Update: The charts in this post have been updated so that each route has its own colour. Thanks to a reader, Brent, who spotted the problem with rendering them only in B&W.
One of the little myths of TTC schedules is that routes with branches, or streets with overlapping routes, actually have something like “blended” service where some care is taken to even out vehicle spacings.
In some cases, the schedules do make an attempt to do this with identical headways on different services, but after that, the service is pretty much left to its own devices to “blend”. For many years, the 502 and 503 services on Kingston Road had similar but slightly different headways. This would mean that there were large scheduled gaps followed by pairs of cars during periods when the departure times at Bingham were almost in sync. As it happened, this problem was at its worst right at the peak of inbound travel. Poor service by design, and in time this was fixed.
An example of overlapping routes where the blend is troublesome lies on Eglinton Avenue east of Yonge where many services run together: 34 Eglinton East, 54 Lawrence East (with two branches of its own), 100 Flemingdon Park, 56 Leaside, 51 Leslie and 103 Mt. Pleasant North. The 103 doesn’t overlap for long and there is no service on the 51 and 56 at some times (although this will change if the proposed Ridership Growth Strategy full-service standards come into effect in November 2008).
On Eglinton there are three major services, each on its own headway. This causes scheduled bunching and wide gaps. Given the different requirements of each route, this is inevitable, but it’s important to remember that many riders will see packs of buses and wide gaps and wonder just what is going on.
Down on Queen Street, there are three services merged westbound between Kingston Road and the Don River, and two services between the Don and McCaul. It’s not uncommon to see cars from different routes running in pairs, and I started wondering just how frequently this happens.
This is another issue of route management: if cars on the less frequent 502 run back-to-back with a 501 by waiting at the merge point for a 501 to slip in front of them, then the 501 does all of the work on the common section of the route. Ridership figures for the 502 would also be affected by this.
After a bit of data crunching, I produced the following two charts showing how services “blend” on Queen Street and on Kingston Road.
Combined Service Westbound at Greenwood
Combined Service Eastbound at University
These charts use a similar layout to the destination charts presented earlier in this series, but in this case, the vertical axis shows which route a vehicle is on rather than its specific destination. Queen cars (dark blue) have a value of “1”, Downtowners (red) “2” and Kingston Road Trippers (green) “3”.
Wherever two cars are running close together, the vertical lines are thicker. This has various effects depending on the combination of routes involved.
- If this represents two (or more) Queen cars, the thick bar will end at the “1” level. If this is a Queen and a Downtowner, there will be a fat line to “1” and a thinner line to “2”.
- Where Downtowners and Kingston Road cars run together, there will be a fat line between “1” and “2” then a thin one above “2”.
- A pair of Queen and a Kingston Road cars will show up as a fat line to “1” and a thin line to “3”.
[Apologies to my readers who write in with suggestions about beautifying these charts. I wrestled with how to display this data in a way that would make bunching immediately obvious, and this is what I came up with. If anyone has a better idea, please let me know via a comment.]
The charts for Greenwood Westbound will include all of the service originating from Russell Carhouse eastward. From this, we can see that bunching does occur, although it varies from day to day. The first few days of January had problems left over from the New Year’s Day snowfall, but the situation is not very pretty even later in the month when things were more or less back to normal.
The 502/503 service does not blend well on many days in the AM peak (when it matters), but Thursday January 10 was a really bad example with almost every pair of trips running together rather than on the scheduled 6 minute headway. There is no reason to explain this situation beyond a complete lack of line management. Through the midday, 501/502 pairs are common although their numbers vary from day to day.
Eastbound at University, there are only 501 and 502 cars (the 503s are down on King), and we are just east of McCaul Street where the 502s merge into the 501 stream coming east. Pairs of cars are evident here too and of particular note are cases where there is a gap inbound on 501 Queen which might have been filled by a 502 Downtowner, but instead it came out together with (and likely behind) the 501 car.
This doesn’t happen every time, but it’s frequent enough that I can conclude there is no attempt to use the Downtowners as “extras” to fill holes in the Queen service at least over the common portion of the route. Moreover, if the Queen car has to carry the gap (rather than an empty Downtowner from McCaul) it will get even more behind schedule on its trip east.
Back in the days when both the Queen and Downtowner cars ran every few minutes (that’s actually long enough ago that the route name was “Kingston Road” and numbers hadn’t yet been assigned to streetcar routes), managing the blend of the various services didn’t matter because they were all so frequent. That’s no longer the case, and the TTC needs to consider whether a supervisor’s time would be well spent actively looking after this sort of problem.
