ARIZONA DEPARTMENT OF TRANSPORTATION REPORT NUMBER: FHWA-AZ91-321 COMPARATIVE ANALYSIS OF LEADING AND LAGGING LEFT TURNS Final Report Prepared by: Jim C. Lee Robert H. Wortrnan David J.P. Hook Mark J. Poppe Lee Engineering, Inc. 2701 East Camelback Road Phoenix, AZ 85016 August 1991 Prepared for: Arizona Deparlment of Transporlation 206 South 17th Avenue Phoenix, Arizona 85007 in cooperation with U.S.Department of Transportation Federal Highway Administration The contents of this report reflect the views of the authors who are responsible for the facts and the accuracy of the data presented herein. The contents do not necessarily reflect the official views or policies of the Arizona Department of Transportation or the Federal Highways Administration. This report does not constitute a standard, specification, or regulation. Trade or manufacturerts names which may appear herein are cited only because they are considered essential to the objectives of the report. The U.S. Government and the State of Arizona do not endorse products or manufacturers. Report No. ~Hw~-~Z-91-321 4. Title and Subtitle 1. Technical Report Documentation Page ( 2. Ciovernment Accession No. 1 3. Recipient's Catalog No. 1 I 5. Report Date August, 1991 6. Performing Organization Code Comparative Analysis of Leading and Lagging Left Turns 3. Author@) 8. Performing Organization Report No. Jim C. Lee, Robert H. Wortman, David J.P. Hook, Mark J. Poppe 9. Performina Oraanization Name and Address Lee ~ngin'eerfhg,Inc. 2701 E. Camelback Road I Phoenix, AZ 85016 12. Sponsoring Agency Name and Address ARIZONA DEPARTMENT O F TRANSPORTATION 206 S. 17TH AVENUE PHOENIX, ARIZONA 85007 10. Work Unit No. 11. Colltract or Grant No. HPR-PG1-35(321) 13. Type of Report & Period Covered Final 8/89 - 8/91 14. Sponsoring Agency Code I 16. Abstract This project was undertaken to compare operational and safety charactcrlstics of leadlng verses lagging protected left turn operation. The measures of effectiveness included fleld measured Intersection delay with leadlng and with lagging left turns. Intersections in Pima County, Glendale, Tempe and Mesa, Arizona were studied wlth bolh leading and lagging operation. Intersection delay studies were also done with flrst car verses third car actuation of leading protected left h r n phases. Signal progression was studied with leading, lagging and combhation of leading and lagging left turns whlch provlded tbe best progression. This evaluation was accomplished with an instrumented vehicle and travel time runs throughout a grid. Accident studies were conducted In Tucson, Pima County and Scottsdale, Arizona. These studies compared accident frequency before and after the convwslon from leading to lagging left turns. A pubilc opinion survey was conducted to obtain motorists1 preference of leading or lagging left turns. Distribution Statement Document is available to the U.S. public through the National Technical information Service, Springfield, Virginia 22161 19. Security Classification (of this report) 20. Security Classification (of this Unclassified page) Unclassified 1 I 15. Supplementary Notes Prepared In cooperatlon with the U.S. Department of Transportation, Federal Highway Administration 17. Key Words Traftlc signals, left turn phasing, third car actuation, accident studies, travel time studies, traffic signal progression, intersection delay, public opinion, motorists1preference J 209 TABLE OF CONTENTS Executive Summary . Part I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 .Introduction Chapter 1 Chapter 2 .lntroductioii . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 .Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Part I1 .Accident Studies Chapter 3 .Scottsdale Accident Analysis . . . . . . . . . . . . . . . . . . . . . . . . 25 Chapter 4 .Pima CountyDucson Accident Studies . . . . . . . . . . . . . . . . . . 33 Part 111 .Traffic Operations Studies Chapter 5 .Phoenix Area Intersection Delay Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Chapter 6 .Pima County Traffic Signal Operation Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 Chapter 7 .Phoenix Area Travel Time Study Analysis . . . . . . . . . . . . . . . 71 Part IV . Public Perception Studies Chapter 8 Part V .Public Awareness and Perception . . . . . . . . . . . . . . . . . . . . . 89 .Conclusions Chapter 9 .Study Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 Chapter 10 .Theoretical Analysis of Leading and Lagging Left T u r n . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 Chapter 11 .Recommendation for Future Work . . . . . . . . . . . . . . . . . . . . 107 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 Appendices Appendix A .Phase Selection Guideline for Left-Turn Phases . . . . . . . . . . . . . . A-1 Appendix B .Scottsdale Accident Analysis . . . . . . . . . . . . . . . . . . . . . . . . . B-1 Appendix C .Pima CountyDucson Accident Analysis . . . . . . . . . . . . . . . . . . . C-1 . . . . . . . . . . . . . . . . . . . . .D-1 E .Pima County Intersection Analysis . . . . . . . . . . . . . . . . . . . . . . E-1 F .Travel Time Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F-1 G .Dallas Area Special Phasing Sequence . . . . . . . . . . . . . . . . . . .G-1 Appendix D .Phoenix Area Intersection Analysis Appendix Appendix Appendix LIST OF FIGURES ....... Phase Overlap with Protected Only Lagging Left . . . . . . . . . . . . . Leading vs . Lagging Left Turn Delay. Phoenix Area . . . . . . . . . . . . Leading vs . Lagging ThroughIRight-Turn Delay. Phoenix Area . . . . . Leading vs . Lagging Total Intersection Delay. Phoenix Area . . . . . . . Fiyre 2-1 Potential Safety Problem: Lagging Left with Phase Overlap 16 Figure 2-2 17 Figure 5-1 Figure 5-2 Figure 5-3 Figure 5-4 Leading vs . Combination Delay. Phoenix Area 52 52 53 ................ 54 .......... 57 Figure 5-5 Third Car vs. First Car Left Turn Delay. Phoenix Area Figure 5-6 Third Car vs. First Car Throughmight-Turn Delay. Phoenix Area .... 57 Figure 5-7 Third Car vs. First Car Total Intersection Delay. Phoenix Area 58 Figure 7-1 Lead-Lag Study Area. City of Glendale ..... .................... 76 ............... ..................... Figure 7-2 Travel Time Study Cost/Hour. City of Glendale 79 Figure 7-3 Lead-Lag Study Area, City of Tempe 80 . . . . . . . . . . . . . . . . 82 Figure 7-5 Lead-Lag Study Area, City of Mesa . . . . . . . . . . . . . . . . . . . . . . 83 Fi y re 8-1 Glendale Questionnaire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 Figure 11-1 Hierarchy of LeEt Turn Uniformity . . . . . . . . . . . . . . . . . . . . . . . 108 Figure 7-4 Travel Time Study Cost/Hour. City of Tempe iii LIST OF TABLES Table 2-1 Table 2-2 Table 2-3 ......................... City of Scottsdale Accident Data . . . . . . . . . . . . . . . . . . . . . . . . City of Scottsdale Delay Study 18 19 Relative Left Turn Accident Rate for Various Left-Turn Signalization Schemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Table 3-1 Accident Analysis Test Intersections. City of Scottsdale 26 Table 3-2 26 Table 3-3 .......... Accident Analysis Control Intersections. City of Scottsdale . . . . . . . . Accident Analysis Summary Count. City of Scottsdale . . . . . . . . . . . Table 3-4 Accident Analysis Test for Comparability. City of Scottsdale ...... 29 Table 3-5 ......... Accident Analysis Results. City of Scottsdale . . . . . . . . . . . . . . . . Number of Left-Turn Accidents. Pima County . . . . . . . . . . . . . . . . Equivalent number of Accidents per Year. Pima County . . . . . . . . . Estimated Approach Volumes. Pima County . . . . . . . . . . . . . . . . . Left-Turn Accident Rate. Pima County . . . . . . . . . . . . . . . . . . . . Left-Turn Accidents. City of Tucson ArterialfArterial Intersections . . . 30 Table 3-6 Table 4-1 Table 4-2 Table 4-3 Table 4-4 Table 4-5 Accident Analysis Test of Treatment. City of Scottsdale 28 31 34 36 37 39 42 Average Left-Turn Accidents per Year. City of Tucson Arterial/Arterial Intersections. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Table 4-7 Left-Turn Accident Rate. City of Tucson ArterialIArterial Intersections 44 Table 4-8 Left-Turn Accidents. City of Tucson ArterialICollector Intersection Table 4-9 Left-Turn Accidents per Year. City of Tucson ArterialfCollector Intersections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Table 4-6 ... Table 4-10 Left-Turn Accident Rate. City of Tucson Arterial/Collector Intersection Table 5-1 Leading vs. Lagging Intersection Delay. Phoenix Area Table 5-2 Leading vs. Combination Intersection Delay. Phoenix Area Table 5-3 Third Car vs . First Car Intersection Delay. Phoenix Area ........... ........ ......... 45 46 51 55 56 . . . . . . . . . . . . . . . . . . . . 62 Pima County . . . . . . . . . . . . 65 Table 6-1 Delay Study Intersections. Pima County Table 6-2 Intersection Total Approach Volumes. . . . . . . . . . . . 66 Vehicle Delay Comparison. Pima County . . . . . . . . . . . . . . . . . . . 67 Average Cycle Length. Pima County . . . . . . . . . . . . . . . . . . . . . . 68 Table 6-3 Percent of Approach Vehicles Stopped. Pima County Table 6-4 Table 6-5 Table 7-1 Travel Time Study Possible Signal Combinations. Phoenix Area .... 72 Table 7-2 Travel Time Study Utilized Signal Combinations. Phoenix Area ..... 72 ................. 73 Table 7-3 Lowest Cost Timing Plans. City of Glendale Table 7-4 Lowest Cost Timing Plans. City of Tempe Table 7-5 Travel Time Results. City of Glendale . . . . . . . . . . . . . . . . . . 74 ..................... 77 ............. 78 ..................... 81 Table 7-6 Travel Time Study Comparisons. City of Glendale Table 7-7 Travel Time Results. City of Tempe . . . . . . . . . . . . . . 82 Travel Time Studies Leading Minus Combination. City of Mesa . . . . 84 Public Awareness and Perception. City of Glendale . . . . . . . . . . . . . 92 Public Awareness and Perception. City of Tempe . . . . . . . . . . . . . . 93 Public Awareness and Perception. Composite . . . . . . . . . . . . . . . . . 94 Public Perception Results. Phoenix. Area . . . . . . . . . . . . . . . . . . . 100 Table 7-8 Travel Time Study Comparisons. City of Tempe Table 7-9 Table 8-1 Table 8-2 Table 8-3 Table 9-1 Table 10-1 Decision Variables for comparing Leading and Lagging Left-Turn Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 EXECUTIVE SUMMARY COMPARATIVE ANALYSIS OF LEADING AND LAGGING LEET TURNS INTRODUCTION In 1985, the City of Tucson initiated an effort to convert the protected left turn signal phases from a leading to lagging operation. It was believed that the use of lagging Icft turns would improve intersection operations and network flows. The limited studies following the conversion from leading to lagging left turn operations suggested that operational andsafety advantages wererealized. Pima County has a number of signalized intersections in proximity to the City of Tucson. In order to provide uniformity in the area, Pima County converted from leading to lagging left turn operations in 1987. Based on the reported Tucson experience, otherjurisdictions in Arizona began to consider changing to the lagging left turn phasing. Scottsdale, which is in thePhoenixmetropolitan area, converted their protected left turn phasing to a lagging operation in early 1989. The evaluation studies by the City of Tucson were rather limited, and there were a number of questions about the operational and safety aspects of the protected left turn phasing alternatives. In an effort to address these questions, a research project was initiated in 1989 by the Arizona Department of Transportation for the purpose of comparing the leading and lagging left turn operations. More specifically, the study was to investigate the following four basic research questions: 1) Is there a difference in intersection delay at isolated intersections between the leading and lagging left turn operation? 2) Is there a difference in signal progression among leading only, lagging only, and mixed left turn operations? 3) Is there a difference in accident experience between leading and lagging left turn operations? 4) Is there a motorist preference between leading and lagging operation? FIELD STUDIES AND ANALYSIS In order to address the research questions, field studies were undertaken in the Phoenix and Tucson metropolitan areas. The Phoenix area portion ofthe research focused primarily on the effects of the various left turn phasing patternson signal coordination and system behavior. Also, the Phoenix area research evaluated the effects of left turn phasing on intersection delay and accidents. Finally, a public opinion poll was performed to obtain information regarding possiblemotorist preferences for the leading or lagging left turn phasing. The Tucson area portion of the research project examined the accident experience resulting fiom the conversion of left turn operationsat intersectionsin the City off ucson and Pima County. In addition, the study compared traffic operations before and after the left turn conversion at selected isolated signalized intersections in Pima County. Accident Studies The City of Scottsdale,in 1988,undertooka sixmonth trial periodoflagging left turn operation. Five intersections were converted fiom leading to lagging operations in June 1988. Based on the trial period experience, the City of Scottsdale converted an additional 45 signals to a lagging operation in the early part of 1989. Due to the brief history of the use of lagging left turns, it was necessary to compare a one year accident experience with the lagging left turns with three years of accident experience with the leading left turns. It is recognized that a multiple year after period would be more desirable due to the random nature of accidents and the multitude of factors which may influence accident patterns. For this reason, the statistical test which was used in the analysis makes use of a control group that serves to discount thc influence of extraneous factors and helps to identify general trends in accidents. Ofthe 50 intersections in the City of Scottsdalethat were converted from leading to lagging left turns, nine intersections were selected for the accident study. The accident experience during the before and after periods were compared with a control group consisting of two phase signalized intersections in the City of Scottsdale. The conversion from leading to lagging left turn operation at signalized intersections under the control of Pirna County occurred in 1987. While 37 intersections were involved in the conversion program, some ofthe intersections were not suitable for accident analysis due to other changes; thus only 21 of the intersections were included. Because some of the approaches at these intersections did not have protected left turn movements, the accident analysis was accomplished by approach. In this way, only the approaches that were affected by the conversion fiom leading to lagging left turns were analyzed. The analysis period consisted of two years prior to the conversion of the signal operations and approximately two years following the change over. At a few intersections, the duration of the before period was less than two years due to the date of signal installation. A "before and after" analysis technique was used where the accident experience at each intersection approach was compared. In the City of'fucson, the conversion from leading to lagging left turns was accomplished in 1985. Again, a "before and after" comparison of the accident experience at individual intersections was undertaken. Data for a before period from 1982to 1984 and an after period fiom 1986 to 1987 were provided by the City from computerized accident reports. A total of 62 intersections were included in the analysis of which 50 were the intersection of major arterial streets. The remaining 12 intersections involved the intersection of major arterial streets with collector streets. Travel Time Studies In an effort to assess the effect of the left turn phasing alternatives on a system of signalized intersections, travel time studies were conducted in three cities in the Phoenix area. Alternative left turn phasing patterns were tested using travel time data along five routes in Glendale and four routes in Tempe. Four Glendale intersections and two Tempe intersections were changed from leading to lagging operation. The patterns tested were a) all leading left turns, b) all lagging left turns, and c) a combination of leading or lagging left turns depending on which best fit the progression along the route. In addition, a combination phasing was tested along one route in Mesa. In order to obtain a true comparison between leading and lagging left turns, it was necessary to use signal timing patterns developed by a common optimization program. Because of the ease of operation and the numerous runs that would be required as part of the combination portion of the study, the computer program known as FORCAST was utilized to determine the optimal signal timings. Intersection Studies Tntersection stopped time delay studies were conducted at six Phoenix area locations to perfo~rna comparative analysis of leading and lagging left turn operations. Manual stopped time delay studies were conducted at each intersection prior to changing from leading to lagging left turns. Five of the six intersections operated with protected/permissiveleft turn phasing on all approaches. The sixth intersection operated with protected only left turns on the northbound and southbound approaches and protected/permissive left turn phasing on the other two approaches. In addition, studies comparing leading left turns with a combination leadingllagging operation were conducted at one intersection in Mesa. At some intersections in the Phoenix area, third car actuation is used on approaches with protected1 permitted left turns. With this operation, the protected left turn phase will not occur unless three or more vehicles are queued in the left turn lane. As part of the research project, delay studies were undertaken for the purpose of comparing the third car verses the first car actuation. This particular part of the research evaluated only the leading left turn condition. Prior to the conversion from the leading to lagging left turn operation, a limited number of intersections under the control of Pima County were selected for a "before and after" comparison of the effect of the change. Nine intersections were filmed with two time-lapse cameras from approximately 3:00 p.m. to 6:00 p.m. Following the change to lagging left turns, the intersections were again filmed during the same time periods. Due to difficulties at two intersections, only seven were included in the final detailed analysis. Using the film record of the intersections, data which represented pertinent operational parameters were extracted. For example, the measures included cycle length, stopped delay, and volumes. In contrast to the intersectionsin the Phoenix area, all ofthe Pima County intersections were isolated and operating with actuated control. Because the intersectionswere in the outlying areas ofTucson, the traffic volumes were not equally distributed in terms of the opposing movements. There were very few cycles in which the approachingtraffic failed to clear the intersection;thus the intersections were generally not operating at saturated conditions. RESULTS AND CONCLUSIONS Based on the field studies, it was found that intersection delay is significantlygreater with the lagging left turn operation. No significant change in total delay was found with third car actuation of leading protected left turns. In addition, no significant differences were found in progression between the leading, lagging, and mixed operations. In terms of the accident experience, no significant differences were found between the leading and lagging left turns. Finally, there was a mixed response From the motorist preference survey. Glendale drivers felt that leading left turns were better while Tempe drivers preferred the lagging left turns. More specifically, the following results were found with respect to each of the questions posed for the research project: 1. Is theread~flerencein intersection delay at isolated intersectionsbetweenthe leading and lagging lefr turn operation? The results of this study indicate significantly greater delay per approach vehicle occurs with lagging operation than with leading operation for the intersections and time period tested. It is important to note that the time periods tested were generally PM peak hour conditions. These would not be as likely to have sufficiently low left turn and through volumes to eliminate many protected left turn phases in the lagging condition. It is conceivable that in off peak conditions more of the left turns could be made in a permissive manner therefore skipping the protected left turn phase. Eliminating the protected phases would likely reduce intersection delay. Intersection delay was also collected for test intersections with both first car and third car actuation. Although there was no significant difference between the two, this test also was only done in the PM peak hour condition. The probable benefit of third car actuation on intersection delay is most likely in off peak conditions. 2. Is there a dlerence in signalprogression among leading only, lagging only, and mixed left turn operations? There were no statistically significant differences in stops, delay or travel time with the different operating conditions. Additionally, the large number of signat timing optimization runs required to evaluate all combinations of leading and lagging operation makes for a cumbersome, time consuming process. The requirement that the Glendale andTempe "mixed" operation was limited to either both leading or both lagging on the same street in order to avoid the "trap" restricted potential progression benefit. An additional limitation was that only four of eight multi-phase Glendale intersections and two of four multi-phase Tempe intersections were considered for change to lagging. The most promise for benefit from laggingor mixed operation was found in the Mesa study where leading left turn operation was utilized for eastbound traffic and lagging for westbound traffic in the after condition. This was the operation which provided the best east-west progression. This mixed operation was possible without the trap condition because of the use of protected only left turns. 3. Is there a digerenee in accident experience between leading and lagging leff turn operations? In all three accident studies - Tucson, Pima County and Scottsdale, therc was no significant difference in left turn accident history between leading an.d lagging operation. 4. Is there a motorist preference between leading and lagging operation.? Lagging left turns seem to be more favorably received in Tempe than in Glendale. This could possibly be due to the close proximity of Tempe to Scottsdale, where lagging left turns are utilized. Public Perception Results More Green Lights With: Leading Lagging Combination No Di fferencemo Response Glendale 38% 16% 24% 22% Tem~e 30% 21 % 27% 22% Left Turns Better With: Leading Lagging No DifferenceNo Response The field studies by the research team provided valuable insight to the understanding of the many variables which influence left turn operations. Within the scope of this study and the conditions at the study sites, it was not possible to collect data for all combinations of the pertinent variables. The results of the field studies together with a somewhat theoretical analysis or understanding yield a comprehensive assessment of leading and lagging left turn operations. In essence, a number of variables have an impact on the effectiveness of left turn alternatives at a specific site. The variables that should be considered fall into the general categories of a) signal control, b) network considerations, c) traffic characteristics, and d) driver perception. PART I INTRODUCTION PART I presents an introduction to the research report,documents the literature search and discusses the research problem statement. CHAPTER 1 gives a background to the consideration and use of a leading or lagging left turn operation. A brief summary of the research results is presented. The overall organization of the research report is also discussed. CHAPTER 2 documents the findings of the literature search, summarizes the theoretical basis for using a leading or lagging operation and reviews actual experience with both operatioris. The research problem statement is presented which was used as the framework for carrying out the individual rcsearch studies. The following four research questions were addressed in the individual studies: Is there a dixerence in intersection delay at isolated intersections between leading and lagging operation ? -1s there a dlfcrence in signal progression among leading only, lagging only and mired operation ? .Is there a dlference in accident experience between leading and lagging operation? .Is there a motorist preference between leading and lagging operation? CHAPTER 1 INTRODUCTION This research project entitled "Comparative Analysis of Leading and Lagging Left Turns" was conducted by Lee Engineering, Inc. in association with Dr. Robert H. Wortman. The rationale for this association in undertaking theresearch is based on the ability ofLee Engineering to conduct field studies in the Phoenix area while Dr. Wortman had information and data related to conditions in the Tucson area. The combined efforts of the two groups permitted a comprehensive investigation of the problem. Lee Engineering was responsible for data collection andanalysesofintersectionsin the Phoenix area, and Robert Wortman conducted the work related to analyzing intersections in the Tucson area. Following the accomplishment of these studies, the research team jointly evaluated the collective findings of the research and prepared this final project report. Probably the most controversial item in the Arizona traffic engineering community for the last few years relates to leading vcrsus lagging left turns. In 1985, the City of Tucson, Arizona initiated an effort to convert the protected left turn signal phases from a leading to lagging operation. It was believed that the use of lagging left turns would improve intersection operations and network flows. The limited studies following the conversion from leading to lagging left turn operations suggested that operational and safety advantages were realized. Pima County has a number of signalized intersections in proximity to the City of Tucson. In order to provide uniformity in the area, Pima County converted from leading to lagging left turn operatior~s in 1987. Based on the Tucson experience, otherjurisdictions in Arizona began to consider changing to the lagging left turn phasing. Scottsdale, which is in the Phoenix metropolitan area, in early 1989 converted their protected left turn phasing to a lagging operation. There are widespread opinions as to the benefits of each of the two methods. A tremendous amount of misinformation and misunderstood information exists. This is particularly critical when one realizes the significant actions which are being taken and considered by cities and coutlties within the state based on this suspect information. It should be noted that most intersections in Arizona with a protected left turn phase also have a permitted phase which allows motorists to turn left through gaps in opposing traffic. At intersections with perrnittedlprotected phasing, simultaneous lagging left turn arrows are used to avoid trapping motorists who have pulled into the intersection while waiting to turn. If more complete information regarding leading and lagging left turns is available and traffic operations decisions are based on that information, the opportunity exists for reduction in automobile delay and number of stops as well as increased safety. Certainly these are worthwhile goals. The need is further enhanced when one considers the reduction in auto emissions and &el consumption associated with such operational improvements. This research project is intended to provide additional information in this area. One of the perceived advantages of lagging left turns in a permissivelprotected operation is the possibility of eliminating some of the protected left turn phases. This would occur when the left turning vehicles find sufficient gaps during the permissive period to reduce the protected green time or skip the phase. Third car detection has been utilized by some Arizona cities to attempt to accomplish the same omission of the protected phase in a protected/permissive (leading) operator. This technique require a vehicle actuation of a detector placed a distance back from the stop line where the third left turning vehicle would be stopped. The protected left turn phase only is called when this "third car detector' is actuated. It is appropriate that a study be undertaken to provide a factual basis for making the determination of the type of left turn phasing needed at individual intersections within Arizona cities. Some ofthe questions addressed in this study include the following: 1 . Does lagging le8 turn operation reduce intersection delay? In other experiments documented in the literature based both on measured and simulated experiments, there generally has been found no significant difference in intersection delay between the two phase set options at isolated intersections. 