Estimation of Operational Benefits of Slow Vehicle Turnouts on Rural Highways in Alaska

Transcription

1 Dunham, Bowie, and Kinney Estimation of Operational Benefits of Slow Vehicle Turnouts on Rural Highways in Alaska Connor Dunham, EIT Kinney Engineering, LLC 0 West Dimond Boulevard, Suite 0 Anchorage, Alaska Phone: 0-- Fax: ConnorDunham@KinneyEng.com Jeanne M. Bowie, PE, PhD (Corresponding Author) Kinney Engineering, LLC 0 West Dimond Boulevard, Suite 0 Anchorage, Alaska Phone: 0-- Fax: JeanneBowie@KinneyEng.com James R. Kinney, PE, PTOE Kinney Engineering, LLC 0 West Dimond Boulevard, Suite 0 Anchorage, Alaska Phone: 0-- Fax: RandyKinney@KinneyEng.com Submitted: August, 0 Revised: November, 0 Word Count: 00 words Tables: (at 0 words each = 0 words) Figures: (at 0 words each = 0 words) Total Words: 0 words TRB 0 Annual Meeting

2 Dunham, Bowie, and Kinney 0 ABSTRACT This paper draws on the results from previous research regarding two-lane highway operations and platooning characteristics to develop an estimate of the operational benefit that will be realized when slow vehicle turnouts are constructed on a 0-mile section of the Sterling Highway on the Kenai Peninsula in Alaska. Speed and volume data were collected at three sites in the study area to quantify the existing conditions. Three measures of effectiveness are presented: the distribution of platoon sizes, the percent following (defined as the percent of all vehicles that are following another vehicle at time headways of seconds or less), and the percent impeded. Percent impeded is calculated as a modification of the percent following measurement to account for vehicles that are following, but are traveling at their desired speed, and are therefore not being impeded. Future operations after the installation of the proposed slow vehicle turnouts are estimated by modifying the percent following to account for lead vehicles pulling into the slow vehicle turnouts and allowing others to pass. The percent impeded was then calculated for this modified value for percent following. A sensitivity analysis was performed to show how percent impeded was affected by differences in the percentage of lead vehicles who use the slow vehicle turnouts to let others pass. It is concluded that the slow vehicle turnouts will provide a measurable benefit to operations on the subject highway. TRB 0 Annual Meeting

3 Dunham, Bowie, and Kinney INTRODUCTION The safety of two-lane rural highways is particularly significant in the state of Alaska because there are no freeway and multi-lane highway facilities located outside of the urban areas. The predominant facilities connecting Alaskan communities are two-lane rural roadways. Experience in Alaska agrees with findings elsewhere that safety can be compromised when drivers become frustrated due to lack of passing opportunities on two-lane rural roads. Passing lanes have been shown to improve operations and safety on two-lane rural roads; however, the installation of passing lanes is not always possible due to topological or cost constraints. In these cases, the construction of slow vehicle turnouts can provide a means for slower moving vehicles to allow faster vehicles to pass. Alaska state law addresses the use of slow vehicle turnouts. According to Alaska state law on a two-lane highway outside an urban district where passing is unsafe because of oncoming traffic or other conditions, the driver of a motor vehicle proceeding at less than the maximum authorized speed of traffic and behind whom five or more vehicles are formed in a line shall turn off the roadway at the nearest place designated as a turnout or wherever sufficient area for a safe turnout exists in order to permit following vehicles to pass. ( AAC 0.00) This report draws on data collected as part of a Highway Safety Improvement Program (HSIP) project that proposes the installation of slow vehicle turnout locations on a 0-mile segment of the Sterling Highway, a two-lane highway on the Kenai Peninsula of Alaska. During the peak traffic period in July, anecdotal evidence suggests that the RV percentage on the Sterling Highway is very high and that the traffic mix includes drivers desiring to travel at high speeds (such as Alaskans who are very familiar with the roads and scenery) as well as drivers desiring to travel at more moderate speeds (such as out-of-state tourists who are traveling the highway for the first time). When safe passing opportunities are limited, driver frustration can become significant and risky passing maneuvers are more likely. PREVIOUS RESEARCH Operational Characteristics of Two-Lane Highways The Highway Capacity Manual (HCM) divides two-lane highways into three categories and defines three different variables to determine the level of service (LOS) of these highway types. () Class I two-lane highways are used primarily by drivers traveling long distances, who expect to travel at high speeds. Class II two-lane highways are used by drivers traveling shorter distances, who don t necessarily expect to travel at high speeds. Class III two-lane highways refer to those portions of two-lane highways that pass through small towns or other areas where the driveway access density is high. In the HCM, the operation of Class I two-lane highways is characterized by a measure of average travel speed (ATS) as well as the average percent of time that vehicles must spend traveling in platoons behind lower speed vehicles because of the inability to pass, or percent of time spent following (PTSF). The operation of Class II two-lane highways is characterized only by PTSF. For Class III two-lane highways, the operation is characterized by the ability of drivers to travel at the posted speed, or percent of free-flow speed (PFFS). As noted in numerous other reports, PTSF is difficult to measure in the field (the HCM does not provide a methodology to measure PTSF in the field, but only a methodology for estimating it). A number of researchers have proposed alternative measures of effectiveness (MOEs) that would be measurable in the field. A 00 paper discussed the German approach to TRB 0 Annual Meeting

