Median Cross-Section Design for Rural Divided Highways (2014)

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.

103 C H A P T E R 6 This chapter discuss the interpretation of the analysis results presented in Chapters 4 and 5 of the report and pre­ sents design guidelines based on those results. 6.1 Median Width In assessing median width effects from the crash analysis for rural divided highways, the results for rural four­lane free­ ways are of primary interest, because the smaller sample sizes for four­lane divided nonfreeways and six­lane freeways were too small to provide useful or consistent results. The fatal­ and­injury crash analysis results for rural four­lane freeways generally indicate that CMCs decrease with wider medians, while rollover crashes generally increase with wider medi­ ans. These two effects are of almost equal magnitude, but in opposite directions. The logical interpretation of this result is that, as median width increases, out­of­control vehicles have more opportunity to roll over before reaching the opposing roadway. Thus, the choice of an appropriate median width depends on a tradeoff between the likelihood of CMC and rollover crashes. The net result of the combined effects of median width on CMC and rollover crashes discussed above (along with the smaller effects on other crash types) is a slight increase in fatal­and­injury crashes as median width increases. The tradeoff between CMC and rollover crashes discussed above is strongly influenced by the difference in severity between these crash types, as shown in Table 6­1. The table shows that for CMCs, fatal crashes constitute 26.7 percent of fatal­and­injury crashes, while for rollover crashes, fatal crashes constitute only 9.0 percent of fatal­and­injury crashes. Thus, because wider medians lead to more of the less severe rollover crashes and fewer of the more severe CMCs, the research results suggest that generally wider medians should be preferred. The crash analysis results indicate that wider medians gen­ erally will have more crashes. But, as indicated in Table E­1 in Appendix E (available on the TRB website), there would be fewer severe crashes as the median gets wider, resulting in a net crash cost savings. The effect shown in Table E­1 might be even more pronounced if crash data that distinguished serious injuries from other injuries were available; however, such data were not available for sites under the jurisdiction of some, but not all, of the participating agencies. The crash analysis shows a monotonic relationship between crashes and median width, suggesting that CMCs would keep decreasing, and rollover crashes would keep increasing contin­ uously as the median width increases. The results of the vehicle dynamics simulation illustrated in Figure 5­24 show a more subtle interpretation of this relationship. Specifically, the vehi­ cle dynamics simulation analysis found that, at a median width in the range from 15 to 18 m (50 to 60 ft), there is a boundary at which the probability of a CMC becomes less than the prob­ ability of a rollover crash. This suggests that, when the lower severity of rollover crashes is taken into account, there are diminishing returns in continuing to make the median wider. Figure 5­24 shows that this boundary is itself a function of median slope; therefore, this effect is examined further in the next section, which addresses median slope effects. 6.2 Median Slope The crash analysis indicates that the median slope ratio also has opposing effects for CMC and rollover crashes, but that these opposing effects for median slopes are opposite to the effects for median width. The models for rural four­lane freeways in Table 4­7 show that higher median slope ratios (i.e., flatter slopes) are associated with more CMCs and fewer rollover crashes. Table 4­7 also indicates that flatter slopes on rural four­lane freeways are associated with fewer fixed­ object crashes. As in the case of median width, the crash anal­ ysis indicates a monotonic effect in which the observed trends continue across the full range of median slope ratios. The results of a benefit­cost analysis for flattened median slopes presented in Table E­2 in Appendix E (available on the TRB website) confirms that providing flatter slopes has a net Interpretation of Results and Design Guidelines

