Median Intersection Design for Rural High-Speed Divided Highways (2010)

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.

4Background A rural expressway is a high-speed (≥50 mph), multilane, divided highway with partial access control. Although design policies vary from state to state, rural expressways are gener- ally a hybrid design between a freeway and a conventional two-lane rural arterial roadway. Like freeways, rural express- ways are typically four-lane divided facilities (i.e., two lanes in each direction separated by a wide, depressed, turf median), which may have grade separations and interchanges. Express- way interchanges are generally limited to locations where the additional expenditure can be justified (i.e., at junctions with major highways, along bypasses, or at intersections with a historically disproportionate rate of serious crashes) because, like a conventional two-lane undivided rural arterial, express- ways have partial access control allowing at-grade intersections and limited driveway access with the potential for signalization. Therefore, most intersections are at-grade. The typical rural expressway at-grade intersection, as shown in Figure 1, is two-way stop controlled (TWSC) with the stop control on the minor (usually two-lane) roadway. This traditional rural expressway intersection design has 42 conflict points as shown in Figure 2. State Transportation Agencies (STAs) have been construct- ing and operating rural expressways since the early 1950s in order to provide many of the safety, mobility, travel efficiency, and economic benefits of freeways at a far lower cost (1). Expressways are less expensive to build because they don’t require the acquisition of as much right-of-way (ROW) or the construction of as many overhead bridges/interchanges and miles of frontage roads necessary to meet the freeway definition. Full access control and the associated grade separations can easily result in freeway construction costing double that of a comparable expressway corridor. Additionally, depending on state design standards, expressways may be designed using a less expensive cross section (i.e., narrower medians and shoulders) (2). As a result, converting undivided rural two-lane highways into expressways has become a popular highway improvement used by many STAs. By providing an extra lane of travel in each direction and a physical separation between opposing traffic flow, expressways make passing easier and drastically reduce the likelihood of dangerous head-on and opposite-direction sideswipe collisions experienced on rural two-lane highways. In addition, medians minimize headlight glare and provide a recovery area for out-of-control vehicles, a stopping area in case of emergencies, space for speed-change lanes, and storage for left-turning and U-turning traffic (3). Minnesota DOT (MnDOT) data has shown that crash rates and severities on rural expressways are indeed lower than on rural two-lane highways: 0.9 crashes per million vehicle-miles (mvm) with a fatality rate of 1.2 deaths per 100 mvm as compared with 1.0 and 1.6, respectively (4). Additionally, when access densities are low (≤ 5 access points per mile), crash rates on rural expressways drop to a level sim- ilar to that on rural freeways: 0.62 crashes per mvm compared with 0.60, respectively (4). Besides the safety benefits, the popularity of conversion is also due to the fact that the high design speed and multilane cross section enable expressways to operate, between intersections, with a capacity approaching that of a freeway; accordingly, expressways are considered more reliable facilities than con- ventional rural arterials. Consequently, expressways attract more trips, especially those made by the freight industry, and are viewed as an essential component for communities seeking to attract industry and economic development. Therefore, because expressways provide many of the safety, mobility, and economic benefits of freeways at a lower cost, expressways have become the fastest growing segment of the nation’s rural highway system (2). The FHWA Office of Highway Policy Information’s (OHPI’s) annual highway statistics reports, Highway Statistics Annual Publications (5), were used to estimate the total rural express- way mileage in the United States on a state-by-state basis in 1995 and 2005. The data obtained is presented in Table 1. C H A P T E R 1 Introduction and Research Approach

5Figure 1. Typical rural expressway at-grade intersection. Examples include the Nebraska State Expressway System and the Iowa Commercial and Industrial Network (CIN). Other agencies, like the Missouri DOT (MoDOT), have simply made a commitment to upgrade rural two-lane highways to express- ways in order to provide better connectivity between regional centers within the state (6). Problem Statement The Nebraska Department of Roads (NDOR) is also in the process of building an extensive program of rural express- ways. This fact is evident in Table 1. Between 1995 and 2005, Nebraska ranked third in rural expressway miles added and concurrently had the largest percentage increase in total rural expressway mileage. As part of a 1988 NDOR Needs Study, engineers used socioeconomic data to designate an expanded State Expressway System. The rural two-lane highways selected for upgrade were chosen • To connect urban centers with a population of 15,000 or more to each other and to I-80; • To add routes with an average daily traffic (ADT) volume of 500 or more heavy commercial vehicles; and • To add additional segments for continuity. Over this decade, rural expressway mileage increased nation- ally by 2,400 miles or 16% with 13 states having added over 100 miles to their rural expressway systems. The majority of this expansion has been done through the conversion of undivided two-lane rural highways into expressways, and this trend is likely to continue as a 2004 survey of STAs indicated that 26 of the 28 responding agencies had plans to expand their state expressway systems over the next 10 years (2). Some states have explicit, strategic programs for upgrading two-lane highways on uniquely identified networks to multilane divided standards. Figure 2. Conflict-point diagram for typical rural expressway intersection.

