Wednesday, March 9, 2016

The case of one-way streets

Another common recommendation from urbanists in North America is the elimination of one-way streets, especially for main streets. The reasoning being that one-way streets reduce the access to commercial main streets, thus reducing their ability to reach certain customers and making businesses lose visibility. So I thought I might give my 2 cents as a traffic engineer on why certain cities have made their main thoroughfare one-way, and the impact on traffic and safety.

The reasons why certain streets have been made one-way

First, let's get a particular case out of the way. Certain residential streets in dense areas have been made one-way not for any traffic engineering reasons, but only because the street is relatively narrow and the planners wanted to have parking on both sides of the street, which didn't leave enough room for more than one travel lane in the center of the street.
8-meter wide street in the Plateau-Mont-Royal with parking on both sides, leaving only about 3,5 to 4,0 meters for a travel lane in the middle of the street (12 to 13 feet)
This is the kind of design that tends to be present only in old, dense areas of cities, where buildings have no off-street parking spots, where streets are relatively narrow and are in a strong grid format, allowing for easy detours. In recent developments, these conditions are pretty rare, streets are rarely in a grid, streets tend to be wider and there tends to be more off-street parking.

Some areas have also learned to use these one-way residential streets as a way to deter through traffic in residential areas, creating a maze of one-ways for cars despite the grid-like street network. Especially when a street's one-way direction changes, forcing people off the street to continue ahead.

Anyway, this type of one-way street tends to be of limited applicability. So let's look at a more contentious application, namely the one-way arterial.

Ste-Catherine street, 2-lane one-way arterial street with parking on either side
Maisonneuve, the "companion" to the above, being one-way in the other direction
OK, so what is the main point explaining why an arterial would be made one-way?

You might have expected it, it's about throughput capacity mainly. The main reason why one-way streets perform better capacity-wise is that what limits street capacity is intersections, and left turns especially can have a deleterious impact on capacity due to the conflict with vehicles coming from the opposite way. Left-turning vehicles therefore have to either wait for a gap in oncoming traffic, or have a separate phase (during which the flow of vehicle coming the other way will be interrupted). This is especially bad when the number of left-turning vehicles is high and when they share a lane with vehicles going straight.

With one-way streets, there is no issue with left-hand turns, they have the equivalent of a permanent priority phase. 
Depending on the number of left-turning vehicles, the second design probably has a capacity 20 to 40% higher than the first
What if you have a lane for left-hand turns only? Like the famous "road diet" design with one lane allowing straight and right-turns and a lane only for left-hand turns? Well, in that case, the effect of going one-way is even greater as it allows to have three lanes in each direction rather than just two, increasing throughput by at least 50%.

In this case, the second design actually increases capacity by at least 50% without requiring any wider street
So that's an obvious reason why traffic engineers find the one-way solution so attractive. When the goal is to move as many vehicles as possible in an hour, it is a very good solution. When the street is not capacity-constrained though, one-ways impose detours and can be annoying for drivers. But what are the effects of one-way streets aside from capacity? Are they safer or more dangerous?

Impact on speed

All else being equal, one-way streets are likely to result in faster speed in vehicles. Why is that? Because you are increasing the number of lanes per direction per street. When there is only the one lane going in one direction, the speed of traffic will actually be determined by the slower drivers, because drivers who tend to drive faster will be stuck behind them, not being able to pass. This keeps speeds low, obviously. When there are two lanes per direction, faster drivers have the ability to pass slower drivers, not only does this allow them to drive faster, but the effect of many cars passing slow drivers might intimidate these to speed up to match the speed of other vehicles on the road.  In effect, traffic speed is determined by the faster drivers.

So the more lanes per direction on a street there are, the higher the speed is likely to be. It's not the only factor, but it's an important one.

Higher speed is actually a mixed bag. For the traffic engineer, higher speed means higher mobility, which is generally an objective of their job. However, higher speeds in urban areas can mean more noise and more serious accidents, because kinetic energy increases with speed to the power of 2.

I know what some will say: why the hell would I want faster streets in my city? That's a good question, but in the end, in the modern economy, you need reasonable fast mobility for some trips in order to be able to tie regions together economically. Not all streets need to be fast, but some reasonably fast streets make sense to facilitate longer distances, and also to enable faster transit. In Montréal, it isn't rare for bus lines on one-way streets to be 50% faster than similar bus lines on similar streets that happen to be two-way streets.

