The summer 2017 conversion of the entire 501 Queen streetcar route to bus operation presents an opportunity to compare the behaviour of the two modes on this route.
Apologies to readers in advance for the length and number of charts, but that’s the nature of the subject.
Background and Data Sources
The raw data for this article comes from the TTC’s vehicle tracking system, CIS, for which much thanks, but the processing and interpretation are entirely my own. The machinery behind the digestion and presentation of TTC data is explained in Methodology for Analysis of TTC’s Vehicle Tracking Data.
In this article, there are data from two separate time periods:
- April 1-30, 2017: At this time, 501 Queen service consisted of two overlapping routes. 501 streetcars operated from Neville Loop to Roncesvalles, while 501L buses operated from Dufferin to Long Branch. A local shuttle, the 501M, provided service on Marine Drive in the Humber Bay area, but it is not part of this analysis.
- May 7-31, 2017: All service on the 501 operated with buses on two branches. 501L buses ran between Neville and Long Branch Loops, while 501P buses ran between Neville and Park Lawn Loops. Buses alternated between the two branches so that, in theory, there would be a 501P half way in between every 501L east of Windermere and The Queensway where the routes diverged.
Many readers will be familiar with charts on this showing the distribution of monthly headways (time between vehicles) and link times (time required to travel between two points). In addition to the detailed data, these charts include summaries of values by hour including averages and standard deviations. The latter values indicate the degree to which actual values differ from the average, and the higher the SD value, the worse the dispersion of individual values. This translates to “bunching” of vehicles which, in the worst case, sees buses running in pairs and trios.
For the purpose of this article, I have created charts pulling together the statistics for streetcar (April) and bus (May) operation. In the case of May, only data from the 7th onward when the route had been converted are included.
Are These Data “Typical” and “Representative”?
In the process of working through the data, I became concerned at the gap between bus and streetcar times. In order to verify whether the April 2017 streetcar values were typical, I also pulled the values for January through March and found that travel times were generally lower for streetcars, although there remain periods (notably evenings) when the bus times over the route are shorter than the streetcar times. However, the difference is not as great as the April 2017 streetcar data implies.
The chart below shows the travel time from Roncesvalles to Silver Birch by month from January to May. (Silver Birch is used as the origin rather than Neville Park because vehicle layovers at the end of the route sometimes occur west of that street, and measurements from that point could include layovers.) May data is bus only, and the other months are streetcar.
This chart shows clearly that April (blue) was an unusual month, and streetcar travel times are higher than for previous months. The May (green) data is for buses which are slightly faster in the evenings, but which lie in the same travel time range as streetcars for the months of 1Q17. The same data can also be shown as a percentage difference relative to the May (bus) data.
Where the values fall below the 0.00% line, the streetcars are faster. As we will see in the detailed charts for April and May below, the advantage varies over the route and by time of day.
The differences westbound are not as striking, but they are still an improvement over the April-to-May comparison.
The moral of the story here is that a data comparison may not be what it seems, and a few weeks’ data are not necessarily representative. For example, if the first part of September were used as a reference, this would be during the traffic mess downtown brought on by TIFF and especially the 504 King diversion. This would not be representative for either mode.
Similarly, the situation under poor weather may not produce the same comparison as under the generally fair weather experienced in May, the base month for the bus comparison here.
Comparison of Link Times April vs May 2017
The following two PDF files contain a set of charts showing the value of link times for various segments of the 501 Queen route.
By way of an introduction, here is one page.
The values shown here are for trips westbound between Silver Birch Ave and Woodbine in April (blue, streetcars) and May (green, buses) 2017. The time on the horizontal scale refers to the hour in which a trip begins (i.e. the point at which a vehicle crossed Silverbirch).
The chart shows that bus trips were slightly faster over this segment, although the degree varies through the day. The standard deviation values are quite close, and the values indicate that most of the travel times varied by only a minute either way from the average. The average is highest for the period beginning at 8:00 am, not unsurprising for an inbound trip in the AM peak.
The descriptions below should be read in conjunction with the sets of charts. Note that the vertical scale is not the same for all charts.
Westbound Travel Times
- Silver Birch to Coxwell: This chart combines the two route segments used in the analysis into one piece embracing all of The Beach. The gap between bus and streetcar sits at roughly a minute or less all day.
- Silver Birch to Yonge: This chart looks at the entire east end of the route. For most of the day, bus times run 4-5 minutes below streetcar times except during the 8:00 AM peak hour.
