Updated May 7, 2010 at 12:10 am: On May 6, the TTC directed staff to report back to the June Commission meeting with further analysis of the Dundas West option for the Long Branch service. With luck, this will give staff a chance to moderate their position and show real tradeoffs between the pros and cons of this proposal.
Updated May 6, 2010 at noon: My presentation to the TTC is linked here for convenience.
Updated May 5, 2010 at 09:50 am: By popular demand, charts sho wing the distribution of headways by length of hadway, time of day and date have been added.
Updated May 4, 2010 at 12:45 pm: Charts have been added showing the proportion of service reaching various locations on the route by date and time of day.
Updated May 3, 2010 at 11:30 am: Comments and charts showing the headways on Lake Shore have been added.
As a follow-up to the question of splitting the Queen car back into separate routes with a dedicated service on Lake Shore Boulevard West, I crunched through the vehicle monitoring data for that route for October and November 2009. These data were obtained to allow an analysis of the Dufferin/Broadview split operation, but I have been pre-occupied with other things and have not published any comments yet on what seemed to be a dead issue.
The analysis here looks at only one aspect of the route, the service on the west end of the 501 Queen line.
October 2009 Destinations
November 2009 Destinations
October 2009 Destinations by Date and Time
November 2009 Destinations by Date and Time
October 2009 Headways Westbound from Humber
October 2009 Headways Westbound at Long Branch
October 2009 Headways Eastbound at Royal York
November 2009 Headways Westbound from Humber
November 2009 Headways Westbound at Long Branch
November 2009 Headways Eastbound at Royal York
October 2009 Headways Distribution WB at Humber (Counts)
October 2009 Headways Distribution WB at Long Branch (Counts)
October 2009 Headways Distribution EB at Royal York (Counts)
October 2009 Headways Distribution WB at Humber (Percentages)
October 2009 Headways Distribution WB at Long Branch (Percentages)
October 2009 Headways Distribution EB at Royal York (Percentages)
How These Charts Were Produced
All streetcars report their position, as measured by GPS units, every 20 seconds to the central monitoring system, CIS. This is the raw data that I receive from the TTC. (Similar data drives the NextBus system, but in real time rather than on an historical basis.)
For analytical purposes, I convert the two-dimensional GPS data to one-dimensional positions with zero at Neville Loop and increments of about 100 per kilometre westward. In effect, I am taking the “wiggly string” that is the Queen route on a map and pulling in out into a straight line. This mapping allows the creation of the time-distance charts familiar to readers of earlier route analyses.
The next step is to select locations on the route as reference points. The one-dimensional data can then be scanned to determine when each car passes each reference point in either direction, and this produces an “as operated” schedule of the service. (Imagine a timetable like GO Transit, but for the Queen car.) This gives headways at each reference point, the times between each point, and a view of how far each car actually went on the line.
Reading the Destination Charts
The charts are arranged with one page per day. Each verticle bar represents a car westbound from Gladstone Ave. (just east of Dufferin). This point was chosen as a reference because, for eastbound service, it is just east of the origin of the Dufferin short-turn loop.
The position of each bar is determined by the time leaving Gladstone, and the length of the bar indicates the last reference point on the line from which the car reported data to CIS. Where bars are thicker, this means two cars left Gladstone very close together. There are a few anomalies.
- A few cars appear to go slightly west of Humber Loop. This is caused by the presence of a reference point in my mapping just beyond the loop. Due to GPS positioning errors, a car may appear to go a bit further west/south than the loop and record data that I pick up for that location.
- A few cars disappear before they get to Kipling. I have a reference point at Royal York, and some Kipling short turns do not report a position far enough west to show up at Kipling on my chart.
Both of these are minor problems affecting a small number of trips. They do not affect the validity of the charts for this analysis.
Reading the Destination by Date and Time Charts
Each set of charts contains seven pages. The first six correspond to 3-hour segments of the day starting at 0600 and running through to 2400. The seventh is an except just for the PM Peak from 1700 to 1900 so that route behaviour during this popular travel time is highlighted.
Each day in the month is one vertical bar, and the colour of each section shows what proportion of the trips reached each destination during the time interval.
The source data are the same as for the destination charts, subdivided by time period.
