Comments to:  Draft of Recommendations of The Access Board on Pedestrian Crosswalks At Roundabouts

Dear Committee Members:

I am since ten years a professor of transportation engineering in the United States.  My training was in Sweden where I in 1982 presented my Ph.D.-thesis on Pedestrian Safety at Signalized Intersections.  I have worked on research relating to pedestrian safety for 25+ years and parallel to this on roundabout safety for 20+ years and would like to give some comments to your proposed guidelines.

It seems like you write that wherever marked (and possibly unmarked) pedestrian crosswalks are provided at roundabouts, each shall meet the requirements set forth in this section, including: (C) Signals. A pedestrian actuated traffic signal complying with Section … shall be provided for each segment of the crosswalk, including at the splitter island. Signals shall clearly identify which crosswalk segment the signal serves.

I will comment on this below, in connection to some direct comments to your discussion section. But first, in http://www.access-board.gov/rowdraft.htm#1106 you write, “Requiring the signal to be pedestrian activated may help limit the impact on traffic flow.”  In reality this may be true, but shouldn’t the responsibility of lawmakers include that the code be made to be followed?  Through education and/or enforcement activities if necessary?  As far as I know, pedestrians, in all U.S. states, have the right-of-way in unsignalized marked crosswalks.  In other words, the primary purpose of signalizing marked crosswalks should be to give automobile drivers the right-of-way part of the time so that automobile capacity does not become too low where pedestrian flows are high.  In Germany, and some other European countries, this is clearly understood and given as the primary reason for signalizing crosswalks.  Still, I acknowledge that we in the U.S. live in a country were many drivers do not stop for pedestrians in crosswalks, even for those carrying white canes, and I understand that we may have to ‘accommodate’ such illegal behavior, and have designs that make it reasonably safe for all pedestrians, including those in wheelchair or visually impaired, even when divers violate codes.

In Discussion (http://www.access-board.gov/prowac/commrept/part3-02-5.htm) you write:

“Modern roundabouts are ….  While this traffic pattern has been an asset to traffic planners in controlling and slowing the flow of traffic at intersections in lieu of having a signalized intersection, the absence of stopped traffic presents a major problem for blind and visually impaired pedestrians when crossing.”

I would like that statement to be backed by facts in the form of crash statistics.  If it were (only) a perceived “major problem” rather than an actual problem, then maybe education rather than engineering changes would be motivated.  I do know that the ‘sole’ serious opposition to roundabouts in Sweden today stems from visually impaired people and their advocacy groups, and I do not mean that this is not a very important subgroup of the pedestrian population, but still, sub-optimization of our traffic environment is one of the reasons that the risk of fatality per mile walked is about ten times higher in the United States than in Sweden, where roundabouts are utilized frequently in the urban environment.  I also know that signalized crosswalks or grade-separated passages are considered at roundabouts in Sweden, where there is a high demand by visually impaired pedestrians…  But, to require signalization of all roundabouts is, in my opinion, definitely unwarranted. At least, it is my opinion, that all crosswalk locations away from roundabouts should be signalized prior to the ones adjacent to single-lane roundabouts getting this type of control.

Rather, you should consider requiring signalization of marked crosswalks (or grade-separated crossings) at multi-lane locations including at multi-lane roundabouts with high pedestrian volumes.  I am fairly convinced that there is no crash data from the U.S. supporting the view that pedestrians are vulnerable to crashes at single-lane roundabouts.  The only pedestrian crash at a U.S. roundabout was, as far as I know, the elderly person hit in Montpelier, VT, and that did not cause any serious injury.  Rather, the roundabout prevented the injury.  There is statistics from Sweden, showing that ‘all’ the country’s (»700) single-lane roundabouts had a total of three pedestrian crashes (with not a single serious injury) in the 1994 to 1997 period.  (Today there are a lot more roundabouts in Sweden, but I do not have any newer statistics.)  If these locations had been signalized, there would have been at least 11 pedestrian crashes according to standard models.  However, the two-lane roundabouts studied, had an actual safety very similar to signalized locations (10.4 predicted and 12 occurred at the 14 locations in Sweden that have considerable pedestrian traffic.)

You write: “Barriers or similarly distinct elements are needed to prevent blind persons from inadvertently crossing a roundabout roadway in an unsafe location. … Because the pedestrian crosswalk is generally placed at least one car length from the entry point, in a location that is not immediately apparent to a blind or visually impaired pedestrian, a cue is needed for crosswalk location.”

Again, I have no objection to the idea that pedestrians are guided to safe crossing points, but crossing outside the crosswalk at a roundabout is probably safer than crossing anywhere away from a roundabout, so there should be million of miles of barriers put up prior to the ones at roundabouts.

You write, “Pedestrians report that vehicles at roundabouts, right slip lanes, and other unsignalized pedestrian crosswalks often do not yield for pedestrians. Pedestrians with disabilities are particularly vulnerable in these situations. People who are blind or visually impaired are unable to make eye contact with drivers making it impossible to 'claim the intersection.' The driver's view of people using wheelchairs is often blocked by other vehicles. Pedestrians with slower than normal mobility may hesitate when entering the street. All of these situations may result in drivers misinterpreting the pedestrian's intention to cross.”

