CHAPTER 9

Conclusions and Future Research

This chapter consists of three sections. The first section presents key findings from the research project. The second section describes several lessons learned during the conduct of the research project. The third section suggests several future research topics to address identified knowledge gaps.

Key Findings

This research developed two categories of CMF. The “case A” CMFs can be used to evaluate the safety effect of ATSPM deployment within an existing signal system. The “case B” CMFs can be used to quantify the change in crash frequency and severity that is associated with ATSPM-based changes to signal timing or operation at one or more signalized intersections. Summary results and key findings for the case A and case B CMFs are presented in this section.

Case A CMF Results

Table 60 summarizes case A CMFs obtained from the evaluation of six arterial streets in three states. The overall CMF values indicate a wide range in safety effect associated with ATSPM deployment, including both reductions and increases in crash frequency. Results showed that both Virginia ATSPM-operated arterials have an overall CMF value that is lower than 1.0, indicating a reduction in crash frequency associated with the ATSPM deployment. On the other hand, the Georgia ATSPM-operated arterials have an overall CMF value that is larger than 1.0, suggesting an increase in crash frequency associated with the ATSPM deployment. In Utah, one ATSPM-operated arterial experienced almost no changes in crash frequency following ATSPM deployment and the other arterial experienced an increase in crash frequency.

Table 60. Summary of case A CMF results.

Bold CMFs exclude 1.0 from the 95^th-percentile confidence interval.

^a – FI: fatal and injury crash; PDO: property-damage-only crash.

^b – Overall CMF value based on all crash severity categories, facility types, and traffic time periods combined.

The trends in the “CMFs by Severity” in Table 60 indicate that CMF values for FI crash frequency are lower than the CMF values for PDO crash frequency at five of the six arterial street study sites. This finding suggests that crashes occurring on ATSPM-operated signal systems are less likely to be associated with an injury or fatality.

The computed CMF values suggest that signal systems operated using ATSPM could be associated with a reduction in crash frequency and severity for some conditions. However, the values also suggest that ATSPM operation could be associated with an increase in crash frequency and severity for other conditions. It is likely that the safety outcomes realized by an agency implementing ATSPMs will reflect how the signal system is operated. For example, changes to the signal operation that result in a change in the (a) operating speed, (b) percentage of vehicles stopping, or (c) cycle length (and related phase termination frequency) could be associated with a change in safety. The findings of this research indicate that there is a potential safety benefit of ATSPM operation; however, the agency should monitor the system after ATSPM implementation (using both safety surrogates and crash history) to ensure that there has been no degradation in safety.

Case B CMF Results

This section describes case B CMFs that were developed for this project. These CMFs can be used to evaluate the operation of an existing signalized intersection, proposed changes in signal operation at an existing intersection, or the operation of a proposed signalized intersection.

The case B CMFs can be used within a crash prediction model (CPM) to evaluate signal operation when multiple factors (with a known association with crash frequency) have changed. These factors can include traffic volume, speed limit, cycle length, phase duration, or left-turn operation. A CPM is defined to include a safety performance function (SPF) that predicts the average crash frequency for a specified set of base conditions, one or more adjustment factors (AFs) that are used to adjust the SPF prediction to account for non-base conditions at a site of interest, and a local calibration factor. The typical CPM has the analytic form shown in Equation 61.

where,

When the case B CMFs are used in a CPM they are renamed as “adjustment factors” (AFs). This change in terminology is in recognition of the naming convention adopted for the forthcoming Highway Safety Manual 2^nd edition. The draft final version of this manual indicates that CMFs used in a CPM should be referred to as AFs. Regardless of the name used, the AFs described in this section are inferred to quantify the change in crash frequency associated with a change in signal operation (as described using one or more ATSPM reports) and have the same practical interpretation as a CMF.

AF for Platoon Ratio

The FI-specific AF for platoon ratio AF_Rp,FI is described by the following equation.

where R_p is the platoon ratio. Platoon ratio is computed using the “percent arrivals on green” measure obtained from the Purdue Coordination Diagram ASTPM report. This AF predicts a value of 1.0 when the platoon ratio equals 1.0 (i.e., random arrivals). It predicts a value of less than 1.0 when platoon ratio is smaller than 1.0 (i.e., favorable progression). In general, the AF indicates that intersection approaches with favorable progression have a lower crash risk than those approaches with unfavorable progression. A PDO-specific version of this AF is described in Chapter 6.

AF for Split Failure

The FI-specific AF for split failure AF_sf,FI is described by the following equation.

where,

The probability of a vehicle experiencing a split failure p_sf is computed using the “proportion of cycles with through phase split failure” and the “proportion of cycles with left-turn phase split failure.” These two measures are obtained from the Purdue Split Failure ATSPM report. This AF predicts a value of 1.0 when there are no split failures for either the left-turn or through movements. It predicts a value larger than 1.0 when split failures occur for either the left-turn or through movements. In general, it indicates that intersection approaches with a higher probability of split failure have a higher crash risk. In addition, for a given probability of split failure, the AF indicates that crash risk is lower when the cycle length is larger. A PDO-specific version of this AF is described in Chapter 6.

AF for Yellow Actuations

The FI-specific AF for yellow actuations AF_ya,FI is described by the following equation.

where,

The variable PVY is computed using the count of yellow actuations obtained from the Yellow and Red Actuation ATSPM report. This AF predicts a value of 1.0 when there are no vehicles entering the intersection during the through yellow indication. It predicts a value larger than 1.0 as the proportion of vehicles entering during the yellow interval increases. In general, it indicates that, for a given approach volume, intersection approaches with a higher probability of intersection entry during yellow have a higher crash risk. A PDO-specific version of this AF is described in Chapter 6.

AF for Left-Turn Gap Availability

The FI-specific AF for left-turn gap availability AF_lt,LT is described by the following equation.

where,

The opposing left-turn permissive capacity variable c_lt,perm is computed using the “proportion of the through green interval with “long” gaps in the through traffic stream.” Left-turn capacity increases with an increase in the proportion of green with long gaps. This measure is obtained from the Left-Turn Gap Analysis ATSPM report. This AF predicts a value of 1.0 when the speed limit is 40 mph and the opposing left-turn permissive capacity is 1,000 veh/h. It predicts a larger value than 1.0 as the left-turn permissive capacity decreases. In general, it indicates that intersection approaches with high left-turn permissive

capacity have a lower crash risk than those approaches with a low capacity. In addition, for a given left-turn capacity, approaches with a high speed limit have a higher crash risk than those with a low speed limit.

Lessons Learned

This section summarizes the insights and lessons gained during the research project that are related to (a) CMF development for ATSPMs and (b) the potential safety benefits of using ATSPMs to manage and monitor signal system operation.

During case A CMF development, one goal was to develop CMF values that vary as a function of ATSPM system characteristics (e.g., the frequency of signal timing adjustments by the agency when using ATSPMs) and signal system characteristics (e.g., speed limit, number of through lanes, signal spacing, volume-to-capacity ratio). However, due to the limitations in data collection and the number of ATSPM-operated arterials included in the CMF development, the research team was unable to quantify reliable CMF functions of this nature.

A second goal of case A CMF development was to produce CMF values that are relatively consistent from arterial-to-arterial and from state-to-state. CMF values that demonstrate this type of consistency could be more reliably transferred to other arterials and states. However, due to the wide variation in safety effects associated with the sites that were studied, the research team chose not to combine the CMF values and, instead, presented the results separately for each state and for each ATSPM-operated arterial.

For case B CMF development, the research team prioritized certain ATSPM reports based on a set of criteria (see Chapter 3 for details of the prioritization process). Based on this prioritization, the team developed case B CMF functions to address the four ATSPMs identified in the following list:

• Percent Arrivals on Green
• Yellow Actuations
• Split Failure
• Left-Turn Gap Analysis

However, other ATSPMs are also utilized by agencies to improve intersection safety. One good example is the Timing and Actuation ATSPM report. This report depicts a detailed view of the vehicle and pedestrian cycles for each phase overlaid with detection actuations (Atkins, 2020). This information can then be reviewed by agencies following the occurrence of a crash to determine factors that may have contributed to the crash. These contributing factors may include a vehicle running a red light or a turning vehicle failing to yield to a crossing pedestrian during the Walk interval. This use of the Timing and Actuation ATSPM report can lead to improvements in intersection safety by giving the operating agency a better understanding of crash contributing factors. However, it may be difficult to develop a CMF value associated with this report because it does not generate specific metrics that can be used as an independent variable during CMF development.

Suggestions for Future Research

Additional research is suggested to address knowledge gaps that were not addressed during this project. The following paragraphs identify suggested research topics that would further improve practitioner’s understanding of the way ATSPMs influence safety at signalized intersections.

This project’s data collection effort for case A CMFs included six ATSPM-operated arterials (two arterials from each of three states). The results obtained from the case A CMF analysis suggested a wide variation in safety effects of ATSPM-operated arterials among arterials and among states. To better understand the relationship between ATSPM-operated arterials and their safety effects, and develop a more

consistent CMF value, additional research is suggested that would expand the number of ATSPM-operated arterials that would be amenable to a before-after study for the purpose of quantifying a CMF that would describe the change in safety associated with ATSPM deployment. The collective set of study sites should have a range of signal operation and geometric design characteristics that would facilitate the development of a CMF function.

As described in Chapter 6, the exploratory analysis of red actuations (i.e., entry the during red interval) found that an increase in the proportion of approach volume entering during the red interval was associated with a reduction in crash rate. This finding is contrary to the results reported in the literature. It is partly explained by the researchers’ selection of study sites that were ultimately found to be suboptimal for the examination of red actuations. Specifically, they unknowingly selected intersections with detection designs that were unable to reliably screen out right-turn-on-red maneuvers. To address this challenge, it is suggested that future research be undertaken to develop a CMF for red actuations using intersections having detection and monitoring equipment that can reliably exclude right-turn-on-red vehicles from the entry-on-red counts.

As discussed in Chapter 3, the research team prioritized existing ATSPM reports for the development of case B CMFs. This prioritization process led to the development of case B CMFs for arrivals on green (platoon ratio), split failures, yellow actuations, and left-turn gap analysis. However, in addition to these four ATSPMs, there are other ATSPM reports (and related measures) that could have a quantifiable relationship with crash frequency or severity. Some of the ATSPM reports that should be the subject of future safety research include Pedestrian Delay and Preemption Details. There is also emerging technology capable of producing an Estimated Pedestrian Conflict report. It is suggested that research be undertaken to explore the relationship between these three ATSPM reports (and related measures) and crash frequency and severity.