6Sorted by Percent Change from 1995–2005 Sorted by 2005 Total Miles Sorted by Miles Change from 1995–2005 % CHANGE RANK STATE * 1995 TOTAL MILES * 2005 TOTAL MILES % CHANGE 2005 MILES RANK STATE * 2005 TOTAL MILES MILES CHANGE RANK STATE * MILES CHANGE (1995- 2005) 1 Nebraska 46 320 595.65 1 Texas 2071 1 New Mexico 484 2 Wyoming 2 8 300.00 2 Virginia 1097 2 Iowa 299 3 Arizona 65 194 198.46 3 New Mexico 830 3 Nebraska 274 4 Iowa 194 493 154.12 4 Mississippi 813 4 Tennessee 241 5 New Mexico 346 830 139.88 5 Ohio 751 5 South Dakota 199 6 South Dakota 196 395 101.53 6 Missouri 733 6 West Virginia 186 7 Tennessee 286 527 84.27 7 Alabama 694 7 Virginia 166 8 West Virginia 243 429 76.54 8 Minnesota 679 8 North Carolina 138 9 Kentucky 214 329 53.74 9 Florida 672 9 Arizona 129 10 Wisconsin 227 348 53.30 10 Georgia 598 10 Wisconsin 121 11 Illinois 204 292 43.14 11 California 587 11 Kentucky 115 12 Arkansas 73 104 42.47 12 North Carolina 580 12 Oklahoma 108 13 Kansas 115 151 31.30 13 Indiana 575 13 Alabama 102 14 North Carolina 442 580 31.22 14 Oklahoma 538 14 Texas 93 15 Oklahoma 430 538 25.12 15 Tennessee 527 15 Illinois 88 16 Washington 188 223 18.62 16 Iowa 493 16 Ohio 73 17 Virginia 931 1097 17.83 17 West Virginia 429 17 Minnesota 61 18 Alabama 592 694 17.23 18 South Carolina 421 18 Kansas 36 19 Ohio 678 751 10.77 19 North Dakota 405 19 California 35 20 Minnesota 618 679 9.87 20 South Dakota 395 20 Florida 35 21 California 552 587 6.34 21 Wisconsin 348 21 Washington 35 22 Florida 637 672 5.49 22 Kentucky 329 22 Mississippi 32 23 Texas 1978 2071 4.70 23 Nebraska 320 23 Arkansas 31 24 Georgia 572 598 4.55 24 Illinois 292 24 Missouri 29 25 Missouri 704 733 4.12 25 Maryland 290 25 Georgia 26 26 Mississippi 781 813 4.10 26 Louisiana 246 26 Indiana 18 27 Indiana 557 575 3.23 27 Washington 223 27 Wyoming 6 28 Alaska 5 5 0.00 28 Pennsylvania 212 28 Alaska 0 29 Maine 0 0 0.00 29 Arizona 194 29 Maine 0 30 Utah 55 54 –1.82 30 New York 185 30 Connecticut –1 31 Nevada 46 41 –10.87 31 Kansas 151 31 Utah –1 32 North Dakota 464 405 –12.72 32 Colorado 149 32 Hawaii –3 33 South Carolina 493 421 –14.60 33 Delaware 106 33 Nevada –5 34 Delaware 127 106 –16.54 34 Arkansas 104 34 Vermont –5 35 New York 222 185 –16.67 35 Oregon 78 35 Massachusetts –7 36 Maryland 360 290 –19.44 36 Michigan 64 36 New Hampshire –10 37 Vermont 25 20 –20.00 37 Utah 54 37 Montana –12 38 Louisiana 317 246 –22.40 38 Idaho 50 38 Rhode Island –13 39 Idaho 65 50 –23.08 39 Nevada 41 39 Idaho –15 40 Pennsylvania 305 212 –30.49 40 Vermont 20 40 Delaware –21 41 Colorado 223 149 –33.18 41 Montana 13 41 New York –37 42 Oregon 125 78 –37.60 42 Wyoming 8 42 Michigan –46 43 Michigan 110 64 –41.82 43 Alaska 5 43 Oregon –47 44 Montana 25 13 –48.00 44 Hawaii 3 44 North Dakota –59 45 Hawaii 6 3 –50.00 45 Rhode Island 3 45 Maryland –70 46 Rhode Island 16 3 –81.25 46 New Jersey 3 46 Louisiana –71 47 New Jersey 98 3 –96.94 47 Maine 0 47 South Carolina –72 48 Connecticut 1 0 –100.00 48 Connecticut 0 48 Colorado –74 49 Massachusetts 7 0 –100.00 49 Massachusetts 0 49 Pennsylvania –93 50 New Hampshire 10 0 –100.00 50 New Hampshire 0 50 New Jersey –95 U.S. TOTALS 14,976 17,379 16.05 U.S. TOTAL 17,379 U.S. TOTAL 2,403 * Federal-Aid Highway Length – Miles by Traffic Lanes and Access Control: Non-Interstate, NHS, Rural Divided Highways (≥ 4 Lanes with Partial or No Access Control) Table 1. Estimated rural expressway mileage by state, 1995 and 2005 (5).