Steve Munro 
Steve: [Apologies to my readers who write in with suggestions about beautifying these charts. I wrestled with how to display this data in a way that would make bunching immediately obvious, and this is what I came up with. If anyone has a better idea, please let me know via a comment.]
I find these black on gray graphs very visually intuitive Steve (aside from looking like UPC codes) they don’t suffer from a lack of colour as it’s clear the thicker lines = bunching; single, thinner lines = single cars with the gray gaps headway approximations. They clearly illustrate your conclusions, that words alone can only approximate.
Steve: I even tried to get Excel to do coloured bars just to see what they would look like, but, like Henry Ford said, any colour you like as long as it’s black. Colour appears to be reserved for other chart formats that don’t show the data the way I want to present it.
As you say, it’s the patterns and the bunching that are important. Maybe the TTC needs a big UPC scanner on top of the CN Tower to monitor service.
LikeLike
I live on Kingston Road and feel poorly served by the 502, 503, 64 [Main], and 12 [Kingston Road Bus] routes.
On Sundays, the only northbound service to the Danforth, between Coxwell and Victoria Park between 645am and 825am, is the 506 along Gerrard from Coxwell to Main Station. TTC route scheduling is still done as if, it is 1954, when no one worked on Sundays but worked 9 to 5, Monday to Friday with all holidays off.
It’s time for the TTC join the real world where people in this city work every day of the week, 24 hours a day, 365 days a year! THANK YOU, STEVE, FOR ALL OF YOUR HARD WORK!
Steve: A small bow to the cheering audience here.
LikeLike
It always bothered me that streetcars couldn’t pass each other like buses. However, bunching does occur with buses, so that is not an answer. Especially with the bus’s narrow front doors, and people not using the rear doors to exit. It would be nice if there were track sidings at the downtown stops, but illegally parked cars and snow would quickly put an end to that.
What should be available, once every streetcar and bus gets GPS, is a monitor screen that tells the driver what their position is in real time. Is the vehicle the lead in a bunch of 2 or 6? Is the one ahead a short turn or not? Where all the vehicles in a radius are going? Are any of them full? But in theory, that should be a job for the supervisor.
And we do need more streetcars, so that headways can become better.
Steve: The sad part in this is that some of what you propose was part of the early claims and hopes for the existing monitoring system, CIS. However, all that the operations folks at the TTC wanted to do was to get the supervisors off of the street and run the system just like the subway from a video console. Alas, the quality of information was much, much worse than even the simplest of subway mimic boards, and I am still wrestling with its shortcomings in producing these analyses. GPS will improve things, but won’t give the tools that should have been available years ago.
Analytical work such as I have presented was supposed to be part of the original CIS, but the TTC never got around to it because its importance wasn’t understood by those who held the pursestrings.
FYI, I have never been asked to present this material in detail to TTC staff even though the initial analysis of 504 King was published a year ago.
LikeLike
If we had revenue service on one-way streets, then passing sidings would be feasible. I think that when we get streetcars coming long distances from Transit City into the downtown core (that debate aside, lets assume it will happen) the best possible solution is to use one-way roads (Richmond/Adelaide, Wellington) and allow passing when multiple routes have to traverse the same stretch.
Steve: No, I doubt you will see Transit City routes running on street through downtown. In any event, one-way streets are far from ideal for pedestrian flows and we certainly should not be encouraging them. Indeed, I would prefer a study of changing the Richmond/Adelaide pair back to two-way, although their perceived role as DVP offramps likely prevents this from happening. It’s ironic that the biggest source of congestion on both street is parking and deliveries.
As for Wellington, it’s not long enough to be bothered with.
Congestion will remain with us on many streets where there is no parallel alternative as you would suggest, and what really needs to happen is that better transit priority be implemented where and when it is actually needed. This was the big flaw with the TTC’s King Street plan — it tried to take over a street where transit service is not particularly frequent except in the peak period, while ignoring the basic issue that King Street should be to move cars, not to store them in the parking lanes. To my mind, moving the cabstands to the parallel streets would make a huge contribution to transit on the main streetcar streets.
LikeLike
Re: coloured bars in Excel — I think what is happening is that the chart is displaying a coloured bar with a black outline, except that the bar is so thin that the outline is all you see. If you set the data series format to remove the border, you should be able to get the bars in colour.
Steve: This fixed the problem. Many thanks.
