News   GLOBAL  |  Apr 02, 2020
 11K     0 
News   GLOBAL  |  Apr 01, 2020
 43K     0 
News   GLOBAL  |  Apr 01, 2020
 6.9K     0 
I'm not wrong, they weren't going to keep the Leslie station. The reason why going underground was considered was because it was easier to start the tunnels at Don Mills and did not cost more.
The tunnels were never going to be started at Don Mills. By the time they started that discussion the TBMs were already in the ground, heading east from Keele.
 
The tunnels were never going to be started at Don Mills. By the time they started that discussion the TBMs were already in the ground, heading east from Keele.
Yes they were, there were separate TBM drives starting from each end, meeting at Yonge. The eastern drives had not yet started.
 
Yes they were, there were separate TBM drives starting from each end, meeting at Yonge. The eastern drives had not yet started.
Huh ... I must have forgotten that. I'd dig back and look - but I think that tunnelling may predate the Internet. :)
 
Screwing up traffic at one intersection (Leslie) to get the surface line to operate as well as a subway all the way to Don Mills is a pretty obvious solution, and a reasonable tradeoff. I doubt that TSP would actually be that impactful.

You're right that TSP wouldn't be that impactful, but it also wouldn't come anywhere near guaranteeing a green light for the LRT.

The longest pedestrian crossing (halfway across Eglinton) is 23 metres, so the Flashing Don't Walk needs to be at least 21 seconds.
Capture4.png


Once the Walk light starts along Leslie, you are locked in for:
7 s Walk
21 s Flashing Don't Walk
3 s Amber
4 s All-Red.
= 35 seconds

The maximum green extension is only 30 seconds so even if you could perfectly predict when the LRT would wish to enter the intersection (which you definitely can't at the westbound near side stop), and you can skip the left turn phase, and go directly back to green for Eglinton after it turns yellow, and shorten the side street to the minimum duration (stranding pedestrians in the median), it would still be impossible to guarantee a green unless you allow green extensions well over 40 seconds. In that case we are in fact talking about very major impacts on delay for everyone else at the intersection, especially with trains arriving every 105 seconds on average (every 3.5 min in each direction) - and not evenly spaced between the two directions.

It takes 30+ s for a green along Eglinton, you'd need 20+ seconds of green for the left turn onto Leslie and you need 35+s for pedestrians to get halfway across Eglinton. If you want a green light for those LRVs arriving on average every 105 seconds you need to take drastic actions.

If you want it to operate as well as a subway you need to use rail pre-emption (the same system they use for GO Trains like at Danforth & Midland). In that case you need to detect the LRT at least 40 seconds in advance and hold the light green indefinitely while the train loads at the near side stop. Which might be manageable for a train that runs every 30 minutes in each direction, but not so much for a train that runs every 3.


In contrast, if they had moved the ROW to the south side of the street, the only thing crossing the tracks would be a 7-metre long pedestrian crossing, independent from the main intersection. In that case you can just leave the ped crossing in Walk all the time, and detect trains 14 seconds in advance so they have time to end the pedestrian crossing before they arrive. In that case it's totally practical to guarantee a green, since the only other thing the crossing needs to do is accommodate one pedestrian phase across two tracks, which it could do for literally all of the time there isn't a train approaching. Which is plenty of time for a ped phase with a 6-second Flashing Don't Walk.
 
Last edited:
You're right that TSP wouldn't be that impactful, but it also wouldn't come anywhere near guaranteeing a green light for the LRT.

The longest pedestrian crossing is 23 metres, so the Flashing Don't Walk needs to be at least 21 seconds.
View attachment 707506

Once the Walk light starts along Leslie, you are locked in for:
7 s Walk
21 s Flashing Don't Walk
3 s Amber
4 s All-Red.
= 35 seconds

The maximum green extension is only 30 seconds so even if you could perfectly predict when the LRT would wish to enter the intersection (which you definitely can't at the westbound near side stop), and you can skip the left turn phase, and go directly back to green for Eglinton after it turns yellow, and shorten the side street to the minimum duration (stranding pedestrians in the median), it would still be impossible to guarantee a green unless you allow green extensions well over 40 seconds. In that case we are in fact talking about very major impacts on delay for everyone else at the intersection, especially with trains arriving every 105 seconds on average (every 3.5 min in each direction) - and not evenly spaced between the two directions.