2 . Does lagging left turn operation provide better signal progression? This study investigates the difference in number of stops and travel time along arterial streets with both leading and lagging operation. 3. Is it necessary or desirable to have consistent phasing (either leading or lagging leJ turns) within any given city, urbanized area, and throughout the state of Arizona? There has been a concern among Arizona traffic engineers, elected officials and citizens that the mixture of leading and lagging operation among jurisdictions created safety or operational problems for motorists. 4 . What i s the egect on accidents of leading versus lagging left turns? Although the City of Tucson has reported a reduction in accident rate with lagging left turn operation, it was based on onIy six months of after period. The Federal Highway Administration (5, p.17) reports a higher accident rate for lagging than leading left turn operation. This may be due to the previously discussed safety problem of phase overlap on permissivelprotected operation. 5. What is the motorkts'perception of the leading versus lagging lefP turns? Based on the experience reported by Tucson (2) and Scottsdale (6) there may be a motorist preference of lagging over leading operation. RESEARCH APPROACH This research project was divided into several subareas for analysis as follows: Effect of leading vs. lagging left tums on intersection delay. Effect of leading vs. lagging left turns on signal system progression. Effect of leading vs. lagging left turns on accident experience. Effect of third car detection actuation on intersection delay. Motorists preference of leading and lagging left turns. Intersection Delay The intersection delay study was conducted both in the Phoenix area and in Pima County. At seven intersections in Glendale, Tempe and Mesa, the intersection delay with leading left turns was compared to that with iagging left turns. At the one Mesa intersection the only after condition involved a leading left turn in one direction and a lagging left turn in the opposing direction. The delay in the Phoenix area was obtainedby counting the queued vehicles at 15 second increments. The Pima County intersection delay was obtained using time lapse photography with both leading and lagging left turns at 9 locations. Signal System Progression The signal system progression was evaluated in Glendale, Tempe and Mesa by driving an instrumented test vehicle down each street to be evaluated for six runs in each direction. The runs were conducted for each of four conditions in Glendale and Tempe: Existing timing (all leading). Optimized all leading timing. Optimized all lagging timing. Optimized combination. It should be mentioned that all-lagging phasing was implemented only at the four Glendale intersections and the two Tempe intersections being changed although there were more intersections being timed and evaluated which were held to leading operation. The combination timing consisted of the best combination (from a system signal progression standpoint) of leading and lagging left tums at the four Glendale and the two Tempe intersections being changed. The theory behind this test is that one can establish the best two-way progression on three streets of a grid by fitting the east-west green into the already established north-south red. This can also be done for the third street, however when attempting to "close" the grid, frequently the bands don't properly fit. It was hypothesized that having the flexibility on oneofthe streets would better permit a good grid closure. The signal progression evaluation in Mesa consisted of the following: Existing leading operation. Combination of leading left eastbound and lagging left westbound. This combination of leading and lagging left turns had been determined to provide better progression than all leading. Accident Experience A before (leading) - after (lagging) accident study was conducted in Tucson, Pima County, and Scottsdale. Although there were varying periods in both conditions among the three jurisdictions, Scottsdale generally had a shorter after period than the other two. This was because lagging left turn operation had been more recently implemented in Scottsdale. Third Car Detection Evaluation A study ofintersection delay with first car versus third cardetection was conducted at 3 intersections in Phoenix and 2 intersections in Tempe. The purpose of the evaluation was based on the premise that if only one or two vehicles were desiring to make a left turn, they could do so during the permissive green period or during the clearance interval. It was hypothesized that eliminating some of the protected left turn phases should provide more green time for through vehicles thereby reducing intersection delay. It should be noted that one of the possible advantages of third car actuation is the ability to continue to have the phase overlap capabilities associated with protected/permissive (leading) operation. Motorists' Preference In an attempt to determine if there was a drivers' preference for either leading or lagging left turns, a questionnaire was sent out to owners of vehicles which had been observed driving on streets being tested in Glendale and Tempe. Approximately 4500 questionnaires were mailed with about half going to drivers in each of the two cities. ADVISORY COMMITTEE An important part of this study involved the formation and use of an advisory committee of municipal, county and state traffic engineers within the state. The philosophy was that these are the individuals who have to operate the signal systems, therefore they should be directly involved in planning, conducting and evaluating the research operation. Advisory Committee Members are as follows: Roger Hatton Robert Pike Al Letzkus Paul Basha Kenneth Shackman Hugo Malanga James Matteson Richard Nassi Ron Krosting Hawey Friedson ADOT Traffic Engineering Section ADOT Research Maricopa County City of Scottsdale Pima County DOT City of Glendale City of Phoenix City of Tucson City of Mesa City of Tempe ORGANIZATION OF REPORT This research report is organized into five parts. Part I presents an introduction to the research, a summary of the literature search and discussion of the research problem statement. Part II presents the accident studies which were carried out in the City of Scottsdale, Pima County and the City of Tucson. Part III presents theresul tsof traffic operations studieswhich wereconducted in thePhoenix area and Pima County. Part IV presents the results of the public awareness and perception analysis conducted for the cities of Glendale and Tempe. Part V discusses the study results, theoretical analysis of leading and lagging left turns, and presents recommendations for future research work. STUDY RESULTS This research study yielded the following results: Lagging operation resulted in greater delay per approach vehicle than leading for the intersections and time periods tested. There was no significantdifference in signal progression between all leading left turns, all lagging left turns and some streets with leading and some streets with lagging. - Although not significant due to limited sample size, there was a notable reduction in delay and travel time on theone street tested with leading left one direction and lagging left in the opposing direction when compared to the leading condition in both directions. There was no significant difference in accident experience between leading and lagging operations. Motorists in Glendale andTempe felt that they experiencedmore green lights with leading than lagging. Glendale motorists felt that left turns were better with leading operation while Tempe motorists felt left turns were better with lagging operation. It should be noted that delay studies for leadingflagging and 3rd/l st car actuation were primarily conducted during the PM peak hour. It is possible that there might be a greater chance to skip protected left turn phases for both lagging and 3rd car actuation in an off-peak period. I CHAPTER 2 LITERATURE REVIEW The concept of a lagging green left turn interval is not new. Neither is the question as to whether leading or lagging left turns is preferable as evidenced by the following excerpt fiom the 1965 version of the Traflc Engineering Eiandbook (1, p.403). While not exactly a third phase, the use of a leading (advance) or lagging (delayed) green may bc helpful in special situations. At an intersection having a fairly heavy left-turn movement, say on Phase A eastbound, the holding of the westbound Phase A green for 5 to 10 sec after eastbound traffic receives its green could be helpful. This is known as giving a "leading" green to Phase A eastbound. Ontheotherhand,if neartheendofphase A,anadditionalS to 1Osecisal!ottedtoPhase Aeastbound, it would be receiving a "lagging" green. The use of either one of these should be approached with extreme caution because a motorist who is receiving the shorter green might not realize it since he sees opposing traffic flowing freely. He may continue to proceed into the intersection against a red signal and may collide with a motorist making a left turn who expected that the opposing motorist would stop. Some authoritiesfeel that the leadinggreen is probably less hazardousthan the lagginggreen because motorists in opposing directions would generally be starting from a stopped position. On the other hand,some authoritiesfavorthe lagging green because of a tendency fortraffic standing at the Stop line in the opposing direction to start when they see the traffic having the leading green begin to move. They feel that the left-turn capacity is increasedbecause the have moved into the intersection during the regular green period and a grcater number of vehicles are able to clear the intersection than when they are starting from the Stop line at the beginning of a leading green interval. Potential for the lagging left turn being more hazardous as mentioned in the previous excerpt refers to what is sometimes called the "trap" of lagging left turns. When lagging left turns have been used with phase overlap (one left turn becomes lagging protected while the opposing permissive left turn terminates) an increase in accidents has been observed in some locations. The driver who is waiting in the intersection to turn left and sees all the traffic lights on the approach change to red and adjacent through traffic stop expects the opposing traffic to also be stopping. The driver waiting in the intersection to turn left either turns unknowingly into the path of opposing traffic which still has the green or gets trapped in the middle of the intersection. The phasing diagram on Figure 2-1 demonstrates how this situation could occur. The potential problem occurs in transition from the 2-6 phase to the 2-5 phase. As the driver of the left turn permissivemovement associated with phase 6 sees theyellow, he might erroneouslyassume phase 2 is also ending and pull out in front of phase 2 traffic still viewing a green indication. One way to alleviate this problem where permissive lagging left turn operation is used, is to require the left turns in opposing directions to be operated simultaneously in a protected manner. This is undesirable where there is a definite imbalance in directional flow. Consider the example of considerably more left turn and through traffic in a northbound direction during a particular period of the day. If the northbound lefts cannot be accommodated during the permissive operation they NEMA PHASE PHASE DIAGRAM - DESCRIPTION Ds 2-6 IL LeR Tlrms Prohlblted l 92T - 2 5 ll , Protected Left Turn Phese 5 ?P562T Dl 1-5 14 , I Protected Left Turns ?P5 T The data on delay, fuel consumption and auto emissions was based on a simulation model used by the Pima Association of Governments (3). Certainly these impressive figures merit fbrther considerationof lagging left turn operation. The fact that this information is based on simulated data rather than field-measureddata may reduce the impact of the findings, however. Additionally, there was a reported reduction in volume during the study which was apparently represented in the simulation model. This volume reduction could account for part of the delay reduction. The simulation run also considered a 10% lower cycle length with the lagging operation than the leading operation. This factor also would result in reduced delay being simulated. It should be noted that this study was based on limited data and only a six month after period. A recent study prepared for the City of Scottsdale presented guidelines for the implementation of leading and lagging left turn phasing. It also considered the potential of third car actuated left turn phasing for protected/permissive operation. These guidelines are presented here (4, p.21): Guidelines for Third Car Actuated - Leading Left Turn Phasing and Permitted/Protected -Lagging Left Turn Phasing: When a traffic engineering study indicates that protectedtpermitted(leading) - permitted/protected (lagging) phasing is appropriate, a second decision is needed regarding lead versus lag. A basic conclusion of this study and premise for the following guidelines is that all leading-protected1 permitted operations should be third car actuated. This will reduce delay by enhancing intersection capacity and arterial travel speed. Recommended guidelines are: 1. Use Leading Protected/Permitted left turn phasing when: intersection is isolated and filly actuated intersection is in a coordinated system and leading left turns will enhance progressive movements, or leading left turns will provide for more efficient trafftc flow than lagging. 2. Use Lagging-PermittedProtectedLeft Turn Phasing when: intersection is in a coordinated system a capacity study and system operational analysis indicates that simultaneous lagging phasing offers operational benefits when compared to leading left turn phasing. (In some systems a combination of lead-lag phasing may be appropriate). There are several signal system simulation programs that can be used to generate key traffic operations "data" for use in the decision process. Specific data which should be considered are intersection capacity, lcvel of service, delay, average speed, fuel consumption and emissions. Care should be taken to evaluate the left turn lane length requirements so as to avoid left turn traffic backing up into the through lanes. Such a backup can offset the other benefits of lagging left phasing. It shouldbenotedthat at locationswith protected left turns, lagging left phasing may offeroperational benefits in comparison with leading left turn phasing. The guidelines discussed above should be followed in evaluating lead versus lag turns in a protected system. Last, but not least, the City should develop a public information program including media notifications and temporary signing in conjunction with the implementation of lagging left tumpemittedfprotected phasing. The CityofScottsdale (5) reported areduction in delay after going tolagging operation in 5 of6 tests. This was based on AM peak, noon peak, and PM peak for both eastbound and westbound direction and is summarized in Table 2-1 (6, p.3). The city's study was based on five intersections along Thomas Road. Four of these five intersections were protected/permissive in the leading condition and were permissive/protected in the lagging condition. One intersection was protected only in the leading condition and pemissive/protected in the lagging. All five intersections had phase overlap for eastbound and westbound approaches in the leading condition and simultaneouslagging arrows in the after condition. This study was based on travel time and delay data. The lagging operation Table 2-1. City of Scottsdale Delay Study Delay Time, Seconds Perlod and With Leading With Lagglng Direction Arrows Arrows AM Peak - Eastbound 46 26 AM Peak - Westbound 14 8 Mid Peak - Eastbound 15 70 Mid Peak - Westbound 44 11 PM Peak - Eastbound 88 59 PM Peak - Westbound 53 30 Dlflerence -20 -6 55 -33 -29 -23 was a best fit offset selection into the previously developed signal plans. The change in delay could possibly be due to the random nature of the lagging left turn fitting into the time space diagram. The system progression was primarily determined by other streets with a higher priority (e.g. Scottsdale Road and Hayden Road). The research team on this current project performed the signal optimization study for the City of Scottsdale and assisted in the hand fitting ofoffsets on Thomas Road to optimize the operation of signals there within the previously determined signal timing patterns. The number of intersection accidents in June and July of 1987 (with leading left turn operation) was compared with the same period in 1988 after lagging operation was implemented. The results are shown in Table 2-2 (6, p.2). Caution should be used in drawing conclusions from this data because of the short time period. Table 2-2. Citv of Scottsdale Accident Data Accident Type Left Turn Rear End Angle Other With Leading Arrows 2 With Lagging Arrows 8 The City of Mesa tested a leadingtlagging operation on Alma School Road in 1988. The intersection of Grove and Holmes was converted from leading left both northbound and southbound to leading left in one direction and lagging in the other based on the time-space diagram. Dramatic improvementwas seen in the noon andeveningperiodsafter implementation ofthe new timing plan. During the noon period, northbound travel time decreased by 106 seconds (a 52% improvement), and southbound travel time decreased by 54 seconds (a 66% improvement). During the evening period, northbound travel time decreased by 61 seconds (a 65% improvement), and the southbound travel time decreased by 85 seconds (a 54% improvement). (7) Research by Fambro and Woods (8) which was included in the Federal Highway Administration publication Guidelinesfor Signalized Left Turn Treatments indicates lagging left operation has an accident rate of twice that of leading. This information is presented in Table 2-3 (9, p. 17). Table 2-3. Relative Left-Turn Accident Rates for Various Left-Turn Signalization Schemes Type of Relative Left-Turn Phasing Accldent Rate Unprotected (Permissive) 1.OO Permissiveffrotected - Lagging Proteded-LeadingPemissive Protected 0.10 It is possible this is due to the previously discussed "trap" of lagging operation where the through movements do not terminate at the same time. This document states (9, p.13): Whcn selecting the type of left-turn signal phasing to install at an intersection, standardization often enters the picture. Some agcncies recommend that only leading left-turn phasing be installed, whilc others recommend lagging left-turn phasing. In fact, there is one large city that uses Wephase (unprotected left-turn) signals exclusively. Uniformity and consistency in the type of signal phasing that is employed has inherent advantages in the area of driver expectancy. The motorists know the phasing arrangement to anticipate and can react accordingly; however, as demonstrated at actuated signals, this same group of drivers has proved adaptable to changes in signalization. Uniformity in left-turn phasing offers no proven safety benefit and in some situations does not result in the most efficient operation. Of particular significance is the last sentence relating to the potential safety benefit of uniformity. This is particularly true when one considers the apparent perceived importance of uniformity demonstrated by the City of Tucson, Pima County, and the City of Scottsdale. TheFHWA publication alsogives rccomrnendationson phase selection. It statesthat there are system considerations of left turn phase selection (9, p.27): Signals placed in a system configuration require consideration of the effects of the left-turn phasing on the system operation. Left-TurnP m . If the time space diagram indicates that traffic on each approach to the intersection arrives at the same time, dual left-turn phasing should be implemented. bad-F.;lg Lefi-Turn Phasiu. If the time space diagram indicates that traffic on each approach to the intersection amves at an appreciable difference in time (10 seconds or more) lead-lag phasing should be implemented. A copy of Table 7 of the FHWA report is included in Appendix A which summarizes the phase selection guideline consideration for left turn phases. Because of the previously discussed potential safety problem of lagging permissive, it was determined that this research project consider these overlap phases and split phases in a protected only lagging operation. A study by Machemehl which was based on a simulation model called TEXAS investigated various left turn sequence pattcrns at an isolated intersection. Machemehl reported (10, p.39): In cases where split left-turn sequences are selected under actuated control, the question of which left-turn movement should lead a through movement green may arise. To determine whether the leading left-turnmovement performs differently than the lagging movement in a split left-turn phase arrangement.20 trafficapproachdemandcombinationswerecomparedforeachofthe twosituations. The results indicate thatthereisno significantdifference in delay toleft-turningortothrough vehicles when a lagging phase is used instead of a leading phase, even though the required phase lengths are very different. This is because the left-turn queue discharges more efficiently with a leading phase minimizingdelay to individual vehicles,but it requiresa 1ongerphasetodos0,causing a longer cycle duration and more &lay at the intersection. On the other hand, because the laggingphase isshorter, the main street green signal must be longer to process the through vehicles that would be processed with the left-turn vehicles with a leading phase. Thus, there is no significant difference between leading and lagging phases with split left turns and actuated control. The literature search does not support the current phasing practices within the state, particularly the apparent need for standardization of either leading or lagging operation within the various governmental jurisdictions. Conversely, the literature generally recommends that the decision for leading versus lagging operation be based on conditions at the specific intersection and the opportunity to provide the best progression. Even though the literature refers to the potential safety problem when terminating one through movement but not the other when going to a lag operation, it is not recognized to be as significant a problem as the local perception. The apparent source of the local importance is the Tucson experience of several lawsuits immediately after implementation of lagging left. PART I1 ACCIDENT STUDIES PART I1 documents accident studies which were performed for three Arizona jurisdictions which have converted from leading to lagging operations. CHAPTER 3 examines the accident experience in the City of Scottsdale. The Scottsdale accident analysis is based on a before and after comparison of the number of left turn accidents at nine intersections. The analysis compares three years of leading left arrow operation with one year of laggingoperation. The Scottsdaleaccident analysis indicated no significant difference in thenumber of left turn accidents between a leading and a lagging operation. CHAPTER 4 examines the accident experience in the Tucson area. The Pima County analysis is based on a before and after cornpatison of 21 intersections. The analysis, in most cases, compares two years of leading operation with two years of lagging operation. The City of Tucson analysis is based on a before and after comparison of50 major arterial intersectionsand 12intersectionsofmajor arterials with collector streets. The analysis compares three years of leading operation with two years of lagging operation. The analysis of left turn accidents in Pima County and the City of Tucson indicated that there were no significant differences resulting from the conversion from leading to lagging left turn operations. CHAPTER 3 SCOTI'SDALE ACCIDENT ANALYSIS INTRODUCTION The City of Scottsdale, in 1988,undertook a 10week trial period of lagging left turn operation. Five intersections were converted from leading to lagging operations in June 1988. Based on the trial period experience, the City of Scottsdale converted an additional 45 signals to a lagging operation during the early part of 1989. Due to the brief history of lagging left turn operation in the City of Scottsdale, a one year lagging experience was compared with a three year leading experience. It is recognized that a multiple year after period would be more desirable due to the random nature of accidents and the multitude of factors which may influence roadway safety. For this reason, the statistical test which was selected for the analysis