2003-09-25_PWETC_AgendaPacket - Laserfiche WebLink

Home

'I ;1 'I Ii JI j j~ I L I" i:i 14 The foregoing findings in actual roadway illumination show such extremes in HFC averages and uniformity (average-to-minimum) that serious doubt is cast on the value of such elements in lighting design. The erratic performance of systems certainly in- validates any analysis of fine differences among various designs. The subject research assignment was to investigate relatively small differences in lighting and to search for nominal design values. However J the extent of variations may be enough to "wash Dut" meaningful analysis. For example, a variation of 50 percent produces in an 0.8-HFC nominal system a range from 0.4 to 1.2 in area averages at random locations. Another system with doubled nominal iliumination of 1.6 HFC will vary from 0.8 to 2.4. This obviously overlaps well into the first system. In order to achieve reasonable separation, a threefold difference appears to be needed in nominal level (such as 0.6 versus 1.8 HFC). Extremes of this magnitude were gen- erally not located for study in the subject research. UGHTED VERSUS UNLIGHTED ROUTES As a group, the lighted freeways for which data are given in Table 1 had an average night-day ratio of 1.43 for all types of accidents. The unlighted freeway average was 2.37. If a freeway experiences 1,000 day accidents during any time period, the expected number of night accide'1ts can be calculated for any assumed night-day rate ratio. The lighted average ratio of 1.43 produces 475 night accidents by use of the equation E, = R &/3, where E, = expected number of night accidents, R = night-day rate ratio, and A. = number of day accidents. The unlighted average ratio of 2.37 produces 790 night accidents. The difference is 315, which represents 40 percent fewer night accidents. If compared with the 1,790 total of day and night accidents, the average overall reduction wonid be 18 percent. This example indicates that the illumination of an unlighted urban freeway could theoretically reduce night accidents by an aver'age of 40 percent or overall accidents by 18 percent. If only the fatal and injury accidents are considered, the night-day ratio is 1.69 for lighted freeways and 3.53 for unlighted ones. A lighted freeway with 1,000 fatal and injury accidents during the day wonid average 560 such accidents at night. If unlighted, however, there would be 1,180 such night accidents. The apparent effect of lighting on the fatal and injury accidents is a reduction of 52 percent in night accidents. The statistical analysis involved a chi-square test for association between accident rate ratios and the presence or absence of lighting. A total of 59 observations were available, 18 of which involved unlighted route sections. (An "observation" is 1 year of data from I section. Thus, 1 route with 3 years of data produces 3 observations.) The data were entirely adequate to allow the test to be performed. The test established that lighted freeways as a group have lower rate ratios for all types of accidents than un- lighted freeways. A highiy significant (1 percent) value of chi-square was found. This means that less than 1 possibility in 100 exists for a chance occurrence. Testing of only the fatal and injury accidents resulted in " similar finding that was significant at the 5 percent level (less than 5 possibilities in 100 exist for a finding due to chance alone). BEFORE-AND-AFTER STUDY In the Chicago area, an opportunity was found to compare accidents before and after relighting 5.3 miles of 6-lane urban freeway. Unfortunately, resurfacing and minor reconstruction along the route were commenced about 1 year after installation of the lighting. The after period of study was thus limited to only 1 year. Data were available for 2 years in the before period, however. The study route was Interstate..94 between 132nd Street and 167th Street. This route had a narrow median (12 ft) between 132nd Street and 146th Street. In the rest of the lighted section, the median width was 33 ft. Study findings from each section are given in Table 3. A direct reduction occurred during the alter period in all types of accidents. A large change in the number of acci- dents per MVM is also fOWld. However, the total number of accidents is less than 600 for all 3 years of data from bolli sections. ,.;