It takes 30+ s for a green along Eglinton, you'd need 20+ seconds of green for the left turn onto Leslie and you need 35+s for pedestrians to get halfway across Eglinton. If you want a green light for those LRVs arriving on average every 105 seconds you need to take drastic actions.

If you want it to operate as well as a subway you need to use rail pre-emption (the same system they use for GO Trains like at Danforth & Midland). In that case you need to detect the LRT at least 40 seconds in advance and hold the light green indefinitely while the train loads at the near side stop. Which might be manageable for a train that runs every 30 minutes in each direction, but not so much for a train that runs every 3.


In contrast, if they had moved the ROW to the south side of the street, the only thing crossing the tracks would be a 7-metre long pedestrian crossing, independent from the main intersection. In that case you can just leave the ped crossing in Walk all the time, and detect trains 14 seconds in advance so they have time to end the pedestrian crossing before they arrive. In that case it's totally practical to guarantee a green, since the only other thing the crossing needs to do is accommodate one pedestrian phase across two tracks, which it could do for literally all of the time there isn't a train approaching. Which is plenty of time for a ped phase with a 6-second Flashing Don't Walk.
With a near-side stop for WB LRT, is the system capable of predicting the WB LRT arrival to the stop in advance, so that the traffic signal optimizes for left-turn movements during the WB LRT dwell time? (≈35s for either SBL or EBL).

In the worst-case scenario, maybe it's better to force cars to do U-turns at the intersection just after the portal (except buses and trucks longer than 10m).
 
Last edited:
With a near-side stop for WB LRT, is the system capable of predicting the WB LRT arrival to the stop in advance, so that the traffic signal optimizes for left-turn movements during the WB LRT dwell time? (≈35s for either SBL or EBL).
Toronto's TSP system? Not really, it can't think that far ahead. It doesn't have complex decision making, it basically just tries to get a green ASAP for the direction requesting priority within the constraints that are programmed (e.g. maximum green extension, maximum truncation, whether phases can be inserted/rotated, etc).

If you detect the LRV extremely far in advance you could kind of manually program an operation that tries to shorten the Eglinton phase in situations where there'd probably be enough time to fit another phase between the current green and the LRV's ETA, but that would require a crazy amount of manual non-standard programming that would be a nightmare to maintain due to the complexity. Someone showing up at the intersection to diagnose a malfunction would need to spend all day figuring out all the unusual programming.

Waterloo's TSP system could probably do it, because it natively predicts LRV arrivals much further in advance and gradually adjusts signal timings to align with the estimated vehicle arrival, without the need for tons of custom programming.

It's also worth noting that "Can it serve X phase while it's loading" is not really the right question. It really depends on when in the cycle the LRV arrives. In my answer above I took a generous interpretation of the question: "Can it serve some other phase (TBD which one) while the LRV is loading".

In the worst-case scenario, maybe it's better to force cars to do U-turns at the intersection just after the portal.
The most effective improvement to TSP would actually be to delete the west side crosswalk, since it's the longest crosswalk and it conflicts with the busy eastbound left turn. The east side crossing can get a Walk light at the same time as the eastbound left turn phase. On the east side, the longest individual crossing is only 15 metres, and the crossing across the tracks is only 7. It's actually a very similar setup to the crosswalks across tram lines in the Netherlands (for which I've also programmed TSP systems). In that case, you actually don't need to hold the green anywhere near as long, since it only takes 10 seconds to clear pedestrians off the tracks. Unless you think the LRV is going to depart in the next ~17 seconds, you can just let the light change to red and turn it back to green later when the LRT is about to depart.
 
Last edited:
I would argue the reverse. With so few level crossings, why increase cost by burying the line, especially under a large watercourse? There need not be a speed penalty if there are no intersections to deal with.

Putting the line in the median is arguably unwise here. Designing the line with an assumption that trains would be short-turned short of Don Mills to mitigate intersection delay, if the data established that Don Mills needed the full service, is an egregious design failure.

