Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
56 Crash Data Analyses to Investigate Contributing Factors This section of the report presents crash data analyses con- ducted to investigate the key factors contributing to driver workload identified in Section 4 and to confirm their role in median-related crashes. The key contributing factors addressed are ramps, sharp curves, and steep grades. 5.1 Crash Analysis Approach The investigation of the key contributing factors was nec- essarily conducted with a much larger set of sites and crashes than was available for the interdisciplinary field reviews pre- sented in Section 4. The interdisciplinary field review sites, which were selected to be studied in detail, included 97.4 miles of divided highways that experienced 1,261 crashes in 5 years. This new crash data analysis was conducted using data for 5 years for the entire rural freeway system in Washington. Although the analysis of interdisciplinary field review sites intentionally focused on sites with high crash frequencies, this analysis looked at the entire freeway system, including sites with high, average, and low crash frequency. Analyses addressed 923.56 miles of directional rural freeway segments that experienced 3,080 median-related crashes in 5 years. The FHWA HSIS data for Washington were used for this analy- sis because this is the only available database with curve and grade data for the entire roadway system. The 5-year period used for the analyses was the same 5-year period (2004 to 2008) used for the Washington interdisciplinary field reviews in Section 4 of this report. The analyses compared the crash frequencies and crash rates for freeway sections adjacent to off-ramps, adjacent to on-ramps, on sharp curves (with radii of 4,500 feet or less), on steep grades (with grades of 4 percent or more), and with none of these features. Combinations of features (ramps and sharp curves, ramps and steep grades, and sharp curves and steep grades) also were considered. Freeway segments adjacent to off-ramps were defined as extending from 0.05 miles upstream of the beginning of the deceleration lane to 0.19 miles (1,000 feet) downstream of the gore area. Freeway segments adjacent to on-ramps were defined as extending from 0.05 miles upstream of the gore area to 0.19 miles downstream of the end of the acceleration lane. The downstream area was included because there might be ramp-related turbulent flow in this area that could lead to median-related crashes. Sharp curves were defined as curves with radii of 4,500 feet or less. The freeway system was divided, based on curvature, into the following categories for analysis: ⢠Curve to left with radius less than 2,000 feet; ⢠Curve to right with radius less than 2,000 feet; ⢠Curve to left with radius from 2,000 feet to less than 3,000 feet; ⢠Curve to right with radius from 2,000 feet to less than 3,000 feet; ⢠Curve to left with radius from 3,000 to 4,500 feet; ⢠Curve to right with radius from 3,000 to 4,500 feet; and ⢠No sharp curve. The limiting value of 4,500 feet for radius of curvature was chosen because this represents a curve that would require a 4 percent superelevation rate on a 70 mph horizontal curve designed with a maximum superelevation rate (emax) of 6 per- cent (50). Consideration of several ranges for curve radius less than 4,500 feet allowed assessment of which ranges should be included in design guidance. Steep grades were defined as grades of 4 percent or more. The freeway system was divided into the following categories for analysis: ⢠Downgrade of more than 5 percent, ⢠Downgrade of 4 to 5 percent, ⢠Upgrade of more than 5 percent, ⢠Upgrade of 4 to 5 percent, and ⢠No steep grade. S E C T I O N 5
57 Generally, the Interstate Highway System in the United States has been designed with a maximum grade of 3 percent, except that grades up to 6 percent are permitted in rolling or mountainous terrain (54). The maximum grade for a 70 mph Interstate freeway in rolling terrain is 4 percent. Therefore, grades steeper than 4 percent were considered in this analysis. Grades above 4 percent were subdivided into categories of 4 to 5 percent and more than 5 percent for assessment of which ranges should be included in design guidance. The next section presents the results from analysis of these curve and grade data. Although no analysis was conducted to determine the statistical significance of the differences shown, the results are meaningful because the data include all crashes that occurred on the entire rural freeway system of a state dur- ing a 5-year period. 