Updated January 31, 2017 at 12:20 pm:
- Saturday vs Sunday travel speeds
- Detailed bus and streetcar speeds
- Terminal layover times
As part of its TOCore studies, the City of Toronto is contemplating changes to King Street to alter the way it serves many users: cyclists, pedestrians, cars, taxis, delivery vehicles and, of course, transit. Recent media coverage latched on to a scheme to remove at least private automobiles from the street completely. This is only one option, but the focus on the “no cars” scheme, probably the most extreme of possibilities, leads to a polarized debate, hardly the way to launch into a proper study.
The primary beneficiary of a “new” King Street is supposed to be the transit service, but a vital part of any proposals and analysis is the understanding of just how the street and its transit work today.
Recent articles related to this post contain background information that I will only touch on briefly here:
- Travel Times on Route 504 King
- How Much Service Actually Runs on King Street (2)
- How Much Service Actually Runs on King Street
- 504 King: Did Eliminating Car Stops Make Any Difference?
- King Street Service Update: June 2016
- Does More Running Time Improve Service?
The basic premise behind improving transit on King is that with less traffic in the way, streetcars (and buses) on the route will move faster, and this will allow better service to be provided without additional resources (vehicles, operators) that the TTC does not have, nor have budget headroom to operate even if they were available.
This sounds good, but it presumes that a large portion of the route is mired in traffic congestion throughout at least the peak periods, and, therefore, there are substantial “efficiencies” to be had by speeding up the service.
Looking at King from Parliament to The Queensway
The study area runs from River to The Queensway, but there is little congestion east of downtown. Moreover, the King route operated via Parliament and Queen for an extended period, and this makes comparisons to past years difficult east of Parliament.
The article on travel times includes links to four charts showing the evolution of travel times on this part of the route over several years, subdivided by week and hour of the day.
For an explanation of these charts, please see the related article.
Note that these charts do not include data for trips operated by buses. I will turn to a comparison of these operations later in the article, but for long-term analysis they have been omitted because:
- Buses are not present in the schedule for all months over the period for which data are presented. Including them could, if they had substantially different characteristics, create variations in the values that did not reflect changes in street conditions.
- Some bus trips do not necessarily serve all stops, especially on outer parts of the route, and this “semi express” operation does not reflect normal “local” trips.
The best case trip westbound across the city occur first thing in the morning when cars crossing Parliament westbound between 6:00 and 6:30 am arrive at Roncesvalles less than half an hour later. This is thanks to very light competition from other road traffic and low passenger demand. Eastbound trips beginning during this period take slightly longer on average, a bit over half an hour.
The worst case occurs in the PM peak when the same trip can take 45-50 minutes, on average, with a considerable number of trips taking longer (as shown by standard deviation values of 5 minutes or more).
This “best case” cannot be achieved under all-day operating conditions, but the “worst case” does not represent typical conditions either. Somewhere in between is a base value to be aimed at.
The late morning (9:00 am to noon) is probably the best one can hope for – a period of moderate demand and without chronic traffic congestion across the route. Travel times during this period are about 35 minutes. This translates to a one-way time saving of 10 to at best 15 minutes, and then only during the PM peak period where trips are now quite long. Putting this in the context of the route as a whole, a 20 minute round trip saving must be compared to the 149 minute scheduled time in the PM peak.
In the morning, the average peak trip is shorter than the afternoon with a scheduled time of 132 minutes, but the potential time saving is also lower because the average trip between Parliament and Roncesvalles differs from midday times by a lower amount than in the PM peak.
In brief, any pro-transit moves implemented on King will have benefits in certain locations, but other parts of the route would either not be affected or have little scope for improvement. A big saving in one location will help there both for speed and reliability, but the effect across the whole route will be smaller.