Reading the Headway Charts
Headways are shown at three locations:
- Just west of Humber Loop on Lake Shore westbound
- Just east of Long Branch Loop westbound
- At Royal York Road eastbound
The first four pages in each case contain data for one week (partial weeks are included with the first or last week of the month as appropriate). Each day has its own colour and trend lines are included.
The fifth and sixth pages show the weekday data separated depending on which schedule was in effect. For October, page 5 shows the days on the “old” schedule, while for November this falls on page 6. The “split schedule” is shown on Page 6 for October and on Page 5 for November.
The seventh and eighth pages show the Saturday and Sunday data. Thanksgiving is included with the October Sundays.
Reading the Headway Distribution Charts
These charts are broken down into seven pages, one for each time period through the day. Note that unlike the destination charts, the Saturdays and Sundays are split off separately.
On the “Counts” charts, each column’s height represents the number of trips. The colour of each section indicates the headways for those trips. For example, westbound at Long Branch on October 1 between 0600 and 0859 there were 17 cars. 3 carried headways of under 5 minutes, a further 6 of 10 minutes, etc.
On the “Percentage” charts, the underlying data are identicle, but they are presented as percentages rather than as absolute counts.
Observations on Destination Charts
A sense of the headway to Long Branch can be obtained by looking only at the bars that reach the top of the charts. The amount of short-turning at Kipling is also easily seen.
Clustering of cars caused by delays is evident by wide gaps between trips at the bottom of a chart, followed by several closely-spaced trips. Until I get into the time-distance charts for the route, I cannot comment on the nature of these gaps which may be caused by actual delays, by congestion, or by poor operation/management on the line. For the purpose of this analysis, it really does not matter because we are concerned only with how reliable, or not, the service is on Lake Shore.
Also visible on the chart is the amount of service short-turning westbound at Roncesvalles. Some short-turns are expected at the end of the AM peak as cars leave service westbound, but that only accounts for a few trips, and for none of the trips later in the day.
Weekends are not a good time to attempt travel on Lake Shore. The scheduled headways are already wider than on weekdays, and short turns have a greater effect. Thanksgiving Day, October 12, is quite well-behaved likely due to less congestion downtown.
Sunday October 18 was the first day of a new schedule period, and service was particularly bad. This is a typical problem when a new set of operators comes onto a route. By contrast, Monday October 19 was the first day of the split route operation, and an astoundingly high proportion of the trips reached Humber and Long Branch. It is possible that there was a deliberate policy of not short-turning anything to see what would happen. This didn’t last long, and short-turning resumes from Tuesday onward, although it does not look as bad as in previous weeks on the standard schedule.
Saturday October 24 was a construction day on The Queensway, but the line was not blocked until after some cars had entered service west of Humber. When the line was severed, these remained as a shuttle service between Humber and Long Branch. They were joined by a few cars that passed through the construction zone in the early afternoon.
Because these trips never passed Gladstone Avenue westbound, they do not appear on the destination charts, but they do show up on the headway charts.
October 31 brought an all-day blockage of service due to an accident on Lake Shore. In this case, the bus replacement trips were not recorded by CIS and, therefore, nothing appears on either the destination or headway charts during the affected period.
Sunday October 25 and November 1 saw many short turns at Roncesvalles and at Kipling. Weekend schedules were not affected by the trial route split, and the pattern seen earlier in October continues through to the end of November.
Through the week of October 26-30, service on the west end of the line is fairly well-behaved. This weekday pattern continues, although some change can be seen after the standard schedules come back into operation on Monday November 23.
Observations on Headway Charts
The data at Long Branch show the effect of short-turns at Kipling both in the position of the trend line compared to the one at Humber, and in the wider range of headway values.
The eastbound data show less scatter, but there remain many cars that are travelling very close together inspite of the recovery time provided at Long Branch Loop.
Provided that most of the service actually operates to the point in question, the trend line will lie at roughly the scheduled headway. (The number of cars per hour is roughly what is expected even though the actual headways may vary considerably from the scheduled value.) However, variation in headways (and hence the unreliability of the service) is shown by the scatter of data points around the trend line. Values close to zero indicate cars travelling close together.
