National Academies Press: OpenBook

Highway Safety Manual User Guide (2022)

Chapter: 3 Integrating the HSM in the Project Development Process

« Previous: 2 Highway Safety Manual Overview
Page 52
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 52
Page 53
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 53
Page 54
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 54
Page 55
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 55
Page 56
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 56
Page 57
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 57
Page 58
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 58
Page 59
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 59
Page 60
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 60
Page 61
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 61
Page 62
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 62
Page 63
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 63
Page 64
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 64
Page 65
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 65
Page 66
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 66
Page 67
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 67
Page 68
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 68
Page 69
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 69
Page 70
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 70
Page 71
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 71
Page 72
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 72
Page 73
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 73
Page 74
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 74
Page 75
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 75
Page 76
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 76
Page 77
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 77
Page 78
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 78
Page 79
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 79
Page 80
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 80
Page 81
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 81
Page 82
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 82
Page 83
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 83
Page 84
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 84
Page 85
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 85
Page 86
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 86
Page 87
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 87
Page 88
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 88
Page 89
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 89
Page 90
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 90
Page 91
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 91
Page 92
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 92
Page 93
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 93
Page 94
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 94
Page 95
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 95
Page 96
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 96
Page 97
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 97
Page 98
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 98
Page 99
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 99
Page 100
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 100
Page 101
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 101
Page 102
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 102
Page 103
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 103
Page 104
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 104
Page 105
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 105
Page 106
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 106
Page 107
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 107
Page 108
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 108
Page 109
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 109
Page 110
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 110
Page 111
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 111
Page 112
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 112
Page 113
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 113
Page 114
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 114
Page 115
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 115
Page 116
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 116
Page 117
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 117
Page 118
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 118
Page 119
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 119
Page 120
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 120
Page 121
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 121
Page 122
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 122
Page 123
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 123
Page 124
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 124
Page 125
Suggested Citation:"3 Integrating the HSM in the Project Development Process." National Academies of Sciences, Engineering, and Medicine. 2022. Highway Safety Manual User Guide. Washington, DC: The National Academies Press. doi: 10.17226/26552.
×
Page 125

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.

52 3 Integrating the HSM in the Project Development Process Program and project decisions are typically based on evaluation of costs, right-of-way, traffic operations, and environmental factors. The HSM provides science-based methods and a reliable approach for quantifying safety impacts in terms of crash frequency and severity, allowing agencies to incorporate it throughout the project development process. This section of the Highway Safety Manual User Guide provides examples for incorporating HSM approaches into each of the stages of the project development process: Planning, Alternatives Development and Analysis, Preliminary Design, Final Design and Construction, and Operations and Maintenance. In the Planning phase, agencies assess conditions, evaluate future multimodal projects, identify locations with potential for crash reduction, and develop policies to address long-term transportation system needs, among other tasks. The planning phase includes development of the agency’s long-range program. The program may account for projects for the next 5 years and that are prioritized based on several factors including safety. Application of safety in decisions at a planning level may include developing statewide policies that incorporate quantitative safety implications to reduce the number and severity of crashes in the long term. Incorporation of safety performance at this stage improves the likelihood of cost-effective resource allocation. HSM Part B provides information for planning applications. Individual projects derived from the agency planning efforts move into the Alternatives Development and Analysis phase. In this phase, multiple alternatives are developed and evaluated. Project decisions are based on evaluation of costs, right-of-way, traffic operations, environmental assessment, and safety. The HSM Part C predictive methods allow agencies to quantify a project’s potential for crash reduction, or to apply the predictive method and compare the safety performance of different alternatives associated with a change in traffic volume, traffic control, or geometrics. After a preferred alternative has been selected, the next phase is the Preliminary and Final Design. Tools provided in the HSM can help designers reach informed decisions throughout Final Design and Construction. Some applications include the incorporation of human factor considerations into design, analysis, decision-making, and documentation of the quantitative safety effects of a proposed design exception. The HSM can also be used in the Operations and Maintenance of an agency’s daily operations. The HSM can be incorporated into processes used to monitor system performance, such as considering the impact of changes or upgrades in mobility, decisions related to access, setting maintenance policies and priorities, and other operational considerations on safety performance. This guide will provide examples of HSM applications in the different phases of the project development process, with the intent to provide agencies with opportunities to use safety performance as a consideration in their decision-making process.

53 3.1 HSM in the Planning Phase 3.1.1 Overview The main goal of system planning is to provide decision-makers with the information needed to make choices about investments in their transportation system. In the planning phase, agencies evaluate the multimodal transportation system and identify priorities, programs, and policies to address long-range transportation needs. The HSM can be used to estimate the safety performance of alternative transportation networks and understand safety implications so that reducing the cost of crashes and saving lives can be compared with other performance metrics. HSM Part B provides the process for planning applications and presents steps to monitor and reduce crash frequency and severity on existing road networks. The HSM Part B includes methods useful for identifying sites for improvement (HSM Chapter 4), diagnosis (HSM Chapter 5), safety countermeasure selection (HSM Chapter 6), economic appraisal (HSM Chapter 7), project prioritization (HSM Chapter 8), and effectiveness evaluation (HSM Chapter 9). The following section includes an application of HSM in planning using the different chapters included in Part B. 3.1.2 Example Problem 1: Planning Application using HSM Part B Introduction A county DOT is working on preparing their long-range safety program and chose to use the HSM roadway safety management process to maximize limited safety resources to save lives and reduce serious injuries for routes and intersections within its jurisdiction. The roadway safety management process from network screening to safety effectiveness evaluation will be applied in this example. Step 1: Network Screening Data Requirements • Crash data for the selected county route system • Roadway network information for the selected county route system Analysis Since the county DOT will identify projects for the county safety program, all sites within the county route system should be screened. Both intersections and roadway segments will be included as elements for the network screening process. The reference populations include rural two-lane undivided highways for roadway segments, as well as all-way stop-controlled intersections and four-leg signalized intersections. The roadway segment mileage information was available, and the selected performance measure needed to account for RTM. The excess expected average crash frequency with EB adjustments was selected as the performance measure for the intersections and roadway segments (Figure 21). The screening methods used for the intersections and roadway segments are the simple ranking method and the sliding-window method, respectively.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 54 Figure 21: Available Performance Measures (HSM Table 4-2 [HSM p. 4-9]) Results and Discussion After the network screening process, the average excess expected average crash frequency (EEACF) with EB adjustments was calculated for all the intersections and roadway segments and ranked in descending order. The top five intersections and top five roadway segments (listed in Table 32) will be selected as candidates for the next step. TABLE 32 Example Problem 1 – Network Screening Process – Intersection and Roadway Segment Rankings Rank Intersection ID EEACF with EB Rank Roadway Segment ID EEACF with EB 1 17 18.2 1 52 7.6 2 83 16.4 2 72 5.1 3 25 15.8 3 105 5.0 4 68 12.2 4 35 3.3 5 46 9.8 5 81 3.5 Notes: EB = Empirical Bayes EEACF = excess expected average crash frequency ID = identification number Step 2: Diagnosis Data Requirements • Crash data for the selected five roadway segments and five intersections • Supporting documentation including current traffic volumes for all travel modes, inventory of field conditions, and relevant photo or video logs

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 55 Analysis Descriptive crash statistics for the selected five roadway segments and five intersections were developed. Information on crash type, crash severity, roadway, and environmental conditions was displayed with bar charts, pie charts, and tabular summaries to gain a better understanding of potential issues. Locations for intersection and roadway segment crashes were summarized using collision diagrams and crash maps, respectively. Additional supporting documentation was reviewed; including traffic signs, traffic control devices, number of travel lanes, and posted speed limits. A field visit was arranged by the county traffic engineers to understand issues identified and to verify opportunities to reduce crash potential. Results and Discussion Based on the analysis results, a large proportion of the roadway segment crashes were roadway departure involving high speed. The percentages of nighttime crashes, during inclement weather conditions, and that were ice/snow-related were also relatively high for roadway segments. For intersections, rear-end and angle crashes are overrepresented. Most of the rear-end crashes occurred on the major road approaches, and a high percentage of the at-fault vehicles were speeding. The angle crashes were overrepresented at some unsignalized intersections, which were the result of vehicles making left turns from minor roads or driveways onto the main road. The field condition inventory indicated that the posted speed limit was 50 mph for most roadway segments. The field visit revealed that no Chevrons were installed for curves on roadway segments, and most of the roadway segments did not have rumble strips. The pavement drainage was operating at less than full capacity, resulting in potential flooding for some roadway segments and intersections. The clearance time at some signalized intersections did not seem to provide enough time for vehicles to clear the intersections, and there were some obstructions (such as bushes) on the roadside limiting the stopping sight distance on the main road, minor roads, and driveways. At intersections, traffic signals did not have backplates, and there was only one signal head for all through travel lanes. Step 3: Select Countermeasures Data Requirements No additional data are required for selecting the proper safety countermeasures. Analysis The contributing factors for roadway segment and intersection crashes were identified based on the information derived from the crash data analysis and the field visit process. The possible contributing factors for prevailing crash types were identified from the perspective of human, vehicle, and roadway before, during, and after the crashes. Results and Discussion The contributing factors and selected safety countermeasures for different facility types and crash types are listed in Table 33.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 56 TABLE 33 Example Problem 1 – Contributing Factors and Selected Safety Countermeasures Facility Type Crash Type Contributing Factor Safety Countermeasure Selected Location ID Intersection Rear-end High approach speed Install automated speed enforcement 83 Slippery pavement Install high-friction surface treatment 25 Poor visibility of signals Install one traffic signal head per lane and add backplates 68, 25 Install flashing beacons as advance warning 25 Angle Limited sight distance Increase sight distance triangle 17, 25 High approach speed Install automated speed enforcement 46, 17 Poor visibility of signal Install one traffic signal head per lane and add backplates 25 Roadway Segment Roadway departure Poor delineation Install Chevrons on curved segment 105, 81 Excessive speed Install automated speed enforcement 35, 105 Drive inattention Install shoulder rumble strips 52, 72 Slippery pavement Install high-friction surface treatment 81 Step 4: Economic Appraisal Data Requirements • Crash data for selected roadway segments and intersections • Current and future AADT values • CMFs for all safety countermeasures under consideration • Construction and implementation costs for each countermeasure • Monetary value of crashes by severity • Service life of the countermeasures Analysis The economic appraisal process outlined in this example only considers changes in crash frequency and does not consider project benefits from travel time, environmental impacts, or congestion relief. The method selected for conducting the economic appraisal of this example is the BCR. The predictive method presented in HSM Part C was applied to determine the expected crash frequency for existing conditions and proposed alternatives. The expected change in average fatal, injury, and PDO crash frequency was then converted to a monetary value using the societal cost of crashes listed in HSM Table 7-1. The annual monetary value was further converted to present value using a discount rate. The costs for implementing the selected safety countermeasures, right-of-way acquisition, construction material costs, utility relocation, maintenance, and other costs were added together to obtain the present values of the project costs. The BCR for each project was calculated based on the present value of the project benefits and project costs.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 57 Results and Discussion The benefits and costs for each proposed project and the relevant BCR are listed in Table 34. TABLE 34 Example Problem 1 – Proposed Projects BCR Project Facility ID Benefit Cost BCR Increase triangle sight distance Intersection 17 $34,500 $9,000 3.8 Intersection 25 $32,000 $11,000 2.9 Install one traffic signal head per lane and add backplates Intersection 68 $26,300 $7,800 3.4 Intersection 25 $28,650 $6,900 4.2 Install flashing beacons as advanced warning Intersection 25 $30,750 $10,600 2.9 Install Chevrons Roadway segment 105 $200,500/mile $80,700/mile 2.5 Roadway segment 81-1 $180,650/mile $59,800/mile 3.0 Install shoulder rumble strips Roadway segment 72 $90,800/mile $38,500/mile 2.4 Roadway segment 52 $102,500/mile $42,980/mile 2.4 Install high-friction surface treatment Roadway segment 81-2 $250,200/mile $190,080/mile 1.3 Intersection 25 $85,650 $59,000 1.5 Install automated speed enforcement Intersection 83 $57,000 $25,000 2.3 Intersection 46 $63,000 $27,500 2.5 Intersection 17 $72,000 $26,000 2.9 Roadway segment 35 $87,000 $23,000 3.5 Roadway segment 105 $92,000 $29,000 3.7 Step 5: Prioritize Projects Data Requirements No additional data are required for selecting the proper safety countermeasures. Analysis An incremental benefit-cost analysis was conducted for the project prioritization. The incremental BCR is an extension of the BCR method. Projects with a BCR greater than 1.0 are arranged in increasing order based on their estimated cost. Then HSM Equation 8-3 (Page 8-11) is applied to project pairs. If the incremental BCR is greater than 1.0, the higher cost project is the preferred one. Conversely, if the incremental BCR is less than 1.0, or is zero or negative, the lower cost project is preferred to the higher cost project. The calculations continue comparing the preferred project from the first pair to the next highest cost. Additional details on this method can be found in HSM Section 8.2.1, Ranking Procedures (HSM p. 8-3). Table 35 illustrates the first sequence of incremental benefit-cost comparisons needed to assign priority to projects. From this table, the improvement project for roadway segment 81 – Install Chevrons receives the highest priority (highlighted in green in the table).

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 58 TABLE 35 Example Problem 1 – Incremental BCR Analysis Comparison Project Project ID PVbenefits PVcosts BCR Incremental BCR Preferred Project 1 Install one traffic signal head per lane and add backplates Int 25 $28,650 $6,900 4.15 (2.61) Int 25 Install one traffic signal head per lane and add backplates Int 68 $26,300 $7,800 3.37 2 Install one traffic signal head per lane and add backplates Int 25 $28,650 $6,900 4.15 2.79 Int 17 Increase triangle sight distance Int 17 $34,500 $9,000 3.83 3 Increase triangle sight distance Int 17 $34,500 $9,000 3.83 (2.34) Int 17 Install flashing beacons as advanced warning Int 25 $30,750 $10,600 2.9 4 Increase triangle sight distance Int 17 $34,500 $9,000 3.83 (1.25) Int 17 Increase triangle sight distance Int 25 $32,000 $11,000 2.91 5 Increase triangle sight distance Int 17 $34,500 $9,000 3.83 2.88 Seg 105 Install automated speed enforcement Seg 105 $92,000 $29,000 3.68 6 Install automated speed enforcement Seg 105 $92,000 $29,000 3.68 0.83 Seg 35 Install automated speed enforcement Seg 35 $87,000 $23,000 3.48 7 Install automated speed enforcement Seg 35 $87,000 $23,000 3.48 (5.00) Seg 35 Install automated speed enforcement Int 17 $72,000 $26,000 2.88 8 Install automated speed enforcement Seg 35 $87,000 $23,000 3.48 (5.33) Seg 35 Install automated speed enforcement Int 46 $63,000 $27,500 2.52 9 Install automated speed enforcement Seg 35 $87,000 $23,000 3.48 (15.00) Seg 35 Install automated speed enforcement Int 83 $57,000 $25,000 2.28

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 59 TABLE 35 Example Problem 1 – Incremental BCR Analysis Comparison Project Project ID PVbenefits PVcosts BCR Incremental BCR Preferred Project 10 Install automated speed enforcement Seg 35 $87,000 $23,000 3.48 0.25 Seg 35 Install shoulder rumble strips Seg 72 $90,800 $38,500 2.36 11 Install automated speed enforcement Seg 35 $87,000 $23,000 3.48 0.78 Seg 35 Install shoulder rumble strips Seg 52 $102,500 $42,980 2.38 12 Install automated speed enforcement Seg 35 $87,000 $23,000 3.48 (0.04) Seg 35 Install high-friction surface treatment Int 25 $85,650 $59,000 1.45 13 Install automated speed enforcement Seg 35 $87,000 $23,000 3.48 2.54 Seg 81 Install Chevrons Seg 81 $180,650 $59,800 3.02 14 Install Chevrons Seg 81 $180,650 $59,800 3.02 0.95 Seg 81 Install Chevrons Seg 105 $200,500 $80,700 2.48 15 Install Chevrons Seg 81 $180,650 $59,800 3.02 0.53 Seg 81 Install high-friction surface treatment Seg 81 $250,200 $190,080 1.32 Notes: Int = intersection PV = present value Seg = roadway segment The process is repeated to assign priorities to the remaining projects. Successive series of incremental BCR calculations are performed, each time removing the projects previously prioritized. Results and Discussion The incremental BCR analysis results are summarized in Table 36. This method provides a priority ranking list of projects based on whether the expenditure represented by each increment of additional cost is economically justified. BCR analysis provides additional insight into priority ranking but does not necessarily incorporate a formal budget constraint. TABLE 36 Example Problem 1 – Ranking Results of Incremental BCR Analysis Rank Project ID Project 1 Roadway segment 81 Install Chevrons 2 Roadway segment 105 Install Chevrons 3 Roadway segment 35 Install automated speed enforcement 4 Roadway segment 105 Install automated speed enforcement 5 Roadway segment 81 Install high-friction surface treatment

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 60 TABLE 36 Example Problem 1 – Ranking Results of Incremental BCR Analysis Rank Project ID Project 6 Roadway segment 52 Install shoulder rumble strips 7 Roadway segment 72 Install shoulder rumble strips 8 Intersection 17 Install automated speed enforcement 9 Intersection 46 Install automated speed enforcement 10 Intersection 83 Install automated speed enforcement 11 Intersection 25 Install high-friction surface treatment 12 Intersection 17 Increase triangle sight distance 13 Intersection 25 Install one traffic signal head per lane and add backplates 14 Intersection 25 Increase triangle sight distance 15 Intersection 25 Install flashing beacons as advanced warning 16 Intersection 68 Install one traffic signal head per lane and add backplates Step 6: Safety Effectiveness Evaluation Data Requirements • Minimum of 10 sites at which the treatment has been implemented • Minimum of 3 years of crash data and traffic volume for the period before implementation • Minimum of 3 years of crash data and traffic volume for the period after implementation • Safety performance function for the facility types being evaluated Analysis An EB before/after safety evaluation method was conducted for the safety effectiveness evaluation. The county DOT decided to upgrade all its signalized intersections to one signal head per travel lane. The safety effectiveness is analyzed using the EB before/after safety evaluation method to assess the overall effect of signal upgrades. To simplify things, the county DOT assumed a constant AADT across all years for both the before and after periods. The county DOT also assumed that all the intersections match the base conditions; therefore, the applicable CMFs and calibration factor are 1.0. The EB method is used to compare crash frequencies at a group of sites before and after a treatment is implemented. The EB method addresses the RTM issue. The process begins by estimating the before and after predicted average crash frequency using the sites’ SPF. Then, an adjustment factor is calculated to account for differences between the before and after periods in number of years and traffic volume at each site. The adjustment factor is obtained by dividing the after predicted crash frequency by the before predicted crash frequency. Next, the expected average crash frequency over the entire after period, in the absence of the treatment, is calculated. The safety effectiveness of the treatment at each site is estimated by dividing the observed crash frequency in the after period by the expected average crash frequency in the after period without treatment. The safety effectiveness is then converted into a percentage crash change for each site. Lastly, the overall unbiased safety effectiveness as a percentage change in crash frequency is obtained using the overall effectiveness of the treatment for all sites, overall variance, and total expected average crash frequency

