White Paper SeriesRoad Diet Conversions:A Synthesis of Safety ResearchMay 2013Libby Thomas, Senior Associate, UNC HSRCFor:Federal Highway AdministrationDTFH61-11-H-00024www.pedbikeinfo.org
This material is based upon work supported by the Federal Highway Administration under CooperativeAgreement No. DTFH610110H-00024. Any opinions, findings, and conclusions or recommendations expressedin this publication are those of the Author(s) and do not necessarily reflect the view of the Federal HighwayAdministration.SummaryThe primary purpose of this review is to assess the available evidence regarding the safety effectiveness of reductionsin the number of motorized traffic lanes, widely known as road diet conversions. Although road diets have beenimplemented since at least the 1970s, earlier reviews and a search of the literature identified no controlled safetyevaluation studies conducted prior to the year 2002. A systematic search of literature dating from 2002 was conducted.Six studies in total were initially identified, with four serving as the basis for most conclusions in this review. Several ofthe studies have used overlapping data from many of the same implementation sites, with the more recent studiesemploying the more robust study methodologies. As a result, the strongest evidence comes from relatively few studiesbuilding on earlier ones. However, a sizeable number of sites have been encompassed in the studies. Studies using datafrom sites in California, Iowa, and Washington provide the strongest evidence of safety effects, with additional reportsproviding corroborating, but somewhat weaker evidence.Road diets can be seen as one of the transportation safety field’s greatest success stories. Total crashes might beexpected to decline by an average of29 percent by converting from four, undivided lanes to three lanes (plus other usessuch as bike lanes). Additionally, the studies determined total crash reductions were higher (47 percent) for treatedsections of more rural thoroughfares passing through smaller towns (Iowa sites) and lower (19 percent) for road dietcorridors in large urban areas (California and Washington sites) (Harkey et al., 2008).Thus far, only a single study from New York City has examined effects on pedestrian crashes (Chen et al., 2013).Although the researchers were unable to use the most robust methodology due to a lack of traffic volume data, theinclusion of 460 road diet sites and large number of comparison locations supports the findings of significant reductionsin total crashes, significant reductions in injurious and fatal crashes, and a trend of lower pedestrian crashes at segments.Total crashes and injurious crashes also declined significantly at intersections abutting the road diet sections.Each potential road diet should be vetted on a case by case basis. Case study and modeling results suggest that addedcaution is warranted before implementing road diets when volumes approach 1,700 vehicles per peak hour or are in therange of 20,000 to 24,000 vehicles per day (HSIS, 2010; Knapp and Giese, 2001; Welch, 1999). However, high qualitydisaggregate estimates of safety effects of road diets for different volume roadways are lacking. Further study ofpotential traffic and safety effects on surrounding roads and access from side streets to the road diet corridor may alsobe needed.
