|
SticksEightySix posted:One of the major motorways here in sydney (M4 for any locals) is a nightmare during peak hour like most major roads, but one thing i have never understood is the fact that most of it is 3 lanes, but at one point it narrows into 2 lanes for the space of about 200 metres before an onramp, becoming 4 lanes after the 2 onramp lanes join the motorway. I can't find the spot you're talking about, could you provide a picture or a map link? It sounds like the cause could be an underpass that's too narrow or an overpass that's too short, providing a temporary width constraint on the motorway. ----- My Least Favorite Interchange You know that, if someone looks at interchanges for a living, the one he hates the most has to be REALLY bad. Let's have a look at it. Ladies and gentlemen, allow me to introduce you to The Olneyville Bypass. It's Rhode Island's oldest freeway (which isn't saying a lot, but it goes a long way toward explaining why it's so messed up.) Now, where to start? Let's begin with the Bypass' original configuration.  (From Wikipedia) What do we have here? Let's see what rules of interchange design this breaks: - Left entrances and exits all over. - More than one exit per mile. - Blind merges. - Insane amounts of weaving. - At-grade intersections at acute angles. - Super low design speeds. - No shoulders. - No acceleration or deceleration lanes. - Every lane is an "Exit Only" lane. Well, that's not very good, is it? Unfortunately, when the expressway was extended north, south, and west of here, it only got worse. Here is the current situation:  So now we have additional ramps, and an incomplete interchange with no access from RI 10 NB to US 6 WB. This is a pretty big deal because it sends traffic from Cranston and southern Providence heading westward through downtown Providence. I had to drive through this thing twice this weekend, so my hatred has only intensified. I'm trying to channel that into some creativity, so here is what we in the biz call a Straight Line Diagram to demonstrate the Bypass' horrible lane balance. Go through it a couple times. Try to stay on Route 6 with as few lane changes as possible. It's like a game you can play twice a day as you struggle through the 5-mile-long backups on the way to and from work! 
|
#
?
Nov 29, 2009 17:35
|
|
|
|
| # ? May 17, 2024 14:45 |
|
|
Cichlidae posted:
I live right near there, and merge onto the overpass right where it splits off into 6 and 10. Of course, I often take Rt 10 South, so I've got to fight across three lanes of traffic (many of whom are fighting to move in the opposite direction) It really is a clusterfuck. It also has some of your favorite "repairs," the wooden support things underneath the overpass. The whole Rt 10 South drive is kind of fun, too. There's so many on and off ramps, I just get in the left lane and stay there until it's time to get off. How would you fix it? It's in such a built up area there's not a lot of room to do anything different, but something need to be done.
|
|
#
?
Nov 29, 2009 19:22
|
|
|
Pagan posted:I live right near there, and merge onto the overpass right where it splits off into 6 and 10. Of course, I often take Rt 10 South, so I've got to fight across three lanes of traffic (many of whom are fighting to move in the opposite direction) RIDOT's got plans, and they mostly involve removing a couple exits and completing the interchange to provide access from RI 10 NB to US 6 WB. Their website is here, at the not-so-auspicious pineapplestudios.com domain. Some left exits and entrances will remain, but Route 10 basically gets a collector-distributor road at the expense of some properties adjacent to the interchange. Hey, it's Olneyville, property values are about as low as they come. Personally, I support what they're trying to do, and think it would be awesome if they connected Route 6 to Route 195 like they'd originally planned to do, even if it means cutting a tunnel under south Providence. Heck, may as well extend 195 west to the state line and bring it all the way to Hartford. Ah, a guy can dream. Of course, it's doubtful that RIDOT will be able to come up with the cash to fix it before the bridges literally fall apart.
|
|
#
?
Nov 29, 2009 19:43
|
|
|
I'm a comp sci student taking an AI class this year, and I've got a major project coming up. (Project: "Do something cool with AI") I thought it might be neat to see if I could use a neural network or genetic algorithm to optimize traffic light timing. After a couple hours in the code, I have cars that queue up, stop at lights, drive straight through intersections, etc. My stoplights, currently, are either east/west or north/south. I'm only a couple hours in. If you don't mind fielding a couple of questions so I can improve my simulation: Are there any bullet-point type rules you follow when designing stoplight timings? I imagine one example might be: Never allow crossing lanes of traffic to both be green simultaneously. How do you measure the success of the timings? I noticed the time/distance diagrams. Would minimizing the horizontal areas be considered a successful benchmark? (I ask this, because the AI technique I'm using needs to know if it has done better or worse every time it runs the simulation.) Assuming two roads of equal traffic volume intersecting, how likely is a driver to want to go straight? Left? Right? (Is this a really hard question?)
|
|
#
?