Service on Lake Shore is subject to widely varying headways, much in excess of the TTC’s “on time” goal of plus of ±3 minutes. If the whole system operated like this, it would be scandalous, but because everything is treated as an inevitable result of “streetcar operation in mixed traffic” and “congestion on Queen Street downtown”, the problem is ignored.
Observations on the Headway Distribution Charts
The values are all over the place (corresponding to the “clouds” of points on the headway charts discussed above). If most service operated at close to the scheduled headway, there should be a wide, central band in the charts showing most headways in the same value range. A substantial amount of data shows very long headways, including values above 25 minutes.
There is a noticeable drop-off in the number of trips operated during the period from 1500-1759. This corresponds to increased short turning of trips destined for Lake Shore that never get past Humber Loop. Scheduled service is not supposed to drop off until after the PM peak.
I will post comparable information for November when I have a chance to build the charts.
October 2010? I’m still stuck in May 2010! Maybe you mean October 2009 😉
Steve: Ooops! My Tardis needs adjusting!
Steve: Ooops! My Tardis needs adjusting!
Any transit fan knows that those type 40’s always need adjusting. Anyone know of a good supplier of Zeiton-7?
If the goal is to get City Council/TTC management to worry about this, I think some sort of a brief executive summary would be very useful in these posts. The detailed graphs are very technical and would not necessarily mean anything to a non-technical audience. I’m thinking of a bar graph that says:
“100% of streetcars reach Roncesvalles
80% of streetcars reach Humber Loop
40% of streetcars reach Kipling
Only 20% of streetcars reach Long Branch.”
(My numbers are made up.)
Steve: I plan to do a summary presentation for the TTC at their meeting on Thursday. However, one can’t do a summary without first having the details, and I wanted this info to be out to the wider community who are interested in this issue as soon as possible.
The Lakeshore Planning Council had a public workshop this past Saturday and the Transportation break-out session identified the reinstated 507 Lakeshore streetcar to Dundas West Station as one of it’s recommended priorities (the others were completing the bike lane network and a half-hourly GO Train service to Mimico & Long Branch stations Saturdays).
The previous year’s Lakeshore Planning Council public workshop (with a different set of participants) also recommended reinstating the 507 Lakeshore streetcar to Dundas West Station, in addition to improving TTC streetcar line management, and more streetcars serving the west end.
More comments on the ‘The Queen Car is Fine as it Is, Thank You’ thread.
“Weekends are not a good time to attempt travel on Lake Shore. The scheduled headways are already wider than on weekdays, and short turns have a greater effect.”
This could use more explanation. Saturday afternoon and early/late evening all have shorter scheduled headways than comparable M-F times. In fact, Saturday afternoon has the shortest scheduled headway of any period throughout the week. Even Sunday afternoon has shorter headways than M-F midday and PM peak!
The only period when weekend headways are clearly longer than M-F headways are: Saturday early morning; Sunday early morning and morning; and Sunday early evening.
Do you mean that the observed headways are much worse on weekends? If observed service is lousier on weekends than M-F, then *that’s* something to note!
Steve: Yes, the headways are even more scattered on weekends. That’s what I was talking about. I suspect that the reason is that on weekends, the techniques used to “minimize short turning” are not in place, and service is even more chaotic than on weekdays.
Taken from service summary March 28 to May 8, 2010:
AM Peak 10 ‘ 20
Midday 11 ‘ 45 ”
PM Peak 11 ‘ 20 ”
Early evening 13 ‘ 30 ”
Late Evening 20 ‘ 00 ”
Early morning 15 ‘ 00 ”
Morning 12 ‘ 30 ”
Afternoon 8 ‘ 40 ”
Early evening 13 ‘ 00 ”
Late evening 18 ‘ 00 ”
Early morning 20 ‘ 00 ”
Morning 16 ‘ 00 ”
Afternoon 11 ‘ 00 ”
Early evening 18 ‘ 00 ”
Late evening 19 ‘ 00 ”
A quick comparison shows these headways don’t change in the upcoming May 9 to June 19, 2010 schedule period.
Next comment: looking at the data for Humber westbound, the tightest-looking scatter graph looks to be for early November M-F (pg 5). This would be the point where some of the split-route bugs were being worked out, no?
The weekend headways are much harder to read because they’re all mixed up.
