Developing a Guide for Quantitative Approaches to Systemic Safety Analysis (2020)

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.

1 Introduction 1.1 Background Highway agencies have traditionally managed the safety improvement process by identifying and correcting high-crash locations (“hot-spots”), where concentrations of crashes and, often, patterns of crashes of similar types, were found. While this approach can identify locations in need of improvement, highway safety managers have found that the search for high-crash locations often focuses safety investments on locations where multiple crashes in a one- to three- year period occurred due to random chance alone. Many such short-term crash patterns would not recur, even if no improvement were made. The statistical phenomenon in which a randomly high number of crashes is typically followed by a lower number of crashes in the next time period is known as regression to the mean. As an alternative (or supplement) to the hot-spot approach, the systemic safety management approach is intended to help agencies base safety investment decisions on estimates of long-term average crash frequencies rather than short-term random fluctuations in crash frequencies. The systemic safety management approach is used to program implementation of proven safety treatments, primarily low cost, across a large number of sites to reduce crash potential using crash prediction models or rating systems based on roadway features correlated with particular severe crash types. While systemic methods do not focus on the crash history at specific sites, methods can consider crash history data in an appropriate way so that potentially valuable information represented by actual crash histories is not ignored. Systemic safety management approaches take a systemwide view of safety improvement needs and, when used in conjunction with benefit-cost analyses, are specifically intended to distinguish sites where safety improvements are needed and economically justified from sites where safety improvements are not needed or are not economically justified. Safety improvements are considered economically justified when the present value of the safety benefits from the improvement exceeds the improvement cost; the greater the amount by which the benefits exceed the cost, the greater the economic justification for the improvement. Thus, benefit-cost analyses, which can establish the economic justification for safety improvements, provide a key complement to systemic safety analyses and can provide assurance that proposed safety improvements will provide safety benefits that exceed the associated implementation and maintenance costs. Benefit-cost analysis can be applied in conjunction with any approach to systemic analysis. Systemic safety management approaches, in conjunction with benefit-cost analyses, can maximize the systemwide safety benefits derived from any given level of safety investment by focusing investments at the locations with the most potential for safety improvement. Systemic safety analysis methods that incorporate benefit-cost analyses, by their nature, emphasize low- cost safety improvements. Higher-cost safety improvements are not likely to be cost effective except at a limited number of sites that have high long-term crash frequencies and/or severities. Low-cost safety improvements are more likely to be cost effective at a broad variety of sites throughout a highway network.

2 As part of this research, the research team developed a comprehensive guidance document that defines quantitative approaches to systemic safety analysis; contrasts systemic safety management approaches to traditional crash-history-based and policy-based safety management approaches; describe the tools available to highway agencies to implement systemic safety analysis; and defines the capabilities, advantages, disadvantages, and data requirements for each tool. The first edition on the American Association of State Highway and Transportation Officials (AASHTO) Highway Safety Manual (HSM) (AASHTO, 2010) does not explicitly acknowledge systemic safety analysis, although many elements of systemic safety analysis are discussed. HSM Part B presents a six-step safety management process that can potentially encompass both crash-history-based and systemic safety management approaches. The distinction between crash-history-based safety analysis methods and systemic safety analysis methods will ideally be made explicit in the second edition of the HSM. HSM Part C procedures make use of crash prediction models but are not considered systemic safety management approaches because the data requirements are generally too extensive for systemwide application. The HSM Part C methodology was developed as a project design tool, not as a project identification or selection tool, since the data needed for the methodology are likely to be available during the project development process for individual projects and not necessarily available for all sites systemwide. This report is intended as a supplement to the comprehensive guidance document developed as part of this research. This report presents background material used to develop the comprehensive guidance document, presents the research approach, and provides recommendations for implementation of the research findings and products as an appendix. The remainder of this section describes the three primary approaches to safety management, presents the research objectives and scope, and provides an overview of the research approach. The remainder of this report is organized as follows:  Section 2 summarizes how agencies in the United States and abroad have implemented systemic safety management approaches within their agencies.  Section 3 summarizes the current state of practice of systemic safety analysis based on a survey conducted early in the project that targeted state, county, and local agencies.  Section 4 summarizes information gathered from state and local/county agencies the research team visited and used to develop the case studies and other sections of the comprehensive guidance document on quantitative approaches to systemic safety analysis.  Section 5 presents conclusions and future research needs based on the research results.  Section 6 provides citations for the resources referenced in this document.  Appendix A provides recommendations on how to best put the research findings/products into practice; identifies possible institutions that might take leadership in applying the research findings/products; identifies issues affecting potential implementation of the findings/products and recommends possible actions to address these issues; and recommends methods of identifying and measuring the impacts associated with implementation of the findings/products.

