Examples of Turns

After writing about turn sizes abstractly, I want to examine some real intersections.  First is the standard Kirkland (and elsewhere) residential intersection.  Kirkland uses wide residential streets (24, 30, and sometimes even 36 feet) and 25 foot curb radii on them.  Parking spots near intersections are rarely used; it’s illegal to park closely and there’s no reason to do so with tons of other parking on almost every street.  If we assume that 10 foot lanes lead to turns closely corresponding to the curb radii, then these lane sizes leave an extra 2, 5, and 8 feet.  Further assuming that both streets are the same width (so, for example, a 24 foot street meeting another 24 foot street), this translates to possible turning radii increases of 7, 17, and 27 feet!  Even the more common middle size pushes that 25 foot curb into the 40 foot classification – suitable for “moderate” speed car turns, or in other words not slow ones.  If the cars encroach a bit into the oncoming lane, these numbers get even bigger.

Next is our local favorite, NE 70th St and 130th Ave NE, and specifically the left turn onto northbound 130th.  The center turn lane is 12′, leaving some extra space.  130th is 30′ across, adding a lot of extra space.  Worse, it’s quite common for drivers making this turn to greatly cut the corner.  I’ve seen cars fully on the stop bar.  Here’s a drawing (map from Bing) with the curb (black), a fairly proper turn track(blue), and an observed turn track (red).  These are intended as tracks of the inside of the turning car.

turns-70-130

It’s interesting that when the corner can’t be cut (blue track), even being far from the curb doesn’t allow a larger turn.  Experience matches this observation; when a car is fully past the stop line (which is necessary for visibility), this turn must be taken pretty slowly.  However, cutting the corner greatly increases the turn size.  A very rough measurement suggests that the 30 foot curb radius turns into a 50 foot effective turning radius.  Again experience matches this; cars on this path are moving quite quickly.  And this is one reason why I prefer to walk up the west side of 130th without a sidewalk over crossing to the sidewalk on the east side.

Last is the site of a Seattle SUV/bicycle collision almost a month ago, Dexter and Thomas:

There was a lot of talk after this one, but for my purposes here I want to draw attention to this image from the video (direct YouTube link for the same video):

dexter-thomas-left-turn

This driver is cutting the corner exactly as described above in the Kirkland intersection.  This leads to a higher speed, which leaves less time to see the cyclist and less time to do anything about it.  It looks like the driver may have sped up and turned more sharply.  With more time, stopping would have been a more viable option.

Fixes

As in the previous post, the simple solution of smaller lane widths for the first scenario is a good one but difficult to achieve politically.  The other two examples are abuses of an additional lane.  Let’s add a simple pedestrian island to the Kirkland example:

turns-70-130-fix

We need to be careful to provide a reasonable but slow left turn path, but with any placement of an island, the red path becomes more obviously ridiculous.

The same thing could be done over in Seattle.  Here’s the same image with a simple median:

dexter-thomas-left-turn-fix1

It would probably need a sign too, but the modification of the vehicle path would be pretty significant.  I think even better would be to push the bike cross back from the intersection:

dexter-thomas-left-turn-fix2

This modifies the vehicle path at least as much and creates a perpendicular crossing of the bike path.  It essentially applies the protected intersection design to a left turn rather than the usual right turn.

I’ve never been to this intersection, so someone with more familiarity would need to turn this into a serious design.  I’ll add, though, that if these roads are too busy or too fast for a treatment such as these, then the problems and dangers extend far past just this intersection design.

Unintended Large Turns

In this post I write about turns at intersections that end up larger than intended.  First I need a bit of background on turning radii, but feel free to skim or skip if you’re already familiar with turn design.

Background

Sizes are generally chosen based on the size and speeds of vehicles to be accommodated.  So, larger turns support larger vehicles and higher speeds.  Smaller turns are cheaper, lead to simpler intersections, and are safer for pedestrians (due to both slower vehicle speeds and shorter pedestrian crossings).  One interesting point is that if large vehicles will be present but infrequent, then on quieter roads it’s reasonable to expect them to encroach on the lane in the other direction.  This gives access to large vehicles, trading an infrequent short delay for the advantages of smaller turns.  Here is a chart of the operational characteristics of various sizes (source: Intersection Channelization Design Guide, NCHRP Report 279, 1985, via FHWA:

Corner Radius Operational Characteristics
<5 Not appropriate for even P-design vehicles
10 Crawl-speed turn for P vehicles
20-30 Low speed turn for P vehicles; crawl-speed turn for SU vehicles with minor lane encroachment
40 Moderate speed turn for P vehicles; low-speed turn for SU vehicles with minor lane encroachment
50 Moderate-speed turns for all vehicles up to WB-50

(P=passenger vehicle, SU=single unit truck, WB=semitrailer, the number is the length in feet)

Roughly every 10 feet added to a turning radius moves it up a classification.