When this program is complete, the Nebraska Expressway System will be approximately 600 miles in length (7). In the midst of this program, NDOR wanted to find out whether upgrading these two-lane highways to expressway standards was improving safety as expected. In 2000, NDOR’s Highway Safety Division compared crash data for 111 miles of rural expressways [i.e., sections with fewer than 8,000 vehicles per day (vpd)] with 324 miles of rural two-lane highways planned for expressway conversion (8). The results of this study are presented in Table 2. For each roadway segment included in the analysis, at least 2 years of crash data were used in the crash rate calculations; 3 years of data were used where possible. As Table 2 shows, crash rates for head-on and opposite-direction sideswipe collisions were substantially lower on rural expressways as anticipated. Unfortunately, this level of improvement did not extend to all crash types. The major concern is the 71% higher right-angle crash rate on expressways. While other intersection-related, multi-vehicle crash types (i.e., rear-end/left-turn) had lower crash rates on expressways, the overall intersection-related crash rate was 2% higher on expressways due to the elevated right-angle crash experience. In addition, the overall crash rate was 5% higher on expressways. As a result of these findings, NDOR concluded that right-angle intersection collisions on their existing rural expressways seem to be negating the safety benefits that should be derived from converting rural two- lane highways into expressways (8). Similarly, Preston et al. (4) reviewed 3 years (2000 to 2002) of rural TWSC intersection crash data for MnDOT and found that rural expressway inter- sections have a greater proportion of right-angle collisions than intersections on rural two-lane highways: 36% to 26%, respectively. Right-angle crashes are potentially more severe on expressways due to the higher speed of mainline traffic and, consequently, are cause for concern, especially as many STAs plan to continue converting two-lane highways into expressways. The right-angle crash problem at rural expressway inter- sections is not specific to Nebraska and Minnesota: Utah data and Iowa data have shown similar trends. Minnesota data and Utah data were presented in NCHRP Report 375 (9). This data showed that 42% of all rural divided highway crashes in Minnesota (34% in Utah) were intersection-related, and 57% of those collisions (69% in Utah) were right-angle or turning crashes. A more recent study showed that 52% of all rural expressway intersection collisions in Iowa were of the right- angle variety (2). These statistics illustrate that right-angle intersection collisions constitute an important issue in rural expressway safety management. During the 1990s, the Iowa DOT upgraded a number of routes on the Iowa CIN to expressway standards. The Iowa CIN is composed of 2,275 miles of primary highways identi- fied by the state legislature to enhance opportunities for the development and diversification of the state’s economy (10). As shown in Table 1, between 1995 and 2005, Iowa ranked second in rural expressway miles added and had the fourth highest percent increase in total rural expressway mileage. Current Iowa DOT long-range plans call for an additional 355 miles of the CIN to be constructed to four-lane divided standards (84 miles of which are included in the 2009–2013 Iowa Transportation Improvement Program). Under the CRASH RATES (crashes/million vehicle miles) CRASH TYPE 324 MI OF RURAL 2-LANE HWYS PLANNED FOR EXPRWY CONVERSION 111 MI OF RURAL 4-LANE DIVIDED EXPRWYS (ADT < 8,000 vpd) PERCENT DIFF. All Crashes 0.942 0.991 + 5.2 Fatal 0.022 0.015 –31.8 Injury 0.324 0.309 –4.6 Property-Damage-Only 0.596 0.668 + 12.1 Single-Vehicle Crashes 0.556 0.661 + 18.9 Run-Off-Road 0.213 0.247 + 16.0 Animal 0.309 0.381 + 23.3 Multiple-Vehicle Crashes 0.386 0.330 –14.5 Sideswipe (Opposite) 0.067 0.005 –92.5 Head-On 0.012 0.004 –66.7 Rear-End 0.131 0.093 –29.0 Sideswipe (Same) 0.064 0.051 –20.3 Left-Turn 0.026 0.025 –3.8 Right-Angle 0.087 0.149 + 71.3 Intersection-Related 0.231 0.236 +2.2 Table 2. Nebraska crash experience: 2-lane highways versus 4-lane expressways (8). 7

