Guidelines for Quantifying Benefits of Traffic Incident Management Strategies (2022)

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.

29   TIM Analysis Step 2— Planning the TIM Analysis Planning the analysis enables the TIM analysts and program managers to develop meaningful results from the analysis within the scope of available data, expertise, funding, and time. At this point, the TIM program manager and/or analyst should have a strong understanding of the problem to be addressed, the goals of the analysis, the desired products from the analysis, and potential resources available for the analysis. For larger TIM programs, the outcome from this step will be a TIM benefits evaluation plan, a document that defines in detail the findings from step 1 (analysis purpose), the analysis approach, the geographic and temporal scope, the types of data available, and potentially to be collected, the basis for parameters and assumptions to be used in the analysis, and the time line and resources for the conduct of the analysis. Further, the plan should include a schedule and products, processes, and data to be delivered at the conclusion of the evaluation. The benefits evaluation plan may be prepared by conducting agencies, such as university research institutes or consulting organizations, in collaboration with the TIM program. For smaller TIM programs, the outcome from this step may be as simple as a briefing or a technical memo that summarizes the decisions. Key components of the evaluation plan typically include the background (from step 1), a proposed methodology, proposed geographic and temporal scope of analysis, and the sources, types, and shortcomings of data for use in the evaluation. Before completing the TIM evaluation plan, the analyst should have accessed samples of data and metadata to understand the expected levels of data processing and cleaning that will be required. In developing the benefits evaluation plan, the key factors must align to select the “right fit” analysis approach: • Methods to be applied or developed as a part of the evaluation. • Geographic and temporal scope of interest. • Data available and acquirable for use in analysis. • Time and resources (expertise and funding) available. This chapter focuses on each of these factors. Note that the time and effort required during this step are not trivial and may require a significant portion of the evaluation budget. In planning the analysis, greater focus must be given to understanding the types, quality, and quantity of data available. Select, Modify, or Develop Analysis Methods As discussed in Chapter 2, there is significant variation in incident reporting procedures and documentation, underlying assumptions, and the methods for estimating TIM outcomes. Each method employs different assumptions and generalizations. As a part of this study, methods C H A P T E R 4

30 Guidelines for Quantifying Benefits of Traffic Incident Management Strategies based on previous empirical and simulation activities or mathematical modeling were com­ pared using common data sets with the goal of understanding the strengths and limitations of these methods, and documenting the level of effort, complexity, and data needs to support the use of these methods. Four methods for estimating secondary incidents, three methods for estimating delay, and two methods for estimating emissions were applied to three corridors: one from Maryland, Washington State, and Texas. The comparison does not indicate the accuracy of one method over another, but rather highlights the considerations in either developing or applying an existing method to estimate benefits. The key considerations in making the decision to apply or modify an existing method, or to develop a local analysis method, are provided below: • Explore what analyses have been conducted in the past. – Are the methods applied during former analyses still relevant and usable? – Do they require updating or enhancements? – Would more robust analyses that use new data or resources yield findings at the level of detail desired? – Would other methods better meet the specific desired outcomes? • Identify what data inputs are required for the method(s) under consideration. • Capture the assumptions and limitations inherent to the method(s) being considered. • Understand what parameters may require “localization” and the feasibility of undertaking localization of parameters. • Consider trade­offs among methods with regard to the types of outcomes captured, the robustness of analyses, data requirements, and the effort to conduct the analysis. Table 6 summarizes the methods considered as a part of this study by TIM outcome category. The table first provides a summary of findings from the analyses conducted as a part of this study. Following the summary are the specific methods and their strengths, limitations, and assumptions based on application to the three corridors. The FHWA TIM­BC tool was also applied to the Maryland corridor, and this activity is summarized in Appendix D. In selecting the method(s) to quantify TIM outcomes, consideration should be given to the geographic and temporal scope of interest, the data available and acquirable for use in analysis, and the time and resources (expertise and funding) available. These factors are covered by the subsequent three sections of this chapter. Refine Geographic Sectioning and Time Period for Analysis The high­level geographic and temporal scope of interest is typically specified during step 1, Specify Analysis Purpose. During step 2, Plan the TIM Analysis, the analyst team must capture with greater detail the region of interest and make decisions on the right­size aggregation level to conduct the analysis. This decision should be made concurrently with the data acquisition and types of analyses to pursue. In other words, the final decisions on segmentation by geography and time should only be made once data and resource availability is understood and analysis methods are selected. If choosing rule of thumb methods for analysis rather than performance data, the selection of geographic and temporal scope can be at a broader level of aggregation. With greater density and completeness of performance data, a more refined geographic and temporal scope will yield more precise and operationally actionable estimates of outcomes. The key considerations in refining and defining the geographic and temporal scope of analysis follow.

