December 5, 2012
Why Bike Boxes Don't Work (Except for Where They're Working)
Here's a fun little brain teaser: Where have you seen this lane configuration:
While you pontificate that, allow me to regale you with a metaphorical vacation slideshow. Two summers ago, I was in Delft, NL as part of a PSU engineering class studying Dutch transportation infrastructure. An assignment offered a list of various pieces of bike infrastructure and asked that we ride around until we find and document an example or two of each. Most facilities were easy enough to track down. As one would expect in the Netherlands, bike facilities from the basic to the exotic were all commonplace, but one item on our checklist was conspicuously hard to locate: a bike box. Eventually, we did manage to find one lonely installation on a tiny side street downtown:
It is worth observing and thinking about this bike box for a moment, given that it's the only instance of that treatment in a sophisticated university city of 100,000 in the unassailable cycling capitol of the world. Why is this the only sight where the treatment was utilized, and why was it the treatment chosen for this location?
One feature that immediately jumps out is that it's on a narrow, brick street. This means that traffic is naturally calmed--cars are limited by the physical characteristics of the street to about the same speed that bikes go (It also invites a chance to demonstrate your city's commitment to bike infrastructure by setting the edges of the bike facility in brick!). Also, note the share of the street's width allocated to bikes versus that allocated to cars. The wide bike lane is indicative of the fact that there are easily as many bikes that use this street as cars. And the street has fairly low traffic volumes in general; this makes it easy for both drivers and cyclists to stay aware of other traffic on the street with them. The cross street is much busier, so most of the time folks have to stop at the signal.
Look carefully at the lane configuration here. Bikes can turn in either direction, although most turn left toward the nearby train station. (Before the construction began, the bikeway to the train station was straight ahead; that configuration can be seen in Google Maps). Cars must turn right. Approaching an intersection on a bike, it's reassuring to know that a car is turning right, whether or not it signals, and whether or not it brakes hard into the turn.
Compare this to a bike box in Portland that has come to typify many of the problems with the treatment. The photo above shows SW Madison Street approaching SW Third Avenue. It is a primary route to the Hawthorne Bridge for both cars and bikes. It carries more cars than bikes, but enough bikes to warrant much more than the tiny sliver of road they've been allotted (an in which they're legally compelled to stay). Most cars go straight, but some turn and not all that turn use signals. While traffic signal timing keeps the speeds reasonable, there's no natural, self-enforcing traffic calming on this road. Tangentially, there is no "No Turn on Red" sign displayed, though the design guidance is unambiguous in recommending (if not outright requiring) one.
Again, take note of the lane configuration. This one should look familiar--it's the lane configuration at the top of the post, except I drew the bike lane as a vehicular lane to fool you (did it work?). But I also wanted to make a point: you will never see this lane configuration used with two vehicular lanes due to the obvious conflicts inherent with adjacent through-right lanes. However, if you swap out the rightward vehicular lane for a bike lane, this is precisely the lane configuration that you see at countless intersections throughout Portland (and beyond).
This is the origin of the right-hook conflict. Complicating matters, the through movement gets the right of way, and in this case, through vehicles are coming from behind, often hard to see, moving at speeds that can be hard to judge, and particularly susceptible to injury in the event of a crash. That's why the bike box here isn't working: unless you arrive at the red phase, it doesn't really address any of these factors. And the signals are timed such that traffic arrives mostly on green.
Intersections like Madison and Third have given bike boxes a bad name, but they can be an effective treatment if we get choosier about where to install them. Above is a bike box on the SE Clinton Street bikeway approaching SE 39th Avenue. The similarities to the bike box in Delft are striking. It's a low-volume, traffic-calmed street, where almost all bikes go straight and all cars must turn. Roadway space is shared until a wide-ish bike lane splits off just before the intersection. Most traffic arrives on red. The bike box does well in this scenario, organizing the queues of bikes and cars as they arrive on red, imparting a natural order to the queue clearance when green arrives. It's a nice bit of infrastructure that's safe and comfortable to ride on. The same conditions exist and the same treatment is used just up 39th Avenue at SE Lincoln Street, but most of Portland's bike boxes have far more in common with the unsuccessful Madison Street installation than this successful one.