It’s a constant battle to ask cities for smaller turning radii.  This is partially due to an incorrect “larger is safer” belief (the same goes for lane widths).  It is also due to pressure from emergency responders (generally fire) for large vehicle access, which is ironic because we then make our roads less safe in the name of safety.  Going further into those arguments is beyond the scope of this post except that it’s worth pointing out that it’s perfectly reasonable for an emergency vehicle to encroach on the opposite lane for a residential turn (even fully, not just a bit) and that it’s certainly possible to design our emergency vehicles to fit safer streets rather than build dangerous streets around large vehicles.

Accidental Size Increases

My real aim with this post is to discuss the difference between the curb radius and the effective turning radius for an intersection and how this leads to larger turns than intended.  The effective turning radius is based on the path that cars actually take.  If there isn’t extra room, then a vehicle closely follows the path of the curb.  The turn will be slightly larger since the vehicle is going around the curb and not on top of it, but the amount is small.

Turn near curb

Turn near curb

For the rest of the post, I will ignore this difference in order to simplify the drawings.

When one side of the turn is wider than necessary, the effective turning radius is larger.  It turns out that the amount is the same as the additional width.  This can be seen in the following diagram:

Turn with one buffer

Turn with one buffer

When both sides of the turn are wider than necessary, the effects combine in a non-linear way.  Instead of increasing the effective turning radius by two times the additional width (one for each side), it turns out that the turn is increased by approximate 3.5 times the additional width!  This is because the vehicle is able to swing in and out of the turn, as shown here:

Turn with two buffers

Turn with two buffers

The math gets more complicated if the additional widths are different from each other, so I won’t go further into it.  We can approximate the effects or calculate numerical results for specific examples (see Wolfram|Alpha).

The next question is whether any of this matters.  Or, do turns have additional width in practice?  I see three causes of this extra space:

  1. The most obvious reason for extra width in a turn is simply a wide lane.  Passengers vehicles are under 7 feet wide with some under 6 feet.  (Examples: Toyota Prius is ~5’10”, Toyota Sequoia is ~6’8″)  Safe streets advocates generally ask for 10 foot lanes in residential/pedestrian areas, though in practice the lanes are often 11′, 12′, or even 14′ wide.  With 10 foot lane, the vehicle will have 1-2′ on each side, so there isn’t room to swing around and we’ll have that ideal scenario from above.  As the lane gets wider, however, there is extra space to be used for increased speed.  This space is present in most intersections, and it adds on to any space from the subsequent reasons.
  2. A less obvious reason is bike lanes or shoulders.  With a standard 5′ bike lane on both sides of the turn, the effective turning radius is increased by about 17 feet, or 1-2 categories in the opening chart!
  3. The last reason is the abuse of a second lane.  When turning onto a two lane road, turning into the far lane greatly increases the turning radius.  Left turns (or right in the UK, AU, etc) will be wider since they are further from the curb, but they are far worse if the vehicle can “cut the corner” by using the oncoming lane for a bit.  Compare the two movements here:

    Two left turn movements

    Two left turn movements

Fixes

We can directly address these issues.  There are other good reasons for smaller lane widths, and this adds one more.  Protected intersections put the curb between the car lane and the bike path, removing the additional space from being used on a turn.  Abuse of additional lanes is a bit harder to address.  Multiple lanes in the same direction shouldn’t exist in pedestrian or bike friendly areas, but they are very difficult (politically) to remove once present.  As far as encroachment in the opposing lanes, simply painting the center line can have good effect.  (I don’t have any sources handy, but I’ve heard that a center line does wonders for reducing speeds outside of intersections by focusing drivers onto their half of the road.  Edit:  I’ve also heard the opposite – that they suggest a high speed corridor.  Unfortunately the only study I’ve seen combined the removal of center lines with bike lanes and a narrow car travel area.)  A stronger solution is some kind of median, which has the added benefit of being a pedestrian island.