Nov 29, 2009 20:30
|
|
|
Pfhreak posted:I'm a comp sci student taking an AI class this year, and I've got a major project coming up. (Project: "Do something cool with AI") I thought it might be neat to see if I could use a neural network or genetic algorithm to optimize traffic light timing. That sounds really cool. For individual signals, it's overkill, but the fact that ideal cycle lengths are so easy to calculate will give you a good baseline to test against. quote:After a couple hours in the code, I have cars that queue up, stop at lights, drive straight through intersections, etc. My stoplights, currently, are either east/west or north/south. I'm only a couple hours in. Yes, that's one. Conflicting movements must never have protected greens at the same time. It's ok to allow conflicts sometimes, like permissive left turns or right turns on red. Another thing is to always go green-yellow-red. You'd be amazed how many consultants mess up this simple rule. Have a protected left turn? Green arrow, yellow arrow, red arrow. Overlap right? Same thing. You can't have a green indication that suddenly turns red. Don't forget to put in adequate red clearance, which is a brief (2 seconds, usually) period where all indications are red. All-red clearances should go between each group of conflicting phases to give cars time to clear the intersection. Just keep in mind how the controllers work, and how few parameters they have. Start with something simple like the following: -Red clearance time -Yellow time (depends on the approach speeds, make it so that 10% of cars will run a red light and 10% will stop when they could have made it) -Maximum green time -Phase order Assuming you're going to include detectors, you'll also need the following: -Detector locations -Minimum green time -Vehicle extension time -More phase order quote:How do you measure the success of the timings? I noticed the time/distance diagrams. Would minimizing the horizontal areas be considered a successful benchmark? (I ask this, because the AI technique I'm using needs to know if it has done better or worse every time it runs the simulation.) We generally use the total delay (per vehicle) as a performance measure. It gets a bit complicated beyond that; I'd recommend getting the Highway Capacity Manual if you want to get into the details. If you want to calculate it yourself (or can't afford the HCM), first find how long it would take a car to go through the network if it were the only car on the road, then subtract that from the actual time it took, leaving you with the delay. Integrate for every vehicle that uses the network within a given 15 minutes or hour or whatever your sample period is. You could get a decent approximation on a signal-by-signal basis by taking the % of the time that a given approach has a red light and multiplying by the volume of that movement, and make it better by calculating the time lost due to queue clearance or waiting for a permissive left. quote:Assuming two roads of equal traffic volume intersecting, how likely is a driver to want to go straight? Left? Right? (Is this a really hard question?) This is something you get a feel for the more you work in the field. You need to take the whole network into account, not just the immediate intersection. If you want a quick approximation, I'd say send 2/3 of the cars through, 1/6 left, and 1/6 right. Any more questions for me? I have a couple for you. First off, where are you generating cars? If you generate them between signals, within your simulated city, volumes will build up very quickly, but you will have trouble tracking individual cars. If you generate them at the ends of roads (at the edge of the city, for example), cars will be easier to track and route, but it will take a long time for the network to build up to its full volume. You're a programmer, so you probably understand a bit better than me, but it seems to be proportional to the square route of a network's linear size. My other question is whether you're routing the vehicles on a per-intersection basis (like in Synchro), or providing each vehicle a path through the whole network when it's spawned (like VISSIM). Synchro's method saves a lot on memory, but VISSIM's is much more realistic, and I've found that Synchro's method leads to weaving and queueing problems at closely spaced intersections.
|
|
#
?
Nov 29, 2009 21:38
|
|
|
Cichlidae posted:Yes, that's one. Conflicting movements must never have protected greens at the same time. It's ok to allow conflicts sometimes, like permissive left turns or right turns on red. This is generally different in Europe, from my experience. Most everywhere I remember in Europe except for the major intersections, there's just two phases: N/S and E/W. On the other hand, turns on red are generally never legal here. Regardless of whether you pass straight through or turn left or right, you enter the intersection on green through. You turn right when there's no crossing pedestrians and left when there's no through traffic in the opposite direction. Only large intersections have dedicated turn phases. This is just to note that things can be done differently. If Pfhreak can generally disregard legislation in his project, maybe implement some adjustable limitations on what is legal, it could be interesting to see what kinds of timings it would end up with, if different levels of aggressiveness were modelled for the traffic. You can also try adding in the European style "red+yellow before green" signalling. (A few seconds before the lights turn from red to green, the yellow lights up together with red to allow motorists to prepare. It is still illegal to start driving until the light has turned green.) (When I visited the US (Maryland) for the first time this summer, I was stumped at how long the phases were and how much waiting time I had as a pedestrian, and I never managed to get a feel for the general traffic signal rules, it seemed like chaos to me. Reading this thread has given me some understanding of what made things feel so different.) nielsm fucked around with this message at 22:59 on Nov 29, 2009 |
|
#
?
Nov 29, 2009 22:56
|
|
|
nielsm posted:This is generally different in Europe, from my experience. Most everywhere I remember in Europe except for the major intersections, there's just two phases: N/S and E/W. On the other hand, turns on red are generally never legal here. We use two-phase on some very low-volume signals, but in an urban state like Connecticut, nearly all signals get some sort of turn phase, even if they don't have a dedicated turn lane. quote:This is just to note that things can be done differently. If Pfhreak can generally disregard legislation in his project, maybe implement some adjustable limitations on what is legal, it could be interesting to see what kinds of timings it would end up with, if different levels of aggressiveness were modelled for the traffic. You can also try adding in the European style "red+yellow before green" signalling. (A few seconds before the lights turn from red to green, the yellow lights up together with red to allow motorists to prepare. It is still illegal to start driving until the light has turned green.) Synchro has different parameters for driver aggression, and VISSIM can model temporary inattentiveness. I've watched people run red lights in both programs, sometimes blatantly. Despite that, aggressive behavior doesn't make a huge impact on overall traffic flow or delay, since it's so rare. Having a mix of different driving behaviors is important, but after a certain point you're spending a lot more time programming and simulating to gain a small amount of accuracy. As to advance yellow-red indications, it would be pretty easy to model in the queue clearance values. Most calculations assume the first car takes 4 seconds to clear the stop bar when the light turns green, then 2 for each subsequent car in the queue. Here in New England, it's more like 2.5 and then 1.5 for each additional car. I imagine it would be similar with yellow-red indications. quote:(When I visited the US (Maryland) for the first time this summer, I was stumped at how long the phases were and how much waiting time I had as a pedestrian, and I never managed to get a feel for the general traffic signal rules, it seemed like chaos to me. Reading this thread has given me some understanding of what made things feel so different.) Cycle lengths need to be longer if you have turn phases, but there's a trade-off. Short cycles have much of their green time eaten up by red and yellow clearance. The shorter the cycle, the less relative green time it can serve. On the other hand, long cycles can cause very long queues and increase delay, even if they have higher capacity. Somewhere in the middle is optimal; most of our cycle lengths are between 60 and 120 seconds.
|
|
#
?
