In previous posts in this series, we have seen where cars on 501 Queen actually go, as well as the gaps and unpredictability of service both downtown and at the outer ends of the line. Now I will turn to the length of time cars take getting from one place on the route to another, and how this varies both over the course of the day, and from day to day through December and January 2007/08.
We hear a lot about traffic congestion and the need for better transit priority. If such a scheme is to benefit riders as a whole, it must address the locations and times when streetcar service is slow. Often this is not the “obvious” time or place– the peak period, downtown — and priority schemes focussed on this narrow time and location will do little to improve service.
This article contains a series of charts that are similar to the headway charts in Part 3. Data are organized into groups by week (for weekdays), Saturdays and Sundays/Holidays. Instead of headways, the times shown are the intervals between a car’s appearance at two locations.
When these times are unvaried and show little scatter, then there is no congestion or variable delay due for stop service, and almost no opportunity to change running times.
When these times vary a lot, but in a predictable way (moving up and down over regular times each day), this shows regular variations in traffic levels and stop service time. Delays caused by traffic signals can be addressed through priority schemes. Delays caused by stop service can be address by increased use of all-door loading. Delays caused by congestion, especially those outside the peak, can be address by traffic restrictions on parking and turns at intersections. These will not be popular in neighbourhoods outside of the core where the main streets are important local commercial strips and the streets are the grid through which drivers access the residential side-streets.
Where these times show unpredictable spikes or move away from a regular pattern, this is the result of some event like a storm, a traffic accident (possibly on a nearby street such as the Gardiner Expressway with a spillover effect), or an unusual rise in traffic (for example around the club district or on New Year’s Eve).
To trace the route’s behaviour, I will move west from Neville to Long Branch, and then return east looking at segments of the Queen line.
As I have discussed before, data in the east end of the Beach can be problematic. On some days, data for cars is missing east of Wineva. We know that a car goes into the Beach, and it comes out again, but what happens while it’s there is mystery. In some cases, we have a departure time from Neville, but no arrival time. This causes the layover to appear as part of the eastbound trip and inflates the actual running time getting to the loop.
To avoid this problem, Woodbine is used as a reliable reference point for service between the Beach and downtown.
Westbound from Neville to Long Branch
Cars that do report a departure from Neville give us some picture of running times through the Beach westbound to Woodbine Avenue.
On Monday, December 3, a long delay blocked service westbound from Neville between roughly 0700 and 0750. This caused the very long link times shown in the graph for that date.
On Monday, December 17, the aftereffects of the snowstorm the previous day are clearly visible with longer than usual running times especially in the pm peak. The storm itself causes a huge swing in the trend line on Sunday the 16th.
Running times are fairly consistent on other days throughout December and January except for some pre-Christmas shopping effects.
Running times from Woodbine to Yonge are fairly consistent through both months with the exception of the December 16 snowstorm, pre-Christmas shopping, and slight rises through the daytime on weekends. This area is not subject to chronic congestion or to major loading delays.
Wilson Park is just east of Roncesvalles, and is a reference point used in these analyses to track service at that junction without the disruptive effects of layovers right at the westbound stop. This section of the route is subject to many types of delay from passenger loads and congestion, and from spillover congestion when the Gardiner Expressway is blocked.
January shows the typical pattern for this section without the effects of Christmas and the December 16th snowstorm. There is a gradual rise in link times leading to the pm peak, but a marked increase falling midafternoon on Saturdays.
In December, the charts show several disruptions including a few blockages during the late am peak and midmorning in week 1, a major disruption on Wednesday, December 12 caused by chronic congestion mainly west of Bathurst (likely a Gardiner event), and a great deal of pre-Christmas congestion. We have seen this pattern for two years in a row, and this is a strong argument for special December schedules on 501 Queen.
This section is about as boring as it gets. It covers the line from just east of Roncesvalles, along the private right-of-way on The Queensway, and through Humber Loop to Lake Shore Boulevard. There is a minor ripple visible in the week before Christmas that is probably all caused by extended times around Roncesvalles.
