Safety evaluation of acceleration and deceleration lane lengths

Institute of Transportation Engineers. ITE Journal, May 1999 by Bared, Joe, Giering, Greg L, Warren, Davey L

THE METHODOLOGY PRESENTED HEREIN ALLOWS A HIGHWAY PLANNER OR DESIGNER TO ASSESS THE BENEFITS OF IMPROVING SAFETY OF INTERCHANGES CONSIDERED FOR RECONSTRUCTION, OR WHEN EXPERIENCING HIGHER THAN AVERAGE ACCIDENT FREQUENCIES, WITHOUT REQUIRING EXISTING ACCIDENT RECORDS.

PRIOR RESEARCH REVEALED THE safety importance of acceleration and deceleration (accel/decel) lane lengths. In particular, deceleration lanes were shown to exhibit higher accident rates than acceleration lanes. A recent research report, entitled Statistical Models of Accidents on Interchange Ramps and Speed-Change Lanes,l documents the development of models relating accidents to ramp characteristics. This feature utilizes the information in this report to estimate accident frequencies for entire ramps, as a function of speed-change lane length, among other variables. An economic analysis procedure is then presented to evaluate the costeffectiveness of extending speed-change lanes. The methodology presented herein allows a highway planner or designer to assess the benefits of improving the safety of interchanges considered for reconstruction, or when experiencing higher than average accident frequencies, without requiring existing accident records.

SCOPE AND ASSUMPTIONS

The feature starts with a Background section that reviews research on the design of accel/decel lane lengths and compares them with the American Association of State Highway and Transportation Officials' (AASHTO) design criteria.2 Similarly, the second section, Safety Experience, reviews related accident analysis literature. The third section, Accident Models, describes the sample size, model form and ramp characteristics used in the formulation of the new model. The fourth section, Cost-Effectiveness of Design, explains a practical procedure to determine economic benefits of lengthening accel/decel lanes. Finally, the Conclusion highlights the advantages of estimating the impact of lengthening accel/decel lanes when the AASHTO criteria cannot be fully met or the proposed length exceeds the minimum requirement (e.g., high accident location).

The procedure presented herein makes the following assumptions:

The new accident models (using negative binomial regression) estimate a reasonable mean accident frequency for a ramp and its adjacent speedchange lane (in Washington state);

The relative effects of statistically significant variables (e.g., length of accel/decel lanes) on mean accident frequencies are reasonable for safety improvement applications and transferable to other states; and

Unit costs presented for construction and land, length of economic life, percent rate of return, accident costs and fixed average daily traffic (ADT) rates are acceptable estimates.

Although the length of the speedchange lane described in this feature varies from AASHTO's definition (refer to Figure 1), the safety impact (in terms of accident frequencies) of a differential change in length is still comparable. This statement is valid only when the taper length and length prior to the gore point (for entrance ramps) or past the gore point (for exit ramps) remain constant. In other words, only if the speed-change lane is lengthened in the area of overlap between the AASHTO and Bauer definitions (shown in Figure 1).

Similar to intersections, accident models for ramps can only explain a moderate percentage of the variation in the data (43 percent in this case). The remaining effects could be random and/or systematic variances due to variables (such as driver behavior) not represented in the models. Despite some reservations about using cross-sectional models (that establish cause and effect relationships), the current negative binomial model is one of the best available tools to evaluate the safety effects of lengthening speed-change lanes.

BACKGROUND

The criteria for the design of accel/decel ramps at interchanges have not changed in the last three decades. Table X-4 and Table X-6 (of AASHTO's 1990 Policy on Geometric Design of Highways and Streets2) provide minimum accel/decel lane lengths respectively. AASHTO defines the length of a speedchange lane as the distance where the lane is 12-feet (ft.) wide to where the ramp design speed is reached on a tangent, or to a curve with a radius less than or equal to 1,000 ft. (Figure 1). Most states use taper-type ramps at exits and parallel design for entrance ramps.3

An unpublished NCHRP Project 3354 report developed a new methodology for determining accel/decel lane lengths based on human factors, trafficflow characteristics and vehicle dynamics. These models divide the AASHTO-defined acceleration lane into areas where distinct driving tasks are performed: 1) initial acceleration and 2) position adjustment (i.e., gap search and acceptance). Likewise, the AASHTO-defined deceleration lane was divided into separate task areas: 1) deceleration in-gear and 2) deceleration while braking. Koepke3 compared AASHTO's minimum values for accel/decel lane lengths (on grades less than 2 percent) to the recommended design values from this new method. Table 1 compares these values for various freeway design speeds and equivalent ramp design speeds.