8best circumstances, if the conditions that result in the most problematic intersections are known during the corridor planning stage of an expressway’s development, those condi- tions can be prevented by highway planners and designers. For existing facilities, locations with problematic conditions can be identified and traffic safety engineers can proactively program the appropriate improvements. In order to better understand the safety performance of TWSC expressway intersections, a number of studies over the years have examined the relationship between the frequency of crashes and traffic volume through the development of safety performance functions (SPFs). The first to do so was McDonald (1) in 1953. He examined 150 unsignalized, divided highway intersections (both three- and four-legged) in rural California. Then, in 1964, Priest (11) studied 316 divided highway intersections in Ohio (the author did not explicitly state the area type, traffic-control type, or the number of intersection legs for the sample intersections). Next, in 1992, Bonneson and McCoy (12) investigated 125 TWSC inter- sections in rural Minnesota using a FHWA Highway Safety Information System (HSIS) data subset (apparently only 17 of these intersections were on divided highways). In 1995, Harwood et al. (9) developed separate SPFs for 153 TWSC and 157 unsignalized, three-legged divided highway intersections in rural California. Finally, in 2004, Maze et al. (2) developed an SPF using 644 TWSC expressway intersections in rural Iowa. The relationships developed as a result of these studies are summarized in Table 3 and Figure 3. Four of these five studies concluded that crash frequency is more sensitive to changes in the minor roadway volume than to changes in the divided highway volume as indicated by the larger values of the minor roadway volume coefficients (1, 2, 11, 12). Maze et al. (2, 13, 14) examined this phenomenon in closer detail for the Iowa DOT by examining how crash rates, crash severity, and crash types are affected by increasing traffic volumes and other various site conditions. REFERENCE SAMPLE EQUATION R2 McDonald, 1953 (1) 150 unsignalized divided highway intersections (both 3- and 4-leg) in rural California 633.0455.0000783.0 MINMAJ VVN Not Given Priest, 1964 (11) 316 divided highway intersections in Ohio (area type, traffic control, and number of intersection legs not specified) Graphical expression as presented in Figure 3 Not Given Bonneson and McCoy, 1992 (12) 125 TWSC intersections in rural Minnesota (108 on two-lane major road, 17 on four-lane divided highways) 7911.02925.0 10001000 6503.0 MINMAJ VVN NotApplicable 153 TWSC divided highway intersections in rural California 575.0672.0498.20 MINMAJ VVeY * 7654321 233.0396.0317.0354.0328.0961.0013.0 XXXXXXXe 0.3914 Harwood et al., 1995 (9) 157 unsignalized, 3-legged, divided highway intersections in rural California 324.0110.1677.12 MINMAJ VVeY * 8743 478.0399.0732.0838.0 XXXXe 0.3454 Maze et al., 2004 (2) 644 TWSC rural expressway intersections in Iowa )*00042.0()*00005.0(02278.0 MINMAJ VVeN 0.381 N = Expected number of crashes per year Y = Expected number of multiple-vehicle crashes over a 3-yr period VMAJ = ADT (vpd) entering from the major road (divided highway) VMIN = ADT (vpd) entering from the minor road (crossroad) X1 = Median width (ft) X2 = Average lane width on major road (ft) X3 = 1 if intersection lighting is present, 0 otherwise X4 = 1 if left-turn channelization is present on the major road, 0 if not X5 = 1 if functional class of major road is 4 or 5; 0 if functional class is 1, 2, or 3 X6 = 1 if major road has 4 lanes in both directions combined, 0 if less than 4 lanes combined X7 = 1 if terrain is rolling or mountainous, 0 if terrain is flat X8 = 1 if partial access control on major road, 0 if no access control Table 3. Summary of SPFs developed for divided highway intersections.