makes use of a control group which serves to discount the influence of extraneous factors and helps to identify general trends in accidents apart from the changes which may be attributed to the implementation of lagging left turn operation. SELECTED INTERSECTIONS The intersection selection process involved the development of an appropriate list for both the test and control intersections. The goal of the analysis is to assess the change in the number of accidents strictly as a function of the conversion from leading to lagging operation apart from any changes in protected or protectedlpemissive left turn phasing. Therefore, the test group was selected on the basis of similar operating conditions in the before and after period. If the intersection was operated in a pemissive/protected phasing during the lagging operation then the intersection must have operated in the protected/permissive mode during the three year leading operation time period in order to be included in the test group. This constraint severely restricted the number of intersections which could be used in the analysis. Of the 50 intersections in the City of Scottsdale which were converted from a leading to a lagging operation only nine met this constraint. The test intersections used in the analysis are shown in Table 3-1. Also shown is the date of conversion from leading to lagging left turn phasing and the mode of operation before and after the conversion. One test intersection (Hayden Road/McDowell) was converted to dual left turn lanes in the last half of 1987. However, the intersection was retained in the analysis because it met the primary criteria of comparable signal operation in the before and after conditions. Similar criteria were used to develop an appropriate set of control intersections. The primary reason for using a set of control intersections is to discount extraneous factors such as changes in traffic volumes, unusually inclement weather over some period of time, changes in accident reporting, etc. The secondary reason for using a set of control intersections is to identify unusual changes in the number of accidents as the result of purely random occurrences which may not be representativeof long term trends. A set of37 two phase intersections in the City of Scottsdale were used as the control group. Intersections which had undergone major reconstruction during the study period were not included in the set of control intersections. A list of the control intersections used in the analysis is presented in Table 3-2. Table 3-1. Accident Analysis Test Intersections, Clty of Scottsdale lntersectron =fore Date of After Condltlon Converslon Condition '61st P1.fI'homas ~ d . YrotectedlYerm~ss~ve Ub/01188 Pent~l~~l~eIProte~ted Scottsdale Rd.fThomas R d ProtectedlPcrmissive 06/01/88 PermissiveProtected Hayden RdJMcDowell Rd. Leading Protected 0 1/27/89 Lagging Protected Miller Rd.Andian School Rd. Protected/Permissivc 02/02/89 PennissiveProtected 68th StAndian School R d Leading Protected 02/07/89 Lagging Protected Hayden Rd./Chaparral N/S Leading Protected 02/08/89 NIS Lagging Protected Scottsdale Rd./Camelback Rd. EIW Leading Protected 02/09/89 E N Lagging Protected Hayden RdJMcDonald Rd. Leading Protected 02/14/89 Lagging Protected Pima Rd./Shea Blvd. Leading Protected 02/22/89 Lagging Protected Table 3-2. Accident Analvsls Control Intersections. Cltv 01Scottsdale Intersection 60th Street i Thomas 64th Street / Camelback 64th Street / Cactus 68th Street / Oak 68th Street / Osborn 70th Street / McDowell 70th Street / Osbom 70th Place / Camelback 7 1st Street / Camelback 71st Place / Shea Swttsdale / Roosevelt Scottsdale / Oak Scottsdale 1 Earl1 Swttsdale / Fifth Avenue Scottsdale / Fashion Square Scottsdale / Jackrabbit Scottsdale 1 Mercer Scottsdale / Cholla Swttsdale I Sweetwater Intersection Scottsdale / Pinnacle Peak Civic Center / Osbom 74th Street / McDowell 75th Street 1 Indian School Miller / Mckellips Miller 1 Chaparral Miller / McDonald Miller / Shea 77th Street / McDowell Hayden I Oak Hayden /Jackrabbit Hayden / Indian Bend 82nd Street I Indian School Granite Reef / Thomas Granite Reef / Camelback Granite Reef / Chaparral Granite Reef / McDonald Pima /Mountain View DATA COLLECTION Accident data and information about each intersection was obtained fiom the Arizona Department of Transportation (ADOT) and the City of Scottsdale's Traffic Engineering Department. The primary source for accident data was the ADOT Accident Location Identification and Surveillance System (ALTSS) data base. The summary reports generated by ALISS were manually reviewed to identify the accidents which were of interest to this study. Police accident reports were also used to supplement the data base. The number of total intersection related accidents and the number of intersection related left turn accidents at each test intersection were recorded for each month from June 1985 to February 1990. For the purposes ofthis study, aleftturn accident wasdefined as those accidentswhich were classified as "left turn" manner of collision or where either vehicle action was classified as "making left turn". Only those left turn accidents on the east and west approachesat the intersection of ScottsdaleRoad Camelback Road and on thenorth and south approaches atthe intersection ofHayden RoadIChaparral were recorded for the left turn accidentanalysis. The other approachesat these two intersections were not considered due to incomparableconditionsin the before and afterperiods. The recorded accident data for the test intersections is shown in Table B-1 and Table B-2 in Appendix B. The total number of intersection related accidents at each control intersection were also recorded for each month from June 1985 to February 1990. The recorded accident data for the control intersections is shown in Table B-3 in Appendix B. ANALYSIS The Scottsdale accident analysis is based on a before and after comparison of total intersection accidents and left turn accidentsat nine test intersections. Due to the low number of accidentsat most intersections it was necessary to develop a pool of intersectionsrather than testing each intersection individually. Two test intersections were converted to a lagging operation as part of the six month trial period starting in June 1988. The remaining seven test intersectionswere converted at various times in January and February 1989. Two sets of test intersections were developed to distinguish between protected/permissiveoperations and protected-only operations. The first group of test intersections are those which have converted from a protectedpermissive to a permissive/protected operation and includes the two intersections which were converted as part of the 10 week trial period. The before conversion time period for Group No. 1 extends fiom 1 June 1985to 3 1 May 1988. The after conversion time period for Group No. 1 extends from 1March 1989 to 28 February 1990. The second group of test intersections are those which were converted in early 1989 fiom a leading protected-only to a lagging protected-only operation. The before conversion time period for Group No. 2 extends from 1January 1986to3 1December 1988. The after conversion time period for Group No. 2 extends from 1 March 1989 to 28 February 1990. Accident data was available only through the end of February 1990 at the time of the analysis. The 37 control intersections were pooled for comparisons with the two test intersection groups. A Table 3-3. Accident Analysls Summary Count, City 01Scottsdale Left Turn Accldeats Group No. 1 Year Test Intersections B3 B2 dl 61st Place / Thomas 0 1 12 16 16 Scottsdale I Thomas 5 7 6 Miller / Indian School 19 23 21 Total Control Intersections Total Accidents 230 286 Group No. 2 Test Intersections 6 6 12 284 Key: B3 = Jun 85 - May 86 B2 = Jun 86 - May 87 B1 = Jun 87 - May 88 Al =Mar 89 -Feb 90 Key: Year 260 I 269 B1 82 4 1 0 1 0 0 6 2 Hayden I McDowell 68h Street / Indian School Rd. Scottsdale / Camelback Hayden 1 Chaparral Hayden 1 McDonald Pima 1 Shea Total Total minus Hayden / McDowell Control Intersections Total Accidents ,"I B3 = Jan 86 - Dec 86 B2 = Jan 87 - Dec 87 B1 = Jan 88 - Dec 88 A1 =Mar 89 -Feb 90 5 1 2 1 2 1 12 7 237 219 - Total Intersection Accidents Group No. 1 Test Intersections Key: B1 61st Place / Thomas Scottsdale I Thomas Miller / Indian School Rd Total 83 0 16 23 39 B2 3 34 16 53 1 27 14 42 A1 1 20 12 33 Control Intersections Total Accidents 230 286 269 219 Croup No. 2 Test Intersections Key: Year B3 Hayden / McDowell 68th Street I Indian School Rd Hayden / Chaparral Scottsdale I Camelback Haydcn I McDonald pima / Shea Total Total Minus Hayden / McDowell 17 17 9 14 9 5 71 54 B2 22 16 10 10 16 5 79 57 Control Intersections Total Accidents 260 284 B3 = Jun 85 - May 86 B2 = Jun 86 - May 87 B1 = Jun 87 -May 88 A1 =Mar 8 9 - F e b 9 0 21 14 10 16 20 13 94 73 7 16 13 8 80 54 237 219 10 B3 = Jan 86 - Dec 86 B2 = Jan 87 - Jan 87 B1 = Jan 88 - Dec 88 A1 = Mar 89 - Feb 90 sumwmy of the number of accidents during the study period for both the test and the control intersections is shown in Table 3-3. The statistical test used in the analysisis achi-square gdness-of-fit test. The anlysis involves two checks: (1) a test for comparability between the control intersections and the test intersections in the before period, and (2) a test of the effect of changing from leading to lagging left turns at the test intersections. A cross product ratio was also calculated to measure the apparent effect of the conversion from leading to lagging left turn phasing relative to the control intersections. The chi-squaretest and cross product analyses are taken fiom a Texas Transportation Institute report: Three Procedures for Evaluating Highway Safety Improvement Programs by Lindsay I . Griffin, 111. (1 1). The test for comparabilitycompares the test andcontrol intersectionsusingthree years ofbeforedata. A test for comparability in the after period could not be performed due to the fact only one year of after data was available. The results of these tests are presented in Table 3-4. The calculated chisquare for Group 1 left turn accidents was 0.03 (p=0.98) which indicates a strong comparability with the control intersections. The calculated chi-square for Group2 left turn accidents was 3.48 e 0 . 1 8 ) which indicates Group 2 is not very comparableto the control group. Therefore, the use of the control group to measure the effect of the lead to lag conversion should be viewed with caution. The comparabilityof Group 2 minus the Hayden/McDowell intersection (Group 2A) was also evaluated. The results indicate an even weaker level of comparability (GZ=4.15, df=2, p=0.13). The calculated chi-square for Group 1 total intersection accidents was 0.59 @=0.75) which again indicates good comparability between Group 1 and the control intersections. The calculated chi-square for Group 2 (G2=5.59, p=0.064) and Group 2A (G2=5.81, p=0.056) indicates that Group 2 total intersection accident results are not very comparable to the control intersections. Therefore, the estimate of the effect of the Group 2 lead to lag conversion relative to the control intersections should be viewed with extreme caution. The test of treatment compares the before and after changes at the test intersections relative to the changes at the control intersections. The results of the test are presented in Table 3-5. Group 1 left turn accidents decreased 32% relative to the control intersections (G2=1-49,d F l , p=0.23) with the conversion h m leading (protected/permissive) to lagging @ermissive/protected)phasing. Group 2 left turn accidents showed an increase of 57% (ff=1.40, ~ 4 . 2 4 )relarive to t?x control Table 3 4 . Accident Analysis Test for Comparability, City of Scotisdale Left Turn Accidents Total Accidents Level of Level of Comparability Significance Comparability SIgniflcance Test Statlstic (P) Group Test Statistic (PI 1 0.03 0.98 0.59 0.75 Table 3-5. Accident Analysis Test of Treatment, City of Scottsdale 1 Left Turn Accidents Total Accidents Level of Level of Treatment Slgnlflcance Change Treatment SLgnlflcance Group Test Statistic (P) (%I Test Statistic (P) 1 2 1.49 1.40 0.23 0.24 -32% 57% 0.3 6 1.08 0.56 0.30 Cbange (%I -12% 17% intersections. With the conversion fiom leading to lagging phasing the results were similar for Group 2 minusHayden/McDowell (Group 2A) with an apparent increase of66% (@= 1-13,p=0.29). Group 1 total intersection accidents decreased 12% relative to the control intersections (@=0.36, p=0.56) with the conversion. Group 2 total intersection accidents showed an increase of 17% (@=I .O8, p=0.30) relative to the control intersections and Group 2A a slight increase of 5% (G2=0.07,p=0.79) with the change for leading protected to lagging protected phasing. The significance ofthese Group 2 changes is questionable due to the lack of comparability between the Group 2 intersections and the control intersections. Therefore, a simple be forelafter test was also performed to evaluate the absolute change in the number of accidents independent of the control intersections. The results are presented in Table 3-6. The reduction in the number of accidents at the Group 1 intersections become more pronounced when evaluated independent of the control intersections. Group 1 left accidents declined 43% (p=O. 12) and total intersection accidentsdeclined 26% (p=O.18). The increase in the number of accidents at the Group 2 intersections becomes less pronounced when evaluated independent of the control intersections. Group 2 left turn accidents increased 32% (p=0.54) and total accidents show a slight increase of 7% to (p=0.66). Group 2A provided similar results except the total intersection accidents show a slight decrease of 12%(p=0.49) when the HaydenhIcDowell intersection is removed h m the group. The high p-values indicate a low probability of any statistically significant difference in the number of accidents at the Group 2 intersections in the before and after periods. DISCUSSION OF RESULTS The Scottsdale accident analysis indicates no statistically significant change at the 90% confidence level in the number of left turn accidents or in the number of total intersection related accidents, with the conversion fiom leading to lagging left turns. The analysis which was applied utilized a comparison group and a test for comparability to evaluate the change in number of accidents. Also, extreme care was exercised in the selection of both the test and control intersections to ensure the analysis isolated strictly on the effects of the conversion from leading to lagging without allowing any extraneous factors to either into the evaluation. The test is admittedly rigorous and the number of intersections which could be analyzed was limited. The small sample size contributed to the finding of no statistically significant change. However, the apparent decrease for Group 1 coupled with the apparent increase in Group 2 seems to indicate that the impact of lagging operation on left turn accidents may be different for protected/ permitted and protected only intersections. The apparent decrease in accidents with the conversion from protectedtpermitted leading operation to a permittdprotected lagging operation indicates people may be less likely to turn across a gap in the permissive phase given the knowledge that a Table 3-6. Accldent Analysis Results, Clty of Scottsdale Left Turn Accidents Group Leading Lagging Absolute Change Test Level of Relative Statlstlc Slgniflcaace Change* Absolute Change Test Level of Relative Statistic Significance Change* Total Intersection Accldents Group Leading Lagging 2A 61.3 54 -12.0% 0.68 Change relative to the control intersections. Group 2A is Group 2 minus the Hayden / McDowell Intersection. 0.49 4.7% protected left turn can be executed at the end of the permitted phase. This same safety advantage would not be realized with the conversion from a protected-only leading to protected-only lagging operation. The availabiiity of accident data did not allow for a three month driver adjustment period at all intersections. However, the lagging left turn was not completely new to City of Scottsdale drivers at the time these intersections were converted. They had just participated in a 10 week trial period of lagging operation at five locations in the City of Scottsdale. Nevertheless, further analysis should be performed as this data becomes available. Finally, it should benotedthat thenumberofreportedaccidents in the CityofScottsdale hasgenerally declined over the past few years. This is evidenced by the general decline in the number of accidents recorded at the control intersections for the last year before conversion and the first year after conversion. The City of Scottsdale's Traffic Engineering Department is not aware of any particular changes which may have brought about this welcomed event. However, in terms of the analysis performed, this put added pressure on the limited number of test intersections analyzed to show commensurate declines. CHAPTER 4 PIMA COUNTY/TUCSON ACCIDENT STUDIES As part of this research project, an analysis of the accident experience in the Tucson area was undertaken. This analysis involved an examination of accidents at signalized intersections before and after the conversion from leading to laggingturns. Accident data from the City of Tucson as well as Pima County were utilized in the study; however the experience of each jurisdiction was analyzed separately. The purpose of this chapter is to document the data collection, analysis, and results of the study. PIMA COUNTY ANALYSIS Introduction The conversion from leading to lagging left turn operation at signalized intersections under the control of Pima County occurred in 1987. At that time, Pima County had a total of 37 intersections which were converted. This constituted virtually all of signalized intersections under the control of Pima County. In addition to the conversion from leading to lagging left turn operation, other operational changes weremade at a number of intersections. These operational changes, lack ofdata, or annexation by the City of Tucson necessitated the elimination of some of the intersections from the accident study. As a result, the analysis included a total of 21 intersections which are listed in Table 4- 1. The type of left turn operation by approach is also shown in the table. As may be noted from the information in Table4- 1, most of the study approachesutilized protected/permitted left turn operations. A limited number of the approaches had protected only left turn movements. Two of the approaches includedin the study had protected/permitted left turn operations in the before period. Changes in intersection signal operations at these two intersections resulted in protected only movements in the after period. Data Collection For the analysis ofthe Pima County signalized intersections, data and information about each of the intersections were obtained fiom the records of the Pima County Department of Transportation. For each of the intersections, the following data and inforrnation were obtained: - date of conversion from leading to lagging left turn operation accident data for the before and after periods signal timing plans estimated traffic volumes Other relevant inforrnation relative to changes in design operation of the intersection. The records maintained by the Pima County Department of Transportation yielded detailed information about the intersections. For example, it was possible to obtain collision diagrams for each intersection as well as accident summaries. In addition, the accidents could be analyzed by type and intersection approach. Table 4-1. Number of Left-Turn Accldents, Plma County Intersection Ajo Way / Palo Verde Rd. Alvernon Way / Irvington Rd. Alvernon Way / Valencia R d Campbell Ave. / River Rd. Craycrofl Ave. / River Rd. Craycroft Ave. / Sunrise Dr. Dodge Blvd. / River Rd. Dos Hombres I Tanque Verde Rd. First Ave. / Ina Rd. First Ave. / Orange Grove Rd. First Ave. / River Rd. Ina Rd. / La Canada Dr. Ina Rd. I La Cholla Blvd. Ina Rd. / Oldfather Rd. Ina Rd. / Thomydale Rd. Kolb Rd. I Valencia Rd. La Cholla Blvd. /Orange Grove Rd. Mission Rd. / Valencia Rd. Orange Grove Rd. I Skyline Dr. River Rd. / Swan Rd. Direetlon Northbound Southbound Southbound Eastbound Eastbound Westbound Northbound (P) Westbound Southbound Westbound Northbound Westbound Eastbound Westbound Northbound (*) Northbound Southbound Northbound Southbound Eastbound Westbound Eastbound Westbound Eastbound Northbound Southbound Eastbound Westbound Northbound (P) Southbound (P) Eastbound (P) Westbound (P) Northbound (P) Southbound (P) Eastbound (P) Westbound (P) Eastbound Westbound Northbound (P) Northbound Westbound Southbound (*) Number of Accidents Before 0 2 Number of Accldents After 0 0 1 4 1 5 3 0 0 1 2 2 1 3 1 1 3 3 1 2 4 0 2 1 2 4 0 0 2 0 0 0 0 1 0 3 8 3 0 Sunrise Dr. / Swan Rd. 2 (P) - Protected only left turns. (*) - Protected / permitted left turns in the before period and protected left turns in the after period. 1 0 0 7 7 0 8 6 4 0 0 1 1 1 0 1 0 3 5 10 2 2 5 3 9 5 1 0 0 0 0 0 0 1 1 0 2 3 13 6 0 0 Because the actual dates of conversion from leading to lagging left turn operation were known, the before period was defined as two years preceding the conversion in 1987. In some cases, it was not possible to obtain a fbl1 two years of accident data at an intersection due to other factors such as the date of signal installation. The actual number of days in the analysis periods for each of the intersections are indicated in a later section of this report. For the analysis, a three month driver adjustment was used; thus the after period commenced three months followingthe conversion at an intersection. The aflcr period then included all accidents from three months after the conversion ofthe signal operation until November 30,1989. This cut-off date wasnecessitatedby thedata collection process; howeverit resultedin an afterperiodofover two years at each of the intersections. Analysis During the process of converting from leading to lagging left turn operation, Pima County reevaluated the need for protected turn movements at the intersections. This resulted in left turn phases being added or deleted at many of the intersections listed in Table 4-1. In order to maintain a reasonable data base, the intersections were analyzed on the basis of the individual approaches. For example, the evaluation focused on left turn accidents on the intersection approaches where the conversion from leading to lagging operation had occurred. Intersection approaches where other changes had been made (such as the addition or deletion of a left turn phase) were not included in the final analysis. In this way, accidents that were directly associated with the leading and lagging left turn operations were isolated for comparison. Table 4- 1 indicates the number of left turn accidents that were reported during the before and after periods at each intersection. The accidents are shown by approach for each of the intersections. For this study, an accident was considered to be a left turn accident if a left turning vehicle on a given approach was involved. As indicated previously, the initial intent was to obtain the accident records foratwo year period prior to the signal operation conversion. In some cases, this was not possible. The after period for all of the intersections was greater than two years with the exception of one which had an after period of a few days less than two years. The number ofdays in the before and afterperiods ofeach intersection is shown in Table C-1 in Appendix C. Using the number of reported accidents and the durations of the before and after periods, the equivalent number of accidents per year was calculated for each study approach. This information is presented in Table 4-2. It should be noted that accidents involving bicycles were not included in the analysis. At all of the intersections, there were only two bicycle related accidents in the before period and one in the after period. The average daily volumes for each of the intersection approaches were obtained from the Pima County Department of Transportation. The estimated average daily approach volumes forthe before and after periods are given in Table 4-3. These values reflect the total volume for an intersection Table 4-2. Equivalent Number 01Accidents per Year, Plma County Intersection Ajo Way / Palo Ver& Rd. Alvernon Way 1 Irvington Rd. Alvemon Way 1 Valencia Rd. Campbell Ave. / River Rd. Craycrofi Ave. / River Rd. Craycrofi Ave. / Sunrise Dr. Dodge Blvd. / River Rd. Dos Hombres / Tanque Verde Rd. First Ave. / Ina Rd. First Ave. / Orange Grove Rd. First Avc. / River Rd. Ina Rd. / La Canada Dr. Ina Rd. / La Cholla Blvd. Ina Rd. / Oldfather Rd. Ina Rd. / Thomydale Rd. Kolb Rd. / Valencia Rd. La Cholla Blvd. /Orange Grove Rd. Mission Rd. / Valencia Rd. Orange Grove R d / Skyline Dr. River Rd. / Swan Rd. Direction Northbound Southbound Southbound Eastbound Eastbound Westbound Northbound (P) Westbound Southbound Westbound Northbound Westbound Eastbound Westbound Northbound (*) Northbound Southbound Northbound Southbound Eastbound Westbound Eastbound Westbound Eastbound Northbound Southbound Eastbound Westbound Northbound (P) Southbound (P) Eastbound (P) Westbound (P) Northbound (P) Southbound (P) Eastbound (P) Westbound (P) Eastbound Westbound Northbound (P) Northbound Westbound Southbound (*) Number 01Accidents Per Year Before After Difference 0.00 0.41 0.4 1 0.80 0.00 0.80 Sunrise Dr. / Swan R d (P) - Protected only left turns (*) - Protected /permitted left tums in the before period and protected left turns in the after period Table 4-3. Estimated Approach Volumes, Pima County Approacb Volumes Before After Intersection Ajo Way / Palo Ver& Rd. Alvernon Way / Irvington Rd. Alvemon Way / Valencia Rd. Campbell Ave. / River Rd. Craycrofi Ave. / River Rd. Craycroft Ave. / Sunrise Dr. Dodge Blvd. / River Rd. Dos Hombres / Tanque Verde Rd. First Ave. / Ina Rd. First Ave. / Orange Grove Rd. First Ave. / River R d Ina Rd. / La Canada Dr. Ina Rd. / La Cholla Blvd. Ina Rd. / Oldfather Rd. Ina Rd. / Thomydale Rd. Kolb Rd. / Valencia Rd. La Cholla Blvd. / Orange Grove Rd. Mission Rd / Valencia Rd. Orange Grove Rd. / Skyline Dr. River Rd. / Swan Rd. Sunrise Dr. / Swan Rd. Direction Northbound Southbound Southbound Eastbound Eastbound Westbound Northbound Westbound Southbound Westbound Northbound Westbound Eastbound Westbound Northbound Northbound Southbound Northbound Southbound Eastbound Westbound Eastbound Westbound Eastbound Northbound Southbound Eastbound Westbound Northbound Southbound Eastbound Westbound Northbound Southbound Eastbound Westbound Eastboimd Westbound Northbound Northbound Westbound Southbound (vpd) 14,709 15,210 7.490 6,104 9,624 10,203 12,062 8,078 7,456 4.23 1 6,450 5,169 16,O 10 16,610 4,763 8,683 5.23 1 11,428 11,002 12,664 12,460 11,612 13,527 12,942 6,740 5,804 11,782 10,525 3,972 9,863 8,429 1,105 8,058 4,219 6.1 13 7.87 1 8,695 11,625 10.57 1 9,644 2,372 4,704 (vpd) 12.127 11,376 13,788 6,256 10,703 7,854 13,568 10,040 7,607 6,415 7,856 8,004 18,959 17,384 5,547 9,059 5,268 11,914 11,710 