How to Design the Perfect Protected Intersection
A protected intersection is about more than just corner islands and two-stage left turns. A truly protected intersection is safer for all modes… Essential elements include the dedicated left and right turn lanes on all vehicle approaches and signal phasing that combines to eliminate all conflicts… Let’s pick it apart a bit…
I’ve passed through this intersection in Utrecht so many times, and for what is a pretty busy street with a lot going on, I've always felt completely safe riding through it in a way that isn’t common in North America. It’s complicated and that’s the point, but it has a lot of the key elements that make Dutch intersections so much safer for everyone, whether they’re driving, walking, or riding a bike… It’s not easy to capture all of that in one photo, so the plan view is shown below along with some key design elements… The photo above is taken on the west leg looking east…
I think that pretty much every single high level planning document I've ever read says we’ll prioritize safety above all else, but few intersection designs in North America actually apply that to every facet of intersection design and operation. Furthermore, it’s not possible to achieve everything at every intersection, but let’s get into what’s so good, and where some elements might not be possible…
Step 1: Allocate space for all modes first
If we’re serious about mode shift, safety, and equity, to name a few reasons often cited for intersection upgrades, and if we prioritize those things over vehicle capacity, then we start to make different choices in our designs. The basic requirements for the perfect signalized intersection that works for everyone should be:
Sidewalk on each side
Bike path on each side (unless bi-directional on one side makes more sense)
Vehicle left turn lane
Vehicle through lane
Vehicle right turn lane
Vehicle receiving lane
Maybe bus lanes
If we have space for a bus lane or even more vehicle lanes like we have in the example above after the basic requirements have been met, then go ahead, but only after all other needs have been accommodated. That might be a hard sell when existing intersection designs create a pattern of traffic that we think we have to accommodate now and into the future, but we don’t always; if there are alternatives, be that alternative routes, alternative modes, or even alternative times to make a trip, people will adjust, traffic evaporation is a real phenomenon…
In terms of space, and considering general North American minimums, we need 1.8m for the sidewalk on each side, I'd say 2.1m for the bike path on each side, 3.3m for each of the four vehicle lanes, that’s around 21 metres already, then throw in the curb and gutter, whatever separation is added between modes, for boulevards, median islands, or on-street parking that may also be provided. A right-of-way of 25-30 metres should accommodate such an intersection comfortably.
Step 2: Adopt No Right Turn on Red on all approaches
No Right Turn on Red (NRTOR) is the default in the Netherlands and many other countries. It used to be the norm in North America too until an oil crisis in the 70’s sought ways to save gas and NRTOR was one of those ways… Interestingly, the Netherlands used that same crisis to build out their bike network and reduce their reliance on the car. North America chose another path… But we are where we are.
NRTOR is becoming more common today, and hopefully one day we see more cities, provinces, or even the country re-adopt it as a standard. Until such time, it’s as simple as adding the NRTOR sign to each approach. NRTOR provides four primary benefits:
It improves safety by reducing conflicts between cars.
It improves safety by reducing conflicts between cars and people walking or using a mobility device.
It improves accessibility for people using a mobility device as drivers should be stopped behind the stop bar and therefore not blocking the crosswalk and curb ramp.
It improves safety by reducing conflicts between cars and people riding bicycles.
There may be some compliance issues early on, but hopefully over time, people get used to it. If compliance issues persist, I’ve seen some LED versions of the NRTOR sign, and something that could work even better would simply be a set of red, yellow and green turn arrows… It would be much more intuitive… The only problem, in BC at least, is that the Motor Vehicle Act acknowledges both green and yellow arrows, but not red like we see in that Dutch example above…
Step 3: Plan Signal Operations for Safety
When we have the above elements in place, it allows the intersection to operate in a way that removes all conflicts. There are a few key principles for signal operation:
Through traffic goes with through pedestrians and bicycles.
Left and right turns are both protected.
Basic signal stages and phases are typically as follows, albeit the order may vary, and stages may be skipped if demand is not detected…
This method of operation removes all conflict if people drive with care and to the law. It’s still possible that drivers will jump a red light, ignore a NRTOR, or get distracted by their phone and rear-end someone. A reminder that safe system design is not about eliminating all collisions, but those that are severe or fatal.
Step 4: Add-ons
Steps 1 to 3 get you to a pretty good place, but there’s more that can be done:
Countdown to green timers let people know their turn is coming, this is most well used in the Netherlands for people riding bicycles, but could equally apply to people walking and driving. When you know your turn is coming, you’re more likely to obey the signals.
