Analytical Procedures for Determining the Impacts of Reliability Mitigation Strategies (2012)

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.

154 C h a p t e r 8 Introduction The predictive models that can be used in transportation modeling and analysis applications are of three kinds: 1. Adjustment factors (percentage reduction) derived from the before-and-after studies; 2. Relationships between the mean Travel Time Index (TTI) and reliability metrics (i.e., the simple or data-poor model); and 3. Direct prediction of reliability metrics as a function of demand, capacity, and disruption characteristics (i.e., the statistical or data-rich model). This chapter provides general guidance on how to apply these relationships. Implementation of the methods within a spe- cific application (e.g., the Highway Capacity Manual [HCM]) will require greater adaptation to the requirements of those methods. Selecting the appropriate relationship The most direct relationships developed for the impact of improvements on reliability are the adjustment factors from the before-and-after studies. However, as with adjustment factors for other forms of transportation analysis (e.g., safety analysis), care must be exercised in their application. Specifically, the base conditions for the before-and-after case studies should roughly match the conditions for the situation at hand. Therefore, the analyst should examine the details provided in Appendix B for the improvement type of interest and decide if the conditions of the case study are relevant. Only then can the adjustment factors be applied. For many planning-level applications, the data-poor models can be used to generate reliability statistics. Because the rela- tionships are based on first knowing the overall mean TTI (i.e., the average TTI over the course of a year that includes all possible sources of recurring and nonrecurring variation), ana- lysts must identify how many nonrecurring events are included in the estimate of the overall TTI produced by their model. Usually, the overall TTI from planning models includes only recurring congestion, so the adjustment provided in Chapter 7 can be used directly. The basic response variable used in this research is the TTI. In some cases, analysts will want different response metrics. TTI can be converted to other measures if the section length and free-flow speed are known. TTI is the result of dividing the actual travel time by the travel time at the free-flow speed. For example, consider a section 1.5 miles long with a TTI of 1.3. The free-flow travel time (at 60 mph, the free-flow speed in this research) is 1.5 minutes, and the actual travel time is 30% higher, or 1.95 minutes. The travel rate is therefore 1.95 divided by 1.5, or 1.3 minutes per mile. Linking Improvements to Model Variables The final stage of model application is to develop linkages between improvement types and the variables in the data-rich model. Table 8.1 presents a general discussion of how the improvements are to be considered and how their effects are to be accommodated by the models. Basically, the effect of improvements is traced to the changes in the independent variables and their determinants. Within the models, improve- ments can affect • Demand (volume for the time period considered); • Capacity (physical capacity, as determined by the HCM); • Lane hours lost due to incidents and work zones. Work zone lane hours lost must be entered directly. Incident lane hours lost can be entered directly or as changes to 44 Incident frequency, a function of both 4▪ Incident rate, and 4▪ Vehicle miles traveled (VMT) (demand); Application Guidelines

155 Table 8.1. General Linkages Between Improvements and Model Variables Action to Improve Reliability Effect on Reliability Model Variable Affected Add Capacity Add new through lanes Increases design capacity. d/c ratio Add other geometric improvements (lane widening, shoulders, and lower grades) Increases design capacity. d/c ratio Modify interchange (new configuration, longer, or additional ramps) Increases design capacity. d/c ratio Add or modify access control, median barriers Modest design capacity increase, significant reduction in probabil- ity of incidents (collisions). d/c ratio; primary incident rate Add managed lane (truck climbing lanes, high-occupancy vehicle [HOV] and high-occupancy toll lanes) Increases capacity in unmanaged lanes by removing trucks, HOVs, toll payers from stream. Improves reliability for vehicles able to switch to managed lanes (d/c of managed lanes will usually be lower than for unmanaged lanes). d/c ratio Add auxiliary lanes Increases capacity by allowing nonthrough vehicles to use auxiliary lanes. d/c ratio Add new interchange Changes demand by changing access to facility; minor effect on design capacity. d/c ratio Add turn lanes Increases capacity by shifting demand out of through lanes and increasing design capacity of through lanes. d/c ratio Convert two-way to one-way streets Reduces demand by shifting one direction of demand to other streets. Increases design capacity for remaining allowed direction. d/c ratio Add safety improvements (median barriers, eliminate visual obstructions, lighting, and wider lanes) Reduces likelihood of collisions and reduces incident frequency. Primary crash rate Operational Improvements Incident Management Improved equipment for incident detection and verification (CCTV) Reduces incident duration. Average incident duration Improved interagency communications for incident detection and verification Reduces incident duration. Average incident duration Improved equipment and service for incident response Reduces incident duration. Average incident duration Improved interagency incident manage- ment coordination Reduces incident duration. Average incident duration Improved responder training Reduces incident duration. Average incident duration Incident command system Reduces incident duration. Average incident duration Crash investigation sites Reduces lane blockage. Shoulder usability factor (in the lanes blocked per incident calculation) Weather Management More effective deployment of snow and ice resources Reduces impact of weather events on pavement and crashes. Capacity reduction not as severe; primary crash rate Snow and ice pretreatment Reduces impact of weather events on pavement and crashes. Capacity reduction not as severe; primary crash rate (continued on next page)