Looking at Daily Variations
The weekly averages presented in the charts above masks a considerable variation in day-to-day route behaviour that is affected by several factors:
- Entertainment District activity and other special events
- Short term delays from construction or loading activities
- Collisions and breakdowns
The detailed data for December 2016 appears in the following sets of charts:
- 504_201612_WB ParliamentKing_EastOfQueensway_Monthlinks
These charts show all of the individual data by hour and day (including data for bus trips). Note that the scale for the “average” pages (the last three in each set) is set with a maximum of 50 while the detail pages use a maximum of 100 to show the higher values.
There is a considerable difference in the values on many accounts:
- Early weekdays (e.g. Monday) tend to have much better conditions than days later in the week.
- Thursday December 15 was a snow day starting in the PM peak.
- Wednesday December 21 was a heavy pre-Christmas shopping day triggering congestion at midday.
- Days from Thursday to Saturday tend to have worse congestion during the evening in the Entertainment District.
- Point delays (typically collisions, disabled vehicles, illness) show up as a short spike in values on one day at a specific time. Congestion and other long-lived problems show up as a “wave” of values rising and then falling during peaks on most days unless the factor causing them (e.g. construction) did not stay in place for an extended period.
An important issue that is not yet fully explored is the degree to which these delays are caused by interference with transit vehicle movement (congestion) or by time spend handling heavy passenger loads and boarding delays. The City is reviewing this as part of their study.
Modifying the Street
Changes to the street layout and operation can affect transit in many ways:
- Removing some types of traffic can reduce total demand on the street and improve travel speeds. This can occur through reduced demand for turning movements at intersections, and fewer delays for parking, not to mention collisions that could block transit service.
- Stop service times can be improved by allowing transit vehicles to reach the stop faster (rather than sitting in a queue of traffic) and, possibly, to handle passengers more quickly if they board from extended sidewalks. That design, however, has implications for other potential road users, notably cyclists.
- Traffic signals can be timed to give benefits to transit, but there are “Catch 22s” depending on the local conditions and priorities, such as:
- Transit vehicles can “waste” green time while serving stops. The question is whether having a nearly-certain “green” when a vehicle is ready to leave is worth the side-effects on other traffic.
- Some locations have traffic signals, but no transit stop. If these are in areas where signals are closely spaced, the “look ahead” capability of transit priority may not achieve its goal because transit vehicles are not detected long enough in advance to ensure priority at a location where they would not stop.
- Some locations have regular transit turns left or right, but no signal priority to assist their moves. “Regular” need not be a scheduled move, but rather one that is commonly made thanks to short turns and diversions.
- Some locations have relatively little cross-traffic, and interrupting it for transit on the main street poses no capacity problem. Other locations are very busy, and there can be capacity problems if the cross street does not get enough green time.
- Transit signals can operate to throttle flow where only one vehicle gets through on a single green cycle. This is a particular problem when service is bunched (and generally late), and it is a design issue if service is scheduled more frequently that the traffic signals will permit.
- Removing parking can allow other traffic to flow around transit vehicles and/or provide space for other uses including taxis, commercial loading, or cycling. However, a free-flowing curb lane can encourage motorists to speed past streetcars at stops in an attempt to leapfrog past them and their pesky passengers. The role of parking is different in commercial areas compared with residential ones.
- Expanding pedestrian and cycling space can produce “friction” with transit operations depending on the degree to which the street layout does not clearly set out and limit access to the transit lanes. Queens Quay West is an example of a situation where the distinction between areas for each type of user is very “porous”, and streetcars cannot operate at full speed when the area is busy with pedestrians wandering onto the tracks. This consideration is more important if buses will operate in semi-reserved streetcar lanes because they are not tethered by the tracks and generally require wider lanes (not to mention more alert pedestrians) in close quarters.
These and other considerations must inform any study and pilot program, with the additional caveat that King Street is far from uniform over its length. A “solution” for the financial district at Bay Street will not work among the condos of Liberty Village or the low-rise districts further from the core.
Comparing Weekdays to Sundays
Another way to examine where and when there are potentials for faster transit service is to compare operating speeds for a series of weekdays with “typical” conditions and speeds on Sundays. (For the purpose of this, Boxing Day Monday was treated as a Sunday.)