Maze et al. (13) found that as expressway volumes increase, crashes more commonly occur at intersections (as opposed to between intersections). These results are presented in Figure 4. In addition, Maze et al. (2) found that intersection crash rates and severity are highly dependent on the volume of traffic entering the intersection from the minor road. As minor road traffic volumes increase, the intersection crash rates increase and the crashes become more severe, as shown in Figure 5. Furthermore, as entering minor road volumes increase, the distribution of intersection crash types changes as shown in Figure 6, with a higher proportion of right-angle crashes occurring. Because right-angle crashes are more likely to be severe, increasing minor road volume results in increased crash severity, as illustrated in Figure 5. When similar bar charts were developed and stratified by entering expressway volumes, intersection crash rates, severity rates, and crash type distri- bution did not tend to change with increasing expressway volumes in Iowa (2). In Minnesota, where some rural express- ways have become heavy commuter routes into the Twin Cities Metropolitan Area and thus experience much larger volumes than in Iowa, Preston et al. (4) observed that right- angle crashes do become more prevalent with increasing expressway volumes as shown in Figure 7. From the set of 644 TWSC rural expressway intersections in Iowa, Burchett and Maze (14) identified the 100 highest- severity and the 100 lowest-severity intersections (in terms of severity index rate as described in Figure 5) from among the 327 intersections that had experienced at least one crash during the study period (1996–2000) for the purpose of conducting a comparative statistical analysis. In their analysis of land use, they found that 75% of the low-severity inter- sections were bordered by agricultural land whereas 86% of the high-severity intersections were bordered by residential or commercial land-use. Further investigation revealed that the fatality rate for intersections located adjacent to residen- tial land use was 79% and 32% greater than for intersections located adjacent to commercial and agricultural land use, respectively. Because residential development serves as a proxy for peak volumes as commuters travel to and from work, Burchett and Maze (14) obtained 24-hr counts for a sample 9 Figure 3. Crash frequency and volume relationship developed by Priest (11).

of 30 intersections with the highest crash severity index rates in order to examine the impact of hourly peaking on safety performance. After determining the peak hours for each inter- section, they found that the 30 highest-severity intersections experienced extreme hourly peaking with 52% of the crashes occurring during the peak hours. In addition, Maze et al. (2) and Preston et al. (4) both found that the distribution of crashes at TWSC rural expressway intersections with the worst safety performance tend to be heavily skewed toward right-angle crashes. Maze et al. found that the 10 highest crash severity intersections in Iowa (those where the actual severity index exceeded the expected index by the greatest amount) had 66.3% right-angle crashes as com- pared with 13.0% at the 10 lowest crash severity intersections (those where the expected severity index exceeded the actual index by the greatest amount). Similarly, Preston et al. observed 53% right-angle crashes at 23 rural expressway intersections in Minnesota where the crash rates exceeded the critical crash rate versus 36% right-angle crashes at 396 rural expressway intersections across the state. Furthermore, Burchett and Maze (14) found that almost 60% of the crashes occurring at the 100 highest-severity and the 100 lowest-severity intersections were of the right-angle variety when vertical curvature, hori- zontal curvature, or intersection skew were present. In an effort to develop a better understanding of the causes of right-angle crashes at TWSC rural expressway intersections, Preston et al. (4) performed a detailed review of the crash reports at three of the intersections over the critical crash rate and found the following: 1. 87% of the right-angle crashes were due to the inability of minor road drivers to recognize oncoming expressway traf- fic and/or select safe gaps in the expressway traffic stream; 2. 78% of the right-angle crashes were “far-side” collisions [i.e., right-angle crashes involving left-turning or crossing minor road vehicles that successfully cross the first (near- 10 Figure 4. Percent intersection crashes on expressways by expressway volume (13).