So a bike box can be a good solution if a specific set of circumstances are met. Used as a cookie-cutter solution to the right-hook conflict, however, bike boxes are the wrong thing used wrongly. So what might work better? A few experiments in the city are worth keeping an eye on. At Couch and Grand, the City is trying to tackle the visibility issue by using an unmissably obnoxious sign to tell drivers when a bike is approaching. More recently and perhaps more intriguingly, "mixing zones" are being utilized on the new bikeway on Multnomah, representing an entirely new (and in my view, more sound) approach to intersection treatments that addresses many of the issues that bike boxes don't. (What about adding the bike box treatment to the end of a traffic-calmed mixing zone?)
In the long term, we should envision physically separate bikeways with separate green phases at our major signalized intersections. In the interim, it's important to acknowledge that we have a uniquely Portlandian problem that will require a uniquely Portlandian solution. Our city's grid system and small block size serves us wonderfully in many ways, but it complicates installation of safe bikeways. A bikeway is only as good as its worst intersection and we've got a lot of intersections in Portland. Like too many pieces of our bike infrastructure, the efficacy of our intersection treatments have not kept pace with our ballooning ridership. Portland can and should be a leader among North American cities in designing and implementing a better solution to this problem.
December 5, 2012 11:17 AM
Since both of those boxes are on my commute home, some nonscientific observations:
The only reason I can see for people getting into the "box" part of the Madison box (as opposed to the extended lane part) is to keep cars from turning on red. It's sort of an awkward (and sometimes aggressive) gesture, especially since you'll have to boogie back to the right and get into the bike lane when the light changes. Re: the Clinton/39th: Crossing the intersection and knowing no one is on your tail is very nice. There, the box lines up and there's no need to get across. Usually 3-4 bikes ride side by side for a while until people's natural speed is clear. The only problem there is, strangely, the wide bike lane. In the evening rush, cars use it as a turn lane because it's wide enough for them to do so. They're not unaware of the rules, of course, but drivers waiting several cycles tend to get impatient (especially when you pass them on a bike get through in one cycle).
December 5, 2012 1:44 PM
Nick Falbo Says:
Your indictment of bike boxes on busy streets is really just an indictment of bicycle lanes in general, and has nothing to do with the box.
When the street warrants a dedicated right-turn-only-lane, you've identified the major solutions:
- A Through Bike Lane, to the left of the right turn only lane.
- A "Mixing Zone" (aka. Combined bike lane/turn lane)
- Dedicated bike signal phase.
The real dilemma is the through/right lanes adjacent to bike lanes. Will that car to your left go straight? Will it turn in front of you? Will is yield to you? Nobody knows.
The vehicular cyclist (anti-bike lane) solution would be to remove the separate lane completely and encourage the bicyclist to occupy a controlling position in the general purpose travel lane. But this is not a very "Portlandian" solution. Bike lanes, and their better cousins, cycle tracks, are here to stay in Portland. So what do we do?
Short of prohibiting right turns at those intersections, the solution has to be about encouraging proper yielding behavior.
- Raised cycle tracks should remain raised through minor intersections. This creates a clear priority of bicycle traffic over turning auto traffic. It also acts as a speed hump.
- Bike actuated signs/lights indicating the presence of approaching bicyclists. Couch/Grand uses a light up sign, but I hope we'll be able to use something more like In-Roadway Warning Lights for a less obtrusive indicator.
- Clear visibility. Prohibiting obstructions (parked cars, etc) is a requirement to let auto-drivers see and yield to through bicyclists.
December 5, 2012 2:41 PM
Chris I Says:
Raised sidewalks and cycletracks should be the standard for all active transportation infrastructure. It calms traffic and makes pedestrian/vehicle interactions more predictable. Unfortunately, this is an expensive option that would likely have a lot of pushback.
December 5, 2012 3:16 PM
Great post, Brian and helpful follow-up, Nick. In addition to keeping cycletracks raised through minor intersections, it'd be great to see those intersections neck down severely so that they cannot be entered or exited at speed. I've even seen some where cross streets are narrowed to only be wide enough for one vehicle even though they're two-way (http://goo.gl/maps/JJtE8).
The NE Cully cycletrack drops the cycletrack at intersections and retains radiused corners for those minor cross-streets. Here to hoping that version 2.0 of a true physically-separated cycletrack will incorporate some of these more subtle but still important elements.
