KCMO Downtown Streetcar

[KCPowercat]

Here's an idea - how about an option where pedestrians must push a button to request the walk light?

I'm in as soon as cars have to do the same.

[DaveKCMO]

I think signal priority for the streetcar is over-blown for such a short route and stop spacing. "Optimization" is how we ended up with one-way freeways downtown.

strictly from the streetcar perspective -- not the automated pedestrian detection -- signal priority is absolutely having a positive effect on every run. you may not have noticed, but a recent schedule change finally allowed 10 minute headways on weekday mornings and evenings.

[grovester]

Headways are important. The trade offs are important. 10 over 11 is not a victory if the pedestrian experience is as described.

[DaveKCMO]

Headways are important. The trade offs are important. 10 over 11 is not a victory if the pedestrian experience is as described.

the pedestrian experience is the way it is because public works believes that giving a walk signal during every signal phase -- which was the case before the auto ped detect system was installed -- would negatively impact streetcar operations. there's no evidence of that.

[KCPowercat]

Headways are important. The trade offs are important. 10 over 11 is not a victory if the pedestrian experience is as described.

the pedestrian experience is the way it is because public works believes that giving a walk signal during every signal phase -- which was the case before the auto ped detect system was installed -- would negatively impact streetcar operations. there's no evidence of that.

okay...so I'm not completely crazy with my rants....who do I have to annoy to give us back our auto-walks?

[grovester]

JFC, what is wrong with those folks? Walk signal every time for some pre-agreed minimum, then allow the streetcar to initiate a change. Why does this have to be so freaking difficult?

[pash]

.

[flyingember]

Similarly, if we had a working real-time tracking system for the streetcar, the lights could be programmed to give a walk sign by default, but switch automatically to using the pedestrian sensors at an intersection when the streetcar is approaching it. ...

Or is that considered genius-level stuff in our smart city?

Look at the perfect no traffic situation with every light green. At 25mph it takes around 5 minutes to go from one end to the other. Union Station to 3rd/Grand is roughly 25 blocks long. So that's 5 block per minute, or 12 seconds per block.

How do we fit a 30 second car/ped signal into this system? The train can go two blocks in the time it takes the ped cycle to complete one block up. You have to know where the train is and stop all ped cycles for the next three blocks (the one closest to the stop and the next two). And you can't immediately flip behind, the green light will be behind the train for 3-4 blocks. With a train coming the other direction too it's possible to have no ped cycle for half the length of the line under this model.

And cars aren't in this model.

The number of total paths of travel downtown is insanely huge. The power of a grid also introduces a huge amount of complexity. There's hundreds of thousands of possible paths downtown just for cars. When you realize thousands of different paths can be in play at once it's an absolutely crazy problem to identify how to account for the paths. One person not behaving perfectly to the system can cause tiny amounts of delay that ripples around downtown for hours.

It's a problem that's beyond genius level. The best algorithms can't account for the human factor and resulting entropy on the flow of the system. And even self-driving won't help with this for some time because there will always be bikes and for decades cars that don't drive themselves. (it could take 30-40 years just to hit 50% saturation of the tech depending on cost)

[pash]

.

[flyingember]

No, we don't need John von Neumann and an ENIAC to figure this out. If you know where the streetcar is, you can adjust the light cycles at the intersections ahead of it to control cross traffic. I'm not sure where you think tracking every moving thing downtown and predicting their paths plays into it.

If you know where the streetcar is doesn't tell you the entire picture. What we want is this system downtown.

http://www.marc.org/Transportation/Comm ... /About-OGL

Imagine there's one train SB leaving the 12th St stop. We don't care where the streetcar is if there's cars trying to turn left onto Truman Rd S backing up into the SB through lane. (I've been on it when this happened) In that situation changing the lights ahead to green does nothing to help train movement.

The idea isn't to prioritize the streetcar 100% of the time but to control the movement of traffic so you get it out of the way. Holding the train at the stop and let more people get on, hold the left turn light onto Truman, hold the green at Grand and Walnut, hold the red at 14th behind the train, hold the ped cycle, etc. The train is delayed but in the right place rather than sitting in traffic.

But we need this system to be implemented on the grid for 3-5 blocks east and west of Main so it can push traffic away from Main.

