[nfitz]

I never have to wait for multiple cars but the wait times are much more than the scheduled levels. So I feel my complaint was legitimate, and this has been echoed by others (including Steve Munro on Twitter). Splitting the 510 into two branches, one stopping at QQ and the other at Union should be a good compromise; how can this be communicated to the TTC is my next question.

They should indeed meet scheduled levels.

Though I'd give them a bit of a pass until they get 3 or 4 more Flexities in service, so that they can end the QQ short-turning by the ALRVs, and see how this new schedule (it's only 3-days old - I don't think there'd been a major 510 AM peak change before that for many months) works.

Perhaps extending to Queens Quay at all times is the solution - though tough to do given how many streetcars they are current short of for peak service.

 

[DSC]

They should indeed meet scheduled levels.

Though I'd give them a bit of a pass until they get 3 or 4 more Flexities in service, so that they can end the QQ short-turning by the ALRVs, and see how this new schedule (it's only 3-days old - I don't think there'd been a major 510 AM peak change before that for many months) works.

Perhaps extending to Queens Quay at all times is the solution - though tough to do given how many streetcars they are current short of for peak service.

I think the problem is that the QQ / Spadina intersection is a real bottle-neck and they do not want to run too many cars through it.

 

[Napoleon]

Ever since this short turn was re-introduced a few weeks ago, it has been faster for me to walk from Bremner to King (a 12 minute walk), then wait less than 30 seconds to catch a car, than to wait for a car to show up at Bremner.
I don't care what the schedule says, cars are not coming every six to seven minutes south of King.

Agreed, I've done the walk from Harbourfront to King a number of times in the past few weeks on a weekend because service both because there were so few cars running to QQ or Union and because the few that were doing that route were completely bunched together. There are times that I don't think I could do a worse job managing the 510 if I tried.

I think the problem is that the QQ / Spadina intersection is a real bottle-neck and they do not want to run too many cars through it.

So they should fix it. I'm sure the right combination of signal priority/timing can make this work.

 

[reaperexpress]

So they should fix it. I'm sure the right combination of signal priority/timing can make this work.

I'm afraid not. Due to the side-of-road arrangement on Queens Quay, streetcars cross pretty much every other movement while making the turn, so they need a dedicated transit phase. That's what they're doing now and it can't be done any other way. There has been some final work going on that will improve efficiency, but mostly for the 509, not the 510.

We could put in more aggressive transit priority there, but it would reduce capacity even further. Given the already-heavy streetcar volumes there, I wouldn't recommend that.

 

[Napoleon]

I'm afraid not. Due to the side-of-road arrangement on Queens Quay, streetcars cross pretty much every other movement while making the turn, so they need a dedicated transit phase. That's what they're doing now and it can't be done any other way. There has been some final work going on that will improve efficiency, but mostly for the 509, not the 510.

We could put in more aggressive transit priority there, but it would reduce capacity even further. Given the already-heavy streetcar volumes there, I wouldn't recommend that.

I remember a really detailed post, possibly by you explaining how this works. I'm wondering what percentage of the transit priority capacity at this intersection is allocated purely to the short-turning of vehicles in the QQ loop. It appears to be a rather signal intensive effort to enter and leave the loop when you're not actually moving passengers. I'm wondering how much better/worse this is than short turning at King. Or is this just terrible design now that the ROW is on the south side.

 

[DSC]

I remember a really detailed post, possibly by you explaining how this works. I'm wondering what percentage of the transit priority capacity at this intersection is allocated purely to the short-turning of vehicles in the QQ loop. It appears to be a rather signal intensive effort to enter and leave the loop when you're not actually moving passengers. I'm wondering how much better/worse this is than short turning at King. Or is this just terrible design now that the ROW is on the south side.

Steve Munro has written a fair bit on transit priority and I think that both the QQ and Charlotte loops cause real problems for everyone! The transit priority does not work well at any of these junctions for either transit or other vehicles. The King/Spadina junction is a particular problem when it is being used as a diversion route and thus sees more transit than usual.

 

[muller877]

I remember a really detailed post, possibly by you explaining how this works. I'm wondering what percentage of the transit priority capacity at this intersection is allocated purely to the short-turning of vehicles in the QQ loop. It appears to be a rather signal intensive effort to enter and leave the loop when you're not actually moving passengers. I'm wondering how much better/worse this is than short turning at King. Or is this just terrible design now that the ROW is on the south side.