3 1.2 Summary of Primary Safety Management Approaches The three primary approaches to highway safety management include crash-history-based, systemic, and policy-based safety management approaches. These approaches vary in terms of the types, quantity, and quality of data required to carry them out; the types of treatments considered for application; and the types of crashes or safety concerns the transportation agency may address. To understand the systemic safety management approach, it is helpful to compare and contrast it to the crash-history-based and policy-based approaches. All three approaches are useful, and many agencies use a combination of approaches to reduce crashes on their network. The general methodologies of these three safety management approaches are described below along with potential advantages and disadvantages of the approaches. Crash-History-Based Safety Management Approach The purpose of the crash-history-based safety management approach, sometimes referred to as a “black-spot” or “hot-spot” analysis, is to identify locations on the system where a high frequency or rate of crashes has occurred and to improve those sites to remedy the situation. Implementing the crash-history-based safety management approach requires high-quality crash data with accurate location data throughout the roadway network. Agencies typically utilize the six-step roadway safety management process when implementing a crash-history-based safety management approach. The primary factors that distinguish the crash-history-based safety management approach from the systemic and policy-based safety management approaches are conceptual approaches to network screening and diagnosis and countermeasure selection. With the crash-history-based approach, sites are ranked for potential safety improvement based on their overall crash experience (i.e., all crash types combined). Crash experience may be estimated based on traditional performance measures such as observed crash frequencies, observed crash rates, or equivalent property damage only (EPDO) observed crash frequencies or other performance measures considered more statistically reliable as described in the HSM such as level of service of safety (LOSS) or expected and excess crash frequencies. For each performance measure, the site-specific information or data are used to estimate the overall crash experience at a site. Sites with higher crash experience based on the selected performance measure are considered to have higher potential for safety improvements, while sites with lower crash experience are considered to have less potential for safety improvement. With the crash-history-based safety management approach, the goal is to reduce crash patterns of interest that occur with a high frequency at individual locations, and to do so at sites with the highest overall crash experience. In most cases, it does not matter that at one site rear-end crashes occur with high frequency, at another site angle crashes occur with high frequency, and at another site head-on crashes occur with high frequency. It simply matters that whatever performance measure is used, sites where the overall crash experience is higher are given a higher priority of potential safety improvement compared to sites with lower overall crash experience. In some cases, though, the crash-history-based safety management approach is also used to identify sites with a high frequency of target crashes. With most performance measures, the crash-history-based safety management approach is considered reactive in nature, as individual sites are identified for potential safety improvement only after having a documented crash history substantial enough to make them rank higher than

