Potential MUTCD Criteria for Selecting the Type of Control for Unsignalized Intersections (2015)

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.

22 CHAPTER 3: LITERATURE REVIEW BACKGROUND The literature review task was subdivided into the following focus areas:  Key reference documents, including the TRB HCM (23), American Association of State Highway and Transportation Officials (AASHTO) Highway Safety Manual (HSM) (24), and ITE Manual of Traffic Engineering Studies (25).  Previous literature that discusses methods for selecting traffic control devices, such as YIELD or STOP signs, for unsignalized intersections.  Previous literature that discusses methods for selecting traffic control devices for unsignalized pedestrian crossings. These methods present additional approaches, such as calculating delay using a series of equations or using a point system, for selecting a traffic control device at an unsignalized intersection. SUMMARY OF KEY REFERENCE DOCUMENTS 2010 Highway Capacity Manual The 2010 HCM (23) Chapter 17 (Urban Street Segments) refers users to the following three documents for guidelines on selecting the appropriate type of traffic control:  Pline, J. (ed.). Traffic Control Device Handbook. Institute of Transportation Engineers, Washington, D.C., 2001 (26).  Koonce, P., L. Rodegerdts, K. Lee, S. Quayle, S. Beaird, C. Braud, J. Bonneson, P. Tarnoff, and T. Urbanik. Traffic Signal Timing Manual. Report No. FHWA-HOP‐08‐024, Federal Highway Administration, Washington, D.C., June 2008 (27).  Federal Highway Administration. Manual on Uniform Traffic Control Devices for Streets and Highways. Washington, D.C., 2009 (1). Chapters 19, 20, and 32 of the 2010 HCM provide methodologies for TWSC capacity calculations (accounting for pedestrians) and capacity analysis methodology for three-lane AWSC approaches. 2000 Highway Capacity Manual The average vehicle control delay can be determined from equations in the HCM (28). These equations have been developed to analyze the capacity, lane requirements, and effects of traffic and design features of unsignalized intersections. Each type of unsignalized intersection has a set of procedures that address the unique elements of its operation. The procedures have been written to focus on the user-defined analysis period under a steady-state condition, meaning that the traffic volumes and units should be relatively stable over the time period being studied. The HCM cautions against using the method for analysis of any transitional period where units within the intersection are changing, leaving that analysis type to the use of simulation models.

23 The HCM defines LOS by computing or measuring control delay for each movement. These delays are based on the priorities of the traffic streams at the intersection, considering the traffic control devices as applied (or proposed) and the availability of acceptable gaps based on the critical gap and follow-up time. The typical analysis period is the peak-hour turning movement volume factored to reflect conditions during the peak 15 minutes using the peak-hour factor. In practice, the traffic volumes are factored from a peak-hour count to assess the warrants identified in the MUTCD. These factors may be based on a 24-hr tube count or a multi-hour manual turning movement count. The HCM procedure and its delay estimations are often used to assess the potential risk for a motorist making a risky move at an unsignalized intersection. Of interest from a multimodal perspective is that the HCM highlights that pedestrians “must use acceptable gaps in major-street traffic streams, but they have priority over all minor-street traffic at a TWSC.” Chapter 18 describes the LOS criteria for pedestrians at unsignalized intersection and highlights that there is a “high” likelihood of risk-taking behavior (acceptance of short gaps) when delays exceed 30 sec and a “very high” likelihood as delays exceed 45 sec. This is reiterated in Chapter 17: “LOS F may also appear in the form of drivers on the minor street selecting smaller than usual gaps. In such cases, safety may be a problem, and some disruption to the major traffic stream may result. Note that LOS F may not always result in long queues but in adjustments to the normal gap acceptance behavior.” The 2000 HCM (28) also includes a graphic (shown in Figure 3) that was adapted from the 1983 edition of the ITE Traffic Control Device Handbook. The figure can be used to forecast the likely intersection control type based on two-way entering traffic volumes. The figure was generated by converting the 8-hr warrants to two-way peak-hour volumes, assuming ADT equals twice the 8-hr volume, peak hour is 10 percent of daily, and the two-way volumes are 150 percent of peak- direction volume. 2010 Highway Safety Manual Crash Prediction The predicted average crash frequency for an intersection can be determined from equations in the HSM (24). These equations, called safety performance functions (SPFs), are regression models for estimating the predicted average crash frequency of individual roadway segments or intersections for a set of specific base conditions. As discussed in the HSM, each SPF in the predictive method was developed with observed crash data for a set of similar sites. The SPFs, like all regression models, estimate the value of a dependent variable as a function of a set of independent variables. In the SPFs developed for the HSM, the dependent variable estimated is the predicted average crash frequency for an intersection under base conditions, and the independent variables are the annual average daily traffic (AADT) of the major and minor intersection legs.

a Roundabo Source: Ad converted t of daily. Tw Figu SPFs and arterials:  Three  Three  Four-  Four- Other typ by the HS Determin of severa severity. at a stop- The pred factors (C uts may be ap apted from Tr o two-way pea o-way volum re 3. Inters adjustment -leg interse -leg signali leg intersec leg signaliz es of interse M Chapter ing the aver l items inclu There is als controlled in icted averag MFs) and a propriate with affic Control D k-hour volum es assumed to ection Con factors hav ctions with s zed intersect tions with st ed intersecti ctions may 12 SPFs. Th age crash fr ding multip o a step for e tersection. e crash freq calibration in a portion of evices Handb es assuming A be 150 percen trol Type a e been devel top control ions (3SG). op control o ons (4SG). be found on e equations equency pre le-vehicle c stimating th Spreadsheet uency for ba factor to adj 24 these ranges. ook (1983 edit DT equals twi t of peak-direc nd Peak-Ho oped for fou on the mino n the minor urban and s for stop con diction for i ollisions by e number o s are availab se condition ust for a pa ion, pp. 4–18) ce the 8-hr vo tion volume. ur Volume r types of in r-road appro -road approa uburban art trol are of i ntersections severity and f vehicle-pe le to assist s is adjuste rticular geog . Peak-directio lume and peak s (Exhibit 1 tersections ach (3ST). ch (4ST). erials but ar nterest to th requires the single-vehi destrian col in the calcul d using cras raphical are n, 8-hr warran hour is 10 per 0-15 in 28) and suburba e not addres is research. determinat cle collision lisions per y ations. h modificati a. ts cent . n sed ion s by ear on