156 Action to Improve Reliability Effect on Reliability Model Variable Affected Weather Management (continued) Microlevel weather forecasting Reduces impact of weather events on pavement and crashes. Primary crash rate Weather monitoring Reduces crash rates due to better traveler information. Primary crash rate Fog warning system Reduces crash rates due to better traveler information. Primary crash rate Work Zone Management Scheduling (accelerated schedules, night time activities) Reduces work zone duration. Work zone duration Use of more durable materials Reduces frequency of work zone occurrence. Work zone duration Improved signing Increases design capacity; decreases crashes. d/c ratio; primary crash rate Increased enforcement Decreases crashes. Primary crash rate Full road and lane closures Decreases design capacity but reduce work zone duration. d/c ratio; work zone duration Traffic control plan development Increases design capacity. d/c ratio Active Traffic Management Traffic signal coordination More green time; responsive cycle increases capacity. d/c ratio Traffic adaptive signal control Through capacity is increased as demand increases. d/c ratio Ramp metering (fixed time, traffic responsive) Increases design capacity. d/c ratio Integrated corridor management Problematic; current FHWA research may reveal impacts; probably reduces demand and/or increases capacity (d/c). Traveler information system improve- ments (pretrip, roadside, and in-vehicle) Problematic; probably reduces demand. d/c ratio Variable speed limits Increases design capacity. d/c ratio Lane controls Increases design capacity. d/c ratio Queue warning Increases design capacity. d/c ratio Truck lane restrictions Increases design capacity of nontruck lanes. d/c ratio Hard shoulder running during peak Increases design capacity, but also increases incident impacts. d/c ratio; shoulder usability factor Access management Increases design capacity. d/c ratio Traveler Information 511 Reduces demand on event-stricken facilities. d/c ratio Variable message signs (VMS) Reduces demand on event-stricken facilities. d/c ratio In-vehicle guidance Reduces demand on event-stricken facilities. d/c ratio Demand Management Telecommuting Reduces demand. d/c ratio Alternative work hours Shifts demand (changes temporal traffic distribution). d/c ratio Land use controls Reduces demand. d/c ratio Road pricing Reduces demand on priced facility. d/c ratio Parking pricing Reduces demand. d/c ratio Shifts to nonauto modes Reduces demand. d/c ratio Table 8.1. General Linkages Between Improvements and Model Variables (continued)

157 44 Lanes blocked per incident, a function of 4▪ Presence of shoulders, and 4▪ Local policy concerning moving lane-blocking inci- dents to the shoulder; or • Average incident duration. Improvements or strategies that affect demand are accounted for twice in the model: in the demand-to-capacity (d/c) ratio and in the incident frequency calculations. The research team also undertook a review of recent studies of reliability improvements. Although none of them deal directly with estimating reliability, they can still be used in the modeling framework presented above (Tables 8.2 through 8.4). In some cases, a recommendation has been provided on how to adapt these results to the modeling framework. In others, the team has not provided a recommendation, but the results are presented because some practitioners might find them useful. As new research becomes available, especially other SHRP 2 Table 8.2. Incident Management Impacts Improvement Impact Improving from no formal IM program to a program that includes detection, verification, and service patrols Atlanta—Average time between first report and incident verification reduced by 74%. Average time between verification and response initiation reduced by 50%. Average time between incident verification and clearance of traffic lanes reduced by 38%. Maximum time between incident verification and clearance of traffic lanes reduced by 60% (1). Houston—Average 30-minute incident duration reduction (2). RECOMMENDATION IDAS model recommends a default reduction in incident duration of 9% for incident detection, 39% for incident response systems, and 51% for combination incident detection and response systems (3). Georgia (NaviGAtor)—Incident clearance time reduced by an average of 23 min- utes. Incident response time reduced by 30% (4). Maryland (CHART)—Blockage duration from incidents reduced by 36%. This translates to a reduction in highway user delay time of about 42,000 hours per incident (5). 