December 5, 2012 3:39 PM
Nick Falbo Says:
Awesome link Carl, that's exactly what we should be doing on our new raised cycle track installations.
December 6, 2012 9:28 AM
Jason McHuff Says:
And if you rotate that view, you can see how they mark a crosswalk across the cycle track. Is something like that needed on Broadway at PSU? (Or maybe a bike symbol aimed at crossing pedestrians?)
December 6, 2012 10:54 AM
Good thoughts. I'd also point out that the Google image shows the ultimate indication of the failure of the 3rd & Madison bike box...a ghost bike.
We can, and must do better.
December 6, 2012 1:11 PM
Roger Geller Says:
On a recent trip to Copenhagen I saw numerous intersections where the cycle track dropped shy of the intersection and introduced cyclists into a mixing zone with right-turning motor vehicles (much as we have constructed on NE Multnomah). "Great," I thought. "We can do this". I then talked to that city's long-term bicycle engineer who told me they were getting away from the mixing zones and would instead replace them with a design that extended their cycle tracks all the way up to the intersection, generally to the right of right-turn lanes. Copenhagen's reason for doing so? People bicycling a) felt uncomfortable being in a travel lane--even for a short distance, and b) because right turning motorists were inevitably held up by pedestrian crossings and cyclists, too were thus delayed.
Let me assure you: from an American cyclist perspective, these mixing zones were not uncomfortable. Danish motorists know well how to drive around people bicycling. Nonetheless, the Danish cyclists didn't like it and the Danish planners felt that it would deter increased growth in bicycle transportation.
What are the treatments they'll use at intersections where the bicycle lane extends all the way up to the crosswalk?: where there are many right turns they'll separate the movements with signals (if there is a dedicated right-turn lane); where there are fewer right turns (or the inability to create a right-turn lane) they'll stripe their big blue swaths through the intersection. I find it intriguing that even at this late date Copenhagen is still tinkering with significant elements in their bikeway design.
The Danish treatment is similar to the approach that has been adopted by Vancouver BC with their two-way cycle tracks. If there are a lot of right turns (more than 100 during the peak) they'll use signals to separate the movements. If there are "very few" right turns, they'll disallow the movement. Inbetween, they'll paint green through the intersection as a visual warning.
Portland, like cities across the world, is constantly looking for the sweet spot that incorporates safe operating conditions with the sense of comfort and security that people need if they are to use a bicycle for urban transportation.
If we could separate right turning cars and through bikes in time with signals at every intersection, we would. All that requires is space and money... I imagine that perhaps the principal reason why there are so few bike boxes in Utrecht is that most of the intersection conflicts have been addressed through signalization.
December 6, 2012 6:23 PM
Dan Berkman Says:
NYC has mixing zones at intersections on it's separated cycle paths on 1st and 2nd Avenues. I ride these cycle tracks everyday and I hate the mixing zones. There are tons of issues:
NYC DOT installed signs at every intersection that says "turning vehicles must yield to bikes." There's a little yield symbol and a bike symbol to drive the point home. And yet, nobody yields....ever.
Every intersection with a mixing zone is a game of chicken to see who will slow first. I see too many cyclists trying to get in front of turning cars to force drivers to stop and too many cars racing to beat the cyclists to the turn. The thing to do is take the "outside" part of the zone to avoid the right/left hook. Unfortunately, doing this puts cyclists near fast moving through traffic.
Theoretically, a little yielding should create a nice mix, allowing the staggering of vehicular and bicycle traffic, but what usually happens at busy intersections, is that cars back up waiting for the crosswalk to clear. Sometimes the cars back up to the point where it's easy to get boxed in on the inside of the mixing zone with no way through the line of cars snaking through the intersection.
NYC does have other bike lanes where there's a split phase signal that allows bike to progress forward while stopping cars from making turns through the cycle track. I do prefer this configuration, but it makes the green signal for bikes shorter since it now has to be split with turning cars.
December 7, 2012 9:32 AM
Nick Falbo Says:
Here's a far-out idea: use a speed hump in the auto-entrance to the mixing zone.
Speed humps can be used in slip lanes and entrances to roundabouts to encourage slower travel and proper yielding. This is precisely what we need from drivers in mixing zones, and it might give bicyclists the upper hand when negotiating for control of the lane.