[My apologies in advance if you are getting tired of reading about the Queen car. In anticipation of the public meeting on this subject on Tuesday, December 4, I am trying to push a lot of material out the door.
[Updated at 7:45 pm December 2: Full-month charts for headways at Woodbine bothways have been added to show the range of values over the entire month.]
This post concerns link times, something I didn’t go into in the previous series on the King Car (I will be adding a post on King link times soon).
First, a bit of background to explain why anyone should care about these charts.
If we break a route up into segments, we can look at the time taken by each vehicle to tavel from “A” to “B”, the beginning and end of the segment. If these times stay fairly steady over an entire day, then it follows that conditions at all times match those at the best of times. In other words there is no improvement to be achieved by relieving “congestion” or any other source of delay unless we can prove that it’s there all day, every day.
A related issue is the degree of scatter in the values. Even if the average stays constant, we could have widely varying individual times for each car. This would indicate something was happening to randomly delay cars over this segment of the line.
In many areas, we will see increases and decreases in the average time, as well as changes in the scatter of times showing conditions as they evolve over the day. Put multiple days’ data together on one chart, and we can see whether there are events on specific days that are out of the ordinary behaviour of the line. Such events cannot reasonably be planned for, although it would be helpful to have a routine strategy to deal with the common types of events (e.g. major events in The Beach, at City Hall, at the CHUM/City building).
This series of posts will look at the line from end to end to review the way each segment actually operated in December 2006.
First, let’s look at the east end of the line from Neville to Yonge.
Each set of charts contains seven pages corresponding to the four blocks of weekdays, all weekdays seen as one set of data, all Saturdays (plus Boxing Day), and all Sundays (plus Christmas Day).
Looking at the Neville to Woodbine link westbound, we can see that for Dec 1-8, the travel times stay in a fairly narrow band averaging 8-9 minutes during the day and dropping slightly in the evening. Moreover, the width of the band containing most data points is about 4 minutes. We can also see specific days and times of the day when the data are scattered more widely around the trend lines.
In week 2 (Dec 11-15), the data are better behaved than in week 1. Although this was a rainy week, for transit, conditions were benign compared to week 1.
In week 3 (Dec 18-22), you can see the effect of the pre-Christmas rush on December 22 where running times almost doubled during midday. We will see this pattern on other links.
In week 4 (Dec 27-29), there is no peak period, and running times are lower than usual except at midday.
Looking at all weekdays on one chart, we can see that under the varying conditions over the month, the majority of the data is concentrated in a band between 6 and 9 minutes, with slight rises in the am and pm peaks.
Saturdays show a very different story. There is a consistent rise starting at 0800 peaking at about 1400 and gradually falling off into the early evening. There is a similar, but not as marked, pattern on Sundays.
In all of these charts there will be occasional outliers such as the high values just after noon on Saturday, December 9. This date was discussed in an earlier post, and these points correspond to congestion and a wide gap in service to the Beach.
The links from Woodbine to Greenwood, and from Greenwood to Broadview show quite flat times on almost all days and times.
From Broadview to Yonge, there is much more scatter, especially on weekdays, and this likely occurs mainly in the stretch from Church to Yonge (very fine-grained link analysis is not possible with CIS data because the raw observations are not closely enough spaced for every car). However, the trend lines for weekdays stay fairly flat. This is also visible in the “all weekdays” chart where the cloud of observations does not move around much with a long arc from early morning to late evening.
Eastbound from Yonge to Broadview, the data in week 1 show the effect of some service delays, notably on December 1. This delay showed up in various ways in the charts for that date discussed in an earlier post. (The near-zero values are an artifact of rogue CIS data that I have not filtered out.) For the other weeks, values in this link are much more settled.
Broadview to Greenwood also shows little variation, although there is an unusually large number of outlier points in week 1.
Greenwood to Woodbine shows a very large amount of scatter. This is due to cars stopping for crew changes at Russell Carhouse. It is worth noting the very large spread in times here showing that significant delays to individual runs are caused by tardy crew changes. The outlying data points are scattered through the day and the main block of data remains at a reasonable level around the trend line. This shows that we are not dealing with persistent traffic congestion that should affect all trips and vary through the day, but with an operating practice that causes very large delays to many runs.
I have not correlated short turn activity with these unusually long trips, but would not be surprised to find that many Woodbine short turns are the direct result of cars leaving Russell Carhouse very late after a long layover. The problem is most common on weekdays, although there are some events shown on the weekend charts.
Calculated times from Woodbine to Neville are not as reliable as other links because for some trips, CIS does not report a car as being “at Neville” until it leaves westbound. For these trips, the terminal layover is included in the eastbound link time. More detailed review of the data is needed to sort out whether the variations in link times are due directly to congestion, or to the availability of longer layovers when travel time over the route as a whole is lower.
Problems with “losing cars” between roughly Wineva and Neville eastbound can be ironed out with the chart of round trip times from Woodbine to Neville. Observations at Woodbine are generally reliable, and the round trip times will include both running time each way and any layover.
The charts show a fair degree of scatter, but most notable is that the longest times show up on weekend evenings. This generally corresponds to periods when layover times at Neville are longer rather than to congestion in the Beach. Conversely, we can also see increased round-trip times on the afternoon of Friday, December 22 when all shopping areas were badly congested. Until we get reliable location data for the eastbound arrivals at Neville Loop, it will be difficult to sort out congestion effects and long layovers.
In Part VI I will turn to service between Yonge and Humber.
Updated December 2, 7:45 pm:
Those who have read the post about the service to Long Branch will know that I concocted a new display of headway data modelled on the link time charts. These charts show the variation in headways for all days and times at one location. Here are the charts for service in the Beach.
Eastbound headways at Woodbine have a wide range of values showing the general disarray of service sent into the Beach. Up to 20 minutes is routine, and over 20 is seen on several occasions except for Christmas Week. The trend line on the charts sits well above the level of the scheduled service in some periods because so many cars are short-turned at Woodbine.
Westbound headways are just as bad. This shows quite graphically that service quality to the Beach is appalling and complaints about irregular, infrequent service are quite justified. The only thing good I have to say is that the service to Long Branch is worse, but both ends of this line are a huge embarrassment.