- Silver Birch to Roncesvalles: This chart reviews the entire route where comparative data exist (there was no streetcar service west of Roncesvalles in April). Buses run slightly faster than streetcars, although some of the improvement seen on the east end of the route is lost when the full width of the line is measured. It is no surprise that for westbound trips, the highest travel times come in the PM peak.
- Woodbine to Coxwell: This is a very short segment at the west end of The Beach where buses make marginally better time, although the values are more dispersed (higher SD values). The big spikes in SD values are probably due to layovers associated with crew changes.
- Coxwell to Greenwood: Again this is a very short segment, and some of the spikiness in the data is probably due to layovers for crew changes. Note that here and in the previous chart, the gap between the modes is about half a minute even though it looks wide because of the small maximum Y value on the chart.
- Coxwell to Parliament: This segment consolidates data for Riverdale and the east side of downtown. The boundary at Parliament is used because this was the eastern limit of a City of Toronto traffic study. As on other parts of the route, the running times are similar for the two modes in the 8:00 hour, with buses running faster at other times. Note that the SD values for bus trips are higher through the AM peak indicating greater variability for buses here.
- Greenwood to West of Broadview: In this segment, buses and streetcars run neck-and-neck through the day with buses getting some advantage in the evening. Again, buses show greater variability than streetcars, although the difference falls off in the evening.
- West of Broadview to Parliament: In this segment, there is a greater drop in bus times as the day goes on, and it is worth noting that the section from River to Parliament tends to have few stops.
- Between Parliament, Jarvis and Yonge (three charts): Buses and streetcars run close to each other over these segments during the daytime.
- Yonge to University: At this point, the relative performance of buses versus streetcars begins to change, particularly in peak hours when the buses are slower.
- Yonge to Bathurst: This takes in a wider slice of the west side of the core with buses and streetcars running close together.
- University to John: This is a short segment, although it can be quite congested. Streetcar and bus times are similar here.
- University to Dufferin: This takes in a long slice of Queen West with buses showing a small advantage, mainly in the evening.
- John to Spadina: This is another congested segment, and streetcars are only slightly slower than buses here.
- Spadina to Bathurst: As we move out of the core, buses pick up a bit of an edge (albeit a very small one).
- Bathurst to Ossington: This area has less congestion, and the buses are slightly quicker.
- Bathurst to Roncesvalles: This is a wider slice embracing all of Queen West. Buses and streetcars show comparable times over this wider section.
- Ossington to Dufferin: Buses are slightly faster in this section.
- Dufferin to Lansdowne: In this section, the stats are roughly even between the modes.
- Lansdowne to Roncesvalles: In this section, the streetcars show a big advantage, although some of the spikiness of the data can be attributed to layovers westbound at Roncesvalles for crew changes.
- Lake Shore/Humber to Long Branch: Data for both months on this segment are for buses which served this part of the route in April and May. It is worth noting that there is a difference in running times between the two months even with the same mode. This shows that care must be taken in making assumptions about small changes.
Overall, the areas where traffic is fairly free-flowing and periods when passenger service times are not too long show the buses coming out better than the streetcars. However, where the street is likely to be congested or when passenger demand is heavy, buses lose their advantage.
Another possible factor, and this from personal observation riding buses on Queen, is that bus operators are generally used to driving at faster speeds and will take advantage of whatever openings there might be in traffic. Streetcar operators are less aggressive, leaving aside issues such as being trapped behind left-turning vehicles. Depending on the degree of overall congestion, turns can also affect buses in locations where general congestion prevents vehicles of either type from advancing in any available lane.
I will return to a more detailed review at the level of individual days later in this article.
Eastbound Travel Times
The eastbound charts are arranged in the same order as the westbound ones (i.e. from east to west), but I will comment on them from west to east. To follow along, go to the last chart and then page backward.
- Long Branch to Lake Shore/Humber: As with the westbound data, there are small differences between April and May times for this section.
- Roncesvalles to Lansdowne: Streetcars have a definite advantage here in the morning, and it declines over the day.
- Roncesvalles to Bathurst: Over the longer segment from Roncesvalles to Bathurst, buses fare slightly better.
- Roncesvalles to Silver Birch: Over the full route, buses are faster, but the amount varies over the course of the day.
- Lansdowne to Dufferin: Buses and streetcars run at comparable speeds until the early evening.