I wonder if they just look at the 510 in an insular fashion when determining the priority here. Do they also consider the back-up it causes on the 504?

And do they have some sort of coordination of the short-turns based on real-time ordering of vehicles? (i.e. at Spadina the driver is told by transit control if they should short turn or not...based on whom was short-turned before which would allow them proper spacing at the south end of the line)

 

[DSC]

I wonder if they just look at the 510 in an insular fashion when determining the priority here. Do they also consider the back-up it causes on the 504?

And do they have some sort of coordination of the short-turns based on real-time ordering of vehicles? (i.e. at Spadina the driver is told by transit control if they should short turn or not...based on whom was short-turned before which would allow them proper spacing at the south end of the line)

There are two types of short-turn. Those that are planned, like Spadina cars turning back north at King and cars short-turned because they are trying to fill gaps in service or break-up 'caravans'. They are, apparently, trying to eliminate the 'surprise' short-turns by better scheduling and management. Steve Munro has just published an article with an interesting table of numbers short-turns. http://stevemunro.ca/2016/01/04/does-more-running-time-improve-service/#more-14188 Presumably this shows the 'surprise' ones NOT the planned ones which are really NOT short-turns as they are scheduled.

 

[reaperexpress]

In my opinion, our best chance at providing reliable service along Spadina is using conditional transit signal priority.

With the extremely frequent service, hitting a red signal can cause a delay nearly as large as a headway (as Steve Munro constantly points out). With such a tiny margin of error and such heavy passenger volumes, there's no practical way for a driver to combat the forces which create bunching. Route supervision might be able to help by holding vehicles at time points to ensure even headways, but that would seriously reduce the already-low average speed.

Traffic signals could be used as an anti-bunching tool by providing a different level of priority to streetcars depending on their headway. Streetcars with headways well below average would get no priority, streetcars with average headways would get the same priority as today, and streetcars with long headways would get a higher level of priority.

From the perspective of other road users, this approach would make no difference. The increased impact from late streetcars getting higher priority than today would be offset by the reduced impact of early streetcars that no longer get priority.

Streetcar average speed would probably improve slightly as a result of priority being allocated. Streetcars with short headways can't make use of priority anyway if there's already a streetcar at the far-side platform, in which case it ends up being a waste of time for everyone. And even when they could have made use of priority, there is the risk that it could result in a late vehicle going the other direction being denied priority.

While it would not make much difference to transit vehicle average speed, the passenger average speed would go way up. The vehicles with long headways are the ones which are absolutely jam packed. It would also sort of compensates for the long wait that many passengers will have had.

 

[Napoleon]

In my opinion, our best chance at providing reliable service along Spadina is using conditional transit signal priority.

Traffic signals could be used as an anti-bunching tool by providing a different level of priority to streetcars depending on their headway. Streetcars with headways well below average would get no priority, streetcars with average headways would get the same priority as today, and streetcars with long headways would get a higher level of priority.

That is an absolutely fantastic idea. I've always been completely stumped by how poorly Spadina service seems to run despite the dedicated ROW. I think part of the problem is the huge number of protected left-turn phases along the route. Letting a bunched vehicle "jump" the left turn phase would certainly help. I can't imagine it would be too hard to implement a pilot project to improve this.

I think the other thing that has to happen is a few stops just have to be removed. The distance between the edge of the platform south of Harbord and the platform north of Wilcocks for a southbound train is only 108 meters, or just over 3 train lengths. Going the other way is even sillier, a mere 80 meters separates the northbound platforms at Harbord and Sussex. 121 meters separate the distance southbound between Dundas and Sullivan. This is just simply inappropriate stop spacing, especially for the new streetcars.

 

[johntauren]

I think the other thing that has to happen is a few stops just have to be removed. The distance between the edge of the platform south of Harbord and the platform north of Wilcocks for a southbound train is only 108 meters, or just over 3 train lengths. Going the other way is even sillier, a mere 80 meters separates the northbound platforms at Harbord and Sussex. 121 meters separate the distance southbound between Dundas and Sullivan. This is just simply inappropriate stop spacing, especially for the new streetcars.

You really can't remove Wilcocks or Sussex stops because then you've got a large gap between the nearest stops (600~m between Harbord and College, 450~m between Harbord and Spadina station), primarily because of Spadina Circle and the loop at the end of the line.