25 Crash Modification Factor A CMF is a multiplicative factor used to compute the expected number of crashes after implementing a given countermeasure at a specific site. The CMFs currently in the HSM for stop-controlled urban and suburban arterials are intersection left-turn lanes, intersection right- turn lanes, and lighting. The HSM (24) includes the potential crash effects of converting a minor- road stop control into AWSC (see Table 6). The safety findings shown in Table 6 are different for urban and rural settings. This observation indicates that perhaps the MUTCD should have different criteria depending on whether the intersection is in an urban setting or a rural setting. Table 6. Highway Safety Manual (24) Table 14-5. Potential Crash Effects of Converting Minor-Road Stop Control into AWSC. Treatment Setting Traffic Volume Crash Type (Severity) CMF a Standard Error Convert minor-road stop control to all- way stop control (MUTCD Warrants Are Met) Urban Unspecified (assumes that MUTCD warrants for all-way stop control are met) Right-angle (All severities) 0.25 0.03 Rear-end (All severities) 0.82 0.1 Pedestrian (All severities) 0.57 0.2 All types (injury) 0.30 0.06 Rural All types (All severities) 0.52 0.04 a CMF=Crash modification factor, bold text is used for the most reliable, and italic text is for less reliable CMFs. From Highway Safety Manual, 2010, by the American Association of State Highway and Transportation Officials, Washington, D.C. Used by Permission. FHWA Crash Modification Factor Clearinghouse Additional potential CMFs are available on the Crash Modification Factors Clearinghouse website (see Table 7 for examples). FHWA has established the Crash Modification Factors Clearinghouse (29) to provide an online repository of CMFs and crash reduction factors (CRFs) with a searchable database. Searching for CMFs related to STOP signs revealed several studies. The studies that may be relevant to this project include the following:  2006 study on the safety evaluation of STOP Sign In-Fill (SSIF) program (30).  2010 study that used a full Bayes (FB) approach to determine the effectiveness of the SSIF program (31).  2010 study that evaluated the conversion from TWSC to AWSC (32).

26 Table 7. Crash Modification Factors from Clearinghouse (29). Countermeasure CRF CMF Seva Major Min Major Max Minor Min Minor Max Num Lane Convert minor-road stop control to all-way stop control (32)b 77 0.23 F, S, M 680 15,400 680 15,400 1 Convert two-way (without flashing beacons) to all-way stop control (without flashing beacons) (32)b 72.4 0.276 F, S, M 680 15,100 680 15,100 1 Convert two-way (with flashing beacons) to all-way stop control (with flashing beacons) (32)b 86.5 0.135 F, S, M 3,550 13,650 3,550 13,650 1 Convert two-way (without flashing beacons) to all-way stop control (with flashing beacons) (32)2 86.6 0.134 F, S, M 1,340 9,900 1,340 9,900 1 Install STOP signs at alternate intersections in residential areas (30)c 54.8 0.45 All NSd NS NS NS 2 Install STOP signs at alternate intersections in residential areas (30)c 72.3 0.28 F, S, M NS NS NS NS NS Install two-way stop- controlled intersections at uncontrolled intersections (31)e 51.1 0.489 All NS NS NS NS NS a Sev = severity, F = fatal, S = serious injury, M = minor injury. b All crash types; before/after using empirical Bayes; four-leg. c All crash types; simple before/after; 380 sites in Vancouver. e NS = not specified. e All crash types; before/after using empirical Bayes or full Bayes; 513 sites. The 2006 study (30) evaluated the safety impacts associated with the SSIF program. The SSIF program was launched by the Insurance Corporation of British Columbia (Canada) in 1998 and consisted of installing STOP signs alternately at every second intersection in residential neighborhoods in the Greater Vancouver Regional District. This alternating pattern provides consistency in the application of STOP signs within a residential neighborhood. The main objective of the program was to reduce the frequency and severity of collisions and thereby reduce insurance claim costs in addition to providing a traffic-calming effect on residential neighborhoods. The evaluation included a time series analysis to investigate the effectiveness of the SSIF program on road safety performance at 380 intersections. The evaluation used comparison groups and three techniques to determine the safety impacts of the SSIF program. The first two techniques were based on the odds ratio methodology, while the third was based on the likelihood method. The results of the three techniques were consistent and showed that injury collisions were reduced 61 percent to 72 percent, while total collisions were reduced 45 percent to 55 percent. It was concluded that the installation of STOP signs at uncontrolled intersections

27 in residential neighborhoods was an effective measure for reducing both the frequency and severity of collisions in urban areas. A later paper (31) evaluated the effectiveness of the SSIF program by using different modeling techniques. The analysis revealed an overall significant reduction in predicted collision frequency of 51 percent. A study in North Carolina (32) evaluated the conversion from TWSC to AWSC with or without flashing beacons using the empirical Bayes method. The purpose of the project was to develop CRFs for the conversion from two-way to AWSC. A total of 53 treatment sites located in urban, suburban, and rural areas were used in the analysis. The authors divided the treatment locations into three groups based upon the presence of an overhead and/or sign-mounted flashing beacon:  Group 1 consisted of 33 intersections without flashing beacons.  Group 2 consisted of 8 intersections with flashing beacons in the before and after period.  Group 3 consisted of 8 intersections where the flashing beacon was installed with the AWSC. The results from the North Carolina study showed a substantial decrease in total, injury, and frontal-impact crashes in the after period. The recommended CRFs from the overall group are a 68 percent reduction in total crashes, a 77 percent reduction in injury crashes, a 75 percent reduction in frontal-impact crashes, and a 15 percent reduction in ran-STOP-sign crashes. ITE Manual of Traffic Engineering Studies The ITE Manual of Transportation Engineering Studies (25) contains several relevant studies of interest to the topic of unsignalized intersection traffic control, including:  Volume studies (Chapter 4).  Spot speed studies (Chapter 5).  Intersection and driveway studies (Chapter 6), including delay, queue length, gap and gap acceptance, and intersection sight distance.  Traffic control device studies (Chapter 7).  Compliance with traffic control devices (Chapter 8).  Pedestrian and bicycle studies (Chapter 12).  Traffic collision studies (Chapter 17). SELECTING TRAFFIC CONTROL DEVICE FOR UNSIGNALIZED INTERSECTION Safety Studies The North Carolina study by Simpson and Hummer (32) included a comprehensive summary of recent literature on stop-controlled intersections. The following is part of their literature summary: Lovell and Hauer’s study [33], which focused primarily on treatment sites located in an urban environment, is regarded as the most comprehensive review of the safety effects of converting intersections to all-way stop control. They reanalyzed