I agree, but this is even truer away from roundabouts at non-signalized locations.

You write, “It is recognized, however, that the purpose of these types of unsignalized crosswalks is to keep traffic moving as continuously as possible.”

That is one reason for constructing roundabouts, but I have for over a decade advocated the use of roundabouts for the primary purpose of improving pedestrian safety.  That delays are reduces is a side effect rather than the primary purpose in my way of thinking.

You write, “Traffic flow can be achieved, while still affording pedestrians with disabilities the opportunity to cross safely, with the use of pedestrian actuated technologies that halt traffic only while the pedestrian is in the crosswalk. An advantage of passive detectors is that, when pedestrians cross slowly, more time can be automatically provided. When a pedestrian crosses quickly, the traffic is stopped only during the time the pedestrian is crossing, thereby eliminating the problem of traffic being held up when no pedestrian is in the crosswalk.”

I agree fully with this strategy.  Wherever it is economically feasible, I support the use of passive or active detection and stoplights.  But, again, roundabouts should not be the first place to implement such systems.

I recently studied the safety of pedestrians at over a hundred locations in Maine.  I counted pedestrian and vehicle volumes and predicted how many crashes there ought to have been if the layout was ‘typical’ (according to TRL models from England and VTI models from Sweden, which in parenthesis gave very similar results) and compared these estimates to actual crash experience involving pedestrians.  I found that the risk of a pedestrian collision is

- roughly 25 times the ‘average’ where pedestrians cross multi-lane streets at unmarked locations with 25 mph speed limits but actual speeds around 30 to 40 mph

- roughly 10 times the ‘average’ where pedestrians cross multi-lane streets in marked crosswalks with 25 mph speed limits but actual speeds around 30 to 40 mph

- roughly 4 times the ‘average’ where pedestrians cross multi-lane streets at unmarked locations with 25 mph speed limits but actual speeds around 25 mph

- roughly 2.5 times the ‘average’ where pedestrians cross multi-lane streets in marked crosswalks with 25 mph speed limits but actual speeds around 25 mph

- signalization of the above listed locations reduces the risks by roughly 50%

- roughly half the ‘average’ where pedestrians cross 2-lane streets as an average for all speeds if the street is posted as 25 mph

- extremely low risk where pedestrians cross 2-lane streets in marked or unmarked crosswalks with 25 mph speed limits but actual speeds around 20 mph

There were no multi-lane streets with actual speeds below 25 mph.

All the figures above include visually impaired people but are not specifically true for that group by itself.  In summary, 4- and 6-lane streets are very dangerous where speeds are high.  Signalization reduces the risk, but the risk is still much higher than at a narrow low-speed location, especially since many of the crashes still occur at very high speeds (jaywalkers or drivers running red lights account for over 50% of the pedestrian crashes). For example, the risk of a pedestrian collision is about 5 to 12 times higher than ‘average’ at a signalized 4-lane crosswalk if cars are driven at 30 mph whereas it is maybe 20% of the average in the vicinity of a single-lane roundabout (2-lane street) handling the same traffic volume.  That is a difference of roughly 50 times.  And that is risk of collision, not risk of serious injury or fatality.  The risk of fatality, for Maine streets and roads, vary as seen in the table here.  In other words, the roundabout may be more than 50 times safer than the signalization.

Table 1    Speed limit and crash severity, Maine data

In conclusion to my comments.  I may be wrong in my assessment that non-signalized crosswalks adjacent to roundabouts are very safe for visually impaired people.  And I do not want to advocate accepting collateral damage.  But, if the design procedures suggested here means that roundabouts will not be constructed, and this means that we will ‘keep’ signalization and see 500 additional pedestrian fatalities a year compared to if roundabouts were utilized, which would have led to (annually) one visually impaired person being killed at a roundabout, should we then celebrate the saving of that one life at a cost of 500?  Maybe?  But, what if I am correct, and there will not be any additional deaths among visually impaired people, and the result of this practice will be 500 more pedestrian fatalities and not a single saved life?  Then we should feel bad about our choice, shouldn’t we?  Especially since some of the 500 will be visually impaired people.

Roundabouts are not the only way of slowing down traffic.  There are other traffic-calming methods that can be used.  Unfortunately, the experience with signalization as a traffic calmer is not encouraging.  Even if the mean speeds are reduced, the top speeds are very high.  And some of those top speeds are found just after the perpendicular walk signal indicates a clear crossing.  And, what the roundabout has in its favor that most other traffic-calming measures don’t have is that it allows for narrow streets, something very important for elderly pedestrians’ safety. 

Now, as my last words, if the suggested design criteria lead to no reduction in the rate of constructing roundabouts, and the proposed signals are such that pedestrians have the absolute right-of-way both when the signal is activated and deactivated (as is the formal rule today if they go blank) then the signalization should cause no bad safety-effects and we would all be winners.  My concern is that the cost for such systems will be prohibitive, and the construction of roundabouts will be delayed. 

Thank you for listening to my thoughts,

Per Gårder,  Professor
Department of Civil Engineering
University of Maine

index    previous comment      next comment