Introduction and Purpose of this ReviewRoad diets, also known as road conversions, are the reallocation of road space through reduction of the number ofmotorized traffic lanes. They are of interest to communities that may be seeking to smooth traffic flows and reducetraffic speeds, improve access management, reduce crashes, improve safety and accessibility for pedestrians or bicyclists,improve parking utilization, improve economic function of the street, or in general gain space for uses more compatiblewith the purposes of the street. Some streets that were built with peak flows in mind, or because the thinking at the timewas that more lanes are better, may have excess capacity for most periods of the day, and roadway space that is not wellutilized. The intent of this review was to assess the available evidence regarding the safety effects of road dietconversions.In addition to the safety effects, roadway managers are interested in knowing what traffic volume and roadwayconditions are most amenable to treatment, while still providing adequate mobility. In general, road diets have beendescribed as maintaining motor vehicle capacity for converted roads with average daily traffic volumes (AADTs) up toapproximately 20,000 vehicles per day under most conditions tested. Road diets may also provide safety and mobilitybenefits to other modes of travel, including bicycling and walking (HSIS, 2010).As mentioned, the reallocation of space can further goals for pedestrian safety and mobility. Even if the reallocatedlane space is not used directly for pedestrian facilities, the use of space to add parking or bicycle lanes increases thebuffer between motorized traffic and pedestrians walking along the sidewalk. Bike lanes may also add to the bufferbetween motorized traffic and fixed objects such as trees and street furniture along the roadside. The fewer number ofmotorized lanes associated with a four-lane to three-lane conversion, or two lanes with median, also reduces pedestrians’exposure to traffic when crossing streets. A center, two-way left turn lane (TWLTL), or median islands also providespace for pedestrian refuge. More capital-intensive conversions include curb realignments or the addition of centermedians or median islands. If curbs are realigned, space may be allocated to green space or other buffers, or to increasesidewalk width.The most commonly studied reallocations of space have been conversions of undivided, four-lane roads into threelanes (one through lane in each direction plus TWLTL), and typically involve reallocation through striping or rechannelization. Many conversions include the addition of dedicated bike lanes in both directions as part of the spacereallocation, a change in parking space allocation, or the addition of both parking and bike lane space. Therefore, thesetypes of road diets rely on relatively inexpensive re-striping, perhaps in association with re-paving. Such actions can bevery cost-effective if combined with regular maintenance activity. Conversions may also incorporate signal timing orphasing changes at intersections to optimize operations and safety benefits. Some communities are also combining roaddiets with roundabout intersection designs.Background on Safety and Operational EffectsWelch (1999), in a paper presented to the Transportation Research Board (TRB) and Institute of TransportationEngineers’ (ITE) Urban Street Symposium in 1999 reported on observations that led to the first four-to three-laneroadway conversions in Iowa. Prior to the mid-1980s, it was common practice to widen two-lane urban arterials to fourlane, undivided roads if traffic volumes exceeded 6,000 vehicles per day. According to Welch, “At public hearings,3
project engineers would state that corridor safety would improve if the two-lane roadway were widened to a four laneundivided roadway. Graphics would be shown to illustrate that additional acceptable gaps in the traffic stream wouldresult, and motorists could avoid rear-end collisions by changing lanes and going around slowing/stopping vehicles.Those in opposition to the widening would argue that travel speeds would increase, pedestrians would have to cross awider street, and noise would increase. In most cases, however, the four-lane undivided cross-section was selected as thepreferred alternative because the only other alternative was generally to do nothing (i.e.: the roadway remains a two-lanefacility).”Welch’s Table 1 shows the actual trends he found following such a two-lane to four-lane conversion of a road withaverage traffic volumes of 10,000 – 14,000 vehicles per day (vpd). It highlights the slight increase in traffic volume, delay,and speed, with a more substantial increase in accident and injury rates, and total value loss.Table 1 –Table from Welch’s 1999 Report following a two-lane to four-lane conversionIn contrast, positive safety and operations effects were experienced by localities that converted some wide, two-laneroads to three-lanes (one narrower lane in each direction with a TWLTL), suggesting the possibility of four to three-laneconversions as a potential way to improve safety and provide acceptable mobility. Welch (1999) reported data from nineconversions in Seattle, Washington and found that volumes were maintained while crash