Identification and Evaluation of the Cost-Effectiveness of Highway Design Features to Reduce Nonrecurrent Congestion (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

60 Objective The research team developed an analysis tool to implement the analytical procedures developed in this research. The purpose of the Analysis Tool is to allow highway agencies to analyze and compare the effectiveness of a range of design treatments at improving travel time reliability for a given highway seg- ment. As part of a quality control review, the research team performed a series of sensitivity analyses using the tool to iden- tify any errors and to assess the reasonableness of the results it provides to users. This exercise was useful in identifying incon- sistencies in the Analysis Tool itself and in identifying inputs or default values that may cause the Analysis Tool to give unreal- istic results. approach Test scenarios were developed to represent realistic condi- tions for typical freeway sections; test scenarios for extreme conditions were also developed. Data representing these vari- ous sets of conditions were entered into the Analysis Tool. Using the default values for user-defined treatment-specific parameters, results were calculated by the Analysis Tool that predicted the delay savings and reliability measures for each scenario. The net present benefit of each scenario was calcu- lated on the basis of the delay savings, safety benefits (direct and indirect), and reliability improvements. This quality con- trol process was iterative: the Analysis Tool generated results for a set of scenarios, and the research team identified par- ticular treatments or input variable combinations that gave unrealistic results. In these cases, the research team reconsid- ered the assumptions and rationale for choosing these default values and made changes as appropriate. In some cases, errors in the calculations were discovered and corrected. The research team devised a two-pronged approach for test- ing the reasonableness of the Analysis Tool: a manual testing process and an automated procedure. Members of the research team who were not involved in the construction of the Analysis Tool conducted manual testing of the tool. These research team members entered data into the Analysis Tool by hand, just as end users would, and recorded results in a separate document. This approach provided the opportunity for an additional check of user friendliness by users unfamiliar with the tool interface. The 16 scenarios shown in Table 7.1 were tested by using the manual method. Because the manual testing was labor intensive, an auto- mated procedure was developed to rerun the 16 scenarios listed in Table 7.1. By using the automated approach, the results of the 16 scenarios could be quickly plotted in vari- ous ways to identify additional or new unrealistic results that were not identified in the previous iteration. Initial results of reasonableness tests The results of the manual testing were plotted for each of the 16 design treatments in the Analysis Tool. Figure 7.1 presents a plot of the results for crash investigation sites. For more information about the scenarios represented by each of the 16 bars in Figure 7.1, refer to Table 7.1. Plots were also created showing all 16 design treatments applied to one scenario. This comparison was useful in iden- tifying design treatments that appeared to yield unrealistically high or low benefits compared with other design treatments. For example, as Figure 7.2 shows, wildlife crash reductions are estimated to provide a very large net present benefit compared with the other design treatments. Although wildlife crash reduction treatments may be very beneficial in some areas, the research team concluded that this result was due to over- estimation of treatment effectiveness at reducing crashes and underestimation of treatment implementation costs. Adjust- ments were subsequently made to the default parameters in the Analysis Tool and incorporated into future iterations of the automated testing. Treatment costs and default values related C h a p t e r 7 Analysis Tool and Underlying Equations: Test for Reasonableness

61 to effectiveness (such as the number of crashes expected to be reduced by the design treatment) can be adjusted by the user to match local conditions. adjustments to Defaults On the basis of the initial testing described above, the research team modified tool input default values in the following ways: • Corrected an error identified in the net present value calculation. • Corrected an error identified in the calculations for dis- tributing crash totals to each hour of the day. • Wildlife crash reduction 44 Reduced the default percentage reduction of property- damage-only, minor-injury crashes, and major-injury crashes (and other noncrash incidents) associated with this design treatment because for most freeway seg- ments, animal–vehicle collisions make up a small pro- portion of total crashes (as the model is very sensitive to crash reductions, the team chose to err on the side of underestimating benefits with default values); 4 Refined initial values used for default installation cost on the basis of the best available information for wildlife crossing treatments; and Table 7.1. Scenarios Tested by Using the Manual Method Scenario No. of Lanes ADT No. of Incidents Location 1 2 52,500 500 Orlando, Fla. 2 4 105,000 500 3 2 30,000 500 4 4 60,000 500 5 2 52,500 100 6 4 105,000 100 7 2 30,000 100 8 4 60,000 100 9 2 52,500 500 Duluth, Minn. 10 4 105,000 500 11 2 30,000 500 12 4 60,000 500 13 2 52,500 100 14 4 105,000 100 15 2 30,000 100 16 4 60,000 100 Note: ADT = average daily traffic. Scenario An nu al D el ay R ed uc tio n (hr ) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 50 0 10 00 15 00 Scenario N et P re se nt B en ef it ($) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 10 00 0 20 00 0 30 00 0 Figure 7.1. Initial results of manual testing for crash investigation sites.