LikeLike
About the use of one-way streets: I was impressed with the way it is implemented in Denver…
Imagine the south curb lane on Richmond for eastbound streetcars and the north curb lane on Adelaide for westbound streetcars. Placing streetcars in a curb lane moving counter to the car traffic on the street has two advantages. First, car traffic does not get in the way, and second, claims have been made that counter-flow LRVs/streetcars result in fewer accidents. The claim is that the LRVs/streetcars are more visible to drivers and their presence does not sink into drivers’ background awareness.
Perhaps this might play a part in improving downtown streetcar flow.
Steve: Have you looked at the amount of parking and deliveries on Richmond and Adelaide? Curb flow (regardless of the direction) for transit is simply not going to happen. Moreover, with this pair of roads seen as part of the DVP ramp structure, there will be pressure to improve flow on these streets for cars and trucks.
LikeLike
Steve wrote about Richmond and Adelaide, “Moreover, with this pair of roads seen as part of the DVP ramp structure, there will be pressure to improve flow on these streets for cars and trucks.”
After I submitted my comment, it occurred to me that if the Gardiner is removed from the DVP to Jarvis, it will all but seal the permanence of one-way traffic and the need for all lanes for cars and trucks.
LikeLike
How did you define bunching? It seems you didn’t compare the observed frequency of bunching to an expected frequency (correct me if I’m wrong). Do you plan to?
Steve: When two cars are for all practical purposes running as a train, I would call that bunching. The shortest scheduled headway on Queen is about 5’30” and that leaves lots of room for any 502 or 503 cars to fit in with a nice space rather than running immediately behind the leader and, likely, doing very little work at all. There are cases in my charts where pairs of 502 and 503 cars, which are supposed to be at least 6 minutes apart, left together on a 12 minute headway. That’s a bunch.
LikeLike
Blending the service only makes sence, but IMHO the city should replace the 503 with a bus, as well as extend the 502 to bathurst station, and replace it with a bus.
Once transit city gets rolling, and miller gets done knocking down various highways, the demand will return and we can re-introduce streetcars in the form of our new LRV’s onto the route.
Steve: Running the 502 to Bathurst Station has been tried before and it was a disaster. Most of the cars never made it past Wolesley Loop. This extension pushes the 502 through a congested part of Queen, adds the requirement to turn at Bathurst bothways, and puts the 502 cars behind 511 Bathurst cars that inevitably take long layovers at the subway.
Combining the 502 and 503 services to one destination downtown will provide a tolerable, if not spectacular, headway for outbound passengers.
LikeLike
I’m not familiar with this type of data, but I’d suggest comparing variances in wait times between the combined routes and a theoretical variance derived by assuming the cars will be equally spaced (F test for heterogeneity of variance). That may sound supererogatory but the effect size might really reinforce the case. Free consulting advice — which may be no good because I’m not familiar with this type of data. I got all the bases covered.
Keep up the good work. Another street with blending and bunching is Eglinton West. Try waiting for a bus there in the evening in the depths of winter.
Steve: I try to keep the measures I am publishing here at a level that the average reader who is not trained in statistical analysis would understand. If I go to a public meeting and say “look at this, your cars are all running in pairs” on this chart, that’s easy to grasp and hard to refute. If I say that the coefficient of some frabjous thingy is 8.34, nobody (especially the TTC) will have the faintest idea of what that means. Even something as comparatively simple as a standard deviation is an abstract value. It will measure the spread of those clouds of data points, but the picture tells the same story so much better.
As for Eglinton West, it like other routes (54 Lawrence East comes to mind) suffers from having branches with uneven running times. The inbound service does not merge properly. Last night on Eglinton East west of Bayview Avenue just after 10 pm, there were five, count them, five inbound buses more or less in a pack, each from its own route whose schedule bears no relationship to the others. I got lucky and caught the last one in the pack.
LikeLike
If the 502 and 503 start and end at the same place, then why not just make them one route?
Steve: Nick, there are times you try my patience. Of course they would be one route, and I don’t care what it’s called.
And in terms of a downtown destination, why not fix the damn (excuse my french) tracks on Adelaide and Richmond, and loop them similar to the express buses that use these streets? For that sake, why not run the 508 here too.
Steve: Adelaide and Richmond will be repaired over the next few years according to the current plans (which I have already published).
No, we won’t want to loop the 502.5 or 508 services there. People waiting for an outbound car need the option of taking whatever shows up first because many riders have alternate ways of getting to their destination. Put the 508 on its once every day or so headway on Adelaide/Richmond, and nobody will ride it.
