I was stuck in traffic yesterday (at this junction right here) - it’s a roundabout that has 4 streams of traffic entering it and there are lights controlling some of the streams (that is switching between traffic going around the roundabout vs traffic entering it).
It’s a common setup - and the precise details of this junction are not important (except to say that this is not just like a four way intersection - it’s different because traffic typically has to traverse more than one set of lights at different parts of the junction… What happened yesterday is that for a couple of the roads entering the junction, traffic was backed up solid and hardly moving at all, because the lights were essentially working against each other - so traffic was either streaming through too fast to enter, or filling up the junction when the lights turned red.
Every time the lights went green (so technically, my lane would be able to go), we had to give way to a stream of traffic that was being allowed into the junction. Every time that flow was stopped, the space we had to drive into was choked with stationary cars.
Now, I know that studies are done on these junctions and traffic flow is modelled - I think in many cases, the lights are programmed to work in pattern that is the least worst compromise, and some junctions have patterns that are different at different hours of the day, or dependent on sensors (these seemed to be very strictly sequenced though).
It made me wonder if some of these junctions would be better if there was a bit of randomness introduced in the timings - in particular, to allow the sets of lights to fall in and out of phase with one another - so the dominant flow of traffic could not completely impede the others (as happened to me yesterday).
So… is this sort of randomness or deliberate phase drift ever used in traffic light sequencing in multi-light junctions?
The trend in traffic engineering is to use sensors to detect traffic and adjust the lights accordingly.
Note that previously only sensors embedded in the road could do this, but now cameras mounted above the road can detect vehicles - much cheaper!
So the thing to do is install sensors there. Also traffic engineers MAY NOT KNOW about the problem if no one tells them! Have you called them to report the problem?
On this particular roundabout, like many others, the main traffic stream is all in the same direction (The A3 in this case), which gives traffic entering from the minor road (Church St) no chance to slip in. The traffic lights are there to allow this, but traffic can be so heavy that it makes little difference.
I know Portsmouth fairly well and it is an overcrowded city built on an island. It is a major ferry, freight and naval port, all of which add to the traffic in this area.
Traffic lights on roundabouts work well at peak times, but in my opinion they should be turned off when traffic is low.
Somewhat related, what I’m often wondered is if traffic engineers assume that drivers will be obeying the law or not.
For example, there shouldn’t be any cars in an intersection (I know the OP is talking about a roundabout but their OP is what made me think of this) after the light turns red because they shouldn’t be there. You’re only supposed to proceed with your turn when you are clear to complete. Of course, very few people do this. Similarly, speed limits generally mean less than 10 over the posted speed limit.
So when doing their calculations do they use the cultural norms or the law?
I run into the scenario all the time. You wait and wait and wait and wait for your turn to go, the light turns green, and the whole Road is filled up with cars you can’t even cross the intersection.
In this scenario, you might even be tempted to run a red light since this might be your only chance to actually get through the intersection. Even if there was a cop nearby he couldn’t get you cuz of the so much traffic.
Yes - roundabouts were originally conceived to be self-regulating systems, however, nearly all busy roundabouts in the UK now have traffic lights added to them, to try to regulate the flow of traffic.
I guess in some cases, the absence of lights would mean that traffic entering from some junctions would just never get a chance to enter the flow of traffic, but as someone who has seen traffic lights go from rarity to ubiquity, it does seem wrong to me. There are many cases of light systems near me that are so designed that they only permit one or two cars through certain routes on each cycle.
Exactly so - and I reckon if the lights were allowed to drift out of phase, there would be cycles when this didn’t happen to you (of course the opposing traffic streams would be inconvenienced, but perhaps only by an amount that balances the permitted flow from your lane - and so it should be acceptable.
There’s a fair bit of voodoo in figuring out the timing of metering lights. Let too few cars through on a cycle, or let one car through per cycle with too long between cycles, and traffic approaching the bridge, freeway or roundabout backs up. Conversely, trying to allow too much traffic through will clog the “downstream” traffic flow.