 

[ehlow]

Spadina doesn't seem that bad from Bloor to King

 

[Napoleon]

You really can't remove Wilcocks or Sussex stops because then you've got a large gap between the nearest stops (600~m between Harbord and College, 450~m between Harbord and Spadina station), primarily because of Spadina Circle and the loop at the end of the line.

I'll take the contrary side of that argument:

• Neither Wilcocks nor Sussex are transfer points, so they're only servicing origin/destination traffic. There's a clear argument for convenient transfers.
• Both of them have rather low ridership. I almost never see anyone board northbound at Sussex, as why would anyone wait to board a streetcar to go just one short stop, especially with the frequent backups in the Spadina Loop. Almost every passenger boarding at Wilcox, which also has low ridership is coming from U of T, so why not direct them to either Harbord or College. Simple to do ridership counts and a cost-benefit analysis. If only 1 rider on average is getting on at peak, every second you delay the 250 person capacity of the new streetcar is rather significant.
• Examine the result of stop removal at Sussex...quoting 450 meters is a bit of a red herring. The worst-case scenario in extra walking time for someone starting directly in the middle between the two is 2.7 minutes (assuming 5 km per hour). There's an interesting mathematical question of how best to calculate the average extra time required to walk to a stop after removal. For all passengers originating between Harbord and Spadina the average extra walking time is 20 seconds.
• The closest subway stop spacing in the downtown is 500 meters.

*A bit of a mathematical aside just because I was entertained. When removing a stop to calculate the average extra time required, one simply has to integrate over the distance between stops and the distribution of riders, which is easily assumed to be uniform in space. The subtlety is that only those passengers closer to the stop you're going to remove will have any extra walking time, and that walking time only grows linearly with the distance. For example if, D, is the distance between Harbord and Spadina and we assume Sussex is smack dab in the middle, only those passengers within a quarter of that distance either way from the original stop are inconvenienced. Integrating over those passengers brings in a factor of 1/2. Finally, only 1/2 of the passengers in the full catchment zone are inconvenienced, and the total extra distance to be walked for all passengers is just D/16. Passengers coming from somewhere off of the Spadina line are proportionally inconvenienced even less by virtue of being able to walk along a diagonal (if buildings along it), so that's the worst case.

I just provide this example because people tend to focus on the total distance between two stops and calculate the extra time for some imaginary rider who wants to walk the total distance between them. The absolute worst case is half that number and the average case is much less extreme.

 

[reaperexpress]

That is an absolutely fantastic idea. I've always been completely stumped by how poorly Spadina service seems to run despite the dedicated ROW. I think part of the problem is the huge number of protected left-turn phases along the route. Letting a bunched vehicle "jump" the left turn phase would certainly help. I can't imagine it would be too hard to implement a pilot project to improve this.

Thanks, it's not my idea, it's actually pretty common for transit priority systems. York Region uses it, though in their case I suspect it's more to limit the impact to other traffic than to improve the quality of transit. Which is of course a challenge with implementation. One group might decide that they want to use conditional priority to hold transit performance constant and reduce impact on other traffic, while other groups might prefer to keep impact on other traffic constant while improving transit performance.

The go-to study on conditional priority is Furth & Muller's 2000 study in TRB (pdf).

Skipping the left turn is exactly what I had in mind with the higher priority at minor intersections. At major intersections I just meant "ordinary" priority given that they don't have any currently.

And no it's not particularly difficult to implement. Just a matter of deciding to do it, really.

I think the other thing that has to happen is a few stops just have to be removed. The distance between the edge of the platform south of Harbord and the platform north of Wilcocks for a southbound train is only 108 meters, or just over 3 train lengths. Going the other way is even sillier, a mere 80 meters separates the northbound platforms at Harbord and Sussex. 121 meters separate the distance southbound between Dundas and Sullivan. This is just simply inappropriate stop spacing, especially for the new streetcars.

I'm actually not too bothered by the spacing on Spadina. It's a bit tight, but given the location of the crossing routes we can have stops either too far apart or too close together (based on my opinion of stop spacing).

The one exception is Richmond. That stop drives me nuts.

 

[LNahid2000]

• The closest subway stop spacing in the downtown is 500 meters.

Huh? I can think of many station pairs downtown that are closer than 500m from each other.