4 other sites on the prioritized list. Sites that rise to the top of the priority list for potential safety improvement tend to be located on urban corridors and at urban intersections where traffic volumes are highest. This can lead to a disproportionally low safety investment in rural locations where crashes tend to be more dispersed around the network. After the network screening step identifies priority locations based on overall crash experience, the diagnostic process is performed on a site-by-site basis. Crash data are reviewed at individual sites to identify trends in crash patterns and types of crashes to be remedied at each location. Some sites may have a documented history of rear-end crashes, while other sites may have a history of angle crashes or head-on crashes. The types of crashes that occur with high frequency at an individual site may not be particularly important during network screening as treatments are tailored to remedy the crash patterns of interest at the individual locations. In many cases, a full range of infrastructure treatments, from low- to high-cost, are considered for potential implementation to address different crash types. Consequently, the diagnosis and countermeasure selection process for the crash-history-based safety management approach is typically much more involved and time consuming than for the other safety management approaches. Economic analyses are then performed at the site level, considering the potential countermeasures identified in the previous step. The costs of recommended countermeasures for the site are compared to the potential economic savings due to crashes prevented by the countermeasures. Results of the economic analysis can be used to prioritize sites for safety investment in such a way that the agency can expect the highest possible benefit from their available safety budget. With the crash-history-based analysis approach, a safety effectiveness evaluation is usually performed using a simple before-after analysis approach to compare the number of crashes before improvement (for X number of years) to the number of crashes that occur after implementation of the countermeasure (for X number of years). The difference between these two values indicates the effect of the countermeasure on safety performance. A safety effectiveness evaluation may also be performed using other approaches such as a before-after study with comparison group or traffic volume correction to address some of the limitations of the simple, before-after analysis approach, an Empirical Bayes (EB) before-after study approach, and other approaches. The benefits of a crash-history-based safety management approach include the ability to:  Focus funds where there is a documented crash history.  Identify treatments that address the specific crash patterns at each site.  Tailor treatments to the specific characteristics of the locations.  Address a wide range of safety conditions and tradeoffs using a quantitative and logical process. Potential limitations or disadvantages associated with the crash-history-based safety management approach include:  Crashes must occur at a site before an improvement is made.

5  Safety improvements may be made at sites to remedy specific crash types that may not occur again, even if no improvements are made.  Implementation of higher-cost safety improvements at a limited number of sites may not effectively reduce crash frequency across the network (Gross et al., 2016).  Crash types that occur frequently but are dispersed across the network may not be effectively addressed (Gross et al., 2016). Systemic Safety Management Approach The purpose of the systemic safety management approach is to be more proactive in programming safety improvements and to address specific crash types not well suited for remedy using a crash-history-based safety management approach by widely implementing primarily low- cost countermeasures. The systemic safety management approach uses safety performance measures related to expected future crashes (such as expected crash frequencies or the presence of crash contributing factors) for network screening and project site prioritization. With the systemic safety management approach, one of the intended purposes is to address crash types not well-suited for remedy using a crash-history-based safety management approach. Crash types that occur with high frequency across the roadway network but not concentrated at individual locations (i.e., crash types that occur with high frequency but are highly dispersed across a roadway network), tend to be overlooked when ranking sites using a crash-history-based safety management approach. The systemic safety management approach can be used to address such crash types by treating many sites that have potential for experiencing that type of crash with low-cost treatments. In many cases, these widely dispersed crash types are consistent with target crashes identified in an agency’s strategic highway safety plan (SHSP). Some of the specific crash types that agencies have focused on when implementing systemic safety management procedures include:  Lane departure  Rollover  Fixed object  Speed related  Younger driver involvement  Impaired driving  Pedestrians  Bicyclists  Nighttime One option for an agency to identify target crash types to address using a systemic safety management approach is to refer to a state or regional SHSP, which documents emphasis areas or target crash types for the state or region’s safety program. Referring to a state or regional SHSP to identify focus crash types for systemic safety analysis does not require the use of any specific type of data (i.e., crash or roadway inventory); however, unless the SHSP specifically states that a target crash type is to be addressed through systemic safety management, it may not be readily apparent which target crash types within a SHSP should be addressed through systemic safety management. An agency may have to infer or deduce which target crash types within the SHSP should be addressed through systemic safety. Referring to the general six-step roadway safety management process, the network screening step in a systemic safety management approach generally uses one of two methods to prioritize sites for potential safety improvement. With one method, crash prediction models or safety