11 Figure 5. Intersection crash rates and severity by minor road entering volume (2). Figure 6. Intersection crash type distribution by minor road entering volume (2).

side) set of expressway lanes, but collide with expressway traffic in the second (far-side) set of lanes after traversing through the median (the concept of near and far-side inter- sections is illustrated in Figure 8)]; and 3. Intersection recognition (i.e., running of the STOP sign) by drivers on the minor, stop-controlled approaches was not a contributing factor in any of the right-angle crashes at these intersections. Similarly, Burchett and Maze (14) found that the ratio of far-side to near-side collisions at 30 TWSC rural expressway intersections with the highest crash severity indices in Iowa was 62% to 38%; however, at 7 of these intersections where horizontal curves were present along the expressway, far-side and near-side collisions were nearly equally distributed at 51% and 49%, respectively. Therefore, horizontal curves on the mainline seem to create a unique hazard for minor road 12 Figure 7. Effect of expressway volumes on right-angle crashes in Minnesota (4). Figure 8. Far-side and near-side intersection definitions.

drivers attempting to select gaps at both the near and far-side intersections. From all of these observations, it appears that the primary safety issue at TWSC rural expressway intersections is right- angle collisions (far-side right-angle crashes in particular). The predominant cause of these crashes seems to be the inability of minor road drivers to judge the speed and distance (i.e., arrival time) of approaching expressway vehicles as they attempt to enter or cross the expressway. Preston et al. (4) speculated that some of these mistakes in judgment may occur because minor road drivers are using a “one-stage” gap selec- tion process. A one-stage gap selection occurs when a minor road driver simultaneously tries to find an acceptable gap in expressway traffic coming from both the left and right, thereby attempting to cross or turn left in a single motion without stopping in the median to re-evaluate whether the gap in traffic coming from the right is still adequate. On the other hand, “two-stage” gap selection is less complex and far less demand- ing on the minor road driver because it breaks down the cross- ing or left-turning process into four easier successive tasks. First, the minor road driver focuses only on finding a gap in expressway traffic coming from the left. Once the minor road driver has crossed the near-side expressway lanes, he or she then stops in the median and focuses on expressway traffic coming from the right. Finally, the minor road driver pro- ceeds to enter or cross the far-side expressway lanes when an acceptable gap is available. Another possible contributing fac- tor to minor road drivers misjudging approaching expressway vehicle speeds and distances may be the fact that many of these intersections are in rural areas where there are no large fixed objects that can be used as points of reference to help gage vehicle arrival times. Other intersection geometric design features (horizontal and vertical curvature on the expressway, intersection skew, median width, etc.) make the task of gap selection more difficult for the minor road driver. In addition, minor road driver age plays a role in their ability to safely navigate through a TWSC expressway intersection. Young drivers have more difficulty due to their inexperience, and elderly drivers have more problems due to their naturally declining visual, motor, and cognitive processing capabilities. A survey of elderly drivers in West Virginia (15) indicated that more than half of the respondents had problems making turning or crossing maneuvers from the minor road at TWSC expressway inter- sections and their greatest difficulty was stated as judging the speed of oncoming vehicles. Maze et al. (2) found that the driver age distribution of those involved in injury and fatal crashes was similar for rural expressway intersections in Iowa versus all rural intersections statewide, but younger drivers (<25 years of age) and older drivers (>55 years of age) were over-represented in right-angle crashes at rural expressway intersections. The safety performance of conventional TWSC rural expressway intersections declines as entering traffic volumes increase, especially from the minor road or where minor roadway volumes are highly peaked. At the most problematic intersections, right-angle collisions are exacerbated as gap selection becomes more of an issue (see Figure 9). As express- way volumes increase, the number of safe gaps in the express- way traffic stream declines and right-angle crashes become more prevalent. As minor road volumes increase, there are more vehicles trying to use the same number of safe gaps; thus, there is an increased probability for right-angle crashes to occur. Furthermore, when more traffic is present on the minor road approaches (i.e., congestion during peak hours) there may be more “peer” pressure on the lead minor road driver to select a gap that he or she would not normally accept, resulting in a higher number of unsafe gaps being selected. Research Objectives As a result of the trend to convert rural two-lane roadways into multilane divided highways, rural expressways are a rapidly growing component of the nation’s