62 44 Set default fatal crash reduction to 0% to err on the side of a conservative benefit estimate, given that most free- way segments will not experience many fatal animal– vehicle collisions. • Snow fence 44 Refined the default installation cost on the basis of the best available information for the installation and main- tenance cost of a typical snow fence; and 44 Set default fatal crash reduction to 0% to err on the side of a conservative benefit estimate, given that most free- way segments will not experience many snow-related fatal crashes that would be alleviated by a snow fence. • Anti-icing systems 44 Refined the default installation cost on the basis of the best available information on the installation and main- tenance costs for such systems; and 44 Set default fatal crash reduction to 0% to err on the side of a conservative benefit estimate, given that most free- way segments will not experience a significant amount of icy conditions. • Drivable shoulder: reduced default shoulder capacity. • Blowing sand: set default fatal crash reduction to 0% to err on the side of a conservative benefit estimate, given that most freeway segments will not experience a significant amount of blowing sand conditions. • Output the benefit–cost ratio for each design treatment for each scenario and created graphics similar to the delay and net present benefit charts shown in Figures 7.1 and 7.2. Although many default values for cost and crash reduction were altered during the validation process to produce con- servative benefit estimates representing a typical site, analysts using the benefit–cost analysis procedures should change these defaults to better represent the specific characteristics of their site and planned treatment implementation. Final results of reasonableness tests After implementing the changes to the Analysis Tool listed above, the automated procedure was used to generate plots showing delay, net present value improvements, and benefit– cost ratios for each of the 16 scenarios with each of the 16 design treatments. These plots are shown in Figures 7.3 through 7.5. The number above the bar in these three fig- ures indicates the number of lanes (2 or 4); L and H indicate low and high crash counts, respectively. Table 7.2 sum- marizes the scenarios and codes presented in Figures 7.3 through 7.5. An nu al D el ay R ed uc tio n (hr ) 0 Ac ce ss ib le S ho ul de r Al te rn at in g Sh ou ld er An ti- ic in g Sy st em s Bl ow in g Sa nd Co nt ro l(G ate d) Tu rna rou nd s Cr as h In ve st ig at io n Si te D riv ab le S ho ul de r Em er ge nc y Ac ce ss Em er ge nc y Cr os so ve rs Em er ge nc y Pu ll- of f Ex tra H ig h M ed ia n Ba rri er In ci de nt S cr ee n M ov ea bl e Ca bl e M B R un aw ay T ru ck R am p Sn ow F en ce W ild fir e Cr as h Re du ct io n Ac ce ss ib le S ho ul de r Al te rn at in g Sh ou ld er An ti- ic in g Sy st em s Bl ow in g Sa nd Co nt ro l(G ate d) Tu rna rou nd s Cr as h In ve st ig at io n Si te D riv ab le S ho ul de r Em er ge nc y Ac ce ss Em er ge nc y Cr os so ve rs Em er ge nc y Pu ll- of f Ex tra H ig h M ed ia n Ba rri er In ci de nt S cr ee n M ov ea bl e Ca bl e M B R un aw ay T ru ck R am p Sn ow F en ce W ild fir e Cr as h Re du ct io n 20 0 60 0 10 00 N et P re se nt B en ef it ($) 2e +0 5 0e +0 0 4e +0 5 6e +0 5 Figure 7.2. Initial results of manual testing: Scenario 1. (text continues on page 70)

63 Note: Number above bar indicates number of lanes (2 or 4); L = low crash count; H = high crash count. Figure 7.3. Delay reduction results.

64 Figure 7.3. Delay reduction results. (Continued)

65 Figure 7.3. Delay reduction results. (Continued)

66 Figure 7.4. Net present benefit results. Note: Number above bar indicates number of lanes (2 or 4); L = low crash count; H = high crash count.

67 Figure 7.4. Net present benefit results. (Continued)

68 Note: Number above bar indicates number of lanes (2 or 4); L = low crash count; H = high crash count. Figure 7.5. Benefit–cost ratio results.

69 Note: Number above bar indicates number of lanes (2 or 4); L = low crash count; H = high crash count. Figure 7.5. Benefit–cost ratio results. (Continued)

70 Table 7.2. Scenarios and Codes Used in Figures 7.3 Through 7.5 Scenario Location No. of Lanes Total Crash Count Total Volume (ADT) ADT Color Code 1 Orlando, Fla. 2 59 30,000 Light 2 2 59 52,500 Dark 3 2 295 30,000 Light 4 2 295 52,500 Dark 5 4 59 30,000 Light 6 4 59 52,500 Dark 7 4 295 30,000 Light 8 4 295 52,500 Dark 9 Duluth, Minn. 2 59 30,000 Light 10 2 59 52,500 Dark 11 2 295 30,000 Light 12 2 295 52,500 Dark 13 4 59 30,000 Light 14 4 59 52,500 Dark 15 4 295 30,000 Light 16 4 295 52,500 Dark or providing areas off the roadway for crash-involved or disabled vehicles (e.g., crash investigation sites), substan- tial delay reductions can be achieved. • Drivable shoulders provide high net present benefits. Driv- able shoulders were found to provide substantial benefits, especially as compared with other design treatments ana- lyzed by the Analysis Tool (on the typical freeway sections analyzed using the 16 scenarios). On investigation of the default parameters of this design treatment, the assump- tions and results appear to be reasonable. • Benefit–cost calculations are not sensitive to local weather conditions. Weather conditions in Duluth, Minnesota, and Orlando, Florida, are substantially different. However, dif- ferences in net present benefits of design treatments applied in these two locations were negligent. Although rain and snowfall affect the TTI curves for both treated and untreated conditions, they appear to affect these curves proportion- ally, so that the difference between the treated and untreated curves does not change substantially. Findings of reasonableness tests The results of the reasonableness testing of the Analysis Tool and underlying equations led to the following conclusions: • Models are very sensitive to crash frequency. The magnitude of treatment benefits is very sensitive to the annual number of crashes. A relatively small reduction in the annual crash total can result in a substantial increase in treatment ben- efits, particularly if the freeway section being analyzed expe- riences moderate to high congestion at some point during a typical weekday. This result makes sense because a reduction in crash frequency not only results in delay savings and reli- ability improvements, but also provides a direct savings of the cost of the crash itself. • Models are very sensitive to incident duration. The duration of lane-blocking time for incidents has a dramatic impact on treatment benefit. By reducing incident clearance time (continued from page 62)