What would be interesting would be some kind of standard deviation measure in the headways. I suspect that the standard deviation would be the least for the early November M-F data. Squinting at a plethora of multi-coloured dots helps only so far.
Eastbound at Royal York doesn’t really depend on westbound data, because there’s plenty of recovery time at Long Branch loop that could theoretically fix up all but the worst headway gaps or bunches. Note that I said *theoretically*.
In sum, from what I can see, the split route didn’t deliver disbenefits to riders west of Long Branch, and on weekdays may have produced benefits had the practice been continued and fine-tuned further.
(One additional benefit to riders going west from downtown was a guaranteed empty streetcar at Yonge street.)
Steve: I included the Royal York data to show how service comes east from Long Branch slightly better than it goes west, but it’s still a mess. There is no reliable line management out there. The extremely long 501 route also instills a sense of entitlement for very long layovers at the terminal, and the schedule is a secondary consideration.
I understand that late last year the TTC introduced some sort of procedure for route 501 so that the driver may be short-turned (replaced by another driver) but the streetcar and its passengers would continue on. I think the object was to keep the driver on schedule and avoid overtime costs without inconveniencing passengers.
Did this not reduce short-turns?
Steve: Yes, it reduced short turns according to TTC stats, but did not eliminate them, especially in the west end. The modified crewing was implemented at the east end of the line, and the technique is not as easy to use when the relief point is long way from the terminal (as Roncesvalles Carhouse is from Long Branch).
The only way the TTC is going to solve this issue is to do the following:
1. Get staff out and chat with the riders and get their opinions on how to solve the issues on this route.
2. Turn all 501 QUEEN cars at “Humber”, reinstate the 507 LONG BRANCH and turn them at “Humber” on weekends, with an extension to “Roncesvalles” and rush hour and midday service to “Dundas West Stn.”
3. Convert the entire Queen/Queensway/Lakeshore service and give the route and give it’s private right of way (this thought would be a million to one shot).
This is one issue will never go away>>unfortunately
I wonder why some Streetcar operators drive quite slowly on The Queensway, while others floor it all the way. I’ve noticed that the slow drivers tend to pump the brake at any sign of track curvature/turning. Do the track curves’ designs allow for Streetcars to operate at regular travelling speeds?
Steve: Yes, that track can be driven at quite decent speeds, but some operators just don’t drive streetcars very aggressively.
Ed wonders about a different way of presenting the headway scatter plots. Another way you could show headway variation is to produce bar graphs of the headway distribution (e.g., xx% of headways were between 3 and 4 minutes long, yy% were from 4 to 5 minutes long), or the cumulative distribution (e.g., xx% of headways were 6 minutes or less, yy% were 7 minutes or less etc.) — say the average of all October workdays between 6-9, 9-12, etc. This might paint a clearer picture of the likelihood of experiencing a headway lower than, greater than, or significantly greater than the average — but at the expense of averaging out conditions over a month instead of comparing trends from one week to the next.
Steve: I have thought of doing something like this in the same manner as the bar graphs showing the distribution of destinations (recently added).
Same goes for buses, or any vehicle. The 103 drivers tend to be very aggressive because that schedule has zero slack (wonder if that was partly behind the death of a pedestrian by a turning 103), whereas on some of the main streets such as Bathurst, I have seen operators run the whole route not exceeding 40 km/h.
I agree with the above posts, a separate 507 route is only part of the solution. Vastly improved route management for schedule adherence & terminus departure is necessary, as well as more streetcars in service, to better handle service disruptions.
The reason for the varying speeds on the Queensway pertain mainly to whether the car is late or on time. If we’re late we can make up time on the right of way, if we’re dogging it, it means we’re on time.
A big part of the problem is our schedule, we have too much running time. I drive a Humber car and I have 60 minutes to make that trip no matter what time of day it is. It’s not required at 2 o’clock in the morning for me to have that time. It leads to us having to go ridiculously slow in order to stick to schedule. Any variation of more than +3 sometimes leads to our dreaded “ease back” text message from Cis. and frustration for both us and our customers. I would hate to be a patron downtown later on weeknights as we have to kill 5 or 6 minutes to keep to time. I would love to have a schedulemaker ride with me to see this insanity.