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 61 in the after period without treatment. A similar example that can be used as a reference can be found in HSM Section 9.10. Results and Discussion Results of this evaluation indicated that there is an overall positive safety effectiveness of 25.5 percent (reduction in total crash frequency) with a standard error of 12.7 percent after the application of the treatment or an overall safety benefit between 12.8 and 38.2. The statistical significance of the estimated safety effectiveness is 2.3 (greater than 2), which indicates that the treatment is significant at the 95-percent confidence level. Tools Available for Part B Application The SafetyAnalyst set of software tools was developed as a cooperative effort by the FHWA and participating state and local agencies. It provides analytical tools for use in the decision-making process to identify and manage a system-wide program of site-specific improvements to enhance highway safety by cost-effective means. SafetyAnalyst software tools are used by state and local highway agencies for highway safety management. AASHTO manages distribution, technical support, maintenance, and enhancement of SafetyAnalyst as a licensed AASHTOware product. 3.2 HSM in the Alternatives Development and Analysis Phase 3.2.1 Overview After the multiyear programs are developed and system-wide and corridor needs are identified, the next step is implementation of the elements of the program. Projects are selected for development. The scope of work and purpose and need are established, and the project moves forward to the Alternatives Development and Analysis phase. In this phase, multiple alternatives are developed and evaluated to address in the project’s purpose and need. Typically, project decisions are based on evaluation of costs, right-of-way, traffic operations, environmental assessment, and safety evaluation. Agencies can now apply the HSM science-based methods to support explicit consideration of quantitative safety. The HSM allows agencies to quantify a project’s potential for crash reduction, or to apply the predictive method and compare the safety performance of different alternatives associated with a change in traffic volume or traffic control. The following section provides examples of HSM application to different facility types for which SPFs have been developed. 3.2.2 Example Problem 2: Rural, Two-Lane, Two-Way Roads and Rural Multilane Highway Introduction A State Route (SR) has been identified by the state DOT as one of the top 5 percent locations in the 2012 Highway Safety Improvement Program (HSIP) report. This 3.9-mile, rural two-lane road, classified as a principal arterial, runs in an east-west direction (Figure 22). The SR has 12-foot lanes with 1-foot gravel shoulders, and the posted speed limit is 55 mph. Trees and vegetation are present along the edge of the road. There are three stop-controlled three-leg intersections located at mileposts (MP) 100.00, 100.78, and 102.95.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 62 Figure 22: SR Rural Two-Lane, Two-Way Road Results from the crash analysis indicate a high proportion of head-on, sideswipe-opposing, and fixed- object crashes along the roadway, particularly in the curve. A high proportion of angle crashes have also occurred at the intersections. In addition, descriptive crash statistics indicate that there could be an issue with drivers speeding on the SR. There are 5 years of observed crash data (2008 to 2012) and traffic volumes available for the analysis. Evaluations of existing and future conditions and the conversion to a four-lane divided roadway are described in the following subsections. Objectives This example was developed to evaluate the existing safety performance of the SR corridor, perform an alternatives analysis, and determine the safety impacts of the conversion from a two-lane rural road to a four-lane divided roadway for future conditions. Similarly, different improvements for intersections were tested. The first part of the example shows how to calculate the predictive average crash frequency for a rural two-lane stop-controlled intersection and a rural two-lane roadway segment; how to combine intersections and roadway segments as part of a corridor study; and the analysis of the two different alternatives. A third alternative involving a conversion from two-lane to four-lane divided is also considered. However, the CMF for conversion from two- to four-lane highways is only applicable to a short length of highway. Longer roadway segments are out of the scope of the two-lane rural roads methodology and can be addressed with the rural multilane highway procedures. The second part shows how to calculate the predictive average crash frequency for a rural multilane stop-controlled intersection and rural multilane divided roadway under existing and future conditions (2030) and how to combine intersections and roadway segments as part of a corridor study. The different rural two-lane and rural multilane alternatives under future conditions (2030) will be compared. The last part is focused on discussing the results of the analysis. The objective of the example is to show how various HSM analysis tools can be applied to assist traffic analysts, engineers, planners, and decision-makers in making sound investment decisions. In some situations, this amount of analysis would not be necessary to make an informed decision, but the issues presented herein should always be considered to assure the final decision is consistent with safety performance objectives.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 63 After reviewing this example, the user should be able to: • Understand what input data are required and the assumptions that are commonly made regarding default values for the HSM procedures • Calculate the predicted and expected crash frequency of rural two-lane two-way road intersections and roadway segments using the HSM • Calculate the predicted crash frequency of rural multilane intersections and segments using the HSM • Understand how to reasonably interpret the results from an HSM analysis, and how these results can be used to support a particular decision • Understand the limitations of the HSM procedures and when it is appropriate to use other models or computational tools 3.2.3 Part 1 – Rural Two-Lane Two-Way Roads Data Requirements for Part 1 The sample corridor was divided into three roadway segment sections (two tangents and one curve), as shown in Figure 23. Crash data were assigned to intersections and roadway segments. The intersections and roadway segments characteristics are summarized in Tables 37 and 38. AADT information provided in the summary table corresponds to year 2012. Figure 23: Example Problem 1 – Sample Rural Two-Lane, Two-Way Road Intersection Data Table 37 lists intersection input data for the example. TABLE 37 Example Problem 2 – Intersections Input Data Intersection Characteristics Input Data Intersection 1 Intersection 2 Intersection 3 Intersection type 3ST 3ST 3ST Traffic flow major road (vpd) 9,000 9,000 9,000 Traffic flow minor road (vpd) 2,500 3,000 1,200 Intersection skew angle (degrees) 0 0 15 Number of signalized or uncontrolled approaches with a left-turn lane 0 0 0

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 64 TABLE 37 Example Problem 2 – Intersections Input Data Intersection Characteristics Input Data Intersection 1 Intersection 2 Intersection 3 Number of signalized or uncontrolled approaches with a right-turn lane 0 0 0 Intersection lighting Not present Not present Not present Calibration factor (Ci) 1.17 1.17 1.17 Observed crash data (crashes/year) 4 5 2 Note: vpd = vehicles per day Roadway Segment Data Table 38 summarizes the roadway segment input data for the example. TABLE 38 Example Problem 2 – Roadway Segment Input Data Characteristics Input Data Roadway Segment 1 Roadway Segment 2 Roadway Segment 3 Segment length (miles) 1.17 0.78 1.95 Traffic volume (vpd) 9,000 9,000 9,000 Lane width (feet) 12 12 12 Shoulder width (feet) 1 1 1 Shoulder type Paved Paved Paved Length of horizontal curve (miles) 0 0.78 0 Radius of curvature (feet) 0 2650 0 Spiral transition curve Not present Not present Not present Superelevation variance (feet/foot) 0 0.02 0 Grade (%) 2 2 2 Driveway density 1.7 0 4.5 Centerline rumble strips Not present Not present Not present Passing lanes Not present Not present Not present TWLTL Not present Not present Not present Roadside hazard rating 5 5 5 Segment lighting Not present Not present Not present Auto speed enforcement Not present Not present Not present Calibration factor (Cr) 1.30 1.30 1.30 Observed crash data (crashes/year) 11 40 11

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 65 Analysis The rural two-lane, two-way predictive method for intersections and roadway segments under existing conditions (year 2012) was applied in the following subsections. For illustrative purposes, detailed calculations are included only for Intersection 3 and Roadway Segment 2. Intersections The first part of the predictive method is focused on defining the limits, facility type, and study period as well as obtaining and preparing input datasets required to apply the predictive models. Detailed information related to data collection can be found in HSM Section 10.4. The data summary for this example is provided in Table 37. Select and Apply SPFs HSM Chapter 10 intersection SPFs are used to calculate the total predicted average crash frequency per year for crashes that occur within the limits of the intersection. To determine the predictive average crash frequency of the sample intersection, select and apply the appropriate SPF for the facility type and traffic control features. The predicted crash frequency for a three-leg, stop-controlled intersection (Intersection 3) can be calculated using HSM Equation 10-8 (HSM p. 10-18): 𝑁 = 𝑒 . . × . × ( ) 𝑁 = 𝑒 . . × ( , ) . × ( , ) = 2.24 𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟 Apply HSM Part C Crash Modification Factors The SPF predictions are then multiplied by the appropriate CMFs to adjust the estimated crash frequency for base conditions to the site-specific geometry and traffic features. Intersection Skew Angle (CMF1i) CMF1i can be calculated using HSM Equation 10-22 (HSM p. 10-31) for a 3ST intersection. The intersection skew angle is 15 degrees: 𝐶𝑀𝐹 = 𝑒( . × ) 𝐶𝑀𝐹 = 𝑒( . × ) = 1.06 Intersection Left-Turn Lanes (CMF2i) No left-turn lanes are present at the example intersection; HSM Table 10-13 (HSM p. 10-32) provides the CMFs for the presence of left-turn lanes. The selected site does not have left-turn lanes; therefore, a CMF of 1.00 is used. Intersection Right-Turn Lanes (CMF3i) No right-turn lanes are present at the example intersection; HSM Table 10-14 (HSM p. 10-33) provides the CMFs for the presence of right-turn lanes. The selected site does not have right-turn lanes; therefore, a CMF of 1.00 is applied. Intersection Lighting (CMF4i) HSM Equation 10-24 and HSM Table 10-15 (HSM p. 10-33) are used to estimate the CMF for lighting. Lighting is not present at the sample intersection; therefore, a CMF of 1.00 is applied.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 66 The combined CMF is calculated by multiplying all the intersection CMFs: 𝐶𝑀𝐹 =𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 𝐶𝑀𝐹 = 1.06 × 1.00 × 1.00 × 1.00 𝐶𝑀𝐹 = 1.06 Apply Calibration Factor The next step is to multiply the results obtained above by the appropriate calibration factor. For this example, the calibration factor for stop-controlled three-leg intersections has been assumed to be 1.17. Users can use a local calibration factor, if available. Obtain the Predicted Crash Frequency for the Site The predicted average crash frequency for Intersection 3 is calculated using HSM Equation 10-3 (HSM p. 10-4), combining results from previous steps: 𝑁 = 𝑁 × 𝐶 × (𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 ) 𝑁 =2.24 × 1.17 × (1.06 × 1.00 × 1.00 × 1.00) 𝑁 = 2.78 𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟 Multiyear Analysis Since 5 years of data are available, all the previous steps need to be repeated four more times. In this example, a growth rate of 2 percent is assumed. Table 39 summarizes the calculations for Intersection 3. TABLE 39 Example Problem 2 – Intersection 3 Multiyear Analysis Results Intersection 3 Year 2008 2009 2010 2011 2012 AADTmajor 8,315 8,481 8,651 8,824 9,000 AADTminor 1,109 1,131 1,153 1,176 1,200 Crashes/year 1 0 4 3 2 Nspf 3ST 2.03 2.08 2.13 2.19 2.24 CMF1i 3ST 1.06 1.06 1.06 1.06 1.06 CMF2i 3ST 1.00 1.00 1.00 1.00 1.00 CMF3i 3ST 1.00 1.00 1.00 1.00 1.00 CMF4i 3ST 1.00 1.00 1.00 1.00 1.00 CMFcomb 1.06 1.06 1.06 1.06 1.06 Ci 1.17 1.17 1.17 1.17 1.17 Npredicted int 2.52 2.58 2.65 2.71 2.78 Notes: AADTmajor = average annual daily traffic on the major route AADTminor = average annual daily traffic on the minor route CMFcomb = combined CMF Nspf = predicted average crash frequency estimated for base conditions Npredicted int = predicted average crash frequency for the intersection

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 67 The average predicted crash frequency for Intersection 3 is obtained by the arithmetic average of the annual predicted crash frequencies (Npredicted int). For this example, this value is 2.65 crashes per year. Roadway Segments Roadway segment data required to apply the predictive method are summarized in Table 38. Roadway Segment 2 is a curve with a radius of 2,650 feet. No spiral transitions are present. Information on different recommendations related to data collection is presented in HSM Section 10.4. Select and Apply SPFs For the selected site, apply the SPF appropriate for rural two-lane, two-way roads. The SPF can be calculated using HSM Equation 10-6 (HSM p. 10-15): 𝑁 = 𝐴𝐴𝐷𝑇 × 𝐿 × 365 × 10 × 𝑒 . 𝑁 = 9,000 × 0.78 × 365 × 10 × 𝑒 . = 1.88 𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟 Apply HSM Part C Crash Modification Factors Multiply the result obtained above by the appropriate CMFs to adjust the estimated crash frequency for base conditions to the site-specific geometry and traffic features. Lane Width (CMF1r) CMF1r can be calculated using HSM Equation 10-11 (HSM p. 10-24): 𝐶𝑀𝐹 = (𝐶𝑀𝐹 − 1) × 𝑝 + 1 CMFra is estimated using HSM Table 10-8 (HSM p. 10-24). For a 12-foot lane width and AADT greater than 2,000, the CMF for the effect of lane width on related crashes (such as single-vehicle run-off-the- road and multiple-vehicle head-on, opposite-direction sideswipe, and same-direction sideswipe crashes) is 1.00. For this example, the default crash severity distribution (HSM Table 10-4 [HSM p. 10-17]) is assumed, yielding the total percent of related crashes as: 𝑝 = % run off road + % head − on + % sideswipe = 52.1 + 1.6 + 3.7 = 57.4% The lane width CMF is then: 𝐶𝑀𝐹 = (𝐶𝑀𝐹 − 1) × 𝑝 + 1 𝐶𝑀𝐹 = (1 − 1) × 0.574 + 1 = 1.00 Shoulder Width and Type (CMF2r) CMF2r can be calculated using HSM Equation 10-12 shown below. For this example, a 1-foot paved shoulder yields a CMFwra of 1.4 (shoulder width, HSM Table 10-9 [HSM p. 10-25]) and CMFtra of 1.0 (shoulder type, HSM Table 10-10 [HSM p. 10-26]). The percentage of related crashes is the same as that calculated for the lane width CMF: 𝐶𝑀𝐹 = (𝐶𝑀𝐹 × 𝐶𝑀𝐹 − 1) × 𝑝 + 1 𝐶𝑀𝐹 = (1.4 × 1.0 − 1) × 0.574 + 1 = 1.23

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 68 Horizontal Curve (CMF3r) For this example, the length of curve is 0.8 mile with a radius of curvature of 2,650 feet and no spiral transitions. Calculate the CMF calculation using HSM Equation 10-13 (HSM p. 10-27): 𝐶𝑀𝐹 = . × . . ×. × 𝐶𝑀𝐹 = . × . . . ×. × . = 1.03 Superelevation (CMF4r) In this example, the superelevation variance is assumed to be 0.02 foot/foot. Calculate the superelevation using HSM Equation 10-16 (HSM p. 10-28): 𝐶𝑀𝐹 . = 1.06 + 3 × (𝑆𝑉 − 0.02) 𝐶𝑀𝐹 . = 1.06 + 3 × (0.02 − 0.02) = 1.06 Grades (CMF5r) A 2-percent grade section falls under the level grade category in HSM Table 10-11 (HSM p. 10-28), resulting in a CMF of 1.00. Driveway Density (CMF6r) Driveway density of less than five driveways per mile leads to a CMF6R of 1.00. Otherwise, the CMF is calculated using HSM Equation 10-17 (HSM p. 10-29): 𝐶𝑀𝐹 = . ×[ . . × ( )]. ×[ . . × ( )] 𝐶𝑀𝐹 = 1.00 Centerline Rumble Strips (CMF7r) The segment example does not include centerline rumble strips; therefore, a CMF of 1.00 is applied. See HSM p. 10-29 for additional details. Passing Lanes (CMF8r) Passing lanes are not available in the example; therefore, a CMF of 1.00 is appropriate. See HSM p. 10-29 for additional details. Two-way, Left-turn Lane (CMF9r) Two-way, left-turn lanes are not present; therefore, a CMF of 1.00 is applied for this example. See HSM p. 10-29 for additional details. Roadside Design (CMF10r) From the data input of this example, a roadside hazard rating of 5 applies to the segment. Using HSM Equation 10-20 (HSM p. 10-30), the CMF is: 𝐶𝑀𝐹 = . . ×. 𝐶𝑀𝐹 = . . ×. = 1.14 Lighting (CMF11r) Lighting is not present along the example segment; therefore, a CMF of 1.00 is applied. See HSM p. 10-30 for additional details.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 69 Automated Speed Enforcement (CMF12r) The example roadway segment does not have automated speed enforcement available; therefore, a CMF of 1.00 is applied. See HSM p. 10-30 for additional details. The combined CMF is then calculated by multiplying all segment CMFs: 𝐶𝑀𝐹 =𝐶𝑀𝐹 × 𝐶𝑀𝐹 × … × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 𝐶𝑀𝐹 =1.00 × 1.23 × 1.03 × 1.06 × 1.00 × 1.00 × 1.00 × 1.00 × 1.00 ×1.14 × 1.00 × 1.00 𝐶𝑀𝐹 = 1.527 Apply Calibration Factor Multiply the predicted average crash frequency and CMFs results obtained in previous steps by the appropriate calibration factor. For this example, the calibration factor has been assumed to be 1.30. Obtain the Predicted Crash Frequency for the Site The predicted average crash frequency is calculated using HSM Equation 10-2 (HSM p. 10-3), combining results from previous steps: 𝑁 = 𝑁 × 𝐶 × (𝐶𝑀𝐹 × 𝐶𝑀𝐹 × … × 𝐶𝑀𝐹 ) 𝑁 =1.88 × 1.30 × (1.00 × 1.23 × … × 1.00) = 3.72 𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟 Multiyear Analysis Since 5 years of data are available, all the preceding steps need to be repeated four more times. In this example, a growth rate of 2 percent is assumed. Table 40 summarizes the calculations for the study period. TABLE 40 Example Problem 2 – Roadway Segment 2 Multiyear Analysis Results Roadway Segment 2 Year 2008 2009 2010 2011 2012 AADT 8,315 8,481 8,651 8,824 9,000 Crashes/year 29 45 48 38 40 Nspf 1.73 1.77 1.80 1.84 1.88 CMF1r 1.00 1.00 1.00 1.00 1.00 CMF2r 1.23 1.23 1.23 1.23 1.23 CMF3r 1.03 1.03 1.03 1.03 1.03 CMF4r 1.06 1.06 1.06 1.06 1.06 CMF5r 1.00 1.00 1.00 1.00 1.00 CMF6r 1.00 1.00 1.00 1.00 1.00 CMF7r 1.00 1.00 1.00 1.00 1.00 CMF8r 1.00 1.00 1.00 1.00 1.00 CMF9r 1.00 1.00 1.00 1.00 1.00 CMF10r 1.14 1.14 1.14 1.14 1.14 CMF11r 1.00 1.00 1.00 1.00 1.00 CMF12r 1.00 1.00 1.00 1.00 1.00

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 70 TABLE 40 Example Problem 2 – Roadway Segment 2 Multiyear Analysis Results Roadway Segment 2 Year 2008 2009 2010 2011 2012 CMFcomb 1.527 1.527 1.527 1.527 1.527 Ci 1.30 1.30 1.30 1.30 1.30 Npredicted seg 3.44 3.51 3.58 3.65 3.72 Notes: CMFcomb = combined CMF Nspf = predicted average crash frequency estimated for base conditions Npredicted seg = predicted average crash frequency for the roadway segment The average predicted crash frequency for Roadway Segment 2 is obtained through the arithmetic average of the annual predicted crash frequencies (Npredicted seg). The average for this example is 3.58 crashes per year. Corridor Analysis (Intersections and Roadway Segments) Analysis results for intersections and roadway segments can be combined into a corridor analysis. This approach combines the predicted crash frequency of the multiple locations to calculate the corridor predicted average crash frequency. This is done by adding the predicted average crash frequency of all roadway segments and intersections, as shown in Table 41. TABLE 41 Example Problem 2 – Corridor Predicted Average Crash Frequency Site Type Predicted Average Crash Frequency (crashes/year) Npredicted (Total) Npredicted (Fatal-and-Injury) Npredicted (PDO) Roadway Segments Roadway Segment 1 4.94 1.59 3.36 Roadway Segment 2 3.58 1.15 2.43 Roadway Segment 3 8.24 2.64 5.59 Intersections Intersection 1 3.57 1.48 2.09 Intersection 2 3.91 1.62 2.29 Intersection 3 2.65 1.10 1.55 Combined (sum of column) 26.89 9.58 17.31 HSM Table 10-3 (HSM p. 10-17) provides default proportions for crash severity level on rural two-lane two-way roadway segments. This is used to separate the crash frequencies into fatal-and-injury and PDO crashes. Fatal-and-injury and PDO default proportions are 32.1 percent and 67.9 percent, respectively. Results from application of severity proportions are included in Table 41. These proportions can be updated using local crash data (refer to HSM Part C Appendix A for details).