15,054 14,488 15,202 15,860 14,808 8,383 8,414 14,242 19,253 3,970 12,418 8,098 1,205 7,610 5,815 6,244 8,119 10,475 12,878 12,417 11,655 3,548 4,23 1 approach. While it would have been desirable to have only the left turn approach volume, this level of detail was not available. It is recognized that the true left turn accident rate should be basedon the volume of left turn vehicles entering the intersection. Because this information was not available, the accident rate was based on the total approach volume. In this way, the influence of exposure and the durations of the study periods were considered. In view of the fact that the before and after periods were separated only by a three month driver adjustment period, it would be expected that the proportion of left turning vehicles in the approach volumes would remain about the same. The accident rate (basedon the total approach volume) for each of the intersection approaches is shown in Table 4-4. A number ofdifferent statistical tests have historically been applied to accident analyses. Typically, these analyses evaluate differences in accidents or the average accident rate. In some cases, differences in accident rates for the before and after periods are compared with the experience at other intersections which were not subjected to a given treatment. After considering the nature ofthe data set, it was decided to apply the Wilcoxen Signed-Ranks Test. Basically, the test examines the direction of the difference within a sample pair as well as the relative magnitude of the difference. It provides a means of analyzing the experience for each of the intersection approaches in addition to examining the collective results of the total sample. An analysis of the total intersection accidents was also undertaken for the purpose of examining any possible effect ofthe change in left turnoperation on the total intersection safety. In order to eliminate possible impacts of other changes at the intersections, only the accidents associated with the approaches included in the left turn study were evaluated. Appendix C contains tables which summarize the data for all accidents on the approaches considered. Discussion of Results The summary results of the Wilcoxen test based on accident rates are presented in Table C-2 in Appendix C. In reviewing Table C-2, it may be noted that some of the intersection approaches are not listed in the table. This is due to the fact that samples with no difference in the before and after periods are dropped from the statistical test. Only the intersection approaches for which there was a difference in the accident rate are shown in Table C-2. Similar information for the analysis of the equivalent number of accidents is given in Table C-3 in Appendix C. Also, information related to the analysis of total intersection accidents is in Appendix C. For the analyses, the null hypothesis was that there is no difference in the accident experience for the before and after periods. At the 95 percent confidence level, the analyses indicate that the hypothesis should be accepted in both cases. In essence, the use of accident rates, the use of the equivalent number of left turn accidents, or the use of total accident data yielded the same results. The conclusion, therefore, is that there was no difference in the accident experience. While the Wilcoxen test did not indicate statistical significance, the actual change in number of accidents was calculated for the left turn accidents and the total accidents. Based on the number of accidents per year, there was a 13.8 percent increase in left turn accidents while the total accidents decreased by 1.5 percent. Table 4-4. Left-TurnAccident Rate, Plma County Intersection Ajo Way / Palo Verde R d Alvernon Way / Irvington R d Dlrectlon Noahbound Southbound Southbound Alvemon Way / Valencia Rd. Eastbound East- Campbell Ave. / River R d Craycrofl Ave. / River R d Craycroft Ave. / Sunrise Dr. Dodge Blvd / River Rd. Dos Hombres / Tanque Verde Rd Fist Ave. / h a R d First Ave. / Orange Grove R d First Ave. / River R d Ina Rd / La Canada Dr. Ina R d / La Cholla Blvd Ina R d / Oldfather R d Ina R d / Thomydale R d Accident Rate (Accidents / MlWon Entering Vehicles) Before After Mereme 0.000 0.180 0.000 0.000 0.142 0.537 0.114 0.848 0.551 0.000 0.000 Westbound No~thbound(P) Westbound !buthbound Westbound Northbound Westbound Eastbound Westbound Northbound (*) Northbound Southbound Northbound Southbound Eastbound Westbound Eastbound Westbound Eastbound Northbound Southbound Kolb Rd / Valencia R d Westbound Northbound (P) Eastbound (P) La Cholla Blvd /Orange Grove R d Mission R d / Valencia R d Orange Grove R d / Skyline Dr. River R d / Swan R d W ~ ~ C P Northbound (P) Mhbound(P) Eastbound (P) westbound(P) Eastbound Westbound Northbound (P) Northbound Westbound Southbound(*) 0.171 0.165 0.288 0.473 0.262 0.379 1.180 0.325 0.1 10 0.236 0.405 0.000 0.407 0.236 0.233 0.521 0.000 0.000 0.730 0.000 0.000 0.000 0.000 0.000 0.000 0.197 0.184 - 0.197 - 0.184 0.384 0.000 0.354 1.037 0.554 0.000 0.466 0.000 0.216 0.264 1.213 0.724 0.000 0.000 + 0384 0.464 Eastbound ) - 0.094 0.094 0.000 0.000 1.274 0.740 0.000 0.683 0.693 0.61 1 0.000 0.000 0.145 0.062 0.067 0.000 0.629 0.000 0.287 0.487 0.783 0.163 0.155 0372 0.245 1.298 0.719 0.085 0.000 0.000 0.000 0.000 Sunrise Dr. 1 Swan R d (P) - Protected only left turns (*) - Proteded /permitted left turns in the before period and protected left tums in the after period + 0.180 - 1.274 - 0.598 + 0.537 - 0.569 + 0.155 - 0.060 + 0.319 + 0.109 + 0.098 + 0.288 -0.156 + 0.262 + 0.092 + 0.693 - 0.458 - 0.053 + 0.081 + 0.033 - 0.245 - 0.891 - 0.483 + 0.148 + 0.521 + 0.730 - 0.216 + 0.090 - 0.176 - 0.170 + 0.466 Because the Wilcoxen test examinesthe experienceat each intersection, the analysis was undertaken by including all of the approaches into a single group. In this way, the collective result of the conversion to lagging left turns was analyzed. It can be argued that the analysis should be accomplished by considering the approacheswith protected/permittedoperations separate from the approaches with protected only operations. Subsequent evaluation of the approaches separated by type of operation also indicated variation in the accident experiencein both group with no statistical difference at the 95 percent confidence level. CITY OF TUCSON ANALYSIS Introduction The conversion from leading to lagging left turn operation in the City of Tucson was accomplished in 1985. At that time, virtually all traffic signals in the City with protected left turn phases were converted to the lagging left turn operation. In the City of Tucson, the practice is to use permitted left turns with the protected movement. For the evaluation of the accident experience in the City of Tucson, a "before and after" type of analysis was again used. Someof thedetailedinformation about accidents as well as the intersections was not readily available; thus a slightly different approach was taken for the analysis. Data Collection The City of Tucson furnished computer summaries of intersection accident data. Because the conversion in signal operation occurred in 1985, that year was eliminated from the analysis. Data for a before period from 1982 to 1984and an after period of 1986to 1987 were compiled by the City from the computerized accident records. The information indicated the total accident rate for an intersection as well as the number of accidents by general types. Again, an accident was considered to be a left turn accident if a left turner was involved. The City also sorts the intersectionsby type. For example,the data was compiled forthe intersection of major arterial streets and for the intersection of major arterials with collector streets. Generally, signals at the intersection of major arterials will have protected left turn phases on all approaches; and signals at intersections with collector streets will have left turn phases on the major arterial approaches. As was the case with the Pima County situation, it was necessary to screen the list of intersections for the purpose of eliminating those where other obvious changes had been made. This resulted in 50 intersectionsinvolving major arterials and 12 intersectionsofmajor arterials with collector streets being included in the study. Analysis For the analysis, the initial problem was to find a method for determining the left turn accident rates. The total approach volumes were available for each intersection; however the left turn volumes were unknown. Thus, it was not possible to directly determine the left turn accident rate at each of the intersections. As a surrogate measure, the left turn accident rate was calculated by multiplying the total intersection accident rate by the ratio of left turn accidents to all accidents. In essence, this resulted in a value that is based on the total left turn accidents within the intersection, the average total entering intersection volumes, and the time period over which the data were gathered. The summaries of information used for both intersection groups are shown in Tables 4-5 through 4-10. The Wilcoxen test was also utilized to statistically evaluate the experiencein the before and after periods. This analysis was accomplished for each of the two groups of intersections using accident rates as well as the number of accidents per year. The results of the statistical tests are in Tables C-4 through C-7. As was the case with the Yima County intersections, it was concluded that there was no significant difference in the before and after accident experience. For the two categoriesof City ofTucson intersections, the total intersection accidents were also compiled. This compilation included an examination of all reported accidents at each of the intersections. The summary tables indicating the total intersection accident experience are included in Appendix C. Again, the analysis of all accidents did not indicate any statistical difference in total accident experience. Discussion of Results The analysis of left turn accidents in Pima County and the City of Tucson indicated that there were no significant differences resulting from the conversion from leading to lagging left turn operations. This finding is somewhat contrary to the comments in the literature that indicate that the lagging left turn operation results in a more hazardous condition. The examination of the change in the number of accidents per year yielded the following results for the before and after periods: - At the intersection of major arterial streets, the number of left turn accidents decreased by 2.8 percent while all accidents decreased by 6.1 percent. For the group of intersections involving major arterials and collector streets, the left turn accidents decrease by 11.3 percent and all accidents decreased by 17.8 percent. Certainly, a review of the accident experience at the individual intersections reveals considerable variation in the results. At some intersections, there were large increases or decreases in accidents. There wasnothing in the available information about the intersections that would explain these variations other than the random nature of accidents. It should be noted that in Pima County as well as the City of Tucson lagging protected left turn phase overlaps are not used in conjunction with permitted left turn operations. This means that the relatively hazardous "left turn trap" condition does not occur. For this reason, it is not surprising that a difference was not found between the leading and lagging conditions. Some traffic engineerswill suggest that the lagging left turn operation is safer because themotoristknows that the left turn can be made at the end of the permitted phase. In this case, the driver is not as pressured to make the permitted left turn. Generally, the City of Tucson uses permitted/protected left turn operations at intersections with protected left turn phases. Again, the datadid not reveal any significant differences. Table 4-5. Left-Turn Accidents, City of Tucson Arterlal I Arterial Interseetlons Number of Left Turn Accidents Intersection 1982 1984 1986 1987 Ajo Way / Mission Rd. 18 13 Ajo Way / Interstate 19 0 0 24 10 Ajo Way / 12th Ave. Alvemon Way 1 Broadway Blvd. 23 9 11 1 41 Alvemon Way / 22nd St. 14 26 Broadway Blvd. / Campbell Ave. Broadway Blvd. / Country Club Rd. 17 14 16 8 Broadway Blvd. / Craycroft Rd. Broadway Blvd. / Kolb Rd. 16 11 16 9 Broadway Blvd. / Swan Rd. 24 7 Broadway Blvd / Wilmot Rd. Campbell Ave. / Fort Lowell Rd. 35 16 28 26 Campbell Ave. / Grant Rd. Campbell Ave. / Speedway Blvd. 31 21 3 3 Congress St. / Granada Ave. Congress St. / Interstate 10 20 9 Country Club Rd. / Grant Rd. 11 13 10 6 Country Club Rd. I Speedway Blvd. 4 6 Country Club Rd. / Valencia Rd. Craycroft Rd. / Golf Links Rd. 9 21 Craycroft Rd. I 22nd St. 46 32 Fort Lowell Rd. / Oracle Rd. 3 16 Golf Links Rd. / Kolb Rd. 16 16 Golf Links R d 1 Wilmot Rd. 24 18 24 18 Grant Rd. / Oracle Rd. Grant Rd. / Stone Ave. 13 21 Grant Rd. / Swan Rd. 30 15 36 27 Grant (Kolb) Rd. / Tanquc Verde Rd. Grant Rd. / First Ave. 21 7 Grant Rd. I Interstate 10 2 13 28 14 Kolb Rd. / Speedway Blvd. Kolb Rd. / 22nd St. 55 27 Main Ave. I Speedway Blvd. 5 13 Miracle Mile 1 Oracle Rd. 2 14 Nogales Highway I Valencia Rd. 28 13 Oracle Rd. I Prince Rd. 34 26 Oracle Rd. / River Rd. 5 8 Oracle Rd. / Wetmore Rd. 3 11 20 10 Speedway Blvd. / Stone Ave. 5 3 Speedway Blvd. / Swan Rd. 12 12 Speedway Blvd. / Wilmot Rd. Specdway Blvd. / Interstate 10 9 3 14 3 St. Mary's R d / Interstate 10 Swan Rd. / 22nd St. 43 5 Valencia Rd. / 12th Ave. 20 12 Wetmore Rd. / First Ave. 5 7 Wilmot R d 15th St. 12 9 Wilmot Rd. / 22nd St. 39 29 12 5 Interstate 10 / 22nd St. 5th Ave. / Interstate 10 5 2 - 42 - Table 4-6. Average I ~ f t - T u r nAcci&nts per Year, City of Tucson Arterial I Atterial lntersectlonr Left Turn Accidents P e r Year Intersectlon Before After Difference Ajo Way / Mission Rd. 6 .OO 6.50 0.50 Ajo Way / Interstate 19 Ajo Way 1 12th Ave. Alvemon Way / Broadway Blvd. Alvernon Way / 22nd St. Broadway Blvd. / Campbell Ave. Broadway Blvd I Country Club R d Broadway Blvd. / Craycroft Rd. Broadway Blvd. I Kolb Rd. Broadway Bivd. / Swan R.d. Broadway Blvd. / Wilmot Rd. Campbell Ave. / Fort Lowell R d Campbell Ave. / Grant Rd. Campbell Ave. / Speedway Blvd. Congress St. / Granada Ave. Congress St. / Interstate 10 Country Club Rd. / Grant Rd. Country Club Rd. I Speedway Blvd. Country Club Rd. / Valencia Rd. Craycroft Rd. I Golf Links Rd. Craycroft Rd. / 22nd St. Fort LoweH Rd. / Oracle Rd. Golf Links Rd. I Kolb Rd. Golf Links Rd. I Wilmot Rd. Grant Rd. / Oracle Rd. Grant Rd. / Stone Ave. Grant Rd. / Swan Rd. Grant (Kolb) Rd. / Tanque Verde Rd. Grant R d / First Ave. Grant Rd. / Interstate 10 Kolb Rd. / Speedway Blvd. Kolb Rd. / 22nd St. Main Ave. / Speedway Blvd. Miracle Mile I Oracle Rd. Nogales Highway I Valencia Rd. Oracle Rd. / Prince Rd. Oracle Rd. / River R d Oracle Rd. / Wetmore Rd. Speedway Blvd. / Stone Ave. Speedway Blvd. / Swan Rd. Speedway Blvd. I Wilmot Rd. Speedway Blvd. / Interstate 10 St. Mary's Rd. I Interstate 10 Swan Rd. / 22nd St. Valencia Rd. / 12th Ave. Wetmore Rd. I First Ave. Wilmot Rd. I 5th St. Wilmot Rd. / 22nd St. Interstate 10 / 22nd St. 5th Ave. / Interstate 10 Total Percent Change -2.84% 43 TabIe 4-7. Left-Turn Accident h t e , City of Tucson Arterial / Arterial Intersections Left Turn Accident Rate Intersection 1982 1984 1986 1987 Ajo Way 1 Mission Rd. 0.453 0.485 Ajo Way 1 Interstate 19 Ajo Way / 12th Ave. Alvernon Way I Broadway Blvd. Alvernon Way / 22nd St. Broadway Blvd. 1 Campbell Ave. Broadway Blvd. / Country Club Rd. Broadway Blvd. / Craycroft Rd. Broadway Blvd. / Kolb Rd. Broadway Blvd. / Swan Rd. Broadway Blvd. / Wilmot Rd. Campbell Ave. I Fort Lowell Rd. Campbell Ave. I Grant Rd. Campbell Ave. 1 Speedway Blvd. Congress St. I Granada Ave. Congress St. I Interstate 10 Country Club Rd. 1 Grant Rd. Country Club Rd. / Speedway Blvd. Country Club Rd. / Valencia Rd. Craycrofl Rd. / Golf Links Rd. Craycroft Rd. / 22nd St. Fort Lowell Rd. I Oracle Rd. Golf Links Rd. 1 Kolb Rd. Golf Links Rd. I Wilmot Rd. Grant Rd. I Oracle Rd. Grant Rd. 1 Stone Ave. Grant Rd. / Swan Rd. Grant (Kolb) Rd. / Tanque Vexde Rd. Grant Rd. / First Ave. Grant Rd. / Interstate 10 Kolb Rd. / Speedway Blvd. Kolb Rd. I 22nd St. Main Ave. / Speedway Blvd. Miracle Mile 1 Oracle Rd. Nogales Highway / Valencia Rd. Oracle Rd. 1 Prince Rd. Oracle Rd. / River Rd. Oracle Rd. / Wetmore Rd. Speedway Blvd. I Stone Ave. Speedway Blvd. / Swan Rd. Speedway Blvd. / Wilmot Rd. Speedway Blvd. / Interstate 10 St. Mary's Rd. / Interstate 10 Swan Rd. 122nd St. Valencia Rd. I 12th Ave. Wetmore Rd. / First Ave. Wilmot Rd. / 5th St. Wilmot Rd. / 22nd St. Interstate 10 / 22nd St. 5th Ave. / Interstate 10 * Accidenls per million entering vehicles - - Table 4-8. Left-Turn Accidents, City of Tucson Arterlal I Collector Intersections Intersection Alvernon Way / 29th St. Auto Mall Dr. /Oracle Rd. Broadway Blvd. / Columbus Blvd. Broadway Blvd / Randolf Way Broadway Blvd. I Rosemont Blvd. Cherry Ave. / 22nd St. Columbus Blvd. / 22nd St. Grant R d / Wilmot Rd. Limberlost Rd. / First Ave. Oracle Rd. / Roger Rd. Santa Clara Ave. / Valencia Rd. Tucson Blvd. / Valencia Rd. Number of Left Turn Accldents 1982 1984 1986 1987 29 14 1 3 13 3 11 7 5 3 11 11 26 8 7 11 5 4 6 1 7 2 16 14 - - Table 4-9. Len-Turn Accldents per Year, City of Tucson Arterial I Collector Intersections Left T u r n Accidents Per Year Intersection Before After Difference Alvernon Way 129th St. 9.67 7.00 -2.67 Auto Mall Dr. / Oracle Rd. 0.33 1.50 1.17 Broadway Blvd. / Columbus Blvd. 4.33 1.50 -2.83 Broadway Blvd. / Randolf Way 3.67 3.50 -0.17 Broadway Blvd / Rosemont Blvd. 1.67 1.50 -0.17 Cherty Ave. / 22nd St. 3.67 5.50 1.83 Columbus Blvd. / 22nd St. 8.67 4.00 -4.67 Grant Rd. / Wilmot Rd. 2.33 5.50 3.17 Limberlost Rd. / First Ave. 1.67 2.00 0.33 Oracle Rd. / Roger Rd. 2.00 0.50 -1.50 Santa Clara Ave. / Valencia Rd. 2.33 1.OO -1.33 Tucson Blvd. I Valencia Rd. 5.33 7.00 1.67 Total Percent Change 45.67 40.50 -5.17 -1 1.32% Table 4-10, Left-Turn Accident Rate, City of Tucson Arterial ICollector Lntersections Intersection Alvernon Way / 29th St. Auto Mall Dr. I Oracle Rd. Broadway Blvd. I Columbus Blvd. Broadway Blvd. I Randolf Way Broadway Blvd. I Rosemont Blvd. Cherry Ave. / 22nd St. Columbus Blvd. 122nd St. Grant Rd. I Wilmot Rd. Limberlost Rd. / First Ave. Oracle Rd. / Roger Rd. Santa Clara Ave. I Valencia Rd. Tucson Blvd. / Valencia Rd. Accidents per million entering vehicles Left T u r n Accident Rate* 1982 1984 1986 1987 - - 0.502 0.017 0.159 0.273 0.1 18 0.211 0.489 0.126 0.161 0.1 83 0.216 0.474 0.351 0.073 0.058 0.236 0.12 0.347 0.233 0.263 0.175 0.042 0.079 0.652 PART III TRAFFIC OPERATIONS STUDIES PART I n of this report discusses the effort undertaken to determine the differences in trafic operation between leading left turn phasing and lagging left turn phasing. CHAPTER 5 discusses the various intersection delay studies that were conducted in the Phoenix area. The first comparison made is that between leading and lagging operation operating at six intersections in the Phoenix area. From the analysis it is shown that total intersection delay is significantly greater with lagging left turns. The second comparison is between leading and combination operation at one intersection in Mesa. Due to the small sample size, a statistical test cannot be performed here. The third comparison is between Third Car and First car actuation operating at five intersections in the Phoenix area. These comparison showed no significant difference in total intersection delay, but a significant increase in left turn delay for the 3rd car actuated condition. CHAPTER 6 discusses the signal operation analysis performed in the Pima County area. Leading left turn operation was compared to lagging left turn operation at actuated-isolated-unsaturated signals in the Pima County area. In a comparison in percent stopped vehicles, there is no change between leading and lagging operation. Vehicle delay increased at all intersections. CHAPTER 7 discusses the travel time and delay studies that were performed in the Phoenix area. In Glendale and Tempe, nine routes were studied with four different timing plans - Existing allleading, optimized all-leading, optimized all-lagging, and optimized combination. Leading and lagging were not permitted in opposing directions due to the trap situation. The results show no consistentresult in the operation ofthese variouspatterns. In Mesa, however, leading left turn phasing was compared to leading eastbound, lagging westbound phasing. This phasing is different in that these phases are protected only and leading and lagging were permitted in opposing directions. The result of the Mesa study showed that delay, travel time and stops were all reduced with the combination phasing, although not significantly. CHAPTER 5 PHOENIX AREA INTERSECTION ANALYSIS INTRODUCTION Intersection stopped time delay studies were conducted to evaluate the difference in performance between leading and lagging left turn arrow operation. One intersection was studied to evaluate the difference between leading and combination leading and lagging operation. Delay studies were also conducted, as part of this research, to evaluate the difference between 3rd car and 1st car actuation. The 3rd car/lst car comparison was performed strictly for the leading operation. DATA COLLECTION The studyofcon~parisonstook the form ofa before and aAer analysis, therefore special care was taken to insure similar conditions existed for each study performed at aparticuiar intersection. Theduration of each study was one hour during the PM peak. Each study was conducted in good weather under normal traffic conditions. Measurement of intersection delay was performed by direct observation of stopped vehicles counted at fifteen second intervals. One observer was assigned to each approach. A turning movement volume count was performed for each study. Vehicles were counted as they entered the intersection. Volume count summaries were generated for each 15 minute interval. The average stopped time delay was calculated using the equation: DELAY = (L: Vs* 15)N where: DELAY Z vs 15 V = average delay, in seconds/vehicle; = sum of stopped vehicle counts; = interval between stopped vehicle counts, in seconds; and = total volume observed during the study period. Average stopped time delay values were calculated for left turn vehicles, throughlight turn vehicles, and total intersection approach vehicles. Summary worksheets for each delay study are presented in Appendix D. The following is a discussion on each of the comparisons that were made. ANALYSIS Leading Versus Lagging Operation A paired comparison was made between the average delay per vehicle in the leading condition and the average delay per vehicle in the lagging condition. Six intersections were used in the analysis: - 5 1st Avenuelo1ive, 5 1st Avenue/Peoria, 48th StreetISouthern, and 48th StreetBroadway. Manual stopped time delay studies were conducted at each intersection prior to any signal timing changes associated with this research. In the before condition, each of the six intersections operated with leading left turns. Five of the six intersections operated with protected,permissive left turn phasing and third car actuation on all approaches. The 48th St./Broadway intersection operated with protected only left turns and first car actuation on the northbound and southbound approaches and protectedlpermissive left turn phasing with third car actuation on the eastbound and westbound approaches. Manual stopped time delay studies were conducted at each intersection with lagging operation. All approaches which were protected/pennissive in the leading condition were pemissive/protected in the lagging condition. The two protected only approaches remained protected only in the lagging operation. Results A before and after difference in the average stopped time delay per approach vehicle was calculated for each intersection. A difference was calculated for left turn vehicles, throughlright turn vehicles, and total intersection approach vehicles. The percent change in delay from the before to the after condition was also calculated. The results ofthe Phoenix area intersection analysis of leading versus lagging left turn operation are presented in Table 5-1 and Figures 5-1,5-2, and 5-3. Average stopped time delay per left turn approach vehicle increased in the after condition at four of the six Intersections studied. The largest change occurred at 51st Ave./Northem, where delay increased by 139% for left turn vehicles. The 48th St./Southern intersection measured essentially no change for left turn vehicle delay with the conversion to lagging left turns, while the intersection of48th StreetJEIroadwayregistereda 5% decrease in delay for left turn vehicles in the after condition. Average delay per through/right turn approach vehicle increased at five of the six intersections studied. The largest increase occurred at 48th St./Southem, with 129% more delay for throughhight turn vehicles in the after condition. The 5 1st Ave.JNorthern intersection was the only intersection which registered a decrease in delay for throughhght turn vehicles in the after condition. Delay decreased approximately 16% at this location. Average delay per total approach vehicle also showed increases in the after condition at the same five intersections, though the changes were not as drastic when total intersection approach vehicles were considered. The large increase in throughhight turn delay at 48th St./Southern was partially offset by no change in left turn delay. However, this intersection still registered the largest increase (85%) in total intersection delay with the conversion to a lagging operation. The 51st Ave./Northern intersection was theonly location which registered an overall improvement in the after condition with a decrease in total intersection delay of approximately 4%. Table 51. Leadlng vs. Lapglng Intersection Delay, Phoenlx Area Delay per Approach Vehicle (seclveh) I Intersection Left Turn ThrulRight 1. 51 st AveIGlendale Before 25.70 22.55 After DiFference Change I 2. 51 st Avernorthem Before After Difference Change 3. 51st Ave/Olive Before After Difference Change 4. 51st AvePeoria Bcfore After Difference Change 5. 48th StISouthern Before After Difference Change 6. 48th StlBroadway Before After Difference Change Total 22.95 Analysis 6 17.16 63.30% 15.62 2.691 yes @=.04) Sample Size Mean of Difference Overall Change Sample Standard Deviation Test Statistic ( t ) Significant @ 95%? 6 9.47 45.54% 11.58 2.002 no @=. 10) I Before Condition: After Condition: Leading Operation Lagging Operation Statistical Analysis Three statistical tests were performed: a difference by intersection left turn movements, a difference by intersection throughtright turn movements, and a difference by total intersection delay. 