4 Dunham, Bowie, and Kinney highway capacity analysis for two-lane highways. () Focusing more on efficiency than on driver comfort, the German HBS 00 uses traffic density (vehicles per mile in both directions) as the MOE for two-lane highway capacity. A study in Madrid, Spain measured the percent of delayed vehicles, defining a vehicle as being delayed if the headway between itself and the previous vehicle was seconds or less. () The researchers were able to develop a regression equation that related the directional volumes on the highway to the percent of delayed vehicles on those highways. A group of researchers in Japan proposed the use of follower density as the MOE for determining LOS of a two-lane highway. () Follower density was estimated by measuring traffic volume, speed, and the percent of vehicles following. The researchers based proposed LOS thresholds for follower density on previous research performed in Japan, South Africa, and Germany. They were able to use a combination of field measurements and simulation to estimate the effect of wintertime conditions (compacted snow on the highway) on these performance measures. A recent study in Montana proposed the measurement of the percent impeded (PI) as a surrogate measure to PTSF, similar to the way in which spot speed measurements can serve as surrogate measurements for space mean speed. () This measurement takes into account that some vehicles driving at their desired speed may be in a platoon, but are not actually being impeded. Only platooned vehicles that are driving less than their desired speed are being impeded. Vehicles that are impeded, then, must be in a platoon and must be traveling at a speed lower than their desired speed. The researchers suggested that the PI could be measured using the following equation. PI = P p P i P p = the probability of a vehicle being in a platoon (measured as the percent of vehicles following a lead vehicle at or below a certain time headway) P i = the probability of a vehicle traveling slower than the desired speed The second term (P i ) is found by developing the desired speed distribution and finding the proportion of these vehicles whose travel speed is greater than the average speed of slow moving vehicles. The desired speed distribution is developed from measurements of the travel speed of vehicles that are either leading a platoon or are not traveling in a platoon. The average speed of slower moving vehicles is determined by measuring the average speed of all vehicles at the head of a platoon. The researchers collected data on two highways in Montana over a one-week period in May and June. In addition to measuring PI, the researchers also considered percent followers, follower density, and the ratio of average travel speed to free flow speed as MOEs. Asserting that an acceptable MOE should detect improvements when there are opportunities for passing, the researchers collected data at several stations near a passing lane on each of the study highways. The data was collected at several different locations in relation to the passing lane, including just prior to the passing lane as well as at several locations downstream of the passing lane. The researchers concluded that PI represented the expected changes in PTSF more consistently than the other measures did. The ratio of average travel speed to free flow speed varied the least over the different locations near a passing lane. The current study compares the measurement of both percent following and percent impeded on the subject highway. TRB 0 Annual Meeting