This is one of those cases where automation hasn’t caught up to human’s innate fuzzy logic.
Yes, it does. I think the correct approach when a roundabout doesn’t work (because of mismatched traffic flows on different inbound routes) is to remove the roundabout and replace it with a signalled junction.
(I realise this is one of those subjects - like grammar - where everyone has an opinion, informed or otherwise. If a traffic engineer tells me I’m wrong about this, I’ll happily defer.)
Engineers are very pragmatic. If there’s a problem they’ll try to fix it rather than ignoring it because they followed the instruction manual while designing it so it’s OK, suggesting drivers have it coming if they’re so stupid, or suggesting the city get some cops patrolling out there.
Pulling into the intersection and waiting if you cannot proceed is the correct way to make a left turn. The additional time it takes to enter an intersection can mean that there’s simply never a gap long enough. Having a short all-red phase is to allow any traffic legally caught in the intersection to clear it. Of course that doesn’t stop those in the intersection from panicking and maybe making a rash move.
The Yellow Trap was pretty notorious (fixed by the Flashing Yellow Arrow) result of this. If the opposite direction was about to show a lagging green arrow, a driver would see the through yellow and panic and make a rash move thinking side traffic was about to be released when it is not. The flashing yellow arrow gives you the option that was previously unavailable of showing a permissive turn with a straight stop.
There’s also the option of extending a flashing yellow arrow for a few seconds after the main phase ends to give stuck traffic a chance to clear without freaking them out with a red, but this is not usually used because of the impact to operations.
I think you’re probably right. Fitting a few sets of lights to an existing roundabout is a cheap and quick workaround, compared to redesigning the whole thing, then digging it up and rebuilding it AND adding lights.
In truth though, as soon as a roundabout has a set of lights on it, it stops being a roundabout and becomes a junction that just happens to have a circular feature in it.
I’ve often wondered if traffic engineers even have a drivers license, or if it’s all theoretical and “should work”. When driving, one encounters many situations in which one wonders of the designers of roads (or even cars, for that mater) have ever actually driven an automobile.
Closer to home, an example right here in this forum. When starting a thread, you are given space for about three words to compose a title. After three words, the first words disappear from the text entry field, so typographical errors are epidemic in threads that are otherwise more carefully proofread. The title is not seen by the poster until the thread has been posted, when it is too late to repair it without the hands-on intervention of a moderator. So,obviously, the web designer of the forum board has never actually used the product he has designed.
I think the problem really is that traffic control and engineering is probably more art than science. I mean, there are standards manuals for things like signage, etc… but I’d bet that actually modeling it would be one of those problems that sounds like it would be simple, but in practice is extraordinarly hard, as it’s not just modeling a half-million cars at once, it’s trying to realistically model driver behavior. And that’s probably not easily or accurately done on a level granular enough to work for an individual stoplight.
I’m personally surprised that the streetlights in a city aren’t starting to be networked, so that if the cameras/inductive loops detect high traffic in one area moving in a certain way, they can self-adjust to keep traffic flowing better. I’d think that knowing what traffic looks like 2-3 intersections in all directions would be an extremely helpful thing for letting a stoplight determine what its timing may be now, and in 5 minutes.
most places I’m familiar with are fairly good about keeping up on traffic studies and optimizing light timings. usually when someone complains about the competence of planners and engineers, it’s just out of frustration when they have to sit at a light longer than they think they should.
Mangetout you’re a Brit? Or do they say roundabout in USA too? I (a non-driver) always thought we say traffic circle.
ETA: Linguistic forensics on speaker: I think, courtesy of some hijack, you’re the snow-pea guy, not as I thought the guy-who-eats-everything guy. Only Brits and Frenchies say mange-tout for snow peas. Ergo,…
Can’t find it now but I’m sure there was a study once in the UK where they removed traffic lights from certain junctions or roundabouts and traffic flow actually improved. The uncontrolled traffic was regulating itself in a far better fashion than the lights.