Nov 29, 2009 23:34
|
|
|
Cichlidae posted:Info... Cichlidae posted:I have a couple for you. First off, where are you generating cars? The system currently places cars at the ends of roads, and allows the simulation to run for a fixed period of time. (Say, 1 minute or 30 seconds.) Why so short? The current goal is to try and take a computer that is totally stupid and allows anything goes signals, and teach it to do things like time two signals in a row for optimal traffic flow. You can generally tell right away if the computer is blocking traffic insanely. It's currently very simple and tile based. The most complex 'road' I've put to test had something like 5 intersections. (Like I said, I've only put a couple hours into it.) Cichlidae posted:My other question is whether you're routing the vehicles on a per-intersection basis (like in Synchro), or providing each vehicle a path through the whole network when it's spawned (like VISSIM). Synchro's method saves a lot on memory, but VISSIM's is much more realistic, and I've found that Synchro's method leads to weaving and queueing problems at closely spaced intersections. Because everything is currently tile based, cars decide which direction they will go upon arriving at an intersection. All roads are two lanes (one each direction) and there aren't any fancy shmancy things like turn lanes. Ideally, I'd like to build a vector based model, where cars are generated at a point with a specific destination in mind. I don't know why you'd need to save on memory, how many cars could you possibly be simulating simultaneously? 100K? 500K? It seems like there can't be that much state to keep track of for every car. So, if this class project goes well, I might tear it down and rebuild it using vectors, and maybe in 3D instead of using my boring 2D model. I've always loved just watching cars in SimCity. :3 If I had a program that would let me play with roads like the diagram you posted a few posts up... Edit: A good example of a genetic algorithm is: http://www.wreck.devisland.net/ga/ You can watch as a computer learns to build a car by failing over and over. I am basically trying to do the same thing, but time a few simple traffic signals instead. Pfhreak fucked around with this message at 02:29 on Nov 30, 2009 |
|
#
?
Nov 30, 2009 02:26
|
|
|
Pfhreak posted:None yet, you've given me some good things to chew on. I suppose it's not so much if you look at it that way. It just seems like it would take a lot of computing resources to have each of thousands of cars track which lane it needs to be in, a mile and 16 intersections before it gets there. Of course, without turn lanes... quote:So, if this class project goes well, I might tear it down and rebuild it using vectors, and maybe in 3D instead of using my boring 2D model. I've always loved just watching cars in SimCity. :3 If I had a program that would let me play with roads like the diagram you posted a few posts up... I'd really recommend VISSIM, then. I hear there's a free trial, and you can have a lot of fun, once you get used to the horrible interface. quote:Edit: Boy, that didn't take long. The very first "car" it made worked perfectly. I've seen genetic algorithms used for things like designing antennae and circuitboards before, and they come up with some really neat stuff that humans could never have designed. I wonder if that's possible with traffic signals, or if they're too simple to lead to that sort of design.
|
|
#
?
Nov 30, 2009 02:37
|
|
|
Cichlidae posted:I suppose it's not so much if you look at it that way. It just seems like it would take a lot of computing resources to have each of thousands of cars track which lane it needs to be in, a mile and 16 intersections before it gets there. Of course, without turn lanes... Yeah, I'm mostly building a software 'toy' at this point. Most of the other people are doing things that are boring and no fun to watch. I'm working on turn lanes, but they are a secondary feature behind actually getting the AI to work. Looking at VISSIM made me boggle a little. That's a hell of a thing! But I bet, like most domain specific applications, it's interface is awful? Cichlidae posted:Boy, that didn't take long. The very first "car" it made worked perfectly. I've seen genetic algorithms used for things like designing antennae and circuitboards before, and they come up with some really neat stuff that humans could never have designed. I wonder if that's possible with traffic signals, or if they're too simple to lead to that sort of design. You can use them anywhere you can 1.) record your conditions in a 'chromosome' and 2.) measure one output objectively against another. (Ie, this solution is better than that one.) Now whether you should use them is another story. Still, they are fun to watch.
|
|
#
?
Nov 30, 2009 02:47
|
|
|
Pfhreak posted:Edit: That is hands-down one of the coolest things I have ever seen. Go little car, go!
|
|
#
?
Nov 30, 2009 14:25
|
|
|
Lobstaman posted:As I approach the overpass for Elm St there is a VMS board and typically it's off and traffic flows nice and smooth. But, whenever it's on (usually saying take Rt 3 for I-84) left lane traffic grinds to a halt. There is a slight curve and a hill leading up to the sign. The I-95 one I can understand since trucks need to slow down a bit, but on the Parkway, people jam on their brakes because they're afraid that going over 45 will send them flying off the road, up the hill, and into the trees.
|
|
#
?