5 Dunham, Bowie, and Kinney Effect of Slow Vehicle Turnouts on Traffic Operations Researchers in New Zealand collected information about the effectiveness of slow vehicle turnouts at eight locations on state highways. () The study sites were chosen to represent locations with a variety of AADTs, heavy vehicle percentages, and grades. Observers at each of the slow vehicle turnout locations recorded the percentage of following vehicles before and after the slow vehicle turnout, vehicle classification, and the percentage of lead vehicles using the slow vehicle turnout (by classification). Overall, % of the platoon leaders used the slow vehicle turnout, with the percentage of platoon leaders using the slow vehicle turnouts varying from to %, depending upon the site. In general, heavy vehicles (including RVs) were more likely to use the slow vehicle turnouts than passenger vehicles. The researchers developed a theoretical equation for determining the percentage following after the slow vehicle turnout: r = q [( q) ( e q ) p] r = the percentage following at the end of the slow vehicle turnout q = the percentage following prior to the slow vehicle turnout p = the proportion of platoon leaders using the slow vehicle turnout This formula doesn t take into account the desired speed of the following vehicles, but it can serve as an initial estimate of the effect of a slow vehicle turnout. In the study in New Zealand, the above formula predicted the actual percent following at the end of the slow vehicle turnout with less than % error for the two sites where it was tested. Effect of Slow Vehicle Turnouts on Traffic Safety Very little is known about the safety effects of slow vehicle turnouts. At the time of writing this report, there were no crash modification factors (CMFs) included in the Crash Modification Factors Clearinghouse ( for slow vehicle turnouts. () The National Cooperative Highway Research Program (NCHRP) Report 00, volume : A Guide for Addressing Head-On Collisions references NCHRP Report 0 which found that constructing turnouts resulted in a 0 percent reduction in total crashes and 0 percent reduction in fatal and injury crashes. () The safety benefit of slow vehicle turnouts could also be estimated based on the safety effects of passing lanes. Unfortunately, there is also little safety information available for passing lanes. The CMF Clearinghouse includes two or three studies with CMFs for passing lanes; however, all of these CMFs are rated as poor, meaning that extreme caution should be used in applying the CMFs from these studies to other locations. () NCHRP 00 suggests that the safety benefits of passing lanes are likely to result in total crash reductions of to 0 percent and may extend outside of the immediate area of the passing lane, similarly to how the operational benefits of passing lanes have been shown to extend for some distance downstream of the passing lane. () Nevertheless, according to NCHRP 00, this strategy is not considered proven because it has not been sufficiently evaluated. STUDY SITE The Sterling Highway is a two-lane highway on the Kenai Peninsula of Alaska that extends from the Seward Highway approximately 0 miles south to Homer, Alaska. The 0-mile section from Soldotna to Homer Hill (MP to MP ) has been nominated as a Highway Safety Improvement Program (HSIP) project by the Alaska Department of Transportation and Public Facilities (ADOTPF) due to a history of high-severity head-on collisions. The proposed TRB 0 Annual Meeting