With regards to weekends being spottier in the west I can believe that, all the events and traffic downtown cause more havoc with fewer cars out there. Perfectly believable. I would go for the 507 as a weekend service, to me that is more practical.
Is there a way of determining how the actual data fits with the specified schedule, for example what percentage of cars are within 1 minute of being on time, 1-2 minutes, 2-3 minutes, 3-4 minutes, 4-5 minutes, 5-10 minutes, more then 10 minutes, short turned, disappeared, mechanical problem, etc.
I’m thinking here a car that is 10 minutes early is as bad as one that is 10 minutes late, especially when the scheduled headway is more then 10 minutes..
The biggest issue, is that the TTC really needs this information in real time, so that when the delays start to increase, then they can dispatch a supervisor to deal with it, when it’s relatively small, say an increase from .7 minutes to 1.7 minutes, and it takes little to fix it. Maybe holding one or 2 cars, and telling a couple of others to only stop for passengers getting off, until a certain stop.
I think what happens is that it gets so far off that it takes hours and hours to get it sorted out again. This isn’t solely a streetcar problem. I overheard a supervisor on a bus route, who was dealing with run numbers 6,15,25 following each other and the currently scheduled bus should have been run 35. This was 8:30AM and I would think it probably took until noon to get all the buses in the right order and on time again.
You would also quickly see delays that keep showing up over and over, so you could look for a more permanent solution.
Steve: Being “on time” is of no importance to customers. Being on a reliable headway is. For me to attempt to calculate the schedule divergence of every car would be a huge task and would not produce meaningful results. It is the lack of regularity and predictability in the service that so infuriates riders.
You may have noticed that the new TTC “Next Car” displays don’t say “the next car is five minutes late”, they say “the next car is expected in five minutes”.
You are correct, headway is often important, for frequent routes, however if vehicles are arriving at the scheduled time, they are going to have the proper headway.
The real issue is that the GPS data should be used in real time, by the TTC. Say for example you have a car due at 11:05, 11:10, 11:15, 11:20 (5 minute headway) at a timed stop. The cars arrive at 11:05, 11:08, 11:17 and 11:20, now with this pattern developing, control calls the car at 11:08 and tells them to hold at the next stop for 2 minutes, they call the 11:17 car and tell them to change their destination sign to not in service, and to only stop to drop off passengers, when they see that car is back on time, they call them and tell them to return to regular service. Now your dealing with problems that are minor, before they become major.
This data is no shock to me. Those of us on the Lake Shore are captive to it. Too many times I’ve stood for 20-35 minutes waiting for a 501 on a random weekend. I’m not sure reinstating the 507 will encourage many commuters to take the streetcar, especially if it requires a transfer.
More 508 cars — and more reliability of all cars in general — would be a welcome addition. Despite being fairly early in the path (Royal York), the 508s are almost never on time in the morning. I’ve given up and coughed up the money for a GO train pass after getting to work late way too many times. No point in taking the 508 West in the pm as the backup at Roncesvalles makes the ride well over an hour.
Given the current level of service in the hood, it’s no surprise that locals predominantly rely on cars. If there was regular, reliable, efficient streetcar service, many of these commuters might leave their cars and the GO train in favor of the TTC.
I’ve been looking at the headway charts (as I was one of the “popular requesters”.
The first level of analysis is “what’s okay, and what’s bad?” For most of the time periods, a headway <5 min is *bad*. Except for Saturday afternoon, headways for much of the day the headways are 11 minutes plus or minus a minute. Headways of under five minutes indicate bunching. (Some exception might be made early in the 0600-0859 period when cars are going into service.) Note that even the <10 minute readings can be viewed with doubt: is that 5'30", or 9'30"?
Likewise, a headway of more than 15 minutes is bad for most of the day.
It's possible to flip back and forth between W/B Humber and Long Branch. Some days like Friday Oct. 9 are hit by erratic service to begin with; then there are a number of cars short-turned at Kipling just to make service to the west end even worse. In the 1500-1759 period that day, 65% of headways at Humber were less than 15 minutes; at Long Branch 25% were under 10 minutes, while 75% were over fifteen minutes.
On the other hand, sometimes shorter headways *increase* between Humber and Long Branch, for example Thursday Oct 15th 1200-1459: 19% <5 min at Humber, 30% <5 min at Long Branch.
