In 2012 the Fatality Analysis Reporting System (FARS) showed that rural intersection collisions accounted for 16% of all fatalities nationwide, with almost 30% of those fatalities occurring at night. The crash rate in Minnesota is slightly lower but still shows a 10% fatality rate for intersection crashes at night for those intersections without roadway lighting. Intersection lighting has been identified as an effective mitigation strategy for reducing nighttime collisions. Intersection lighting illuminates the rural intersection areas providing drivers with additional visual information prior to making a turn decision. The roadway lighting also provides drivers access to otherwise low contrast information when approaching a rural intersection. While rural intersection lighting has been connected to a reduction in rural intersection nighttime crashes to some extent, limited information is available about how the quality or quantity of that roadway lighting effects nighttime crashes. Lighting warrants and lighting level recommendations are provided by the American National Standard Institute and the Illuminating Engineering Society (e.g., ANSI/IES, RP-8-14) and a number of other national and statewide reports. These recommendations suggest a minimum amount of horizontal illuminance required at intersection locations (and for other roadway lighting applications) that allows drivers sufficient illumination for object visibility. The horizontal illuminance was utilized as a method to provide sufficient and appropriate lighting levels at intersection locations. A number of other parameters are included when calculating an appropriate lighting design for an intersection including, pavement classification, glare, luminaire type, headlights, and foliage. These recommendations provide an excellent foundation for establishing new lighting at intersections, yet little is understood about the benefits of any amount of roadway lighting at intersection locations. The objective of this study was to identify isolated rural intersections of interest, build a lighting data collection system, take lighting measurements at these intersections, and then analyse that data with respect to the factors within the crash data. Finally, the recommendations from the analysis would provide some insight into lighting levels for rural intersections and future research directions given the data analysed. A number of steps were implemented to achieve these efforts: 1) Surveying county engineers about intersection lighting 2) Identifying intersections where data collection could be completed 3) Identifying and building a data collection system 4) Collecting horizontal illuminance data at those intersections 5) Comparing nighttime crash ratios using illuminance data 6) Providing results and conclusions A county engineering intersection lighting survey was created and then administered to all county engineers within the Minnesota. Of the 87 counties, a total of 45 county engineers responded by completing the survey. The survey asked the country engineers (or designates) if they maintained any intersection lighting, what type of lighting they maintained, estimate of the cost of installing and maintaining the intersection lighting, questions about any recent lighting projects, and whether they had any specific intersections that would be appropriate for horizontal illuminance measurement. Of the total counties that responded, 27% of them concluded they did not currently maintain any rural intersection lighting. The remaining respondents identified approximate totals of lighting, installation, maintenance, and additional costs associated with rural intersection lighting. The respondents also provided the researcher with a number of locations where intersection lighting measurements could be taken. These suggestions were accumulated and then compared with the crash database information to identify appropriate intersections of interest. During the intersection identification process, a crash database was used to examine the number of crashes at the intersection location, the status of the intersection lighting (e.g., lighted or unlighted), intersection geometry (e.g., near a curve), and intersection configuration (e.g., ‘T’ or crossroad). A final list of intersections was generated prior to building the data collection equipment. Upon review of the intersections a number were removed to streamline the data collection process. In addition, a number of intersections were added around the greater metro area in an effort to reduce data collection time and expedite the project. A final list of intersections from a total of six different counties identified 63 intersections of interest to be measured by the data collection system. To collect the required horizontal illuminance measurements in a reasonable amount of time using an effective method, the researcher identified an effective measurement apparatus as previously created by the Center for Infrastructure Based Safety Systems at the Virginia Tech Transportation Institute. The data collection device utilizes a number of Minolta illuminance meters that are positioned on the roof of a vehicle in such a manner as to collect horizontal illuminance data while the vehicle is in motion. The data collection technique allows a researcher to collect data at multiple intersections without having to stop the vehicle and take measurements using a hand held device. The technique also allowed the researcher to take horizontal illuminance measurements from all approaches for each of the rural intersection locations in order to collect as much data as possible. The device utilized a Trimble R7 Global Positioning System attached to the vehicle to map the data to the actual location of the intersection. After data collection was complete, the researcher cleaned and checked the data for errors before combing the information into an overall data file. The data was then analysed using a number of negative binomial regression models that reviewed all the data, just those intersections that contained roadway lighting, and just those intersections that had no lighting installed. The results showed a significant benefit of horizontal illuminance in reducing nighttime crash risk. The first negative binomial model that used lighted and unlighted intersection data showed that an increase in 1-lux (0.1 fc) from the average illuminance measured (3.91 lux) had a 9% reduction in the nighttime crash ratio. For a second model only looking at lighted intersection data, a 1-lux (0.1 fc) increase in lighting level resulted in a crash ratio reduction of 20%. Finally, a third model used only unlighted intersection data. A 1-lux (0.1 fc) increase in the average illuminance (0.20 lux/0.01 fc) resulted in a crash reduction ratio of 94% for unlighted intersections. The unlighted intersections were essentially increasing the light level to a lighted point or threshold as defined in the current project, therefore the benefits are higher. Contrary to other research, ‘T’ intersections were identified as providing reductions in the nighttime crash ratio (e.g., 200%) compared to crossroad intersections. Also, those intersection locations that were near a curve were found to have reduced nighttime crash ratios (e.g., 178%) compared to those that were not located near a curve. A number of interactions also occurred between the predictor variables that influenced the impact of intersection type and also the average horizontal illuminance. The research concluded and re-confirmed the beneficial nature of roadway lighting at isolated rural intersections. Beneficial levels increase above the average illuminance levels as defined in each model. For combined lighted and unlighted intersections studied, increases of 1-lux (0.1 fc) from the average of 3.91 lux (~0.36 fc) reduced crash rate ratios by 9%. For lighted intersections alone the increase of 1-lux (0.1 fc) to the average (6.41 lux/0.6 fc) reduced crash rate ratios by 20%. However, the largest impact came from installing lighting at unlit intersections. Finally, future research beyond installing lighting at unlighted intersections should identify a baseline for lighted isolated intersections. The average obtained in this study was 6.41 lux (0.6 fc) across lighted intersections. A level/benefit ratio should be researched that can provide the greatest impact and minimal lighting levels required. For example, future research could investigate minimum horizontal illuminance levels, perhaps starting at 5-lux (0.5 fc) as a baseline for testing nighttime visual performance and for comparisons with crash rates. Additional research would grow the current illuminance database and provide validation for both minimum and maximum horizontal illuminance, vertical illuminance, and luminance levels at rural intersection locations. A final recommendation is to investigate the impact of newer roadway lighting technologies (e.g., Light Emitting Diode) as a method to reduce lighting levels but still maximize driver visual performance. Newer technologies can also provide cost savings at the reduced levels in addition to maintenance and longevity benefits, which may in turn counteract any high upfront costs. (Author/publisher)
Samenvatting