15% to 38% reduction in all secondary crashes, 4% to 30% reduction in rear-end crashes, and 21% to 43% reduction in severe secondary crashes (4). RECOMMENDATION Based on CHART, reduce incident lane hours lost by 36%. Improved equipment for incident detection and verification (CCTV) Brooklyn—Average time required to clear incident from roadway reduced by 66% (6). San Antonio (TransGuide)—20% improvement in response time (21% reduction for major incidents and 19% for minor incidents) (7). RECOMMENDATION Based on TransGuide and assuming that incident response time is 20% of incident duration time, reduce incident duration by 4%. Improved interagency communications for incident detec- tion and verification Minneapolis–St. Paul (Highway Helper)—Automatic tow truck dispatch program is credited with a 20-minute reduction in incident response and removal times (85% improvement) (8). RECOMMENDATION Assuming that response time is 20% of incident duration time, reduce incident duration by 17%. Improved equipment and service for incident response Hayward, California—38% reduction in incident duration, 57% reduction in breakdown duration (9). Northern Virginia—Reduction in duration for all incidents is 2 to 5 minutes for cell phone in response vehicles, 2 to 5 minutes for CAD screens in response vehicles, and 4 to 7 minutes for GPS location for response vehicles (10). Oregon—Duration of delay-causing incidents decreased by approximately 30% on Highway 18 and 15% on Interstate 5 (service patrol addition) (11). Pittsburgh—Service patrol reduced response time to incidents from 17 to 8.7 minutes (12). Washington State—Average freeway incident clearance time for large trucks reduced to 1.5 hours from 5 to 7 hours without incident response team (13). RECOMMENDATION For implementation of service patrols, reduce incident duration by 38%.

158 research projects currently underway, their results can be adapted to the modeling framework in a similar manner. relationship Between Incident Management efficiency and Model Variables The incident management factors in Table 8.2 relate primar- ily to the technological (physical) aspects of incident manage- ment (i.e., equipment deployed to detect, verify, and respond to incidents). However, effective incident management depends not only on equipment but how efficiently the equipment is used and how well responders work together on the incident scene; institutional arrangements and programmatic aspects will determine the level of efficiency. Although it is thought that these attributes influence incident duration, quantifying them for inclusion in a statistical model is a challenging task. Originally it was thought that Traffic Incident Management Self-Assessment scores, which rank the level of sophistication and/or aggressiveness of incident management programs, could be used for this purpose. However, these self-assessment scores were available from only three of the cities used in the urban freeway analysis. A few other key aspects of incident management programs were identified; these were available for six locations. Table 8.5 presents the results; cities are not identified because to obtain this information the research team had to maintain anonym- ity. There appears to be a loose relationship between self- assessment scores and incident duration: higher scores, which indicate greater sophistication or aggressiveness, generally correspond to lower incident duration. However, the sample size here is so small that it is impossible to say with certainty that a mathematical relationship exists. These limited results do suggest that additional work including many more loca- tions is warranted. Induced Demand effects of Improvements It has long been observed that transportation improvements that reduce travel times, especially those related to capacity expansion, become victims of their own success: lower travel times spur increased demand for the improved facility. This phenomenon, known as induced demand, has both short-run and long-run components. In the short run, trips will divert from nearby congested facilities to take advantage of the new lower travel times, and travelers who previously avoided a congested peak period will be drawn back to the peak. In the long run, reductions in travel time are thought to increase the amount of travel (VMT) as lower congestion allows both Table 8.3. Weather Management Impacts Improvement Impact More effective deployment of snow and ice resources Idaho DOT, U.S. Route 12—Mobile anti-icing operations reduced average winter accident frequency by 83% compared with the past 3 years (14). Snow and ice pretreatment Finland (Finnish National Road Administration)—Duration of slippery road conditions estimated to decrease by 10 to 30 minutes per deicing activity, decreasing the chance for accidents caused by slipperiness. Estimated average time saved was 23 minutes per deicing activity (15). Minneapolis, I-35W and Mississippi River Bridge—2000–2001 season had a 50% reduction in total number of crashes over comparison season (1996–1997), even with an increase in average daily traffic of 9.3% (16). Microlevel Weather Forecastinga Weather monitoring Idaho Storm Warning System—Mean speeds in southbound lanes drop from 47.0 mph without dynamic message signs (DMS) to 41.2 mph with DMS warnings (~12% reduction). When high winds occurred with snow- covered pavement, mean speeds in southbound lanes dropped 35% from 54.7 mph to 35.4 mph compared with a 9% decline from 48.4 to 44.1 mph in northbound lanes (17). Fog warning system London Orbital Motorway, M25—Fog messages were followed by a statis- tically significant overall net reduction in mean vehicle speeds of about 1.8 mph. (18). Utah Fog Warning System, I-215—Average vehicle speed measured during fog events increased from 54 to 62 mph after system was deployed. Speed increase was partly attributable to reduction in the number of excessively slow drivers during fog events (19). Salt Lake Valley—15% increase in speeds and 22% decrease in standard deviation of those speeds under foggy conditions (20). a No recommendations made for weather strategies’ impacts on reliability.

