Now that we’ve had a view at what Spadina looks like on the best of days, Christmas, we turn to a sampling of regular weekdays and quite a variation in the character of service.
Generally speaking, the service between King and Bloor is quite frequent, although there are occasional gaps for which there is little obvious reason. A few genuine traffic delays (yes, they do happen even with a right-of-way) show up, and the stairstep pattern we saw at major intersections appears quite regularly showing the delays at these locations.
Friday December 1, 2006
December 1st, as we have seen in analyses of other routes, was not a good day — there was heavy rain and it was cold leading into what would be a snowy weekend. Although other routes showed serious effects from congestion, Spadina is reasonably well-behaved north of King. However, this is due more to the sheer number of cars on the route than on any deliberate management.
Down on Queen’s Quay, we can see that many cars destined for Union actually short-turned at the foot of Spadina with one gap of over half an hour between roughly 17:45 and 18:20 pm at Union. This was not the only long gap on December 1, and more show up in the late evening.
As I mentioned in the first installment, I had thought initially that there was something wrong with my data analysis or with the raw CIS data to yield such inconsistent headways to Union. However, data of the holidays and Sundays are very well-behaved indicating that the basic data collection and charting process is not the problem.
Although Spadina has extremely frequent service, less than a two minute headway at times, the service charts consistently show aggressive short-turning. I can only assume that this is to get cars on time as it does nothing obvious to fill gaps in service northbound from King — there are so many cars that gaps are rare and would simply be normal service headways on any other line.
Cars holding for their place at Spadina Station also show up in these charts. For example, at about 9:25 am there is a light pink line at Spadina Station that takes a very gradual trip out to the street appearing at about 9:50. This car actually sat within the station and the loop, but because CIS does not record these movements, it simply appears to take a long time getting to its first surface observation point. Other similar cases can be seen through the day.
Note also the gap arriving northbound just before 9:25. For some reason, possibly a blockage at Charlotte Loop, several cars went south to Queen’s Quay or to Union that would probably have been expected to turn at King.
At around 16:30, there is a blockage at Harbord holding service bothways. This shows up as extended times to get beyond Harbord in botrh directions.
There are several traces that bounce back and forth vertically. These are “noise” created by CIS’ confusion about the location of a car and the same run is reported as being alternately in two different places on the line in quick succession. It is difficult to filter all of these out of the source data, and I hope that when the TTC moves to GPS-based tracking, this sort of phenomenon will disappear.
Monday, December 4, 2006
December 4 had rain with snow late in the evening, but we have seen on other routes that service was generally well-behaved. This is probably due to it being Monday when there is less casual traffic on the road.
In any event, the service is somewhat better behaved than on December 1, although we still see short-turning at the south end of the route.
Wednesday, December 13, 2006
December 13 is a midweek day with early morning rain, but no other weather to mention thereafter. There is one small delay southbound between Harbord and College at the end of the AM peak, and a few extraordinarily long layovers at Spadina Station through the morning.
A gap opens up northbound from King around 16:30 and this may be related to congestion around the Charlotte loop. Most notable about this gap is that nothing was done to space southbound service from Spadina Station, and the gap bounces back from that location after 17:00. Everything is finally sorted out when this collection of runs returns northbound.
Service to Union is still plagued by short-turns during the day, but these vanish once the PM peak is out of the way.
Friday, December 22, 2006
I included December 22 because on every other route analysis, that day has proved a complete disaster for service. The afternoon peak starts at about noon, and the combination of early rush hour and mid-day parking/turning restrictions made a total mess of routes like King. The weather was rainy off and on.
By contrast, Spadina starts with an extremely well-behaved AM peak, moves through the midday and afternoon with some bunching, but nothing too serious. Even the PM peak, although it has some early short-turning at the south end, has far better service than we normally see on the line.
This is a good example of the benefit of the right-of-way because operating conditions are consistent with little regard for the chaos that might exist in the traffic lanes.
Short Turns
Spadina’s not so little secret is the amount of unscheduled short-turning at the south end of the line and the unreliable service along Queen’s Quay to Union Station. The regulars down at Queen’s Quay tell me that they never wait and it’s always faster to walk down Bay, but I try to have faith. It’s sort of like clapping to save Tinker Bell — if enough of us really believe, then a Spadina car will actually show up. Alas, that rumble coming down the tunnel is a car on the vent grate, not a streetcar.
These charts, one page for each day of the month, show the destinations of all southbound cars from Queen Street on the 510. Because the reference point, Queen, is very near the regular Adelaide/Charlotte/King turnback point, many trips are represented by short lines. Some cars were accurately reported by CIS as going only to Queen’s Quay Loop, while others get to about York Street (in the data) before CIS wakes up and discovers that they are really northbound on Spadina. The lines that make it up to the top of the data are the trips that actually went to Union.
Thick dark lines are for two or more runs that run almost together and which map as a single bar.
In this chart, you can see that on some days there was a great deal of short-turning at Queen’s Quay and some major gaps in the service to Union Station. This is a classic example of how the TTC mismanages service and forgets that there are riders on the “outer” part of lines. We’re supposed to be showing the flag for LRT on the Harbourfront services, but you would never know it from the service to Union on the 510.