Turn calming by forcing a tighter radius can slow turning traffic and improve safety. This is less important with the perfect intersection as conflicts such as walk with traffic are removed, but slower turning traffic is never a bad thing and supports that safe system strategy of reducing the likelihood of a serious injury or fatality if someone does something they shouldn’t.
Add bicycle lean rails to make the wait for green on a bicycle a bit more comfortable.
Directional pedestrian and bike push buttons allow for efficiencies, if the signal knows a person is only crossing in one direction, they can shorten the intergreen with a conflicting traffic phase. If the signal knows there are only people biking and not walking, it can skip the long walk time.
Adopt near side signals that discourage drivers creeping over the stop bar, keeping crosswalks car-free and accessible.
Take care to keep curbs low or tapered to eliminate the chances of clipping a pedal.
Add Tactile Warning Surface Indicators (TWSIs) to guide anyone with sight loss walking through an intersection.
Step 5: If you can’t do all that…
If it’s not possible to provide separate right turn lanes, and separate turning traffic from crossing pedestrians and cyclists, start to think about these mitigations to reduce conflicts:
Leading pedestrian and bike intervals can give people walking or on bicycles a head start on traffic.
Provide additional signage to better alert drivers. The Dutch use the flashing ‘Let Op’ sign, ‘Pay Attention’ in English, while Vancouver has an example of RT Yield to Cyclists sign with flashing LED’s activated by approaching cyclists.
Bend-out bike paths and crosswalks to improve sight lines between turning drivers and people walking and cycling.
This isn’t a dream…
We have at least one example of such an intersection in North America in Canmore, Alberta… Let’s do more…
I’ve posted this video before from a couple of years ago, but it might help with understanding of the signal phasing and removal of conflicts…
An update…
After the above post, there were a lot of comments and questions about how it operated, or even how hostile it was for pedestrians. I was trying to keep it high level and focussed on the key elements, the benefits of the separate turn lanes, the no-right-turn-on-red, and the signal phases that are key to the safe operation. I chose a large complex intersection for a reason which we’ll get into a bit below. Some additional thoughts…
The Crossings are Too Wide!
This comment came up a lot, and many people focused on the fact that people were crossing many lanes of traffic, which was considered too hostile an environment. I intentionally chose this intersection for a couple of reasons… I had the main photo from a previous trip demonstrating the right turners being held back while people biked through, but also because it showed that a large complex intersection could be completely safe for all users. The example is urban with a reasonable number of pedestrians and high numbers of cyclists passing through, all without conflict.
Relating this to North America, we have plenty of wide multi-lane intersections that are a horrible experience to walk or ride through. The design proposed above can make them safe, so showing it’s possible on larger complex intersections was kind of the point. If you can get away with fewer lanes, all good, but if we can make our larger intersections safer, why not?
Why are we prioritizing road design for motorists?
This somewhat related comment came up a few times too. Adding more space for vehicles must be bad, right? But the reality is it can allow for safer methods of operation. It’s not as simple as counting the number of lanes and saying we’re prioritizing motorists. With conventional methods of intersection operation, we ask pedestrians to cross the road while allowing traffic to turn across them. We allow traffic to turn right on red. It’s our existing methods of operating that treat pedestrians as an afterthought for the sake of vehicle capacity. Adding turn lanes might seem like it’s prioritizing vehicles, but in reality it’s enabling the signal phasing that makes the intersection safer for everyone.
Too much faith placed in drivers obeying signals?
Like any signal-controlled intersection, a driver can run a red light, be distracted, etc. This doesn’t really solve that issue, but it’s still a better way of operating compared to allowing drivers to make their own decisions on when it’s safe to turn in gaps or across pedestrians. So while not foolproof, it’s still much safer than typical North American methods of signal operation.
How much capacity does it lose?
It’s hard to answer this easily, and if I were, the answer would be… It depends… The actual answer depends on many things and it’s not necessarily a case of losing capacity… Are there multiple high, conflicting traffic volumes? Can some high traffic movements such as left and right turns operate together? How much pedestrian or bike traffic is there? If capacity were to be reduced, could traffic divert to another route, and does that route have capacity? Let’s consider a few scenarios:
Lane capacity is typically 1600-2100 vehicles per hour. Lower end if there’s more turning traffic or trucks. A conventional shared through and right turn lane restricts access to the right turn, and when vehicles get to the stop line, they might have to yield to pedestrians crossing, now blocking through traffic. One or two vehicles might squeeze through on a red if it’s permitted, but right turn capacity can be pretty low in this scenario, and this can impact through capacity, so this common configuration is not always that great for vehicle capacity.
