Online Newsletter of the UC Berkeley Traffic Safety Center: Volume 3, No. 2, Summer 2006
Putting Safety into Road Planning: a New Diagnostic Approach
Visitors from the Colorado DOT discuss their system to address safety in roadway design—a Traffic Safety Seminar presentation.
Roadway designs' effectiveness in moving traffic is traditionally measured by the Level of Service (LOS), which rates speed and ease of travel, volume of traffic that can be accommodated and other indicators of roads' convenience and efficiency for users. However, LOS is not connected to a roadway's safety.
The 1998 federal highway bill, the Transportation Equity Act for the 21st Century, TEA-21, required that safety be explicitly considered in the planning process (similar to the way that environmental impacts must be considered). However, much remains to be understood about the best way to measure safety and how to decide where money for safety improvements would be best spent.
In order to address this issue, the Colorado Department of Transportation has devised a system that expands the LOS ranking to include safety. They call it Level of Service of Safety (LOSS).
Two engineers involved in devising the system, Jake Kononov and Bryan Allery, from the Safety Engineering and Analysis Group of the Colorado DOT, explained key aspects of the new system at the Traffic Safety Seminar in April. Their talk was titled "Achieving Road Safety Through Highway Design."
 |
They demonstrated the system by showing how it worked on a two-lane mountain road leading to the town of Black Hawk, Colo. (map, at left). As a result of a 1991 state law permitting limited stakes gaming in Black Hawk and two other towns on Colorado's Front Range, casinos opened, creating a significant increase in traffic.
The introduction of gaming, and the traffic it created, presented an opportunity to test their new diagnostic method in "before" and "after" scenarios. The intention is to identify "sites with promise," locations where counter-measures will have the greatest impact on safety at the least cost.
Drinking and Driving and Gambling = Safety?
Kononov began by showing how traditional methods of computing accident risks would produce misleading results when applied to the entire 5.85-mile segment leading to the casinos. Using traditional accident rate figures, that is, the number of accidents per trip made on the segment, as traffic increased (along with alcohol-related crashes), the road's average crash rate improved dramatically.
From 1991 to 1994, the trip rate rose more than fourfold, that is the number of trips made per mile of road. Additionally, drivers who had been drinking rose substantially, as indicated by a rise in accidents where alcohol was involved. Despite these two trends that would suggest greater danger on that strip of road, the accident rate, the number of accidents per trip made, dropped by half. (Nothing had substantially changed about the road's design or construction.)
(Much of the presentation can also be found in a paper they wrote with Zane Znamenacek. A link to the pending version is provided in the sidebar.)
"Is drinking and driving in concert with gambling good for safety?" they asked rhetorically.
"This shows the fallacy of using accident rates without adjustment. We are not measuring safety," Kononov said.
The "adjustment" relates changes in the accident rate to changes in traffic exposure through statistical analysis. This relationship is reflected by the Safety Performance Functions (SPF) calibrated for various road types.
Development of the SPF lends itself well to the conceptual formulation of the Level of Service of Safety (LOSS). The concept of level of service uses qualitative measures that characterize safety of a roadway segment in reference to its expected performance.
The Colorado DOT has identified four types of roadway LOSS:
- LOSS-I, for roadways that have a low potential for accident reduction;
- LOSS-II, with a better-than-expected safety performance;
- LOSS-III, with a worse-than-expected safety performance, and
- LOSS-IV, for roads that have a high potential for accident reduction.
But, Kononov pointed out, a roadway's LOSS status can't be determined without knowing the "expected norm."
"There is no way to get a consensus on the safety of a road without a common measure," Kononov said. "Physicians agreed what normal blood pressure should be, but that's not true of traffic engineers when it comes to safety."
If the LOSS shows that the roadway would benefit from improvements, the next step is uncovering the causes. The Colorado DOT team has developed diagnostic techniques and a method for applying a pattern recognition algorithm to tease out contributing factors. They also use accident diagramming, along with site visits and reviews of design and construction plans for the segment.
Another example presented at the seminar examined a case history where the roadway overall seemed sufficiently safe (LOSS-II), but closer analysis suggested that there might be a potential for accident reduction.
The DOT team broke down the roadway into segments one-tenth of a mile long and analyzed those segments with the algorithm and patterning sequences. They found that 15 of the 18 accidents recorded happened at night. A site visit showed that the crash locations did not have adequate reflectivity devices and striping. After those relatively inexpensive improvements were made, the accident rate was cut substantially.
This approach can be used to make the most out of scarce safety dollars, they noted, as well as to garner support for improvements that might not otherwise seem necessary.
For additional information, see the links on the sidebar.