Some operators like to use Lake Shore as a raceway, pulling into Long Branch well ahead of schedule as shown on CIS (-3, -5, sometimes -8). On the other hand, there are a few operators who go slower on Lake Shore than they do on Queen! I suspect that west of Humber it's "out of sight, out of mind".
You can see speedy operators in particular by flipping back and forth between the 2100-2359 charts. In a time when M-F headways are 13 minutes up to about 10:45 PM, and 20 minutes after than, look how many <5 minute headways pop up for Long Branch. I've been on quite a few of these runs, and I certainly appreciate getting home quickly. People waiting along Lake Shore may not appreciate this, though.
The 1700-1859 charts are interesting. This is rush hour. A headway of over fifteen minutes 1) sucks; 2) will result in a crowded streetcar. At Humber loop, <15 minutes can be expected for about 75% of the runs. 60% seems to be mostly a lower bound for headways 18 minutes (way late).
The problem is that the long headways influence waiting time much more than their count would indicate. As an extreme example, if one streetcar showed up every minute for five minutes, and then there was a 55-minute gap:
— 80% of the headways are 1 minute
— 20% of the headways are 55 minutes
— average headway is 10 minutes
— if you go to the stop at a random time, unfortunately you have a 91% chance of arriving at the stop during the 55-minute gap
Also, the time intervals chosen for analysis could be tweaked, since the 11-or-so-minute “standard” headway is just over the boundary between the 5-10 and 10-15 minute intervals. What might work:
0-8 minutes (early)
8-13 minutes (roughly scheduled headway)
13-18 minutes (late)
>18 minutes (way late)
We’d be looking for most headways to be in the 8-13 minute range, since this is plus or minus a couple of minutes of most scheduled headways.
I know that managing to headways is obviously the way to go if the customer is to be satisfied – I know we’ve discussed before that the surface routes aren’t done this way for a number of reasons – costs, overtime etc.
I wonder if on some of the more problematic lines the TTC could be convinced to do a “test” for a month or two to see just what the difference is (both in terms of cost and quality), I don’t think anyone is expecting the entire system to move from scheduled service, but perhaps the streetcars, or the longer lines could be?
You know you posted your address for the entire world to see eh?
Steve: I am in the phone book, and my address is on many, many documents in the public realm.
I’m not really familiar with the Queen/King/Roncesvalles traffic congestion situation, but is the backup only along King, or Queen as well? If it were only King that was so backed up, then the 508 (and maybe 504?) cars could be rerouted along Dufferin or even Shaw and Queen, but again, I don’t know the area.
Steve: Going via Dufferin is not possible because there are no curves in the SW quadrant of the Queen/Dufferin intersection, and they cannot be added due to grades. Shaw is rather a long way east to be sending cars up to Queen, but in any event, they would then face turns at Queen & Ronces that do not exist if they stay on King. Queen & Ronces needs a transit priority scheme that reacts to the presence of streetcars and adjusts traffic signals to optimize flow. As things now stand, the signals cycle on a fixed time pattern.
Jonathan asks about traffic delays at Queen/King/Roncesvalles.
Westbound King is much worse than westbound Queen. (Currently, the 504 shuttle buses add a bit of delay on Queen.) However, even if Dufferin could be used as you propose, the 508 cars would hit one of the slow spots on Queen, through Lansdowne/Jaimeson which is an offset intersection with complicated interlinked traffic signals. It would still be faster, but not as much as you’d think.
Steve proposes: “Queen & Ronces needs a transit priority scheme that reacts to the presence of streetcars and adjusts traffic signals to optimize flow. As things now stand, the signals cycle on a fixed time pattern.”
I am not sure this is possible. Queen/King/Roncesvalles is a three-way signal: E/W Queen, followed by separate phases with left turns for northbound King and southbound Roncesvalles (and the right-side pedestrian signal for each direction).
Giving more time to northbound King might help a bit. However, streetcars returning to the carhouse back up already because of switching delays into the carhouse. And cars which turn onto Queen/Queensway on the same phase either prevent streetcars from turning into the yard, or are backed up by streetcars that grab the right-of-way.
Taking away some time from southbound King again might help, but there are plenty of southbound King cars, as well as pedestrian crossing.