28 data from three previous safety studies in San Francisco, Philadelphia, and Michigan and added a new data set from Toronto, Canada. Intersections were converted from either two-way stop control or one-way streets to all-way stop control. Reference sites were used to account for regression to the mean. The San Francisco data consisted of one-year before and after comparisons of crashes occurring at 49 urban intersections converted from two-way to all-way stop control between 1969 and 1973. The San Francisco reference data was obtained for a different time frame than the treatment data, from 1974 to 1977. The unbiased results for the San Francisco data showed a 62 percent reduction in total crashes, an 83 percent reduction in right-angle crashes, and a 74 percent reduction in injury crashes. The Philadelphia data contained the largest treatment sample, with 222 urban intersections. The data contained only intersections converted from one-way streets to all-way stop control between 1968 and 1975 and used 2-year before-and-after comparisons. The unbiased results for the Philadelphia data showed a 43 percent reduction in total crashes, a 77 percent reduction in right-angle crashes, and a 73 percent reduction in injury crashes. Along with the data from San Francisco and Philadelphia, the Toronto data contained only urban intersections. The Toronto data analyzed 79 intersections converted from two-way to four-way stop control between 1975 and 1982. The unbiased results for the Toronto data showed a 40 percent reduction in total crashes, a 50 percent reduction in right-angle crashes, and a 63 percent reduction in injury crashes. The Michigan data was the only group pertaining to low-volume, high-speed rural roads and contained a set of 10 intersections. The Michigan data used 2- and 3-year before- and-after periods for intersections converted from two-way to all- way stop control between 1971 and 1977. The reference data was obtained from 1974 through 1976. The unbiased results for the Michigan data showed a 53 percent reduction in total crashes, a 65 percent reduction in right-angle crashes, and a 61 percent reduction in injury crashes. Lovell and Hauer’s study [33] revealed consistent safety effectiveness for all-way stop conversion. In the four data sets, total crashes were reduced by 40 percent to 62 percent, right-angle crashes were reduced by 50 percent to 83 percent, and injury crashes were reduced by 61 percent to 74 percent. Likelihood functions were then used to merge the four sets of results into joint estimates of crash reduction factors. After combining results, they found that the conversion to all- way stop control reduced total crashes by 47 percent and right-angle and injury crashes by 72 percent and 71 percent, respectively. Persaud [34] used the Philadelphia sample converted from one-way streets to all- way stop control in a study that examined how traffic volumes and other issues play a role in crash reductions at urban all-way stops. The results show that the effectiveness of all-way stop conversion in urban areas is not limited to a certain range of entering volumes that follow MUTCD warrants. When analyzing total and right angle crashes, it “can be just as effective for total entering volumes less than 6,000 per day as it is for higher volumes” [34]. The study also showed that for total and right-angle crashes, all-way stop conversion in urban areas is no less

29 effective when approach volumes are unbalanced as when they are equal on all approaches. For rear-end crashes, which make up a small percentage of total crashes, the effectiveness decreases as total entering volumes increase and as the minor road volume drops below 25 percent. The study examined whether there is an increase in crashes in the acquaintance period immediately after conversion and found there is no significant difference in crashes during the first six months after conversion to all-way stop. The study also suggests that the effectiveness of all-way stop control does not decrease as its use becomes commonplace (32). A 2006 paper (35) examined the proper level of traffic control on low-volume rural roads. The authors used 10 years of crash data for more than 6,000 rural, unpaved intersections in Iowa. Stop-controlled intersections were compared to uncontrolled intersections. Crash models were developed with logistic regression and hierarchical Poisson estimations. For ultralow-volume intersections, those used by fewer than 150 vehicles per day, results indicated no statistical difference in the safety performance of each level of control. The authors’ review of the literature found that the most frequent crash factor was not STOP sign violations but failure to yield right of way from the stop position (36, 37, 38), and that other research found that available sight distance at low-volume intersections might have negligible effect on safety and operations (39, 40). NCHRP Report 320 (36) discusses the conversion of stop to yield control. The report found that converted intersections experienced an increase in crashes, the severity and distribution of crashes did not change significantly, and converted intersections had higher crash rates overall than unchanged intersections. According to the study, candidates for conversion to yield control should have adequate sight distance, volume less than 1,800 ADT (1,500 ADT for major roads and 600 ADT for minor roads), and fewer than three crashes in 2 years. A 1983 study (41) compared crash experience at stop-controlled and no-control intersections in rural Michigan and found that there was no statistical difference for intersections with major street volumes less than 1,000 vpd. Polus’s before-after study (42) of hazardous urban intersections where level of control was increased because of crash history (no control to yield control, no control to stop control, and yield control to stop control) showed that increase in control often resulted in more vehicular crashes (although the changes were mostly statistically insignificant), and introducing traffic control at an uncontrolled intersection resulted in reduction in pedestrian crashes. To understand the increase in vehicular crashes with increase in traffic control, Polus studied the gap and lag acceptance characteristics at stop and yield control movements. He concluded that the increase in mean accepted gap value at movements controlled by STOP signs (compared to yield controlled movements) was significant and probably reduced the safety at such movements. A 2000 study (43) that reviewed the effectiveness of various strategies in reducing crashes and concluded an accident modification factor (AMF) of 0.53 for conversion from two-way to all- way stop for total intersection crashes. This value was based on the Lovell and Hauer study discussed earlier (33). In the rural expressway intersection safety toolbox developed for the Iowa Department of Transportation, Hochstein et al. (2011) (44) note the effectiveness of converting