Impact on accident risks

So one-way arterials are likely to have higher speeds compared to similarly wide two-way arterials, does that mean they're more likely to have crashes? Not really. Higher speeds tend to increase the likelihood of severe injuries in crashes, but not really increase the likelihood of crashes.

For pedestrians, one-way streets are actually not that bad, because they simplify vehicle movements, so they only have to look in one direction before crossing, whether at intersections or mid-block. This increases predictability for pedestrians, which is crucial for safe decision-making.

Also, one-way streets avoid one major issue that is often seen at intersections with traffic lights, the left hook. The issue here occurs during regular green lights that allow left turns while oncoming traffic still has its green, meaning the regular round solid green light. Drivers turning left thus have to wait in their lane for a gap in the traffic in front of them, which is where their attention will be concentrated.

When the driver sees a gap in the traffic in front, especially if traffic is heavy and he is frustrated, he will quickly exploit that gap and accelerate quickly to turn left across oncoming traffic. Meanwhile, not only was his attention not to his left, but the traffic is likely to have impaired his visibility of the pedestrian crossing to his left...

So if a pedestrian was crossing the street at the same time (or worse, a cyclist), you will have a dangerous situation where the left-turning driver is accelerating quickly towards a pedestrian he has not yet seen, which may result in a dangerous crash.  Even if the driver manages to stop before hitting the pedestrian, he becomes vulnerable to a side crash by the oncoming vehicle to his right.

With one-way streets, since there is no oncoming traffic, left-turning drivers will tend to keep their attention to their left, where they are going, which puts any crossing pedestrian well within their line of sight. Since they also don't need to exploit a gap, they may also accelerate slower, without a feeling of urgency.

So this is all theoretical, do we have some empirical data? Well, more or less. Recently, a graph of pedestrian crash hot spots was made of Montréal, and I used my own knowledge of the streets to identify one-ways and two-way streets, superposing both to see if a pattern can be found:
In yellow, one-way streets, in green, two-way streets
I know from experience too that there are tons of pedestrians on all streets I've identified, so you can't explain away the discrepancy with the number of pedestrians. At worst, you can say there is no great difference in accidents between one-way and two-way streets, at best, you can notice a trend toward a lesser number of accidents on one-way streets. So I guess it confirms my intuition that one-way streets, for a similar number of lanes, do not decrease pedestrian safety, but are likely to increase it somewhat, at least at intersections.

Impact on trucks

First, say hello to my "friend" the WB-20 (Which due to my job I have dubbed in French the MCAM.... Maudit Camion À Marde... Goddamn Fucking Truck):
Turning template of the WB-20
What is a WB-20? It is the term used to describe the design vehicle for which all roads that receive Federal funding in the US have to be designed. It represents a semi-truck pulling a 53-foot trailer behind it. This is the usual truck size in the US, and since Canada is the US' toy poodle, the US standard has also been adopted north of the border.

This truck is the death of old towns. Why? Because to simplify freight, trucking companies much prefer to stick to the 53-foot trailer size, even for local deliveries. But city streets aren't made for such huge behemoths, even North American main streets. This might be part of the reason why many stores prefer the periphery over a main street location, to avoid the hassle of having 53-foot trailers making deliveries down narrow streets.

In comparison, most of Europe limits trailer sizes to 12 meters (40 feet), which North American norms call a WB-15, which turns much better:
In orange, the turning template of the WB-15 superposed on that of the WB-20
Note that an alternative might have been to adopt a standard of trucks pulling two shorter trailers, which, like articulated buses, result in a better turning area:
The WB-20D, same capacity as a WB-20, but much better turner

In red the turning area of a WB-20D, superposed over a WB-20.
Anyway, the WB-20 standard also explains why most highways have extremely wide curbs at intersections, exactly to allow this kind of vehicle to turn.

Unfortunately, these trucks are the lifeblood of the economy. They deliver materials to factories then deliver goods to businesses. It sucks, but it is what it is, if you want a city to be economically viable, it is best to find a way to welcome these huge trucks on streets. Yes, I know it would be best to force truck companies to use smaller trucks in cities, but unless the regulation is made at the national level, I doubt it would be enforceable.

So let's look what a turning WB-20 would look like on a "road diet" street with 1 lane per direction plus a center lane for left turns and space for parked cars on either side (about 13 meters wide minimum, 44 feet):

Notice the problem? A truck in its right lane turning right would need to get on the wrong side of the street for nearly 20 meters (nearly 70 feet). Since there may be cars stopped at the stop line, that stop line would have to be pushed WAY back. It's obviously not a comfortable situation for anyone. You could also widen the street with a big curb at the corner...