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 71 Empirical Bayes Adjustment Method The next step in the process is to update predictions based on the observed/reported crashes. A total of 62 roadway segment crashes and 11 intersection crashes occur each year. Using the predictive models, the total predictive average crash frequencies for roadway segments and intersections are 16.76 crashes and 10.13 crashes per year, respectively. Empirical Bayes Adjustment Method The predicted average crash frequency is then adjusted using the EB method by applying the following steps. In this example, crashes can be assigned accurately between intersections and roadway segments; therefore, the site EB method is applicable. Refer to HSM Sections A.2.4 and A.2.5 (HSM p. A-19 and A-20) for additional details on the different EB methods. The expected number of crashes for either roadway segments or intersections is calculated using HSM Equation A-4 (HSM p. A-19): 𝑁 = 𝑤 × 𝑁 + (1 − 𝑤) × 𝑁 To complete this calculation, weighting adjustment factors are needed for the sample roadway segment and intersection. Calculate using the previous crash predictions, with HSM Equation A-5 (HSM p. A-19): 𝑤 = ×∑ For this calculation, the overdispersion parameter (k) from each of the applied SPFs is needed. The overdispersion parameter associated with Roadway Segment 2 is 0.303. The closer the overdispersion parameter is to zero, the more statistically reliable the SPF. On a per-mile basis, the overdispersion parameter is found by using HSM Equation 10-7 (HSM p. 10-16): 𝑘 = . 𝑘 = .. = 0.303 The overdispersion parameter associated with the three-leg, stop-controlled intersection is 0.54: 𝑘 = 0.540 Using these overdispersion parameters, the weighting adjustment factors are found to be 0.156 for Roadway Segment 2 and 0.123 for Intersection 3: 𝑤 = . ×( . . . . . ) 𝑤 = 0.156 𝑤 = . ×( . . . . . ) 𝑤 = 0.123

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 72 For this example, there were an average of 40 observed/reported crashes per year on Roadway Segment 2 and an average of 2 observed/reported crashes per year at Intersection 3. The expected number of crashes for roadway segments and intersections is then calculated as follows: 𝑁 = 𝑤 × 𝑁 + (1 − 𝑤) × 𝑁 𝑁 = 0.156 × 3.58 + (1 − 0.156) × 40 𝑁 = 34.3 𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟 𝑁 = 0.123 × 2.65 + (1 − 0.123) × 2 𝑁 = 2.1 𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟 Similar analyses are performed for all roadway segments and intersections. Results of the analysis can be found in the sample spreadsheets provided with the Highway Safety Manual User Guide. The total expected average crash frequency for the corridor is the sum of the expected crashes along the roadway segments and intersections. Calculate this sum using HSM Equation 10-4 (HSM p. 10-10): 𝑁 = ∑ 𝑁 + ∑ 𝑁 𝑁 = (9.99 + 34.32 + 10.54) + (3.96 + 4.91 + 2.08) = 65.79 Table 42 presents a summary of the predictive method calculations. Columns 2 through 4 contain the predicted average crash frequency for total, fatal-and-injury, and PDO crashes. The fifth column contains the observed/reported number of crashes per year. Columns 6 and 7 contain the overdispersion parameter and weighted adjustment to be used to obtain the expected average crash frequency (last column). HSM Table 10-3 (HSM p. 10-17) provides default proportions for crash severity level on rural two-lane two-way roadway segments. This can also be used to separate the expected average crash frequencies into fatal-and-injury and PDO crashes. The fatal-and-injury and PDO default proportions are 32.1 percent and 67.9 percent, respectively. TABLE 42 Example Problem 2 – Predicted and Expected Crash Frequency Calculations Summary (2008 to 2012) Site Type Predicted Average Crash Frequency (crashes/year) Observed/ Reported Crashes (Nobserved) (crashes/ year) Overdisper sion Parameter (k) Weighted Adjustment (w) (Equation A-5 from HSM Part C, Appendix A) Expected Average Crash Frequency (Nexpected) (Equation A-4 from HSM Part C, Appendix A) Npredicted (Total) Npredicted (Fatal-&- Injury) Npredicted (PDO) Roadway Segments Roadway Segment 1 4.94 1.59 3.36 11 0.202 0.167 9.99 Year 1 4.75 1.52 3.22 10 0.202 Year 2 4.84 1.55 3.29 12 0.202 Year 3 4.94 1.59 3.35 14 0.202 Year 4 5.04 1.62 3.42 8 0.202 Year 5 5.14 1.65 3.49 11 0.202

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 73 TABLE 42 Example Problem 2 – Predicted and Expected Crash Frequency Calculations Summary (2008 to 2012) Site Type Predicted Average Crash Frequency (crashes/year) Observed/ Reported Crashes (Nobserved) (crashes/ year) Overdisper sion Parameter (k) Weighted Adjustment (w) (Equation A-5 from HSM Part C, Appendix A) Expected Average Crash Frequency (Nexpected) (Equation A-4 from HSM Part C, Appendix A) Npredicted (Total) Npredicted (Fatal-&- Injury) Npredicted (PDO) Roadway Segment 2 3.58 1.15 2.43 40 0.303 0.156 34.32 Year 1 3.44 1.10 2.34 29 0.303 Year 2 3.51 1.13 2.38 45 0.303 Year 3 3.58 1.15 2.43 48 0.303 Year 4 3.65 1.17 2.48 38 0.303 Year 5 3.72 1.20 2.53 40 0.303 Roadway Segment 3 8.24 2.64 5.59 11 0.121 0.167 10.54 Year 1 7.91 2.54 5.37 11 0.121 Year 2 8.07 2.59 5.48 15 0.121 Year 3 8.23 2.64 5.59 12 0.121 Year 4 8.40 2.70 5.70 10 0.121 Year 5 8.57 2.75 5.82 7 0.121 Intersections Intersection 1 3.57 1.48 2.09 4 0.540 0.094 3.96 Year 1 3.40 1.41 1.99 2 0.540 Year 2 3.48 1.45 2.04 6 0.540 Year 3 3.57 1.48 2.09 5 0.540 Year 4 3.66 1.52 2.14 4 0.540 Year 5 3.76 1.56 2.20 3 0.540 Intersection 2 3.91 1.62 2.29 5 0.540 0.087 4.91 Year 1 3.71 1.54 2.17 8 0.540 Year 2 3.81 1.58 2.23 3 0.540 Year 3 3.91 1.62 2.28 4 0.540 Year 4 4.01 1.66 2.34 6 0.540 Year 5 4.11 1.71 2.40 4 0.540 Intersection 3 2.65 1.10 1.55 2 0.540 0.123 2.08 Year 1 2.52 1.04 1.47 1 0.540 Year 2 2.58 1.07 1.51 0 0.540 Year 3 2.65 1.10 1.55 4 0.540 Year 4 2.71 1.13 1.59 3 0.540 Year 5 2.78 1.16 1.63 2 0.540 Total 26.89 9.58 17.31 73 − − 65.79

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 74 Alternatives Analysis The previous section demonstrated the application of the predictive method for rural two-lane, two-way roadway segments and intersections under existing conditions. The predictive method can also be applied to alternatives analysis. This process is more detailed and specific about the impacts of implementation of project improvements. The agency develops potential alternatives and compares performance across the alternatives. The two- lane, two-way rural roads predictive method can be applied to compare alternatives, as described in the following paragraphs. Calculations and formulas are the same used in the previous sections, and the results are summarized into tables. Tables 43 and 44 contain the input data for the current conditions, along with two alternatives to improve the corridor existing safety performance. For simplicity, only geometric elements that are being improved or upgraded are listed in the tables. Alternative 1, as compared to the No Build scenario, consists of: • Shoulder widening from 1- to 6-foot shoulders • Adding an uncontrolled left-turn lane to each intersection For demonstration purposes, it is assumed the AADT remains the same and the road does not attract any additional traffic. Alternative 2, in addition to those improvements listed in Alternative 1, consists of: • Improve roadside hazard rating to Level 3 by removing vegetation along the road • Install lighting along the roadway segment and at intersections • Implement auto speed enforcement Similarly, it is assumed the AADT remains the same, and the road does not attract any additional traffic. TABLE 43 Example Problem 2 – Roadway Segment Alternatives Input Data Roadway Segment 1 Characteristics Input Data by Alternative No Build Alternative 1 Alternative 2 Shoulder width (feet) 1 6 6 Roadside hazard rating 5 5 3 Segment lighting Not present Not present Present Auto speed enforcement Not present Not present Present Roadway Segment 2 Characteristics Input Data by Alternative No Build Alternative 1 Alternative 2 Shoulder width (feet) 1 6 6 Roadside hazard rating 5 5 3 Segment lighting Not present Not present Present Auto speed enforcement Not present Not present Present

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 75 TABLE 43 Example Problem 2 – Roadway Segment Alternatives Input Data Roadway Segment 3 Characteristics Input Data by Alternative No Build Alternative 1 Alternative 2 Shoulder width (feet) 1 6 6 Roadside hazard rating 5 5 3 Segment lighting Not present Not present Present Auto speed enforcement Not present Not present Present TABLE 44 Example Problem 2 – Intersection Alternatives Input Data Intersection 1 Characteristics Input Data by Alternative No Build Alternative 1 Alternative 2 Number of signalized or uncontrolled approaches with a left-turn lane 0 1 1 Intersection lighting Not present Not present Present Intersection 2 Characteristics Input Data by Alternative No Build Alternative 1 Alternative 2 Number of signalized or uncontrolled approaches with a left-turn lane 0 1 1 Intersection lighting Not present Not present Present Intersection 3 Characteristics Input Data by Alternative No Build Alternative 1 Alternative 2 Number of signalized or uncontrolled approaches with a left-turn lane 0 1 1 Intersection lighting Not present Not present Present The effect of the multiple treatments (such as widening shoulders, lighting the roadway segments and intersections, adding left-turn lanes) is reflected in the decrease of predicted average number of crashes. All these different adjustments are considered through the CMFs, which are used to adjust the SPF estimate of predicted average crash frequency for the effect of these different individual geometric design and traffic control features. The CMF for the SPF base condition of each geometric design or traffic control feature has a value of 1.00. Calculations for the No Build scenario are the same as the first part of the example. Table 45 summarizes the results for the No Build scenario and Alternatives 1 and 2. Total predicted, observed, and expected average crash frequencies are in bolded text. As shown in the table, the expected number of crashes under existing conditions is higher (65.79 crashes per year) than for Alternatives 1 and 2. As anticipated, the expected number of crashes for Alternative 2 is lower than Alternative 1. However, an economic evaluation should be conducted to determine which alternative is more cost-effective. Detailed

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 76 calculations are provided in the sample spreadsheets provided with the Highway Safety Manual User Guide. TABLE 45 Example Problem 2 – Alternatives Analysis Results Summary Alternative Site Type Npredicted Nobserved Overdispersion Parameter (k) Weighted Adjustment (w) Nexpected No Build Roadway Segment 1 4.94 11 0.202 0.167 9.99 Roadway Segment 2 3.58 40 0.303 0.156 34.32 Roadway Segment 3 8.24 11 0.121 0.167 10.54 Intersection 1 3.57 4 0.540 0.094 3.96 Intersection 2 3.91 5 0.540 0.087 4.91 Intersection 3 2.65 2 0.540 0.123 2.08 Total 26.89 73 − − 65.79 Alternative 1 Roadway Segment 1 4.02 11 0.202 0.198 9.62 Roadway Segment 2 2.91 40 0.303 0.185 33.14 Roadway Segment 3 6.70 11 0.121 0.198 10.15 Intersection 1 2.00 4 0.540 0.156 3.69 Intersection 2 2.19 5 0.540 0.145 4.59 Intersection 3 1.48 2 0.540 0.200 1.90 Total 19.30 73 − − 63.08 Alternative 2 Roadway Segment 1 3.01 11 0.202 0.248 9.02 Roadway Segment 2 2.18 40 0.303 0.232 31.21 Roadway Segment 3 5.02 11 0.121 0.248 9.52 Intersection 1 1.80 4 0.540 0.170 3.63 Intersection 2 1.97 5 0.540 0.158 4.52 Intersection 3 1.34 2 0.540 0.217 1.86 Total 15.33 73 − − 59.76 Discussion of results of this section is provided after the rural multilane highway alternative analysis. 3.2.4 Part 2 – Rural Multilane Highways The agency also decided to analyze the safety performance of converting the rural two-lane, two-way roads into a four-lane divided roadway. The analysis will be performed for existing and future (2030) conditions. To understand the methodology, the example first shows how to calculate the predictive average crash frequency for a rural multilane stop-controlled intersection and rural multilane divided roadway under existing conditions, and how to combine intersections and roadway segments as part of a corridor study. Next, the predicted average crash frequency for future conditions will be compared with the 2030 rural two-lane road predicted crash frequency. Finally, a discussion of the results will be provided.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 77 Data Requirements for Part 2 Figure 24 shows the different facility types included in this example. Since the rural multilane predictive method does not include a CMF for curves, there is no need to break the corridor into multiple segments. Crash data are available for years 2008 to 2012. The roadway segment and intersections characteristics are summarized in Tables 46 and 47. Figure 24: Example Problem 1 – Sample Rural Multilane Highway Intersection Data Table 46 summarizes the input data required to apply the predictive method for all intersections. TABLE 46 Example Problem 2 – Intersections Input Data Characteristics Input Data Intersection 1 Intersection 2 Intersection 3 Intersection type 3ST 3ST 3ST Traffic flow major road (vpd) 9,000 9,000 9,000 Traffic flow minor road (vpd) 2,500 3,000 1,200 Intersection skew angle 0 0 15 Number of signalized or uncontrolled approaches with a left-turn lane 1 0 0 Number of signalized or uncontrolled approaches with a right-turn lane 1 0 0 Intersection lighting Not present Not Present Not Present Calibration factor (Ci) 1.20 1.20 1.20 Observed/reported fatal-and-injury crashes (crashes/year) 2 2 1 Observed/reported PDO crashes (crashes/year) 2 3 1 Roadway Segment Data Table 47 summarizes the multilane rural divided roadway segment input data.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 78 TABLE 47 Example Problem 2 – Roadway Segment 1 Input Data Characteristics Input Data Roadway Segment 1 Roadway type Divided Segment length (miles) 3.90 Traffic volume (vpd) 9,000 Lane width (feet) 12 Shoulder width (feet) 8 Shoulder type Paved Median width (feet) 30 Sideslopes --- Segment lighting Not present Auto speed enforcement Not present Calibration factor (Cr) 1.08 Observed/reported fatal-and-injury crashes (crashes/year) 19 Observed/reported PDO crashes (crashes/year) 43 Analysis The rural multilane highways predictive method for intersections and roadway segments under existing conditions (year 2012) is applied in the following sections. For illustrative purposes, detailed calculations are included only for Intersection 1 and Segment 1. Intersections The first part of the predictive method is focused on defining the limits, facility type, and study period as well as obtaining and preparing input datasets required to apply the predictive models. Detailed information related to data collection can be found in HSM Section 11.4. The data summary for this example is provided in Table 46. Select and Apply SPFs The intersection SPFs in HSM Chapter 11 estimate the total predicted average crash frequency for intersection-related crashes within the intersection limits and on the intersection legs. SPFs are provided for different intersection types and severity levels. To determine the predictive average crash frequency of the sample intersection, select and apply the appropriate SPF for the facility type and traffic control features. The total and fatal-and- injury predicted crash frequencies for a stop-controlled intersection (Intersection 1) can be calculated using HSM Equation 11-11 with coefficients from Table 11-7 (HSM p. 11-20 and 11-21): 𝑁 = 𝑒 × × ( ) 𝑁 = 𝑒 . . × ( , ) . × ( , ) 𝑁 = 1.327 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 79 𝑁 ( ) = 𝑒 . . × ( , ) . × ( , ) 𝑁 ( ) = 0.633 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 A separate set of fatal-and-injury SPFs are also available for agencies that do not wish to consider severity level C (possible injury) on the KABCO scale (see Appendix B of this guide). Apply HSM Part C Crash Modification Factors Calculate the appropriate CMFs to adjust the predicted crash frequency for base conditions to the site-specific geometry and traffic features. For this example, all the intersection CMFs are equal to 1.00. Intersection Skew Angle (CMF1i) CMF1i can be calculated using HSM Equations 11-18 (total) and 11-19 (fatal-and-injury) for a 3ST intersection (HSM p. 11-33). Intersection 1 is not skewed; therefore, the CMF is 1.00: 𝐶𝑀𝐹 = . ×. . × + 1 = 1.00 𝐶𝑀𝐹 ( ) = 0.017 × 𝑠𝑘𝑒𝑤0.52 + 0.017 × 𝑠𝑘𝑒𝑤 + 1 = 1.00 Intersection Left-Turn Lanes (CMF2i) One left-turn lanes is present at the example intersection; therefore, CMF of 0.56 for total crashes, and 0.45 for fatal-and-injury crashes are applied. HSM Table 11-22 (HSM p. 11-34) presents CMFs for the presence of left-turn lanes for total and fatal-and-injury crashes. Intersection Right-Turn Lanes (CMF3i) Similarly, a right-turn lane is present at the example intersection; therefore, a CMF of 0.86 for total crashes, and 0.77 for fatal-and-injury crashes are applied. HSM Table 11-23 (HSM p. 11-35) presents CMFs for the presence of right-turn lanes for total and fatal-and-injury crashes. Intersection Lighting (CMF4i) Lighting is unavailable in this example; therefore, a CMF of 1.00 is appropriate. HSM Equation 11-22 (HSM p. 11-35) is used to estimate the CMF for lighting. The combined CMF is then calculated by multiplying all the intersection CMFs: 𝐶𝑀𝐹 =𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 Apply Calibration Factor The next step is to multiply the results obtained above by the appropriate calibration factor. For this example, the calibration factor has been assumed to be 1.20. Obtain the Predicted Crash Frequency for the Site The predicted average crash frequency for Intersection 1 is calculated using HSM Equation 11-4 (HSM p. 11-4), combining results from previous steps:

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 80 𝑁 = 𝑁 × 𝐶 × (𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 ) 𝑁 = 1.327 × 1.20 × (1.0 × 0.56 × 0.86 × 1.0) 𝑁 = 0.767 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 0.633 × 1.20 × (1.0 × 0.45 × 0.77 × 1.0) 𝑁 ( ) = 0.263 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 0.767 − 0.263 = 0.503 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Multiyear Analysis Since 5 years of data are available, all the previous steps need to be repeated four more times. In this example, a growth rate of 2 percent is assumed. Table 48 summarizes the calculations for the study period. Only calculations for total crashes are shown in the table. Calculations for fatal-and-injury crashes and other detailed calculations are provided in the sample Highway Safety Manual User Guide spreadsheets. TABLE 48 Example Problem 2 – Intersection 1 Multiyear Analysis Results Intersection 1 Year 2008 2009 2010 2011 2012 AADTmajor 8,315 8,481 8,651 8,824 9,000 AADTminor 2,310 2,356 2,403 2,451 2,500 Crashes/year 2 6 5 4 3 Nspf 3ST 1.184 1.218 1.253 1.290 1.327 CMF1i 3ST 1.00 1.00 1.00 1.00 1.00 CMF2i 3ST 0.56 0.56 0.56 0.56 0.56 CMF3i 3ST 0.86 0.86 0.86 0.86 0.86 CMF4i 3ST 1.00 1.00 1.00 1.00 1.00 CMFcomb 0.48 0.48 0.48 0.48 0.48 Ci 1.20 1.20 1.20 1.20 1.20 Npredicted int 0.68 0.70 0.72 0.75 0.77 The average predicted crash frequency for Intersection 1 is obtained by the arithmetic average of the annual predicted crash frequencies (Npredicted int). For this example, this value is 0.725 crash per year.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 81 Roadway Segments Roadway segment data required to apply the predictive method are summarized in Table 47. For this example, the analysis corridor consists of only one four-lane divided segment. Information on different recommendations related to data collection is presented in HSM Section 11.4. Select and Apply SPFs Separate SPFs are available for undivided (HSM Equation 11-7 and Table 11-3 [HSM p. 11-15]) and divided (HSM Equation 11-9 and Table 11-5 [HSM p. 11-18]) rural multilane highways for total and fatal-and-injury severity levels: 𝑁 = 𝑒 × ( ) ( ) 𝑁 = 𝑒 . . × ( , ) ( . ) 𝑁 = 6.60 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 𝑒 . . × ( , ) ( . ) 𝑁 ( ) = 3.48 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 As with the intersection SPFs, a separate set of fatal-and-injury SPFs are available for agencies that do not wish to consider Severity Level C (possible injury) on the KABCO scale (see Appendix B of this guide). Apply HSM Part C Crash Modification Factors Calculate the applicable CMFs to adjust the estimated crash frequency for base conditions to the site-specific geometry and traffic features. Lane Width (CMF1r) For a 12-foot lane width, the effect of lane width on related crashes (such as single- vehicle run-off-the-road and multiple-vehicle head-on, opposite-direction sideswipe, and same-direction sideswipe crashes) is 1.00, as shown in HSM Table 11-16 (HSM p. 11-30). Shoulder Width and Type (CMF2r) CMF2r can be calculated using HSM Table 11-17 (HSM p. 11-31). The SPF base condition for the right shoulder is 8 feet. NOTE: The CMFs provided in HSM Table 11-17 only apply to paved shoulders. For an 8-foot right shoulder width, the applicable CMF is 1.00. Median Width (CMF3rd) The base condition assigned to the median width CMF is 30 feet, assuming no traffic barrier. These base conditions match the characteristics of the example roadway segment, resulting in a CMF of 1.00. HSM Table 11-18 (HSM p. 11-31) contains CMFs for different median widths on divided roadway segments. Lighting (CMF4r) Lighting is not present along the example roadway segment; therefore, a CMF of 1.00 is applied. See HSM p. 11-31 for additional details.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 82 Automated Speed Enforcement (CMF5r) The example segment does not have automated speed enforcement available; therefore, a CMF of 1.00 is applied. See HSM Page 11-32 for additional details. The combined CMF is then calculated by multiplying all the intersection CMFs: 𝐶𝑀𝐹 =𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 Apply Calibration Factor Multiply the results obtained in the two previous steps by the appropriate calibration factor. For this example, the calibration factor is assumed to be 1.00. Obtain the Predicted Crash Frequency for the Site Lastly, the predicted average crash frequency is calculated using HSM Equation 11-3 (HSM p. 11-4), combining results from the preceding steps: 𝑁 = 𝑁 × 𝐶 × (𝐶𝑀𝐹 × 𝐶𝑀𝐹 × … × 𝐶𝑀𝐹 ) 𝑁 = 6.60 × 1.08 × (1.0 × 1.0 × 1.0 × 1.0 × 1.0) 𝑁 = 7.13 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) =3.48 × 1.08 × (1.0 × 1.0 × 1.0 × 1.0 × 1.0) 𝑁 ( ) = 3.76 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 7.13 − 3.76 = 3.37 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Multiyear Analysis Since 5 years of data are available, all the steps above must be repeated four more times. In this example, a growth rate of 2 percent is assumed. Table 49 summarizes Roadway Segment 1 multiyear calculations. The average predicted crash frequency for Roadway Segment 1 is obtained through the arithmetic average of the annual predicted crash frequencies (Npredicted seg). The average for this example is 6.84 crashes per year. TABLE 49 Example Problem 2 – Roadway Segment 1 Multiyear Analysis Results Roadway Segment 1 Year 2008 2009 2010 2011 2012 AADT 8,315 8,481 8,651 8,824 9,000 Crashes/year 50 72 74 56 58 Nspf 6.07 6.20 6.33 6.47 6.60 CMF1ru 1.00 1.00 1.00 1.00 1.00

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 83 TABLE 49 Example Problem 2 – Roadway Segment 1 Multiyear Analysis Results Roadway Segment 1 Year 2008 2009 2010 2011 2012 CMF2ru 1.00 1.00 1.00 1.00 1.00 CMF3ru 1.00 1.00 1.00 1.00 1.00 CMF4ru 1.00 1.00 1.00 1.00 1.00 CMF5ru 1.00 1.00 1.00 1.00 1.00 CMFcomb 1.00 1.00 1.00 1.00 1.00 Cr 1.08 1.08 1.08 1.08 1.08 Npredicted seg 6.56 6.70 6.84 6.98 7.13 Corridor Analysis (Intersections and Roadway Segments) Analysis results for intersections and roadway segments can be combined into a corridor analysis. This approach combines the predicted crash frequency of the multiple locations to come up with corridor predicted average crash frequency. Table 50 summarizes the predicted crash frequency of all roadway segments and intersections and provides the corridor results. TABLE 50 Example Problem 2 – Corridor Predicted Average Crash Frequency Site Type Predicted Average Crash Frequency (crashes/year) Npredicted (Total) Npredicted (Fatal-&-Injury) N predicted (PDO) Roadway Segments Divided Roadway Segment 1 6.84 3.62 3.22 Intersections Intersection 1 0.72 0.25 0.48 Intersection 2 1.57 0.76 0.81 Intersection 3 1.51 0.78 0.73 Combined (sum of column) 10.65 5.41 5.24 HSM Tables 11-4 (HSM p. 11-17), 11-6 (HSM p. 11-20), and 11-9 (HSM p. 11-24) provide default proportions of crashes by collision type and crash severity level for rural multilane undivided, divided roadways, and intersections. These proportions can be applied to the predicted crash frequencies for selected collision types. These proportions can be updated using local crash data (refer to HSM Part C, Appendix A for details). Empirical Bayes Adjustment Method The next step in the process is to update predictions based on the observed/reported crashes. First, it must be determined whether the EB method is applicable to this example. HSM Section A.2.1 (HSM p. A-16) provides guidance on how to determine the applicability of

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 84 the EB method. Since this project upgrade involves the development of a new alignment for a substantial portion of the project length, the EB method is not applicable. The main reason is the historical observed/reported crash data may not be a good indicator of the crash experience that is likely to occur in the future after the implementation of a major change. If the conversion from two-lane to four-lane divided was done in a short section of the corridor (less than 2 miles) to allow more passing opportunities, then the expected number of crashes could have been calculated using the observed/reported number of crashes from the rural two-lane two-way facility. Details on the application of the predictive method can be found in HSM Appendix A, Section A.2 (HSM p. A-15). Alternatives Analysis Part 1 of this example demonstrated the application of the predictive method for the rural two-lane, two-way roadway segment and intersections under existing conditions, and for alternative analysis. The second part of the problem was focused on the application of the predictive method for the rural multilane rural roadway segment and intersections under existing conditions. The next step in the process is to compare the different alternatives under consideration. Since the construction of the four- lane divided roadway is a future project, predicted average crash frequencies for year 2030 (opening year) will be calculated for the following scenarios: rural two-lane, two-way road No Build, Alternative 1, Alternative 2, and proposed rural multilane corridor. According to the local metropolitan planning organization, the projected 2030 AADT for the SR corridor is 13,500 vehicles per hour. This translates to a growth factor of about 2.28 percent per year. Minor road AADTs are obtained by applying the growth factor to the existing AADTs. Table 51 summarizes the traffic volumes for the different facility types. TABLE 51 Example Problem 2 – Year 2030 AADT for Rural Two-Lane and Rural Multilane Facilities Facility Roadway Segment 1 Roadway Segment 2 Roadway Segment 3 Rural two-lane roads 13,500 13,500 13,500 Rural multilane road 13,500 Intersection 1 Intersection 2 Intersection 3 Rural two-lane major road 13,500 13,500 13,500 Rural two-lane minor road 3,750 4,500 1,800 Rural multilane major road 13,500 13,500 13,500 Rural multilane minor road 3,750 4,500 1,800 Using the same input data (rural two-lane Tables 37, 38, 43, and 44, and rural multilane Tables 46 and 47), proposed safety countermeasures, and 2030 AADTs, the predictive average crash frequency is calculated for rural two-lane No Build, Alternative 1, and Alternative 2, and the four-lane divided road (Alternative 3). Table 52 includes the predicted crash frequencies for total, fatal-and-injury, and PDO crashes.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 85 TABLE 52 Example Problem 2 – Future Conditions Alternative Analysis Summary (2030) Alternative Site Type Npredicted Total Npredicted FI Npredicted PDO Rural Two-lane: No Build Segment 1 7.71 2.47 5.23 Segment 2 5.58 1.79 3.79 Segment 3 12.85 4.12 8.72 Intersection 1 6.31 2.62 3.69 Intersection 2 6.90 2.87 4.04 Intersection 3 4.68 1.94 2.74 Total 44.04 15.82 28.22 Rural Two-Lane: Alternative 1 Segment 1 6.27 2.01 4.26 Segment 2 4.54 1.46 3.08 Segment 3 10.45 3.35 7.10 Intersection 1 3.54 1.47 2.07 Intersection 2 3.87 1.60 2.26 Intersection 3 2.62 1.09 1.53 Total 31.28 10.98 20.30 Rural Two Lane: Alternative 2 Segment 1 4.70 1.51 3.19 Segment 2 3.41 1.09 2.31 Segment 3 7.84 2.52 5.32 Intersection 1 3.19 1.32 1.86 Intersection 2 3.48 1.45 2.04 Intersection 3 2.36 0.98 1.38 Total 24.98 8.87 16.11 Rural Multilane: Alternative 3 Segment 1 10.91 5.54 5.37 Intersection 1 1.37 0.46 0.91 Intersection 2 2.98 1.40 1.58 Intersection 3 2.87 1.45 1.43 Total 18.14 8.84 9.29 Results and Discussion From Part 1 of the problem, it was concluded that the proposed countermeasures for rural two-lane Alternative 2 produced the lower predicted and expected crash frequency. However, conducting an economic evaluation was recommended to make a cost-effective decision.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 86 The state DOT also considered modifying the rural two-lane corridor to a four-lane divided facility. The analysis was conducted for future conditions (design year 2030); however, since the EB method was not applicable, the comparison with the other alternatives was conducted using the predicted average crash frequency. (NOTE: If the conversion from two-lane to four-lane divided was done in a short section of the corridor [less than 2 miles] to allow more passing opportunities, then the expected number of crashes could have been calculated.) The analysis results indicate that the four-lane divided alternative reduces the total crash frequency by 59 percent in comparison to the 2030 No Build scenario. Alternatives 1 and 2 would reduce the total crash frequency by 329 percent and 43 percent, respectively. Predicted crash frequencies for fatal-and- injury and PDO crashes are also provided. (NOTE: Fatal-and-injury and PDO crash frequencies for rural two-lane are calculated based on proportions, and for rural multilane are calculated using fatal-and- injury SPFs.) Alternative 2 and Alternative 3 have the lowest predicted fatal-and-injury crash frequencies. Based on these results, the four-lane conversion would potentially provide the greatest reduction in crash frequency along the corridor, but an economic evaluation is required to better understand which alternative is the most cost-effective. Refer to HSM Chapter 7, Economic Appraisal, for methods to compare the benefits of potential crash countermeasures to crash costs. Tools Available for HSM Part C Application The Interactive Highway Safety Design Model (IHSDM) and spreadsheet tools are available to assist in the HSM Part C predictive method calculations. The HSM Part C spreadsheet tools can be downloaded from the HSM website under the Quick Links section (http://www.highwaysafetymanual.org). In addition to analyzing safety performance, IHSDM has a design consistency module that may be helpful to users for planning or design. 3.2.5 Example Problem 3: Urban and Suburban Arterials Introduction The example facility is a 0.3-mile urban arterial with commercial development. The corridor has two 12-foot lanes in each direction, and a TWLTL that provides access to the driveways along the road. Most of the properties adjacent to the corridor have multiple direct access points. Parallel on-street parking is available along the corridor. The posted speed limit is 35 mph. The corridor is bounded by a four-leg signalized intersection on the north and a three-leg stop-controlled intersection on the south (Figure 25). A high proportion of rear-end, angle, and sideswipe crashes have occurred at the facility in the past few years. In addition, a couple fatalities and a serious injury crash were reported. The city has decided to evaluate alternatives to mitigate the safety issues, to improve traffic operations, and make it more pedestrian friendly. Five years of crash data are available for this study (2008 to 2012). Analysis details are provided in the sections below. Figure 25: Sample Urban and Suburban Arterial

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 87 Objectives This example is focused on evaluating the crash reduction potential of various design alternatives for an urban arterial. Several improvements were considered as part of the project, including providing a physical median along the corridor in one section of the corridor, providing dedicated bus pullout areas, widening the sidewalk, and providing a median separation. This example demonstrates the quantitative safety analysis of the existing facility and two additional alternatives along the corridor. The first part of the problem illustrates how to calculate the predictive average crash frequency for a signalized intersection (Intersection 1) and an urban roadway segment. The second part of the problem illustrates how to combine all intersections and roadway segments as part of a corridor study, and the analysis of the two different alternatives. The objective of each of the problems is to show how various HSM analysis tools can be applied to assist traffic analysts, engineers, planners, and decision- makers in making sound investment decisions. In some situations, this amount of analysis would not be necessary to make an informed decision, but the issues presented herein should always be considered to assure the final decision is consistent with safety performance objectives. After reviewing this example, the user should be able to: • Understand what input data are required and the assumptions that are commonly made regarding default values for HSM procedures • Calculate the predicted and expected crash frequency of urban and suburban intersections and roadway segments using HSM • Understand how to reasonably interpret the results from an HSM analysis, and how these results can be used to support a particular decision • Understand the limitations of the HSM procedures and when it is appropriate to use other models or computational tools Data Requirements Intersection Data Table 53 summarizes the input data required to apply the predictive method for urban and suburban arterials at Intersections 1 and 2. TABLE 53 Example Problem 3 – Intersections Input Data Characteristics Input Data Intersection 1 Intersection 2 Intersection type 4SG 3ST Traffic flow major road (vpd) 23,000 23,000 Traffic flow minor road (vpd) 14,000 1,500 Intersection lighting Not present Not present Calibration factor( Ci) 1.15 1.15 Data for Unsignalized Intersections Only Number of major-road approaches with left-turn lanes 0 0 Number of minor-road approaches with right-turn lanes 0 0 Data for Signalized Intersections Only Number of approaches with left-turn lanes 0 −

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 88 TABLE 53 Example Problem 3 – Intersections Input Data Characteristics Input Data Intersection 1 Intersection 2 Number of approaches with right-turn lanes 0 − Number of approaches with left-turn signal phasing 0 − Type of left-turn phasing Not applicable − Number of approaches with right-turn-on-red prohibited Not present − Intersection red-light cameras Not present − Sum of all pedestrian crossing volumes 400 − Maximum number of lanes crossed by a pedestrian 5 − Number of bus stops within 1,000 feet (300 meters) of the intersection 1 − Schools within 1,000 feet (300 meters) of the intersection (present/not present) Not present − Number of alcohol sales establishments within 1,000 feet (300 meters) of intersection 1 − For this example, intersection crash data disaggregated by year and collision type are available. Table 54 shows the crash data details for Intersections 1 and 2. TABLE 54 Example Problem 3 – Disaggregated Intersection Crash Data for the Study Period Collision Type Intersection 1 2008 2009 2010 2011 2012 Sum Average Multiple-Vehicle Nondriveway 3 6 4 7 4 24 4.8 Single-Vehicle 0 1 0 0 0 1 0.2 Total 3 7 4 7 4 25 5 Collision Type Intersection 2 2008 2009 2010 2011 2012 Sum Average Multiple-Vehicle Nondriveway 2 6 5 3 4 20 4 Single-Vehicle 0 0 0 0 0 0 0 Total 2 6 5 3 4 20 4

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 89 Roadway Segment Data TABLE 55 Example Problem 3 – Arterial Roadway Segment Input Data Characteristics Input Data Roadway Segment 1 Roadway type 5T Segment length (miles) 0.3 Traffic volume (vpd) 23,000 Type of on-street parking Parallel (commercial/industrial) Proportion of curb length with on-street parking (0.5 x Lpk/L) 0.4 Median width (feet) − Segment lighting Present Auto speed enforcement Not present Major commercial driveways 2 Minor commercial driveways 8 Major industrial/institutional driveways − Minor industrial/institutional driveways − Major residential driveways − Minor residential driveways 2 Other driveways − Speed category Greater than 30 mph Roadside fixed-object density 20 Offset to roadside fixed objects (feet) 10 Calibration factor (Cr) 1.1 Note: Lpk/L = proportion of curb length with on-street parking Similarly, roadway segment crash data disaggregated by year and collision type for the study period are shown in Table 56. TABLE 56 Example Problem 3 – Disaggregated Roadway Segment Crash Data for the Study Period Collision Type Roadway Segment 2008 2009 2010 2011 2012 Sum Average Multiple-Vehicle Nondriveway 5 7 6 8 9 35 7 Single-Vehicle 0 2 1 1 1 5 1 Multiple-Vehicle Driveway-Related 6 5 3 4 2 20 4 Total 11 14 10 13 12 60 12