5.2 Analysis Results for Median-Related Crashes on Rural Freeways Table 5-1 presents a comparison of median-related crash frequencies and crash rates for freeway segments adjacent to and not adjacent to ramps. The table shows that the median- related crash rates for segments adjacent to off- and on-ramps are both higher than the crash rates for segments where there is no ramp present. Off-ramps have slightly higher median- related crash rates than on-ramps, although the observed dif- ference is minimal when the comparison is based on crash rate per million vehicle-miles of travel. Table 5-1 confirms that both off- and on-ramps are associated with elevated median- related crash rates and are confirmed as contributing factors to median-related crashes. Table 5-2 presents a comparison of median-related crash frequencies and crash rates for freeway segments by curvature categories. The table shows that median-related crash rates are highest for the sharpest curves (radius less than 2,000 feet) and that median-related crash rates decrease as the curve radius increases. Median-related crash rates are lowest of all for freeway segments where no sharp curve is present; such sections may be either tangent sections or curves with radius greater than 4,500 feet. However, there is very little difference in median-related crash rates for freeway segment curves in the 3,000 to 4,500 feet radius category and freeway segments with no sharp curve. Therefore, a decision was reached that design guidance for sharp horizontal curves should apply to curve radii less than 3,000 feet. For a given curve radius cat- egory, the difference in median-related crash rates between curves to the left and curves to the right is small, although curves to the right have slightly higher crash rates than curves to the left. This result is logical, because motorists are most likely to run off the outside of sharp curves and the median is on the outside of a curve to the right. Table 5-3 presents a comparison of median-related crash frequencies and rates for freeway segments by grade catego- ries. The table shows that median-related crash rates are high- est for the steepest grades (more than 5 percent), lower for 4- to 5-percent grades, and lowest for grades less than 4 percent. Based on this finding, a decision was reached to include both the 4 to 5 percent grade category and the more than 5 percent grade category in design guidance for steep grades. The differ- ences in median-related crash rates between downgrades and upgrades are small, particularly for fatal-and-injury median- related crashes. Table 5-4 shows the median-related crash frequencies and rates for specific combinations of ramp and curve catego- ries. The results in Table 5-4 generally confirm the results presented in Tables 5-1 and 5-2. There do not appear to be any circumstances in which specific combinations of ramps and curves have substantially higher or lower crash rates than suggested by their independent effects. However, it should be noted that the sample sizes for specific ramp and curve com- binations are generally very small. Table 5-5 shows the median-related crash frequencies and rates for specific combinations of ramp and grade categories. The results in Table 5-5 generally confirm the results pre- sented in Tables 5-1 and 5-3. There do not appear to be any circumstances in which specific combinations of ramps and grades have substantially higher or lower crash rates than suggested by their independent efforts. However, as for the analysis of ramp and curve combinations, the sample sizes for specific ramp and grade combinations are small. The sample sizes for specific combinations of sharp curves and steep grades are generally too small to provide meaningful results. Ramp category Freeway segments Number of median- related crashes (2004â2008) Total crash rate FI crash rate Cumulative length (mi) Exposure (MVMT) Total FI PDO per mi per year per MVMT per mi per year per MVMT Off-ramp 92.91 2544.00 404 144 260 0.870 0.159 0.310 0.057 On-ramp 69.76 1983.92 289 105 184 0.829 0.146 0.301 0.053 No ramp present 760.89 17344.07 2387 903 1484 0.627 0.138 0.237 0.052 All segments combined 923.56 21871.99 3080 1152 1928 0.667 0.141 0.249 0.053 Table 5-1. Crash frequencies and rates by ramp category for rural freeways in Washington.