6 10.38 42.17% 9.00 2.825 yes e.04) Table 5-2. Leading vs. Comblnalion intersection Delay, Phoenix Area Delay per Approach Vehlcle Intersection Southern/Stewart Before Condition: After Condition: Left Turn ThruIRJght Total Before After Difference 37.86 33.25 -4.61 10.76 -1.13 14.34 13.02 -1.32 Change -12% -11% -9% 9.63 Leading Operation Combination (leading EBIlagging WB) phasingpattern. A separate statistical analysis was not performed due to the limited sample size. The primary motivation for this study was to evaluate possible improved progression rather than intersection delay. The results of the progression analysis are included in Chapter 7 - Phoenix Area Travel Time Analysis. Third Car Versus First C a r Actuation A paired comparison was also made between the average delay per vehicle in the 3rd car actuated condition and the 1st car actuated condition. Five intersections were used in the analysis: - 48th Street/Southern, 48th Stree~roadway, 35th Avenue/Dunlap, 43rd AvenueNorthern, and 5 1st Street/Elliot. Manual stopped time delay studies were conducted at each intersection prior to any signal timing changes associated with this research. In the before condition, all five intersections operated with leading left turns. Three ofthe five intersections operated withprotected/permissiveleftturn phasing and 3rd car actuation on all approaches. The 48th Street/Broadway intersection operated with protected only left turns and 1st car actuation on the northbound and southbound approaches. The eastbound and westbound approaches operated with protected/permissive left turn phasing and 3rd car actuation. The 5 1st StreetElliot intersection operatedin afivephasemode in the before condition with protected/permissive left turn phasing and 3rd car actuation on the eastbound and westbound approaches. Manual stopped time delay studies were conducted at each intersection with 1st car actuation. All approaches which were 3rd car actuated in the before condition were converted to 1st car actuation in the after condition. The two protected only approaches remained protected only with 1st car actuation in the after condition. Results A before and after difference in the average stopped time delay per approach vehicle was calculated for each intersection. A difference was calculated for left turn vehicles, throughfright turn vehicles, and total intersection approach vehicles. The percent change in delay from the before to the after condition was also calculated. The results of the Phoenix area intersection analysisof 3rd car versus 1st car left turn actuation are presented in Table 5-3 and Figures 5-5,s-6, and 5-7. Average stopped time delay per left turn approach vehicle decreased in the after condition at four of the five intersections studied. The largest decrease occurred at 48th St./Southern, where delay decreased by approximately 32% for left turn vehicles. The 51st St./Elliot intersection recorded a 9% increase in delay for left turn vehicles in the after condition. Table 5-3. Third Car VS. Flrst Car Intersection Delav. Phoenir Area Intersection Delay per Approach Vehlcle (seclveh) Left Turn ThruIRigbt Total 1. 48th StlSouthern Before 54.95 2 1.56 27.23 37.43 After 27.66 29.35 Difference -17.52 6.10 2.12 Change -32% 28% 8% 2. 48th StBroadway Before After Difference Change 3. 35th Ave/Dunlap Before After Difference Change 4. 43rd Avernorthem Before After Difference Change 5. 51st St~Elliot Before After Difference Change 29.41 31.97 2.56 9% 9.55 11.11 1.56 1 6% 14.72 16.34 1.62 11% Analysis 5 5 Sample Size Mean o f Difference -9.80 0.3 5 7.5 1 Sample Standard Deviation 5.17 Test Statistic ( t ) -2.918 0.150 Significant @ 95%? yes @=.04) no (p=.89) Before Condition: 3rd Car Actuated Leading Operation 1st Car Actuated Leading Operation After Condition: 1 5 -1.19 4.47 -0.597 no w . 5 8 ) 70 -f 60 50 \ 40 H ./ i 32 0 10 0 0 I 2 1 3 4 5 hforsecflon 3rd Car = 1st C a Figure 5-5. Third Car VS.First Car Left Turn Delay, Phoenix Area 45 40 35 3 30 >: > 6 25 20 8 15 10 5 0 t 2 3 htessbctlon m 3 r d Car 4 5 1st l k Figure 5-6. Thlrd Car vs. Flrst Car ThroughlRight-Turn Delay, Phoenix Area 57 50 45 40 -35 >I3025 u -$a 2 0 a 15 10 5 0 1 2 3 Inlersection 3rd Car Figure 5-7. Thlrd Car vs. 4 5 1 st Car First Car Total Intersection Delay, Phoenix Area Average delay per through/right turn approach vehicle increased at three of the five intersections studied. The largest increase occurred at 48th St./Southern with a difference of 6.1 seconds per approach througwright turn vehicle. This represents an increase in delay of approximately 28%. Two intersections actually show a decrease in delay for throughlright turn vehicles with the conversion to 1st car actuation. The largest decrease occurred at 48th St./Broadway which recorded 7.8 1 seconds less delay or a decrease of 20% in the after condition. Average delay per total intersection approach vehicle increased at three of the five intersections studied. The largest percent increase occurred at 5 1st St./Elliot where delay increased 1 1% with the conversion to 1st car actuation. The largest overall increase in delay occurred at 48th St./Southem where total intersectiondelay increased 2.12 secondsper approach vehiclc in the after condition. This represents an increase in delay of 8% with the conversion to 1st car actuation. A decrease in total intersection delay occurred at two intersections. Delay decreased approximately 20% at 48th St./ Broadway and approximately 6% at 35th Ave./Dunlap. Statistical Analysis The same three statistical testsused in the leading versus lagging analysis were performed to evaluate the difference in delay per vehicle for 3rd car and 1st car actuation. The null hypothesis, once again, is that there is no significant difference in delay for the two conditions. A two tail test was performed at a 95% level of confidence. The results of the paired data analysis are also presented in Table 5-3. The critical t-value for the test is 2.776. The mean of the difference for left turn vehicles is -9.80 seconds per approach vehicle. The calculated t-value is -2.918. (p.=04) The absolute value of the calculated t-value is greater than the critical t-value therefore the test indicates a statistically significant decrease in delay at the 95% confidence level for left turn vehicles with 1st car actuation. The mean of the difference for through1 right turn vehicles is 0.35 seconds per approach vehicle. The calculatedt-value is 0.150 (p=.89).The calculated t-value is less than the critical t-value therefore the null hypothesis can not be rejected on the basis of this test. The mean of the difference for total intersection delay is -1.19 seconds per approach vehicle. The calculated t-value is -0.597 (p=.58). The absolute value of the calculated tvalue is less than the critical t-value therefore the null hypothesis cannot be rejected. On the basis of this test, it is concluded that total intersection delay is not significantly different at the 95% level in the before and after conditions although there is a significant difference in delay for left turn vehicles. DISCUSSION OF RESULTS The results of the leading versus lagging analysis in the Phoenix area indicate a leading operation tends to be more efficient in terms of intersection delay. This finding is somewhat contrary to other experiments documented in the literature (10) However, the results of this current study are not surprising given the current phasing practices for implementation of lagging left arrow operations in the State of Arizona. In this research there were no left turn phase overlaps with the lagging left turn operations. Therefore, the potential benefits associated with phase overlap were lost in the conversion to a lagging operation. The results of the leading versus combination lead~lagphasing indicate there may be reduced intersection delay with a combination phasing. However, the limited scope of the investigation precludes making any strong conclusions regarding the merits of combination phasing as related to intersection delay. The results of the 3rd car versus 1st car actuation are perhaps the most interesting results associated with the Phoenix area intersection delay studies. Asmight be expected, thedelay for left turn vehicles was less with 1st car actuation. However, the 5 1st St./Elliot intersection did record a 9% increase in delay for left turn vehicles in the after condition. However, this intersection does not provide a left turn phase for the north and south approaches. Therefore, the conversion to 1st car actuation did not serve as an advantage for left turners on the north and south approaches. More surprising,however, were the results recorded for the througwght turn and total intersection delay. The analysis indicates no statistically significant difference at the 95% confidence level in delay per through/right turn vehicle or total approach vehicle with the change to 1st car actuation. ?'his finding may, perhaps, be explained by examining the conditionsunder which the delay studies were performed. The 48th St./Broadway intersection recorded a 20% decrease in intersection delay with the conversionto 1st car actuation. This intersectionoperates very close to capacity during the PM peak. The intersection also recorded slightly higher volumes during the 3rd car actuation study. This increased volume could have pushed the intersection into cycle failure during the course of the 3rd car actuation study. All the delay studies were performed during the PM peak hour. Five intersections were used in the 3rd car versus 1st car analysis, or a total of 20 left turn approaches. All the intersections used in the analysis were running between 30 and 45 cycles per hour. Left turn volumes exceeded 150 vehicles per hour at 15ofthe 20 approachesduring the 3rd car actuated studies. These volumes would indicate the left turn arrow is being actuated a large proportion of cycles in the PM peak regardless of 1st car or 3rd car actuation. The City ofPhoenix collected data on thenumber of left arrow actuations andleft turn volumes during the time the intersections of35th Ave./Dunlap, 43rd Ave./Northern, and 5 1st St./Elliot wereoperated in the 1st car actuated mode. The same information was also collected for a typical weekday under 3rd car actuation. The data does indicate an increase in the number of times the amow is actuated with the conversion to 1st car detection. However, the proportional increase in the number of actuation is much more acutein the off-peak hours than during thePM peak. Therefore, greaterdelay reductions should be expected in the off-peak rather than the PM peak with 3rd car actuation. CHAPTER 6 PIMA COUNTY TRAFFIC SIGNAL OPERATION ANALYSIS INTRODUCTION As part of this project, an operational analysis of selected intersections in Pima County was undertaken. This analysis involved an examination of traffic and signal operation parameters before and after the conversion from leading to lagging turns. The purpose of this chapter is to document the data collection, analysis, and results of the study. In 1984, the City of Tucson initiated a program to convert exclusive left turn signal phases from leading to lagging operation. The actual conversion of the signal phases began in 1985. Basically, the general rationale for this change was attributed to the potential for improving the operation of the signalized intersections with exclusive left turn phases. More specific reasons were related to: the need for the exclusive turn phase only when the demand in a signal cycle exceeded the capacity of the permitted left turn movement, and the influence of the leading left turn on arterial progression. Given the conversion of the traffic signal operation by the City of Tucson, there was some lack of uniformity in the left turn phasing because the traffic signals under the jurisdiction of the Arizona Department ofTransportation and the Pima County Department of Transportation continued to use the leading left turn phase. In order to eliminate the confusion to the motorists, all governmental agencies subsequently converted to the lagging left turn phase when exclusive turn phases are used. The Pima County Department of Transportation converted to the lagging left turn type of operation in the spring and summer of 1987. The conversion from leading to lagging left turn signalsby Pima County in 1987representedaunique opportunity to examine the effect of the operational change. The time schedule for the conversion prevented the development of a formal funded research program on short notice; however a cooperative data collection effort was organized by Dr. Robert H. Wortman. With the cooperation of the Pima County Department of Transportation and the Arizona Department of Transportation, a "before and after" data collection effort was undertaken at selected intersections. This data set was then used as the basis of the comparative analysis of the left turn phasings in this research study. Selection of Intersections The conversion program in Pima County involved a total of thirty-seven signalized intersectionsin the Tucson area. At some of these intersections,various modifications to the signal operations were made in addition to the conversion of the left turn phasing. At a limited number of intersections, the only planned change was to switch fiom the leading to the lagging left turn operation; thus these intersections were selected for the "before and after" data collection. The intersections studied are listed in Table 6-1. Ultimately, the intersection of First Avenue and Ina Road had other changes in the signal phasing Table 61. Delay Study Intersections, Pima County Intersectlon Type of Control (a,b) Ajo Way / Alvernon Way 4 Phase (c) Alvernon Way / Irvington Rd. 4 Phase (Protected/Pennissive) Campbell Ave. / Skyline Rd. 3 Phase (Protected) First Ave. / Ina Rd. 4 Phase (d) First Ave. / Orange Grove Rd. 3 Phase (ProteddPermissive) First Ave. / River Rd. 3 Phase (ProtededlPermissive) Ina Rd. / Thomydale Rd. 4 Phase (ProtectedIPermissive) Kolb Rd. / Valencia Rd. 4 Phase (Protected) Palo Verde Rd. / Valencia R d 3 Phase (Protected) - (a) The number of phases reflects the basic operation of the intersection. Phase overlaps were used in situations with opposing leading protected left turns. (b) In the "aftern wndition, the "protected / permissive" left turn operation obviously becomes "permitted / protected". (c) At the intersection of Ajo Way and Alvernon Way, a combination of types of control were used. For example, some approaches had protected left turn operations. (d) At the intersection of First Avenue and Ina Road, a Cphase signal operation was used in the before condition with protected / peimissive left turns on the northbound and westbound approaches. For the after wndition, the northbound and southbound approaches on First were treated RS separate phases. In addition, the lane use on the northbound approach was changed. as well as modifications in the lane use. While field data were collected at the site, the intersection was eliminated from the comparative analysis for this reason. In addition, the initiation of construction in the area of Ina Road and Thomydale Road significantly changes the traffic at that location prior to an opportunity to collect the "after" data. While the changes in signal phasing at these intersections were limited to the conversion of the left turn operations, it must be recognizedthat there were somemodifications in the signal timing. These modifications included the adjustment of green time allocation for specific movements as well as a re-evaluation of the time for the clearance or change interval. Signal Phasing Pima County uses actuated control for traffic signals; thus all of the intersections in the study utilized fill actuated control. In addition, each of the intersections operated on an isolated basis with no interconnection with adjacent signals. There was some variation in the treatment of left turn movements in the study intersections. Some of the intersections had permitted plus protected left turn movements while other intersections had protected left turn movementsonly. Table 6- 1 identifiesthe operation ofthe left turn signal phasings at each of the study locations. As has been indicated previously, the intersectionofFirst Avenue and InaRoad ultimatelyunderwent changes in signal operation as well as lane use. These modifications significantly changed the operation of the intersection; thus the intersection was later eliminated from the analysis. It should be noted that phase overlaps were used for the leading left turn conditions; however, the overlaps were not used with the lagging left turn operations. For example, given an intersection with left turn arrows on all approaches, the signal would operate as an eight phase signal (or four phase with phase overlaps for the turn movements) with the leading left turns. With the lagging left turns, the signal would operate as four phase control (without any overlaps). This type of operation was standard in the Tucson area. In essence, this resulted in the loss of the use of the phase overlap when there were differences in the demand for the left turn movement. At a limited number of intersections which utilized the protected only left turns, a phase overlap condition would occur wit11the lagging left turn operation. For example, one intersection had very low westbound approach volumes. For some cycles, the eastbound through and left turn movements would occur at the same time. At another intersection, the side street had low volumes on the north and south approaches. In addition, the westbound approach had very few left turns. Because ofthe lack of westbound left turns and cycles without traffic on the side street, the eastbound through movement continued along with the eastbound protected left turn. In essence, there was no need for a protected westbound left turn phase in some cycles. This type of operation only occurred at intersections with protected only left turn operations. With respect to the actual signal timing, the study utilized the signal settings employed by Pima County for the before and after conditions. There was no attempt by the research team to evaluate the signal timing settings used at the intersections. Certainly, it was necessary for Pima County to adjust some of the signal timing settings in addition to changing from the leading to the lagging operation. Part of the reason for the necessity of adjusting the signal timings was associated with the fact that the loss of the left turn phase overlap had an impact on the phasing for the through movements. DATA COLLECTION For the field data collection, two timelapse super 8mm movie cameras wereused to film theoperation ofeach of the intersections. The location of the cameras was elevated by using trucks with elevating platforms or raised vantage points from nearby terrain. With the use of two cameras, it was possible to simultaneously film all of the intersection approaches. In addition, the cameras had internal clocks with digital displays; thus the filming with the two cameras could be coordinated. All filming was accomplished with the cameras operating on a speed of one frame per second. The filming of each intersection occurred during the period from 3 PM to 6 PM on weekday afternoons. There was an attempt to schedule the before and after data collection at a specific intersection on the same day of the week even though all intersections were not filmed on the same day of the week. While filming was scheduled for the period from 3 PM to 6 PM, it was not possible to obtain a three hour data set in all cases. Equipment problems in addition to the time required to change the film cartridges resulted in some lost time. The time period fiom 3 PM to 6 PM was selected for several reasons. First, this period permitted makingobservationsovera peak hourperiod. Second, it was possible to schedule theuse ofthe trucks with elevating platforms at the end of the normal workday. Finally, it was difficult to schedule personnel for data collection at other times. While it may have been desirable to collect data at other time periods, it must be recognized that the data collection was accomplished with resource limitations on a cooperative basis. Nevertheless, a rather extensive data set was collected for study. The before data were collected during the period from middle of March 1987 to the middle of May 1987. This time schedule was necessitated and constrain& by the timing of the signal conversions by Pima County. The after data collection began in early October 1987; thus there was a transition period of several months before the collection of the after data. During the fall of 1987, difficulties were encountered which served to disrupt and extend the data collection effort. For example, adverse weather and other demands for the use of the trucks made it impossible to film on some days. In addition, the shortening of the daytime period finally made it necessary to cease the data collection efforts until the spring of 1988. Data collection resumed in the spring of 1988. Difficulties were encountered with the films taken at the intersection of Kolb Road and Valencia Road; thus data collection was repeated at that intersection during the summer of 1990. ANALYSIS Using the film record of the intersections during the before and after periods, data which reflected operational parameters were extracted. This operational dab for each intersection was then used for the comparative analysis of the leading and lagging left turn phasing. The discussion that follows presents the analysis and results of each of the operationaI parameters. Intersection Volume In the design of the data collection effort, it was recognized that significant changes in volume can have a potential impact on the operational measures of intersection performance. For this reason, a number of precautions were taken in an attempt to minimize the possibility of major changes in volume between the before and after study periods. For example, the initiation of the data collection for the after period was undertaken within several months of the completion of the before data collection. The after data collection was delayed until the fall of 1987 to avoid the possible effect of the summer period. In addition, an effort was made to collect the before and after data at a given intersection on the same day of the week. The turn movement volumes for each intersection were obtained from the film records for the before and after periods. These volumes were then used to determine the actual changes in approach volumes for the two study periods. Because of the loss of some time during the data collection process, the raw turn movement volumes were expanded to the equivalent of a three hour period. The total volume was then divided by three to provide an average hourly volume. This procedure resulted in a number that could be used for comparing the before and after periods. Table 6-2 presents the average approach volumes for each intersection. At most of the study intersections, only minor differences in traffic volumes were observed. Given a relatively short period between the before and after data collection, only small differences would be expected. Table 6-2. Intersection Total Approach Volumes, Plma County Average Approach Volume (vpb)* Intersection Before After Difference Ajo Way / Alvemon Way 3 644 3523 -3 % AIvemon Way / Irvington Rd. 2788 2882 3% Campbell Ave. I Skyline Rd. 2527 3070 21% First Ave. I Orange Grove Rd. 2519 2472 -2% First Ave. / River Rd. 3379 3 107 -8% Ina Rd. / Thomydale Rd. 3 495 * Kolb Rd. / Valencia Rd 7052 5950 -16% Palo Ver& Rd. / Valencia Rd. 2560 2472 -3 % The average approach volumes are for the entire intersection. The value in the table reflects the sum of all approaches. ** After values not available for Ina R d / Thomydale R d Two exceptions to small changes in the approach volumes can benoted. Theintersection of Campbell Avenue and Skyline Drive, however, had a significant increase in traffic volume for which there is no explanation for the cause. The before data set was collected in April 1987, and the after data set was taken the following October. While there was only six months between the data collection periods, there was a 2 1 percent increase in the approach volumes at that intersection. This increase generally occurred on all approaches and throughout the study period. In essence, there was amajor increase in the use of the intersection. In the second case, there was a 16 percent decrease in the approach volumes at the intersection of Kolb Road and Valencia Road. At this location, there had been a major change in employment in the vicinity of this intersection; thus the after condition was influenced by the reduction in employment. In terms of later analyses of operational performance, it should be noted that significant increases in traffic volumes were found at only one intersection. In most cases, there was a reduction in the approach volumes. Arrival of Vehicles In addition to approach volume, the actual time of arrival of vehicles at an intersection during the signal cycle has an influence on overall performance measures. For this reason, the amval of vehicles was examined as part of the analysis of before and after conditions. All of the intersections in the study were operating on an isolated control basis. While there may be some platooning from adjacent signals, the overall arrival pattern should be random in terms of the time that vehicles arrive during a cycle. The arrival pattern of vehicles for a given intersection was examined by determining the percent of the approach vehicles that had to stop due to the operation of the traffic signal. Basically, the review of the film revealed the approach vehicles that were required to stop as well as the vehicles that were able to pass through the intersection without stopping. The percent vehicles stopped was then calculated by comparing the number of vehicles that stopped to the total approach volume. Table 6-3 summarizes this information for each of the intersections. A review of Table 6-3 reveals that there was little change in the percent vehicles stopped at half of the intersections. There was increaseof about five percent in the stopping vehicles atone intersection, and a decrease of about six percent at another. The greatest measuredchange was at the intersection of Kolb Road and Valencia Road where there was an increase of ten percent. This difference was based an only the east and west intersection approaches. Another interesting aspect of the information in Table 6-3 is the consistency of the values for the various intersections. At most ofthe intersections the percent of stopped vehicles was in the general range of fifty to fifty-five percent. The main exception was the intersection of Palo Verde Road and Valencia Road where the percentage for the before and after conditions was significantly lower than at other intersections. This lower value can be explained by the fact that there is a free flow right turn lane on one of the approaches. Vehicle Delay One of the main indicators of intersection performance is vehicle delay. Delay is specified in the Highway Capacity Manual (12) as the measure for determining intersection level of service. For this study, the stopped time delay was determined for each of the