LikeLike
I’m not advocating that you publish recondite stuff here. But honest, I’ve found that when you present results and can cite a test for your conclusions, it really helps, not only to persuade people but to give them useful information. Testing also helps establish where the big problems are. But that’s not all! You can do the test I proposed in Excel!! Wow!!! Plus, if you act now….
Im starting to sound like Benny Hinn. “I have a letter from Steve Munro in Toronto. He prayed for streetcar bunching to end on Queen Street, and Queen service was HEE-yulled!”
LikeLike
Further to my last comment, statistics also help you separate the effects of blending from the bunching you would expect on an unblended line. The bunching on 512 before the reconstruction was the norm during rush hour, without blending. Sure, you can see that the problems on 502 probably contribute to the bunching in the blended Queen service, but how much do they contribute? That’s my last comment about statistics.
I do have a comment about line management, though. What line management? On 512 it seems to me the drivers make the decisions that a line manager should be making.
LikeLike
There is a simple fix to the 501/502/503 issue.
The 502/503 service should be changed to bus service, and the buses would be local from Bingham to Kingston and Queen. Buses would than use Lakeshore Blvd and Eastern Ave into downtown Toronto, totally bypassing Queen East and all the stops and traffic.
Not only do riders get a much faster and better ride, but you take away the bunching problem on Queen by having to many routes along one street.
Steve: You assume that everyone on Kingston Road is going to wherever the bus is headed. A basic issue in transit network design is that we cannot and should not provide point-to-point services except where there is a very strong sustained demand. A good example of this is the link from Downsview Station to York University.
If you now force anyone who is coming from someplace other than the bus stops downtown to transfer at Queen and Kingston Road, you must add that time and inconvenience to the “slower” trip on a streetcar on Queen Street.
LikeLike
Actually I have ridden the 502/503 service more than a couple times on my commute back home to Scarborough, and at least over 85% if not more like 90% of the riders get on downtown and do not get off still Kingston and Queen and points north.
That is why there is a 502/503, because people use it to get downtown.
A transfer at Kingston/Queen for the very few who take it to places along Queen is not that big a deal, when the ride is improved for thousands more riders, not to mention the potential for this route to serve South Scarborough residents as well, who want a speedy trip downtown.
LikeLike
It may be a bit of a divergence from the 501/02/03 discussion underway here but I wanted to make some comments about “blended” services in general.
I have done some studying of the service summary reports and always am a bit confused when I start thinking about the “statistics” they cite about the headways, etc. of the blended services. It would seem evident that routes like Eglinton West that have seemingly no common factor (so to speak) in run times of the various branches will end up causing a unpredictable flow of buses from the merge point to the subway station. If this is the case I question the validity or importance of even listing such information in their service summary because it doesn’t even attempt to represent the real thing. Is this a case where a number, any number, even if not actually representative of the situation is better than no number at all?
Do the people in planning realize that there may be ten minute gaps in the early evening on the busiest part of the Eglinton West route or is the output of an effective headway for the blended service from a spreadsheet enough to convince them that the service is acceptable? I am not sure I would expect a whole lot of number crunching being successful or even possible to space the common service from different routes along Eglinton East but couldn’t there be better planning along single, more controllable routes?
And to be real honest I can’t even figure out how they arrive at their numbers for that common part (the blended part) of routes, taking into account the information provided on the Service Summary – round trip times, number of vehicles, distances, targeted headways, etc. Presumably it must be taking into account the length of the common or blended part, which I don’t believe appears on the summary?
And it may be even more out of the scope of this discussion but does planning attempt to fill the short turning buses first before scheduling the ones going the full distance to arrive at the station? As an example, would a route like Finch West that always leaves the subway full be set up so that, say a Jane or Milvan (Weston Road) bus arrived first and gathered up the riders heading between the two points leaving the Kipling or Humberwood riders to board a bus arriving next that was actually heading that far? I ask this because I remember observing replacement bus services along Queen setup this way in the past. Standing at Yonge heading westbound a wave of buses would arrive east, turn around at Church and a bus to Roncesvalles would arrive westbound first, followed shortly by a bus to the South Kingsway area and finally a couple going all the way to Long Branch. I thought this seemed to work well, whether it was intentional or not. Of course plans like this are by no means guaranteed given the unpredictable things that happen over the course of minutes to throw a plan all out of whack but is such an idea like this used at the TTC?
Steve: This is the sort of thing that is possible when vehicles operate “as directed” rather than to a fixed schedule. The TTC has not even included that option formally as part of their Queen car review. Of course, it all assumes a clever Route Supervisor who takes the trouble to organize the service this way.
LikeLike