58 Curve category (radius/direction) Freeway segments Number of median- related crashes (2004â2008) Total crash rate FI crash rate Cumulative length (mi) Exposure (MVMT) Total FI PDO per mi per year per MVMT per mi per year per MVMT Radius < 2,000 ft (left) 12.27 295.81 145 57 88 2.363 0.490 0.929 0.193 Radius < 2,000 ft (right) 12.27 295.81 155 67 88 2.526 0.524 1.092 0.226 Radius 2,000 to < 3,000 ft (left) 20.49 494.91 104 40 64 1.015 0.210 0.390 0.081 Radius 2,000 to < 3,000 ft (right) 20.49 494.91 108 53 55 1.054 0.218 0.517 0.107 Radius 3,000 to 4,500 ft (left) 23.45 526.72 79 25 54 0.674 0.150 0.213 0.047 Radius 3,000 to 4,500 ft (right) 23.45 526.72 66 26 40 0.563 0.125 0.222 0.049 All curves < 2,000 ft 24.54 591.62 300 124 176 2.445 0.507 1.011 0.210 All curves 2,000 to < 3,000 ft 40.98 989.82 212 93 119 1.035 0.214 0.454 0.094 All curves 3,000 to 4,500 ft 46.90 1053.44 145 51 94 0.618 0.138 0.217 0.048 All left curves 56.21 1317.44 328 122 206 1.167 0.249 0.434 0.093 All right curves 56.21 1317.44 329 146 183 1.171 0.250 0.519 0.111 All sharp curves 112.42 2634.88 657 268 389 1.169 0.249 0.477 0.102 No sharp curve 811.14 19237.11 2423 884 1539 0.597 0.126 0.218 0.046 Table 5-2. Crash frequencies and rates by curve category for rural freeways in Washington. Grade category (percent grade/direction) Freeway segments Number of median- related crashes (2004â2008) Total crash rate FI crash rate Cumulative length (mi) Exposure (MVMT) Total FI PDO per mi per year per MVMT per mi per year per MVMT 5%+ downgrade 14.80 256.66 87 27 60 1.176 0.339 0.365 0.105 4% to 5% downgrade 24.68 499.26 117 37 80 0.948 0.234 0.300 0.074 5%+ upgrade 14.80 256.66 71 27 44 0.959 0.277 0.365 0.105 4% to 5% upgrade 24.68 499.66 102 39 63 0.827 0.204 0.316 0.078 All 5%+ grades 29.60 513.32 158 54 104 1.068 0.308 0.365 0.105 All 4% to 5% grades 49.36 998.92 219 76 143 0.887 0.219 0.308 0.076 All steep downgrades 39.48 755.92 204 64 140 1.033 0.270 0.324 0.085 All steep upgrades 39.48 756.32 173 66 107 0.876 0.229 0.334 0.087 All steep grades 78.96 1512.24 377 130 247 0.955 0.249 0.329 0.086 No steep grade 844.60 20359.75 2703 1022 1681 0.640 0.133 0.242 0.050 Table 5-3. Crash frequencies and rates by grade category for rural freeways in Washington.
Ramp/curve combination (ramp type/radius/ direction) Freeway segments Number of median- related crashes (2004â2008) Total crash rate FI crash rate Cumulative length (mi) Exposure (MVMT) Total FI PDO per mile per year per MVMT per mile per year per MVMT Off-ramp/< 2,000 ft (left) 0.59 22.85 9 5 4 3.051 0.394 1.695 0.219 Off-ramp/< 2,000 ft (right) 1.19 40.22 27 9 18 4.538 0.671 1.513 0.224 Off-ramp/2,000 to < 3,000 ft (left) 1.82 38.31 12 6 6 1.319 0.313 0.659 0.157 Off-ramp/2,000 to < 3,000 ft (right) 2.58 65.94 15 8 7 1.163 0.227 0.620 0.121 Off-ramp/all < 2,000 ft curves 1.78 63.07 36 14 22 4.045 0.571 1.573 0.222 Off-ramp/all 2,000 to < 3,000 ft curves 3.01 78.53 39 15 24 2.591 0.497 0.997 0.191 Off-ramp/no sharp curve 86.73 2376.68 341 116 225 0.786 0.143 0.267 0.049 On-ramp/< 2,000 ft (left) 1.41 38.79 20 8 12 2.837 0.516 1.135 0.206 On-ramp/< 2,000 ft (right) 1.11 35.43 18 9 9 3.243 0.508 1.622 0.254 On-ramp/2,000 to < 3,000 ft (left) 2.29 63.04 19 4 15 1.659 0.301 0.349 0.063 On-ramp/2,000 to < 3,000 ft (right) 1.85 59.16 12 4 8 1.297 0.203 0.432 0.068 On-ramp/all < 