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 90 Analysis The urban and suburban arterial safety analysis differs from the previous two predictive methods since pedestrian and bicycle collisions must be accounted for, with respect to intersections and roadway segments. Each collision type will be analyzed in detail. The first part of the analysis will be focused on understanding how to apply the predictive method to one signalized intersection (Intersection 1) and one roadway segment (Roadway Segment 1) independently using 2012 data. Next, these steps will be repeated for each year for which data are available. These results will be combined to perform a corridor analysis consisting of one segment and two intersections. Lastly, two additional alternatives for roadway improvements will be analyzed as part of an alternatives evaluation. Intersections The first part of the predictive method is focused on obtaining input data required to apply the predictive model. Detailed information on different recommendations related to data collection can be found in HSM Section 12.4 (HSM p. 12-6). The intersection data summary for this example is provided in Table 53. Select and Apply SPF for Multiple- and Single-Vehicle Collisions For the four-leg signalized intersection, SPF values for multiple-vehicle, single-vehicle, vehicle-pedestrian, and vehicle-bicycle collisions are determined. The general functional form of the multiple- and single-vehicle collision SPFs is shown in the following equations. The SPF for multiple-vehicle collisions is applied to calculate the predicted average crash frequency (total, fatal-and-injury, and PDO crashes) using HSM Equation 12-21 and HSM Table 12-10 (HSM p. 12-29 and 12-30, respectively). The SPF for single-vehicle crashes is applied to calculate the predicted average crash frequency (total, fatal-and-injury, and PDO crashes) using HSM Equation 12-24 and HSM Table 12-12 (HSM p. 12-32 and 12-33). Multiple-Vehicle Collisions by Severity Level for Intersection 1 𝑁 = 𝑒𝑥𝑝 𝑎 + 𝑏 × 𝑙𝑛 𝐴𝐴𝐷𝑇 + 𝑐 × 𝑙𝑛(𝐴𝐴𝐷𝑇 ) 𝑁 = 𝑒𝑥𝑝 −10.99 + 1.07 × 𝑙𝑛(23,000) + 0.23 × 𝑙𝑛(14,000) 𝑁 = 7.04 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 = 𝑒𝑥𝑝 −13.14 + 1.18 × 𝑙𝑛(23,000) + 0.22 × 𝑙𝑛(14,000) 𝑁 ( ) ′ = 2.25 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 = 𝑒𝑥𝑝 −11.02 + 1.02 × 𝑙𝑛(23,000) + 0.24 × 𝑙𝑛(14,000) 𝑁 ( )′ = 4.55 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 91 Single-Vehicle Collisions by Severity Level for Intersection 1 𝑁 = 𝑒𝑥𝑝 −10.21 + 0.68 × 𝑙𝑛(23,000) + 0.27 × 𝑙𝑛(14,000) 𝑁 = 0.45 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 = 𝑒𝑥𝑝 −9.25 + 0.43 × 𝑙𝑛(23,000) + 0.29 × 𝑙𝑛(14,000) 𝑁 ( )′ = 0.12 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 = 𝑒𝑥𝑝 −11.34 + 0.78 × 𝑙𝑛(23,000) + 0.25 × 𝑙𝑛(14,000) 𝑁 ( )′ = 0.33 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 The following adjustments are applied to the predicted average crash frequency for fatal-and-injury crashes and for PDO crashes to ensure the sum matches the total predicted number of crashes. Multiple-Vehicle Collisions by Severity Level for Intersection 1 𝑁 = 𝑁 ( ) + 𝑁 ( ) 𝑁 ( ) = 𝑁 × ( )′ ( )′ ( )′ 𝑁 ( ) = 7.04 × 2.252.25 + 4.55 = 2.33 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 7.04 − 2.33 = 4.71 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Single-Vehicle Collisions by Severity Level for Intersection 1 𝑁 ( ) = 0.45 × 0.120.12 + 0.33 = 0.12 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 0.45 − 0.12 = 0.33 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Select and Apply SPF for Vehicle-Pedestrian and Vehicle-Bicycle Collisions at Signalized Intersections Vehicle-Pedestrian Collisions at Signalized Intersections Vehicle-pedestrian collisions at signalized and unsignalized intersections are estimated using a different set of SPFs. For signalized intersections, use HSM Equation 12-29 (HSM p. 12-36) with coefficients from HSM Table 12-14 (HSM p. 12-37): 𝑁 = 𝑒𝑥𝑝 𝑎 + 𝑏 × 𝑙𝑛(𝐴𝐴𝐷𝑇 ) + 𝑐 × 𝑙𝑛 + 𝑑 × 𝑙𝑛(𝑃𝑒𝑑𝑉𝑜𝑙) + 𝑒 × 𝑛 𝑁 = 𝑒𝑥𝑝 −9.53 + 0.40 × 𝑙𝑛(37,000) + 0.26 × 𝑙𝑛 14,00023,000 + 0.45 × 𝑙𝑛(400) + 0.04 × 5 𝑁 = 0.078 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Vehicle-Bicycle Collisions at Signalized Intersections Vehicle-bicycle collisions are accounted for in HSM Equation 12-31 with intersection adjustment factors taken from HSM Table 12-17 (HSM p. 12-38).

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 92 Before calculating the vehicle-bicycle collisions, the predicted average crash frequency of multiple- and single-vehicle crashes must be calculated: 𝑁 = 𝑁 + 𝑁 𝑁 = 7.04 + 0.45 = 7.49 Apply HSM Part C Crash Modification Factors to Multiple- and Single-Vehicle Collisions CMFs are applied to adjust the estimated crash frequencies for base conditions to account for the effect of site-specific geometry and traffic features. Intersection Left-Turn Lanes (CMF1i) The CMF for left-turn lanes is found in HSM Table 12-24 (HSM p. 12-43). Since Intersection 1 does not have left-turn lanes, a CMF of 1.00 is recommended. Intersection Left-Turn Phasing (CMF2i) HSM Table 12-25 (HSM p. 12-44) provides the CMF for various phasing types. The applied CMF is the product of each leg. For Intersection 1, no protected left-turn phasing is present; therefore, the CMF is equal to 1.00. This CMF does not apply to stop-controlled intersections. Intersection Right-Turn Lanes (CMF3i) CMFs for installation of right-turn lanes are found in HSM Table 12-26 (HSM p. 12-44). Intersection 1 does not have right-turn lanes, yielding a CMF of 1.00. Intersection Right-Turn-on-Red (CMF4i) There are no right-turn-on-red prohibitions in Intersection 1; therefore, a CMF of 1.00 is applied. This CMF is applied by using HSM Equation 12-35 (HSM p. 12-44): 𝐶𝑀𝐹 = 0.98(  ) This CMF does not apply to stop-controlled control intersections. Intersection Lighting (CMF5i) Intersection lighting is not present at the intersection; therefore, the CMF is equal to 1.00. To modify the crashes because of intersection lighting, HSM Equation 12-36 and HSM Table 12-27 (HSM p. 12-45) are used. 𝐶𝑀𝐹 = 1 − 0.38 × 𝑝 Intersection Red-Light Cameras (CMF6i) Red-light cameras are not present at this intersection; therefore, a CMF of 1.00 is recommended. CMF can be estimated using HSM Equations 12-37 through 12-39 (HSM p. 12-45). The combined CMF is then calculated by multiplying all the intersection-related CMFs: 𝐶𝑀𝐹 =𝐶𝑀𝐹 × 𝐶𝑀𝐹 × … × 𝐶𝑀𝐹

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 93 Apply Part C Crash Modification Factors for Vehicle-Pedestrian and Vehicle-Bicycle Collisions at Signalized Intersections Vehicle-Pedestrian Collisions at Signalized Intersections Bus Stops (CMF1p) Intersection 1 has a bus stop within 1,000 feet. The appropriate CMF from HSM Table 12-28 (HSM p. 12-46) is 2.78. Schools (CMF2p) The CMF for the presence of schools near intersections is presented in HSM Table 12-29 (HSM p. 12-46). No schools are present at Intersection 1; therefore, a CMF of 1.00 is applied. Alcohol Sales Establishments (CMF3p) Because of alcohol sales in proximity to Intersection 1, a CMF of 1.12 is applied. HSM Table 12-30 (HSM p. 12-47) provides the CMF values. The combined CMF is then calculated by multiplying all the pedestrian-related CMFs: 𝐶𝑀𝐹 =𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 𝐶𝑀𝐹 =2.78 × 1.00 × 1.12 = 3.11 Vehicle-Bicycle Collisions at Signalized Intersections The sum of the base conditions SPFs (N spf int) is multiplied by the CMFs to obtain the predicted crash frequency (Nbi). This value will be later multiplied by a bicycle adjustment factor. 𝑁 = 𝑁 × (𝐶𝑀𝐹 × 𝐶𝑀𝐹 × … × 𝐶𝑀𝐹 ) 𝑁 = 7.49 × (1.00 × 1.00 × … × 1.00) = 7.49 Apply Calibration Factor The next step is to multiply the results obtained above by the appropriate calibration factor. For this example, the intersection calibration factor is 1.15. Obtain the Predicted Crash Frequency for the Site Lastly, the predicted average crash frequency is calculated using HSM Equations 12-5, 12-6, and 12-7 (HSM p. 12-5 and 12-6), which combine the predicted average crash frequency, crash modification factors, and calibration factors: 𝑁 =𝐶 × (𝑁 + 𝑁 + 𝑁 ) 𝑁 =𝑁 × (𝐶𝑀𝐹 × 𝐶𝑀𝐹 × … × 𝐶𝑀𝐹 ) 𝑁 =𝑁 + 𝑁 Multiple-Vehicle Collisions by Severity Level for Intersection 1 𝑁 = 7.04 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 2.33 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 4.71 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 94 Single-Vehicle Collisions by Severity Level for Intersection 1 𝑁 = 0.45 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 0.12 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 0.33 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 = 𝑁 + 𝑁 = 7.04 + 0.45 = 7.49 𝑁 =𝑁 × (𝐶𝑀𝐹 × 𝐶𝑀𝐹 × … × 𝐶𝑀𝐹 ) 𝑁 = 7.49 × (1.00 × 1.00 × … × 1.00) = 7.49 Vehicle-Pedestrian Collisions at Signalized Intersections 𝑁 = 𝑁 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 𝑁 = 0.078 × (2.78 × 1.00 × 1.12) = 0.242 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Vehicle-Bicycle Collisions at Signalized Intersections The predicted crash frequency Nbi is multiplied by the bicycle crash adjustment factor (fbikei) from HSM Table 12-17 (HSM p. 12-38) using the following equation: 𝑁 = 𝑁 × 𝑓 𝑁 = 7.49 × 0.015 = 0.112 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 The intersection predicted crash frequency is then calculated by multiplying the calibration factor by the sum of the multiple-, single-vehicle, pedestrian, and bicycle predicted crash frequencies. 𝑁 =𝐶 × 𝑁 × 𝑁 × 𝑁 = 1.15 × (7.491 + 0.242 + 0.112) 𝑁 = 9.02 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 𝐶 × 𝑁 × 𝑁 × 𝑁 = 1.15 × (2.447 + 0.242 + 0.112) 𝑁 ( ) = 3.22 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) =𝐶 × (𝑁 ) = 1.15 × (5.044) 𝑁 ( ) = 5.80 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Multiyear Analysis Since 5 years of data are available, all the steps above need to be repeated four more times. In this example, an AADT growth rate of 2 percent is assumed. Table 57 summarizes the calculations for the signalized intersection.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 95 TABLE 57 Example Problem 3 – Intersection 1 Multiyear Analysis Results Intersection 1 Year 2008 2009 2010 2011 2012 AADTmajor 21,248 21,673 22,107 22,549 23,000 AADTminor 12,934 13,193 13,456 13,725 14,000 Crashes/year 3 7 4 7 4 Nbrmv 6.354 6.520 6.690 6.860 7.040 Nbrsv 0.416 0.420 0.430 0.440 0.450 Npedbase 0.075 0.076 0.076 0.077 0.078 Npedi 0.234 0.236 0.238 0.240 0.242 Nbikei 0.102 0.104 0.107 0.110 0.112 CMF1i 4SG 1.00 1.00 1.00 1.00 1.00 CMF2i 4SG 1.00 1.00 1.00 1.00 1.00 CMF3i 4SG 1.00 1.00 1.00 1.00 1.00 CMF4i 4SG 1.00 1.00 1.00 1.00 1.00 CMF5i 4SG 1.00 1.00 1.00 1.00 1.00 CMF6i 4SG 1.00 1.00 1.00 1.00 1.00 CMFcomb 1.00 1.00 1.00 1.00 1.00 CMF1p 2.78 2.78 2.78 2.78 2.78 CMF2p 1.00 1.00 1.00 1.00 1.00 CMF3p 1.12 1.12 1.12 1.12 1.12 CMFped comb 3.11 3.11 3.11 3.11 3.11 Ci 1.15 1.15 1.15 1.15 1.15 Npredicted int 8.171 8.376 8.586 8.801 9.022 The average predicted crash frequency for Intersection 1 is obtained by adding the arithmetic 5-year average of multiple- and single-vehicle, vehicle-pedestrian, and vehicle-bicycle annual predicted crash frequencies. For this example, this value is 8.59 crashes per year. Roadway Segments The first step in applying the predictive method is collecting the data required to apply the SPFs. Detailed information on different recommendations related to data collection can be found in HSM Section 12.4 (HSM p. 12-6). Select and Apply SPF for Single-Vehicle Collisions and Multiple-Vehicle Driveway- and Nondriveway- Related Collisions For urban and suburban arterials, SPFs values are calculated to multiple-vehicle nondriveway, single-vehicle, multiple-vehicle, vehicle-pedestrian, and vehicle-bicycle collisions.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 96 The general functional form of the roadway segment multiple- and single-vehicle collision SPFs, excluding the driveway-related SPF (HSM Equation 12-16 [HSM p. 12-22]), is taken from HSM Equations 12-10 (multiple-vehicle collisions [HSM p. 12-18]) and 12-13 (single-vehicle crashes [HSM p. 12-20]), with appropriate regression coefficients selected from HSM Tables 12-3 and 12-5: 𝑁 = 𝑒 × ( ) ( ) Multiple-Vehicle Nondriveway-Related Collisions by Severity 𝑁 = 𝑒 . . × ( , ) ( . ) 𝑁 = 2.33 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁′ ( ) = 𝑒 . . × ( , ) ( . ) 𝑁 ( )′ = 0.65 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁′ ( ) = 𝑒 . . × ( , ) ( . ) 𝑁 ( )′ = 1.78 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Single-Vehicle Collisions by Severity 𝑁′ = 𝑒 . . × ( , ) ( . ) 𝑁 = 0.55 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁′ ( ) = 𝑒 . . × ( , ) ( . ) 𝑁 ( )′ = 0.12 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁′ ( ) = 𝑒 . . × ( , ) ( . ) 𝑁 ( )′ = 0.40 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 After calculating the initial SPF, adjustment factors are applied to ensure the summation of the fatal- and-injury and PDO annual crashes match the total annual crashes. This is done using the following functional form, HSM Equations 12-11 and 12-14 for fatal-and-injury crashes, and taking the difference for PDO crashes using HSM Equations 12-12 and 12-15 (HSM p. 12-20 and 12-21): 𝑁 ( ) = 𝑁 × ( )′ ( )′ ( )′ Multiple-Vehicle Nondriveway-Related Collisions by Severity 𝑁 ( ) = 2.33 × 0.650.65 + 1.78 = 0.63 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 2.33 − 0.63 = 1.71 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 97 Single-Vehicle Collisions by Severity 𝑁 ( ) = 0.55 × 0.120.12 + 0.40 = 0.13 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 0.55 × 0.400.12 + 0.40 = 0.42 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Multiple-Vehicle Driveway-Related Collisions The number of driveway-related collisions is calculated using HSM Equation 12-16 (HSM p. 12-22). Crashes per driveway type and traffic volume adjustment values are from HSM Table 12-7 (HSM p. 12-24). For the project roadway segment, there are two major commercial driveways, eight minor commercial driveways, and two minor residential driveways: 𝑁 = ∑ 𝑛 × 𝑁 × , 𝑁 = 2 × 0.165 × 23,00015,000 . + 8 × 0.053 × 23,00015,000 . + 2 × 0.016 × 23,00015,000 . = 1.30 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 The driveway-related fatal-and-injury and PDO crashes are calculated by applying proportions found in HSM Table 12-7 (HSM p. 12-24) is HSM Equations 12-17 and 12-18 (HSM p. 12-27). For this example, for a five-lane arterial with a TWLTL, the proportions of fatal-and-injury and PDO crashes are 0.269 and 0.731, respectively. 𝑁 ( ) = 𝑁 × 𝑓 𝑁 ( ) = 1.30 × 0.269 = 0.35 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 𝑁 × 𝑓 𝑁 ( ) = 1.30 × 0.731 = 0.95 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Apply HSM Part C Crash Modification Factors to Single-Vehicle Collisions and Multiple-Vehicle Driveway- and Nondriveway-Related Collisions CMFs are applied to the estimated crash frequencies to adjust for base conditions, to account for the effect of site-specific geometry and traffic features. On-Street Parking (CMF1r) The CMF for on-street parking is calculated using HSM Equation 12-32 with the factor read from HSM Table 12-19 (HSM p. 12-40): 𝐶𝑀𝐹 = 1 + 𝑝 × 𝑓 − 1 𝐶𝑀𝐹 = 1 + 0.40 × (1.709 − 1) = 1.28

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 98 Roadside Fixed Objects (CMF2r) For this CMF, HSM Equation 12-33 (HSM p. 12-40) is applicable, using the fixed-object offset factor from HSM Table 12-20 and the proportion of fixed-object collisions from HSM Table 12-21 (HSM p. 12-41): 𝐶𝑀𝐹 = 𝑓 × 𝐷 × 𝑝 + 1 − 𝑝 𝐶𝑀𝐹 = 0.087 × 20 × 0.016 + (1 − 0.016) = 1.01 Median Width (CMF3r) This CMF is applied to represent the effect of median width in reducing cross-median crashes. However, it is not applicable to medians serving as TWLTL. For this example, a CMF of 1.00 is appropriate, for all other conditions, use HSM Table 12-22 (HSM p. 12-42). Lighting (CMF4r) The effect of adding lighting along the roadway segment is calculated using HSM Equation 12-34, with proportions from HSM Table 12-23 (HSM p. 12-42): 𝐶𝑀𝐹 = 1 − 𝑝 × 1 − 0.72 × 𝑝 − 0.83 × 𝑝 𝐶𝑀𝐹 = 1 − 0.274 × (1 − 0.72 × 0.432 − 0.83 × 0.568) = 0.94 Automated Speed Enforcement (CMF5r) Automated speed enforcement is not present at the study segment; therefore, a CMF of 1.00 is appropriate. More information can be found on HSM p. 12-43. The combined CMF is then calculated by multiplying all the segment-related CMFs. 𝐶𝑀𝐹 = 𝐶𝑀𝐹 × 𝐶𝑀𝐹 × … × 𝐶𝑀𝐹 𝐶𝑀𝐹 = 1.28 × 1.01 × 1.00 × 0.94 × 1.00 = 1.22 Multiple-Vehicle Nondriveway-Related Collisions by Severity 𝑁 = 2.33 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 0.63 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 1.71 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Single-Vehicle Collisions by Severity 𝑁 = 0.55 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 0.13 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 0.42 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 99 Multiple-Vehicle Driveway-Related Collisions by Severity 𝑁 = 1.30 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 0.35 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 0.95 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 = 𝑁 + 𝑁 + 𝑁 𝑁 = 2.33 + 0.55 + 1.30 = 4.18 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 0.63 + 0.13 + 0.35 = 1.10 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 1.71 + 0.42 + 0.95 = 3.08 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 = 𝑁 × (𝐶𝑀𝐹 × 𝐶𝑀𝐹 × … × 𝐶𝑀𝐹 ) 𝑁 = 4.18 × (1.22) = 5.10 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 1.10 × (1.22) = 1.34 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑁 ( ) = 3.08 × (1.22) = 3.75 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Vehicle-Pedestrian and Vehicle-Bicycle Collisions for Urban Roadway Segments The predictive method for the urban and suburban arterials does not include SPFs for pedestrian- and bicycle-related crashes. Vehicle-pedestrian and vehicle-bicycle collisions for urban segments are calculated as a proportion of the predicted multiple-vehicle nondriveway-related, single-vehicle, and multiple-vehicle driveway-related crashes (𝑁 ). Pedestrian and bicycle adjustment factors are provided in the HSM. Vehicle-Pedestrian Collisions along Segments The SPF associated with vehicle-pedestrian collisions along segments is governed by HSM Equation 12-19 with the adjustment factor from HSM Table 12-8 (HSM p. 12-27): 𝑁 = 𝑁 × 𝑓 𝑁 = 5.10 × 0.023 = 0.12 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Vehicle-Bicycle Collisions Along Segments The SPF associated with vehicle-bicycle collisions along segments is similarly calculated, governed by HSM Equation 12-20 with the adjustment factor from HSM Table 12-9: 𝑁 = 𝑁 × 𝑓 𝑁 = 5.10 × 0.012 = 0.06 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 100 NOTE: These factors apply to the methodology for predicting all severity levels combined. All results obtained by applying these pedestrian and bicycle adjustment factors are treated as fatal-and-injury crashes. The adjustment factor does not apply to PDO crashes. Apply Calibration Factor The final step is to multiply the results obtained above by the appropriate calibration factor. For this example, the calibration factor has been assumed to be 1.10. The predicted average crash frequency is calculated using HSM Equation 12.2 (HSM p. 12-4), which combines the predicted average crash frequency for base conditions, CMFs, and calibration factors: 𝑁 = 𝐶 × 𝑁 + 𝑁 + 𝑁 𝑁 = 1.10 × (5.10 + 0.12 + 0.06) = 5.81 𝑁 ( ) = 1.10 × (1.34 + 0.12 + 0.06) = 1.68 𝑁 ( ) = 1.10 × (3.75) = 4.13 Multiyear Analysis Since 5 years of data are available, all the steps above need to be repeated four more times. In this example, an AADT growth rate of 2 percent is assumed. Table 58 summarizes the calculations for the roadway segment. TABLE 58 Example Problem 3 – Roadway Segment 1 Multiyear Analysis Results Roadway Segment 1 Year 2008 2009 2010 2011 2012 AADT 21,248 21,673 22,107 22,549 23,000 Crashes/year 11 14 10 13 12 Nbrmv 2.125 2.175 2.226 2.278 2.332 Nbrsv 0.526 0.531 0.537 0.543 0.548 Nbrdwy 1.182 1.210 1.238 1.267 1.297 Nped 0.108 0.110 0.112 0.115 0.117 Nbike 0.056 0.057 0.059 0.060 0.061 CMF1ru 1.28 1.28 1.28 1.28 1.28 CMF2ru 1.01 1.01 1.01 1.01 1.01 CMF3ru 1.00 1.00 1.00 1.00 1.00 CMF4ru 0.94 0.94 0.94 0.94 0.94 CMF5ru 1.00 1.00 1.00 1.00 1.00 CMFcomb 1.22 1.22 1.22 1.22 1.22 Cr 1.10 1.10 1.10 1.10 1.10 Npredicted seg 5.33 5.45 5.56 5.68 5.81

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 101 The average predicted crash frequency for the study segment is obtained by adding the arithmetic 5-year average of multiple- and single-vehicle, multiple-vehicle driveway-related, vehicle-pedestrian, and vehicle-bicycle annual predicted crash frequencies. For this example, this value is 5.57 crashes per year. Corridor Analysis (Intersections and Roadway Segments) Analysis results for intersections and roadway segments can be combined into a corridor analysis. This approach combines the predicted crash frequency of multiple locations to come up with corridor predicted average crash frequency. Table 59 summarizes the predicted crash frequency of all roadway segments and intersections and provides the corridor totals. TABLE 59 Example Problem 3 – Corridor Predicted Average Crash Frequency Site Type Predicted Average Crash Frequency (crashes/year) Npredicted (Total) Npredicted (Fatal-&-Injury) Npredicted (PDO) Roadway Segment 1 5.57 1.61 3.96 Intersection 1 8.59 3.07 5.53 Intersection 2 2.74 1.05 1.69 Project Total 16.90 5.72 11.18 HSM Tables 12-4 (HSM p. 12-20) and 12-6 (HSM p. 12-22) provide default distributions of crashes by collision type and severity level for multiple-vehicle nondriveway and single-vehicle roadway segment crashes, respectively. HSM Tables 12-11 (HSM p. 12-32) and 12-13 (HSM p. 12-36) provide default distributions of crashes by collision type and severity level for multiple-vehicle and single-vehicle intersection crashes, respectively. These proportions can be applied to the predicted crash frequencies for selected collision types. The HSM provides information on how to update these values using local data (refer to HSM Part C, Appendix A for details). Empirical Bayes Adjustment Method For this example, observed crash data is available by location; therefore, the predictions can be adjusted using the EB method. Details on the applicability of the EB method can be found in HSM Section A.2.1 (HSM p. A-16). After making the adjustments, the expected average crash frequencies for roadway segments and intersections can be combined to come up with a corridor expected average crash frequency. In this example, crash data is available by site; therefore, the site EB method is applicable. Refer to HSM Sections A.2.4 and A.2.5 (HSM p. A-19 and A-20) for additional details on the different EB methods. Available observed crash data for segments and intersections has been broken down into multiple- vehicle and single-vehicle crashes, as shown in Table 60.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 102 TABLE 60 Example Problem 3 – Disaggregated Roadway Segment and Intersection Crash Data for the Study Period (2008 to 2012) Collision Type Intersection 1 2008 2009 2010 2011 2012 Sum Average Multiple-Vehicle Nondriveway 3 6 4 7 4 24 4.8 Single-Vehicle 0 1 0 0 0 1 0.2 Total 3 7 4 7 4 25 5 Collision Type Intersection 2 2008 2009 2010 2011 2012 Sum Average Multiple-Vehicle Nondriveway 2 6 5 3 4 20 4 Single-Vehicle 0 0 0 0 0 0 0 Total 2 6 5 3 4 20 4 Collision Type Roadway Segment 2008 2009 2010 2011 2012 Sum Average Multiple-Vehicle Nondriveway 5 7 6 8 9 35 7 Single-Vehicle 0 2 1 1 1 5 1 Multiple-Vehicle Driveway-Related 6 5 3 4 2 20 4 Total 11 14 10 13 12 60 12 There were 60 roadway segment crashes and 45 intersection crashes for the study period. The expected number of crashes for either segments or intersections is calculated using HSM Equation A-4 (HSM p. A-19): 𝑁 = 𝑤 × 𝑁 + (1 − 𝑤) × 𝑁 The weighting adjustment factors for each collision type for the sample roadway segments and intersections are needed to complete these calculations. HSM Equation A-5 (HSM p. A-19) is used to obtain the weighting factors: 𝑤 = ×∑ Overdispersion parameters are also estimated for each SPF set. The overdispersion parameter associated with segment SPFs are found in HSM Table 12-3 (multiple-vehicle nondriveway-related [HSM p. 12-19]), Table 12-5 (single-vehicle [HSM p. 12-21]), and Table 12-7 (multiple-vehicle driveway- related [HSM p. 12-24]). Intersection overdispersion parameters for multiple- and single-vehicle collisions can be found in HSM Tables 12-10 (HSM p. 12-30) and 12-12 (HSM p. 12-33), respectively. Intersection overdispersion parameters for vehicle-pedestrian collisions can be found in HSM Table 12-14 (HSM p. 12-37). The following is an example of how to calculate the weighting adjustment factor for segment multiple- vehicle nondriveway collisions using the weighting factors equation. The segment overdispersion

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 103 parameter for this collision type is 0.81, and the sum of all the predicted roadway segment crashes is 14.96: 𝑤 = 11 + 0.81 × (2.86 + 2.92 + 2.99 + 3.06 + 3.13) = 0.076 The segment predicted average crash frequency for this collision type is 2.99 crashes per year. The expected number of crashes is calculated as follows: 𝑁 = 0.076 × 2.992 + (1 − 0.076) × 7 = 6.70 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Expected average crash frequencies are presented in Table 61. Columns 2 through 4 contain the predicted average crash frequency for total crashes, fatal-and-injury, and PDO. The fifth column contains the observed/reported number of crashes per year. Columns 6 and 7 contain the overdispersion parameter and weighted adjustment to be used to obtain the expected average crash frequency (last column). TABLE 61 Example Problem 3 – Predicted and Expected Crash Frequency Calculations Summary (2008 to 2012) Collision Type/ Site Type Predicted Average Crash Frequency (crashes/year) Observed/ Reported Crashes (Nobserved) (crashes/ year) Overdisper sion Parameter (k) Weighted Adjustment (w) (Equation A-5 from HSM Part C, Appendix A) Expected Average Crash Frequency (Nexpected) (Equation A-4 from HSM Part C, Appendix A) Npredicted (Total) Npredicted (Fatal- &- Injury) Npredicted (PDO) Roadway Segments Multiple-Vehicle Nondriveway Roadway Segment 1 2.99 0.80 2.19 7 0.810 0.076 6.69 Single-Vehicle Roadway Segment 1 0.72 0.17 0.55 1 0.520 0.348 0.90 Multiple-Vehicle Driveway-Related Roadway Segment 1 1.66 0.45 1.22 4 0.100 0.546 2.73 Intersections Multiple-Vehicle Intersection 1 7.70 2.54 5.16 5 0.390 0.062 4.98 Intersection 2 2.38 0.87 1.51 4 0.800 0.095 3.85 Single-Vehicle Intersection 1 0.50 0.13 0.37 0 0.360 0.528 0.36 Intersection 2 0.26 0.08 0.18 0 1.140 0.406 0.10 Total 16.21 5.03 11.18 21 − − 19.61 The total expected crashes for the site is the sum of the roadway segment and intersections. Table 62 summarizes the predicted values for bicycle and pedestrian crashes. Agencies with observed/reported crashes for these types can likewise calculate expected crashes with an overdispersion parameter. In the

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 104 absence of observed/reported pedestrian and bicycle crashes, the total and fatal-and-injury (not applicable to PDO) predicted crash frequencies for roadway segments and intersections are calculated by adding the multiple-vehicle and single-vehicle crashes to the pedestrian and bicycle predicted crashes. TABLE 62 Example Problem 3 – Predicted Pedestrian and Bicycle Average Crash Frequency (2008 to 2012) Site Type Predicted Average Crash Frequency Nped Nbike Roadway Segments Roadway Segment 1 0.124 0.065 Intersections Intersection 1 0.274 0.123 Intersection 2 0.064 0.042 Combined 0.461 0.230 Lastly, the total expected average number of crashes for the corridor is 20.3 crashes per year, as shown in Table 63. Results of the analysis can be found in the sample spreadsheets provided with the Highway Safety Manual User Guide. TABLE 63 Example Problem 3 – Corridor Predicted and Expected Crash Frequencies Crash Severity Level Npredicted Nped Nbike Npredicted Total Nexpected (vehicle) Nexpected Total Total (3)+(4)+(6) 16.21 0.46 0.23 16.90 19.61 20.30 Fatal-and-Injury (6)Total × (2)FI / (2) Total (3)+(4)+(6) 5.03 0.46 0.23 5.72 6.09 6.78 PDO − − (6)Total × (2)PDO / (2) Total (3)+(4)+(6) 11.18 0.00 0.00 11.18 13.52 13.52 Alternatives Analysis The first section of this example demonstrated the application of the predicted method for urban and suburban arterial roadway segments and intersections. The predictive method can also be applied to alternative analysis. This process is more detailed and specific about the impacts of the project. The agency develops potential alternatives and compares performance across alternatives, as shown in Figure 26. The next example shows how to apply the urban and suburban arterials predictive method to compare alternatives. Calculations and formulas are similar to the previous example, and results are provided in summary tables.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 105 No Build. The facility is an urban arterial with commercial development. A TWLTL provides for left-turn movements to and from the median. The current configuration allows for left-turn movements at any location along the corridor. The corridor has on-street parallel parking. The posted speed limit is 35 mph. Properties adjacent to the facility have multiple direct access points to the corridor. The pedestrian sidewalk is restricted to 3 feet at some locations along the corridor. Alternative 1. No land use changes are anticipated in the facility. A physical 14-foot median is added in one section of the corridor. The remainder of the corridor remains as TWLTL. Bus pullout areas are provided at the existing bus stop. The alternative provides a 12-foot sidewalk with a 2-foot buffer. Single left-turn lanes and phasing are added to the major road at the signalized intersection. Intersection lighting is added to both intersections. Alternative 2. No land use changes are anticipated along the corridor. Right-of-way acquisition led to the addition of a median separation and dedicated high-occupancy vehicle lane. A 12-foot pedestrian sidewalk is provided with a 3-foot buffer. The TWLTL is replaced with exclusive left-turn lanes, which provide limited access to left-turning vehicles at dedicated locations across the corridor. Minor commercial driveways in large parking areas are consolidated along the corridor, going from eight to four in total. Left-turn lanes and protected left-turn phasing are provided for all four legs at the signalized intersection. Exclusive right- and left-turn lanes are provided at the three-leg unsignalized intersection. Figure 26: Example Problem 2 – Project Alternatives It is assumed that AADT remains the same in each alternative, and the road does not attract any additional traffic. Tables 64 and 65 contain the input data for the different scenarios. Only geometric elements that are being upgraded are listed in Tables 64 and 65. The previous example is the No Build condition.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 106 TABLE 64 Example Problem 3 – Intersection Alternatives Input Data Characteristics Input Data by Alternative No Build Alternative 1 Alternative 2 Intersection 1 Intersection type 4SG 4SG 4SG Intersection lighting Not present Present Present Data for Signalized Intersections Only Number of approaches with left-turn lanes 0 2 4 Number of approaches with left-turn signal phasing 0 2 4 Type of left-turn signal phasing for leg 1 Protected/ Permitted Protected Type of left-turn signal phasing for leg 2 Protected/ Permitted Protected Type of left-turn signal phasing for leg 3 Protected/ Permitted Type of left-turn signal phasing for leg 4 (if applicable) Protected/ Permitted Maximum number of lanes crossed by a pedestrian 5 5 7 Intersection 2 Intersection type 3ST 3ST 3ST Intersection lighting Not present Present Present Data for Unsignalized Intersections Only -- -- -- Number of major-road approaches with left-turn lanes 0 0 2 Number of major-road approaches with right-turn lanes 0 0 1 TABLE 65 Example Problem 3 – Roadway Segments Alternatives Input Data Segment Characteristics Input Data by Alternative No Build Alternative 1 Alternative 2 Roadway type 5T 5T 4D Type of on-street parking Parallel (Commercial/ Industrial) None None Proportion of curb length with on-street parking 0.4 0 0 Median width (feet) – for divided only Not present Not present 10 Minor commercial driveways 8 8 4 Offset to roadside fixed objects (feet) 10 2 15 The effect of the multiple safety countermeasures (such as lighting and adding left-turn lanes) is reflected in the decrease of predicted average crash frequency.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 107 These safety improvements are all considered by the application of CMFs, which are used to adjust the SPF base condition estimate of predicted average crash frequency for the effect of the individual geometric design and traffic control features. The CMF for the SPF base condition of each geometric design or traffic control feature has a value of 1.00. Calculations for the No Build scenario are the same as the first part of the example. Table 66 summarizes the results for all alternatives. Total predicted, observed, and expected average crash frequencies are provided. TABLE 66 Example Problem 3 – Alternative Analysis Summary Results Alternative Site Type Npredicted Nobserved Nexpected No Build Roadway Segment 1 5.6 12 10.5 Intersection 1 8.6 5 5.8 Intersection 2 2.7 4 4.1 Total 16.9 21 20.4 Alternative 1 Roadway Segment 1 4.5 12 10 Intersection 1 6.2 5 5.5 Intersection 2 2.5 4 4 Total 13.3 21 19.5 Alternative 2 Roadway Segment 1 1.5 12 9.2 Intersection 1 4.6 5 5.3 Intersection 2 1 4 3.4 Total 7.1 21 17.9 Results and Discussion The use of the HSM in alternative evaluation allows the agency to quantify the impact of safety improvements such as removing on-street parking, consolidating driveways, installing a raised median, and adding left-turn lanes and phasing. This gives the agency a tool that provides valuable information in the decision-making process. NOTE: The HSM does not require any agency to implement a particular alternative based solely on the safety performance evaluation, and it is not intended to be a substitute for the exercise of sound engineering judgment. The No Build predicted crash frequency is lower than the observed crash frequency. This indicates that more crashes are occurring on the site than the average site with similar characteristics. The results in Table 66 also indicate that implementation of Alternatives 1 and 2 would reduce the predicted number of crashes by 22 percent and 58 percent, respectively. However, after the EB adjustment using observed crash data, the expected number of crashes for Alternatives 1 and 2 are 4 percent and 12 percent lower, respectively, than the No Build scenario. Overall, the different improvement projects are anticipated to reduce the total crashes for both alternatives. However, an economic evaluation is required to better understand which alternative is the

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 108 most cost-effective. Refer to HSM Chapter 7, Economic Appraisal, for methods to compare the benefits of potential safety countermeasures to crash costs. 3.3 HSM in Design 3.3.1 Overview Historically, the highway design process was based on the application of established design criteria. Adherence to design standards was viewed as the means to establish an acceptable level of safety. With the release of the HSM, designers are provided with tools to perform safety performance-based design. This allows development of solutions based not just on design standards, but also on quantifying the safety performance of different design considerations. For instance, designers can establish the safety impact of changing a design parameter, evaluate the impact of design exceptions on safety performance, assess the interactions of the road user with the highway, and evaluate design solutions based on user abilities and limitations using the human factors information included in the manual. 3.3.2 Example Problem 4 Evaluation of Curve Realignment versus Design Exception Introduction The example is a rural two-lane road that is being upgraded from a posted speed limit of 40 mph to 60 mph. Several changes in the roadway alignment are expected, particularly around curves. However, one curve location is adjacent to a high-quality wetland, and reconstructing such a curve may present a challenge from permitting, constructability, and cost points of view. The other option is to leave the existing curve geometry untouched and request a design exception. To mitigate the potential adverse effects of the design exception, some improvements are considered, including shoulder widening and paving shoulders. To better understand the safety benefits of reconstruction versus design exception, an analysis for both alternatives was conducted and is described in the following sections. Five years of crash data are available (2008 to 2012). In this example, the existing curve will be referred as Roadway Segment 1, the proposed curve will be referred to as Roadway Segment 2, and the existing curve with mitigation measures will be referred to as Roadway Segment 3. Objectives This example is focused on determining the safety performance of two design alternatives of a curve location to help design engineers with the decision-making process. The problem illustrates how to calculate the predictive and expected average crash frequency for two curve locations with different radii. After reviewing this example, the user should be able to: • Understand what input data are required and the assumptions that are commonly made regarding default values for the HSM procedures • Calculate the predicted and expected crash frequency of rural two-lane curve segments using the HSM • Understand how to reasonably interpret the results from an HSM analysis and how these results can be used to support a particular decision • Understand the limitations of the HSM procedures and when it is appropriate to use other models or computational tools

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 109 Data Requirements Roadway Segment Data Table 67 contains the input data for this analysis. TABLE 67 Example Problem 4 – Curve Segments Input Data Characteristics Input Data Roadway Segment 1 Roadway Segment 2 Segment length (feet) 0.24 0.30 Traffic volume (vpd) 13,500 13,500 Lane width (feet) 12 12 Shoulder width (feet) 2 6 Shoulder type Gravel Paved Length of horizontal curve (feet) 0.24 0.30 Radius of curvature (feet) 1,600 2,000 Spiral transition curve Not present Not present Superelevation variance 0.02 0 Grade 2 2 Driveway density 0 0 Centerline rumble strips Not present Not present Passing lanes Not present Not present TWLTL Not present Not present Roadside hazard rating (RHR) 4 3 Segment lighting Not present Not present Auto speed enforcement Not present Not present Calibration factor (Cr) 1.23 1.23 Observed crash data (crashes/year) 12 12 Analysis Calculations for Roadway Segments 1 and 2 shown next are for year 2012. Details on the multiyear analysis are provided in the following sections. Roadway Segments Segment data required to apply the predictive method are summarized in Table 67. Roadway Segment 1 length is 0.24 mile with a curve radius of 1,600 feet. Roadway Segment 2 length is 0.3 mile with a curve radius of 2,000 feet. As part of the new realignment, Roadway Segment 2 shoulder type is upgraded from gravel to paved and widened from 2 feet to 6 feet. The superelevation and RHR for Roadway Segment 2 are

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 110 also upgraded. All remaining parameters are the same for both locations. Information on different recommendations related to data collection is presented in HSM Section 10.4. Select and Apply SPFs For the selected site, apply the appropriate SPF for rural two-lane, two-way roads. The SPF can be calculated using HSM Equation 10-6 (HSM p. 10-15): 𝑁 = 𝐴𝐴𝐷𝑇 × 𝐿 × 365 × 10 × 𝑒 . 𝑁 = 13,500 × 0.24 × 365 × 10 × 𝑒 . 𝑁 = 0.87 𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟 𝑁 = 13,500 × 0.30 × 365 × 10 × 𝑒 . 𝑁 = 1.08 𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟 Apply HSM Part C Crash Modification Factors Multiply the result obtained above by the appropriate CMFs to adjust the estimated crash frequency for base conditions to the site-specific geometry and traffic features. Lane Width (CMF1r) CMF1r can be calculated using HSM Equation 10-11 (HSM p. 10-24) shown below: 𝐶𝑀𝐹 = (𝐶𝑀𝐹 − 1) × 𝑝 + 1 CMFra is estimated using HSM Table 10-8 (HSM p. 10-24). For a 12-foot lane width and AADT greater than 2,000, the CMF for the effect of lane width on related crashes (such as single-vehicle run-off-the- road and multiple-vehicle head-on, opposite-direction sideswipe, and same-direction sideswipe crashes) is 1.00. For this example, since the lane width is the same as the base conditions, the applicable CMF for both roadway segments is 1.00. Shoulder Width and Type (CMF2r) CMF2r can be calculated using HSM Equation 10-12 (HSM p. 10-27). For this example, a 2-foot gravel shoulder yields a CMFwra of 1.18 for Roadway Segment 1 and 1.00 for Roadway Segment 2 (shoulder width HSM Table 10-9 [HSM p. 10-25]) and CMFtra of 1.0 (shoulder type HSM Table 10-10 [HSM p. 10-26]). The percentage of related crashes is the same as that calculated for the lane width CMF: 𝐶𝑀𝐹 = (𝐶𝑀𝐹 × 𝐶𝑀𝐹 − 1) × 𝑝 + 1 𝐶𝑀𝐹 = (1.3 × 1.01 − 1) × (0.521 + 0.016 + 0.037) + 1 𝐶𝑀𝐹 = 1.18 𝐶𝑀𝐹 = (1 × 1 − 1) × (0.574) + 1 𝐶𝑀𝐹 = 1.00

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 111 Horizontal Curve (CMF3r) For this example, Roadway Segment 1 length is 0.24 mile with a radius of curvature of 1,600 feet, and Roadway Segment 2 length is 0.30 mile with a radius of curvature of 2,000 feet. The CMF is calculated using HSM Equation 10-13 (HSM p. 10-27): 𝐶𝑀𝐹 = . × . . ×. × 𝐶𝑀𝐹 = 1.55 × 0.24 + 80.21,600 − 0.012 × 01.55 × 0.24 𝐶𝑀𝐹 = 1.13 𝐶𝑀𝐹 = 1.55 × 0.30 + 80.22,000 − 0.012 × 01.55 × 0.30 𝐶𝑀𝐹 = 1.09 Superelevation (CMF4r) The superelevation variance for Roadway Segment 1 is 0.02 feet per foot, and for Roadway Segment 2 it is 0. Therefore, the superelevation is calculated using HSM Equation 10-16 (HSM p. 10-28): 𝐶𝑀𝐹 . = 1.06 + 3 × (𝑆𝑉 − 0.02) 𝐶𝑀𝐹 . = 1.06 + 3 × (0.02 − 0.02) 𝐶𝑀𝐹 = 1.06 𝐶𝑀𝐹 . = 1.00 Grade (CMF5r) A 2 percent grade section falls under the level grade category in HSM Table 10-11 (HSM p. 10-28), resulting in a CMF of 1.00 for both roadway segments. Driveway Density (CMF6r) Driveway density of less than five driveways per mile leads to a CMF6R of 1.00. Otherwise, the CMF is calculated using HSM Equation 10-17 (HSM p. 10-29): 𝐶𝑀𝐹 = . ×[ . . × ( )]. ×[ . . × ( )] 𝐶𝑀𝐹 = 1.00 𝐶𝑀𝐹 = 1.00 Centerline Rumble Strips (CMF7r) The roadway segments do not have centerline rumble strips; therefore, a CMF of 1.00 is applied. See HSM p. 10-29 for additional details. Passing Lanes (CMF8r) Passing lanes are not present in the example; therefore, a CMF of 1.00 is appropriate for both roadway segments. See HSM p. 10-29 for additional details. Two-Way, Left-Turn Lane (CMF9r) TWLTLs are not present; therefore, a CMF of 1.00 is applied for this example. See HSM p. 10-29 for additional details.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 112 Roadside Design (CMF10r) The data in this example indicate a RHR of 4 for Roadway Segment 1, and a rating of 3 for Roadway Segment 2. The CMF is calculated using HSM Equation 10-20 (HSM p. 10-30): 𝐶𝑀𝐹 = . . ×. 𝐶𝑀𝐹 = . . ×. 𝐶𝑀𝐹 = 1.07 𝐶𝑀𝐹 = . . ×. 𝐶𝑀𝐹 = 1.00 Lighting (CMF11r) Lighting is not present at this location; therefore, a CMF of 1.00 is applied. See HSM p. 10-30 for additional details. Automated Speed Enforcement (CMF12r) The site does not have automated speed enforcement available; therefore, a CMF of 1.00 is applied. See HSM p. 10-30 for additional details. The combined CMF is then calculated by multiplying all the intersection CMFs: 𝐶𝑀𝐹 =𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 × ⋯× 𝐶𝑀𝐹 𝐶𝑀𝐹 =1.0 × 1.18 × 1.13 × 1.06 × 1.0 × 1.0 × 1.0 × 1.0 × 1.0 × 1.07 ×1.0 × 1.0 = 1.517 𝐶𝑀𝐹 =1.0 × 1.0 × 1.09 × 1.0 × 1.0 × 1.0 × 1.0 × 1.0 × 1.0 × 1.0 × 1.0 × 1.0 = 1.086 Apply Calibration Factor Multiply the predicted average crash frequency and CMF results obtained in previous steps by the appropriate calibration factor. For this example, the calibration factor has been assumed to be 1.23. Obtain the Predicted Crash Frequency for the Site The predicted average crash frequency is calculated using HSM Equation 10-2 (HSM p. 10-3), combining results from previous steps: 𝑁 = 𝑁 × 𝐶 × (𝐶𝑀𝐹 × 𝐶𝑀𝐹 × … × 𝐶𝑀𝐹 ) 𝑁 = 0.87 × 1.23 × (1.52) 𝑁 = 1.62 𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟 𝑁 = 1.08 × 1.23 × (1.09) 𝑁 = 1.45 𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟 Multiyear Analysis Since 5 years of data are available, all the previous steps need to be repeated four more times. In this example, a growth rate of 1.5 percent is assumed. Table 68 summarizes the calculations for the study period.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 113 TABLE 68 Example Problem 4 – Roadway Segment 1 Multiyear Analysis Results Roadway Segment 1 Year 2008 2009 2010 2011 2012 AADT 12,719 12,910 13,104 13,300 13,500 Nspf 0.816 0.828 0.84 0.853 0.866 CMF1r 1.00 1.00 1.00 1.00 1.00 CMF2r 1.18 1.18 1.18 1.18 1.18 CMF3r 1.13 1.13 1.13 1.13 1.13 CMF4r 1.06 1.06 1.06 1.06 1.06 CMF5r 1.00 1.00 1.00 1.00 1.00 CMF6r 1.00 1.00 1.00 1.00 1.00 CMF7r 1.00 1.00 1.00 1.00 1.00 CMF8r 1.00 1.00 1.00 1.00 1.00 CMF9r 1.00 1.00 1.00 1.00 1.00 CMF10r 1.07 1.07 1.07 1.07 1.07 CMF11r 1.00 1.00 1.00 1.00 1.00 CMF12r 1.00 1.00 1.00 1.00 1.00 CMFcomb 1.52 1.52 1.52 1.52 1.52 Cr 1.23 1.23 1.23 1.23 1.23 Npredicted seg 1.52 1.54 1.57 1.59 1.62 TABLE 69 Example Problem 4 – Roadway Segment 2 Multiyear Analysis Results Roadway Segment 2 Year 2008 2009 2010 2011 2012 AADT 12,719 12,910 13,104 13,300 13,500 Nspf 1.019 1.035 1.050 1.066 1.082 CMF1r 1.00 1.00 1.00 1.00 1.00 CMF2r 1.00 1.00 1.00 1.00 1.00 CMF3r 1.09 1.09 1.09 1.09 1.09 CMF4r 1.00 1.00 1.00 1.00 1.00 CMF5r 1.00 1.00 1.00 1.00 1.00 CMF6r 1.00 1.00 1.00 1.00 1.00 CMF7r 1.00 1.00 1.00 1.00 1.00 CMF8r 1.00 1.00 1.00 1.00 1.00

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 114 TABLE 69 Example Problem 4 – Roadway Segment 2 Multiyear Analysis Results Roadway Segment 2 Year 2008 2009 2010 2011 2012 CMF9r 1.00 1.00 1.00 1.00 1.00 CMF10r 1.00 1.00 1.00 1.00 1.00 CMF11r 1.00 1.00 1.00 1.00 1.00 CMF12r 1.00 1.00 1.00 1.00 1.00 CMFcomb 1.09 1.09 1.09 1.09 1.09 Cr 1.23 1.23 1.23 1.23 1.23 Npredicted seg 1.36 1.38 1.40 1.42 1.45 The average predicted crash frequency for Roadway Segments 1 and 2 are obtained through the arithmetic average of the annual predicted crash frequencies (Npredicted seg). The average for Roadway Segments 1 and 2 are 1.57 and 1.40 crashes per year, respectively. Empirical Bayes Adjustment Method The next step in the process is to update predictions based on the observed/reported crashes. Twelve roadway segment crashes occurred per year. The predictive models indicate that the total predicted average crash frequencies for Roadway Segments 1 and 2 are 1.57 and 1.40 crashes per year, respectively. The predicted average crash frequency is then adjusted using the EB method by applying the following steps. In this example, the proposed geometric upgrade represents a minor change in alignment; therefore, the EB method is applicable. Refer to HSM Section A.2.1 (HSM p. A-16) for additional details about the applicability of the EB method. The site EB method is applicable. Refer to HSM Section A.2.5 (HSM p. A-20 to A-22) for additional details on the different EB methods. The expected number of crashes for segments is calculated by HSM Equation A-4 (HSM p. A-19): 𝑁 = 𝑤 × 𝑁 + (1 − 𝑤) × 𝑁 To complete this calculation, weighting adjustment factors are needed for the samples. Calculate using the previous crash predictions with HSM Equation A-5 (HSM p. A-19): 𝑤 = ×∑

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 115 For this calculation, the overdispersion parameter from each of the applied SPFs is needed. The overdispersion parameter for Roadway Segment 1 is 0.983 and for Roadway Segment 2 is 0.787. The closer the overdispersion parameter is to zero, the more statistically reliable the SPF. On a per-mile basis, the overdispersion parameter is calculated by using HSM Equation 10-7 (HSM p. 10-16): 𝑘 = . 𝑘 = 0.2360.24 𝑘 = 0.983 𝑘 = 0.2360.30 𝑘 = 0.787 Using these overdispersion parameters, the weighting adjustment factors are found to be 0.115 and 0.153 for Roadway Segments 1 and 2, respectively: 𝑤 = 11 + 0.983 × (1.52 + 1.54 + 1.57 + 1.59 + 1.62) 𝑤 = 0.115 𝑤 = 11 + 0.787 × (1.36 + 1.38 + 1.40 + 1.42 + 1.45) 𝑤 = 0.153 Twelve observed/reported crashes per year were reported in the curve. The expected number of crashes for the roadway segments is then calculated as follows: 𝑁 = 𝑤 × 𝑁 + (1 −𝑤) × 𝑁 𝑁 = 0.115 × 1.57 + (1 − 0.115) × 12 𝑁 = 10.8 𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟 𝑁 = 0.153 × 1.40 + (1 − 0.153) × 12 𝑁 = 10.4 𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟 Results of the analysis can be found in the sample spreadsheets provided with the Highway Safety Manual User Guide. Table 70 presents a summary of the predictive method calculations. Columns 2 through 4 contain the predicted average crash frequency for total crashes, fatal-and-injury, and PDO. The fifth column contains the observed/reported number of crashes per year. Columns 6 and 7 contain the overdispersion parameter and weighted adjustment to be used to obtain the expected average crash frequency (last column).

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 116 TABLE 70 Example Problem 4 – Predicted, Expected, and Observed Crash Frequency Calculations Summary (2008 to 2012) Site Type Predicted Average Crash Frequency (crashes/year) Observed/ Reported Crashes (Nobserved) (crashes/ year) Overdisper sion Parameter (k) Weighted Adjustment (w) (Equation A-5 from HSM Part C Appendix A) Expected Average Crash Frequency (Nexpected) (Equation A-4 from HSM Part C Appendix A) Npredicted (Total) Npredicted (Fatal-&- Injury) N predicted (PDO) Roadway Segment 1 1.568 0.503 1.065 12 0.983 0.115 10.8 2008 1.522 0.488 1.033 12 0.983 2009 1.545 0.496 1.049 12 0.983 2010 1.568 0.503 1.065 12 0.983 2011 1.591 0.511 1.080 12 0.983 2012 1.615 0.518 1.097 12 0.983 Roadway Segment 2 1.404 0.451 0.953 12 0.787 0.153 10.4 2008 1.362 0.437 0.925 12 0.787 2009 1.383 0.444 0.939 12 0.787 2010 1.403 0.450 0.953 12 0.787 2011 1.424 0.457 0.967 12 0.787 2012 1.446 0.464 0.982 12 0.787 Details about the predictive method calculations can be found in the Highway Safety Manual User Guide spreadsheets. Comparison of the predicted and observed crash frequencies shows that the site is experiencing more crashes than the average site with similar characteristics. Application of Mitigation Measures The next step is to calculate the safety effects of the mitigation measures on the existing roadway. The existing curve gravel shoulders are 2 feet wide. The proposed improvements include paving the shoulders and increasing the width to 4 feet. Since the changes only involve shoulder type and width, all the other steps shown in the previous section are the same. Shoulder Width and Type (CMF2r) CMF2r can be calculated using HSM Equation 10-2 (HSM p. 10-3). For this example, a 4-foot paved shoulder yields a CMFwra of 1.15 (shoulder width HSM Table 10-9 [HSM p. 10-25]) and CMFtra of 1.0 (shoulder type HSM Table 10-10 [HSM p. 10-26]). The percentage of related crashes is the same as that calculated for the lane width CMF: 𝐶𝑀𝐹 = (𝐶𝑀𝐹 × 𝐶𝑀𝐹 − 1) × 𝑝 + 1 𝐶𝑀𝐹 = (1.15 × 1 − 1) × (0.574) + 1 𝐶𝑀𝐹 = 1.09 Table 71 summarizes the results of the three scenarios. Roadway Segment 3 refers to the existing curve with the addition of mitigation measures to request the design exception.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 117 TABLE 71 Example Problem 4 – Predicted, Expected, and Observed Crash Frequency Calculations Summary for the Three Scenarios (2008 to 2012) Site Type Predicted Average Crash Frequency (crashes/year) Observed/ Reported Crashes (Nobserved) (crashes/ year) Overdisper sion Parameter (k) Weighted Adjustment (w) (Equation A-5 from HSM Part C Appendix A) Expected Average Crash Frequency (Nexpected) (Equation A-4 from HSM Part C Appendix A) Npredicted (Total) Npredicted (Fatal-&- Injury) N predicted (PDO) Roadway Segment 1 1.568 0.503 1.065 12 0.983 0.115 10.80 2008 1.522 0.488 1.033 12 0.983 2009 1.545 0.496 1.049 12 0.983 2010 1.568 0.503 1.065 12 0.983 2011 1.591 0.511 1.080 12 0.983 2012 1.615 0.518 1.097 12 0.983 Roadway Segment 2 1.404 0.451 0.953 12 0.787 0.153 10.37 2008 1.362 0.437 0.925 12 0.787 2009 1.383 0.444 0.939 12 0.787 2010 1.403 0.450 0.953 12 0.787 2011 1.424 0.457 0.967 12 0.787 2012 1.446 0.464 0.982 12 0.787 Roadway Segment 3 1.444 0.463 0.980 12 0.983 0.123 10.70 2008 1.401 0.450 0.951 12 0.983 2009 1.422 0.456 0.966 12 0.983 2010 1.443 0.463 0.980 12 0.983 2011 1.465 0.470 0.995 12 0.983 2012 1.487 0.477 1.010 12 0.983 The results, summarized in Table 72, indicate that the proposed curve will reduce the total expected crash frequency by about 4 percent (0.4 crash per year). The existing curve with mitigation measures reduces the total expected crash frequency by only 1 percent (0.1 crash per year). TABLE 72 Example Problem 4 – Analysis Results Summary Site Type Length (miles) Nobserved Crash Frequencies (crashes/year) Npredicted Nexpected Roadway Segment 1 – Existing Curve 0.24 12 1.6 10.8 Roadway Segment 2 – Proposed Curve 0.3 12 1.4 10.4 Roadway Segment 3 – Existing Curve with Mitigation Measures 0.24 12 1.4 10.7