Nov 30, 2009 19:12
|
|
|
GWBBQ posted:The greatest enemy of the CT driver, the gentle curve Curvature is very difficult to accurately gauge from the road. I'm sure you've all noticed that curves seem much sharper on the road than they do on maps. Alignment and insufficient superelevation can really increase the perceived curvature. For example, if the beginning of a horizontal curve coincides with a peak or valley, the curve will seem a lot steeper. Your example on Route 15, from Street View, seems to have a long, slow downgrade beginning there, which makes the curve seem sharper. Additionally, it doesn't look like it has much superelevation. We can put up to a 10% cross-grade on curves to ease a car around them, and the difference is very noticeable (especially when you're stopped in traffic!) The Green Book has a great section on alignment and aesthetics, but, alas, it's published by AASHTO. They also offer the latest MUTCD (without revisions) for just $96, along with a link to the FHWA website where it can be downloaded in its entirety gratis! Oh, but you only have to pay $80 if you're an AASHTO member.
|
|
#
?
Nov 30, 2009 22:54
|
|
|
Pfhreak posted:Because everything is currently tile based, cars decide which direction they will go upon arriving at an intersection. All roads are two lanes (one each direction) and there aren't any fancy shmancy things like turn lanes.
|
|
#
?
Dec 1, 2009 01:09
|
|
|
Nesnej posted:It depends on how realistic you want to be. If you just randomly assign routes to the cars or use pre-defined ones, it doesn't need to be computationally very expensive, but the second you introduce some route finding, poo poo gets real. I've been running some traffic assignments on the north-eastern part of Zealand (basically Copenhagen metropolitan region & Co.) and the reasonably new i5 computers we have in the lab take about 10 hours ruminating the data before they spit out the databases. This is done with a static trip matrix so there are no trip generation/attraction calculations to be done. This is of course an entirely different level of aggregation of data, no low-poly cars, just numbers. I've kind of wondered if some serious heuristics coupled with an ant-routing-esque engine would produce more realistic results. In reality, human drivers don't find the shortest or best route, they pick routes that involve a lot of simple roads (routes tend to involve three stages: "get on major road", "navigate near destination", "get off major road") and then iteratively improve from there via their own efforts and via communicating with their friends. The vast majority of little roadlets are simply never inspected in any manner by drivers. At least, out here in the Bay Area, which may have a different general road layout than Copenhagen. I kind of feel like I could spend months, if not years, just mucking about with traffic simulators.
|
|
#
?
Dec 1, 2009 17:59
|
|
|
Not sure if you know, but is there some reason that you see LED stoplights with LEDs that are out? I thought those things were supposed to last forever. Also, Route 8 is so nice now that they used stimulus money to repave it. I wish they had bought some 65 MPH signs to go with it, since I know one day I'm going to be stopped for 80 in a 55 and it will hurt.
|
|
#
?
Dec 1, 2009 20:30
|
|
|
ZorbaTHut posted:I've kind of wondered if some serious heuristics coupled with an ant-routing-esque engine would produce more realistic results. In reality, human drivers don't find the shortest or best route, they pick routes that involve a lot of simple roads (routes tend to involve three stages: "get on major road", "navigate near destination", "get off major road") and then iteratively improve from there via their own efforts and via communicating with their friends. The vast majority of little roadlets are simply never inspected in any manner by drivers. Hopefully these will get built into our traffic software someday. The cars in those programs only have one route they will take, and even if they're stuck in gridlock for hours, they'll go that way instead of taking another street. Of course, you can manually program them in VISSIM to use different routes at different times, but that's cheating! smackfu posted:Not sure if you know, but is there some reason that you see LED stoplights with LEDs that are out? I thought those things were supposed to last forever. LEDs don't last forever, but they last collectively much longer than incandescent lights and use much less power. With so many small light sources, if a few goes out, the red light is still visible. With an incandescent, it goes out all at once and the Conflict Monitor Unit goes into all-out panic mode. quote:Also, Route 8 is so nice now that they used stimulus money to repave it. I wish they had bought some 65 MPH signs to go with it, since I know one day I'm going to be stopped for 80 in a 55 and it will hurt. Route 8 is a beautiful drive, and I can't blame you for wanting to enjoy more of it in the same time! If you like the resurfacing job, go to its recovery.org site: code:(can't link to it for whatever reason) and give it a positive vote. People are saying it's too costly, but the low bid came at under $2.8 million. In fact, all of my jobs lately have bid at 1/2 to 1/3 the estimated price. Nifty! And finally, check out this dedicated and attentive officer I found on Photolog today: 
|
|
#
?
Dec 1, 2009 22:18
|
|
|
Awesome thread! I've got a question for you: What's going on here? http://tinyurl.com/yfn99bo Basically, there are ramps from South Loop 1 -> West 290 & from East 290 -> North Loop 1, but none from North Loop 1 -> East 290 & West 290 -> South Loop 1. This results in people who are northbound on Loop 1 that want to go east on 290 having to exit, turn right onto the 290 frontage road (fighting with the people who  HAVE TO GO TO WALMART RIGHT NOW ), then drive a ways down the frontage road, through two traffic lights, and *then* they have an on-ramp to 290. I presume that the ramps that aren't there weren't built due to budget constraints (as the bases are built, they just never got completed), but why was it *those* ramps that weren't built as opposed to the ones that were? On any given day, I see far less traffic going from E290 -> N Loop 1 than from N Loop 1 -> E290. In fact, at most times, it's actually faster to take Loop 1 on north to 360, then take 360 to 290
|
|
#
?