2,000 ft curves 2.52 74.22 38 17 21 3.016 0.512 1.349 0.229 On-ramp/all 2,000 to < 3,000 ft curves 3.40 98.47 37 13 24 2.176 0.376 0.765 0.132 On-ramp/no sharp curve 63.10 1787.50 220 80 140 0.697 0.123 0.254 0.045 Table 5-4. Crash frequencies and rates by ramp and curve category combinations for rural freeways in Washington. Ramp/grade combination (ramp type/percent grade/direction) Freeway segments Number of median- related crashes (2004â2008) Total crash rate FI crash rate Cumulative length (mi) Exposure (MVMT) Total FI PDO per mile per year per MVMT per mile per year per MVMT Off-ramp/5%+ downgrade 0.00 0.00 0 0 0 -- -- -- -- Off-ramp/5%+ upgrade 0.39 9.35 12 5 7 6.154 1.283 2.564 0.535 Off-ramp/4% to 5% downgrade 1.10 21.80 8 1 7 1.455 0.367 0.182 0.046 Off-ramp/4% to 5% upgrade 0.74 21.06 7 3 4 1.892 0.332 0.811 0.142 Off-ramp/all 5%+ grades 0.39 9.35 12 5 7 6.154 1.283 2.564 0.535 Off-ramp/all 4% to 5% grades 1.84 42.86 15 4 11 1.630 0.350 0.435 0.093 Off-ramp/no steep grade 92.52 2534.65 392 139 253 0.847 0.155 0.300 0.055 On-ramp/5%+ downgrade 0.06 21.16 17 5 12 56.667 0.803 16.667 0.236 On-ramp/5%+ upgrade 0.23 3.20 1 1 0 0.870 0.313 0.870 0.313 On-ramp/4% to 5% downgrade 0.92 28.76 3 0 3 0.652 0.104 0.000 0.000 On-ramp/4% to 5% upgrade 1.19 26.26 8 3 5 1.345 0.305 0.504 0.114 On-ramp/all 5%+ grades 0.29 24.36 18 6 12 12.414 0.739 4.138 0.246 On-ramp/all 4% to 5% grades 2.11 55.02 11 3 8 1.043 0.200 0.284 0.055 On-ramp/no steep grade 69.47 1959.56 271 99 172 0.780 0.138 0.285 0.051 Table 5-5. Crash frequencies and rates by ramp and grade category combinations for rural freeways in Washington.
60 5.3 Comparison of Results for Median-Related Crashes to Other Crash Types Table 5-6 summarizes the crash data for rural freeways in Washington by crash type (median-related crashes, run-off- road right crashes, and on-road crashes). The table shows that the overall proportions of total crashes, ramp crashes, and grade crashes for these crash types are very similar (roughly 28 percent median-related crashes, 28 percent run-off-road right crashes, and 44 percent on-road crashes). The corre- sponding proportions for fatal-and-injury (FI) crashes on ramps, steep grades, and all site types combined are roughly 32 percent median-related crashes, 38 percent run-off- road crashes, and 30 percent on-road crashes. There is a more even distribution of crash types for total crashes on sharp curves (roughly 34 percent median-related crashes, 32 percent run-off-road crashes, and 34 percent on-road crashes) than for ramps and steep grades. The proportions for FI crashes on sharp curves (roughly 36 percent median- related crashes, 29 percent run-off-road crashes, and 35 percent on-road crashes) vary slightly more than for total crashes on sharp curves. Furthermore, it is evident that, as the curve radius decreases, the proportion of median- related crashes increases and the proportion of on-road crashes decreases, for crashes of all severity levels combined, as follows: ⢠Curve radius from 3,000 to 4,500 feet (25 percent median- related crashes); ⢠Curve radius from 2,000 to less than 3,000 feet (33 percent median-related crashes); and ⢠Curve radius less than 2,000 feet (40 percent median- related crashes). Table 5-7 compares crash rates by crash type for crash rate per mile per year and crash rate per million