intersections. The standard procedure for determining stopped delay was used where the number of stopped vehicles is counted at a set interval. The intersections were filmed at one second intervals; thus thedelay was determined based on observations taken from the film. While it is common to select fifteen second intervals for observations, a ten second interval was used in this study. The shorter interval was used to improve the accuracy especially for the movements with low approach volumes. Table 6-4 summarizes the results of the delay analysis and indicates the average stopped delay for the stopped vehicles as well as the total approach vehicles. These values reflect the overall delay for Table 6-3. Percent of Approach Vehicles Stopped, Plma County Percent Stopped Before After Intersection 54.1 53.0 Ajo Way / Alvemon Way Alvernon Way / Irvington Rd. Campbell Ave. / Skyline R d First Ave. / Orange Grove Rd. First Ave. 1River R d Ina Rd. / Thornydale Rd. Kolb Rd. 1Valencia Rd. Palo Verde Rd. I Valencia Rd. 31.3 33 3 * Atter value not available for Ina Rd. 1 Thornydale R d ** At the Kolb Rd. / Valencia Rd. intersection, the values are for the eastbound and westbound approaches only. For the before condition, the percent vehicles stopped for all approaches was 49.2 percent. The after condition value for all approaches was not available. - Table 64.Vehicle Delay Comparison, Pima County Delay per Stopped Intersection Vehicle (Sec) Ajo Way / Alvemon Way Before 32.68 After 39.68 Difference 7.00 (2 1%) Delay Per Approach Vehicle (Sec) 17.75 2 1.04 3.29 (19%) Alvemon Way I irvington Rd. Before After Difference Campbell Ave. 1 Skyline Dr. Before Afier Difference First Ave. / Orange Grove Rd. Before After Difference First Ave. / River Rd. Before After Difference Ina Rd. 1 Thomydale Rd. Before After Difference Kolb Rd. 1 Valencia Rd. Before After Difference Palo Verde Way 1 Valencia Rd. Before After Difference Average Change * After value not available +20% +30% an intersection. More detailed information for each approach and the movement at each approach is contained in Appendix E. At all of the intersections where delay was actually measured, there were increases in the averagedelay per vehicle. Even for the intersections where there weredecreascs in the approach volume, the average vehicle delay increased. In considering the results of the delay analysis, it is important to recognize that the data collection was during the period from 3 PM to 6 PM. For this reason, a true offpeak condition was not included in the analysis. Cycle Length The final parameter that was included in the analysis of the before and after conditions was cycle length. Again, it should be noted that the research team did not attempt to evaluate the adequacy of the signal timing. The study simply utilized the signal timings that were in place prior to and after the conversion of the left turn signal operation. Upon the implementation of the lagging left turn signals, the Pima County staff did make some adjustments in the signal timing. These adjustments were necessary for the efficient operation of the intersection. The average signal cycle lengths for the before and after periods for each intersection are given in Table 6-5. A general review of the table reveals that the differences in the cycle lengths vary from intersection to intersection with increases at some of the sites and decreases at other locations. If the differences in cycle length are considered in terms of the type of left turn treatment, there is a trend that can be noted. At intersections where there was a decrease in the cycle length, the permitted/ protected left turn was utilized. The increases in cycle length were at intersections where protected only left turns were utilized. Changes in cycle length, therefore, were a function ofwhether left turns were permitted along with the through movement or not. The exception to an increase in cycle lengths with protected only lagging left turns occurred at the intersection of Palo Verde Rd. and Valencia Rd. At this intersection, the average cycle lengths remained virtually the same even with the protected left turn operations. Because ofthelow approach volumes for some movements, this is one of the intersections that resulted in a phase overlap type of operation. Because of this condition, the average cycle length remained the same. Table 6-5. Average Cycle Length, Pima County Intersection Ajo Way 1 Alvernon Way Alvemon Way I Irvington Rd. Campbell Ave. I Skyline Rd. First Ave. I Orange Grove Rd. First Ave. / River Rd. Ina Rd. 1Thomydale Rd. Kolb Rd. 1 Valencia Rd. Palo Verde Rd. I Valencia Rd. * After value not available for Ina Rd. /Thomydale Rd. Average Cycle Length (Sec) Dlffereace Before After 95.3 72.6 79.9 77.3 95.6 85.8 65.7 62.1 1143 70.4 90.3 71.9 90.7 76.7 62.6 19.0 -2.2 10.4 -5.4 -4.9 11.0 0.5 DISCUSSION OF RESULTS In considering the results ofthe anatysisofthePimaCounty intersections, it must be recognized that: all of the study locations were operating with actuated control; the signals were basically isolated from other intersections, and there was no coordination with adjacent intersections at the time of the data collection; the intersections were not operating at what could be considered as saturated conditions; and - vehicle queues generally cleared during each cycle. There was some variation in themeasured approach volumes at the study intersections; however only major changes occurred at two intersections. Because the intersections were not operating at saturated conditions, increases in volumes would not necessarily result in significant increases in delay. Generally, there was little change in the percent vehicles stopped. This would suggest that arrival pattern was random in terms of the signal cycle. For this reason, the effect of platooning should not be a factor with respect to delay calculations and measurements. It is significant to note that the reduction in cycle length was associated with intersections where permitted left turns were allowed. On the other hand, intersections with protected left turns only had increases in cycle length with the lagging left turn operation. This result is reasonable because of the fact that the opportunity for phase overlap was lost when the lagging left turn was used. In considering this general statement, it must be recognized that low traffic volumes for some movements can result in phase overlap operations with lagging protected only left turns. For this type of condition, the average cycle length did not increase. The interesting result of the analysis is that vehicle delay increased at all intersections. At the study intersections, there was an average increase of 20% in the delay per stopped vehicle and an average increase of 30%in the delay per approach vehicle. The finding of delay increases is consistent with the results of the Phoenix area studies. Even when there was a decrease in approach volumes, there were increases in delay. Delay might be expected to increase with longer cycle lengths; however delay also increased at intersections with reductions in average cycle length. CHAPTER 7 TRAVEL TIME STUDY INTRODUCTION As part ofthis research project alternativephasing sequences were tested using travel time dataalong five routes in Giendale and four routes in Tempe. The patterns tested were: - All Leading AllLagging A combination of leading or lagging depending on which best fit the progression. In addition, a combination phasing was tested along Southern Avenue in Mesa. This chapter documents the timing, data collection, analysis, and results of these travel time studies. SIGNAL TIMING In order to obtain a true comparison between leading and lagging left turns, it was necessary to use signal timing patterns developed by a common optimization program. Because of the ease of operation and the numerous runs that would be required as part of the combination portion of the study, FORCAST was utilized to optimize the signals. The first signal timing - all leading left turns, wasperfotmedusing FORCAST operatingon the City ofscottsdale computer. Subsequentruns were performed on the PC-based version of the FORCAST program. Initial travel time runs were performed along eachroute to determinetravel speeds and link distances. Existing intersection phasing and minimum times were obtained fiom the City of Glendale and City of Tempe Traffic Engineering staff. Traffic volumes were obtained from the city staff and the Maricopa Association of Governments Transportation Planning Division and were supplemented with turning movement counts made by Lee Engineering. FORCAST optimizes by calculatinga cost for each cycle length. Arange of acceptable cycle lengths is input into the program. Based upon the phasing, volumes, and progression priority, FORCAST creates an optimum timing plan for each cycle length. FORCAST then calculates the motorists' cost of each timing plan based upon the main street delay, side street delay, and stops. A stop is equivalent to 20 secondsof either main street or side street delay. FORCASTuses a simple procedure to increase the cost due to a saturated intersection. If the intersection is saturated, FORCAST adds one cycle length to both the main street and side street delay value. Timing plans wereimplemented in the study area for AMpeak,PM peak andoff-peak trafficpatterns. Since FORCAST allows for different phasing patterns, the coding and optimization of all leading or all lagging was straightforward. The most difficult part of the timing portion of the project was in determining which combination of lead and lag at each intersection would produce the optimum combination timing. For the Glendale timing, at each intersection along 5 1st Avenue there are 16 timing combinations theoretically possible as listed in Table 7-1. Table 7-1. Travel Time Study Possible Slgnal Comblnatlons, Phoenix Area Approach Combination Nortb South East West 1 Lead Lead Lead Lead 2 Lead Lead Lag Lead 3 Lead Lag Lead Lead Lead 4 Lag Lead Lag 8 Lead Lead Lead Lead Lag Lag Lag Lag Lead Lead Lag Lag Lead Lag Lead Lag 9 10 11 12 Lag Lag Lag Lag Lead Lead Lead Lead Lead Lead Lag Lag Lead 13 14 Lag Lag Lag Lag Lag Lag Lag Lag Lead Lead Lag Lag Lead Lag Lead Lag 5 6 7 15 16 Lead Lag If the four signals along 5 1st Avenue could be timed with any of the sixteen timing patterns, then in order todetermine the optimumphasingcombination at each intersection, timing plans would have to be generated for 160=65,536possible combinations. If there are three timing patterns- AM, MID, and PM, possible, the number jumps to 65,536 * 3 = 196,608 possible combinations. Since all of the signals along 5 1st Avenue operate in a protective-permissivemode, only four of the sixteen combinations can be utilized in order to avoid the trap. These four combinations are shown in Table 7-2. This reduced the number of patterns in Glendale to 256 plans per time period or a total of 768 combinations. In Tempe only two signals were changed from leading to lagging requiring the generation of only 48 timing plans. Table 7-2. Travel Time Study Utilized Signal Combinations, Phoenix Area Approach South East West Combination North Lead Lead Lead 1 Lead 4 Lead Lead Lag Lag 13 16 Lag Lag Lag Lag Lead Lag Lead Lag There was one additional constraint in that whatever combination was chosen, it had to be implemented for all three time periods of the day. Therefore if Peoria W B had lagging left turn in the AM peak, it had to remain that way for the off-peak and PM peak periods. Once the timing plans were generated the outputs were scanned for the lowest cost cycle length. This was not necessarily the same cycle length for each combination. A spreadsheet was then created which showed each combination and the costs associated with the AM, PM and off-peak plan. This Table 7-3. Lowest Cost Timing Plans, City of Glendale Intersection Phasing Glendale Olive Peorla Northern 2 2 2 2 2 2 2 2 2 2 2 2 2 2 1 2 2 2 2 3 2 3 3 2 3 3 2 2 2 2 4 2 2 2 3 2 2 1 3 1 Legend: 1 All Lagging 2 All Leading 3 Leading N-SLagging E-W 4 Leading E-WLagging N-S 1340 Welghted Cost Mid Total 1820 1804 1838 1824 1827 1840 1826 1948 1969 1821 1806 1806 1968 1840 1902 1826 1844 1841 1829 1815 1827 1837 1834 1849 1838 2246 1868 1872 1870 1851 2355 1872 1870 1874 5028 5041 5103 5118 5133 5165 5245 5265 5295 5302 5316 53 17 5357 5395 5396 5410 5410 5414 5425 5450 5462 5522 5542 5556 5558 5592 5639 5654 5684 5862 5874 5897 5972 5988 2774 was weighted by the volume to determine the lowest cost plan. The thirty lowest cost plans for Glendale are shown in Table 7-3 and the lowest cost plans for Tempe are shown in Table 7-4. It should be noted that the lowest cost timing plan for both cities was very similar to the all-leading timing plan. There are two reasons this occurred. By implementing lagging left turns, these must be tied together preventing an overlap scenario. Since there is a loss of effaciency associated with this type of phasing, FORCAST will only choose lagging if the left turn volumes are nearly identical. If lagging left turns are implemented, the greatest benefit is that should the left turn vehicles make their maneuvers during the through movement, ie. finding available gaps in the traffic stream, then it is possible that the protected left turn phase will not be necessary. FORCAST does not have an algorithm which determines if this scenario will occur, therefore, it cannot recognize the benefits of lagging left turns. There may be situations where the left turn traffic volume is light but unbalanced andFORCAST would choose leading operation, while lagging might be a better choice both because it fits better in the progression scheme and because lagging operation might avoid the need for some of the protected phases, which provides more time for the through movement. DATA COLLECTION Once the timing plans were implemented into the street, travel time runs were performed using the 'floating car' method. The TIMELAPSE Travelog data collection computer was utilized to collect this data. Once the routes are entered into the computer, the driver simply pushes a button at the beginning of the route and drives to the end of the route. Table 7-4. Lowest Cost Timlng Plans, City of Tempe Intersection Phasing Weighted Cost PM AM Mld Broadway Soutbern 97 1 444 3 820 3 3 4 Legend: 1 All Leading 2 All Lagging 3 Leading N-SLagging E-W 4 Leading E-WLagging N-S 861 1649 394 Total 2234 2904 - Six travel time runs were performed for each route in each direction for three time periods AMpeak, PM peak, and off-peak. One driver collected all the data in Glendale and a separate driver collected the travel time data for Tempe. The same driver was used for all runs in each city in order to eliminate the variability of different drivers. Once all of the travel time runs were colIected, they were up loaded to ASCII files on the IF3M PC. TIMELAPSE has developed a software program which reads these data files and computes the following information: . - Travel Time Time in Queue (delay time) Stops Average speed Cruise speed Fuel Consumption Carbon Monoxide emissions Hydrocarbons emissions Nitrous oxides emissions An example output for the software program is found in Appendix F. The six runs were averaged for each route to determine the average stops, delay time and travel time for each route. Each ofthe estimates for the routes wasmultipliedby its respective volume to produce a weighted point estimatebasedupon the route volume. A paired Student's t-test was then performed between each sample. The fotlowing comparisons were made: Existing leading minus FORCAST optimized leading Existing leading minus FORCAST optimized lagging Existing leading minus FORCAST optimized combination FORCAST leading minus FORCAST lagging FORCAST leading minus FORCAST combination FORCAST lagging minus FORCAST combination The estimates of stops, delay and travel time produce three distinct variables for each timing plan. In order to comparethe timingplans, it was felt that aggregatingthesethree variables into one variable would be helpful. Because FORCAST develops timing plans which weights the benefit of reduced stops with reduced delay and travel time, a representative cost for each timing plan was developed using the information in A Manual on User Benefir Analysis of Highway and Bus Transit Improvements, (13) published by the American Association of State Highway and Transportation Officials, 1977.These values have been updated to 1988 dollars by using the transportation portion of the Consumer Price Index. The following values were utilized. Parameter Stops Delay Travel Time c!Sl $4 1.oo / 1000 stops $0.616 / Vehicle-hour idling $ 3.35 / Vehicle-hour traveling ANALYSIS Glendale Travel Time Study The study area for the Glendale portion of the project was bounded by Grand Avenue to the south, Cactus Road to the north, 43rd Avenue to the east, and 67th Avenue to the west as shown in Figure 7-1. In this study area, all the major arterial - major arterial intersections were operating in a protected-permissive leading left turn mode in the before condition. Optimization of all ofthe signals within the study area was performed using the FORCAST signal timing program but only the signals along 5 1st Avenue had the phasing patterns changed during the course of the study. The five routes chosen for the Glendale study were the following: 5 1st Avenue 59th Avenue Peoria Avenue Olive Avenue Northern Avenue I ; V + NORTH V ' 0 ~ K S w t 1 € E ~ T D e E ~ ~ P M S M T O B E A R 1 1 8 T E D Flgure 7-1. Lead-Lag Study Area, Clty of Glendale 76 B ~ Table 7-5. Travel Time Results, City 01Glendale I Welgbted Delay (Veblcle Hours) - Route and The 51s Ave. NB 51st Ave. NB 51st Ave. NB 51st Ave. NB 51st Ave. SB 51st Ave. SB 51st Ave. SB 51 st Ave. SB 59th Ave. NB 59th Ave. NB 59th Ave. NB 59th Ave. NB 59th Ave. SB 59th Ave. SB 59th Ave. SB 59th Ave. SB MID OFF PM AM MID OFF PM AM MID OFF PM AM MID OFF PM Northern Ave. EB AM Northern Ave. EB MID Northern Ave. EB OFF Northern Ave. EB PM Northern Ave. WE3 AM Northern Ave. WB MID Northern Ave. WB OFF Northern Ave. WB PM Olive Ave. EB Olive Ave. EB Olive Ave. EB Olive Ave. EB Olive Ave. WB Olive Ave. WB Olive Ave. WB Olive Ave. WB Peoria Ave. EB Peoria Ave. EB Peoria Ave. EB Peoria Ave. EB Peoria Ave. W B Peoria Ave. WB Peoria Ave. WB Peoria Ave. WB Total: AM MID OFF PM AM MID OFF PM AM MID OFF PM AM MID OFF PM FOR had. 45 20 25 n 40 106 81 42 64 24 49 54 25 14 79 I I Welghted Travel TLme FOR Lead. 162 188 417 383 382 210 426 285 I Wel&ted Stops (Thousand Vehicle-Stops) Erlst. F O R MIR FOR LeodLead. Lag. Comb. 2 6 6 5 3 3 8 3 12 8 8 8 15 15 12 6 9 12 9 12 5 5 5 5 8 12 8 8 6 10 12 10 The data from these five travel time runs are shown in Table 7-5. Comparisons were made between the different phasing patterns. The comparison between (1) existing leading (2) FORCAST optimized leading (3) FORCAST optimized lagging and (4) FORCAST optimized combination shown in Appendix D with the results shown in Table 7-6. An equivalent costhour which shows an equivalent motorists' cost based upon stopped time delay, travel time and stops is shown in Figure 7-2. This information is broken into the four travel time periods AM peak, midday peak, PM peak, and an off-peak period. As Table 7-6 suggests, there is a significant difference in travel time and delay between both the FORCAST leading FORCAST lagging and between the FORCAST leading -FORCAST combination plans. There doesn't seem to be any discernible pattern. If the cost parameter alone is looked at, then it appears that the existing timing plan works best for the AM peak, the combination plan works best for the midday and PM peak, and the lagging plan works best for the off-peak. In the AM peak, the lagging plan also works better than the FORCAST leading or the combination. - It appears, at least from this information, that lagging left turns work best in situations such as an off peak period where left turn volumes are relatively light. In this instance, the extra time that is saved from sometimes avoiding the left turn phase can begiven to the through movements resulting in better progression. Table 7-6. Travel Time Study Comparisons, City of Glendale Level of Slgnlfkanee Comparison Existing Leading FORCAST leading - Least Delay Existing ~eading (P) .07 Level of Level of Last Travel SlgnUIcame Significance Time (P) h a s t Stops (PI Existing FORCAST Lading .16 Leading .27 Existing Leading FORCAST lagging Existing Leading .08 FORCAST Lagging .34 Existing Leading .73 Existing Leading FORCAST Combination FORCAST Combination -86 FORCAST Combination .27 FORCAST Combination .26 FORCAST Leading FORCAST lagging FORCAST Lagging .03 FORCAST Lagging .01 FORCAST Leading .43 FORCAST Leading FORCAST Combination FORCAST Combination .02 FORCAST Combination .O1 FORCAST Combination -87 FORCAST Lagging FORCAST Combination FORCAST Combination .47 FORCAST Combination .58 FORCAST Combination .29 f 1-10 V) F ' n 0 1') 'f >J BROADWAY RD. z! W E D A DR. \-f G F 8 Es? d (\ f\ t NORTH d b kh ! 9 SOUTHERN AVE. LEGEND 0 S I O W WHERE PROORESSIONTO BE STUDIED S I Q W WHERE PHASINO TO BE ADJUSTED Figure 7-3. Lead-Lag Study Area, City of Tempe An equivalent cost~hourwhich shows an equivalent motorists' cost based upon stopped time delay, travel time and stops is shown in Figure 7-4. This information is broken into the three travel time periods AM peak, midday peak and PM peak. The results shown in Table 7-8 indicate that there is only one significant result in the Tempe travel time data. FORCAST leading had significantly fewer stops than FORCAST lagging. In viewing Figure 7-4, it is noted that lagging has a higher cost than FORCAST leading or combination in the midday and PM peak, but combination has a higher cost in the AM peak. The cost difference between leading andlagging is least in the AM peakand greatest in thePM peak. In Tempe, at the two intersections where lagging left turns were implemented, there is a very great directional split between left turns at these two intersections in the PM peak. By forcing these two movements together, it has greatly increased the motorists' cost in the PM peak. Table 7-7. Travel T h e Results,City of Tempe Welgbted Delay (Vehicle-Hours) IEX&. Weigbted Travel Ttme (Vehlde-Hours) FOR FOR FOR I Exist. FOR. Lead had. Lag. Comb. Lead. L e d AM 17 4 3 15 117 101 MID 28 4 8 38 164 141 PM 25 40 44 42 % 112 AM 36 22 14 5 79 48 MID 49 68 29 14 2M 191 Rvl 44 11 87 48 202 157 AM 12 16 12 9 55 54 MID 26 NIA* 18 6 209 NIA* PM 54 63 21 5 241 237 AM 15 42 19 58 149 198 MID I I NIA* 47 33 187 N/A+ PM 23 17 27 13 103 % AM 6 15 14 23 104 110 MID 49 106 167 49 483 540 PM 22 117 68 120 nl 35s 30 AM 202 208 34 52 38 MID 117 72 52 57 PM 100 43 137 79 Aiest Rd NB AM 31 11 14 9 Priest Rd NB MID 17 6 3 31 Priest Rd NB PM 8 21 16 25 Priest Rd.SB AM 1 4 22 21 Priest Rd.SB MID 1 1 1 1 120 130 Priest Rd.SB PM 19 11 6 16 99 113 Total (Mmus SauhemMid) 652 741 865 706 3894 3859 * -Not able to collect data due to consbuctionalong S o d em Ave. Route and Tlme 48th St.. NB 48th St.. NB 48th St.. NB 48th St.. SB 48th St.. SB 48th St.. SB Southern Ave. EB Southern Ave. EB Southern Ave. EB Southern Ave. WB Soutlsern Ave. WB ScutkmAve.WB Broadway Rd. EB Broadway Rd. EB Broadway Rd.EB Broadway Rd WB Broadway Rd W B BroadwayRdWB Weighted Stops (Il~ousandVehlebSbps) FOR FOR 1 Exist. FOR M)R. FOR Lag. Comb Lead. Lead. Lag. Comb. W 117 3 3 3 7 142 176 5 5 5 5 3 3 3 3 117 115 66 58 1 3 3 4 2.50 203 6 0 1 2 6 14 5 14 5 239 184 50 47 3 2 2 2 197 168 9 NJ'A* 4 4 199 167 1 3 9 4 4 181 151 3 10 10 3 0 NJ'A* 9 5 229 207 104 89 6 2 2 4 2 2 2 7 109 119 21 10 21 10 598 466 6 17 17 17 310 359 205 204 1 2 4 8 8 21 10 10 10 458 467 12 4 12 8 301 236 83 85 6 3 3 3 150 170 6 6 6 1 1 82 88 2 2 2 2 1 0 1 1 53 56 12A 116 0 0 0 0 99 107 3 3 3 3 4016 3777 151 104 146 126 Mesa Travel Time Study The City of Mesa changed the phasing at Southern and Stewart from Leading east-west to leading east and lagging west. This study area is shown in Figure 7-5. Lee Engineering collected travel time data along Southern Avenue in the AM, Midday and PM peak time periods to determine the effect of this changeover. The results of this change are shown on Table 7-9. While not significant,these numbers do show a substantial reduction in delay, stops, and travel time due to the changing fiom an all leading phasing pattern to a combination leading-lagging phasing pattern. Scottsdale Travel Time Study In the Spring of 1988, Lee Engineering performed an optimization of the signals within the City of Scottsdale.Optimization was performed using the city's FORCAST computerprogram. At that time, several travel timeruns wereperformedusingtheTIMELAPSETravelog data collectortodetermine stops, delay, and travel time. These travel time runs were performed by membersof the City Council, Transportation Commission and city staff. Flgure 7-5. Lead-Lag Study Area, City of Mesa 83 Table 7-9. Travel T h e Studles Liending Mnus ComMnatIoa,City of Mesa Welgbted Delay 1 Weighted Travel Ttme I Wetghted Stops (Vehidehours) (Vehkle-hours) (RIOIISIIIWI Vehicle Stops) ROUE mwlLeading La@g Difference IarUng Lagging Difference Leadlog Lag& Merence Southern Ave. EB AM 5 0 5 n 22 6 0.8 0.0 0.8 h d k m Ave. EB MID 52 1 51 129 71 58 5.2 0.0 5.2 Southem Ave. EB I34 33 132 90 42 41 8 5.9 2.9 2.9 Soulhem Ave. EB AM 10 3 8 68 60 8 21 2.1 0.0 Southern Ave. EB MID 5 5 0 76 74 2 24 2.4 0.0 Southern Avee El3 PM 0 5 -5 74 66 8 0.0 2.5 -2.5 Total 113 506 21 Samplesize 6 Mean Difference 15.197 Std Deviation 21371 Test Stat. 1.702 Significant Level of Significance (p) N .15 382 Sample Size 6 Mean Difference 24l.579 Std Deviaticm 23.436 Test Stat. 2.151 Significant N Level of Significance (p) .W 16.4 10.0 Samplesize 6 MeanDiffemce 1.069 Std Deviation 2.675 Test Stat. 0.979 Significant N Level of Significance@) .37 Once the city converted all their signals to lagging phasing, city staffperformed another retiming of their signal system in the Spring of 1990. Assuming the volumes along the streets stayed relatively constant over this two year period, another set of travel time runs were performed by city staff. The results of the leading and lagging phasing is shown in Table D-4 in Appendix D. In some cases, the travel time studies for both the leading and lagging left turns were conducted at different times ofthe day, specifically the PM peak runs were made between 4 PM and 6 PM in 1988 and generally between 4 PM and 5 PM in 1990. The results shown in Figure F-1 (Appendix F-4) however, have been analyzed for both leading and lagging in those time periods which were common to both plans. This information must be used carehlly for several reasons: - 2 year time period between studies. City conversion of numerous intersections from protected to permissive -protected between the two studies. Reduction of extension time throughout system. Refinement of leading timing patterns after the before studies. More recent traffic counts in timing lagging condition. - Different drivers in before and after studies. DISCUSSION OF RESULTS It is difficult to determine if either leading or lagging left turns are a better operation for a given situation. While not statistically