But in many ways the cure is worse than the disease, especially since pedestrian crossing length would jump from about 14 meters to 22 meters if done on all corners. If you think this is just theoretical, think again, why do you think intersections like these ones are made?

Meanwhile, let's look at that same WB-20 turning with one-way streets of the same width:

Here, there is no problem, even if the truck goes into the center or left lane, as they also go in the same direction, it can easily go there without any danger to anyone. So one-way streets are a way to allow big trucks into cities without transforming intersections into the abominations I just showed, while maintaining adequate pedestrian safety.


One-way streets are certainly not the best solution everywhere in a city. However, I do believe that it would be foolish to turn our back on them, they are a good solution to a particular problem, namely how to deal with long North American trucks and provide adequately fast streets for longer distances, both for cars and for buses. In fact, I think that this design in urban areas is far preferable to the huge 6-lane arterial which we unfortunately see too often in suburbs.

The kind of suburban arterial that could and should be replaced by a pair of 3-lane one-way streets separated by 100 meters
However, on commercial main streets where the focus should be on pedestrians and slow traffic, two-way streets are the better solution.


  1. After your photos of Sainte-Catherine and your map of dangerous intersections, how do you come to the conclusion in your last sentence that two-way is better for commercial main streets?

    1. Two-way streets increase accessibility and favor lower speeds, so commercial main streets that depend at least somewhat on consumers coming by car are likely to benefit more from a two-way street. One-way streets are more appropriate for streets that serve primarily for through traffic, not for shopping. If most customers come by transit, on foot or on bikes, then one-way streets are fine. Indeed, Ste-Catherine street is one-way but is full of people because it is very well served by the metro system.

    2. What about surface transit? Jarrett Walker argues that one way streets are bad for surface transit because people have to use two stops for any origin/destination and it's thus less likely that their origin/detination within walking distance of both stops.

    3. I can see how it could be good and bad for transit. As you say, on the bad side, it could force people to walk one more block to get to their stop and somewhat complicate transfers. On a positive side however, one-way streets would likely result in faster bus speeds, so though it would take more time to get to the stop, travel time in the bus could be shortened. Furthermore, if we compare it to the scenario of two two-way streets each with its own bidirectional bus line, the one-way streets would likely have shorter headways because the two lines would be condensed into one. Meaning that instead of having bus lines heading west on street A and street B, everyone who would have been using them would have to use buses on street A.

      Jarrett has probably access to specific data, so maybe he has analyzed similar cases and found a reduction in transit use, the negative overwhelming the positive, but I would be interested in data, if it existed.

    4. Here's Jarrett's post. It doesn't contain any data, just his geometrical argument, which I've already repeated here.

      One-way pairs are certainly not a problem in a dense area where streets are close by, and there is a lot of service in every street. It might start to be an issue in a lower density setting, where arterials are far apart. The only way the service can be useful there is by running both ways on the same (fast) street.

      (Sorry about the deleted comment, I had no experience commenting on a blogger platform before)

    5. Yes, if the one-way pairs are separated by 200 or 400 meters, it starts affecting coverage significantly. However, I spoke here more of one-way pairs that are closer together, 80 to 150 meters from one another.

      I'll note a few things too. A very wide arterial can be extremely difficult to cross for a pedestrian, it's nearly impossible to cross a 6-lane street mid-block, but a 1-way 3-lane street is much easier to cross, and even at intersections, the 1-way street will have shorter traffic light cycles and be easier to cross. That is what I call the "barrier effect" of very wide streets/roads that can effectively inflict a 3-4 minute detour to users to be able to cross it. So his 400-meter limit is a bit of a simplification.

      Also, even he has pointed out that people are willing to walk longer distances for faster and more frequent transit lines. I think this could somehow mitigate the effect of distancing parallel unidirectional bus lines.

    6. Yeah, I'm not convinced this is such a big problem for transit. Sure, if you have arterials spaced very far apart like in the suburbs, don't make them one-way, but in more downtown type settings, it's less of an issue.

      Ex, you live right in between route 1 and route 2. 1 and 2 both run on one-way couplets. Going from north to south, the routes are 1E, 1W, 2E, 2W... E for east-bound, W for west-bound. So with this set up, 1W might be closer to you than with 2-way streets, but 1E might be further. However, 2E would be closer, so you can just take 2E instead of 1E.