Dec 2, 2009 18:37
|
|
|
grnberet2b posted:Awesome thread! It's a sad fact of building roads that sometimes, things just don't get finished. The usual suspects: Inability to obtain right-of-way: Doesn't seem to be the case here, as the land is cleared for the ramp piers. Environmental issues: I don't see any waterways there. Cost: Could be the problem here; it's stopped several projects in Connecticut. We have to set priorities very carefully, and sometimes shifting priorities leave a project (or part of one) in the dust. NIMBY: Always an issue. Maybe the nearby businesses were against the ramps for some reason. People aren't always logical. So, why those two ramps and not the others? It could be because of staging, that the first two were already built when budget cuts hit. It could be a structural issue that needs to be redesigned, or a subsurface problem. Just as likely is the possibility that the contractor went out of business between the two. Hopefully, the condition is temporary! Route 1 is still under construction in that picture, so there's still hope. ----- I'm getting married this weekend, so I'll be off on my honeymoon for 2-3 weeks. Feel free to keep posting questions/comments, and I'll get back to them (no doubt with a page full of posts) when I return!
|
|
#
?
Dec 2, 2009 22:03
|
|
|
Cichlidae posted:I'm getting married this weekend, so I'll be off on my honeymoon for 2-3 weeks. Feel free to keep posting questions/comments, and I'll get back to them (no doubt with a page full of posts) when I return! Hey, congratulations! And thanks again for the awesome thread.
|
|
#
?
Dec 3, 2009 00:37
|
|
|
Cichlidae posted:If you like the resurfacing job, go to its recovery.org site:
|
|
#
?
Dec 4, 2009 22:42
|
|
|
Funktor posted:1. Michigan Lefts. They seem to work pretty well to me. What's your take? So that's what they are called. In Mississauga, ON [suburb of Toronto] there's a divided roadway called the Queensway a few km north of the Gardener Expressway/QEW (which is also called the queens way to add more confusion). It runs between commercial business parks and residential areas to the south. It's only used in a few segments of the road ie. the Dixie Road / Queeensway Interchange [where there is a lot of truck traffic]. At least it used to be that way in the 80's, it looks like they added in left turn lanes after the fact. http://maps.google.com/maps?f=q&sou...008079&t=h&z=17 It looks like Peel County/Mississauga used that road as a test case back in the 70's for design, as it quickly terminates near the borders on one end and has different style intersections at almost every light.
|
|
#
?
Dec 5, 2009 02:59
|
|
|
Are you as amused by the curves some of the I-84 on / off ramps in Vernon as I am? My parents used to drive them all the time when I was little, and at 27 it still blows my mind how sharp they are. I joke the former is kept as is to make sure SUV drivers are a little more aware of their surroundings than they might otherwise be. It pretty much dares your vehicle to flip. (The latter might not seem that sharp to people who haven't driven it, but trust me on this one.)
|
|
#
?
Dec 5, 2009 20:37
|
|
|
The on ramp acts more like a 2 lane on ramp with the huge right side shoulder. The Exit 63 EB off ramp is rather sharp too, and as the image shows, 90% of the time is filled with cars. the fact that there hasn't been a massive chain reaction accident here is beyond me.
|
|
#
?
Dec 6, 2009 03:34
|
|
|
Is cost the primary reason why concrete is used in more and more viaducts/flyovers than steel girders? Do the concrete viaducts that have a "V" cross sections have a particular name that they're called?
|
|
#
?
Dec 7, 2009 01:45
|
|
|
Firstly, awesome thread, I powered through it in a few days. My dad used to work for the transporation government of Quebec. I'll ask him more later, but from what I remember he had a role in pioneering the use of roundabouts a lot in Quebec. And speaking of Quebec I wanted to know what you think of this giant clusterfuck, and maybe if you could somehow make it better. I don't think I've heard of a single traffic report where it wasn't clogged to hell. Also fun is 15/20/720 branch going northbound there is actually 2 levels. It's quite possible to end up right below where you want to be, and the GPS doesn't know any better until it's far too late. Also come to think of it, there's another odd entrance ramp that I take on the way back from work. The southbound entrance to the 404 from hwy 7 just seems really odd whenever I take it, I don't think I've ever seen another entrance ramp that diverges from the highway, then merges with another entry ramp, and then merges with the highway.
|
|
#
?
Dec 7, 2009 04:28
|
|
|
I know that you are mainly a highway guy, but I was wondering if you could give an overview of street parking design guidelines. What kind of thoughts go into marking or not marking individual spaces? We have both marked and unmarked street parking here in Seattle. When is it appropriate to have zero street parking in front of a building? How much space do you want to leave in between the street and the lane divider (the key being whether or not it is impossible to get in or out of the car when there is traffic passing by)? Given no right of way constraint, what is your ideal city street design (arterial and non-arterial)?
|
|
#
?