159 longer and more trips to be made. Longer trips can result from the location decisions for place of residence and business. The converse is that congestion suppresses these aspects of travel. Short-run induced demand can be studied via travel demand models that account for diversion of traffic from parallel facili- ties to an improved highway, shifts of travelers from other modes, and (depending on how the models are applied) the role of improved highways in causing people to shift the desti- nations of their trips. However, the models usually do not account for the effects of highway improvements on the total number of trips made and shifts in the locations of households, businesses, and other activities. In previous studies, the induced demand effect has been quantified as elasticities of VMT with respect to highway travel time or lane miles. Travel time elasticities have been used in sketch planning analyses to estimate the aggregate response of travelers to transportation system improvements that provide time savings. The elasticities indicate the per- centage change in VMT expected to result from a 1% change in travel time or lane miles. Cohen provided a summary of these studies (Table 8.6) (27). The results of Barr and Gorina are especially relevant because of the use of travel time as the causal factor. Their elasticities were in the -0.1 to -0.4 range, indicating that a 10% decrease in travel rate would cause a 1% to 4% increase in household VMT. These increases in VMT include the effects of modal diversion, trip distribution (in this case, substituting longer trips for shorter trips), and increases in the total number of person trips made. For an individual facility, it would be expected that time sav- ings would cause a greater increase in VMT than those sug- gested by the above elasticities. VMT increases occur because traffic increases on individual facilities include not only the Table 8.4. Active Traffic Management Impacts Improvement Impact Traffic signal coordination Phoenix—6.2% to 8% average increase in trip speeds (21). IDAS model uses a capacity increase of 14 to 20%. Actual increase value is sensitive to traffic variability and frequency of retiming (3). RECOMMENDATION Decrease mean TTI by 7%. Traffic adaptive signal control Los Angeles (ATSAC)—Travel time reduced by 12% to 18%, delay reduced by 44%, speed increased by 16% (22). Minneapolis (SCOOT)—Installation in 56 intersections showed 19% reduction in delay during special events, 8% during peaks (12). Oakland County, Michigan (SCATS)—Corridor travel time reduced from 7% to 32% over optimized fixed-time signal control. Average travel time reduction of 8% (average speed increased from 25 to 27 mph) (12). IDAS model recommends a default capacity increase of 8 to 14%. Actual increase value is sensitive to traffic variability. Assumes upgrade from coordinated preset timing (3). Dallas (North Central Expressway)—15% increase in speed, 15% decrease in delay (23). RECOMMENDATION Reduce mean TTI by 12%. Ramp metering (fixed time) Portlan, Oregon—25% increase in volume (24). Portland, Oregon—43% reduction in peak period accidents (13). Houston—29% increase in speed (25). IDAS model uses a default mainline capacity increase of 9.5% offset by a ramp capacity decrease of 33%. IDAS also suggests a reduction in accidents of 30% on ramp and adjacent freeway links (3). Minneapolis–St. Paul—14% average increase in throughput, 7% increase in corridor speed, 26% decrease in peak period accidents (26). Denver—19% increase in volume (24). Seattle (I-405 in 1997)—5% to 6% increase in volume (24). IDAS model uses a default mainline capacity increase of 13.5% offset by a ramp capacity decrease of 28%. IDAS also suggests a reduction in accidents of 30% on ramp and adjacent freeway links (3). RECOMMENDATION Based on the Seattle before and after study presented in Chapter 6, use the following adjustments: 11% reduction in average travel time and 12% reduction in Planning Time Index. VMS/DMS Austin—7% to 12% reduction in upstream lane volumes of an incident (13). RECOMMENDATION For peak hour and peak period only, reduce demand volume by 3.5% (assumes 9% reduction in volumes during an incident and that incidents comprise 40% of total delay).