significant, lagging left turns appeared to operate better for three time periods in Glendale (based on FORCAST plans). The combination timing plan worked better than leading or lagging in Glendale for only the Midday and PM peak. In Tempe, the combination was never the lowest-cost plan. This was surprising, for it was felt that by having theopportunity for leading or lagging at a particular intersection would help improve progression. It should be stressed again that the FORCAST timing plan must overcome two obstacles in order to choose lagging left turns for intersection phasing. The fact that it does not recognize left turns made on the permissive period results in it not determining the true best combination plan. The combination timing plan fared best in Mesa where stops, delay, and travel time were all three reduced substantially. This type of combination phasing is different than those tested in either Glendale or Tempe. The Mesa combination plan was leading eastbound and lagging westbound. In Tempe the phasing tested was leading north-south and lagging east-west. It would appear that a substantial reduction in motorist cost is not very apparent with this type of phasing, but is very apparent with the Mesa phasing. It is important to realize that to implement the Mesa phasing, it is necessary to have either protected only operation, or programmed visibility traffic signal heads as is currently being used in Texas. In conclusion the following points should be mentioned. - One of the greatest benefits of lagging left turns is in decreasing the need for a protected left turn phase. This increases the opportunity for larger progression bands through the intersection. In order for a timing program to implement the best phasing, it is necessary for that program to evaluate the left turn movement in conjunction with gaps in the opposing traffic stream. Since FORCAST does not do this, it is not a good program for optimizing the combination phasing. A program which evaluates the gaps in the opposing traffic stream would be a better program. It is likely that FORCAST may pick the wrong timing plans for an area when considering the combination of leading and legging left turns. Combination timing seems to work best when leading and lagging are implemented for opposing directions, i.e. leading eastbound and lagging westbound. There does not appear to be much benefit when leading and lagging are implemented for perpendicular directions, i.e., lagging north-south and leading east-west. It is possible however, that the latter condition could result in improvement if the optimization sofhvare recognized the left turns made in a permissive manner. One benefit for lagging operation may be in locations where left turns are actuated by the first car in the left turn bay. Although this same benefit may be realized in some situations with third car actuation, there are those where sufficient gaps occur in the opposing traffic stream to permit more than two vehicles turning left on the permissive phase. In those cases, lagging would skip the protected phase and 3rd car actuated leading wouldnot. Each time this occurs in a coordinated system, the time saved from omitting the protected phase goes to one of the through green phases thereby increasing the opportunity for improved progression. But, the time saved by omitting the protected phasemay be offset by increased delay to through traffic during the simultaneous protected phase when there are few, if any, left turn vehicles on the opposing approach. - In locations like Tempe, where there is a high directionality with opposing left turn volumes, there is substantial delay associated with lagging operation due to the loss of phase overlap. At locations where lagging is implemented, programmed visibility signal heads might permit phase overlap with permissive-protected operation. PART IV PUBLIC PERCEPTION STUDIES PART N documents public opinion surveys which were conducted to determine if motorists have a preference for a leading or a lagging operation. CHAPTER 8examines theresultsofpublic opinion surveys conducted in the Cities of Glendale and Tempe. Both surveyswere conducted by taking arandom sample of motorists at intersections which had been converted, as part of this research, from leading to lagging left turn operations. The survey instrument chosen for this study was a mail-in questionnaire which contained four questions. The first two questionsprovided information concerningdriver awareness of various left turn operations. The second two questions provided information concerning driver preference for leading or lagging left turn operations. There were 802 responses received from the Glendalequestionnairemailing and 633 From the Tempe mailing. Approximately 49% of the Glendale motorists which responded to the survey indicated the signals are better with leading left arrows, while approximately 42% preferred the lagging operation. The results of the Tempe questionnaire indicated a nearly two-toone preference for the lagging left arrow operation. One possiblereason Tempe motorists may prefer the lagging operation is the close proximity of Tempe and Scottsdale, where lagging left turns are utilized. CHAPTER 8 PUBLIC AWARENESS AND PERCEPTION INTRODUCTION One of the important aspects of this research effort, especially fiom the perspective of an elected official, deals with the possible preference of either leading or lagging operation. Information concerningmotorist preferenceof leading or lagging left turn operation was obtained through public opinion surveys conducted in the Cities of Glendale and Tempe. Both surveys were conducted by taking a random sample of motorists at intersections which had been converted, as part of this research, from leading to lagging left turn operations. DATA COLLECTION The survey instrumentchosen for this study was amail-in qucstionnaire. A questionnaire containing four multiple choice questions and a space for additional comments wasprepared. The questionnaire was designed to be contained on a nine inch by four inch postage paid post card. A copy of the questionnaireused for motorists in the City ofTempe is shown in Figure 8-1. A similar questionnaire was prepared for the City of Glendale survey. The only difference between the two questionnaires were those changes which were necessary to reflect a change in study area location. A cover letter, which accompaniedthe questionnaire,briefly explained the nature of the study and the various types of left turn operation: leading, lagging, and combination leading and lagging. The first two questions provided information concerning driver familiaritywith the study area andleft turn operations. The second two questions provided information concerning driver perception of and preference for the various types of left turn operations examined in this research effort. Two lists, each with 2400 recorded license plates, were generated. The first list was generated from the four intersections converted in the City of Glendale: - 5 1st Ave./Glendale, 5 1st Ave./Northem, 5 1st Ave./Olive, and 5 l st Ave./Peoria. The number of plates recorded was evenly distributed among the four intersections and among the four approaches at each intersection. Ten percent of the plates recorded were from left turning vehicles and 90% from throughor right turningvehicles. The samplewas alsodistributed throughout the day with 20% of the plates being recorded between 7:00 and 9:00 AM, 60% between 9:00 AM and 4:00 PM, and 20% between 4:00 and 6:00 PM. The second list was generated from the two intersections converted in the City of Tempe: - 48th St./Southem, and 48th St./Broadway. QUESTIONNAIRE Note: This questionnaire is a part of a research project designed to study traffic signal operation. Your response could help the City of Tempe and the Arizona Department of Transportation improve traffic flow. !Recently the City of Tempe changed the left turn arrows at 48th Street and Broadway and 48th Street and Southern, There have been three types of left turn arrow operation: a. All leading left turn arrows (May, 1990 and before) b. All lagging left turn arrows (early June, 1990) c. Some leading and some lagging left turn arrows (late June, 1990) 1. How many times per week do you drive in this area? 4. With respect to left turns, are the signals: - 0-5 - better with leading left turn arrows - 6-10 - better with lagging left turn arrows - 11-15 - no diffcrcnce - 16+ 2. Had you noticed this change in left turn amws? - yes - no 5. If one (leading or lagging) is better, why? 3. Please think about your past and present travel in this area, When driving in this area, do you encounter: - more green lights with all leading left turn arrows - morc green lights with all lagging left turn arrows - morc green lights with some leading and some lagging - no difference Figure 8-1. Tempe Questlonnahe The Tempe sample was distributed in the same manner as the Glendale sample with the only exception being that the number of plates recorded was evenly distributed between two rather than four intersections. The license plates were recorded on a typical week day during the time the lagging operation was being tested at these intersections. Each list was searched to eliminateduplicate entries. The Arizona Department of Transportation Motor Vehicle Division then generated two mailing lists of registered vehicle owners for the two samples. Two lists containing approximately 2250 names each were generated fiom the original two lists of 2400 license plates. The questionnaireswere mailed during the time the combination leading and lagging operation was being tested. ANALYSIS There were 802 responses received from the Glendalequestionnairemailing. This represents a return rate of approximately36%. Approximately 100 questionnaireswere not deliverable due to incorrect addresses, change of addresses, and similar situations. The results for the Glendale questionnaireare presented in Table 8-1. A 95% confidence interval was calculated, treating each possible response as a binomial parameter. The data indicates a relatively even distribution in the fiequency respondents drive in the study area. The data also indicates a high level ofrecognition ofthe changes which had occurred to the left arrow operations over the course ofthe study, with over 85% noticing the change in left turn arrows. Approximately 38% of the Glendale motorists responding indicated they encountered more green lights with the all leading left turn operation. This compares to approximately 16% for the all lagging operation and 24% for the combination leading and lagging operation. Approximately 17% indicated there was no difference in the three types of operations studied. With respect to left turns, approximately 49% indicated the signals are better with leading left arrows, while approximately 42% preferred the lagging operation. There were 633 responses received from the Tempe questionnairemailing. This represents a return rate of approximately 28%. Approximately 90 questionnaireswere not deliverable. The results for the Tempe questionnaireare presentedin Table 8-2. A 95% confidence interval was also calculated. The Tempe data also indicates a relatively even distribution in the fiequency respondents drive in the area and a high level of recognition of the changes which had taken place. Approximately 30% of the Tempe motorists responding indicated they encountered more green lights with the all leading operation. This compares to approximately 21% for the all lagging operation and 27% for the combination leading and lagging operation. Again, approximately 17% indicated no difference. With respect to left turns, approximately 30% expressed a preference for the leading operation and approximately 61% preferred the lagging operation. The two surveys taken together represents a composite of 1435 motorists responses. A composite of the Glendale and Tempe questionnairesis presented in Table 8-3. Nearly twice as many motorists believe they encountered more green lights with the all leading operation as compared to the all lagging operation (34.4% versus 18.3%). However, 25.3% reported they encountered the best progression with the combination leading and lagging operation. With respect to left turns, the composite results indicate a slight motorist preference for the lagging left arrow operation. Approximately 50% expressed a preference for lagging and approximately 4 1% prefer the leading operation. Approximately 9% expressed no preference. Table 8-1. Publlc Awareness and Perception, Clty of Glendale Qucstlon How many times per week do you drive in this area? Count Percent 0-5 610 11-15 16+ No response Had you noticed this change in left turn arrows? Yes No No response Please think about your past and present travel in this area. When driving in this area, do you encounter: More green lights with all leading left turn arrows More green lights with all lagging lef? turn arrows More green lights with scimc leading and some lagging No difference No response With respect to left turns, are the signals: Better with leading left turn arrows Bettcr with lagging left turn a m w s No difference No response Questionnaires Mailed: Number of Responses: 2250 802 36% 95% Conf. Interval Table 8-2. Publlc Awareness and Perception, City of Tempe 95 % Conf. Question How many times per week do you drive in this area? Count Percent 0-5 6-10 11-15 16+ No response Had you noticed this change in left turn arrows? Yes No No response Please think about your past and present travel in this area. When driving in this area, do you encounter: Mom green lights with all leading left turn arrows More green lights with all lagging left turn arrows More green lights with some leading and some lagging No difference No response With respect to left turns, are the signals: Leading Better with leading left turn arrows Better with lagging left turn arrows No difference No response Questionnaires Mailed: Number of Responses: 2250 633 28% Interval Table 8-3. Public Awareness and Perception, Composite Question How many times per week do you drive in this area? Count Percent 4500 1435 32% 0-5 6-10 11-15 16+ No response Had you noticed this change in left turn arrows? Yes No No response Please think about your past and present travel in this area. When driving in this area, do you encounter: Mom green lights with all leading left turn arrows More green lights with all lagging left turn arrows More green lights with some leading and some lagging No difference No response With respect to left turns, are the signals: Leading Better with leading left turn arrows Better with lagging left turn arrows No difference No response Questionnaires Mailed: Number of Responses: 95 % Conf. Interval DISCUSSION OF RESULTS The results of the Glendate and Tempe surveys indicate that the largest proportion of motorists feel they encountered more green lights with the all leading left arrow operation. The composite results indicate 34.4% of the motorists responding believe they encounter more green lights with the all leadingoperation. This would tend toindicate that a greaternumberofmotoristsperceive traffic flow is better with the all leading left arrow operation. The two surveys are also consistent with regard to the perception of the relative merits of a combination leading and lagging left arrow operation. Both surveys indicate a greater number of motorists encountered more green lights with the combination phasing than with the all lagging operation. The two surveys are not consistent with regard to motorist preference of leading or lagging left arrows. The Glendale survey indicates that, with respect to left turns, the leading left arrow sequence is favored. However, Tempe motorists expressed a preference for the lagging operation. One possible reason Tempe motorists may prefer the lagging left arrow operation is the proximity ofthe City of Scottsdale and the City of Tempe. The City of Scottsdale implemented lagging left arrows on a city-widebasis in early 1989. City ofTempemotorists aremuch more likely to drive the streets of Scottsdale than are City of Glendalemotorists. Therefore, City ofTempemotorists were in a better position to adjust to the conversion to a lagging operation. A number of City of Tempe motorists included references to the City of Scottsdale in their comments. The comments were generally instructive. A common theme for those who preferred the leading left arrow operation was consistency and driver expectancy. Lane blockage due to queued left turn vehicles was also a comment commonly made by those who expressed a preference for the leading operation. Safety was a reason given by both those who preferred leading and those who preferred the lagging operation. Some people simply commented on the need for left turn arrow phasing in general. Finally, it should be noted that this sampling technique does not guarantee an unbiased sample of motorists in the Cities of Glendale and Tempe. The fact that the survey required the respondents to take the initiative to fill out and mail the questionnaire introduces some bias to the sample. But it does provide some indication of the sentiments of those people who feel as though the issue is important enough forthem totake the time to fill out thequestionnaire and return it. Thereisno reason to believe this group of people would be more inclined than the public at large to favor a leading or lagging left arrow operation. Therefore, it is felt that this sample does fairly represent the sentiments of motorists in the Cities of Glendale and Tempe. PART V CONCLUSIONS PART V presents the results and conctusions of the research project. CHAPTER 9 presents the results of the four research questions identified at the initiation of the project. It was found that intersection delay is significantlygreater with lagging left turn operation. No significant change in total delay was found with third car actuation of leading protected left turn operation. No significant difference in progression was found between leading, lagging and mixed operation. No significant difference was found in accident experiencebetween leading and lagging operation. There was a mixed response fi-om the motorist preference survey. Glendale drivers felt left turns were better with leading while Tempe drivers felt it was better with lagging. CHAPTER 10 presents some observations made by the research team during the conduct of the study. CHAPTER 11 identifies future work which would be of value in this research area. CHAPTER 9 STUDY RESULTS In response to the five research questions stated in Chapter 1, the results of the study are as follows: 1. Is there a dzrerence in intersection delay at isolated intersections between leading and lagging operation? The results ofthis study indicate significantly greaterdelay per approach vehicle occurs with lagging operation than with leading operation for the intersections and time periods tested. The Phoenix area studies reflected a 42% increase in delay conversion per approach vehicle with conversion fi-om leading to lagging operation. The same conversion in Pima County resulted in a 30% increase in delay per approach vehicle. It is important to note that the time periods tested were generally PM peak hour conditions. These would not be as likely to have sufficiently low left turn and through volumes to eliminate many protected left turn phases in their lagging condition. It is conceivable that in off peak conditions more of the left turns could be made in a permissive manner therefore skipping the protected left turn phase. Eliminating the protected phases would likely reduce intersection delay. Intersection delay was also collected for test intersections with both first car and third car actuation. Although there was no significant difference between the two, this test also was only done in the PM peak hour condition. The probable benefit of third car actuation on intersection delay is most likely in off peak conditions. 2 . Is there a diflerence in signal progression among leading only, lagging only and mixed operation ? There were no statistically significant differences in stops, delay or travel time with the different operating conditions. Additionally, the large number of signal timing optimization runs required to evaluate all combinations of leading and lagging operation makes for a cumbersome, time consuming process. The requirement that the Glendale and Tempe "mixed" operation was limited to either both leading or both lagging on the same street in order to avoid the "trap"restrictedpotentia1progression benefit. An additional limitation was that only four of eight multi-phase Glendale intersections and two of fourmulti-phaseTempe intersections were considered for change to lagging. The most promise for benefit from lagging or mixed operation was found in the Mesa study where leading left turn operation wasutilized for eastbound traffic and lagging for westbound trafic in the after condition. This was the operation which provided the best east-west progression. This mixed operation was possible without the trap condition because ofthe use of protected only left turns. 3. Is there a need to have conshtent left turn phasing (leading or lagging) within any given city, urbanized area and throughout the State of Arizona? No evidence was found in this study supporting the need to have the same phasing consistency. Although many of the cities prefer consistency within theirjurisdiction, a straw poll of the representatives on the Advisory Committee found unanimous agreement that the state should not pass legislation mandating either operation everywhere. 4 . Is there a diflerence in accident experience between leading and lagging operation? In all three accident studies - Tucson, Pima County and Scottsdale, there was no significant difference in left turn accident history between leading and lagging operation. 5 . Is there a motorist preference between leading and lagging operation? Table 9- 1 presents a summary of the information obtained in the motorists' survey. Lagging left turns seem to be more favorably received in Tempe than in Glendale. This could possibly be due to the close proximity of Tempe to Scottsdale, where lagging left turns are utilized. Table 9-1. Public Perception Results, Phoenix Area Question More Green Lights With: Leading Lagging Combination No Difference I No Response Left Turns Better With: Leading Lagging No Difference I No Response City Glendale 3 8% Tempe 30% 16% 24% 22% 21% 27% 22% 49% 30% 42% 61% 9% 10% CHAPTER 10 THEORETICAL ANALYSIS OF LEADING AND LAGGING LEFI' TURNS INTRODUCTION As a result of this study, an increased knowledge of the subtle aspects of leading verses lagging left turns was gained both by the research team and the Advisory Committee. Many of the observations made werenot supported by statistical analysis due to limited sample sizesor other factors; however, it was believed to be important to try to identify some of the issues related to the question of leading and lagging left turns. This chapter is intended to serve that purpose. The value of the field data collection and the analyses ofthat data can be emphasized by the fact that it provided insight to the understanding of the many variables which influence left turn operations. Within the scope of this study and the conditions at the study sites, it was not possible to collect data for all possible combinations of the pertinent variables. The results of the field studies together with a somewhat theoretical analysis yield a comprehensive assessment of leading and lagging left turn operations. This section identifies a number of variables that have an impact on the effectiveness of left turn alternatives. The variables that should be evaluated when considering leading or lagging left turn operations are generally associated with the signal system, traffic characteristics, as well as the driver. The effectiveness of the application of leading or lagging left turns then becomes a function of the conditions at a specific intersection or location. Table 10-1 presents a general summary of the variables that should be considered in comparing leading and lagging left turn operations. The variables listed irr Table 10-1 are related to operational aspects of the problem area. It must be recognized that safety impact is the result of the decisions Table 10-1. Decision Variables for Comparing Leading and Lagging Left Turn Operations Signal Control Application of actuated control Fixed versus variable cycle length Left-turn operations Use of phase overlap Network Considerations Offset requirements Allocation of unused green time Traff~cCharacteristics Approach volumes Directional distribution of opposing flows Acceptable gaps in opposing flows Peak versus off peak volume variations Driver Perception Need for uniformity Driver compliance and acceptance 101 related to many of the variables. For this reason, safety was not included in the list. The discussion that follows provides a more detailed explanation of the variables. SIGNAL CONTROL The variables associated with signal control focus on the traffic signal operation at a spccific intersection. Generally, these variables are related to use of green time or have an impact on the application and allocation of green time. Included in this category are: application of actuated control, fixed versus variable cycle length, left turn operation, use of phase overlaps, and lost time. This discussion assumes that exclusive left turn signals are operated on an actuated basis; thus the issue is whether the through and right turn movements are also operated on actuated control. Ifthe intersection operates with full actuated control, there is the opportunity for the skipping of phases and the adjustment of green time in relation to the trafiicdemand. These factors will ultimately affect the effectiveness of the left turn operations. Obviously, full actuated control results in a variable cycle length unless a background cycle is utilized; thus the advantage is that the signal will operate in response to the demand with the potential of reducing wasted green time. As a result, inefficiencies in left turn operations can potentially be offset by efficiencies from the actuated control. With the fixed cycle length, the need and duration of the lei3 turn arrow can have a different result. The left turn operation variable reflects the specific operation of left turn movements. In essence, the issue is whether left turns occur a) only in a protected phase or b) on a permitted basis in conjunction with a protected phase. One of the noted advantages of the lagging left turn is that the left turn demand may be satisfied by pennittedmovements; thus the need or duration ofthe protected movement is reduced. This advantage would not be available for the protected only operation. One of the major advantages of the leading left turn operation is associated with the use of phase overlaps. In essence, once the minor left turn movement is satisfied, the through trafic on the approach of the heavier left turn movement is released. Given differences in opposing directional