6 Dunham, Bowie, and Kinney mitigation for this crash history is the installation of slow vehicle turnouts at locations. Currently, there are no existing passing lanes or officially designated slow vehicle turnouts on this section of highway. Slow vehicle turnouts are being promoted rather than passing lanes primarily due to the amount of budget immediately available and the length of highway that can be treated. There are also some topographic and environmental constraints in some areas. When funds are available, passing lanes are planned for the future. The annual average daily traffic (AADT) on the Sterling Highway in the study area varies from around,000 vehicles per day (vpd) at the north end of the segment to about,00 vpd at the south end of the segment; however, traffic volumes are very seasonal, with volumes peaking in the month of July at about 0 percent of AADT and dropping to 0 to 0 percent of AADT in December through February. () Based on the National Cooperative Highway Research Program (NCHRP) Report 00, volume : A Guide for Addressing Head-On Collisions, a total crash reduction factor of 0% was used for each turnout, assuming an upstream influence area of,000 feet and a downstream influence area of,000 feet. () Under these assumptions, the benefit-to-cost ratio that has been calculated for this project is 0.:. (0) In addition to the expected reduction in crashes, the proposed slow vehicle turnouts are likely to have a significant impact on operations. To understand the existing operations on the Sterling Highway in the study section, traffic volume and speed data was collected at three locations in the study area for a total of complete hour periods days in May and 0 days from the end of June to the beginning of July including Memorial Day (Monday, May, 0) and Independence Day (Wednesday, July, 0). Data was collected using continuous radar traffic data collectors that recorded time, speed, and length for vehicles traveling northbound and southbound separately. The radar traffic data collectors must be placed on vertical posts of adequate diameter to hold the weight of the data collector; therefore, sites with two-post signs were chosen as data collection sites. Site was located south of Anchor Point, Alaska and north of Homer, Alaska at approximately mile point on the Sterling Highway. The radar detector at this site was positioned on a road sign with inch diameter tube steel approximately feet above the road surface. The AADT for this site is 0 vehicles per day (vpd), with peak recorded summer traffic of 0 vpd. The site was located in a no passing zone, with the nearest passing zones being. miles to the north and 00 feet to the south. The speed characteristics of all vehicles with a time gap greater than seconds is shown in TABLE. Site was located north of Ninilchik and south of Clam Gulch, Alaska at approximately mile point on the Sterling Highway. The radar detector at this site was positioned on a road sign with inch diameter tube steel approximately feet above the road surface. The AADT for this site is vpd, with peak recorded summer traffic of vpd. The site was located in the middle of a passing zone approximately 00 feet in length. The speed characteristics of all vehicles with a time gap greater than seconds is shown in TABLE. Site was located north of Kasilof and south of Reflection Lake, Alaska at approximately mile point 0 on the Sterling Highway. The radar detector at this site was also positioned on a road sign with inch diameter tube steel approximately feet above the road surface. The AADT for this site is 0 vpd, with peak recorded summer traffic of vpd. The site was located in a no passing zone, with the nearest passing zones being 00 feet to the north and 0 feet to the south. The speed characteristics of all vehicles with a time gap greater than seconds is shown in TABLE. TRB 0 Annual Meeting

7 Dunham, Bowie, and Kinney 0 0 TABLE Speed Characteristics of Study Locations Study Location Speed Limit Median Speed th Percentile 0 MPH Pace (mph) (mph) Speed (mph) (mph) Site. 0. to Site.. to Site.. to Daily recorded volumes at the three sites varied from a minimum of 00 vehicles per day (vpd) in mid-may to 00 vpd the early part of July. Recorded hourly volumes varied from to vehicles per hour (vph). On weekends and holidays, traffic volumes peaked around midday, whereas on weekdays, traffic peaked between and pm. Vehicle classification was not collected as part of this study; however, vehicle classification data taken by a permanent traffic recorder within the study area reveals that the heavy vehicle percentage on this roadway is approximately 0%. () PLATOONING Prior to measuring the MOEs described above, it is necessary to develop criteria for determining whether or not any given vehicle is platooned. One such criteria is the time headway at or below which vehicles are assumed to be following each other, or the critical headway. Earlier versions of the HCM suggested that seconds should be used as the critical headway, but later versions shortened this distance and suggest that vehicles following each other with a -second headway or less should be considered to be in a platoon. Studies of vehicle platooning on two-lane highways use a variety of critical headway thresholds from to seconds. A study of platooning on signalized arterials suggested a method of finding the critical headway by plotting the coefficient of variance of platoon size for different values of critical headway and looking for an inflection point. () The researchers indicate the inflection point indicates stability of the measurement and suggest that this value be chosen as the critical headway for the definition of platooning. The coefficient of variance of platoon sizes is defined as the standard deviation of all of the platoon sizes given an assumed critical headway divided by the average platoon size at that assumed critical headway. FIGURE shows how coefficient of variation varies with critical headway for the data collected on the Sterling Highway. On the graph, an inflection point can clearly be seen at a critical headway of seconds. Based on this analysis, a critical headway of seconds was chosen. TRB 0 Annual Meeting