vehicle-miles of travel on rural freeways in Washington. Type of segment Number of crashes by crash type (2004â2008) Percent of crashes by crash type Median related ROR right On road Total Median related ROR right On road Ramp Category Off ramp 404 391 730 1525 26.5 25.6 47.9 On ramp 289 297 488 1074 26.9 27.7 45.4 No ramp present 2387 2442 3788 8617 27.7 28.3 44.0 All segments combined 3080 3130 5006 11216 27.5 27.9 44.6 Curve Category Radius < 2,000 ft (left) 145 101 111 357 40.6 28.3 31.1 Radius < 2,000 ft (right) 155 142 94 391 39.6 36.3 24.0 Radius 2,000 to < 3,000 ft (left) 104 85 91 280 37.1 30.4 32.5 Radius 2,000 to < 3,000 ft (right) 108 122 124 354 30.5 34.5 35.0 Radius 3,000 to 4,500 ft (left) 79 80 116 275 28.7 29.1 42.2 Radius 3,000 to 4,500 ft (right) 66 99 135 300 22.0 33.0 45.0 All curves < 2,000 ft 300 243 205 748 40.1 32.5 27.4 All curves 2,000 to < 3,000 ft 212 207 215 634 33.4 32.6 33.9 All curves 3,000 to 4,500 ft 145 179 251 575 25.2 31.1 43.7 All left curves 328 266 318 912 36.0 29.2 34.9 All right curves 329 363 353 1045 31.5 34.7 33.8 All sharp curves 657 629 671 1957 33.6 32.1 34.3 No sharp curve 2423 2501 4335 9259 26.2 27.0 46.8 Grade Category 5%+ downgrade 87 84 129 300 29.0 28.0 43.0 4% to 5% downgrade 117 149 169 435 26.9 34.3 38.9 5%+ upgrade 71 83 123 277 25.6 30.0 44.4 4% to 5% upgrade 102 100 156 358 28.5 27.9 43.6 All 5%+ grades 158 167 252 577 27.4 28.9 43.7 All 4% to 5% grades 219 249 325 793 27.6 31.4 41.0 All steep downgrades 204 233 298 735 27.8 31.7 40.5 All steep upgrades 173 183 279 635 27.2 28.8 43.9 All steep grades 377 416 577 1370 27.5 30.4 42.1 No steep grade 2703 2714 4429 9846 27.5 27.6 45.0 Table 5-6. Comparison of crash type distributions by ramp category, curve category, and grade category.
61 Type of segment Crash rate per mile per year (2004â2008) Crash rate per MVMT (2004â2008) Median- related crashes ROR right crashes On- road crashes Total crashes Median- related crashes ROR right crashes On- road crashes Total crashes Ramp Category Off ramp 0.870 0.842 1.571 3.283 0.1588 0.154 0.287 0.599 On ramp 0.829 0.851 1.399 3.079 0.1457 0.150 0.246 0.541 No ramp present 0.627 0.642 0.996 2.265 0.1376 0.141 0.218 0.497 All segments combined 0.667 0.678 1.084 2.429 0.1408 0.143 0.229 0.513 Curve Category Radius < 2,000 ft (left) 2.363 1.646 1.809 5.819 0.490 0.341 0.375 1.207 Radius < 2,000 ft (right) 2.526 2.315 1.532 6.373 0.524 0.480 0.318 1.322 Radius 2,000 to < 3,000 ft (left) 1.015 0.830 1.201 3.045 0.210 0.172 0.249 0.630 Radius 2,000 to < 3,000 ft (right) 1.054 1.191 1.210 3.455 0.218 0.247 0.251 0.715 Radius 3,000 to 4,500 ft (left) 0.674 0.682 0.989 2.345 0.150 0.152 0.220 0.522 Radius 3,000 to 4,500 ft (right) 0.563 0.844 1.151 2.559 0.125 0.188 0.256 0.570 All curves < 2,000 ft 2.445 1.980 1.671 6.096 0.507 0.411 0.347 1.264 All curves 2,000 to < 3,000 ft 1.035 1.010 1.205 3.250 0.214 0.209 0.250 0.673 All curves 3,000 to 4,500 ft 0.618 0.763 1.070 2.452 0.138 0.170 0.238 0.546 All left curves 1.167 0.946 1.245 3.359 0.249 0.202 0.266 0.717 All right curves 1.171 1.292 1.256 3.718 0.250 0.276 0.268 0.793 All sharp curves 1.169 1.119 1.251 3.539 0.249 0.239 0.267 0.755 No sharp curve 0.597 0.617 1.061 2.275 0.126 0.130 0.224 0.480 Grade Category 5%+ downgrade 1.176 1.135 1.743 4.054 0.339 0.327 0.503 1.169 4% to 5% downgrade 0.948 1.207 1.370 