transportation network. As these facilities experience growth in traffic, at-grade intersection collisions begin to reduce the safety benefits that should be achieved as a result of conversion, bringing into question the assumption that expressways are an effective alternative to full access-controlled facilities at high volumes. When the safety performance of at-grade intersections begins to deteriorate, the traditional approach taken by STAs is to consider improvements one intersection at a time. Often, the improvement path starts with the application of several signing, marking, and/or lighting improvements followed by the implementation of traffic signals and, ultimately, grade separation. The problem with this method is that it is reactive to problems at intersections with historically poor safety records. Substantial time is required to fully determine, with 13 Figure 9. Crash causation comparison for rural TWSC intersections in Minnesota (4).

confidence, that a safety problem exists and then the design, construction, and/or implementation of a solution may take many years, during which the problem may continue or worsen. For example, the environmental work, preliminary and final design, ROW acquisition, and construction of a rural inter- change may be a 5 year project or longer. Moreover, the high cost of constructing a grade separation or an interchange limits their use on expressways. When an interchange is not economically justified, when the funding is not available, or while an interchange is being developed, the traditional interim corrective measure for a TWSC expressway intersection is signalization. However, rural expressway intersections often experience safety problems long before they meet traffic signal volume warrants. Furthermore, AASHTO’s Policy on Geometric Design of Highways and Streets (a.k.a. the Green Book) (3) and NCHRP Report 500, Volume 5: A Guide for Addressing Unsignalized Intersection Collisions (16) guard against using signalization as a safety device and state that intersection control by traffic signals should be avoided whenever possible. On rural expressways in particular, signals are dangerously inconsistent with the expressway driver’s expectation of a free-flow roadway for high-speed travel, thus creating high potential for rear-end crashes and red-light running. Research on the safety effects of signalizing divided highway intersections has shown mixed results (17–21). In general, signalization leads to an increase in crash rates with reduced severity due to a shift in crash types (16), but large variability in the safety effectiveness of signalization at indi- vidual locations has been observed (21). In addition, Bonneson and McCoy (12) concluded that the costs of stopping express- way traffic are so high that a very heavy minor road demand must be present to economically justify installing a traffic signal, and when volumes reach those levels, a diamond inter- change is a more economically viable alternative. Thus, some states have established policies that prohibit signalizing express- way intersections, preferring to design interchanges where large minor roadway traffic volume levels are anticipated. Of course, in the long-run, this practice will be expensive and impractical if widely applied. Therefore, designers must have other options besides signalization and grade separation to address rural expressway intersection safety. STAs have experimented with a wide range of intersection safety treatments at problematic rural expressway intersections to improve their safety performance while avoiding signal- ization and grade separation. A number of these strategies are identified and briefly described in NCHRP Report 500, Volume 5 (16). Unfortunately, few states have adequately exam- ined the safety effects of these treatments as most are listed as “tried” or “experimental” strategies. In order to determine the safety effectiveness of these alternatives and classify them as “proven,” the “tried” and “experimental” strategies will need to be appropriately implemented and evaluated scientifically through rigorous before-after studies. However, some STAs are reluctant to implement and test these strategies because design guidance is lacking and is generally silent on their application. The objective of this project is to review and document expressway intersection countermeasures implemented by various STAs and to recommend improvements to the rural median intersection design guidance currently provided in the Green Book (3) and the Manual on Uniform Traffic Control Devices (MUTCD) (22) for high-speed (50 mph and faster) divided highways with partial or no access control (express- ways). The Green Book and the MUTCD should address the conditions under which unconventional geometric and traffic- control treatments are warranted based on safety considerations and should provide guidance on how STAs need to proactively plan for expressway intersection safety during the initial cor- ridor planning process as well as throughout the entire life cycle of an expressway intersection as it reaches specific volume thresholds. At low volumes, ordinary TWSC expressway inter- sections can provide very good safety performance and may be the most appropriate design in most locations when a two-lane roadway is first converted into an expressway. How- ever, as volumes are projected to grow