30 TWSC to AWSC to be about 47 to 64 percent, and they include references to the 2000 Harwood et al. study (43), 1984 Briglia study (45), and the CMF Clearinghouse (29). A 2009 study (46) identified about 2,500 unsignalized intersections under 60 categories (based on traffic control, number of lanes, median type, entering volume, etc.), representing nearly all possible types of unsignalized intersections existing. Annual crash profile tables for these categories were developed that can be used as reference values that can assist in identifying unsignalized intersections with specific problems, such as a high number of fatal crashes or high number of rear-end crashes. This information is presented in the form of a database application for easy access. Charbonneau (47) developed modified TWSC warrants in 1995. The modified warrants include: 1. Where a street enters a through street. 2. Where an unsignalized intersection is in a signalized area. 3. Where the safe approach speed is less than 10 mph due to unremovable visibility obstructions, such as a building or topography. 4. Where the crash history indicates three or more reported crashes for the last 3 years that might be corrected by the use of STOP signs. 5. Where an engineering study indicates the application of the normal right of way is unduly hazardous. The study found that two-way STOP sign warrants may not adequately address crash problems, and all-way warrants do not distinguish the wide variation in risk associated with the range of volumes between different levels of streets (i.e., local, collectors, and arterials). It was also observed that crashes decrease at warranted all-way stops and increase at unwarranted stops. A 1998 paper (48) discussed research that developed a method where the safety of a two-way, stop-controlled intersection could be estimated based on parameters such as intersection geometry, traffic volume, pavement conditions, traffic composition, and available sight distance. They used a simulation model to estimate the frequency of potential conflicts or collisions resulting from sight distance restrictions. Table 8 summarizes the LOS categories and the equations that can be used to determine the numeric value. The crash warrants for signals were investigated as part of an NCHRP project (49). A procedure was developed for quantifying the safety effect of signal installation based on the predictive methods in the HSM. The procedure was used to develop revised content for the crash signal warrant. Application of the procedure to a range of typical intersection conditions indicated that there is a threshold volume of observed crashes beyond which signal installation is likely to improve safety. The threshold values were found to vary by area type, intersection legs, and number of lanes on each intersection approach. Table 9 shows the threshold values recommended in the research and recommended for the next edition of the MUTCD.

31 Table 8. LOS for Two-Way Stop-Controlled Intersections Developed Using Simulation (48). Total Number of Conflicts per Crossing Vehicle, Con1 Total Hazard per Crossing Vehicle, HZ2 ([kg-m2/sec2]/104) LOS <0.05 <1.46 A 0.05–0.10 1.49–2.93 B 0.10–0.15 2.93–4.39 C 0.15–0.20 4.39–5.85 D 0.20–0.25 5.85–7.32 E >0.25 >7.32 F 1 Con = 43.1 – 0.092 (AVSDR) + 0.89 (ADT) + 2.30 (Speed) – 0.063 (AVSDL) + 9.45 (T) 2 HZ = –15924 + 1551 (Speed) – 16 (AVSDR) – 10 (AVSDL) + 1.467 (ADT) + 979 (T) Where: ADT = average daily traffic on the major road (thousands of vehicles/day). AVSDL = average sight distance from the left (m). AVSDR = average sight distance from the right (m). Con = total number of conflicts per year per 1,000 crossing vehicles. HZ = total hazard per year per crossing vehicle, used to account for severity and measured as the potential kinetic energy per year per vehicle conflict. Speed = prevailing speed on the major road (km/h). T = trucks on the major road (percent). Table 9. Recommended Crash Numbers from Bonneson et al. (49). Area Type Number of Through Lanes on Each Approach Minimum Number of Reported Crashes in One-Year Period and Three-Year Period Total of Angle Crashes and Pedestrian Crashes (All Severities)b Total of Fatal-and-Injury Angle Crashes and Pedestrian Crashesb Major Minor Four Legs Three Legs Four Legs Three Legs Urban 1 1 5 (6)c 4 (5) 3 (4) 3 (4) 2+ 1 5 (6) 4 (5) 3 (4) 3 (4) 2+ 2+ 5 (6) 4 (5) 3 (4) 3 (4) 1 2+ 5 (6) 4 (5) 3 (4) 3 (4) Rurala 1 1 4 (6) 3 (5) 3 (4) 3 (4) 2+ 1 10 (16) 9 (13) 6 (9) 6 (9) 2+ 2+ 10 (16) 9 (13) 6 (9) 6 (9) 1 2+ 4 (6) 3 (5) 3 (4) 3 (4) a Rural values apply to intersections where the major-road speed exceeds 40 mph or intersections located in an isolated community with a population of less than 10,000. b Angle crashes include all crashes that occur at an angle and involve one or more vehicles on the major road and one or more vehicles on the minor road. c Reported crashes for the three-year period appear in parentheses. Capacity and Volume Studies A 1983 ITE paper by Upchurch (50) developed a procedure for selecting the most economical type of sign control at an intersection. The guidelines were developed based on an economic analysis that quantified the effect of each sign type (yield, two-way stop control, and four-way

32 stop control) in terms of intersection operation costs. These costs include fuel costs; vehicle operating costs; the cost of delay to motorist and passengers; air pollution costs; crash costs; and sign material, installation, and maintenance costs. The costs were evaluated for various intersection conditions using a traffic simulation model and published crash prediction equations. Based on the crash rates used in the study, yield control was found to be more economical than the two types of stop control. Both stop controls were found to have capacity limits beyond which they did not provide a satisfactory LOS. The paper estimates that by using the proposed more efficient sign control selection procedure, the nationwide intersection operating costs could be reduced by as much as $15.1 billion per year. A 1988 ITE paper (51) reviewed issues related to traffic management in residential areas and developed a decision-making framework for uniform and effective traffic control implementation. Specific criteria for traffic control installation at urban residential (low-volume) intersections were not included in the 1980 MUTCD. The authors proposed a set of criteria (shown in Table 10) based on network consideration, traffic volume, crash history, sight distance, and speed patterns. Table 10. Criteria for Various Traffic Control (Table 1 in 51). Traffic Control Network Function Traffic Volume Crash History Sight Distance Minimum SASa No Control Local/Local <1,500 vpd intersection volume 0–2 crashes per year Posted speed limit, all approaches Yield Local/Collector Local/Local 1,500–3,000 Pattern ≥ 2 per year in 3 years ≥10 mph Two- Way Local/Local Local/Collector Collector/Collector ≥3,000 ≥3 per year with pattern <10 mph Multi- way Collector/Collector See MUTCD ≥5 per year with pattern <10 mph, highly restricted visibility on opposing approaches a SAS = safe approach speed In 1995, Box (52) developed guidelines for use of traffic control signs at low-volume urban intersections. He recommended consideration of roadway classification, crash history, and safe approach speed in determining the most appropriate control mode. Box’s recommendations were incorporated into a table by Bonneson et al. (53), which is reproduced in Table 11. Box (52) indicates that this table should only be used for intersections with a total entering traffic volume of 300 veh/hr or less during the peak hour. He also cautions that the no-control or yield-control options may not work well when the total entering volume exceeds 100 veh/hr.

Roadway Classifica Major Local Local Collector Collector Collector Collector a The tabl during the b Collision on the low c Approac d None m hour. A 1997 I hour inte delay, an HCM me all appro overall in for TWS the split b and the m pedestria Table 11 tiona Minor Local Local Local Local Collector Collector e is only appli peak hour. Tw s susceptible er-volume ap h speed for mi eans no contro TE paper (5 rsection volu d intersectio thodologies aches, and in tersection v C, A stands etween TW inor-street ns and bicyc Figure 4. . Candidat Crash Historyb 1 yr 3 y <2 <4 ≥2 ≥4 <2 <4 ≥2 ≥4 <2 <4 ≥2 ≥4 cable to interse o-way stop, m to correction b proach. Two c nor-road drive l at intersectio 4) provides mes. Three n queue len . Intersectio tersection q olume. In th for AWSC, SC and AW volume is ap lists. Optimal In e Control fo Minor-R Speedc ( r <10 Stop Stop Stop Stop Stop Stop ctions in urba ulti-way stop y stop or yield ollisions in a 1 rs; based on a n. May be limi a guide for s graphs wer gth. The gra n LOS and d ueue length e graphs (sh and S stand SC occurs w proximately tersection C 33 r Minor-R oad Control mph) of… 10 to Stop Stop Stop Stop Stop Stop n areas with a , or signal con control (e.g., 2-month perio n evaluation of ted to a total e electing inte e developed phs incorpo elay were c was compu own in Figu s for a traffi hen the ma 210 veh/hr ontrol Bas oad Approa for Minor-Ro 20 2 Y Y Y total entering v trol should be right-turn, left d or four in a 3 their sight dis ntering volum rsection tra based on in rate 1988 M alculated us ted from av re 4, Figure c signal. Fig jor-street vo . The study ed on LOS ch by Box ad Sight Dist 1 to 30 ield ield or Stop ield Stop Stop Stop olume of 300 considered for -turn, and righ -year period. tance to major e of 100 veh/h ffic control tersection L UTCD warr ing the weig erage interse 5, and Figu ure 5 and Fi lume is abo did not expl (Figure 1 i (52). ance Approac ≥30 Noned Yield Yield Yield Yield or S Yield or S veh/hr or less higher volum t-angle collisio -road vehicle. r during the pe based on pe OS, intersec ants and 19 hted averag ction delay re 6), T stan gure 6 show ut 1,000 veh icitly consid n 54). h top top es. ns) ak ak- tion 94 e of and ds that /hr er

Figure 5 Figure Another performa selection (CORSIM character vehicle w combinat right) and . Optimal 6. Optimal ITE paper fr nce of stop- of appropri ) was used istics and ob as chosen a ions (10 left 50 volume Intersection Intersection om 2004 (5 controlled in ate stop cont to analyze d tain the ass s the best m -80 through combinatio Control B (Figu Control B 5) found tha tersections; rol based on ifferent com ociated dela easure of ef -10 right, 20 ns for each t 34 ased on Ave re 2 in 54) ased on Av t turning per the paper c turning mo binations o y in each ca fectiveness f left-60 thro urning distr rage Delay . erage Queu centages ha ontained dis vements. C f vehicular se. Average or the study ugh-20 righ ibution for e (5-sec Sign e Length (F ve a major i cussion of g orridor Simu volumes and control dela . Three turn t, and 30 le ach type of ificance Le igure 3 in mpact on th uidelines fo lation road y in second ing distribu ft-40 throug stop contro vel) 54). e r s per tion h-30 l

were run stop is pr increases point wh point is t show the Figure A 2008 s Highway 5,000 cas methodo The study choice of 10 percen recomme The auth is found t minor str streets se should be , resulting in eferred at lo , delay valu ere either co ermed the tr transition p 7. Major S tudy (56) ve Capacity S es. The resu logies, and i also found intersection t, 15 percen ndations fro ors conclude hat if deman eets, two wa e low to me favored.” a total of 3 wer traffic v es for two-w ntrol can be ansition poin oint volume treet–Mino and a On rified HCM oftware) for lts showed t was recom that the per control typ t, and 20 pe m the study d the follow d is unbala y-stop cont dium traffic 00 simulatio olumes due ay and four used, beyon t in the stud s for major r Street Vo e-Lane Mi 2000 Exhib the estimati that Exhibit mended tha centage of l e. Graphs w rcent left tu and Exhibit ing: “On th nced betwee rol should b , all-way-sto 35 ns. The ana to lower de -way stop co d which fou y. The grap and minor st lume Relati nor Street ( it 10-15 usi on and com 10-15 was i t the graphs eft-turning v ere develop rns. Figure 8 10-15 of H e basis of th n major and e used; if de p control is lysis showed lays, and as ntrol becom r-way stop h shown in reets for var onship for Figure 6 in ng HCM 20 parison of c nconsistent developed i ehicles has ed for no lef shows a co CM 2000, w e criterion o minor stree mand is som preferred; o that the us the intersec e closer un control is pr Figure 7 wa ious turning a Two-Lane 55). 00 methodo ontrol delay with the res n the study b a significan t turns as w mparison o ith 10 perce f minimizin ts and if the ewhat bala therwise, si e of a two-w tion volume til they reac eferred. Thi s developed distribution Major Str logies (and t for more th ults from HC e used inste t impact on ell as 5 perc f control typ nt left turns g delay alon traffic is lo nced and mi gnal control ay h a s to s. eet he an M ad. the ent, e . e, it w on nor

Figure 8 In 2012, with the performa and left-t 5, 10, 15 roundabo was prod intersecti A numbe TWSC an Some of length ca NCHRP improvem models to step-wise modes at the LOS movemen delay and Figure 10 effectiven . Comparis Jiang et al. ( options of tw nce index fo urn volumes , and 20 perc ut (yield) co uced as the on operation r of studies d AWSC in these impro lculation). Report 457 ent alterna evaluate th process for a problem i (threshold o t or 5 vehic LOS in ter ). Finally, t ess and its on of Contr 57) also dev o-way stop r type selec ; they comp ent of the v ntrol. The r basis for cho s. have focuse tersection c vements are (53) docume tives and foc e operationa evaluating ntersection. f LOS D) or le-hours for ms of averag he guide em other, non-m ol Type wi Turns ( eloped a set , signal, and tion. The re leted over 2 olume. Each esulting set osing inters d on develop apacity anal incorporate nts the step uses on the l impacts o the operatio The effectiv total delay multi-lane m e control de phasizes tha otorist-rela 36 th HCM 20 Figure 7 in of charts fo roundabout searchers co 4,000 simul scenario w of charts, an ection contr ing method ysis method d in the 201 s involved in use of capa f traffic cont nal effects o eness of an (threshold o ovement). lay, based o t the best al ted effects. 00 Exhibit 56). r selecting i . The charts nsidered 8,1 ation runs, u as run for si example of ol in light o ologies that ologies prov 0 HCM (e.g the formal city analysis rol alternati f alternative alternative i f 4 vehicle-h A graph com n HCM 200 ternative is 10-15, with ntersection were based 60 combina sing left-tu gnal control which is sh f anticipated improve or ided in the ., 95th perce engineering procedures ves. The gu geometrics s identified ours for sin bining the 0 is provide selected on t 10 Percent control type on LOS as tions of dem rn percentag , TWSC, an own in Figu benefits fo supplement 2000 HCM ntile queue study of and simulat ide provides and control based on eit gle-lane effect of bo d (shown in he basis of Left s, the and es of d re 9, r the . ion a her th its

Figure 9 Fi . Intersectio gure 10. Ac n Control ceptable Op Type and P Tr erating Co 37 eak-Hour V affic (57). nditions at olumes wi Unsignaliz th 10 Perce ed Intersec nt Left-Tur tions (53). ning

38 In a 2000 paper, Wu (58) discusses a new capacity analysis methodology based on the additional-conflict-flow method (developed from graph theory) as an alternative to the traditional gap acceptance method. This methodology takes into account the number of pedestrians per approach, which is not included in the HCM AWSC methodology. Brilon and Miltner (59) developed a method for evaluating capacity at unsignalized intersections based on the influence of pedestrians and bicyclists. Called the conflict technique, their method allows practitioners to consider the influence of nonmotorized road users on motor vehicle operations. Moreover, the method simplifies the theoretical approach. Different modalities of operation, such as a pedestrian crossing at the entries to an intersection, can be considered, as can the fact that some road users do not comply with priority rules. To calibrate the calculation method, traffic at several intersections was observed by video and analyzed for traffic volume, delay, compliance with priority rules, and other parameters. With these field measurements, the calculation method was calibrated to actual road-user behavior. Comparison of the conventional calculation concept based on gap acceptance and the new conflict technique showed that they provide similar results. In particular, the authors concluded that consideration of pedestrians and limited priority effects is a considerable benefit of the new method. Gard (60) developed empirical equations to predict the maximum queue length for major-street left turns and minor-street movements at TWSC intersections. The regression equations were found to closely fit the data (40 percent of the 184 observed maximum vehicle queues were correctly predicted, and 85 percent were predicted within one vehicle). Tian and Kyte (61) also developed an empirical model for estimating the 95th percentile queue length for AWSC approaches and showed that the methodology for predicting queues at TWSC intersections can be applied to AWSC intersections. This finding is incorporated in the 2010 HCM. Kirk et al. (62) conducted a study to use operational characteristics to determine the size and the design of intersections based upon a targeted level of operation. This approach was designed to allow for a preliminary evaluation of a broader range of possible designs, by screening out those designs considered less desirable or inappropriate on the basis of operational performance. An intended benefit of this approach was to also allow for a more objective comparison of all alternatives because all options targeted the same operational service level. The use of the critical lane analysis method was considered an appropriate approach for developing size estimates for intersections. Similar methods for stop-controlled and yield-controlled intersections were also identified because it was necessary to expand these methods to include unsignalized designs as well. The result of the project was the development of the Intersection Design Alternative Tool, capable of evaluating 13 intersection alternatives and identifying preferred lane configurations from more than 12,000 available configurations. The tool identifies the most efficient design (minimum number of lanes) that is capable of meeting a targeted level of operation. A designer is presented with several options that meet the minimum operational requirements, allowing examination of other trade-offs such as right-of-way impacts, safety considerations, and the like. This approach eliminates the need to compare alternatives with varying operating levels across different types of traffic control. The proposed approach aims to provide greater efficiency in the

evaluatio operation appropria typical de SELECT CROSSI 2009 Ma Figure 11 based on develope NCHRP NCHRP crossing to check 20 pedes exceeds 3 (rather th extension Traffic si NCHRP n and conce al efficiency te and prop signs. ING TRAF NGS nual on Un shows part major-stree d based upo Figure 11 Report 562 Report 562/ treatments. A the pedestria trians per ho 5 mph). If f an signs, sig s are alterna gnal warran Report 562 ptual design and impro erly customi FIC CONT iform Traff of the 2009 t volume, pe n the finding . 2009 MU /TCRP Rep TCRP Repo fter selecti n volume. T ur for both ewer pedest nals, or mar tives that ca ts in the MU analysis. If o of intersect ved safety p zed design ROL DEV ic Control D MUTCD p destrian vol s from NCH TCD Pedes ort 112 rt 112 (22) p ng the prope he minimu directions (1 rians are cro kings) such n be consid TCD can al ne or more 39 ion alternati erformance. for each inte ICES FOR evices edestrian hy ume, crossin RP Report trian Hybri resents guid r speed cate m pedestrian 4 pedestrian ssing the st as traffic ca ered. so be evalu signal warra ves, with th The approa rsection, av UNSIGNA brid beacon g length, an 562/TCRP d Beacon G elines for th gory, the gu volume for s per hour i reet, then ge lming, med ated for the nts are met e intent to a ch allows fo oiding the u LIZED PE guidance. T d speed. Th Report 112 uidance Fi e applicatio idelines cal a peak-hou f the major- ometric imp ian refuge is intersection , then a sign chieve great r a more se of standa DESTRIAN he decision e guidance (22). gure. n of pedestr l for the eng r evaluation road speed rovements lands, and c to support t al can be er rd or is was ian ineer is urb he

considere pedestria  Cross signs  Enha locati categ  Activ displa  Red: beaco  Signa The guid appropria crossing spreadsh location. Figu City of T The City pedestria their PHB priority e based on d; otherwis n delay. The walk: This , as opposed nced: This c on and pede ory are pres e: Also call y a warning This categor n) to motor l: This categ elines also p te for a give distances, an eet tool are a re 12. Guid ucson of Tucson, n crossing tr installation valuation fo the characte e, the engine report disc category enc to unmarke ategory incl strians wait ent or active ed active wh only when y includes t ists at the pe ory pertain rovide a ser n location. d walking s lso availabl elines Plot, Arizona, (63 eatments. B policy is o rm consistin ristics of th er can cons usses five ca ompasses s d crossings. udes those d ing to cross. at the cross en present, pedestrians hose device destrian loc s to traffic c ies of plots t The plots co peeds (see F e for a user 34-ft Pavem ) has a num ecause they ften referred g of several e site under 40 ider other de tegories of tandard cros evices that Warning si ing location this category are present s that displa ation. ontrol signa o assist the rrespond to igure 12 fo to enter the ent, ≤ 35 m ber of guide developed t to by other questions, consideratio vices in the devices: swalk mark enhance the gns, markin at all times includes th or crossing t y a circular ls. engineer in specific com r an exampl specific cha ph, 3.5 ft/s lines in plac he pedestria jurisdiction with points n. context of e ings and ped visibility of gs, or beaco . ose devices he street. red indicatio determining binations o e). Paper wo racteristics o ec Walking e for the ins n hybrid be s. Their pol assigned to stimated estrian cros the crossin ns in this designed to n (signal or which devi f speeds, rksheets an f a particul Speed (22) tallation of acon (PHB) icy used a each answer sing g ce is d a ar . ,

41 City of Phoenix The City of Phoenix, Arizona, (64) adopted guidelines similar to Tucson’s for installation of PHBs. Key differences between guidelines in Tucson and Phoenix are:  Crashes receive twice as many points in Phoenix as in Tucson.  Phoenix gives additional points for very high (>40) crossing counts.  Phoenix has more subdivisions of distances to the nearest controlled crossing.  Phoenix accounts for the number of through lanes.  Phoenix provides for “unique circumstances.” The Phoenix guidelines state that locations with fewer than 30 total points should not be considered for PHB installation. Unmarked locations should be considered for signing/striping enhancements before PHB installation is considered. Locations where a signal warrant exists will not be considered for PHB installation. Arizona Department of Transportation The Arizona Department of Transportation (ADOT) has developed a set of draft PHB installation guidelines based on existing guidance in Tucson and Phoenix, as well as ADOT’s own Pedestrian Safety Deficiency Index (65). ADOT’s draft guidelines state that there are many possible treatments to improve pedestrian crossings, including, but not limited to, marked crosswalk, high-visibility crosswalk, two-stage crosswalk, median refuge, street lighting, in- pavement lights, rectangular rapid flash beacon (RRFB), PHB, and pedestrian signal. A comprehensive evaluation of pedestrian crossing safety should be conducted in order to identify the most effective treatment. A minimum total score of 35 points merits consideration of a PHB, and ADOT advises that PHBs should not be installed on roadways with speed limits greater than 45 mph. The draft guidelines are shown in Table 12.

42 Table 12. Arizona DOT PHB Evaluation Draft Guidelines (65). Question Points 1. Motor vehicle crashes correctable by installation of PHB (most recent 5 years of data) involving pedestrians, bicyclists, wheel chairs, skateboards, motorized scooters, or golf carts crossing within 500 feet on either side of the proposed PHB location, or half the distance to the nearest signal (whichever is less): 5 points per crash 2. Average peak hour pedestrian crossing volume within 500 feet on either side of the proposed PHB location, or half the distance to the nearest traffic signal (whichever is less): 0–10 0 points 11–20 2 points 21–39 4 points 40 + 6 points 3. Location of nearest existing traffic signal or existing PHB: Less than 500 ft −5 points 500–1000 ft 0 points Over 1000 ft 10 points 4. Posted speed limit: Under 30 0 points 30 and 35 2 points 40 and 45 4 points 5. Roadway traffic volume (ADT): Less than 5000 0 points 5000–9999 2 points 10000–14999 4 points 15000 + 6 points 6. If the roadway does not have a raised median with a minimum width of 6 feet: 5 points. 7. If a designated, maintained, and permitted shared-use path or walkway crosses the road at the proposed PHB location: 5 points 8. If the proposed PHB location is within 500 feet of a senior center, medical facility, community center, school or other pedestrian activity generator: 5 points 9. If the proposed PHB location does not have roadway illumination: 5 points 10. If the crossing distance is greater than 36 feet: 5 points. If a raised median with a minimum width of 6 feet is present, the crossing distance is measured to the median. TOTAL Additional factors to be considered when a crossing merits PHB consideration: • Is the location within a coordinated signal network? • Does the roadway environment support the installation of the PHB? Does the street have adjoining sidewalks and/or pathways that will result in a logical utilization of the PHB? • Is right-of-way needed? Are there utility conflicts? Is there significant potential for environmental or cultural issues? • Is funding of the PHB available? • Is 120/240 single phase power available at a reasonable cost? Does the local jurisdiction support the installation of a PHB? Is the local jurisdiction willing to pay for the power for the PHB? Is the local jurisdiction willing and capable of accepting the maintenance and operation of the PHB? Will the local jurisdiction pay the power for lighting the crosswalk? City of Boulder The City of Boulder, Colorado, has pedestrian crossing treatment installation guidelines that use the minimum pedestrian volume thresholds for the installation of any pedestrian crossing treatment (e.g., marked crosswalks, RRFB crossings, and underpasses) (66). A unique element of

the Bould threshold The city on the pr state that both high be an inc In these c crosswalk pedestria been prep a similar T Minimum 20 peds pe 18 peds pe 15 peds pe 10 school- * Young, e ** School Figure 1 Virginia The Virg as follow er criteria i s (Table 13) revised their ocedures an the use of R traffic volu rease in traf ases, the us (HAWK) b n crossing s ared to aid figure exist able 13. Bo Pedestrian Vol r hour* in any r hour* in any r hour* in any aged pedestrian lderly, and dis Crossing defin 3. City of B Departmen inia Departm s: s that young . guidelines d considerat RFBs may mes and hig fic crashes a e of convent eacons may hould be bas in this deter s for high-sp ulder, Colo ume: one hour, or two hours, or three hours s traveling to/ abled pedestri ed as a crossin oulder Ped b t of Transp ent of Tran , elderly, an in Novembe ions for inst not be appro h pedestrian nd/or traffic ional pedest be more ap ed on engin mination. Fi eed roadwa rado, Mini from school** ans count twic g location wh estrian Cro y Fox, Tutt ortation sportation ( 43 d disabled p r 2011; thos alling pedes priate in loc volumes. I delay that m rian traffic propriate. W eering judgm gure 13 is fo ys. mum Pedes in any one ho e towards volu ere ten or more ssing Treat le, Hernan VDOT) (66 edestrians c e guidelines trian treatm ations wher n these extre akes the us signals or hi hile the de ent, the lim r low-speed trian Volum ur me thresholds student pedes ment Instal dez (67). ) has propos ount twice t provide mu ents (67). Th e there is a me conditio e of RRFBs gh-intensity cision not to it line in Fi (35 mph or e Thresho trians per hou lation Guid ed RRFB in owards volu ch greater d e guideline combination ns, there m inappropria activated use RRFBs gure 13 has less) roadw lds (66). r are crossing elines prep stallation cr me etail s of ay te. at a ays; ared iteria

44  At least 20 pedestrians crossing in the highest hour and,  There is a marked crosswalk existing or justified at the location and,  Other applicable pedestrian options have been reviewed and determined by engineering judgment to not be applicable. Pedestrian counts, a crossing gap study, and other key pieces of data must be obtained before and after installation. Washington County, Oregon In 2010, commissioners in Washington County, Oregon, changed their policy on midblock crossings (68). Previously, Washington County had approved pedestrian crossings only at road intersections, with few exceptions. However, with the increasing demand for pedestrian and bicycle facilities (e.g., trails) that cross the street network at locations other than intersections, the county decided it was appropriate to review and change the county’s policy and practice. The new policy authorizes the county engineer to approve a modification or design exception under the appropriate county code for a midblock crossing. The application for a midblock crossing requires the applicant to describe the need for the crossing, document the current and anticipated characteristics of the roadway and adjacent area (including transit service, land use, and nearby pedestrian generators), and conduct a pedestrian and vehicle volume count and a gap analysis. Midblock crossing treatments are organized into a progressive tier system shown in Table 14. Table 14. Washington County, Oregon, Tiered Midblock Crossing Treatments (68). Tier Standard Additional Treatments Considered Tier One Crosses a 2-lane street with or without an island/refuge – install high-visibility mounted signs and markings Refuge islands, curb extensions, staggered pedestrian refuges Tier Two Crosses a 3-lane street with an island/refuge – install high- visibility signs and markings Flashing beacons, pedestrian- actuated signal/beacon Tier Three Crosses a 2-lane street without an island/refuge or a 4-lane street with island/refuge – install high-visibility signs and markings or pedestrian-actuated signal Pedestrian-actuated signal/beacon Tier Four Crosses a 4-lane or greater street without an island/refuge – install pedestrian-actuated signal or beacon Pedestrian-actuated signal, pedestrian over- or undercrossing County guidelines include the use of the table produced by Zegeer et al. (69) for FHWA that provides recommendations for installing pedestrian treatments at uncontrolled locations based on ADT. Texas Department of Transportation In December 2012, the Texas Department of Transportation (TxDOT) distributed guidelines regarding PHBs (70) and guidelines regarding RRFBs (71). All of the following conditions must be met before one of these devices can be considered:  An engineering study must be performed and meet the guidelines detailed in Chapter 4F of the Texas MUTCD.  The location has an established crosswalk with adequate visibility, markings, and signs.

45  The posted speed limit is 40 mph or less (does not include school speed zones).  The location has 20 or more pedestrians crossing in 1 hr.  The location is deemed a high-risk area (e.g., schools and shopping centers).  The crosswalk is more than 300 ft from an existing traffic-controlled pedestrian crossing.