Decision-Making Guide for Traffic Signal Phasing (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.

55 9 Safety Considerations 9.1 Understanding the Safety Impacts Associated with Signal Phasing By one definition, signal phasing mode is the determination of allowable conflicts within an intersection, or in other words, the determination of which vehicle movements and/or pedestrian crossings may proceed concurrently. As such, signal phasing mode selection has a large impact on the safety of an intersection; inappropriate selection of a certain phasing mode could lead to an unreasonable safety risk. As mentioned in Chapter 4 of this Guide, three safety reasons (sight distance, lane configuration, and crash history) should influence the selection of left-turn phasing mode. These three safety reasons are elements that could by themselves dictate the left-turn phasing mode; for example, the lack of adequate sight distance necessitates the use of protected-only phasing while the lack of a dedicated left-turn storage bay necessitates the use of permissive-only or split phasing. In some cases, safety considerations allow multiple phasing determinations. These cases are where this Guide offers new tools based on conducted research. 9.2 Methods of Safety Analysis 9.2.1 Traditional Method - Qualitative Crash History Prior to 2010, when the American Association of State Highway and Transportation Officials (AASHTO) published the Highway Safety Manual (HSM), the traditional method for any transportation safety analysis was a review of crash history. As Chapter 4 of this Guide outlines, crash history remains a key data input to determine the appropriate signal phasing. The use of crash data is built on the institutional knowledge of transportation professionals that protected-only left-turn phasing will alleviate left-turn crashes at an unprotected left-turn movement. At signalized intersections operating under protected-only conditions, the crash history may not provide a full perspective of the latent crashes that may occur under permissive-only phasing. In this case, the practitioner may choose to use the crash history of an unsignalized two-way stop control intersection of similar characteristics (e.g., lane use, traffic volumes, urban/suburban context) as a proxy. By nature of an unsignalized intersection, the uncontrolled left-turn movements operate as permissive and thus may be an appropriate comparison. Engineering judgment must be applied because certain unsignalized intersections may operate differently than the intended signalized intersection characteristics. For example, adjacent signalized intersections may create gaps in traffic flow and facilitate left-turning movements at the unsignalized proxy intersection, more closely resembling protected-permissive left- turn phasing. This scenario would not properly mimic an isolated signal operating under permissive-only left-turn phasing. If the unsignalized intersection is operating outside of the influence area of a signalized intersection, it may be a truer comparison to an isolated signalized intersection.

56 9.2.2 HSM Method – Quantitative Crash Prediction The HSM transitioned safety analysis from a reactive, descriptive analysis to a proactive, predictive analysis. Two of the predictive tools introduced in the HSM are SPFs and CMFs. SPFs are equations that model the number of crashes that are expected to occur at a specific intersection or road segment based on various geometric and volume parameters such as number of lanes or AADT. A hypothetical example of an SPF follows: 𝐿𝑒𝑓𝑡 − 𝑇𝑢𝑟𝑛 𝐶𝑟𝑎𝑠ℎ𝑒𝑠 = 𝛼𝑀𝐴𝐽𝐴𝐴𝐷𝑇 𝑀𝐼𝑁𝐴𝐴𝐷𝑇 𝑒𝑥𝑝 where, MAJAADT = major road AADT MINAADT = minor road AADT PROT = 1 if left turns are provided a fully protected phase; 0 otherwise LTAADT = the left-turn AADT α, β1…β5 are parameters estimated during the model calibration CMFs are derivative values of SPFs that quantify the expected change in crash frequency as a result of implementing a specific countermeasure. In other words, CMFs allow engineers and planners to predict the change in expected crash frequency before a modification is made to the transportation system. A CMF is simply computed as follows: 𝐶𝑀𝐹 = 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝐶𝑟𝑎𝑠ℎ 𝐹𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 𝑤𝑖𝑡ℎ 𝑇𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝐶𝑟𝑎𝑠ℎ 𝐹𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 𝑤𝑖𝑡ℎ𝑜𝑢𝑡 𝑇𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 In practice, a CMF is most often reported as a singular value. CMF values have a baseline of 1.0; a value of 1.0 indicates that the treatment is expected to have no impact on crash frequency. A value less than 1.0 indicates that the treatment is expected to reduce crash frequency, while a value greater than 1.0 indicates the opposite – that the treatment is expected to increase crash frequency. CMFs sometimes apply to all crashes at an intersection or segment and other times apply only to certain types of crashes, such as severe crashes, angle crashes, or crashes during hours of darkness. For example, a CMF value of 0.7 indicates that a treatment is anticipated to reduce 30% of the relevant crashes. A large quantity of CMFs was included in the original publication of the HSM; as more CMFs are researched, FHWA reviews and accepts them into the CMF Clearinghouse database. Example: A traffic engineer wants to understand the impact of installing Flashing Yellow Arrows on an arterial corridor. The most appropriate CMF for FYA installation for this scenario is 0.92 for left-turn crashes. Over the past three years, there have been 100 left-turn crashes along the length of this corridor. Based on the CMF value, the engineer can expect that installation of FYA will eliminate 8% of the left-turn crashes (8 crashes). Note: 0.92 is the accepted CMF for FYA installation in Virginia. Many other studies have reported different CMF values for FYA installation. It is important to consider the intersection and study characteristics to choose the most appropriate CMF for each scenario.

57 9.3 New Research Research conducted as part of this project sought to develop new SPFs and CMFs to help answer several outstanding research questions. The research team used data from across the country to develop these new equations and values. The analysis was conducted at the intersection level, i.e., crashes were predicted for the intersection as a whole and not for an individual approach. Analysis was conducted for several crash types, but the only statistically significant results found were for left-turn opposing (LTOPP) crashes and LTOPP fatal+injury (LTOPP INJ) crashes. For additional background on the development of these CMFs and SPFs, refer to Appendix B. The CMFs and SPFs shown were developed using a cross- sectional analysis. Additionally, the CMFs and SPFs were developed on an intersection level and should be applied to the total number of crashes at an intersection, not only a single approach. 9.3.1 Developed CMFs Table 5 includes the new CMF values produced in this project. The results indicate that LTOPP and LTOPP INJ crashes are both expected to decrease with the conversion of left-turn phasing on the major and/or minor road from permissive-only to protected-only. The converse results are also true: converting from protected-only left-turn phasing to permissive-only left-turn phasing yields an expected increase in both LTOPP and LTOPP INJ crashes. A larger expected reduction in intersection crashes is found from converting the left-turn phase mode on the major road due to the higher vehicle volumes on the average major road in comparison to the average minor road. Additionally, given that the LTOPP INJ CMF is lower than the LTOPP CMF, these results show that left-turn phase mode has large implications on fatal+injury crash types, which are the focus of many agencies’ Vision Zero and Road to Zero plans. No significant change in safety performance was found for changing from permissive-only to protected-permissive left-turn phase mode. Table 5 - CMFs Derived from Research Treatment Crash Type CMFs Estimate Standard Error Convert from permissive-only or protected-permissive left-turn phase mode to protected-only left-turn phase mode on major road LTOPP 0.507 0.070 LTOPP INJ 0.443 0.073 Convert from permissive-only or protected-permissive left-turn phase mode to protected-only left-turn phase mode on minor road LTOPP 0.702 0.098 LTOPP INJ 0.687 0.109 Convert from protected-only left-turn phase mode to permissive- only or protected-permissive left-turn phase mode on major road LTOPP 1.973 0.273 LTOPP INJ 2.259 0.374 Convert from protected-only left-turn phase mode to permissive- only or protected-permissive left-turn phase mode on minor road LTOPP 1.424 0.198 LTOPP INJ 1.457 0.231

58 Applying These CMFs As mentioned, the CMFs computed in this study are intersection-based, i.e., converting from permissive- only to protected-only left-turn phase mode on the major road has an expected LTOPP crash reduction of 49.3% (1-0.507) for the intersection crashes. These CMFs should not be applied to only the crashes on one approach. These CMFs can be used to predict the safety impact of adjusting the left-turn phase mode at a single intersection or along the entire length of a corridor. There are two applications of this information. First, an agency looking to convert major or minor street signal approaches to protected-only phase mode can use these CMFs to quantify the effect on safety performance. Second, an agency looking to convert a signal from protected-only to either permissive-only or protected-permissive phase mode for operational benefits can quantify the expected increase in crashes. Further discussion on balancing safety and operational impacts of signal phasing can be found in Chapter 11 of this Guide. Several additional CMFs can be derived from those in the table above. Converting both the major and minor roads to protected-only left-turn phase mode yields the following CMF for LTOPP crashes of all severities: 𝐶𝑀𝐹 = 𝐶𝑀𝐹 ∗ 𝐶𝑀𝐹 = 0.507 ∗ 0.702 = 0.356 Converting just one approach of a major/minor road to protected left-turn phasing yields the following CMFs, assuming that both approaches have similar characteristics: 𝐶𝑀𝐹 , = 1 − 0.5 ∗ (1 − 𝐶𝑀𝐹 ) = 1 − 0.5 ∗ (1 − 0.507) = 0.754 𝐶𝑀𝐹 , = 1 − 0.5 ∗ (1 − 𝐶𝑀𝐹 ) = 1 − 0.5 ∗ (1 − 0.702) = 0.851 9.3.2 Developed SPFs Below are two SPFs developed by the research accompanying this project. The first estimates the total annual left-turn approach crashes of all severities. The second estimates the total annual left-turn approach fatality/injury crashes (K+ABC). 𝐿𝑇𝑂𝑃𝑃= 𝑒 . 𝑀𝐴𝐽𝐴𝐴𝐷𝑇 . 𝑀𝐼𝑁𝐴𝐴𝐷𝑇 . 𝑒( . . . . ) where, LTOPP = Left-turn opposing crashes (all severities) per year MAJAADT = major road AADT (sum of both approaches) MINAADT = minor road AADT (sum of both approaches) RATIOLTMAJ = Ratio of major road traffic turning left to total traffic on the major road in peak hours RATIOLTMIN = Ratio of minor road traffic turning left to total traffic on the major road in peak hours PROTMAJ = 1 if major road has protected-only left-turn phasing; 0 otherwise PROTMIN = 1 if minor road has protected-only left-turn phasing; 0 otherwise

59 𝐿𝑇𝑂𝑃𝑃𝐼𝑁𝐽 = 𝑒 . 𝑀𝐴𝐽𝐴𝐴𝐷𝑇 . 𝑀𝐼𝑁𝐴𝐴𝐷𝑇 . 𝑒( . . ) where, LTOPPINJ = Left-turn opposing crashes (fatal+injury only) per year Other variables as previously defined Applying These SPFs As with this study’s CMFs, these SPFs are intersection-based, and the annual crash estimates produced apply to intersections as a whole, not approaches. Use of the SPFs requires information on all approaches of the intersection. The primary application of the SPFs is for locations with no crash history to which to apply a CMF. If considering a phase mode change at an existing signal, these SPFs may be used to estimate the expected annual crashes of the alternative phase mode. Chapter 11 provides further information on comparing signal phase mode alternatives.