8 Dunham, Bowie, and Kinney Coefficient of Variation Site Site Site Critical Headway (s) FIGURE Change in coefficient of variation with critical headway. Vehicles that are considered to be following as defined by time headway are not necessarily impeded if the lead vehicle is traveling at a speed that is similar to the desired speed of the following vehicle. The methodology for determining percent impeded acknowledges this fact directly by developing the distribution of desired speeds and comparing that distribution to the speed of the slower vehicles (those leading a platoon). Two other methods for including desired speed were considered in this study. One possible method is to limit vehicles that are labeled as following to those that are traveling below the speed limit or another set speed threshold such as the th percentile speed. This method recognizes that vehicles that are traveling at a relatively high speed are unlikely to feel impeded. Another methodology for including speed in the definition of platooned vehicles is to look at the relative speeds between the lead and following vehicles. Vehicles that have been following for some time are likely to be closely following the lead vehicle in terms of both time head way and speed as they look for opportunities to pass. On the other hand, vehicles that have recently joined the queue and do not yet feel impeded are less likely to be closely matching the speed of the lead vehicle. A comparison was made between using only critical headway to define platooning, using th percentile speed in addition to critical headway to define platooning, and using a mph speed differential in addition to critical headway to define platooning. The difference in determination of percent following using these three methods was only to percent. Therefore, in the remainder of this study, critical headway alone was used to determine whether or not a vehicle was platooned for the purpose of calculating percent following. [Note that the methodology for determining percent impeded uses vehicle speed as a criterion for determining whether or not a vehicle is impeded and this calculation has a much larger effect.] TRB 0 Annual Meeting

9 Dunham, Bowie, and Kinney Existing Platooning Characteristics As described above, vehicle platooning was identified based on a critical time headway of seconds or less. Based on this definition, the distribution of platoon sizes and the percent following were calculated. Additionally, the percent impeded was calculated using the same time headway of seconds or less and using the desired speed distribution for each site to determine what percent of the following vehicles desired a higher speed than the lead vehicles. As can be seen in FIGURE, the majority of platoons contain or fewer vehicles, with only % of vehicles in platoons of or more vehicles and thus subject to the Alaska statute which requires the lead vehicle to pull into a turnout to allow other vehicles to pass. The percent following (defined as the percent of vehicles following another with a time headway of seconds or less) as it relates to hourly volume for each site is shown in FIGURE. As would be expected, the percent following increases as the hourly volume increases. In order to determine the percent impeded, it was necessary to develop the desired speed distributions for each site. The distribution of desired speeds includes only those vehicles who are not following another vehicle. This includes all platoon leaders and all vehicles that are not in a platoon. The distributions of desired speeds for each site are shown in FIGURE. As can be seen in the figure, each site has a distinct speed distribution curve; therefore, the percent impeded calculation for each site used the desired speed distribution particular to that site. The final piece of information needed to calculate the percent impeded is the average speed of slower vehicles. For each data collection hour, the average speed of platoon leaders (the slower vehicles) was compared to the distribution of desired speeds for that site to determine the probability of a vehicle traveling slower than the desired speed (P i ) for that hour. FIGURE shows two different sample calculations of P i for site at two different times of the day. The percent impeded (defined as the percentage of vehicles following another with a time headway of seconds or less at a speed less than their desired speed) as it relates to hourly volume for each site is shown in FIGURE. As with percent following, the percent impeded increases with hourly volume. Note that percent impeded is less than percent following. This is expected, as some vehicles that are following a lead vehicle may be comfortable in that position and may not desire to pass the lead vehicle. TRB 0 Annual Meeting