104 positive effect on safety. This effect is cost­effective, even though earthwork/grading costs increase. However, a supplementary analysis considering differences in the severity distributions between crash types found that flatter slopes still had a positive effect on safety, but the benefit­cost ratios were less than 1.0. The crash analysis and benefit­cost results should be taken only as a general indication of the desirability of flattening slopes, both because the crash analysis results may over simplify a complex relationship (see the discussion of the vehicle dynamics simula­ tion results below) and because of variability in the earthwork/ grading costs needed to achieve flatter slopes. The vehicle dynamics simulation analysis again provides a more complete understanding of the subtleties of median slope effects, as it did for median width effects. In this case, the vehicle dynamics simulation results indicate an interac­ tion between median slope and median width not evident in the crash analysis results. For median slopes in the range from 1V:4H to 1V:7H, the boundary between medians for which CMCs are most prevalent and those for which rollover crashes are most prevalent falls in the median width range from 15 to 17 m (50 to 55 ft). For median slopes of 1V:8H or flatter, that boundary falls at 18 m (60 ft). Thus, the vehicle dynamics simulation results indicate that the concerns about high­severity CMCs are of greatest concern for median widths less than 18 m (60 ft) and for median slopes steeper than 1V:8H. Furthermore, the vehicle dynamics simulation results suggest that the likelihood of CMCs does not continue increasing as the median slope becomes flatter than 1V:8H. Chapter 1 of this report noted that there has been specula­ tion that flatter median slopes may contribute to an increase in CMCs. The crash analysis indicates that this is true to an extent, but may be counterbalanced by a decrease in rollover crashes. The vehicle dynamics simulation results indicate the conditions under which CMCs become less probable than rollover crashes; as the median width increases, less severe rollover crashes become more likely than more severe CMCs, whatever the median slope. The vehicle dynamics simulation results also indicate that the most favorable median shape from the standpoint of roadside safety is a trapezoidal shape, sloping down from the inside shoulders, with the center of the median being flat. Practical drainage considerations make it undesirable to grade the center of the median as completely flat, but slopes near the center of the median flatter than those closer to the traveled way appear desirable. It appears that the most desir­ able median slope should be 1V:8H or flatter immediately outside the traveled way and, where practical, still flatter near the center of the median. 6.3 Median Barriers The cross­sectional (regression) models developed in the crash analysis for traversable and barrier medians can be used to estimate the safety differences between traversable and barrier medians with various geometric characteristics and barrier types. However, this approach is likely to be less accurate than using the crash reduction factors (CRFs) for median barriers developed in the EB before­after evaluation and documented in Tables 4­30 and 4­31. Table 6­2 presents a summary of these CRFs. The CRFs for median barriers can also be expressed as CMFs, the form of countermeasure/treatment effectiveness measure used in the AASHTO Highway Safety Manual (63). CMFs have a nominal value of 1.0. CMF values less than 1.0 indicate crash types whose frequency is reduced by a counter­ measure or treatment. CMF values greater than 1.0 indicate crash types whose frequency is increased by a countermea­ sure or treatment. Table 6­3 presents a summary of the effec­ tiveness of median barriers that is equivalent to Table 6­2, but expressed as CMFs, rather than CRFs. Tables E­3 through E­5 in Appendix E present benefit­cost analyses based on the crash prediction models for fatal­and­ injury crashes on rural four­lane freeways presented in Table 4­7 and the median barrier effectiveness estimates shown in Table 4­31. This analysis focused on fatal­and­injury crashes because there are no explicit CMFs for property­damage­ only (PDO) crashes. A supplementary analysis showed that PDO crashes were unlikely to substantially affect the benefit­ Traversable medians Barrier medians Crash severity level CMCs NCMCs Rollover crashes Hit-fixed-object crashes Other median- related crashes CMCs NCMCs Rollover crashes Hit-fixed-object crashesa Other median- related crashes Crashes by crash severity level as a percentage of fatal-and-injury median-related crashes Fatal 26.7 4.8 9.0 5.0 4.4 40.5 14.3 9.9 2.7 3.0 Injury 73.3 95.2 91.0 95.0 95.6 59.5 85.7 90.1 97.3 97.0 Crashes by crash severity level as a percentage of total median-related crashes Fatal 19.5 1.8 6.1 1.4 1.7 28.8 5.3 6.8 0.7 1.4 Injury 53.5 37.5 61.9 26.6 36.7 42.4 31.6 61.9 25.9 45.3 PDO 27.0 60.7 32.0 72.0 61.6 28.8 63.2 31.3 73.3 53.3 a for barrier medians, includes hit-barrier crashes. Note: Based on data for all agencies combined from Tables 4-4 and 4-21. Table 6-1. Summary of crash severity distributions by median type and crash type.