6 performance functions (SPFs) are used to calculate predicted and/or expected crash frequencies of target crash types at specific sites. Predicted crash frequencies are estimated directly from the SPFs, while expected crash frequencies are calculated using statistical procedures to combine observed crash frequencies and predicted crash frequencies from SPFs. With the publication of the HSM and recent emphasis on the use of SPFs for roadway safety management, many agencies have developed their own SPFs or calibrated existing SPFs using their own crash and inventory data. The development of agency-specific SPFs has allowed agencies to implement systemic safety management approaches within their Highway Safety Improvement Program (HSIP). With systemic safety and the use of SPFs, the emphasis is on calculating predicted, expected, or excess crash frequencies for target crash types. With the second network screening method, a rating system is developed to represent crash potential at sites within the network. Crash potential is generally assessed by identifying crash contributing factors present at each site. Crash contributing factors may be identified based on published research or through a quantitative analysis of site characteristic data to determine which characteristics are overrepresented at sites where certain crash types occur. Sites identified as having the highest crash potential are given the highest priority for programming safety improvements. Similar to the policy-based safety management approach, low-cost countermeasures proven to effectively reduce crashes are generally the first treatment types considered for implementation as part of a systemic safety management approach (although higher-cost countermeasures can be considered as well). Through the use of low-cost safety improvements, more sites can be improved which can lead to a greater reduction in target crashes across the network. Types of countermeasures that agencies have implemented as part of their systemic safety management projects include: Roadway segments:  Rumble strips (both shoulder and centerline)  Cable median barrier  SafetyEdgeSM  High friction surface treatments  Enhanced pavement markings  Curve warning signs  Chevrons/delineators  Lane/shoulder widening  Speed feedback signs  Tree/clear zone removal Intersections:  Signal backplates  Crosswalk enhancements – striping, signing, rapid rectangular flashing beacons  Countdown pedestrian signals  Pedestrian refuge islands  Curb extensions  Reflective strips on sign posts  Mini-roundabouts

7  Lighting Countermeasures are chosen to remedy target crash types and in conjunction address the crash contributing factors identified for the specific crash types of interest, and are then implemented at many sites where those crash contributing factors are present, regardless of previous crash history. Typically, one, two, or three proven, low-cost countermeasures are initially identified for consideration to address a particular target crash type. With the systemic safety management approach, much of the diagnosis and countermeasure selection process is actually done ahead of time in terms of identifying crash patterns of interest and potential countermeasures. Consequently, in relation to the six-step safety management approach, the diagnosis and countermeasure selection process for systemic safety management typically involves less effort and is less time consuming in comparison to the diagnosis and countermeasure selection process for the crash-history-based safety management approach. In a systemic safety management approach, the economic evaluation is often fairly simple. Since specific countermeasures have already been identified for implementation and a prioritized list of implementation locations has been developed, agencies tend to either determine a specific rating score threshold, for which countermeasures are implemented at all sites that score at or above the threshold, or simply begin countermeasure implementation at the top of the list and work down until funding is exhausted. Generally, in a systemic safety management approach, the countermeasures chosen for implementation already have a well-documented and reliable estimate of their safety effectiveness. Therefore, a safety effectiveness evaluation is not always important to the agency, but it is also useful to inform future decision making so agencies may have interest in evaluating the success of their systemic treatment applications. Quantitative impacts of systemic treatment applications are most commonly analyzed using a trend analysis, simple before-after study method, an Empirical Bayes before-after study method, or a shift of proportions method. However, the methodological approach to evaluation depends on the type and amount of data available, the goals of the evaluation, and the agency resources available to complete the evaluation. The potential benefits of implementing a systemic safety management approach include:  This approach can be used in the absence of high-quality historical site-level crash data (Gross et al., 2016).  This approach is proactive because countermeasures can be programmed for implementation at locations that may not have a history of crashes (Gross et al., 2016). In particular, even sites with zero crash history can be prioritized for potential safety improvement using a systemic safety management approach.  The approach helps agencies broaden their traffic safety efforts and consider the potential for future crashes as well as crash history when identifying where to make safety improvements (Preston et al., 2013).  This approach provides the ability to program projects further into the future as projects can be based on the presence or absence of crash contributing factors (i.e., roadway characteristics) that do not change frequently from year to year.