Dec 7, 2009 22:32
|
|
|
Awesome thread, it's taken me more than a few days to get through it all but it's pretty drat cool information. I have to hand it to the engineers who did the newest interchanges along the Red Mountain section of Loop 202 - I believe you'd call them dumbbells, but they all are essentially overpasses over the freeway with roundabouts instead of signals. Even Arizona's monumentally stupid drivers can deal with them at the low traffic loads they encounter, and it's so much faster than waiting for a signal. I'd link you but the Google Maps info and satellite images are out of date and don't really show it. I love that you called out Phoenix for using the SPUI, our engineers have a major hardon for them in certain areas (Phoenix proper and Scottsdale, mostly) and honestly, they work really well. I drive through one daily, SR143 and University. Of course, if you know Phoenix's freeway system at all, you know that driving through that interchange means I'm probably dealing with arguably the worst / most overloaded interchange in the whole state, and you'd be right - the Broadway Curve. Technically, that pretty much only refers to the actual curve itself where I10 turns from a N/S alignment to an E/W alignment, but the whole problem area is here on Google Maps. You have US60, which is currently being widened and serves as the largest single path for drivers to get from Tempe / Chandler / Mesa / Gilbert and towns further east and south to get to anywhere near the central or west parts of town, dumping onto I10 (which, officially, is also US60 at that point). You also have I10, which takes everyone who comes from the more southern extremes (Maricopa, Tucson, areas of the east that are close to the southern half of Loop 202) into the same area. 143 doesn't feed a whole lot of traffic in or out, but it is the primary route from there to the airport, and especially in the evenings, serves as essentially a major (and permanently backed up) onramp to I10. I think there have been freeway-within-a-freeway concepts discussed for possible solutions - to try and separate traffic that enters and exits the complex on I10 from everyone coming to/from US60 and SR143, as well as the study mentioned above where they want to make that bastard 24 lanes wide. Any thoughts on other solutions? The biggest issue with it, as someone who drives it daily, is that there's no real alternate option. If you have to take it, you have to take it; unless you're going all the way downtown to the mini-stack where the 10, 51, and 202 all intersect, there is no viable alternate route. Much easier question: care to identify the interchange type where US60 and Loop 101 meet? Most of the ramps are normal long, sweeping jobs, like at all of our other stacks, except the ones from 60W to 101S and 60E to 101N - they're much narrower and have 25MPH recommended speeds. They're also great fun to take in a sports car at about 50, and surprisingly cause almost no major issues, despite the fact that the barriers are nearly solid black with skidmarks. Also, I've seen you bitch about left-exits / entrances regularly. Does that still apply when they are HOV-only ramps? We have a few of those, most notably between US60 and I10 and also I10 and Third Street. One last question, I swear: use this PDF for reference. The South Mountain section of Loop 202 is still in planning, and causing all sorts of NIMBY bitching, but it will most likely be built (we REALLY loving love our cars here!) and the alignment shown there is the one that has the most backing. Why the gently caress do they think it's a good idea to bring it up to the 10 so far from where the 101 already is? I10 west of Central Ave is already a giant clusterfuck, and now you'll have people trying to get between 101 and 202 on that side of town clogging the 10 up even more. For the record, Loop 303 looks far more impressive on that PDF than it is in real life - it's essentially a trumped up four-lane road with a median.
|
|
#
?
Dec 8, 2009 00:57
|
|
|
IOwnCalculus posted:Much easier question: care to identify the interchange type where US60 and Loop 101 meet? Most of the ramps are normal long, sweeping jobs, like at all of our other stacks, except the ones from 60W to 101S and 60E to 101N - they're much narrower and have 25MPH recommended speeds. They're also great fun to take in a sports car at about 50, and surprisingly cause almost no major issues, despite the fact that the barriers are nearly solid black with skidmarks. I'm not the engineer, but it looks to me like those two ramps you mentioned would have been your average stack ramp, but there's buildings right where they would have gone, so I'd chalk it up to ROW.
|
|
#
?
Dec 8, 2009 15:54
|
|
|
grnberet2b posted:Awesome thread! Yeah S. Austin! At least 71 and 35 isn't an at grade interchange anymore.
|
|
#
?
Dec 10, 2009 16:08
|
|
|
IOwnCalculus posted:Much easier question: care to identify the interchange type where US60 and Loop 101 meet? Most of the ramps are normal long, sweeping jobs, like at all of our other stacks, except the ones from 60W to 101S and 60E to 101N - they're much narrower and have 25MPH recommended speeds. They're also great fun to take in a sports car at about 50, and surprisingly cause almost no major issues, despite the fact that the barriers are nearly solid black with skidmarks.
|
|
#
?
Dec 10, 2009 21:49
|
|
|
Great thread! I love that bits of Denmark have been covered, Copenhagen in particular. Looking over maps of the city really hammers home that the older parts predate the invention of the automobile with a fair margin. Even in the never parts of the greater Copenhagen area, we have beauties like this, a 5-leg intersection I pass through on my commute. They've put in lines to guide traffic now, but it used to be just a big blank empty expanse of asphalt. Unfortunately, generally road maintenance in the Copenhagen area is severely lacking. At work, the road was completely dug up to lay new pipes and fiber etc. which did cause some traffic problems, but at least those were temporary. Now, what was once a perfectly fine piece of road has been patched over with no concessions to quality, it's the closest thing you can get to going off-road without actually removing the asphalt completely. However, this is an effect of the city council choosing to forego road maintenance in favor of better public transport, so it's a tradeoff. Luckily, the rest of Denmark is a bit more reasonable. We even get cool poo poo like this (Danish, sorry) (Google maps):  People were unconciously swerving to "avoid" the V-shaped pillars on the rail bridge even though they're on the other side of the guard rails. Someone realized that this was because people were looking at the top of the pillars and assumed that they were vertical and would therefore be very close to the road at the bottom. So they decided to light the bridge up to accentuate the shape of the pillars. Plus, it looks really cool  More pictures (PDF)
|
|
#
?
Dec 13, 2009 00:22
|
|
|
Please tell me why my city's (Toronto) Traffic Engineers think that inserting a whole bunch of "right-turn only" signs in the middle of the financial district makes traffic go faster? I usually have to make 5 forced right turns before I can finally make a left, and by that time I've driven several blocks away from my destination.
|
|
#
?
Dec 13, 2009 00:42
|
|
|
Cichlidae posted:It's not just your imagination. Putting a message on a VMS will cause people to slow down significantly to read it all. We try to make our messages as short as possible, like so: Why don't you guys do what we do in Michigan and keep the signs on all the time? If there isn't some kind of traffic jam or construction, the signs usually have a message about fastening your seatbelt, or not driving drunk. It helps drivers get used to the sign and not surprised when it's on and trying to tell you something. ESPECIALLY if you guys are worried about suddenly blinding someone at night. A lot easier than installing a camera and looking for a break in the traffic, thats for sure.
|
|
#
?