The proposed configuration with a dedicated right-turn lane, phase, and no-right-turn-on-red the rest of the cycle might seem much worse, but unlike the scenario in the bullet above, this way of operating may actually improve right turn capacity by giving it a dedicated phase with no conflicts. You could argue the conventional way just struggles at peak times, and off-peak, it works better, but off-peak, capacity isn’t an issue in the first place, so it doesn’t hurt to wait a little if the primary goal is safety.
If the high right-turn can operate with a non-conflicting high left-turn, these two movements can operate together and take up a larger share of green time if other movements have less need for it.
Don’t forget that traffic volumes today are a product of the system in place today, and there is potential for patterns to change, for people to take other routes, or make different choices. That might be a hard sell politically, but should at least be considered when evaluating options.
Are there any examples of directional Pedestrian crossings?
What is a directional crossing? One where if you push the push button on one side, it only activates the walk signal for that direction. Most intersections today, if you approach a signal on foot and see someone push the button on the far side, you’d most likely assume you don’t have to also push it on your side. With directional crossings, you would have to trigger it for each direction. Taking beg buttons to the next level, you might say… I’m not advocating for this everywhere, but there may be instances where it helps sell the other other improvements…
There are no such directional crossings that I'm aware of in North America, and our signals aren’t really set up to make use of it. Our signal controllers are another area that really needs to evolve with these more complex intersections. Let’s explore the idea a little more anyway… The pedestrian phase has a few components… There is a walk phase that’s usually not enough to cross the entire road, say 7 seconds, then a flashing don’t walk phase which is long enough to cross the road if you just stepped onto the crossing at, say, the 6 or 7 seconds mark, let’s say that’s 14 seconds, but it varies with crossing distance, then potentially an all-red period between the pedestrian phase ending and the next conflicting vehicle phase starting. If the signals are smart enough, you can reduce the time for the pedestrian phase to the next vehicle phase once pedestrians are halfway across the street and out of the conflict zone with the next vehicle phase.
To make this easier to understand, let’s look at an example, let’s say there is a westbound vehicle left turn after the east and westbound pedestrian phase. If the traffic signal assumes a pedestrian could be crossing in both directions, it must provide the maximum flashing don’t walk signal just in case before allowing the left turn. However, if the signal knows there is only a pedestrian crossing eastbound, that pedestrian will be out of the conflict area sooner, and we can buy a little left turn capacity improvement for the left turn.
As I said above, I'm not suggesting this is always a great way of operating a signal in an urban environment where we should be more heavily prioritizing vulnerable modes, but I think in an area where pedestrian demand is less frequent, this way of operating could help convince some.
Can the same Outcome be achieved with Two-stage crossings?
Yes, sort of. The above example accommodates a one-stage crossing, which is obviously more convenient, whereas a two-stage crossing can be less convenient with more pedestrian delay, depending on how it operates. However, there are examples of two-stage crossings in North America, so this might be the way to achieve similar efficiencies.
In the example above, that westbound vehicle left turn may get its green arrow while pedestrians can cross the east side of the two-stage crossing, but the west side can only be called when the east and westbound through traffic is going. If there’s not much vehicle demand, we could end this phase plus the walk phase earlier and give capacity to other movements, so like the capacity question above, it depends…
The two-stage crossing can provide refuge for those slower to cross, which isn’t a bad thing either.
THe crossings need to be set back more!
This might not have been a comment on my article, but stems from a recent YouTube video about the Canmore intersection mentioned above and I thought it was worth speaking to here. The YouTuber praised the intersection for its safer operation, but kind of missed the key benefits of the intersection achieved with the separate turn lanes, signal phasing and no-right-turn-on-red, and suggested it could be improved with larger crossing setbacks.
Crossing setbacks are something that have been touted as a key requirement of protected intersections for safe operation for a while, especially in North American guidance, but these setbacks only provide a benefit when traffic turns across the pedestrian or bike phase. If there’s no conflict, there’s no need for the larger setbacks that are intended to improve sight lines in the conflict area.
Never stop improving…
I last saw the intersection in the main example above in mid June 2025, as I was sketching it up for the article at the end of 2025, I noticed the lines on the map seemed to show the north leg had been removed. Interestingly, what was a major four leg intersection on the Western City Boulevard became three, what was a four lane road plus turn lanes, was now just the turn lanes…
Utrecht is famous for reducing vehicle capacity and prioritizing other modes and this is just one more example. The work on the Western City Boulevard has been going on for several years now, working its way towards this intersection and reducing space for traffic with each step. I found the city’s construction update website (https://debouw.app/projects/westelijke-stadsboulevard-utrecht/updates) and pulled in a few of the photos showing the closure of the north leg and the intersection reconfiguration… I guess I need to go back…
Even the perfect protected intersection can be improved upon…