10 Dunham, Bowie, and Kinney 0 Frequency of vehicles with given platoon size Thousands 0 Site Site Site 0 0 > 0 Number of Vehicles in Platoon FIGURE Distribution of platoon size for each data collection site. Percent Following % 0% % 0% % 0% % 0% % Site Site Site 0% Hourly Volume (vehicles per hour) FIGURE Percent following by hourly volume for each data collection site. TRB 0 Annual Meeting

11 Dunham, Bowie, and Kinney % 0% Percent of Vehicles % % % % Site Site Site 0% 0000 Desired Speed (mph) FIGURE Distribution of desired speeds for each data collection site. FIGURE Sample calculation of P i for two different time periods at site. TRB 0 Annual Meeting

12 Dunham, Bowie, and Kinney % 0% Percent Impeded % 0% % Site Site Site 0 0 0% Hourly Volume (vehicles per hour) FIGURE Percent impeded by hourly volume for each data collection site. Estimated Effect of Slow Vehicle Turnouts When vehicles traveling in a platoon encounter a slow vehicle turnout, there are three possible results: The lead vehicle does not use the slow vehicle turnout. This results in no operational benefit due to the turnout, and no reduction in PI. The lead vehicle uses the slow vehicle turnout, freeing a following vehicle who desired to pass. This results in an operational benefit due to the turnout and can be seen as a reduction in PI. The lead vehicle uses the slow vehicle turnout, but the following vehicle was traveling at or near their desired speed and therefore had not been impeded. This results in no operational benefit due to the turnout, and no reduction in PI. Traffic performance is only improved when the following vehicle is being impeded and desires to pass, and the lead vehicle uses the slow vehicle turnout. Combining the equation for PI developed by Al-Kaisy and Freeman with the equation for reduction in percent following due to a slow vehicle turnout developed by Koorey, the following equation can be used to estimate the PI after a slow vehicle turnout. PI = P p P p ( e P p) P svt P i PI = percent impeded P p = the probability of a vehicle being in a platoon prior to the slow vehicle turnout (measured as the percent of vehicles following a lead vehicle at or below a certain time headway) P i = the probability of a vehicle traveling slower than the desired speed prior to the slow vehicle turnout P svt = the probability of the lead vehicle using the slow vehicle turnout In the research done by Koorey, the use of slow vehicle turnouts by platoon leaders was percent on average, with a range between 0 to 0 percent. FIGURE through FIGURE TRB 0 Annual Meeting

13 Dunham, Bowie, and Kinney show the resulting reduction in PI for the three sites studied on the Sterling Highway, using the range of values for P svt found in the Koorey study. Percent Impeded 0.00%.00%.00%.00%.00% 0.00%.00%.00%.00%.00% 0.00% Hourly Volumes (vehicles per hour) PI before SVT 0% Leaders use SVT % Leaders use SVT 0% Leaders use SVT FIGURE Estimation of change in PI due to presence of slow vehicle turnout for site. Percent Impeded 0.00%.00%.00%.00%.00% 0.00%.00%.00%.00%.00% 0.00% Hourly Volumes (vehicles per hour) PI before SVT 0% Leaders use SVT % Leaders use SVT 0% Leaders use SVT FIGURE Estimation of change in PI due to presence of slow vehicle turnout for site. TRB 0 Annual Meeting