and site conditions are expected to change, innovative intersection designs should be programmed well in advance of any safety and/or operational problems. Research Approach and Report Organization The approach to this research was divided into four major tasks. The first task involved reviewing and summarizing the existing guidance for expressway intersection design currently provided in the Green Book and the MUTCD. Areas where guidance is lacking or needs to be expanded were identified. The results of this task are summarized in Chapter 2 with a more detailed review of the Green Book guidance and the identified limitations included as Appendix A (Appendices A and B as submitted by the researchers are not published herein, but are available on the TRB website at www.TRB.org by searching for NCHRP Report 650). The second task was to perform a literature review to iden- tify safe and effective median intersection design treatments. In order to identify the intermediate intersection design treat- ments between an ordinary TWSC rural expressway inter- section and an interchange, two important research questions must be addressed. First, the safety effectiveness or the expected improvement resulting from the implementation of each expressway intersection strategy must be quantified. Second, how traffic volume levels impact the safety and operational performance of each design strategy must be understood to determine the volume thresholds that trigger the implemen- 14

tation of the next intersection improvement. Once more is known about the safety benefits of the various expressway intersection strategies and the impact of volume levels, a more proactive and systematic approach to expressway intersection safety planning can be developed. Thus, the goal of the liter- ature review was to determine what research has previously been conducted in these areas. A summary of the literature review findings is presented in Chapter 3 with a full literature review provided as Appendix B. Ten of the most promising expressway intersection safety strategies identified in the literature review were selected for further study. The third task in the research effort involved conducting “case studies” to investigate and document the experience of STAs that have experimented with these strate- gies. The case studies are presented in Chapter 4, and their objective was two-fold. The first objective was to interview knowledgeable staff from the respective agencies to determine 1. The circumstances surrounding the treatment’s imple- mentation (i.e., reasons for, cost, etc.); 2. Intersection site conditions (i.e., type and intensity of land use, traffic volumes and patterns, geometry including horizontal/vertical curves and skew, etc.); 3. Public reaction (complaints, elderly driver issues, indica- tions of erratic driving behavior, etc.); 4. Issues the agency encountered; 5. Lessons learned including design guidance and general advice; and 6. Whether the agency had performed any subjective or objective evaluations of the operational and/or safety per- formance of the treatment. The second objective of the case studies was to obtain crash data from the agency, where possible, and to conduct naïve before-after analyses of the intersection safety treatments in order to begin to understand their potential for improving expressway intersection safety. By no means are these before- after analyses meant to be scientifically rigorous evaluations that develop reliable estimates of their safety effectiveness. They are simply observational before-after studies that compare the count of the before-period crashes with the after-period crashes. This naïve before-after analysis approach has two major limitations. First, it does not take regression-to-the-mean into account (23). The sites selected for treatment were not randomly selected. They were likely selected because they were high crash locations. Therefore, it is possible that, in the after period, there would have been a reduction in crashes due to simple chance even if nothing had been done. Only a Bayesian analysis can correct for the regression-to-the-mean bias (23), but a secondary way to attempt to account for regression-to- the-mean is to use as many years of before and after crash data as possible, which is the approach taken in this research. Although some of the sites examined in the case studies had limited data available, where more than 3 years of before and after data were obtained, statistical evaluations were performed. The second limitation of the naïve before-after analysis approach is that the noted change in safety does not represent the true effect of the treatment. It also reflects the effect of other external factors such as changes in traffic volume, vehicle fleet mix, weather, driver behavior, and so on. It is not known what part of the change in safety can be attributed to the treatment and what part is due to the various other influences (23). The fourth and final task of this research effort was to iden- tify and describe expressway intersection topics warranting further study. A focus group was held in December 2006 to prioritize future research interests regarding the 10 counter- measures examined in the case studies. The results of this focus group prioritization as well as the conclusions and recom- mendations of this research effort are presented in Chapter 5. 15