160 three effects noted above (modal diversion, trip distribution, and trip frequency), but also route diversion (in which travel- ers shift the routes they use but do not alter their origins or destinations). These previous studies considered only changes in average travel times and did not include the effect of reliability on induced demand. However, it has been noted that travel time reliability has additional value to travelers beyond consider- ation of average or typical conditions (28). To the extent this is true, improvements in reliability may have an additional effect on induced demand. One approach to this issue may be to convert reliability improvements to equivalent travel time units. For example, Bates et al. measured variability as the standard deviation of travel time and found the value of vari- ability reductions to be equal to 0.8 to 1.3 times the value of mean travel time reductions (29). Brownstone and Small measured variability as the difference between the 90th and 50th percentile travel times and found the value of variability reductions to be roughly equal to the value of mean travel time reductions (30). However, the merit of adding a reliability factor to changes in mean travel time may be dubious. If elasticities are based on empirical data collected over a sufficiently long period of time so that they include the effect of disruptions, then add- ing a reliability factor would be double counting. That is, to the extent that observed travel times are overall mean travel times that include both recurring and nonrecurring sources, then the relationships identified in Chapter 7 indicate that an improvement in the overall mean also means that reli- ability has improved. If this is the case, then the reliability effect is already embedded in the observed increases in travel activity. Table 8.5. Institutional and Programmatic Characteristics on Incident Management Programs in Study Locations Urban Area Traffic Incident Management Self-Assessment Quick- Clearance Law Property Damage Only Move-to- Shoulder Law Can a Fatality Be Moved with Medical Examiner Present? Average Peak Period Incident Duration (min) Overall Score Programmatic and Institutional Operational Communications and Technology 1 85.9 27.5 32.1 26.3 Yes Yes Yes 32.1 2 82.0 25.5 32.1 24.4 Yes Yes Yes 43.5 3 74.0 21.3 29.3 23.4 Yes Yes Yes 45.0 4 NA NA NA NA No No No 47.3 5 NA NA NA NA No Yes No 52.0 6 NA NA NA NA No Yes No 61.5 Note: NA = not available. Table 8.6. Summary of Elasticities Used for Induced Demand (27) Change in Long-Run VMT Elasticity (%) Study Primary Data Source Travel Time Lane Miles Comment Barr and Gorina 1990 and 1995 Nationwide Personal Transportation Survey (NPTS) -0.3 to -0.5 NA Elasticities may be overstated because of the tendency for longer trips to have higher average speeds than shorter trips. Reanalysis suggests elasticities of -0.1 to -0.4. SACTRA Fuel price elasticities -1.0 NA Elasticity may be overstated because of differences in opportuni- ties available to motorists to reduce travel time and fuel costs. Noland Highway statistics NA -0.8 Elasticity may be overstated because (a) of shifts of VMT and lane miles among highway systems and (b) highways that are widened have more VMT/lane mile than other highways. Strathman 1995 NPTS, TTI Urban Mobility Study data set NA -0.32 Elasticity includes direct effects of lane miles on household VMT and indirect effects due to changes in density. Marshall TTI Urban Mobility Study data set NA -0.76 to -0.85 Elasticity may be overstated because of roadway classification issues and diversion from outside urban areas.