There isn’t nearly as much need for a left-turn for southbound Roncesvalles, so maybe the two separate phases can be rolled together into a long green arrow for northbound King, then a shorter green for both directions. However, this doesn’t solve the problem of left-turning traffic on northbound King, including left-turning streetcars, being lined up back to Wilson Park.
The green time for Queen eastbound could be reduced, but this doesn’t buy much. And this will start to back up westbound Queen, which already backs up a bit because of the extened red — not as bad as King, of course.
What’s really needed is a both reconfiguration of the intersection and a better entry to Roncesvalles yard.
Just for clarification, the traffic light cycle at Queen&Queensway/Roncesvalles&King is something like this:
1. West/East green light.
2. North and north-to-west left turn green light.
3. South and south-to-east left turn green light.
(2 and 3 might be in the other, but regardless, this is a complicated light cycle.)
On the King/Queen/Ronces intersection: could we do a transit scramble phase, where any streetcar can go, and they manage it like an all-way stop?
May 7, 2010 at 8:54 pm
“On the King/Queen/Ronces intersection: could we do a transit scramble phase, where any streetcar can go, and they manage it like an all-way stop?”
Have you really thought this out? How does a street car go from King to the Queensway while another goes from Roncesvalles to King or Queen and another goes from the Queensway to Roncesvalles. This would result in total chaos. An ALRV is a little larger than a person and they cannot side step one another.
The answer is, no, I had not thought it out at all – I was just asking! I haven’t been to Roncesvalles since I lived on Fermanagh back in the early 90’s… That interchange has always been a mess but perhaps now the traffic is that much worse. Aside from yard movements, my memory is that the intersection was strictly Queen to Queensway and King to Roncesvalles.
Why not have one of the traffic engineers who undoubtedly read this blog put some times and volumes to the motions and figure out how long an all-way transit phase would have to be?
Any transit-only phases will be useless without further extensive changes to the intersection and Roncesvalles carhouse trackage. Only eastbound Queen would benefit as things stand now, and eastbound streetcars never back up unless there’s an operator change and a transit-only phase is no help there.
In particular, northbound King has mixed traffic on the streetcar tracks, with streetcars and many other vehicles all wanting to make a left turn. A streetcar-only phase is a waste of time as soon as the first non-streetcar hits the head of the line.
There are also two major reasons for backups beyond the intersection feeding backwards:
1) streetcars going into the yard (out-of-service King, Bathurst, Spadina, Harbourfront, etc. cars) must pause at all the switches and often have to wait for the correct switch to be thrown by the attendant
2) cars making the left turn pass to the right of the right-turning streetcars — presto instant conflict
You could take care of 2) by prohibiting vehicular left turns from King to Queensway, but that will just cause other problems, such as all the drivers now going north to Queen via residential streets and backing up Queen.
In any case, unless yard operations speed up, 1) remains a problem.
Furthermore, an “all way scramble phase” for streetcars makes as much sense as an all-way scramble phase for cars at Yonge and Bloor.
Comparing streetcars at Roncesvalles/Queen to cars at Bloor and Yonge is a straw man argument. The numbers don’t add up; the volume of cars, multiple lanes in each direction preclude the functioning as an all-way stop (which is essentially what a scramble for transit might behave as).
I accept that having mixed use lanes allowing automobile left turns would effectively bar the scheme from working, and you’ve convinced me that clearing the switches induces a much greater time constraint than would be acceptable. You’d have to deny left turns for automobiles (in any direction), provide streetcar-only queues at the intersection, and improve the yard operations and switching, and I suppose that’s just too much to ask.
Having westbound 501 Queen streetcars stopping at the north-east corner on Queen Street and westbound 507 Long Branch streetcars stopping at the north-west corner on Roncesvalles, would also be a problem for people looking for the first streetcar going along Queensway. Especially with the 3 phase traffic signals at the interesection.
Would be better if both westbound streetcars could use the north-west corner on the Queensway, on a safety island. However, the tracks going into the Roncesvalles yard would be problematic with the current track configuration.
Steve: Yes, I thought about this, and there’s no easy solution to that problem. One could always use (grin) NextBus to see whether a 507 or 501 was expected to arrive next, but of course, you wouldn’t know whether the 501 was going to short turn at Ronces.