numbers went down.Burden and Lagerwey (1999) provided traffic volume data for 18 road diets from four states and Canada, includingthe nine Seattle locations described by Welch. “Before” traffic volumes from the 18 road diets described in Burden andLagerwey were between 9,700 and 23,000. In each case except one, traffic volumes were maintained or increased afterthe conversion. Lake Washington Blvd., in Kirkland, WA, with an initial volume of 23,000 vehicles per day, increased tonearly 26,000 after the conversion, and during one period was carrying about 30,000 vehicles per day. The one exceptionwhere the after volume decreased somewhat carried an initial 20,000 vehicles per day, which dropped to 18,000 in theafter period suggesting the potential for traffic diversion to other routes. Huang, Stewart, and Zegeer (2002) alsomentioned two road diet examples where traffic volumes decreased on the treated streets and increased somewhat on4
nearby routes: A conversion of Polk St from three lanes to two lanes in San Francisco was followed by an ADT decreaseof 2% and ADT increases on two nearby parallel streets by 8% and 15% (actual volumes not reported). However, PolkStreet also saw an increase in bicycle traffic from 37 to 52 during the peak hour. The conversion of Valencia Street inSan Francisco also resulted in a 10% decrease in ADT, to 19,979, while the ADTs on four parallel streets increased by2% to 8%. The number of bicycles, however, more than doubled on Valencia during the peak hour, from 88 to 215.The City of Orlando, Transportation Planning Bureau carried out a variety of studies prior to a road diet conversionof Edgewater Drive. Following the conversion, traffic volumes dropped initially from 20,500 vehicles per day to 18,000,but then recovered to 21,000. Pedestrian volumes increased by 23 percent, and bicycle volumes increased by 30 percent.Parking use on Edgewater Drive also increased, and motorized traffic did not increase on nearby roads. All of theseoutcomes helped to meet eight of nine community objectives for the road diet conversion (Edgewater Drive – Before andAfter Re-striping Results, 2002).In the case study examples described above, it is unclear how long after the implementation that traffic volume orother data were collected, or whether volumes had stabilized in the after period.Knapp and Giese (2001) also carried out detailed simulation sensitivity analyses to attempt to confirm the reportedoperational impacts of four to three-lane road diet conversions. The results of the simulation analyses primarilyconfirmed the case studies assessments. Knapp and Giese (2001) noted that there may be some impacts to motorvehicles during peak periods of greater than about 1750 vehicles per hour. Simulation results indicated that a decrease inaverage arterial speed (ranging from 0 to 4 mph) for through vehicles would be expected for a three-lane configurationcompared with a four-lane design, given a large range of peak-hour volumes, access densities, and access-point left-turnvolumes (Knapp and Giese, 2001). The simulated arterial level of service for a converted roadway began to show adecrease when the bi-directional peak-hour volume was about 1,750 vehicles per hour (or 17,500 vehicles per day if 10percent of the daily volume is assumed to occur in the peak hour) (Knapp and Giese, 2001). The models found that athigher access point densities (around 40 - 50 points per mile), four-lane undivided roadways with high left turningvolumes begin to operate more like de-facto three-lane roadways in terms of speeds, as drivers avoid the left lanes onsuch four-lane roads.Considering past research by others, Knapp recommended that a four-lane undivided to three-lane conversion beconsidered as a feasible option (with respect to volume only) when bi-directional peak-hour volumes are less than 1,500vehicles per hour, but that some caution be exercised when the roadway has a bi-directional peak-hour volume between1,500 and 1,750 vehicles per hour. However, Knapp pointed out that the results are for one idealized simulation casestudy, which included optimization of signal timing to minimize vehicle delay along the corridor.Knapp and Giese (2001) also reported on 13 road diets in California, Montana, Minnesota and Iowa. A number ofsites had volumes of around 20,000 to 24,000 ADT. Observed crashes decreased at all of the sites reported on byKnapp and Giese (for which data were available, see p. 28). Note that the observed crash reductions could have beenaffected by random effects due to regression toward the mean, although such effects would not be expected at all sites.Over all, travel speeds decreased at three of 10 sites with speed data available. Average speed increased at one location.The largest effects were on high end speeders (more than 5 mph over the limit, or above the 85th percentile). Maximum5
85th percentile or average speed reductions noted were three to four miles per hour. However, well-controlled estimatesof crash reductions and operational data for different volume roadways were not provided.Burden and Lagerwey (1999) indicated that road diet conversions prior to 1999 maintained capacity by keeping thesame number of lanes (often as intermittent turn lanes/pockets) at intersections, where capacity and delay are