
The length of the boxes shows the degree to which values are spread out and how this evolves from hour to hour. However, this does not tell the entire story because the Week 1 data contain two days of normal operation and two days of TIFF diversion.
When we break out the non-TIFF days (Tuesday-Wednesday September 4-5) from the TIFF days (Thursday-Friday September 6-7), the real difference emerges. Travel times are considerably higher for TIFF days and the values have a greater range. The non-TIFF days in Week 1 (as in other parts of the month) show a similar pattern to those in Week 1 with a relatively small range of values in the 2nd and 3rd quartiles and short “whiskers” above and below the boxes for the 1st and 4th quartiles. This shows how the King Street Pilot area is producing quite reliable travel times through the core area.
(Note that the low value at 13:00 in the first chart was caused by a GPS error on one car that appeared to travel more quickly across downtown than all others. I have not filtered events such as this out of the data as they are comparatively rare.)
Full sets of updated travel time charts including detailed week-by-week breakdowns are linked below.