Dec 17, 2009 08:11
|
|
|
Zero One posted:Why don't you guys do what we do in Michigan and keep the signs on all the time? You know, this never really made sense to me before but it does now. I see those signs all the time in Michigan with some safety messages like those, and I always thought "what a waste of money just to tell people to not drive drunk". But if there was an accident ahead, it wouldn't be a sudden shock to see the message change from the safety slogan to the warning. Why isn't this the case with your signs, Cichlidae? Energy savings?
|
|
#
?
Dec 17, 2009 11:25
|
|
|
I know this isn't quite your field, but do you have any ideas about super efficient car parking lots? I can understand the grid of boxes layout makes sense in terms of storing big square things, but when you factor in turning circles for the 90 degree turns making the driving lanes wider and the generally bad drivers wouldn't there be a more efficient model for laying out car parking spaces? Ever heard anything about this even? Thanks for the great thread!
|
|
#
?
Dec 17, 2009 14:46
|
|
|
BigKOfJustice posted:So that's what they are called. Pretty ballsy to build power pylons right down the middle of the road, eh? Michigan Lefts work quite well, as long as people understand what's going on and there's plenty of room for U-turners. Crackpipe posted:Are you as amused by the curves some of the I-84 on / off ramps in Vernon as I am? My parents used to drive them all the time when I was little, and at 27 it still blows my mind how sharp they are. I joke the former is kept as is to make sure SUV drivers are a little more aware of their surroundings than they might otherwise be. It pretty much dares your vehicle to flip. The on-ramp was designed back in the late 40s/early 50s when I-84 used to be part of Route 15. You can see the wetlands all around it; space was pretty constrained. I'm sure it's not up to modern design standards, though it wouldn't be tremendously hard to fix if we had the cash. As to the off-ramp, all the ramps in that area are weird because of the old Buckland Circle that used to exist at the intersection of 30 and 83. I don't know how long ago it was removed, but you can see it in the old (1960s) aerial photos. By the way, have you noticed that we've done a little rearranging of the I-84 off-ramps onto 83 and put in a median? That was kind of a stealth job, and we knew we'd get a lot of complaints from people who have to flip a bitch to get to Dunkin Donuts  Lobstaman posted:The on ramp acts more like a 2 lane on ramp with the huge right side shoulder. The Exit 63 EB off ramp is rather sharp too, and as the image shows, 90% of the time is filled with cars. the fact that there hasn't been a massive chain reaction accident here is beyond me. We have an ongoing project at this off-ramp to try to help alleviate the queueing problem. What's nuts is that I once drove through there and the signal was broken. They put up temporary stop signs. I'm sure you can imagine the carnage, but for someone who doesn't live there, try to imagine how a double-left works when it's stop-controlled and every approach lane (all 14 of them) is over capacity. I learned there that Synchro is a very poor predictor of how people REALLY act at multi-lane stop-controlled approaches. Cichlidae fucked around with this message at 17:55 on Dec 17, 2009 |
|
#
?
Dec 17, 2009 17:35
|
|
|
Prince of Dicks posted:Is cost the primary reason why concrete is used in more and more viaducts/flyovers than steel girders? Do the concrete viaducts that have a "V" cross sections have a particular name that they're called? Cost is the primary reason, with maintenance concerns coming in a close second. Steel needs to stay painted or else it will corrode, and a large bridge can take tens of millions of dollars just to paint. The paint itself generally contains chromium and lead among other nasty things. As to the V cross-section, we call them tub girders here. They're very easy to inspect because there's enough room for a human to walk inside the larger girders. ReaperUnreal posted:Firstly, awesome thread, I powered through it in a few days. My dad used to work for the transporation government of Quebec. I'll ask him more later, but from what I remember he had a role in pioneering the use of roundabouts a lot in Quebec. Yeah, I've actually looked at that freeway before. It's so bizarre because the lanes are "backward," so the interchanges would have to be rebuilt entirely. I seem to recall reading that they would be rebuilt soon, or are in studies to do so. quote:Also come to think of it, there's another odd entrance ramp that I take on the way back from work. The southbound entrance to the 404 from hwy 7 just seems really odd whenever I take it, I don't think I've ever seen another entrance ramp that diverges from the highway, then merges with another entry ramp, and then merges with the highway. It's a pretty standard way of consolidating ramps to minimize high-speed merges and, in this situation, removing what would otherwise be a very nasty weaving area. You're right, though, it is somewhat rarer to do it with on-ramps; usually it's one off-ramp that splits instead of two on-ramps that merge.
|
|
#
?
Dec 17, 2009 17:55
|
|
|
Basho posted:I know that you are mainly a highway guy, but I was wondering if you could give an overview of street parking design guidelines. I definitely avoid parking lots when I can, and when I do have to deal with them, I use our highway design manual to get the specs. I imagine it'd vary from state to state. As to when and where it should be used, two things are required: 1 is the 8-foot width required to put in a parking space, 10 if we have it, and the other is a request from the town, since the state of Connecticut doesn't administer on-street parking itself. quote:Given no right of way constraint, what is your ideal city street design (arterial and non-arterial)? If we're talking a brand new city, I'd probably churn out something like Chandigarh. It's made up of several modular areas that are each relatively self-sufficient, which minimizes travel, and these modules have a hierarchy of roads. Of course, that's relying on a car-oriented society and 50-year-old urban planning. The theory is sound, but I'd like to integrate mass transit (with high-density urban centers and plenty of rail). I'd stick with the orthogonal grid, though it's terribly boring, because square, four-way intersections are the safest. Any freeways, though, would be built with gentle curves to keep drivers awake and aware.
|
|
#
?
