|This report is an archived publication and may contain dated technical, contact, and link information|
Publication Number: FHWA-RD-99-074
A Safety Evaluation of UVA Vehicle Headlights
This analysis will focus on the percentage reduction in crashes required to recover the increased cost of fluorescent roadway delineation and auxiliary ultraviolet (UVA) headlights. There are currently no studies of crash reductions due to UVA vehicle headlights that would allow us to compute a "true" cost/benefit ratio. Instead, we computed the estimated costs of implementing UVA /fluorescent technology and compared these to the costs of crashes that could potentially be reduced using this technology. We then computed the reduction in crashes needed to pay for the intervention costs, what we call the "break-even point."
This analysis used costs for UVA/fluorescent devices provided by the Federal Highway Administration (FHWA). Highway striping paint costs are $0.36/m ($0.11/ft) for regular paint and $0.49/m ($0.15/ft) for fluorescent paint. UVA headlights would be an additional $50 per vehicle. The number of vehicle registrations were obtained from Highway Statistics 1994 (FHWA).
The miles striped were computed using data for U.S. lane miles and U.S. road miles, excluding local roads (FHWA). The combination of these will give us the total line miles to be striped. This includes lane markings and centerline markings. We computed total relevant U.S. line miles, which includes lane lines and centerlines, to be 1,753,815 km (1,089,800 mi).
Table 8 summarizes the intervention costs of UVA/fluorescent technology using the above estimates.
We identified crash geometries where the use of UVA headlights might reduce the frequency or severity. Because UVA headlights will not reduce daylight crashes, they were excluded from our analysis. The lighting conditions included were dark; dark, but lighted; dusk; and dawn. We found six crash types that would potentially benefit from the use of UVA headlights at night. These crash geometries included: (1) crashes where a pedestrian or bicyclist was involved; (2) crashes occurring in a construction or maintenance zone; (3) crashes occurring on entrance and exit ramps of interstates; (4) single-vehicle roadway-departure crashes; (5) two-vehicle opposite-direction crashes where collisions include head-on and offset frontal; and (6) sideswipe crashes with both vehicle traveling straight, where a driver fails to hold his/her lane.
To estimate injury incidence, we followed the procedure in Miller, Lestina, and Spicer (1998), which is similar to Blincoe and Faigin (1992), Miller and Blincoe (1994), Miller, Galbraith et al. (1995), and Blincoe (1996). We began with a sample of nonfatal motor vehicle-related injury counts of relevant crashes by sampling strata and police-reported injury severity from NHTSA's 1996 General Estimates System (GES) (NHTSA, 1994). GES does not contain the information on body region injured and Maximum Abbreviated Injury Scale (MAIS) needed to apply costs. NHTSA's Crashworthiness Data System (CDS) (NHTSA, 1995) and National Accident Sampling System (NASS) (since renamed the National Automotive Sampling System, NHTSA, 1987) describe these crash injury details. We used 1988-1991 CDS for the description of injuries to passenger vehicle occupants involved in tow-away crashes. The most recent medical description available of non-CDS nonfatal crash victims (passengers of vehicles other than towed passenger vehicles) came from 1982-1986 NASS. Multi-year data were needed to obtain large enough samples of injury victims by body region and MAIS severity to accurately determine the incidence of rare injuries such as paralyzing spinal cord injuries. We used 1996 GES data to weight CDS and NASS data so that they would represent the 1996 nonfatal injury total. Since the years the CDS and NASS data were collected, many changes have come about to influence passenger and non-occupant traffic safety, such as airbags and increased seat-belt use. In developing GES weights to apply to these data, we controlled for police-reported injury severity, age of the victim, and restraint use (belted, unbelted, unknown, in a child seat). Thereby, we had a hybrid CDS/NASS file with weights that summed to estimated annual GES nonfatal incidence by police-reported injury severity, restraint use, and vehicle type. Fatality counts came from the 1996 Fatal Analysis Reporting System (FARS) (NHTSA, 1994).
We then estimated the total cost of these crashes. Costs per crash-involved person by MAIS and body region were merged in the file. The crash costs used are described in Miller et al. (1997), Miller (1997), and Blincoe (1996). Total crash costs include direct costs, which are actual dollar expenditures related to crash injury and damage. Direct costs include amounts spent for hospital, physician, rehabilitation, prescriptions, and related medical costs. Also included in direct costs are short-term work loss, employer productivity loss, travel delay, and cost to repair or replace damaged vehicles and property. Also included are the costs attributed to quality-adjusted life years (QALYs) lost. Table 9 presents total crash cost by crash geometry.
* Crashes include those in lighting conditions other than daylight.
Only UVA headlights would be required in order to increase the visibility of pedestrians and possibly reduce crashes. We found that a 9.6-percent reduction in nighttime pedestrian crashes will pay for the additional headlight costs. Alternatively, a 3-percent reduction in all relevant nighttime crashes will pay for the costs of UVA headlights and fluorescent highway paint combined. We also looked at other combinations and found that a 5.5-percent reduction in nighttime pedestrian crashes and a 2-percent reduction in the remaining relevant nighttime crashes will pay for the costs of UVA headlights and fluorescent highway paint. At effectiveness levels above the break-even point, the benefit/cost ratio rises linearly with effectiveness. For example, a 6-percent effectiveness of UVA /fluorescent technology corresponds to a benefit/cost ratio of 2, while a 9-percent effectiveness level results in a benefit/cost ratio of 3.
The evaluation of UVA/fluorescent technology found that detection and recognition distances of pedestrians and bicyclists increased by 33 percent to 117 percent. Unfortunately, in the current literature, a relationship between increased detection and recognition distances and crash reduction is not shown. However, if we assume that pedestrian crashes are reduced by 17 percent (half of the value of the lowest recognition improvement, i.e., 33 percent), we would achieve a benefit/cost ratio of 1.8 for UVA headlights. Table 10 presents the benefit/cost ratio for nighttime pedestrian crash reductions between 10 percent and 50 percent.
in nighttime pedestrian crashes.
Improvements in detection and recognition were also found for roadway delineation. These improvements ranged from 14 percent for right-curve delineation to 48 percent for a no-passing zone. Again, a literature search for the crash effects of improved delineation detection found nothing. If we assume that nighttime crashes, including incidents where pedestrians are involved, are reduced by 7 percent (half of the value of the lowest recognition, i.e., 14 percent) we would achieve a benefit/cost ratio of 2.4 for UVA headlights and fluorescent striping. Table 11 presents the benefit/cost ratio for the different scenarios of reductions in nighttime crashes.
by percentage of reduction in nighttime crashes.*
TRT Terms: Pavements--Performance--Mathematical models, Pavement distress, Pavement performance, Bias (Statistics), Precision