  2. You say that one-way streets simplify pedestrian crossing at mid-block because they only have to look one way. But this is only an improvement for single-lane one-way streets.

    On multi-lane one-way streets, or any time there are multiple lanes moving in the same direction, a person crossing is at-risk from one of the most deadly types of crashes: what I call the double-threat. A vehicle traveling at hazardous speed can be masked by the presence of another, closer vehicle, creating a blind spot for the person crossing. Or perhaps the driver of the rear car, not seeing the pedestrian, decides to try and pass the car in front, leading to a collision course. This is illegal behaviour in every place I've studied, but it still happens all the time.

    I have a hypothesis that the relatively good safety record of UK streets is partly due to the fact that it is very rare for streets to have more than one lane in one direction. Even out in the suburbs.

    Where the streets are wider, well, I've personally witnessed some near-tragedies. And just from observations, it's obvious that UK street design in most places has largely been crap for decades, and that's only starting to change nowadays. People mostly get by due to the small width of streets. If the typically-UK style of street design was implemented on American-sized wide streets, it would result in even more of a bloodbath than currently takes place in the USA.

    One way streets are pretty rare here as well, even on streets that seem like they should be way too small to support cars passing. "One way systems" are mostly kept to city centres or densely populated residential neighbourhoods.

    The truck issue is well-taken, but I think that cities are going to have to insist on smaller trucks for deliveries. That's how it goes here. Fire trucks and even ambulances are smaller as well. Cargo bikes are growing in popularity for some types of deliveries too; one company even has an "intermodal centre" for off-loading larger HGVs onto cargo bikes. Besides avoiding the turning radius problem, they also tend to get to their destinations more quickly than any motor vehicle can, because they can use all the bike shortcuts and avoid traffic. And it's not like shops have any place to unload goods from a truck anyway; "pavement parking" of large HGVs is unfortunately far too common and also needs to be stamped out for safety reasons.

    1. The problem you mention, of a car hiding another, is more valid at pedestrian crossings where one car may stop to allow a pedestrian to cross, but another car behind that one doesn't see the pedestrian because of it and goes on at full speed. I doubt it applies for a mid-block crossing outside of marked crossings, because the speed differential of the two vehicles must be really, really high for that situation to occur.

      From my calculations, on a 2-lane 1-way street, with a nearest car traveling at 40 km/h, the car in the further lane needs to go at minimum at 60 km/h to mathematically risk hitting the pedestrian that starts crossing, thinking he has the time. Not only that, but at the moment the pedestrian starts crossing, if everyone stays at constant speed, it will take 7 seconds for a crash to occur, however, after 1 second of starting to cross, the pedestrian will notice the car if he's looking in the right direction, after 2 seconds, the driver himself will notice the pedestrian.

      At that point, calculations show that even a truck in slippery conditions would be able to do an emergency stop and stop well short of hitting the pedestrian. Even a tram or a subway's deceleration potential would be sufficient to slow down enough to avoid hitting the pedestrian. And that's if the pedestrian doesn't simply step back after starting to cross after noticing the car.

      So, you would need to have the following situation:

      1- A car driven at 40 km/h in the near lane at around 60 meters of where the pedestrian wants to cross
      2- A car driven at 60 km/h about 60 meters behind the other car in the far lane
      3- The pedestrian starts to cross, but stops looking toward oncoming traffic while doing so, crossing at a leisurely pace of 1 m/s
      4- The driver of the far lane not noticing the pedestrian that is visible for him and right in front of him for at least 7 seconds

      Then a severe crash would occur. The likelihood of it seems remote to me.

      Now, if there is a crossing and one car in the near lane stops completely, then that is the danger, with the pedestrian starting to cross and then a car passing the stopped car at 50 or 60 km/h without slowing down. Even then, a pedestrian looking in the right direction would be at least likely to see the car just before entering the second lane.

      BTW, I cross a similar crossing (2-lane, 1-way because of a median) every morning and evening to get at work, I've never had any problem seeing or being seen by cars. The biggest problem is when a bus is parked in the near lane, I and others then stop momentarily just before entering the second lane and check if it's clear.

      A city requiring freight trucks to be smaller in a North American context would only succeed in weakening or killing its industries and businesses inside its limits, pushing them to the periphery where bigger trucks are allowed. Smaller trucks would need to be mandated at MINIMUM at the State/provincial level for the rule to stick, and even then they would face huge pressures from the trucking industry lobby.

      Fire trucks and the like are much easier to pull off.