The charts that follow compare the average speed over the entire route in each direction.
- Each set of charts has 19 pages, one for each hour from 6:00 am to midnight.
- The data show the average speeds of streetcars at each point along the route (at a resolution of 10m increments). The vertical scale is in kilometres/hour.
- The blue lines show the speed for vehicles from Monday, December 5 to Friday, December 9, 2016.
- The orange lines show the speed for vehicles on the four Sundays in December plus Monday, December 26.
- The “East” charts cover from Broadview Station to west of Spadina, while the “West” charts cover from York to Dundas West Station.
- Eastbound charts should be read from right to left, the direction of travel.
The weekday data (blue) sits below the Sunday data at many locations, but certainly not all of the route. In other words, the places where the blue line hangs below the orange are targets for improvement, although getting all the way to Sunday conditions could be a challenge simply because of differences in demand.
Although it is not part of the study area, the section on Broadview Avenue shows effects in both directions (southbound to the core and northbound to the Danforth subway) during the AM peak. A similar pattern shows up on Roncesvalles southbound in the AM peak, and within the study area through Parkdale and points east .
Where the problem is general congestion, travel speeds are lower over the distance between stops. Where it is primarily stop service time, the difference shows up as a sharp downward peak at the stop itself.
There is a particularly striking location of congestion west of Spadina in the evening periods. This is an example of an area that has become extremely busy both with residents and with the restaurants and clubs in the area. Daytime patterns are completely different, and this will be a challenging section of King to redesign while keeping competing groups of users happy.
During peak periods, eastbound traffic to Spadina was also affected by Queen cars making the turn northward on Spadina without benefit of any transit priority. Depending on how badly bunched service might be, this could trigger delays to several streetcars, not to mention other traffic.
Another notable area of congestion lies east of The Queensway where traffic queuing for the left turn holds up transit service, sometimes severely.
Probably the most important aspect of these charts is the degree to which weekday and Sunday travel speeds are very close to each other. This implies that there is little to gain in many locations without changes that will speed up transit vehicles at all times of the day. Equally important, if changes have a negative effect for transit in some locations, the combined effect over the route could be at best a break even situation as occurred on Queens Quay.
Saturdays Versus Sundays (added January 31, 2017)
Saturdays are much less congested on King than Sundays, but where this does occur, notably in the Entertainment District, any new street design should take this into account. The following charts are in the same format as the weekday comparisons in the section above.
Buses Versus Streetcars (Updated January 31)
A question often comes up about the speed of buses operating on King mixed in with the streetcars. The following charts show bus and streetcar data separately for the period of December 5-9, 2016. The streetcar data are the same as in the “Weekday vs Sunday” charts above. Only the periods when buses operate are shown, and only the central part of the route where they overlap with streetcars.
- 504_201612_WB SCarVsBus AM
- 504_201612_WB SCarVsBus PM
- 504_201612_EB SCarVsBus AM
- 504_201612_EB SCarVsBus PM
The buses and streetcars have similar speeds westbound except as they approach The Queensway. Two factors are likely at play here:
- Buses can scoot into the curb lane and bypass left-turning traffic queues, provided the lane is available.
- The buses are near the end of their trips, and therefore lightly loaded. The streetcars by contrast begin to pick up riders destined for points north of Queen as they pass through Parkdale.
Eastbound, the buses manage a faster speed in some locations, but notably not in the most congested areas.
A particular note is that the buses may achieve a higher top speed at times, but this is not sustained for long because they will reach another stopping location.
The following sets of charts compare the average speed of buses and streetcars between Parliament and The Queensway. In each set, the first page shows the average and standard deviation values for each mode on an hourly basis through the day. There is only a small difference between the two modes over most periods and directions with neither having a clear advantage. Eastbound bus trips in the AM peak do show an advantage for the earlier part of the period. The locations where speeds are different can be seen in the charts above.