- Dufferin to Ossington: Buses run about half a minute faster through this section which is comparatively uncongested, but also short.
- Dufferin to University: Buses better the streetcar times only in the early evening over this wider view of Queen West.
- Ossington to Bathurst: Times for this segment are roughly equal.
- Bathurst to Spadina: Streetcars have shorter times at mid-day when there is a lot of congestion here, and otherwise the modes are roughly the same.
- Bathurst to Yonge: The two modes are roughly the same with each being slightly faster/slower at different times of the day.
- Spadina to John: In this congested area, streetcars generally have the shorter times.
- John to University: The two modes run neck-and-neck here.
- University to Yonge: Buses are faster over this section in the PM peak.
- Yonge to Jarvis/Parliament: Over both of these sections, buses are slightly faster than streetcars during some periods, equal otherwise.
- Yonge to Silver Birch: For the east end of the route, buses run 2-4 minutes faster than streetcars except for the 17:00 hour of the PM peak.
- Segments east of Yonge tend to be uncongested, and as we saw for the westbound data, the buses tend to run a bit faster than the streetcars with the smallest difference coming in the PM peak.
- Greenwood to Coxwell: As on the westbound trips, times in this segment are affected by crew changes, and so the SD values bounce around more.
- Coxwell to Woodbine: Bus times in the evening here are unusually high. A more detailed review of the data is needed to determine why.
- The Beach: Bus times are very slightly better than streetcars.
Comparison of Headways
The scheduled level of service with buses is considerably better than with streetcars because buses are smaller, and the TTC puts more of them out when they replace service. It is not unusual for riders to “love” the bus replacements because the wait times, an important component in service attractiveness, are shorter.
Any plans to widen headways on streetcar routes needs to take this into account, and offset the effect by providing more reliable service with few gaps and bunches. However, the TTC appears to have abandoned this goal with the adoption of “standards” that are so lax even the most appallingly bad service can meet them.
The following charts are similar in format to the link time charts above, but with the difference that they give the headway at a point, not the travel time over a segment.
A few points are noteworthy here:
- For much of the route, the standard deviation of headways is roughly the same for streetcars and buses, but since the scheduled bus headway is shorter, this actually represents more badly-bunched service.
- As we move westward across the route, the bus SD values converge with the average headways indicating that by about Broadview, most buses are running in pairs. This situation continues all day long and into the evening. For streetcars, the SD values are in the same range, but they tend to stay at least a bit below the average implying that although unevenly spaced, the streetcars are not running nose-to-tail.
- Headways shown between Dufferin and Roncesvalles for April include the 501L shuttle buses, and so this is a mix of streetcar and bus service. (I have not filtered out the 501L service for these locations in April data.)
- From Royal York westward, only the 501L bus service is included. With SD values close or equal to the headway, this shows that even with only half of the service operating through to Long Branch, it tends to do so in packs of two buses.
- Eastbound from Long Branch (last page of the eastbound charts), the SD values are lower, at least through daytime hours when there is some supervision present to dispatch vehicles. The evening is quite another matter.
- As with the westbound data, the eastbound stats include buses for April on the 501L shuttle between Roncesvalles and Dufferin.
- From Ossington eastward, the bus SD values track closely to the average headways indicating that buses are running in pairs all the way across the city.
For the record, the TTC’s service standard for routes operating more frequently than every 10 minutes is that vehicles can be up to 50% above or below their scheduled time (i.e. for a scheduled 6 minute service, vehicles can arrive every 3 to 9 minutes) and up to 40% of all trips can be outside of this range. Although service on Queen is ragged, it generally meets the standard, especially thanks to the large blanket exemption for nearly half of the service.
The situation is likely compounded by the TTC’s tendency not to tightly manage replacement bus services because, after all, the route is under construction and disruptions are inevitable. This really does not hold water for May 2017 because most of the construction projects that will affect the route have not had much of an effect yet. (Track construction on Lake Shore, for example, consists of nothing more than two piles of rail, and no actual track reconstruction had started as of the beginning of June.)
A Note About Weekends
A preliminary review of weekend data shows that buses do not fare as well then especially on the congested western half of the route. Some of the streetcar data I have includes weekends when service was disrupted or diverted, and this must be filtered out before a proper comparison is possible.
Coming in Part II
In the interest of cutting off a long post, I am ending this installment here. A second section will review data at a greater level of detail to see the degree to which “averages” mask wider swings in travel times and headway reliability. Stay tuned!