    2. I initially assumed a formal crossing -- in the informal case there are also several dangerous scenarios where there are sight-line obstacles, a long stream of traffic with few breaks, or a driver who yields to the informal crossing, for example.

      In London I saw a couple (on Feb 14th, no less), get frustrated waiting at an informal crossing of a large road for a long time. They made a sudden dash for it, as one car slowed down, but not seeing the extra car coming up. Luckily the driver saw them in time, with a few metres to spare.

      This is something that could happen in other contexts, also 2-way roads, and might be more of a lesson in the failings of British street design in general (why was such a major desire line so poorly served?).

      Regarding trucks, I don't think that it has to weaken or kill those businesses. Obviously if a business is particularly reliant on 5-axle trucks then perhaps it is best not to be located within a residential, retail or walkable mixed-use area where street design standards are oriented towards people. That still leaves plenty of land for industrial estates and the like, elsewhere.

      In Europe, there are plenty of heavy goods vehicles -- larger ones than the USA -- but they are not used for deliveries in city centres nor expected to navigate small streets. Remember, the rail network barely handles freight at all compared to the USA, so it must all go by truck (although I did see an impressive amount of freight go by while out and about along the Munich S-Bahn -- electrified too).

  3. Not all deliveries are made by semi tractor trailers though, especially not in central city areas. It would be important to categorize exactly which businesses would both want to locate in an urban neighborhood and also want to accommodate large delivery vehicles (supermarkets for instance) to figure out how to accommodate them specifically. Coffee shops, offices, small grocers, bakers, etc. are already served by the relatively smaller delivery trucks used by UPS and FedEx.

    Another factor is that not all intersections need to be "turning intersections" either. This is partly where truck routes came from in the first place, providing a route that allows them to navigate without getting stuck at turns or dead-ends or because of height clearances. So you could theoretically have a city with super tight street corners that's still navigable by large vehicles by providing strategically placed generous intersections (something along the lines of a European ring road) and requiring those vehicles to drive straight through the narrower side street until they come out the other side. It's much easier to pull to the side of the street to make a delivery than to try to jockey into a loading dock, which you don't see much of in urban neighborhoods anyway.

  4. I see the main disadvantage of one-way couplets being that they encourage high-speeds on streets where it's undesirable. Aside from noise, there is still a potential safety issue but that will manifest itself mostly with mid-block crossings rather than intersections which as you showed as no less safe than with 2-way streets.

    Why are high speeds dangerous?

    1) As you said, greater kinetic energy/force when there's a collision.

    2) Greater stopping distance. Stopping distance includes the distance travelled during the reaction time for the driver, plus the distance travelled while braking. Both are greater the faster you travel. People say you should have 20mph speed limits/design speeds vs 35mph because the risk of dying when struck at 20mph is much lower, but honestly, most collisions on streets with 20mph travel speeds will probably occur at much slower speeds like maybe 5-10mph because drivers will have had time to brake. Meanwhile on roads with 35mph travel speeds many of the collisions will probably be at 25-30mph.

    3) Pedestrians have a hard time judging when it's safe to cross when travel speeds are high. Maybe on Strong Towns but somewhere someone said humans are build to withstand falls/crashes at speeds that are equivalent to running speed, and that's why fatality rates increase so much from 20 to 30-35mph. I don't know about that (do faster animals fare differently?)... But I do think that humans find it quite difficult to judge how much time they have to cross when a car is coming at 50mph vs 20mph. A pedestrian frustrated with waiting for a safe gap might misjudge how much time they have on a high-speed arterial in a way that they wouldn't with slower traffic. And this applies for turning drivers too, which might get impatient and frustrated with their difficulty to estimate when it's safe to turn and forget to watch out for pedestrians (or just get t-boned).

    1. Can't we make left-turning movements less dangerous for pedestrians (and everyone else, really) by setting back the pedestrian crossing by one car-length? This way, it's possible to separate "waiting for a gap in the oncoming traffic" and "yielding to pedestrians". Of course, this doesn't get rid of the problem for large trucks.

  5. this is a great illustration of when to, and when not to, use one way streets. Like beauty and art, using one way streets depends on the context.

    Mr. (Ms.) Urban Kchoze,
    Could you comment on making transportation better for kids?

    kid city
    Joseph Lambke, AIA, LEEDap | Feb 4, 2017

    Of course you know that if transit works well for kids, it makes the city way better for the elderly too.

    Thanks for the good work!

  6. Excellent post! See also these posts at StrongTowns: