4 SAFETY 4.1 Purpose and Overview The purpose of this document is to provide guidance on safety analysis procedures for specific transportation planning and project development applications with a safety component. All planning and project development efforts need to be individually scoped as there are a number of different tools and techniques that can be applied. APM Section 4.1.2 identifies the recommended safety analysis procedures for common planning and project development applications. The primary goal of any safety analysis presented in this chapter is to promote a proactive approach to reducing the frequency of fatal and serious injury (Injury-A) crashes. This is consistent with the Oregon Transportation Plan (OTP) that states “it is the policy of the State of Oregon to continually improve the safety and security of all modes and transportation facilities for system users including operators, passengers, pedestrians, recipients of goods and services, and property owners.” The Oregon Transportation Safety Action Plan implements the OTP policy. The first edition of the Highway Safety Manual (HSM) provides the technical foundation for many of the procedures discussed in this chapter. However, this chapter does not replicate the entire guidance of the HSM, and the reader is encouraged to consult the HSM directly where appropriate. The HSM is published by the American Association of State Highway and Transportation Officials (AASHTO) with support from the Federal Highway Administration (FHWA), the Institute of Transportation Engineers (ITE), and the Transportation Research Board (TRB) Highway Safety Performance Committee (ANB25). The HSM is a national guide—and the first of its kind—providing science-based methods, procedures, and measures that integrate quantitative estimates of crash frequency and severity into roadway planning, evaluation, and project development. Prior to the HSM, crash analysis for planning and project development was typically limited to simple evaluations of crash data and somewhat subjective analysis. Evaluations of future safety performance were primarily limited to meeting design standards, with few options for comparing alternatives. In contrast, the tools in the HSM allow safety to become a meaningful performance measure that can be implemented at any stage of the transportation decision-making process. HSM methodologies are provided to assist agencies in their effort to integrate safety into their decision-making processes, but are not intended to be a substitute for the exercise of sound engineering judgment. No standard of conduct or any duty toward the public or any person shall be created or imposed by the publication and use or nonuse of the HSM. The HSM does not supersede publications such as the MUTCD, the AASHTO Green Book, or other AASHTO and agency guidelines, manuals and policies. Analysis Procedure Manual Version 2 4-1 Last Updated 09/2019 As stated in the HSM, it is neither intended to be, nor does it establish, a legal standard of care for users or professionals. HSM screening tools provide a robust methodology for objectively evaluating historical crash data based on frequency, severity, collision type, and other crash characteristics. The screening tools identify locations with the highest potential for reducing the frequency and severity of crashes and, by identifying factors contributing to the crashes, help choose effective potential countermeasures. Screening tools are discussed in APM Section 4.3. The HSM Predictive Method is the first comprehensive model for estimating the frequency and severity of crashes based on traffic, roadway, and roadside characteristics. Predictive analysis can be applied to quantify the safety impact of design alternatives and forecast scenarios, using the understandable language of crash frequency and severity. Predictive analysis can also be applied in conjunction with historical data analysis to overcome statistical limitations inherent in historical data analysis. Predictive tools are discussed in APM Section 4.4. The HSM framework also provides local agencies a methodology to expand on the foundation of the HSM by calibrating to local conditions and developing custom safety performance functions (SPFs). ODOT’s webpage on the HSM includes information on completed and future research projects. The APM does not address ODOT highway safety program procedures or traffic operations-level safety analysis. This includes road safety audits, collision diagrams, detailed safety investigations, and benefit-cost analyses. Contact the Traffic-Roadway Section for procedures related to those programs. 4.1.1 Statewide Crash Rate References Statewide average crash rates are used in the critical crash rate analysis method and are useful resources for informal discussions of crash frequency. The Oregon State Highway Crash Rate Tables are published annually by the ODOT CAR Unit. Crash Rate Table II shows statewide average crash rates for each of the last five years, by urban and rural area and by roadway classifications for state highways. These crash rates are based on overall crash frequency and total vehicle miles traveled on mainline state highways. Federal functional classifications can be found on the ODOT Federal Functional Classification (FC) webpage. Exhibit 4-1 shows intersection crash rates by land type and traffic control, based on a 2011 assessment of data from 2003-2007. The crash rates here are based only on crashes that occurred at an intersection or because of an intersection and are given as a rate per million vehicles entering the intersection [million entering vehicles (MEV)]. Analysis Procedure Manual Version 2 4-2 Last Updated 09/2019 Intersection crash rates also need to be compared to the published statewide 90th percentile intersection crash rates in Exhibit 4-1. Any rates close to or over the 90th percentile rates need to be flagged for further analysis. The intersection crash rate is calculated by the following formula: 6 𝐴𝐴𝐼𝐼𝐼𝐼𝐴𝐴𝐶𝐶𝐴𝐴 𝑁𝑁𝐴𝐴𝑁𝑁𝑁𝑁𝐼𝐼𝐼𝐼 𝐼𝐼𝑜𝑜 𝐶𝐶𝐼𝐼𝐶𝐶𝐼𝐼ℎ𝐼𝐼𝐼𝐼 𝑥𝑥 10 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐶𝐶𝐼𝐼𝐶𝐶𝐼𝐼ℎ 𝑅𝑅𝐶𝐶𝐼𝐼𝐼𝐼 𝑝𝑝𝐼𝐼𝐼𝐼 𝑀𝑀𝑀𝑀𝑀𝑀 = The values shown in Exhibit 4-1 represent the 90th per(c𝐴𝐴en𝐴𝐴t𝐴𝐴il𝐴𝐴e )c𝑥𝑥r a(s3h6 5ra 𝑑𝑑te𝐶𝐶s𝑑𝑑 f𝐼𝐼r/o𝑑𝑑m𝐼𝐼𝐶𝐶 a𝐼𝐼 )study of 500 intersections in Oregon. The crash rates are grouped by rural/urban, signalized/unsignalized, and three-leg/four-leg intersections. Intersections with crash rates that exceed the 90th percentile values shown in the table should be flagged for further analysis. For more information on crash rates and using this table, see Section 4.3.4 Critical Crash Rate. Exhibit 4-1: Intersection Crash Rates per MEV by Land Type and Traffic Control Rural Urban 3SG 3ST 4SG 4ST 3SG 3ST 4SG 4ST No. of Intersections 7 115 20 60 55 77 106 60 Mean Crash Rate 0.226 0.196 0.324 0.434 0.275 0.131 0.477 0.198 Median Crash Rate 0.163 0.092 0.320 0.267 0.252 0.105 0.420 0.145 Standard Deviation 0.185 0.314 0.223 0.534 0.155 0.121 0.273 0.176 Coefficient of Variation 0.819 1.602 0.688 1.230 0.564 0.924 0.572 0.889 90th Percentile Rate 0.464 0.475 0.579 1.080 0.509 0.293 0.860 0.408 Source: Assessment of Statewide Intersection Safety Performance, FHWA-OR-RD-18, Portland State University and Oregon State University, June 2011, Table 4.1, p. 47. Note: Traffic control types include 3SG (three-leg signalized), 3ST (three-leg minor stop-control), 4SG (four-leg signalized), 4ST (four-leg minor stop-control). For intersections other than the configurations shown in Exhibit 4-1, there are usually too few locations with that intersection configuration to provide statewide statistics. There are some stop controlled intersection configurations that could be approximated as indicated in Exhibit 4-2 and Exhibit 4-3 below. Any other intersection configurations not in Exhibit 4-1, Exhibit 4-2, or Exhibit 4-3 should by default be flagged for further analysis, since the unusual configuration is likely to warrant a closer look at the crashes. Analysis Procedure Manual Version 2 4-3 Last Updated 09/2019 Exhibit 4-2: 3 Legged Stop Control, with Driveway(s) into Intersection 3 legged stop control, with driveway(s) into intersection at what would be a fourth leg location. Crash rate higher than the reference rate could indicate that the driveway volumes are affecting safe operation of the intersection. If the driveway volume is low compared to the opposite minor leg, the 3ST reference crash rate could be applied. Example: Rogue River, E. Main at Broadway St. If the driveway volume is high compared to the opposite minor leg, the 4ST reference crash rate could be applied. Example: Bend, Cooley Rd at NE Hennell Rd, Bend Rogue River, E. Main at Broadway St. Bend, Cooley Rd at NE Hennell Rd Analysis Procedure Manual Version 2 4-4 Last Updated 09/2019 Exhibit 4-3: 4 Legged Intersection, 3 Way Stop 4 legged intersection 3 way stop This configuration could apply the 4ST reference crash rate, since the minor legs have left turn conflicts similar to the 4ST. Example: Salem, Argyle at Missouri Salem, Argyle at Missouri Street view, looking west. 4.1.2 Tools and Procedures by Application Type Safety analysis tools should be specified during the scoping process. Model scoping language is provided in APM Chapter 2. In order to facilitate scoping of tools, this section describes the recommended safety analysis tools and procedures for common application types as shown in Exhibit 4-4. Safety analysis tools are identified in order of increasing level of effort in the chart going from left to right. Plan or project level of detail is shown in increasing level of detail in the chart going from top to bottom. Best practice/recommended methods are shown with closed circles, while open circles identify optional or supplemental methods. Analysis Procedure Manual Version 2 4-5 Last Updated 09/2019 Exhibit 4-4: Applicability of Safety Analysis Tools by Plan or Project Type The tools and procedures recommended here describe the crash analysis appropriate for the scope and scale of typical applications. A balanced approach to safety analysis should also consider queues, sight distances, safe and convenient crossing opportunities, and other safety- related techniques where appropriate for the application context. • All Applications: Safety Priority Index System (SPIS) – Identify top 5% or 10% locations from the o most recent three (3) SPIS Site listings • RTPs, Transportation System Plans (TSPs) and High-Level Corridor Plans: Critical Crash Rate – required o Excess Proportion of Specific Crash Types – required o Crash rate comparison – minimum requirement when other methods can’t be o applied. Compare intersection crash rates to the 90th percentile crash rates (Exhibit 4-1) and segment crash rates to Table II in the CARS crash rates tables. PLANSAFE – optional, system-wide predictive method. Recommended if there o are regional transportation network changes proposed or if there is an expectation of significant demographic changes. Crash Modification Factors (CMFs) – Optional. Use to estimate potential crash o reduction of alternatives. Analysis Procedure Manual Version 2 4-6 Last Updated 09/2019 • Multimodal Mixed-Use Areas (MMAs) MMAs are governed by OARs which require that public safety is not o compromised. The primary safety analysis methods recommended for MMAs are the OAR requirements:  If the area has a crash rate above statewide averages. Crash rates are generally considered on the mainline and the crossroad.  If the area includes a top 10% SPIS site  If existing or future traffic queues will create a safety concern on the mainline highway exit The new safety analysis options could provide additional value for MMA o analysis. The Critical Crash Rate could be a complement to crash rates as described in the o OARs. Excess Proportion of Specific Crash Types can be used to identify crash patterns o and multimodal safety concerns. The HSM Predictive Method can also be used to evaluate an MMA location and o identify mitigation needs before an MMA is granted, or could be incorporated into guidelines for evaluating plan amendments within MMAs. • Facility Plans/Refinement Plans: HSM Predictive Method – required if valid model exists o  Excess Expected Crash Frequency – use for existing conditions evaluation  Net Change in Predicted Crash Frequency – use for alternatives evaluation If no valid model exists: o  Use Critical Crash Rate, Excess Proportion of Specific Crash Types, and statewide crash rate comparisons for existing conditions  Use CMFs for alternatives evaluation • Development Review: HSM Predictive Method – recommended if valid model exists o  Excess Expected Crash Frequency – use for existing conditions evaluation  Net Change in Predicted Crash Frequency – use for alternatives evaluation If no valid model exists: o  Use Critical Crash Rate, Excess Proportion of Specific Crash Types, and statewide crash rate comparisons for existing conditions  Use CMFs for alternatives evaluation • Project Development / National Environmental Policy Act (NEPA) Work: HSM Predictive Method – required if valid model exists o  Excess Expected Crash Frequency – use for existing conditions evaluation  Net Change in Predicted Crash Frequency – use for alternatives evaluation If no valid model exists: o  Use Critical Crash Rate, Excess Proportion of Specific Crash Types, and statewide crash rate comparisons for existing conditions  Use CMFs for alternatives evaluation Analysis Procedure Manual Version 2 4-7 Last Updated 09/2019 • Countermeasure Development: Using site characteristics and analysis results, identify contributing factors o Document Crash Modification Factor(s) o Perform additional geometric safety assessments, as applicable: o  Intersection functional area  Sight distance  Conflict points  Access management 4.2 Crash Data Crashes are used as the basis of safety analysis presented in this chapter. Crash frequency (number of crashes per year) and crash severity (rating based on most severe injury sustained in a crash) are fundamental indicators of the “safety” of a roadway. Observed crash data provides information for describing and analyzing crash frequency and severity for historical time periods. Predictive methods proactively estimate crash frequency and severity for situations where observed crash data are not available, such as for future conditions. This chapter focuses on analysis of “objective” safety, which is based on quantitative measures that are independent of the observer. Objective safety is not directly experienced by a traveler, unless he or she is involved in a crash. In contrast, analysis of “subjective” safety involves the perception of how safe a person feels while using the transportation system. An assessment of subjective safety for the same location may vary between observers, and techniques for assessing the subjective safety of a location are not covered in this chapter. Subjective safety is an important component of many design and policy decisions. A person’s choice of mode and route is strongly affected by how safe and comfortable the mode and route feels. 4.2.1 ODOT Crash Data Sources The ODOT Crash Analysis and Reporting Unit (ODOT CAR) maintains a statewide crash record database that includes all reported crashes involving a motor vehicle on public roads. These data are collected by the Department of Motor Vehicles from police and driver reports then provided to ODOT CAR for quality assurance, standardization, and distribution. ODOT CAR produces a variety of publications annually that summarize crashes throughout the state of Oregon. Crash data only include reported crashes to the Department of Motor Vehicles (DMV). Many crashes are not reported because they fall under the $2,500 reporting threshold or are just not reported (i.e., single vehicle incident, or on tribal lands). Errors can still occur in the coded crash records so it is important to carefully review and note any anomalies. Reporting errors on officer reports and DMV forms and officer reports such as crash location are common and while the crash reporting technicians attempt to reconcile them, sometimes data are not available or is incomplete. This is why sometimes perceived crash issues from local residents differ from available crash data. See Chapter 3 for more information. Analysis Procedure Manual Version 2 4-8 Last Updated 09/2019 The Crash Data System provides analysts access to detailed crash reports for a custom study area and time period. The Crash Data System can be accessed online on the Internet through ODOT’s Unified Access Gateway and ODOT’s Intranet site. The Crash Data System provides tools for querying crash data by jurisdiction, location, and time. Crash data formatting and querying is different for state highways, city streets (non-state roads within city limits), and county roads (non-state roads outside city limits). Crash data can be downloaded in print-formatted or spreadsheet-formatted reports. Report download options include: • Summary by Year CDS150: A general summary of crashes for the queried location, displayed by year, collision type, and generalized severity (fatal, nonfatal injury, property damage only). • Crash Location CDS390: A detail report with a single line of data for each crash, including location, date, collision type, injury severity, and contributing factors. • Comprehensive (PRC) CDS380: A detail report with at least three lines of data for each crash, including a row for every vehicle and participant in the crash. Summary includes location, date, collision type, injury severity, contributing factors, and more. Data extracts are also available, providing unformatted full access to records for every crash, vehicle, and participant individually. Data extracts are available as a comma-delineated text document or as an Access database. The Access database includes code definitions and pre- defined report queries, making it a valuable resource for analysts familiar with database software. The Comprehensive (PRC) CDS380 report or data extracts are recommended for use with the analysis procedures listed here, as these are the only formats that include the full injury severity scale (KABCO) and identify crashes involving pedestrians or bicyclists. The summary reports identify crashes involving pedestrians, but not bicyclists, and only include severity as fatal, nonfatal injury, or property damage. Summary reports may not include sufficient information for all analysis methods described in this chapter. The crash data reports are heavily code-based and use of the Statewide Crash Data System Motor Vehicle Traffic Crash Analysis and Code Manual is required for a full understanding of the crash data. This manual and the online help documents provide additional important information about the crash data and reports and are available through the ODOT CAR Publications page or from within the Crash Data System. Crash data are geocoded with latitude and longitude coordinate values. This allows for easy display on a map using a Geographic Information System (GIS) such as ArcMap, QGIS, or other online tools. If crash data are mapped using coordinate values, care should be taken to identify any records with the “Unlocatable_Flag” indicating that the coordinates are nonspecific default values. Caution should be exercised when identifying actual crash locations from reported data. Crashes may be reported at the nearest integer milepoint or intersection even if they occurred hundreds of Analysis Procedure Manual Version 2 4-9 Last Updated 09/2019 feet away. Crash data should be checked for discrepancies, such as where a crash occurred on a curve but where the reported location is a straightaway section. 4.2.2 Crash Characteristics, Trends, and Patterns Crash analysis involves identifying trends and patterns on facilities. Analyzed crash types or severities may be all crashes or a more specific subset of crashes, such as fatal and serious injury crashes. These trends can then be used to identify applicable countermeasures for future mitigation. For guidance on performing a detailed on-site investigation and diagnosis of a safety trend, refer to the ODOT Safety Investigations Manual. ODOT crash data includes many characteristics that can be used to identify trends and patterns. The analysis methods in this chapter primarily use the following characteristics: • Crash Location: Geographical crash location is described by milepoint, distance to nearest intersection, and latitude and longitude coordinates. • Intersection-Related: Crashes are identified as located at an intersection or related to the functioning of an intersection. • Driveway-Related: Crashes are identified that are related to the use of a driveway. • Severity: Crash severity is equal to the most serious injury sustained by anyone involved in the crash, based on the on-scene assessment (but which may not align with final medical determination of injuries). Severity is ranked on the KABCO scale: K – Fatal injury, an injury that results in death o A – Incapacitating injury, a nonfatal injury that prevents the person from walking, o driving, or doing activities they were capable of before the injury B – Non-incapacitating evident injury, an injury that is evident to observers at the o scene of the crash C – Possible Injury, an injury or claim of an injury that is not evident to observers o at the scene of the crash O – No injury, also described as Property Damage Only (PDO) o • Collision Type: This field describes the intended movements of the vehicle(s) at the time of collision. Crashes are coded as one of the following collision types: Angle – Vehicles collided while traveling on crossing or perpendicular paths, o such as would occur if a vehicle ran a red light and crashed into a vehicle traveling on the crossing roadway Head-On – Vehicles collided while traveling in opposite directions, their forward o movement impeded while attempting to occupy a location simultaneously Rear-End –Vehicles collided while traveling in the same direction, with one o vehicle hitting the rear end of the second vehicle Sideswipe-Meeting –Vehicles collided while traveling in opposite directions, with o the side of at least one vehicle involved Sideswipe-Overtaking – Vehicles collided while traveling in the same direction, o with the side of at least one vehicle involved Turning Movement – Collision involved one or more vehicles turning, originally o traveling on parallel paths Analysis Procedure Manual Version 2 4-10 Last Updated 09/2019