Dec 18, 2009 00:15
|
|
|
|
| # ? May 17, 2024 14:45 |
|
|
IOwnCalculus posted:Awesome thread, it's taken me more than a few days to get through it all but it's pretty drat cool information. Yes, SPUI can handle loads of capacity, especially those with double-lefts like in Phoenix. However, I saw quite a few that were built incorrectly (probably retrofitted diamonds) so that the left turns all had to operate independently. 44th Street under Loop 202 is an example. Of course, it was under construction, probably to widen the bridge and complete the SPUI. quote:Of course, if you know Phoenix's freeway system at all, you know that driving through that interchange means I'm probably dealing with arguably the worst / most overloaded interchange in the whole state, and you'd be right - the Broadway Curve. Technically, that pretty much only refers to the actual curve itself where I10 turns from a N/S alignment to an E/W alignment, but the whole problem area is here on Google Maps. You have US60, which is currently being widened and serves as the largest single path for drivers to get from Tempe / Chandler / Mesa / Gilbert and towns further east and south to get to anywhere near the central or west parts of town, dumping onto I10 (which, officially, is also US60 at that point). You also have I10, which takes everyone who comes from the more southern extremes (Maricopa, Tucson, areas of the east that are close to the southern half of Loop 202) into the same area. 143 doesn't feed a whole lot of traffic in or out, but it is the primary route from there to the airport, and especially in the evenings, serves as essentially a major (and permanently backed up) onramp to I10. If the traffic projections are correct, that's getting up with the busiest freeway in the world, which has an 18-lane cross-section. I actually drove through that stretch several times last week (got married in Phoenix). quote:I think there have been freeway-within-a-freeway concepts discussed for possible solutions - to try and separate traffic that enters and exits the complex on I10 from everyone coming to/from US60 and SR143, as well as the study mentioned above where they want to make that bastard 24 lanes wide. Any thoughts on other solutions? The biggest issue with it, as someone who drives it daily, is that there's no real alternate option. If you have to take it, you have to take it; unless you're going all the way downtown to the mini-stack where the 10, 51, and 202 all intersect, there is no viable alternate route. Making it dual-divided is about the only way they can add capacity without introducing some MAJOR weaving problems. Of course, it'll be tremendously expensive, and require revamping every single interchange in the area. The whole situation is a result of the major bottleneck in the area, as you noted. Completing the Loop 202 south of the mountains would help, but at this point we're talking the difference between 18 lanes and 24 lanes. Of course, there's always the option of moving people to other modes. I saw the new light rail systems downtown, and they're actually not bad, but I'm still not optimistic. Phoenix just isn't dense enough to support effective mass transit, and nothing short of bulldozers or bombs is going to change that. quote:Much easier question: care to identify the interchange type where US60 and Loop 101 meet? Most of the ramps are normal long, sweeping jobs, like at all of our other stacks, except the ones from 60W to 101S and 60E to 101N - they're much narrower and have 25MPH recommended speeds. They're also great fun to take in a sports car at about 50, and surprisingly cause almost no major issues, despite the fact that the barriers are nearly solid black with skidmarks. It's just a stack with two of the flyovers replaced with semidirect connections. Pretty rare, so it doesn't have a name. The 25 mph advisory speed doesn't mean much; it's probably not even the design speed. To calculate those advisory speeds, we go around the curve with a little ball in a circular track. When the ball slides 4 quote:Also, I've seen you bitch about left-exits / entrances regularly. Does that still apply when they are HOV-only ramps? We have a few of those, most notably between US60 and I10 and also I10 and Third Street. Left exits/entrances on HOV lanes are just fine for two reasons. First off, the driver expects to exit/enter on the left when he's in an HOV facility. Secondly, since there's only one through lane, you don't have the problems with merging into high-speed traffic that you normally have with left entrances; You'd be facing the same merge if you entered from the right. There are still minor problems like the difficulty of looking out your passenger window when you merge on from a ramp, but the cost savings of building ramps on the left override that. quote:One last question, I swear: use this PDF for reference. The South Mountain section of Loop 202 is still in planning, and causing all sorts of NIMBY bitching, but it will most likely be built (we REALLY loving love our cars here!) and the alignment shown there is the one that has the most backing. Why the gently caress do they think it's a good idea to bring it up to the 10 so far from where the 101 already is? I10 west of Central Ave is already a giant clusterfuck, and now you'll have people trying to get between 101 and 202 on that side of town clogging the 10 up even more. For the record, Loop 303 looks far more impressive on that PDF than it is in real life - it's essentially a trumped up four-lane road with a median. I don't see why they couldn't connect 202 with 101's current terminus like you say. It's mostly farmland, and there's plenty of room for an interchange. My guess would be political pressure. Either someone important lives south of that interchange, or it's some politician's pet project to bring 202 closer to downtown. There would be some benefit to having 202's terminus near downtown, but the gains in continuity from having it end at 101 would be much nicer. As an aside, it's nice to see they have right-of-way protection. If we'd had that kind of foresight here 50 years ago, our roads would look a lot less congested today. Edit: One thing I REALLY didn't like about Phoenix was the use of ramp metering on freeway-freeway ramps. That is EXTREMELY dangerous, but I wouldn't expect much from a city that uses non NCHRP 350-compliant devices in its work zones. What's that? Warning lights with 10 lb battery packs mounted at head level adjacent to traffic going 70 mph? Nothing bad could come of this! Cichlidae fucked around with this message at 18:27 on Dec 21, 2009 |
|
#
?
Dec 18, 2009 04:21
|
|