TIM Analysis Step 2—Planning the TIM Analysis 31   Safety: Identifying Secondary Crashes  When possible, rely on direct reporting or logging of secondary crashes from law enforcement, TMC operator, and service patrol before and after TIM implementation.  After-the-fact methods to identify which incidents are secondary are provided below.  The highest fidelity in identifying an incident as secondary is achieved from tailored temporal and spatial boundaries based on duration or severity of the incident and potentially even by peak, night, and off- peak, based on localized conditions.  Reduction in secondary incidents from TIM is typically estimated as linear with the reduction in incident duration. Method Strengths Limitations Data Needs Raub, 1997 “15% of all crashes are secondary” Ease of application because it does not require any "search" method for identifying a subset of incidents as secondary incidents Does not consider unique time and space impacts of individual incidents. Based on 1 month of data from Chicago, Illinois, using 15- minute and 1,600-meter search parameters • Total number of incidents Hirunyanitiwattana and Mattingly, 2006 “incidents 1 hour & 2 miles upstream of crash” Constant secondary crash "search" function Does not consider unique time and space impacts of individual incident severity • Incident start time • Incident location • Incident direction Chou, 2010 “Time and Space boundary based on incident duration, volume, and # lanes blocked” Creates unique time and space "search" boundaries for each incident. Based on 6 months from New York HELP program Delay relationships only formulated for up to two lanes closed • Incident location and direction • Volume and lane count • Number of lanes closed • Incident duration Sun et al., 2010 “Time and Space boundary based on Incident type and v/c ratio” "Search" boundaries defined by event characteristics Requires details on injury severity. Does not accommodate events with more than three lanes closed • Incident start time, location and direction • Volume to capacity ratio • Injury severity Mobility: Quantifying Reduction in Delay  If data, expertise, time, and resources allow, highest fidelity is achieved from empirical and simulation- augmented localized estimation of delay prior to and after TIM implementation.  The FHWA TIM-BC tool and others listed in Chapter 2 offer more streamlined, less time- and resource- intensive options.  Most TIM programs apply a combination of rules of thumb and regression or a similar mathematical model to estimate delay savings based on incident duration and other factors such as number of lanes closed. Method Strengths Limitations Data Needs Chang and Igbinosun, 2015 “Simple equation to estimate delay savings” Directly estimates delay savings without explicit estimation of TIM effect Does not consider effect of shoulder events Inherent in equation is TIM effectiveness in reducing incident duration • Traffic volume • Incident duration • Number of lanes and lanes blocked Table 6. Lessons learned from quantifying TIM outcomes. (continued on next page)

32 Guidelines for Quantifying Benefits of Traffic Incident Management Strategies Method Strengths Limitations Data Needs Sun et al., 2010 “Equation based on change in incident duration and excess demand” Simple to apply Does not require incident duration data Considers events up to three lanes blocked Assumes no delay if demand is less than the reduced capacity from the incident • Incident duration with TIM • Incident duration without TIM • Traffic demand • Reduced capacity Khattak and Rouphail, 2004 “Delay a function of volume capacity ratio” Different equation parameters based on urban/rural, incident duration, and lane blockage Assumes non-TIM is 1.25 times TIM incident duration Covers incidents blocking up to two lanes Computes delay for incidents up to 60 minutes in duration • Area type—urban or rural • Number of lanes • Incident duration rounded to 15 minutes • Number of lanes and lanes blocked • Traffic volume and capacity Environment: Quantifying Reduction in Emissions and Fuel Use  A secondary but important benefit from TIM.  Consider use of rules of thumb for this estimation based on delay.  Estimates are only as accurate as estimates of delay. Many tools include this as an output.  Higher-fidelity tools can be applied with additional data such as elevation, road curvature, and fleet mix. Method Strengths Limitations Data Needs Chang and Igbinosun, 2015 “Direct estimation from delay” Simple to apply and only requires estimate of delay reduction from TIM Based on Maryland DOT local calibrated emission rates • Delay reduction achieved through TIM Morris and Lee, 1994 “Emissions savings per safety assist” Simple to apply and only requires the number of safety assists Assumes all assists have equal emission savings and based on data 2+ decades ago • Number of SSP assists Efficiency Outcomes  This can only be performed with localized detailed records for program costs before and after the TIM program or project implementation.  Once efficiency savings are established for a period of time, these estimates can be adjusted and extrapolated for future periods of time.  Efficiency outcomes were not examined for the three corridor data sets. Traveler Satisfaction Outcomes  Comment cards or survey analysis is the best method for quantifying these outcomes.  Traveler satisfaction outcomes were not examined for the three corridor data sets. Table 6. (Continued). Confirm Changes in the Region Consider what has changed and remains the same over the period of time selected to analyze the TIM activity. This may include the construction or enhancement of roadways, growth or decline in regional demand or population, an atypical weather season, or the implementation of other ITSs. In the course of the analysis, it is important to understand whether performance changes and benefits are because of the TIM activity or other confounding factors. For example,

TIM Analysis Step 2—Planning the TIM Analysis 33   ICT may have increased during the first winter with TIM, when compared to the previous winter without TIM, because of multiple major snowstorms, resulting in more complex incidents than the previous winter. Define Homogeneous Road Segments Typically, data available for analysis will not have full incident­specific information such as demand and speed prior to and after each incident. If data are available, the effort and time in cleaning and analyzing incident­specific outcomes may not meet analysis time and funding constraints. Consequently, it is advisable to create segment­level roadway demand and capacity estimates. In defining road segments for analysis, first, consider the segments of the road network with common levels of TIM service. For example, if the analysis centers on the assessment of an SSP program with greater frequency of patrolling on one facility over another, then the benefits of the program should be differentiated by facility. Second, consider road segments that have similar demand and capacity characteristics. Because the key driver for most TIM benefits is the reduction in delay, and delay is a function of roadway demand and capacity, segmenting by similar levels of demand and capacity is advisable. In doing this, consider the locations of major freeway ramps or interchanges for road segment boundaries. Third, consider the geographic location of data sources to be used in the analysis. For example, the locations of road sensors may not allow the level of performance measure detail needed to differentiate benefits for shorter road segments. Define Homogeneous Time Segments The effects of TIM activities are observed over time. Some outcomes are immediately visible (e.g., SSP), and others may take more time to show value (e.g., public outreach). Likewise, outcomes may prove more significant during peak versus off­peak times of day. When selecting a before and after analysis, it is important to make this selection, such that: • The period prior to TIM has equivalent seasonal trends as the period after TIM is in place. • The period after TIM is in place does not include the time for stabilization of the TIM implementation (i.e., getting the kinks out). • Data are available over time, both with and without the activity or outcome of interest. As with defining road segments, consider grouping the time periods of the day and days of the week (weekday/weekend) into time periods with similar traffic conditions. This segmentation is most critical when making the decision on how best to align roadway demand to incidents. Typical segmentation includes four time periods: weekday AM peak, weekday PM peak, weekday off­peak, and weekend. More detailed segmentation may focus on the day of week and night off­peak. More detailed temporal segmentation will offer greater fidelity in benefits estimates but at greater data collection and analysis expense. Inventory and Examine Data Sets Available A plan for a comprehensive analysis that is not grounded in the realities of available data is not likely to succeed. During step 1, the analysts will have identified the specific outcomes of greatest interest and value. These outcomes define the basis for data needed for analysis.

34 Guidelines for Quantifying Benefits of Traffic Incident Management Strategies In summary, data sets central to TIM analysis are those with variables pertinent to the outcomes of interest. The basic data sets and data fields needed include: • Incident data—start time, stop time, location, lanes blocked, primary or secondary classification, type (crash, injury, fatality, etc.). • Response data—detection time, response time, time at the incident by responder, equipment and materials use, etc. • Aggregate incident statistics—frequency of incidents by type and time period, the average time to clear the roadway and shoulder lanes (roadway clearance and incident clearance times), etc. • Demand and speed data—vehicles per hour per lane, miles per hour. • Facility­level data—number of lanes, left and right shoulder, capacity, road curvature/grade, elevation, fleet mix, occupancy, etc. A more detailed framework oriented toward the establishment of a TIM performance measurement program is offered in Guidance for Implementation of Traffic Incident Management Performance Measure- ment (Pecheux et al., 2014). The guidebook describes considerations, requirements, case studies, and a data framework example for establishing a TIM performance measurement program. Based on the maturity and focus of the TIM program, the above five types of data may be merged through a common referencing system, may be available but separate, or may not all be available. Merged data are nearly never the case, but it is the ultimate goal for those who perform the analysis. For unified data, the analysis will require some approximation or use of measures based on peer TIM program experience. This section presents data availability considerations for quantifying mobility, safety, efficiency, environmental, and traveler satisfaction outcomes from TIM activities. Data Required to Estimate Key Mobility Outcome—Delay The analysis of a TIM program’s effect on reducing traveler delay requires, at a minimum, the time and location of incident events in the desired analysis period. This information allows the mapping of traffic demand to each event, which is a critical input to nearly every delay estimation method. Other incident data fields include duration and the number of lanes blocked, the latter serving to estimate capacity loss from the incident. In addition to the incident data, most TIM benefit estimation strategies require demand/ volume as an input. Thus, there is also a need to reasonably estimate the traffic volume at the time and location of each incident. Clearly, location­specific and high­resolution temporal volume counts reduce the number of required assumptions needed for volume estimation and potentially for capacity loss estimation. However, there is a tradeoff between the effort required to apply detailed volume data at the incident level and the relative improvement in precision and accuracy of TIM benefit estimation. The previous section, “Refine Geographic Sectioning and Time Period for Analysis,” discusses considerations in aggregation for demand and incident data. The key consideration in accurately estimating delay savings is the incident duration with and without TIM activities. Ideally, empirical data prior to and after the implementation of the TIM program will be available. For legacy TIM programs where facility demands have changed dramatically, data on corridors with TIM can be compared with comparable corridors without TIM. In this latter case, incident duration, lane closure, and demand will be required for both corridors. Data Quality and Availability Dictate Analysis Approach • Clean and comprehensive data are critical to high-fidelity estimates of benefit. • The absence of other data may mildly reduce the fidelity of TIM benefits estimates. • In the absence of key data, such as incident duration, sketch-level analysis methods are advised.

TIM Analysis Step 2—Planning the TIM Analysis 35   In Table 7 are strategies to overcome the absence of specific data, and strategies to fill gaps from missing data. Note, without incident duration data, the estimate of benefits is closer to a sketch­level analysis as compared to an operations assessment. Based on the method chosen or developed to estimate the delay savings from TIM, other data may be required, such as roadway grade, weather conditions, or area type (urban or rural). Transforming the reduction in delay to monetized outcome requires additional data, mainly the value of time as measured in dollars per hour. The following types of data at a facility or regional level, as well as seasonal or temporal levels, allow the agency to estimate a more accurate valuation of delay: • Percentage of traffic that is commercial vehicle. • Vehicle occupancy rates. • Rates of fuel consumption in gallon per hour by vehicle type (commercial and motorist). • Fuel costs per gallon of fuel. The above data are typically available at the metropolitan planning organization or state agency level. Appendix E offers sources for regional or national rates for the above data types. Data Required to Estimate Safety Outcomes— Secondary Incidents and Risk Reduction Though TIM programs cannot prevent all incidents from occurring, they can mitigate the occurrence of secondary crashes by efficiently responding to and clearing primary incidents. The data required vary by methods chosen to estimate the number of prevented secondary incidents. • All methods require incident count data. • Most methods require incident location and direction. • Some methods require traffic volume, lane count, lane closure, incident duration, or incident severity data. Additional safety benefits stem from limiting the duration of time responders are present in high­speed traffic. Data required to capture this outcome are: • Injury and fatality count for responders while dispatched to manage incidents. If regional or state­level data are unavailable, national data can be used. • Reduction in time responding to incidents through TIM activities. The surrogate for this can be a reduction in ICT through TIM. However, more accurate estimates will account for the reduction in time out of vehicle and exposed to traffic. Data Required To Estimate Environmental Outcomes Environmental impacts are an important consideration in estimating TIM benefits. Though several emissions estimation methods exist, the core input for most is the estimate of delay savings. Simpler methods only require the number of safety patrol assists. For more rigorous mobility benefits analysis employing simulation models, emissions data are output from these processes and tools. Data Required To Estimate Efficiency Outcomes Efficiency outcomes data will vary significantly by the TIM activity and program and can include the time and resources (equipment, fuel) spent fielding incidents and after­incident reporting by various responders as well as other positions coordinating TIM response.

36 Guidelines for Quantifying Benefits of Traffic Incident Management Strategies Data Type What If Data Are Not Available? How To Overcome Incomplete or Missing Data Incident Duration Consider using state or regional incident duration estimates based on incident severity and lane closure types. If regional or state information is unavailable, refer to national rates. The following table of national incident duration is provided as a reference. Average Minutes Incident Duration (FHWA, 2013b) Severity Shoulder 1 Lane 2+ Lanes Non-crash 29.8 29.1 47.4 Property damage only (PDO) 38.1 42.3 56.9 Injury 57.4 43.9 46.8 Fatal 229.6 175.5 187.1 Consider estimating average incident duration by incident type, severity, time of day, and road segment using the data that are available and assign the respective average value for missing data. Without severity data, consider sketch planning methods, such as Quick Benefits Analysis Method presented in Appendix B. When more than 50% of data are missing for a given grouping, consider a broader aggregation first through longer road segments, then merging groups that are most similar in the duration attribute. Number of Lanes Blocked If incident duration and severity data are available, use the table presented above or a similar local table to approximate lanes blocked. If only incident type information is available, the below table can be used to randomly assign the number of lanes blocked to align with the percentages in the table below or a comparable local estimate. Lane Blockage by Incident Type (FHWA, 2013b) Incident Shoulder 1 Lane 2+ Lanes Crash 55.8% 27.8% 16.4% Non-crash 83.7% 14.8% 1.6% To fill missing data, use existing data on incident duration and lane blockage to assign entries with missing lanes blocked. The table below is an example of a Maryland I-495 data set applied within this study. Maryland I-495 Incident Look-Up Type Duration (minutes) Annual Frequency Shoulder 18.0 2,591 1 Lane 19.9 6,764 2 Lanes 33.2 1,206 3 Lanes 43.1 340 4+ Lanes 45.8 137 Facility Demand If incident-specific volume or temporal and road segment-level volume data are unavailable, consider using annual average daily traffic (AADT) or seasonal facility volumes from regional data. Apply a peaking factor for the AADT to estimate demand during peak and off-peak times of day. Where demand is missing, use the average demand on the same road segment with equivalent time of day and day of week as the missing data. Effective Capacity Factors Nearly all agencies apply the HCM freeway capacity factors to estimate the effective capacity of road segments that block either shoulders or mainlines. As a part of this work, incidents across three state corridors were analyzed to confirm and expand HCM factors. They are available in Appendix C. n/a Table 7. Strategies to compensate for missing or incomplete data to support quantifying mobility outcomes.

TIM Analysis Step 2—Planning the TIM Analysis 37   Data Required To Estimate Traveler Satisfaction Outcomes Typically, traveler satisfaction is measured through surveys. In addition, agencies that allow travelers to post feedback using social media or websites can use that information as part of the assessment of traveler satisfaction. Build an Analysis Schedule By creating a timeline and resource allocation to key tasks, the TIM staff or consultants leading the evaluation understand the pace and resource expectations. The evaluation schedule should estimate the effort and time required in terms of the following factors: • Data preparation (collection, cleaning, and merging). • Model or method development and calibration. • Fine­tuning of model, methods, parameters, or processes. • Performing the analysis and aggregating results. • Developing communications products. The schedule should assign responsibility for specific tasks and provide downstream activities such that the effect of expediting or delaying one activity can be readily understood in terms of subsequent activities and the completion of the analysis.