8  With this approach, it may be easier to more equally distribute safety funds regionally or across jurisdictions compared to programming safety improvements based solely on a crash-history-based safety management approach.  This approach is adaptable based on available data. The potential limitations or disadvantages associated with the systemic safety management approach include:  Available software that can be used for quantitative systemic safety analyses is used on a limited basis either because it is considered expensive, data intensive, or agencies do not invest the time and resources to collect the necessary data for use with the software.  Project prioritization and the process for evaluating the benefits of a systemic safety management approach (i.e., project evaluation) are not well understood because of the lack of before crashes at improvement sites.  Staff may be reluctant to incorporate a systemic safety management approach within their safety management procedures because it deviates from the traditional crash-history- based safety management approach (i.e., hot-spot analysis).  Agencies have little guidance on how to allocate funding to support programming safety improvements identified through a combination of crash-history-based and systemic safety management approaches. Policy-Based Safety Management Approach The purpose of the policy-based safety management approach is to bring design or operational features of sites up to a specified standard or policy. The policy-based safety management approach is intended to reduce both liability and crash potential. The countermeasures implemented using this safety management approach are often those proven to effectively reduce crashes. After a countermeasure has been shown to effectively reduce crashes, agencies may develop a policy to implement the countermeasure as part of their regular design, construction, operations, and maintenance programs, including new construction and reconstruction. The agency may also choose to improve existing facilities for safety purposes only (Gross et al., 2016). A policy-based safety management approach promotes the application of low-cost countermeasures proven to effectively reduce crashes. Examples of countermeasures installed as part of a policy-based safety management program include:  Installation of retroreflective backplates on all new signal installations and signal upgrades.  Installation of shoulder rumble strips/stripes on all two-lane roads with a shoulder of sufficient width.  Installation of the SafetyEdgeSM treatment for all asphalt paving projects without curbs. In a policy-based safety management approach, detailed economic analyses are generally not performed. Once the policy for treatment implementation has been determined, the agency budgets the necessary funding into future projects as part of construction and/or maintenance

9 costs. Expected benefits may be estimated using a reliable crash modification factor (CMF) for the treatment and applying it to the number of crashes experienced in a given time period over the portion of the system to which the treatment is going to be applied. This approach provides general systemwide estimates of crash reductions that can be used to justify the expense of the treatment. If a safety effectiveness evaluation is performed, it is generally conducted over many similar sites within the network. Often, a before-after analysis will be conducted for all, or a large number, of the sites where the treatment was implemented. Alternatively, an agency may choose to evaluate how effective their systemwide implementation of a treatment was by considering the change in crash frequency or rate for a targeted crash type over the entire system after treatment implementation. The potential benefits of implementing a policy-based safety management approach include:  Countermeasures proven to effectively reduce crashes are implemented.  This approach can be used in the absence of high-quality historical site-level crash data.  This approach is proactive because countermeasures can be programmed for implementation at locations that may not have a history of crashes.  This approach is easily understood.  This approach serves to reduce liability and crash potential.  Safety improvements can be incorporated into different types of projects (e.g., new construction, reconstruction, rehabilitation, and maintenance).  Implementation costs may go down when installation is programmed into scheduled construction and maintenance projects, eliminating separate project start-up and traffic control costs. The potential limitations or disadvantages associated with the policy-based safety management approach include:  It may take years to bring the targeted facility types up to the desired standard/policy.  This approach can be perceived as increasing the total cost of projects by adding countermeasure implementation costs that may not have otherwise been included.  Resources may not be allocated as efficiently as possible because treatments may be implemented at locations with low potential to reduce crashes. 1.3 Research Objective and Scope The objectives of this research were to develop a guide and training materials to assist state departments of transportation (DOTs), metropolitan planning organizations (MPOs), local agencies, and other safety practitioners to better understand, use, and implement quantitative approaches to systemic safety analysis. The results of the research have been developed into a draft chapter for consideration in the second edition of the HSM. The final deliverables for this research include:

10  This final report, which documents the research process and findings.  A comprehensive guidance document on the implementation of quantitative approaches to systemic safety analysis (intended as a companion document to this final report).  A technical memorandum on implementation of research findings and products (Appendix A). Collectively, these deliverables:  Define quantitative approaches to systemic safety analysis and distinguish them from other approaches for identifying safety improvements such as the traditional crash- history-based (e.g., “hot-spot”) and policy-based approaches.  Communicate the benefits of quantitative approaches to systemic safety analysis for identifying safety improvements.  Provide a review of existing methods and tools to conduct quantitative approaches to systemic safety analysis, including the FHWA Systemic Safety Project Selection Tool, U.S. Road Assessment Program software (usRAP Tools and ViDA), and AASHTOWare Safety Analyst.  Provide a review of how other guidebooks, tools, software, resources, and ongoing research may be applicable to quantitative approaches to systemic safety analysis.  Present the data needs for the various methods and tools to successfully implement quantitative approaches to systemic safety analysis.  Specify and define appropriate applications (e.g., roadway functional classification, crash type, segments vs. intersections) for the various methods and tools for quantitative approaches to systemic safety analysis.  Provide a critique and capabilities assessment of each method and tool that agencies could use for quantitative approaches to systemic safety analysis.  Recommend current best practices and potential revisions that would increase the effectiveness or improve the ease of implementation of the methods and tools for quantitative approaches to systemic safety analysis.  Recommend or develop methods that agencies may use to evaluate the results (i.e., safety impacts) of improvements that were implemented using quantitative approaches to systemic safety analysis.

11 1.4 Overview of Research Methodology The research in Project 17-77 was conducted in seven tasks as follows: Phase I Task 1: Review Literature Task 2: Review Current Practice Task 3: Develop Outline and Write Portions of Draft Guidance Document Task 4: Develop Work Plan Task 5: Prepare Interim Report Phase II Task 6: Execute Approved Work Plan Task 7: Prepare Final Deliverables In Task 1 the research team reviewed and summarized literature relevant to the objectives of this research. As part of this review, the research team (1) summarized existing methods and tools to conduct quantitative approaches to systemic safety analysis, (2) critiqued and assessed the capabilities of each method and tool for conducting quantitative approaches to systemic safety analysis, (3) identified the data needs for the various methods and tools to successfully implement quantitative approaches to systemic safety analysis, and (4) summarized how other guidebooks, tools, software, resources, and ongoing research may be applicable to quantitative approaches to systemic safety analysis. In Task 2 the research team conducted a web-based survey of highway agencies and interviewed staff from selected agencies to better understand how agencies currently implement quantitative approaches to systemic safety analysis. In addition, several NCHRP and Federal Highway Administration (FHWA) research projects related to systemic safety analysis were being conducted at the time of this research. The research team reached out to the contractors leading these efforts to gain knowledge concerning the direction of their research and discuss how best to coordinate efforts. Key findings from the survey of practice are presented in Section 3 of this report. In Task 3 the research team developed an outline of a guidance document for implementing quantitative approaches to systemic safety analysis and wrote portions of the guidance document. In Task 4 the research team prepared recommended work plans for gathering information to develop a set of comprehensive guidance and training materials to assist DOTs, MPOs, local agencies, and safety practitioners to better understand, use, and implement quantitative approaches to systemic safety analysis. The work plans described the research team’s recommended approaches to gathering information for the key deliverables of this research. Activities included conducting on-site visits with highway agencies and facilitating a focus group of selected stakeholders to review and comment on the comprehensive guidance document. In Task 5 the research team prepared and submitted an interim report presenting the results of Tasks 1 through 4. The interim report included the results of the literature review (Task 1) and survey (Task 2), a detailed outline of the draft guidance document (Task 3), the recommended Phase II work plans (Task 4), and an updated schedule and budget for the remaining Phase II

12 tasks. The interim report also included, as a standalone appendix, the draft guidance document as it stood at the completion of Task 3. The research team met with the NCHRP project panel in July 2017 to review the interim report and discuss the Phase II work plans. In Task 6 the research team executed the recommended work plans, as approved by NCHRP. In Task 7 the research team prepared the final project deliverables which included (a) this final report; (b) a comprehensive guidance document on the implementation of quantitative approaches to systemic safety analysis; (c) a flyer to serve as marketing material to promote the implementation of quantitative approaches to systemic safety analysis (an appendix included in the guidance document); (d) a document highlighting the benefits of systemic safety analysis to decision makers (an appendix included in the guidance document); and (e) a technical memorandum titled “Implementation of Research Findings and Products,” (Appendix A in this report). In addition, the research team developed a PowerPoint presentation that describes the background, objectives, findings, and guidelines from the research. In addition to the original project scope, at the request of the research team leading the production of the second edition of the HSM and in consultation with the panel overseeing this research, the research team prepared a draft chapter on systemic safety analysis for consideration in the second edition of the HSM.