3.525 0.234 0.298 0.339 0.871 5%+ upgrade 0.959 1.122 1.662 3.743 0.277 0.323 0.479 1.079 4% to 5% upgrade 0.827 0.810 1.264 2.901 0.204 0.200 0.312 0.716 All 5%+ grades 1.068 1.128 1.703 3.899 0.308 0.325 0.491 1.124 All 4% to 5% grades 0.887 1.009 1.317 3.213 0.219 0.249 0.325 0.794 All steep downgrades 1.033 1.180 1.510 3.723 0.270 0.308 0.394 0.972 All steep upgrades 0.876 0.927 1.413 3.217 0.229 0.242 0.369 0.840 All steep grades 0.955 1.054 1.461 3.470 0.249 0.275 0.382 0.906 No steep grade 0.640 0.643 1.049 2.332 0.133 0.133 0.218 0.484 Table 5-7. Comparison of crash rates by type of segment and crash type for rural freeways in Washington.
62 ⢠For sites with curves, the percentage of median-related crashes on wet and snow-covered pavements increases as the curve radius decreases. ⢠Sites with steep grades experience a higher percentage of median-related crashes on wet or snow-covered pavement than sites not on steep grades. 5.5 Discussion of Contributing Factors The crash analysis for rural freeways in Washington con- firmed that the following contributing factors, identified in the interdisciplinary field reviews, are, in fact, overrepresented in median-related crashes: ⢠On- and off-ramps; ⢠Sharp curves (with radii of less than 3,000 feet); ⢠Steep grades (4 percent or more); and ⢠Wet and snow-covered pavement conditions. Other contributing factors, including combinations of the factors noted here, did not have sufficient sample sizes for their contributions to be formally tested. Type of segment Percent of crashes on wet or snow- covered pavement by crash type Median related ROR right On road Ramp Category Off ramp 68.1 62.7 39.3 On ramp 58.8 57.9 37.5 No ramp present 61.8 59.7 33.3 All segments combined 62.3 59.9 34.6 Curve Category Radius < 2,000 ft (left) 82.1 63.4 51.3 Radius < 2,000 ft (right) 70.3 75.4 57.5 Radius 2,000 to < 3,000 ft (left) 69.2 60.0 45.3 Radius 2,000 to < 3,000 ft (right) 48.2 58.2 28.8 Radius 3,000 to 4,500 ft (left) 73.4 57.5 36.2 Radius 3,000 to 4,500 ft (right) 50.0 57.6 29.6 No sharp curve 61.0 59.1 33.6 Grade Category 5%+ downgrade 73.6 61.9 51.9 4% to 5% downgrade 71.8 56.4 31.1 5%+ upgrade 77.5 74.7 37.4 4% to 5% upgrade 73.5 82.0 39.1 No steep grade 60.8 58.7 33.7 Table 5-8. Comparison of percent of crashes on wet or snow-covered pavement by crash type for rural freeway crashes in Washington. 5.4 Involvement of Road Surface Condition in Median-Related Crashes and Other Crash Types The percentage of crashes on wet and snow-covered pave- ments was compared for median-related crashes and other crash types on rural freeways in Washington. The climate in Washington includes both extremely wet and extremely dry regions, and rural freeways are found in both environments. Table 5-8 shows that median-related and run-off-road right crashes include a substantially higher percentage of crashes on wet or snow-covered pavements than on-road crashes (62 and 60 percent vs. 35 percent). This clearly implies that loss of con- trol on wet or snow-covered pavements has a more important role in crashes in which a vehicle leaves the road than in on- road crashes. For median-related crashes, the table indicates that ⢠The percentage of median-related crashes on wet or snow- covered pavements is slightly higher for off-ramps than for on-ramps or locations where no ramp is present.