14 Dunham, Bowie, and Kinney Percent Impeded 0.00%.00%.00%.00%.00% 0.00%.00%.00%.00%.00% 0.00% Hourly Volumes (vehicles per hour) 0 0 PI before SVT 0% Leaders use SVT % Leaders use SVT 0% Leaders use SVT FIGURE Estimation of change in PI due to presence of slow vehicle turnout for site. As can be seen in the figures, the slow vehicle turnout is expected to have a measurable effect on the operations of the highway, at least for a short distance downstream of the slow vehicle turnout location. CONCLUSIONS Previously published research on the platooning and passing characteristics of two-lane highways was drawn upon to determine the expected operational benefit of constructing slow vehicle turnouts on rural two-lane highways in Alaska. The measures of effectiveness used in the HCM are ATS (average travel speed) and PTSF (percent time spent following); however, it is difficult to directly measure PTSF on an existing roadway. A measurable parameter that has been linked to PTSF in previous research is PI (percent impeded). A number of parameters had to be estimated in order to determine the PI of existing sites on the Kenai Peninsula of Alaska. The critical headway for platooning was found by plotting the coefficient of variation of platoon sizes against a range of headway assumptions. The critical headway for the subject highway was found to be seconds. This value agrees with the value used in the HCM and other research. After PI was calculated for each of the existing sites, the expected reduction in PI due to the construction of a slow vehicle turnout was estimated using the technique employed by Koorey in New Zealand. The slow vehicle turnouts are expected to have a measurable effect on operations of the subject highways. FUTURE RESEARCH This paper models the desire for passing on two-lane highways using a simplified approach. In addition to the conditions stated in this paper, there may be other circumstances under which a following vehicle may desire to pass, even if they are traveling at their desired speed. One such TRB 0 Annual Meeting

15 Dunham, Bowie, and Kinney condition may be when the lead vehicle is larger than the following vehicle, especially if the lead vehicle is a truck or an RV. The following vehicle in this situation may desire to pass simply to improve their ability to see ahead or to allow for a change of scenery. After the slow vehicle turnouts are constructed, the researchers plan to collect additional data to determine the accuracy of the estimation performed in this paper. If vehicle classification data is also collected, the size of the lead vehicle could also be examined as a possible parameter in the model. REFERENCES ) Highway Capacity Manual 00, Transportation Research Board, 0. ) Brilon, W. and F. Weiser. Two-Lane Rural Highways: the German Experience. In Transportation Research Record: Journal of the Transportation Research Board, No., Transportation Research Board of the National Academies, Washington, D.C., 00, pp. -. ) Romana, M.G. and G. Lopez. Estimation of Percentage of Delayed Vehicles Based on Traffic Variables for Rural Highways. In Transportation Research Record: Journal of the Transportation Research Board, No., Transportation Research Board of the National Academies, Washington, D.C.,, pp. -. ) Munehiro, K., A. Takemoto, N. Takahashi, M. Watanabe, and M. Asano. Performance Evaluation for Rural + lane Highway in a Cold, Snowy Region. Presented at st Annual Meeting of the Transportation Research Board, Washington, D.C., 0. ) Al-Kaisy, A. and Z. Freedman. Estimating Performance of Two-Lane Highways: Case Study Validation of a New Methodology. In Transportation Research Record: Journal of the Transportation Research Board, No., Transportation Research Board of the National Academies, Washington, D.C., 00, pp. -. ) Koorey, Glen. Passing Opportunities at Slow-Vehicle Bays. In Journal of Transportation Engineering, Vol., No., American Society of Civil Engineers, New York NY, 00, pp. -. ) Crash Modification Factors Clearinghouse. U.S. Department of Transportation, Federal Highway Administration. Accessed July 0, 0. ) Neuman, T.R., R. Pfefer, K.L. Slack, H. McGee, F. Council. Volume : A Guide for Addressing Head-On Collisions. In Guidance for Implementation of the AASHTO Strategic Highway Safety Plan, NCHRP Report 00, National Cooperative Highway Research Program, Washington, D.C., 00. ) Annual Traffic Volume Report: Central Region. Alaska Department of Transportation and Public Facilities, 00. 0) FFY0 Highway Safety Improvement Program Candidate Description. Alaska Department of Transportation and Public Facilities Central Region Traffic and Safety Section, 0. ) Jiang, Y., S. Li, D.E. Shamo. Development of Vehicle Platoon Distribution Models and Simulation of Platoon Movements on Indian Rural Corridors. Publication FHWA/IN/JTRP-00/. Joint Transportation Research Program, Indiana Department of Transportation and Purdue University, West Lafayette, Indiana, 00. TRB 0 Annual Meeting