Incorporate the on train ped counter historic data averages in too, like at 5pm on a Saturday X people get on/off at stop Y. There should be known coordinating dwell time data. This helps determine dwell time at a stop given real time information from the on train counter. ex. it's near 5pm, 70 people are on the train at the next stop it's likely 3 will get off, 2 on and it needs 30 seconds to do this

So the signal system knows the next stop wait time, it knows how many cars are at what spots and heading what spot. It can auto-adjust the light cycle across 6-8 lights to match this. Changing when the train leaves the stop is way too late, it should know what to do 2-3 blocks before it gets to the stop because 60 seconds to dwell plus 30 seconds to get to the next stop is longer than the light cycle back to green. 15 seconds to the next block is less than the entire cycle for that light.

You're not literally figuring the path of everyone downtown, but indirectly through the movement of traffic. It doesn't need to know there's a pedestrian crossing against a green or know there's a car is going to pull out of a garage. It just needs to see where cars are in the through lane.

Now because it's dynamically adjusting car/train movement it can dynamically adjust ped movement too. It knows there's no worries about adding 5 seconds to a crosswalk signal because it knows it can't change the light to green a block over any quicker.
In that situation where the train is blocked at Truman Rd S can give two long ped cycles on the W and S side of the intersection and flip it to don't walk on the other two to help clear cars. Behind the train stuck at 14th stop it does a long E-W ped cycle.

[KCPowercat]

Just give me a walk signal

[pash]

.

[DaveKCMO]

Similarly, if we had a working real-time tracking system for the streetcar, the lights could be programmed to give a walk sign by default, but switch automatically to using the pedestrian sensors at an intersection when the streetcar is approaching it. ...

Or is that considered genius-level stuff in our smart city?

the problems you see with real-time arrival estimates do not actually affect operations. it's strictly a problem with customer-facing information.

and, no, we do not need OGL. there are multiple systems already in place, it's just matter of programming and policy.

[flyingember]

Similarly, if we had a working real-time tracking system for the streetcar, the lights could be programmed to give a walk sign by default, but switch automatically to using the pedestrian sensors at an intersection when the streetcar is approaching it. ...

Or is that considered genius-level stuff in our smart city?

the problems you see with real-time arrival estimates do not actually affect operations. it's strictly a problem with customer-facing information.

and, no, we do not need OGL. there are multiple systems already in place, it's just matter of programming and policy.

When you refer to other systems, do you mean across all of downtown at every streetlight or just on Main? Because the exact tech/system is irrelevant, OGL was just an example.

Improving traffic flow downtown is a key step to doing away with the freeway loop. It's clearly possible to move traffic on a parkway just as easily (see Ward Pkwy) we just don't tune lights so we don't end up with a number of traffic crunches downtown.

The easiest place to tell that we haven't tuned downtown correctly yet is the train should never have a two block long line or cars behind it. Something is off about how we use the grid downtown.

[DaveKCMO]

latest ridership info: http://kcstreetcar.org/wp-content/uploa ... v-2016.pdf

[longviewmo]

latest ridership info: http://kcstreetcar.org/wp-content/uploa ... v-2016.pdf

Page not found?

[grovester]

did that once for me, then loaded after a refresh. (chrome)

[Highlander]

latest ridership info: http://kcstreetcar.org/wp-content/uploa ... v-2016.pdf

Page not found?

So what's the trend? If you read the propaganda put out by Citizens for Responsible Government, streetcar ridership would dwindle to nothing in the winter months. For a person like me who prefers to just walk when the weather is nice, I'd use the streetcar far more in inclement weather. Any signs of usage dropping off?

[chingon]

latest ridership info: http://kcstreetcar.org/wp-content/uploa ... v-2016.pdf

Page not found?

So what's the trend? If you read the propaganda put out by Citizens for Responsible Government, streetcar ridership would dwindle to nothing in the winter months. For a person like me who prefers to just walk when the weather is nice, I'd use the streetcar far more in inclement weather. Any signs of usage dropping off?

Yael is crowing that usage has "fallen 40%".

[DaveKCMO]

everyone predicted ridership would fall off when it got cold. kansas citians cower in their homes at the first sign of cold weather. then they eventually come out again after winter settles in.

of course, yael picked the highest ridership month (july) to compare to the lowest (november). it's still well above the estimated daily average (2,700). it'll take more than six months to be confident in any trend. by comparison, KCATA sets ridership goals on an annual basis.