105 cost analysis results. The analysis results show that flexible barriers (i.e., cables), semi­rigid barriers (i.e., steel guardrail), and rigid barriers (i.e., concrete) can all be cost­effective in reducing crashes under appropriate conditions. As shown in Tables 6­2 and 6­3, each of these barrier types reduces the more severe CMCs while increasing less severe hit­fixed­ object crashes (including hit­barrier crashes). Rigid barriers generally are used only in narrow medians. The benefit­cost analysis results in Appendix E show that flex­ ible and semi­rigid barriers are generally more cost­effective than rigid barriers and generally should be preferred where the median is wide enough to accommodate the deflection that occurs when a vehicle strikes a flexible or semi­rigid barrier. Figure 5­25, based on vehicle dynamics simulation results, provides guidance on appropriate barrier heights so that the barrier and vehicle bumpers interact appropriately when a collision occurs. 6.4 Design Guidelines The following design guidelines have been derived from the research results: • The AASHTO Green Book (1) recommends 1V:6H slopes within medians, with 1V:4H slopes considered adequate in some cases. Based on the research results, it is recom­ mended that the Green Book be changed to recommend 1V:8H slopes within medians, with 1V:6H slopes consid­ ered adequate in some cases. It also is recommended that slopes flatter than 1V:8H be considered near the center of the median, where practical. • It is recommended that the median barrier warrants in the AASHTO Roadside Design Guide (2) be changed to indicate that barrier be considered for median widths up to 18 m (60 ft) where the median slope is less than 1V:8H. • It is recommended that the CMFs for median barrier instal­ lation shown in Table 6­3 be considered for inclusion in the AASHTO Highway Safety Manual (63), potentially in conjunction with the SPFs for median­related crashes presented in Tables 4­6 and 4­7. These CMFs are suitable for planning of roadside design policies that would be applied over many sites or to analyses conducted with a combination of an SPF for median­related crashes and the application of the EB method. However, these CMFs are probably not a suitable tool for application to indi­ vidual sites without use of an SPF and the EB method, because individual sites are unlikely to have experienced a sufficient number of CMCs to make application of the CMFs accurate. • Benefit­cost analysis suggests that flexible median barriers may be cost­effective even at lower traffic volumes than suggested in AASHTO median barrier warrants. Crash type Crash reduction factor (%) Flexible median barriers Semi-rigid median barriers Rigid median barriers Total median- related crashes F & I median- related crashes Total median- related crashes F & I median- related crashes Total median- related crashes F & I median- related crashes Rural four-lane freeways All median-related crash types combined –227 –60 –152 –50 –140 –8 CMCs 96 92 98 100 100 100 CMCs + NCMCs 55 62 89 83 100 100 Rollover crashes 74 57 88 69 100 100 Hit-fixed-object crashesa –720 –132 –426 –113 –892 –23 Other median-related crashes –128 –67 –72 –47 –40 14 Rural six-lane freeways All median-related crash types combined –66 –17 –– –– –– –– CMCs 73 69 –– –– –– –– CMCs + NCMCs 74 69 –– –– –– –– Rollover crashes 23 31 –– –– –– –– Hit-fixed-object crashesa –209 –128 –– –– –– –– Other median-related crashes 16 14 –– –– –– –– a Increases in crash frequency include hit-barrier crashes. Note: Statistical significance and standard errors are shown in Tables 4-30 and 4-31. Table 6-2. Summary of CRFs for median barrier installation.

106 Crash type Crash modification factor Flexible median barriers Semi-rigid median barriers Rigid median barriers Total median- related crashes F & I median- related crashes Total median- related crashes F & I median- related crashes Total median- related crashes F & I median- related crashes Rural four-lane freeways All median-related crash types combined 3.27 1.60 2.52 1.50 2.40 1.08 CMCs 0.04 0.08 0.02 0.00 0.00 0.00 CMCs + NCMCs 0.45 0.38 0.11 0.17 0.00 0.00 Rollover crashes 0.26 0.43 0.12 0.31 0.00 0.00 Hit-fixed-object crashesa 8.20 2.32 5.26 2.13 9.92 1.23 Other median-related crashes 2.28 1.67 1.72 1.47 1.40 0.86 Rural six-lane freeways All median-related crash types combined 1.66 1.17 –– –– –– –– CMCs 0.27 69.00 –– –– –– –– CMCs + NCMCs 0.26 69.00 –– –– –– –– Rollover crashes 0.77 31.00 –– –– –– –– Hit-fixed-object crashesa 3.09 1.28 –– –– –– –– Other median-related crashes 0.84 14.00 –– –– –– –– a Increases in crash frequency include hit-barrier crashes. Note: Statistical significance and standard errors are shown in Tables 4-30 and 4-31. Table 6-3. Summary of CMFs for median barrier installation.