161 The situation is further clouded because no empirical studies have been done on the induced demand effect of operational treatments. Unlike capacity expansions (the basis of previous elasticity work), which improve recurring congestion every day, operational treatments only affect those conditions when dis- ruptions occur (e.g., incidents and work zones). Although the effect of operational treatments can be tracked to a reduction in overall mean travel times, which in theory should have an induced demand effect, it is still not known if an improvement in travel times on a few days affects travel behavior in the same way as travel time improvements that affect every day. These issues are sufficiently complex to warrant additional study. This project did not attempt to address these issues, but focused on the immediate or first-order impacts of improve- ment strategies on reliability. As new research becomes available that quantifies induced demand effects, it can be incorporated with the relationships developed in the present study. This process would involve three steps: 1. Estimate the first-order change in mean travel time and reliability measures. 2. Increase demand using elasticities from new research. The pivot point formulation is a convenient way to implement elasticities; for example, V V T T= ( )0 0 β where V = new volume, including induced demand; V0 = original volume, before the improvement; T = travel time after the improvement; T0 = travel time before the improvement; and b = elasticity. 3. Reestimate the mean travel time and reliability measures using the new (increased) demand values. references 1. Booz-Allen & Hamilton. Executive Summary: 1996 Olympic and Paralympic Games: Event Study. Federal Highway Administration, U.S. Department of Transportation, 1997. http://ntl.bts.gov/lib/ jpodocs/rept_mis/3006.pdf. Accessed May 17, 2012. 2. Intelligent Transportation Systems Joint Program: Benefit. Research and Innovative Technology Administration, U.S. Department of Transportation, 1997. www.itsbenefits.its.dot.gov/its/benecost.nsf/0/ 0B37A6D584E620B2852569610051E268. 3. Cambridge Systematics, Inc. IDAS User’s Manual. Federal Highway Administration, U.S. Department of Transportation, 2000. http:// idas.camsys.com/documentation.htm. Accessed May 21, 2012. 4. Institute of Transportation Engineers. 1996 ITS Tour Report: East- ern North America and 1996 ITS World Congress: Volume I. 1997. 5. Chang, G.-L., and S. Rochon. Performance Evaluation and Benefit Analysis for CHART in Year 2007. Final report. 2007. http://chartin put.umd.edu/reports/chart2007final.pdf. 6. Stough, R. R. (ed.). Intelligent Transport Systems: Cases and Policies. Edward Elgar Publishing, United Kingdom, 2001. 7. Turner, S., W. R. Stockton, S. James, T. Rother, and C. M. Walton. ITS Benefits: Review of Evaluation Methods and Reported Benefits. Report 1790-1. Texas Department of Transportation, Austin, October 1998. http://tti.tamu.edu/documents/1790-1.pdf. 8. ATA Foundation, and Cambridge Systematics, Inc. Incident Manage- ment: Challenges, Strategies, and Solutions for Advancing Safety and Roadway Efficiency. Report. Research and Innovative Technology Administration, U.S. Department of Transportation, February 1997. www.itsbenefits.its.dot.gov/its/benecost.nsf/ID/19F4564DD66A22 F6852569610051E2E7?OpenDocument&Query=BMeasure. 9. Skabardonis, A., H. Noeimi, K. Petty, D. Rydzewski, P. P. Varaiya, and H. Al-Deek. Freeway Service Patrol Evaluation. Research report UCB-ITS-PRR-95-5. California PATH, University of California, 1995. http://escholarship.org/uc/item/36r1t2m2. 10. Maas, G., M. Maggio, H. Shafie, and R. R. Stough. Incident Manage- ment and Intelligent Transportation Systems Technology: Estimat- ing Benefits for Northern Virginia. In Intelligent Transport Systems: Cases and Policies (R. R. Stough, ed.), Edward Elgar Publishing, United Kingdom, 2001. 