The remaining pages show granular detail for operations over the month. These are subdivided into pairs of pages with bus and streetcar data, for the AM and PM peak periods, for each of the five weeks of December 2016. The format has been chosen to allow comparison by flipping back and forth between pairs of pages.
Linear trendlines are included as a general guide but they should be viewed with care because the streetcar data has more points especially at the edges of the peak periods. This affects the slope of the trendlines. Curved lines (moving averages or polynomial interpolations) have not been used because these are not well behaved when the number of data points is low and a small number of values away from the average can have a big effect on the shape of the curves.
The point of these detailed charts is to show that the relative “advantage” of one mode over the other is not consistent from day to day or hour to hour, and both are subject to the varying conditions found along King Street. Note that Thursday, December 16 was a snow day in the PM peak.
The Effect of Schedules
A fairly well-known problem to transit riders and streetcar operators is the effect of schedule changes intended to reduce short turns. This has been achieved (to some extent) by giving streetcars more running time for their trips. They are less likely to be late, and therefore less likely to require a short turn. However, this is a mixed blessing because:
- Under benign conditions (good weather, modest loads, no unusual traffic events), streetcars will tend to run early. The typical response, indeed the one enforced by Transit Control, is that they will dawdle and take occasional siestas along the route. This is very frustrating to everyone. (A friend of mine now retired from the TTC claimed that he simply could not drive that slowly.)
- Queues of streetcars pile up at the terminals because they arrive early and have scheduled layovers. It is not unusual to see streetcars waiting on the street to get into a station that is already full of cars waiting to depart.
If the King Street Pilot proceeds and does reduce travel times, this will only make the “running early” problem worse. The benefits may not be fully realized if operators must drag their way across the now-speedy King Street just to stay on schedule. Conversely, if schedules are changed to reduce travel times in anticipation of benefits from the pilot, and these do not materialize, then the service will be chronically late just as it was before the running times were padded.
If a true reduction does materialize, the benefit of any streetcars released will be limited. (A shorter round trip requires fewer cars to provide the same level of service.) The problem is that any headway improvements apply over the entire route (or at least to a major short-working portion such as Dufferin to Parliament), and so a large benefit in a problem area is stretched out over a much longer piece of the route.
For example, if a service now operates every 2’30”, a 130 minute round trip will require 52 vehicles. Saving 10 minutes each way (20 minutes total) is a big challenge, but it would reduce the vehicle requirement only by 8. Factoring the 52 vehicles into the shorter trip would bring the headway down to 2’07”. That’s a reasonable improvement, but it depends on a substantial saving in trip time that could be difficult to achieve.
If the saving is only 5 minutes each way, 10 in total, then the 52 cars would appear every 2’18”. This is a small enough change that it is unlikely to be noticed especially if service remains erratic. The question of irregularly spaced vehicles has been examined many times on this site, and it remains a chronic problem that the TTC has not addressed in spite of many fine words on improved service quality.
Terminal Layovers (Added January 31, 2017)
In an attempt to reduce short turns, the TTC has added running time to the 504 King schedules so that even if they are delayed, they have sufficient time to recover from this for their next trips. However, the offsetting effect is that if trips are not delayed, operators must either dawdle along the route, or they arrive well ahead of schedule at terminals causing backlogs of cars. This is a common sight at both Broadview and Dundas West Stations.
The following charts show the time taken at terminals for the month of December, 2016. These periods are measured from the intersection just south of the stations (Danforth and Broadview in the east, Bloor and Dundas in the west) so that queuing time on street when the loop is full is counted. The lowest values of observed times indicate situations where cars arrived and left as quickly as possible with almost no layover. Many observed values are well beyond the level that would allow a reasonable period at the terminal for a short break for the operator. Where these times are consistently high, this shows an opportunity for recovering time in the schedule. However, these periods do not correspond to the peaks when possible headway reductions could improve capacity.
Redesigning King Street for a new hierarchy of users has many implications for everyone on that street (not to mention the network of streets nearby). Some benefits are possible for transit, but these must be clearly understood and not oversold as if to claim that streetcars will now truly “rocket” across the city.