Safety improvement

The decision to install a roundabout as a safety improvement should be based on a demonstrated safety problem of the type susceptible to correction by a roundabout. A review of crash reports and the type of accidents occurring is essential. Examples of safety problems include:

• High rates of crashes involving conflicts that would tend to be resolved by a roundabout (right angle, head-on, left/through, U-turns, etc.);
• High crash severity that could be reduced by the slower speeds associated with roundabouts;
• Site visibility problems that reduce the effectiveness of stop sign control (in this case, landscaping of the roundabout needs to be carefully considered); and
• Inadequate separation of movements, especially on single-lane approaches.

Chapter 5 should be consulted for a more detailed analysis of the safety characteristics of roundabouts. There are currently a small number of roundabouts and therefore a relatively small crash record data base in the U.S. Therefore, it has not been possible to develop a national crash model for this intersection type. Roundabout crash prediction models have been developed for the United Kingdom (3). Crash models for conventional intersections in the United States are available (4, 5). Although crash data reporting may not be consistent between the U.K. and the U.S., comparison is plausible. The two sets of models have a key common measure of effectiveness in terms of injury and fatal crash frequency.

Therefore, for illustrative purposes, Exhibit 3-5 provides the results of injury crash prediction models for various ADT volumes of roundabouts versus rural TWSC intersections (6). The comparison shown is for a single-lane approach, four-leg roundabout with single-lane entries, and good geometric design. For the TWSC rural intersection model, the selected variables include rolling terrain, the main road as major collector, and a design speed of 80 km/h (50 mph). Rural roundabouts may experience approximately 66 percent fewer injury crashes than rural TWSC intersections for 10,000 entering ADT, and approximately 64 percent fewer crashes for 20,000 ADT. At urban roundabouts, the reduction will probably be smaller.

Also for illustration, Exhibit 3-6 provides the results of injury crash prediction models for various average daily traffic volumes at roundabouts versus rural and urban signalized intersections (6). The selected variables of the crash model for signalized (urban/suburban) intersections include multiphase fully-actuated signal, with a speed of 80 km/h (50 mph) on the major road. The 20,000 entering ADT is applied to single-lane roundabout approaches with four-legs. The 40,000 ADT is applied to double-lane roundabout approaches without flaring of the roundabout entries. In comparison to signalized intersections, roundabouts may experience approximately 33 percent fewer injury crashes in urban and suburban areas and 56 percent fewer crashes in rural areas for 20,000 entering ADT. For 40,000 entering ADT, this reduction may only be about 15 percent in urban areas. Therefore, it is likely that roundabout safety may be comparable to signalized intersections at higher ADT (greater than 50,000).

These model comparisons are an estimation of mean crash frequency or average safety performance from a random sample of four-leg intersections from different countries and should be supplemented by engineering judgment and attention to safe design for all road users.