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 118 Results and Discussion The application of the HSM in the design stage provides engineers with valuable information in the decision-making process. NOTE: The HSM does not require agencies to implement specific alternatives based solely on safety performance evaluation but instead provides the means to make an informed decision. The analysis conducted to determine the crash reduction impacts of upgrading the existing curve to comply with the current roadway design guidance indicates that more crashes are occurring at the site than the average site with similar characteristics. The predicted crash frequencies for curve Roadway Segments 1 and 2 are 1.6 and 1.4 crashes per year, respectively. However, the observed annual crash frequency for the site is 12 crashes per year. After application of the EB adjustment, the expected crash frequency resulted in 10.8 and 10.4 crashes per year for Roadway Segments 1 and 2, respectively. To request a design exception, mitigation measures were applied to the existing curve Roadway Segment 1. Results of the application of the predictive method to the curve with mitigation measures show predicted and expected crash frequencies of 1.4 and 10.7 crashes per year, respectively. For the analysis, the proposed curve alignment would reduce expected crash frequency by 4 percent, or 0.4 crashes per year. This reduction may not seem significant, so the analyst may need to look at other factors such as crash severity and specific collision types that are being addressed with the improved alignment or mitigation measures. In addition, this might be only one element of the entire corridor project, and significant differences may become obvious when reviewing the corridor as a whole. Results may not always be favorable. A treatment (such as concrete median barriers) may increase the total crash frequency but reduce the severe crashes. Sound engineering judgment is ultimately the main driver of the decision-making process. Results from this analysis offer engineers additional information to make an informed decision. The next step is to conduct an economic evaluation to determine the most cost-effective investment. Refer to HSM Chapter 7, Economic Appraisal, for methods to compare the benefits of potential crash countermeasures to crash costs. 3.3.3 Example Problem 5: Intersection Skew Angle Introduction A four-leg stop-controlled intersection on a rural multilane highway has a leg on the minor road with a skew angle of 40 degrees. Due to an increase in crash frequency at this location, the local jurisdiction has considered removing the skew angle (perpendicular). They would like to assess the potential change in expected average crash frequency. Objectives This example is focused on determining the change in expected average crash frequency because of realigning an intersection approach. The problem shows how to apply a CMF from the HSM Part D. After reviewing this example, the user should be able to: • Understand what input data are required to apply the HSM Part D procedures • Calculate the change in expected average crash frequency using the HSM

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 119 • Understand how to reasonably interpret the results from an HSM analysis, and how these results can be used to support a particular decision Data Requirements The existing skew angle is 40 degrees. The expected average crash frequency for this site is 12 crashes per year. The applicable CMF is calculated using HSM Equation 14-3 (HSM p. 14-19). The CMF applies to total intersection crashes: • Expected average crash frequency: 12 crashes per year • Existing skew angle: 40 degrees Analysis The first step in the analysis is calculating the CMF for the existing condition. The skew angle is 40 degrees. The skew angle CMF is calculated using the following equation: 𝐶𝑀𝐹 = . ×( . . × ) + 1 𝐶𝑀𝐹 = 0.053 × 40(1.43 + 0.053 × 40) + 1 = 1.60 Then calculate the CMF for the after condition. The skew angle is 0 degrees: 𝐶𝑀𝐹 = 0.053 × 𝑠𝑘𝑒𝑤(1.43 + 0.053 × 𝑠𝑘𝑒𝑤) + 1 = 1.00 𝐶𝑀𝐹 = 0.053 × 0(1.43 + 0.053 × 0) + 1 = 1.00 The treatment CMF is then calculated by dividing the CMF for the after condition by the CMF for the existing condition: 𝐶𝑀𝐹 = 1.001.60 = 0.63 This result is used to quantify the difference between the existing condition and the change after the application of the treatment. The CMF treatment is applied to the expected crash frequency without the treatment: 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑐𝑟𝑎𝑠ℎ𝑒𝑠 𝑎𝑓𝑡𝑒𝑟 𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 = 0.63 × 12 = 7.5 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Lastly, the change between the expected average crash frequency with and without treatment is calculated: 𝐶ℎ𝑎𝑛𝑔𝑒 𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑐𝑟𝑎𝑠ℎ𝑒𝑠 = 12.0 − 7.5 = 4.5 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑟𝑒𝑑𝑢𝑐𝑒𝑑 Results and Discussion The example shows how to compute the change in expected average crash frequency after implementation of a treatment. The reduction in skew angle from 40 degrees to 0 degrees yielded a reduction of 4.5 crashes per year. This CMF did not have a standard error available; therefore, a confidence interval for the reduction could not be calculated.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 120 3.3.4 Example Problem 6: Deceleration Ramp Lengthening Introduction As part of a rehabilitation project, a local jurisdiction is considering to make improvements to an urban grade-separated diamond interchange. One of the improvements is the lengthening of an existing eastbound off-ramp deceleration lane. The current length is 450 feet, which is planned to be lengthened by 350 feet. The engineers would like to assess the change in average crash frequency by implementing this improvement. Objectives The following example is focused on determining the change in average crash frequency caused by a deceleration ramp lengthening. The problem shows how to apply a CMF from HSM Part D. After reviewing this example, the user should be able to: • Understand what input data are required for applying HSM Part D procedures • Calculate the change in crash frequency and apply standard error using the HSM • Understand how to reasonably interpret the results from an HSM analysis and how these results can be used to support a particular decision Data Requirements The existing ramp 5-year average crash frequency is 19 crashes per year. The applicable CMF can be found in HSM Table 15-4, Potential Effect of Extending Deceleration Lanes (HSM p. 15-6). The CMF applies to all collision and severity types. The desired level of confidence for this example is 95 percent: • The CMF is 0.93 • CMF standard error is 0.06 Analysis The first step in the analysis is to calculate the 95th percentile confidence interval estimation of crashes with the treatment in place by using HSM Equation 3-8 (HSM p. 3-22): 𝐶𝑜𝑛𝑓𝑖𝑑𝑒𝑛𝑐𝑒 𝐼𝑛𝑡𝑒𝑟𝑣𝑎𝑙 (𝐶𝐼%) = [𝐶𝑀𝐹 ± (𝑆𝐸 × 𝑀𝑆𝐸)] where: CMF = the crash modification factor to be applied SE = the standard error of the CMF MSE = the multiple of standard error for the desired level of confidence A low desired level of confidence yields a confidence interval of 65 to 70 percent; a medium desired level of confidence yields a confidence interval of 95 percent; and a high desired level of confidence yields a confidence interval of 99.9 percent. HSM Section 3.5.3 (HSM p. 3-19) provides details about CMFs and detailed explanation of standard errors. Then, the estimation of crashes with the treatment in place is calculated as follows: 𝐶𝑟𝑎𝑠ℎ𝑒𝑠 𝑤𝑖𝑡ℎ 𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 = [𝐶𝑀𝐹 ± (𝑆𝐸 × 2)] × 𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟 𝐶𝑟𝑎𝑠ℎ𝑒𝑠 𝑤𝑖𝑡ℎ 𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 = [0.93 ± (0.06 × 2)] × 19 = 15.39 𝑜𝑟 19.95 𝑐𝑟𝑎𝑠ℎ𝑒𝑠/𝑦𝑒𝑎𝑟

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 121 An MSE value of 2 yields a 95-percent probability that the true value is between 15.39 and 19.95 crashes per year. The change in average crash frequency is calculated as follows: 𝐿𝑜𝑤 𝐸𝑠𝑡𝑖𝑚𝑎𝑡𝑒 = 19.95 − 19.00 = 0.95 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑒 𝐻𝑖𝑔ℎ 𝐸𝑠𝑡𝑖𝑚𝑎𝑡𝑒 = 19.00 − 15.39 = 3.61 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑒 Results and Discussion The range of values suggests that lengthening the deceleration ramp by 350 feet may potentially increase, decrease, or cause no change in the average crash frequency at the study location. 3.4 HSM in Operations and Maintenance 3.4.1 Overview Agencies are responsible for providing a reasonably safe and efficient transportation system for users on their daily operations. Typical operation activities include minimizing recurring congestion, managing incidents, weather-related events, work zones, handling special events, and managing daily traffic operations of the roadway network. Typical maintenance activities include improving pavements, roadside elements, and bridge facilities. The HSM provides users data-driven and science-based methods to supplement system monitoring, identify opportunities for improvement, and assess safety impacts of operations and maintenance activities. Examples of the HSM application to improve operations include changes in signal timing, addition of passing lanes, and addition of left- and right-turn lanes. Systemic safety treatments can also be evaluated. Maintenance improvements such as signs, guardrail, and lighting upgrades and work zone closures can also be quantified by applying the HSM tools. 3.4.2 Example Problem 7: Adding Protected Left-Turn Phases Introduction An urban four-leg signalized intersection with permissive left-turn phases in all four approaches is experiencing left-turn queuing issues on the major road. The city is evaluating the addition of exclusive left-turn phases on the major road and would like to assess the change in expected average crash frequency due to this improvement. Objectives The following example is focused on determining the change in average crash frequency caused by the addition of protected left-turn phases on the major road. The problem shows how to apply a CMF from the HSM Part D. After reviewing this example, the user should be able to: • Understand what input data are required for applying HSM Part D procedures • Calculate the change in crash frequency using the HSM Part D • Understand how to reasonably interpret the results from an HSM analysis and how these results can be used to support a particular decision

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 122 Data Requirements The signalized intersection expected average crash frequency is 28 crashes per year. The intersection has four permissive left-turn phases, and the improvement considers upgrading the major movement approaches to protected left turn. Analysis The first step in the analysis is calculating the CMF for the existing condition. The permissive left-turn phase CMF is equal to 1.00 (HSM Table 14-24 [HSM p. 14-36]). The CMF for left-turn phasing is applied to each approach and multiplied together. The existing CMF for the intersection is calculated as follows: 𝐶𝑀𝐹 = 𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 × 𝐶𝑀𝐹 𝐶𝑀𝐹 = 1.00 × 1.00 × 1.00 × 1.00 = 1.00 The next step is to calculate the CMF for the after condition. The protected left-turn phase CMF is 0.94 for each protected approach (HSM Table 14-24 [HSM p. 14-36]): 𝐶𝑀𝐹 = 0.94 × 0.94 × 1.00 × 1.00 = 0.88 The treatment CMF is then calculated by dividing the future CMF by the existing CMF: 𝐶𝑀𝐹 = 0.881.00 = 0.88 This result is used to quantify the difference between the existing and future condition after the application of the treatment. The CMF treatment is applied to the expected crash frequency without the treatment: 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑐𝑟𝑎𝑠ℎ𝑒𝑠 𝑎𝑓𝑡𝑒𝑟 𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 = 0.88 × 28 = 24.7 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Lastly, the change between the expected average crash frequency with and without treatment is calculated as follows: 𝐶ℎ𝑎𝑛𝑔𝑒 𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑐𝑟𝑎𝑠ℎ𝑒𝑠 = 28.0 − 24.7 = 3.3 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑟𝑒𝑑𝑢𝑐𝑡𝑖𝑜𝑛 Results and Discussion The example shows how to compute the change in expected average crash frequency after the implementation of a treatment. The change of left-turn signal phasing from permissive to protected- only in the major road led to a reduction of 3.3 crashes per year. The CMF did not have a standard error available; therefore, a confidence interval for the reduction could not be calculated. 3.4.3 Example Problem 8: Work Zone Analysis Introduction Predicting crashes under existing and proposed conditions can be challenging if the site is very different from base conditions and the calculations may be more complicated if the site is under construction. There are many factors that may be considered to predict crashes in a work zone. These may include work zone length, work zone duration, type of construction work (reconstruction, rehabilitation, etc), contracting limitations (area available for contractor work operations), construction season (winter, spring), available right-of-way (available shoulder width), barrier type (drums, concrete barriers). The HSM simplifies the potential factors by focusing on the work zone length and duration.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 123 The HSM provides two work zone CMFs that take into account work zone length and duration. Although more information is needed for a comprehensive work design, the following example is intended to illustrate the use of such CMFs and illustrate how maintenance of traffic (MOT) designers can obtain additional information to make informed decisions during the work zone design process Example A 5-mile rural freeway corridor is scheduled to undergo rehabilitation. The MOT team is designing the work zone layout and assessing the likely change in crash frequency between three work zone length and duration scenarios. The scenarios under consideration include constructing the overlay using one 5- mile work zone in 60 days; the second scenario involves two 2.5-mile work zones with a total duration of 90 days; and the third scenario involves five 1-mile work zone sections with a total duration of 120 days. Objectives This example is focused on determining the change in average crash frequency because of the increase of work zone length and duration. The example shows how to apply a CMF from HSM Part D. After reviewing this example, the user should be able to: • Understand what input data are required for applying HSM Part D procedures. • Calculate the change in crash frequency using these CMFs. • Understand how to interpret the results from CMF calculations to support a particular decision. Data Requirements The sensitivity analysis scenarios include work zones of 5-, 2.5- and 1-mile lengths with durations of 60, 90, and 120 days, respectively. The corridor expected average crash frequency under base conditions is 4.0 crashes per year. Analysis The first step in the analysis is calculating the CMF for the increase in work zone length using HSM Equation 16-2 (HSM p. 16-7). 𝐶𝑀𝐹 = 1 + % × . 𝐶𝑀𝐹 = 1 + 880 × 0.67100 = 6.90 𝐶𝑀𝐹 . = 1 + 390 × 0.67100 = 3.61 𝐶𝑀𝐹 = 1 + 96 × 0.67100 = 1.64 The next step is to calculate the CMF for increase in work zone duration using HSM Equation 16-1 (HSM p. 16-6): 𝐶𝑀𝐹 = 1 + % × . 𝐶𝑀𝐹 = 1 + 275 × 1.11100 = 4.05

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 124 𝐶𝑀𝐹 = 1 + 463 × 1.11100 = 6.13 𝐶𝑀𝐹 = 1 + 650 × 1.11100 = 8.22 Next, calculate the combined effect of work zone length and duration under the proposed work zone condition: 𝐶𝑀𝐹 = 𝐶𝑀𝐹 × 𝐶𝑀𝐹 𝐶𝑀𝐹 = 6.90 × 4.05 = 27.96 𝐶𝑀𝐹 = 3.61 × 6.13 = 22.17 𝐶𝑀𝐹 = 1.64 × 8.22 = 13.50 This result is used to quantify the expected number of crashes under the proposed work zone scenario: 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑐𝑟𝑎𝑠ℎ𝑒𝑠 𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜 1 | = 27.96 × 4 = 111.8 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑐𝑟𝑎𝑠ℎ𝑒𝑠 𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜 2 . | = 22.17 × 4 = 88.7 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑐𝑟𝑎𝑠ℎ𝑒𝑠 𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜 3 | = 13.5 × 4 = 54.0 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 Lastly, the change in expected crash frequency under the proposed work zone scheme is calculated as follows: 𝐶ℎ𝑎𝑛𝑔𝑒 𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑐𝑟𝑎𝑠ℎ𝑒𝑠 𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜 1 = 111.8 − 4.0 = 107.8 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑒 𝐶ℎ𝑎𝑛𝑔𝑒 𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑐𝑟𝑎𝑠ℎ𝑒𝑠 𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜 2 = 88.7 − 4.0 = 84.7 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑒 𝐶ℎ𝑎𝑛𝑔𝑒 𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑐𝑟𝑎𝑠ℎ𝑒𝑠 𝑆𝑐𝑒𝑛𝑎𝑟𝑖𝑜 3 = 54.0 − 4.0 = 50.0 𝑐𝑟𝑎𝑠ℎ𝑒𝑠𝑦𝑒𝑎𝑟 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑒 Results and Discussion The work zone example shows how to compute the change in expected average crash frequency for three proposed work zone scenarios. The results indicate that the varying conditions of the scenarios are likely to increase the crash frequency significantly. From the results, Scenario 1 with the longest work zone length and shorter duration yields the highest CMFlength and a lowest CMFduration. However, when combined, the total CMF for Scenario 1 (5-mile, 60 day) yields highest annual average crash frequency with respect to the base condition among other scenarios. Similarly, Scenario 3 (1-mile, 120- day) yields lowest annual average crash frequency with respect to the base condition among other scenarios.

SECTION 4 – PART D: CMF APPLICATIONS GUIDANCE 125 As a result, the combined effect of CMF related to length and duration yields to the lowest increase in the expected annual average crashes. The standard errors for these CMFs were not available; therefore, a confidence interval in the estimate could not be calculated. Figure 27: CMF related to Length and Duration of Work zone with Expected Annual Average Crash Increase for Work Zone Scenarios Although Scenario 1 having 5-mile of length (farthest from the baseline CMF for length) and 60-day of duration (closest from the baseline CMF for duration), this option yields the highest expected annual average crash increase (108 crashes per year from the baseline). On the contrary, scenario 3 having 1- mile of length (closest from the baseline CMF for length) and 120-days of duration (farthest from baseline CMF for duration) yields the lowest expected annual average crash increase (50 crashes per year) among other scenarios. With the respect to baseline work zone conditions (CMF of 1.0 for length and duration 1.0), scenario 3 yields the lowest expected annual average crash increase among all other scenarios.

Next: 4 HSM Part D: CMF Applications Guidance »
Highway Safety Manual User Guide Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

The Highway Safety Manual can be used to identify sites with the most potential for crash frequency or severity reduction; identify contributing factors to crashes and mitigation measures; and estimate the potential crash frequency and severity on highway networks, among other uses.

The TRB National Cooperative Highway Research Program's NCHRP Web-Only Document 323: Highway Safety Manual User Guide is a user-friendly, companion and reference document that helps safety analysts use the Highway Safety Manual.

Supplemental to the document are three Peer Exchange Reports from Irvine, California; Baltimore, Maryland; and Nashville, Tennessee; and three Tech Briefings (1, 2, and 3).

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!