11. Bertini, R. L., S. Tantiyanugulchai, E. Anderson, R. Lindgren, and M. Leal. Evaluation of Region 2 Incident Response Program Using Archived Data. Report No. PSU-CE-TRG-01-01. Oregon Department of Trans- portation, Salem, 2001. www.its.pdx.edu/pdf/IR_FinalReport.pdf. 12. What Have We Learned About Intelligent Transportation Systems? Fed- eral Highway Administration, U.S. Department of Transportation, 2000. http://ntl.bts.gov/lib/jpodocs/repts_te/13316.pdf. 13. Hall, R. Incident Dispatching, Clearance and Delay. Working paper UCB-ITS-PWP-2000-14. California PATH, University of California, 2000. www.path.berkeley.edu/PATH/Publications/PDF/PWP/2000/ PWP-2000-14.pdf. 14. Breen, B. D. Anti Icing Success Fuels Expansion of the Program in Idaho. Idaho Department of Transportation, Boise, 2001. http://ops .fhwa.dot.gov/weather/publications/antiicingidaho.pdf. 15. Pilli-Sihvola, Y., K. Toivonen, and J. Kantonen. Road Weather Service System in Finland and Savings in Driving Costs. In Transportation Research Record 1387, TRB, National Research Council, Washington, D.C., 1993, pp. 196–200. 16. Johnson, C. I-35W & Mississippi River Bridge Anti-Icing Project: Operational Evaluation Report. Minnesota Department of Trans- portation, Saint Paul, July 2001. www.transportation.org/sites/ sicop/docs/Anti-icing%20evaluation.pdf. Accessed Aug. 13, 2012. 17. Kyte, M., P. Shannon, and F. Kitchener. Idaho Storm Warning System Operational Test. Final report. Idaho Transportation Department, Boise, December 2000. http://ntl.bts.gov/lib/11000/11100/11191/ cc01.pdf. 18. Cooper, B. R., and H. E. Sawyer. Assessment of M25 Automatic Fog- Warning System: Final Report. Safety Resource Centre, Transporta- tion Research Laboratory, Crowthorne, Berkshire, United Kingdom, 1993. www.trl.co.uk/online_store/reports_publications/trl_reports/ cat_intelligent_transport_systems/report_assessment_of_m25_ automatic_fog-warning_system_-_final_report.htm. 19. Martin, P. T., J. Perrin, and B. Coleman. Adverse Visibility Information System Evaluation (ADVISE): Interstate 215 Fog Warning System: Final Report. University of Utah and Utah Department of Transportation, Salt Lake City, 2003. http://ops.fhwa.dot.gov/weather/best_practices/ 1024x768/transform_param2.asp?xslname=pub.xsl&xml name=publications.xml&keyname=62. 20. ITS Benefits Database. Research and Innovative Technology Administration, U.S. Department of Transportation, 2004. www.

162 itsbenefits.its.dot.gov/its/benecost.nsf/ID/E4BFDFEF91B742D285 256FB90052C4BB?OpenDocument&Query=Home. 21. Van Aerde, M., and H. Rakha. The Metropolitan Model Deploy- ment Initiative: Evaluating Throughput and Efficiency Impacts. Presented at Intelligent Transportation Systems of America (ITSA) Conference, 1999. 22. ITS Benefits: Continuing Successes and Operational Test Results. Federal Highway Administration, U.S. Department of Transporta- tion, 1997. www.fhwa.dot.gov/publications/research/operations/ its/98002/contsuccess.pdf. 23. Turnbull, K. F., and R. H. Henk (eds.). Transportation Research Cir- cular 474: 2nd National Symposium on Integrated Transportation Management Systems. TRB, National Research Council, Washing- ton, D.C., 1997. http://onlinepubs.trb.org/onlinepubs/circulars/ circ474/newcover.html. 24. Piotrowicz, G., and J. Robinson. Ramp Metering Status in North America: 1995 Update. Federal Highway Administration, U.S. Department of Transportation, 1995. www.oregon.gov/ODOT/ HWY/ITS/pdfs/benefitsofits/rampmeteringstatus1995.pdf. 25. Turner, S., W. R. Stockton, S. James, T. Rother, C. Michael. ITS Ben- efits: Review of Evaluation Methods and Reported Benefits. Texas Department of Transportation and Federal Highway Administra- tion, U.S. Department of Transportation. 26. Cambridge Systematics, Inc. Twin Cities Ramp Meter Evaluation: Final Report. Minnesota Department of Transportation, Saint Paul, 2001. http://ntl.bts.gov/lib/jpodocs/repts_te/13425.pdf. 27. Cohen, H. The Induced Demand Effect: Evidence from National Data. Presented at Conference on Working Together to Address Induced Demand, Eno Transportation Foundation Policy Forum, 2001. 28. Small, K. A., R. Noland, X. Chu, and D. Lewis. NCHRP Report 431: Valuation of Travel-Time Savings and Predictability in Congested Conditions for Highway User-Cost Estimation. TRB, National Research Council, Washington, D.C., 1999. 29. Bates, J., J. Polak, P. Jones, and A. Cook. The Valuation of Reliability for Personal Travel. Transportation Research Part E, Vol. 37, No. 2, 2001, pp. 191–229. 30. Brownstone, D., and K. A. Small. Valuing Time and Reliability: Assessing the Evidence from Road Pricing Demonstrations. Trans- portation Research Part A, Vol. 39, No. 4, 2005, pp. 279–293.