Note that on Saturday, December 16, there is a gap in data between 13:00 and 14:00. This is caused by a CIS outage where there is no source data for the period. Similarly, in the early afternoon of December 29, there is a gap in the CIS data.
In the next chapters of this analysis, I will turn to weekend service, headway reliability and link times. Also, I plan a quick review of service on 509 Harbourfront and a look at the combined 509/510 service at Union Station.
Steve Munro 
Thanks for all your work on this, Steve. I honestly don’t know where you get the data or the time to crunch it all. Either way, your work shows us some of the problems the TTC has, and how easy it would be to fix.
Streetcars specifically are not easy to run on a schedule, but rather are easier to run as a headway. Anything on rails that runs less then 10 minutes apart is the same, take the subway for example, most riders likely dont even know that they actually do run on a schedule, because all of the public information just details the headway. If the TTC wants to run buses though the downtown, then let them run buses. If they want to run LRT, then run LRT, but running a bus based service schedule with LRT based vehicles is just not going to work.
Steve: The data comes from the TTC’s Communications and Information System (CIS), the vehicle monitoring system that is now a few decades old, and I was given the info by the TTC. This shows the much more collegial relationship I have with them these days even though some of the output from my analysis isn’t quite what they had hoped to see. Now that streetcar operations have been merged with subway ops, they are starting to look at headway-based management.
LikeLike
Looking at the Dec 25 short turn chart and seeing how the line could/should work is quite surprising. I’ve had the misfortune several times of waiting at the Union loop for an eternity for a 510. It’s sad to know that it doesn’t have to be that way. I don’t expect perfect schedule adherance like Dec. 25 all the time, but your charts make it clear that the short turning is far too agressive at King and Queen’s Quay.
On your short-turn charts, the legend shows 185 as the Queen’s Quay short turn, but on the charts it looks like that is supposed to be at 135.
Steve: Thanks for pointing out this nit. I will replace the charts later today when I have access to the source tables and can regenerate the PDF.
Updated Feb 23: I finally got around to regenerating the PDF with the correct legend.
LikeLike
But the subway operates on a time-based schedule also, and it always has.
Let me be devil’s advocate here — on the one hand you favor semi-exclusive ROWs that are vulnerable to delays at intersections, and then on the other, you have very unrealistic expectations about schedule adherence on these ROWs. Semi-exclusive ROWs and perfect schedule adherence don’t go together.
A time-based schedule of some kind will always be needed, because people are driving these trains, and they want to go home on time. Even if we ignore the human factor, do you expect a streetcar will never have bad luck on a run and hit a string of reds, forcing a short-turn?
If these ROWs were exclusive, the chances that a short-turn would be needed would be almost zero. How many short-turns happen in the subway these days? Very few.
Steve: First off, the variations on Spadina are primarily due to two factors:
delays at signals
variation in loading times
This is further complicated by the fact that the signal spacing on Spadina is extremely close, and the signals are not set up optimally for transit priority. There are probably two or three more cars in service than are actually needed simply to allow for the delays at traffic lights.
Loading delays will always exist, but for new lines with low-floor all-door loading, we won’t have as large a variation in dwell times during periods when there are many people at major stops. Even the subway has variations in dwell times at major stations, but not nearly as bad as if everyone got on at the front door, climbed up three steps and showed their transfer to the driver.
In the stats I have reviewed, actual delays (as in service blockages) at intersections are rare. I pointed out one in the data just to show people what it looked like, and can assure you that it’s a rarity.
As for time-based scheduling, there is nothing preventing the TTC and the ATU from entering into an agreement that says “I will work 8 hours per day, plus or minus 15 minutes” on the understanding that under “normal” circumstances, crews would end early. This would give up to half an hour’s padding to deal with an entire line running late. This “overhead” in payroll would be part of the cost of doing business.
On the subway, “short turns” are often performed by swapping crews between trains. The crew gets back on time, but the riders are not turfed off, and operationally it’s much simpler than taking trains into pocket tracks or across crossovers. At times, the whole line just runs late. As long as everything is in roughly the right place, crew changes and breaks still work properly, just later than scheduled.
This is an issue for the TTC and the ATU to haggle out, but the end result should be something that reduces annoyance to riders and, through that, the potential for hard feelings between riders and operators.
LikeLike
You mentioned how subway crews are swapped rather than having the passengers turfed off. I’ve actually been one at least one streetcar ride in my day where they swapped operators rather than dump us passengers. Obviously we were running late though I don’t know the reason but the operator was whipping that puppy down South on Broadview like you wouldn’t believe until some clown driving either a Cavalier or a Celebrity pooped my party by being in the same lane as the tracks.
Steve: On street change-overs are actually part of the schedule on routes that operate from both Roncesvalles and Russell carhouses. Some runs start at one but end up at the other, and operators between a pair of such cars will swap so that at the end of the day the cars run in with an operator from the correct division. This allows a car to make an odd number of half-trips rather than an even one to wind up back in the east or west end as appropriate.
From time to time, operators will miss their changeovers, and then you will see a short-turn in the middle of downtown to put the car (and the operator) in the correct place going the opposite direction.
LikeLike