Screwing up traffic at one intersection (Leslie) to get the surface line to operate as well as a subway all the way to Don Mills is a pretty obvious solution, and a reasonable tradeoff. I doubt that TSP would actually be that impactful.

But subway just for its own sake is wasting money.

- Paul
I would be inclined to agree here, forcing a small trickle of cars wishing to turn left at Eg/Leslie to plan alternate and slightly less convenient routes is 100% a worthy payoff for extending subway-like service to Don Mills. A pedestrian crossing could be retained with full railway gates to let passengers cross between platforms, they do similar in Chicago on the L system.
 
The maximum green extension is only 30 seconds so even if you could perfectly predict when the LRT would wish to enter the intersection
You need proper TSP like you see with ION, not just extended greens.

I've certainly seen traffic lights in Toronto that go to flashing countdown in a lot less than 7 seconds!
 
You need proper TSP like you see with ION, not just extended greens.
Toronto's TSP system is not just extended greens. It can do everything a signal could possibly do to prioritize transit: extend greens, shorten greens, change the order of phases and add additional phases.

The difference between the systems is in the logic used to determine how/when to apply these actions. Like I said in the post you're quoting.

I've certainly seen traffic lights in Toronto that go to flashing countdown in a lot less than 7 seconds!
First of all, no you didn't - the minimum Walk in Toronto is 7 seconds. A signal will literally shut itself down if it detects a Walk duration shorter than that.
Second of all, what does that number prove about signal priority? Just because the countdown started doesn't mean you can give a green light to transit. There are still pedestrians in the intersection!

The issue is the TOTAL minimum duration of a pedestrian phase. That includes:
- Minimum Walk (almost always 7 s, but sometimes 8)
- Flashing Don't Walk (1.2 m/s, sometimes 1.1)
- Pedestrian All-red (at least 4 seconds, but usually equal to the amber+all red of the adjacent vehicle phase)

If you're just clearing pedestrians across the tracks, this is pretty short and it's totally practical to guarantee a green even with a TSP system like Toronto's that only detects trams within the same block (or occasionally an adjacent block). But if you're trying to get pedestrians all the way acros Finch Avenue it could take ages. So if you want to guarantee a green you either need to detect trams minutes in advance (like Waterloo does), do massive green extensions (like some US and Alberta light rail systems do), or cut the crossing up into multiple stages (like the Netherlands and France do).
 
Last edited:
First of all, no you didn't - the minimum Walk in Toronto is 7 seconds.
Pretty sure I have ... my ever failing memory says it was one of the Lakeshore crossings that used to be 2-stage ... and the countdown started in the 30s.

I'll keep my eyes open ...
 
Pretty sure I have ... my ever failing memory says it was one of the Lakeshore crossings that used to be 2-stage ... and the countdown started in the 30s.

I'll keep my eyes open ...
What are you talking about? In your previous post you said the signals "go to flashing countdown in a lot less than 7 seconds". That means that the Walk light was displayed for less than 7 seconds, which is impossible.

Now you're telling me about the duration of the countdown? What does that have to do with the Walk duration?
 
What are you talking about?
I believe we were talking about pedestrian signals.

In your previous post you said the signals "go to flashing countdown in a lot less than 7 seconds".
Yes. I said "I've certainly seen traffic lights in Toronto that go to flashing countdown in a lot less than 7 seconds!"

That means that the Walk light was displayed for less than 7 seconds, which is impossible.
You have made that claim.

I've only seen it once ... perhaps there was a fault. Which happens. I've tried to cross a road on a white pedestrian signal before, dodged a car, and realised it managed to show green in both directions. I thought that was impossible too ... and yes, I reported it straight away to 311.

Now you're telling me about the duration of the countdown?
Yes. Now I'm telling you you about the duration of the countdown.

What does that have to do with the Walk duration?
I'm just providing further details of what I once observed; I was surprised by both the quick procession to flashing, and the length of the flashing. At the time I'd assumed that they jumped quickly to the flashing because it was such a huge amount of time that they needed to do it flashing, having to walk cross 8 lane plus a median. Now I'm wondering if it there was a fault.
 

Back
Top