This section covers the line from the Humber Loop exit to Lake Shore all the way out to Long Branch Loop. The running time here is a fairly consistent 20 minutes with some minor fluctuations over the two-month period.
Eastbound from Long Branch to Neville
This section takes us from Long Branch back to the Lake Shore entrance to Humber Loop. As with the outbound trips, the running time is fairly consistent at about 20 minutes, but it tends to be slightly more scattered than for outbound trips, and there are a few disruptions.
On Tuesday, December 4, cars were held inbound on Lake Shore around 1100, and there was another delay inbound at about 1700.
The pre-Christmas week brings longer running times, but the effect is not as marked as in Parkdale.
The Saturday chart for December is interesting because the blue trendline (December 1) covers a period when this part of the route operated with buses. Note that they took considerably longer to make the trip than the streetcars they replaced, although this may be partly due to the leisurely way in which unscheduled replacement services tend to operate.
There is an unusual condition on Saturday, December 24 when at regular intervals there is one outlier with a long running time. Looking at the underlying data, this appears to be a CIS malfunction. The same car (run 15) on every trip shows no layover at Long Branch, but a quite leisurely run eastbound to Kipling. This indicates that there was no CIS data indicating a departure from the loop and so the layover was included in the calculated running time to Humber. The pattern continues all day over a period where at least three different operators would have had the car.
Sunday shows a slight increase in running time during the December 16 snowstorm, but note that there are few data points as much of the service was bogged down elsewhere on the line.
As on the westbound trip, this section of the route is uneventful. A slight increase in running times is caused by the snowstorm on December 16.
The trip through Parkdale into downtown is, like its westbound counterpart, the most congested part of the route. Running times can double during certain periods and on certain days, and this requires special provision in the schedules as I mentioned earlier. The effect is far more pronounced in December than in January.
There is a rise in link times into the pm peak that takes the trend lines up to 40 minutes during weeks 1 and 2, and above this in the week before Christmas. On some days, the trip from Wilson Park to Yonge consumes almost as much time as the scheduled trip from Humber to Neville. Coupled with a similar problem westbound, this completely overwhelms the schedules and much short turning is needed. Saturdays have the same pattern as weekdays, but the peak comes earlier in the afternoon.
The effect of New Year’s Eve in the club district is quite noticeable on the page for week 4.
Link times from Yonge to Woodbine are relatively well-behaved although some days just before Christmas do show major congestion effects.
Times in January are more settled than in December.
As I mentioned earlier, reliable times for cars at Neville are not available on all days and several trips never report an arrival at Neville at all. Such trips don’t appear in this chart. Those trips that do report times at Neville may not report a distinct arrival and departure, and their layovers are counted as part of the eastbound journey.
This accounts for the scatter in values, and for the fact that the times are generally longer in the evening and on weekends when the time for layovers is more generous. This is particularly obvious in January when there are no other effects to soak up time on the eastbound trips.
As I have mentioned before, the overwhelming location for delays on 501 Queen lies between Yonge Street and Roncesvalles, and anyone who rides the line regularly will know this. However, the charts show the magnitude of the problem and the degree of variation in running times that makes scheduling extremely difficult on two counts.
First, there is so much variation, especially during busy seasons like December, and it is spread over such a long distance, that cars passing through this will often be quite late. Any line management based on a schedule is doomed. Headway-based management coupled with extras to pad the round trip times when needed stand a better chance.
Second, since all Queen service passes through this route section, service to the east end and to the Lake Shore neighbourhoods is always at the mercy of whatever happens from Yonge to Parkdale. If the route is split into sections with:
- a Beach service that stays mainly east of Bathurst,
- a Queen service running from Humber to somewhere east of Yonge, and
- a Lake Shore section that stays off Queen all together,
there is much better chance that service on all three sections can be properly managed.