flows, the phase overlap can increase the efficiency of the signal operation. There is concern about the ''left turn trap" that is created with the application ofphase overlaps with lagging left turns. For this reason, jurisdictions frequently operate lagging left turns simuItaneousIy without a phase overlap. Some work is currently being done with signal displays which will potentially eliminate the left turn trap problem. Nevertheless, the use ofphase overlaps is a major consideration in terms of signal operation. One of the elements in the analysis of intersection capacity is lost time. Basically, this is the time during a cycle that is lost due to start up delay or clearance time. Generally, the total lost time in a cycle is a function of the signal phasing. With the leading left turns and phase overlaps, there is less lost time during a cycle than with simultaneous lagging left turns. With the lagging left turnoperation, there are more situations that require the stopping and restart of traffic streams. NETWORK CONSIDERATIONS The variables associated with network considerations reflect signal coordination concerns. In this group, the variables are: offset requirements, and the allocation of unused green time. In thecoordination of signal networks, oneoftheconcernsis the offset, or the time difference between the beginning of the green phase at successive intersections. Because of this concern, the issue in the application of the exclusive phase is related to the release of the through traffic at a given intersection. With the leading left turn, the through movement will be released early if the left turn phase is not fi~llyutilized.This means that the through traffic will arriveearly at thenext intersection; thus the progression is affected. It has been argued that with the lagging left turn, the duration ofthe left turn phase does not influence the start of the green phase. In theory, this should resolve the problem of controlling the offset; thus the lagging left turn offers an advantage. For some network configurations, it ispossiblethat amixedoperation ofleading and lagginglefi turns may provide improved network coordination. The mixed operation could use leading left turns in one direction and lagging left turns in the other or simply mix the use of leading and lagging left turns depending on the offset requirement. A factor that complicates the signal coordination problem is the allocation of unused left turn green time. If the left turn phase is not fully utilized, then the unused time for that phase must be allocated to someotherphase ifa fixedcyclelength ismaintained. With the leading leftturn, the through traffic is generally released early; thus the offset with the next intersection is affected. The allocation of unused green time with the lagging left turn will vary depending on the operation of the signal by ajurisdiction. Somejurisdictions will add the unused left turn time to the beginning of the next phase. In essence, unused left turn time on the main street will be added to the beginning of the side street through movement. Also, the unused side street time will be added to the beginning of the main street through green phase. Otherjurisdictions may accumulate all unused time and add it to the beginning ofthe main street through movement. In either case, it results in the possible early release of the through movement which disrupts the planned offset. Basically, both types of left turn operations have a potential adverse impact with respect to the planned offset. For agiven location, it isnecessaryto evaluate the probability ofhaving unused green time and the effect on network operations. TRAFFIC CHARACTERISTICS Whether the intersection signals utilize left turn actuations only or include actuated control for some or all through movements, there are a number of traffic variables that can influence the left turn operations. These variables include: - approachvolumes, directional distribution of opposing flows, acceptable gaps in opposing flows, and peak versus off peak volume variations. The approach volumes reflect the general magnitude of the traffic movements as well as the percentage of left turns at a given intersection approach. Basically, the magnitude of the approach volumes will have an impact on the need for the protected left turn movement. As the volumes increase at an intersection, it is less likely that the left turn demand can be accommodated by the permitted operation. Forthis reason, the probability that the laggingleftturn phase will not beneeded decreases. In addition, increases in volumes will potentially increase the probability that the left turn phase will be fully utilized. The directional distribution of the opposing flows affects the possibility of phase overlap operation as well as the advantage of the phaseoverlap accommodating the through movements. With balanced left turn and through movement approach volumes, the advantage of the phase overlap with the leading left turn operation is eliminated. With the elimination of phase overlap operation, the eficiency for the leading operation is similar to the lagging operation. The number of acceptable gaps in traffic opposing a permitted left turn movement will influence the permitted left turn capacity. For both a protected/permissive and a permissive/protected operation, the larger the permitted left turn capacity, the less time which must be dedicated to the protected left turn phase. A large left turn volume coupled with a large permitted left turn capacity may overcome some of the disadvantages associated with a lagging operation under the conditions of high directionality. With a lagging operation, the permitted left turn capacity is first filly utilized then the protected phase is used to satisfy the residual left turns. The opposite is true for a leading operation, where the capacity of the protected phase is first exhausted. An accurate evaluation of the permitted left turn capacity is much more critical to the efficient timing of the lagging left arrow operation. I Typically, traffic engineers are concerned with the peak conditions; however the off peak traffic volumes may yield a different set of results in terms of demands on signal operations. For example, the left turn demand in the offpeak periods could possibly be satisfied without the need for protected turn phases. With this condition, there would be a potential advantage to the lagging left turn operation due to the fact that the left turn phase would only be used if needed. Similarly, third car actuation could result in more delay reduction in the off peak periods, than in the peakperiods. Thus, it is possible that the operation of the intersection could be improved. One of the factors to be considered, therefore, is the difference in peak and off peak traffic conditions. Although traffic engineers are typically concerned with the peak conditions, they should be just as concerned with what happens during the other 22 hours of the day. DRIVER PERCEPTION An important element in the operation of any signal system is the perception ofthe driver. Variables related to driver perception are: the need for uniformity, and driver compliance and acceptance. One ofthe basic considerations for the application oftraffic control devices is uniformity. Certainly, there arc arguments that leading and lagging left turn operations should not be mixed because of the lack of uniformity. On the other hand, some mixed operations even in Arizona apparently go unnoticed by the motoring public. If the driving population perceives the need for uniform left turn operations, then the effectiveness of mixing the operations will potentially be affected. The true question to be resolved is the importance of uniformity for a particular area. While not totally unrelated to the uniformity issue, another driver variable is compliance and acceptance. For example, if the drivers comply with the signal display, uniformity may not be as great an issue. Anticipation on the part of the driver, as a result of uniformity, can be dangerous as well. Certainly, it is likely that differing driver populations may yield differing responses in terms of compliance and acceptance. CHAPTER 11 RECOMMENDATIONS FOR FUTURE WORK Through the conduct of this study, several questions have arisen which are outside the scope of this project. These are identified here as possible items for future research. Egectiveness of laggingleft turns and third-car actuation in off-peak conditions. The studies of laggingandthird car operation conductedin this project were primarily duringthePMpeak periods. A major potential benefit of both lagging left turns and third car actuation lies in the possibility of eliminatingprotectedleft turn phases. In alaggingoperation, if theleft turns can be accommodated on the permissive green thereby eliminating the protected phase, a reduction in intersection delay should result. Similarly, if fewer than three cars arrive prior to the beginning of a protected left turn phase, delay should be reduced with third car actuation. For this reason evaluation of off-peak delay comparing leading with lagging left turns and third car actuation with first car actuation would be valuable. Efectiveness ofleading left turns inoriedirection and lagging left turns in opposingdirection. As tested at one Mesa intersection, there is the potential for improvement to progression at locations where the platoons arrive at differenttimes tohave the left turns leadin the direction the platoon first amves and lag in the opposing direction. The sample size in this research project was inadequateto make a conclusive statement,however, the Mesa results were very promising. Feasibility of overcoming "trap" but allowing combination ofpermissive leading or lagging with phase overlap. Current experimentation in Texas is evaluating the use of a 5 section programmed visibilityhead (Dallas signal face display) for left turn drivers. This head would continue to display a circular green indication to left turn drivers whose concurrent through movement is terminated but whose opposing through movement continues. Although this method has good results (14) the signal display violates section 4B-12 (3.2.)(15, p.4B-12), of the MUTCD which states: 2. During the permitted lejl turn movement, all signal indications on the approach shall display all a.c. CIRCULAR GREEN indication. A formal request has been made by the Texas State Department of Highways and Public Transportation to revise the MUTCD to allow the Dallas signal face display. Appendix G shows the typical phasing and special (Dallas) phasing being tested. Signaloptimizationsofharewhich evaluates options of combinations ofleading and lagging left turns and which predicts protected lagging phase duration based on permissive left turn phase capacity. Although PASSER I1 will evaluate the combination of leading and lagging left turns on arterial streets, the research team is unaware of software which will accomplish this for a grid. In this research project the FORCAST optimization software developed by Computran Systems, Inc. was used with all possible combinationsof leading and lagging left turns within the constraints of either both leading or both lagging on the two approaches of the same street. In the lagging situation however, FORCAST does not consider the permissive left turns which can be made during the through movements, therefore it overestimates the green time required for the protected lagging phase. This results in a higher cost prediction than actually would result. It is impossible therefore to determine the true lowest cost phase combination. A software program should be developed which: Evaluates all possible combinations of leading and lagging left turns for a grid, and, Includes a gap acceptance algorithm which will predict the number of left turns which can be made during apermissiveperiod and the resultingrequiredprotectedlagging phase durations. Evaluation of "trade o f ' of un*rrnity verses the benefit of varying left turns between leading and lagging under different conditions. There is a perceived value of having a uniform left turn treatment. It is unlikely that this could or should be made completely uniform throughout the country, state or even an individual city. On the other hand, it is logical that the greatest efficiency of system operation (i.e. least delay) would come from the ability of varying between leading and lagging left turns not only from intersection to intersection, but also from approach to approach at a given intersection. The optimum performance likely would result from analyzing all combinations ofleading and lagging left turns at each approach for each time period evaluated. In this case, an approach might be leading in one period of the day and lagging in another. Figure 1 1- 1 graphically portrays this trade off. This analysis would necessitate evaluating a very large number of combinations of phase options. It may be desirable to develop software which would perform a two step optimization process. - Coarse level analysis to determine best combination of leading and lagging left turn phase, and A more detailed analysis similar to TRANSYT 7F Maximum Unlformlty Either all leading or all lagging left turns throughout the state. Consistent treatment within crty - all leading or all lagglng. Both leading and lagging at the same Intersection depending upon progression. Van/ing between leadin and la glng at the same approach of intersectionby pattern I m e of ky variation). Greatest Efficiency Figure 11-1. Rlerarchy of k f t Turn Unlformlty 108 Additional work would include an evaluation of driver expectancy of left turns vary between leading and lagging by time of day. As stated previously, it is expected that a reduction in delay should result from considering all possible combinations of phasing for each time period being optimized. A study should be made of possible driver confusion resulting from such a treatment. REFERENCES 1. Traflc Engineering Handbook, Institute of Traffic Engineers, Washington, D.C., 1965. 2. City of Tucson Informational Memorandum From Joel D. Valdez, City Mhnager to Mayor and Council, May 10, 1985. 3. Nassi, Richard B., City of Tucson Traffic Engineer Telephone Conversation, December 19, 1988. 4. Lag Left- Turn Phasing, Third CarActuated, Lead Left-Turn Phasing, Entranco-Mann-Johnson, Inc., Phoenix, Arizona, April, 1988. 5. Basha, Paul B., City of Scottsdale Traffic Engineering Manager, Letter to Jim Lee, September 2 1, 1988. 6. Memorandum from Paul Basha, Traffic Engineering Manager to Thomas J. Wilson, Acting City Manager, August 25, 1988. 7. Memorandum from Alan Sanderson, Traffic Studies Supervisor to Ron Krosting, Traffic Engineer, June 29,1988. 8. Fambro, D.B. and D.L. Woods, Safety Benefits of Left- Turn Signal Phasing Alternatives. Submitted to the Institute of Transportation Engineers for possible publication in the ITE Journal, June, 1981. 9. U. S. Department of Transportation, Federal Highway Administration, Guidelines for Signalized Left Turn Treatments, Washington, D.C., November, 1981. 10. Machemehl,RandyB.andAnnM.Mechler,ComparativeAnalysis oflefi- Turn PhaseSequencing, TransportationResearch Record 956, Transportation Research Board, Washington, D.C., 1984. 11. Grifin, Lindsay I., l%reeProceduresforEvaluatingHighway Safety Improvement Programs, A presentation at the American Society of Civil Engineers 1982 Annual Convention, New Orleans, Louisiana, October 29,1982. 12. Transportation Research Board, Highway Capacity Manual, Special Report 209, Washington, D.C., 1985. 13. American Association of Staff Highway and Transportation Officals, A Manual on User BeneJt Analysis of Highway and Bus Improvements, 1977. 14. Brookes, Joni; Collins; Rich and Haenel, Herman E., Texas State Department of Highways and Public Transportation, Study ofMVTCD and City ofDallas Permissive/Protected Lead/Lag Left Turn Phasings, June, 1990. 15. U.S. Department of Transportation, Federal Highway Administration, Manual on Uniform Trafic Control Devices,Washington, D.C. 1988. APPENDIX A PHASE SELECXION GUIDELINE FOR LEm-TURN PHASES APPENDIX A PHASE SE1,ECTION GUIDELINE FOR LEFT-TURN PHASES UGGl NG LEFT LEADING LEFT Best S u i t e d t o 1 - l n t e r s e c t l o n s or where opposing l e f t - t u r n s a r e p r o h l b l ted. Can be used w \ t h e f t h e r pretlmed or actuated c o n t r o l . LEA01 Kj DUAL LEFT-TURN Use s b u l d bo kept t o a mlnlmum where no l e f t t u r n lane ex\ s t s due t o p o s s l b i l i t y o f one rnover e n t b l o c k l n g tho o t h e r . L LAGGI ffi DUAL LEFT-TURtI BhSlC TWEE W h S E S l n p l e s t and l e a s t exponstve. Can be provided w l t h prettmed or s l n g l e r l n g actuated control lers. L Showl d be kept t o mlnlmum slnce bo?h l e f t - t u r n s r e c e l v e e x a c t l y t h e same green time. Lagging l e t t shoul d on1 y be used when system o p w a t l o n needs donlnate s a f e t y c o n s l d c r a t l o n . 0Vf??lhD PVASI ?G Best s u l t e d t o dual r l n g uated c o n t r o l l e r s . bct- Most ef t i c l e n t nathod o f provldIng p r o t e c t e d on1 y l e f t - t u r n phase. __c Through mvement Is a l towed to move when t h e r e I s no opposl ng left-turn traffic. LEAD-LKi LEFT-TURN P M S I N; - SPLIT FHASING Best when l e t t - t u r n vo!ume I s very heavy I n b o t h d l r e c t l o n s and Is equal o r g r e a t e r than the canpanion t h o u g h movenent. Especially e t f e c t l v e when one of t h e through Isnes must be used as an o p t l o n a l l e f t - t u r n lane o r where a separate l e f t - t u r n lane can n o t be provided. OVERLAP RlASlNG E l t h e r pretlmed or dual r l n g actuated c o n t r o l l e r can be used. As e f f l c l e n t as dual o v e r l a p phasl ng. ,f- l e t t with Should be used on1 y f o r Interconnected s i g n a l s where I t a s use uIII Increase t h s w i d t h of the progressive band. APPENDIX B SCOTI'SDALE ACCIDENT ANALYSIS TABLE B1 SCOTTSDALE ACCIDENT ANALYSIS TEST IMERSECTIONLEFT TURN ACCIDENT SUMMARY TABLE 6-1 CONIlNUED TABLE 8 2 SCOllSOALE ACClDENT ANALYSIS TEST INTERSECTION TOTAL ACCIDENT SUMMARY TABLE B-2 CONTINUED INTERSECTION SCOTTSOALElTHOMAS TABLE 6-3 SCOTTSOALE ACCIDENT ANALYSIS CONTROL INTERSECTION ACCIDENT SUMMARY TABLE B-3 CONTINUED TABLE B-3 CONnNUED TABLE B-3 CONTINUED TABLE B3 CONTINUED TABLE 5 3 CONnNUEO APPENDIX C PIMA COUNTY/TUCSBN ACCIDENT ANALYSIS TABLE C-1 DURATION OF STUDY PERIODS NUMBER OF DAYS IN ANALYSIS PERIODS INTERSECION AJO WAY IPAL0 VERDE RD. ALVERNON WAY IIRVINGTON RD. ALVERNON WAY IVALENCIA RD. CAMPBEU AVE. I RIVER RD. CRAYCROFT AVE. IRIVER RD. CRAYCROFT AVE. I SUNRISE DR. DODGE BLVD. IRIVER RD. DOS HOMBRES1TANQUE VERDE RD FIRST AVE. / INA RD. FIRST AVE. 1ORANGE GROVE RD. FIRST AVE. IRIVER RD. INA RD. 1LA CANADA DR. INA RD. / LA CHOUA BLVD. INA RD. 1 OLDFATHER RD. INA RD. I THORNYDALE RD. KOLB RD. IVALENCIA RD. LA CHOLLA BLVD. IORANGE GROVE RD. MISSION RD. 1VALENCIA RD. ORANGE GROVE RD. I SKYLINE DR. RIVER RD. / SWAN RD. SUNRISE DR. 1SWAN RD. BEFORE 730 311 730 730 730 730 417 730 730 730 231 730 730 730 730 325 331 730 730 562 913 AFTER 878 878 884 863 861 857 862 856 857 877 877 848 847 826 827 823 871 883 863 711 793 TABLE C-2 WILCOXEN SIGNED - RANKS TEST PlMA COUNTY ACCIDENT LEFT TURN RATE ANALYSIS 1NTERSECTION AJO WAY IPAL0 VERDE RD. NORTHBOUND SOUTHBOUND ALVERNON WAY / IRVINGTON RD. SOUTHBOUND EASTBOUND ALVERNON WAY / VALENCIA RD. EASTBOUND WESTBOUND CAMPBELL AVE. / RIVER RD. NORTHBOUND WESTBOUND CRAYCROFT AVE. IRIVER RD. SOUTHBOUND WESTBOUND CRAYCROFT AVE. I SUNRISE DR. NORTHBOUND (P) DODGE BLVD. IRIVER RD. WESTBOUND DOS HOMBRES / TANQUE VERDE RD EASTBOUND WESTBOUND FIRST AVE. / INA RD. NORTHBOUND * FIRST AVE. / ORANGE GROVE RD. NORTHBOUND SOUTHBOUND FIRST AYE. I RIVER RD. NORTHBOUND SOUTHBOUND INA RD. I LA CANADA DR. EASTBOUND WESTBOUND INA RD. ILA CHOLLA BLVD. ACCIDENT RATE RANK DIFFERENCE W/ SIGN TABLE C-2 CONT. INTERSECION INA RD. ILA CHOUA BLVD. EASTBOUND WESTBOUND INA RD. IOLDFATHER RD. EASTBOUND INA RD. IMORNYDALE RD. NORTHBOUND SOUTHBOUND EASTBOUND WESTBOUND KOLB RD. IVALENCIA RD. NORTHBOUND (P) SOUTHBOUND (P) EASTBOUND (P) WESTBOUND (P) LA CHOLIA BLVD. /ORANGE GROVE RD. NORTHBOUND (P) SOUTHBOUND (P) EASTBOUND (P) WEST8OUND (P) MISSION RD. / VALENCIA RD. EASTBOUND WESTBOUND ORANGE GROVE RD. ISKYUNE DR. NORTHBOUND (P) RIVER RD. I SWAN RD. NORTHBOUND WESTBOUND SUNRISE DR. 1SWAN RD. SOUTHBOUND * ACCIDENT RATE RANK DIFFERENCE WI SIGN - 0.216 + 0.090 +5 - 0.176 - 14 - 0.1 70 - 13 + 0.466 + 298 + 25 T(+) = - - 18 P Protected only left-turns * - Protectedlpermittedleft-turnsin the before period and protected left turns in the after period. T(-) = -297 TABLE C 3 WILCOXEN SIGNED RANKS TEST PlMA COUNTY ANALYSIS OF NUMBER OF LEFT-TURN ACCIDENTS - INTERSECTION AJO WAY / PAL0 VERDE RD. NORTHBOUND SOUTHBOUND ALVERNON WAY / IRVINGTON RD. EASTBOUND ALVERNON WAY I VALENCIA RD. EASTBOUND WESTBOUND CAMPBELL AVE. I RIVER RD. NORTHBOUND WESTBOUND CRAYCROFT AVE. I RlVER RD. SOUTHBOUND DODGE BLVD. I RIVER RD. WESTBOUND DOS HOMBRES / TANQUE VERDE RD EASTBOUND WESTBOUND FIRST AVE. IINA RD. NORTHBOUND FlRST AVE. I ORANGE GROVE RD. NORTHBOUND SOUTHBOUND FIRST AVE. / RlVER RD. NORTHBOUND SOUTHBOUND INA RD. / LA CANADA DR. EASTBOUND WESTBOUND INA RD. / LA CHOUA BLVD. EASTBOUND WESTBOUND INA AD. / OLDFATHER RD. EASTBOUND ACCIDENT DIFFERENCE RANK TABLE C 3 CONT. INTERSECTION INA RD. / THORNYDALE RD. NORTHBOUND SOUTHBOUND EASTBOUND WESTBOUND KOLB RD. I VALENCIA RD. EASTBOUND (P) LA CHOUA BLVD. / ORANGE GROVE RD. SOUTHBOUND (P) EASTBOUND (P) WESTBOUND (P) MISSION RD. / VALENCIA RD. EASTBOUND WESTBOUND ORANGE GROVE RD. SKYLINE DR. NORTHBOUND (P) RIVER RD. /SWAN RD. NORTHBOUND SUNRISE DR. / SWAN RD. SOUTHBOUND ACCIDENT DIFFERENCE RANK 2.96 1.70 -0.56 -2.00 -26.5 -2.25 -28 0.42 0.42 -1.10 +9.5 +9.5 -21 0.83 -0.26 +18 -5 1.51 +24 1.13 +22 0.80 +17 T(+) = 330 +34 +25 -14 T(-) = -265 TABLE C-4 WLLCOXEN TEST BASED ON LEFT TURN ACClDENT RATES (CITY OF TUCSON ARTERIAL/ARTERIAL INTERSECTIONS) INTERSECTION AJO WAY IMISSION RD. AJO WAY IINTERSTATE 19 AJO WAY I12TH AVE. ALVERNON WAY IBROADWAY BLVD. ALVERNON WAY I22ND ST. BROADWAY BLVD. ICAMPBELL AVE. BROADWAY BLVD. ICOUNTRY CLUB RD. BROADWAY BLVD. ICRAYCROFT RD. BROADWAY BLVD. IKOLB RD. BROADWAY BLVD. ISWAN RD. BROADWAY BLVD. IWllMOT RD. CAMPBELL AVE. IFORT LOWELL RD. CAMPBEU AVE. IGRANT RD. CAMPBEU AVE. ISPEEDWAY BLVD. CONGRESS ST. IGRANADA AVE. CONGRESS ST. IINTERSTATE 10 COUNTFIY CLUB RD. IGRANT RD. COUNTRY CLUB RD. ISPEEDWAY BLVD. COUNTRY CLUB RD. IVALENCIA RD. CRAYCROFT RD. I GOLF UNKS RD. CRAYCROFT RD. I22ND ST. FORT LOWEU RD. IORACLE RD. GOLF UNKS RD. IKOLB RD. GOLF UNKS RD. IWlLhtlOT RD. GRANT RD. IORACLE RD. GRANT RD. ISTONE AVE. GRANT RD. ISWAN RD. GRANT (KOLB) RD. 1 TANQUE VERDE RD. GRANT RD. / FIRST AVE. GRANT RD. IINTERSTATE 10 KOLB RD. ISPEEDWAY BLVD. KOLB RD. I22ND ST. MAIN AVE. ISPEEDWAY BLVD. MIRACLE MILE /ORACLE RD. ACCIDENT RATE DIFFERENCE RANK Wl SIGN TABLE C-4 CONT. INTERSECTION NOGALES HIGHWAY IVALENCIA RD. ORACLE RD. / PRINCE RD. ORACLE RD. 1 RIVER RD. ORACLE RD. / WETMORE RD. SPEEDWAY BLVD. ISTONE AVE. SPEEDWAY BLVD. 1 SWAN RD. SPEEDWAY BLVD. 1 WILMOT RD. SPEEDWAY BLVD. I INTERSTATE 10 ST. MARY'S RD. / INTERSTATE 10 SWAN RD. I22ND ST. VALENCIA RD. 1 12TH AVE. WETMORE RD. IFIRST AVE. WILMOT RD. 1 5 M ST. WILMOT RD. I22ND ST. INTERSTATE 10 I22ND ST. 5TH AVE. I INTERSTATE 1 0 ACCIDENT RATE DIFFERENCE RANK W I SIGN + 0.238 t 40 - 19.5 - 25 - 0.077 - 0.127 - 0.225 - 38 + 0.159 + 31 - 0.017 -3 - 24 - 0.108 + 35.5 + 42 + 48 + 0.177 + 0.265 + 0.596 - 0.002 + 0.072 - 0.084 - + 17.5 - 21 - 0.069 + 0.134 + 0.063 T (+) = + 669.5 1 - 16 - T (-) = 553.5 + 26 + 15 TABLE C-5 WILCOXEN TEST BASED ON LEFT-TURN ACCIDENT RATES (CITY OF TUCSON ARTERIAIJCOLLECTOR INTERSECTIONS) INTERSECTlON ACCIDENT DIFFERENCE + 0.151 ALVERNON WAY 129TH ST. AUTO MALL DR. 1ORACLE RD. BROADWAY BLVD. ICOLUMBUS BLVD. BROADWAY BLVD. IRANDOLPH WAY BROADWAY BLVD. I ROSEMONT BLVD. CHERRY AVE. I 22ND ST. COLUMBUS BLVD. 122ND ST. GRANT RD. 1 WlLMOT RD. LIMBERLOSTRD. IFIRST AVE. ORACLE RD. IROGER RD. SANTA CLARA AVE. 1VALENCIA RD. TUCSON BLVD. IVALENCIA RD. - 0.056 + 0.101 + 0.037 - 0.002 - 0.136 + 0.256 - 0.137 - 0.014 - 0.141 - 0.178 0.474 T (+) + 46.5 I RANK WI SIGN + 10 -4 +5 +3 - 1 6 12 - 7.5 2 9 7.5 -I1 T (-) = - 31.5 + + + TABLE C-6 WILCOXEN TEST BASED ON NUMBER OF LEFT-TURN ACCIDENTS (CITY OF TUCSON ARTERlAtJCOLLECTOR INTERSECTIONS) INTERSECTION ALVERNON WAY I 29M ST. AUTO MALL DR. IORACLE RD. BROADWAY BLVD. I COLUMBUS BLVD. BROADWAY BLVD. 1RANDOLPH WAY BROADWAY BLVD. / ROSEMONT BLVD. CHERRY AVE. I22ND ST. COLUMBUS BLVD. / 22ND ST. GRANT RD. / WILMOT RD. UMBERLOST RD. / FIRST AVE. ORACLE RD. / ROGER RD. SANTA CLARA AVE. 1VALENCIA RD. TUCSON BLVD. IVALENCIA RD. T (+) = 33 T (-) = 4 5 DIFFERENCE IN ACCIDENTS PER YEAR -2.67 1.17 -2.83 -0.17 -0.17 1.83 -4.67 3.17 0.33 -1.50 -1.33 1.67 RANK WI SIGN -9 +4 -10 1.5 - 1.5 +8 -12 11 3 -6 -5 7 - + + TABLE C-7 WILCOXEN TEST BASED ON NUMBER OF LEFT-TURN ACCIDENTS (CITY OF TUCSON ARTERIAIJARTERIAL INTERSECTIONS) INTERSECTION AJO WAY / MISSION RD. AJO WAY / 12TH AVE. ALVERNON WAY I BROADWAY BLVD. ALVERNON WAY /22ND ST. BROADWAY BLVD. I CAMPBELL AVE. BROADWAY 8LVD. 1 COUNTRY CLUB RD. BROADWAY BLVD. I CRAYCROFT RD. BROADWAY BLVD. I KOLB RD. BROADWAY BLVD. I SWAN RD. BROADWAY BLVD. IWILMOT RD. CAMPBELL AVE. / FORT LOWELL RD. CAMPBELL AVE. I GRANT RD. CAMPBELL AVE. / SPEEDWAY BLVD. CONGRESS ST. / GRANAOA AVE. CONGRESS ST. / INTERSTATE 10 COUNTRY CLUB RD. / GRANT RD. COUNTRY CLUB RD. / SPEEDWAY BLVD. COUNTRY CLUB RD. /VALENCIA RD. CRAYCROFT RD. / GOLF LINKS RD. CRAYCROFT RD. / 22ND ST. FORT LOWELL RD. / ORACLE RD GOLF LINKS RD. / KOLB RD. GOLF LINKS RD. I WILMOT RD. GRANT RD. / ORACLE RD. GRANT RD. ISTONE AVE. GRANT RD. / SWAN RD. GRANT (KOLB) RD./TANQUE VERDE RD. GRANT RD. / FIRST AVE. GRANT RD. / INTERSTATE 10 DIFFERENCE IN LEFT-TURN ACCIDENTS RANK W/ SIGN TABLE C-7 CONT. INTERSECTION KOLB RD. ISPEEDWAY BLVD. KOLB RD. 122ND ST. MAIN AVE. / SPEEDWAY BLVD. MIRACLE MILE / ORACLE RD. NOGALES HIGHWAY IVALENCIA RD. ORACLE RD. / PRINCE RD. ORACLE RD. / RIVER RD. ORACLE RD. / WETMORE RD. SPEEDWAY BLVD. / STONE AVE. SPEEDWAY BLVD. / SWAN RD. SPEEDWAY BLVD. / WILMOT RD. SPEEDWAY BLVD. / INTERSTATE 10 ST. MARY'S RD. / INTERSTATE 10 SWAN RD. / 22ND ST. VALENCIA RD. / 12TH AVE. WETMORE RD. / FIRST AVE. WILMOT RD. / 5TH ST. WILMOT RD. /22ND ST. INTERSTATE 10 / 22ND ST. 5TH AVE. / INTERSTATE 10 DIFFERENCE IN LEFT-TURN ACCIDENTS RANK W/ SIGN TABLE C-8 TOTAL EQUIVALENT ACCIDENTS PER YEAR PlMA COUNTY INTERSECTIONS NUMBER OF ACCIDENTS BEFORE AFTER DIFFERENCE INTERSECTION AJO WAY IPAL0 VERDE RD. NB/SB % CHANGE 15.50 7.47 -8.03 -51.81 SBlEB 3.53 9.54 6.01 170.25 EBWB 9.00 7.02 -1.98 -22.00 NBWB 11.OO 13.56 2.56 23.27 SBWB 4.50 7.63 3.13 69.56 NB (PI 2.50 1.28 -1.22 1.75 0.85 -0.90 12.00 8.09 -3.91 3.00 2.55 -0.45 8.00 7.08 -0.92 23.81 10.42 -13.39 10.00 15.52 5.52 12.50 16.38 3.88 1.50 4.42 2.92 17.50 23.79 6.29 13.48 11.11 -2.37 6.59 10.04 3.45 13.50 6.61 -6.89 ALVERNON WAY I IRVINGTON RD. ALVERNON WAY IVALENCIA AD. CAMPBELL AVE. IRIVER RD. CRAYCROFT AVE. I RIVER RD. CRAYCROFT AVE. I SUNRISE DR. DODGE BLVD. IRIVER RD. WB DOS HOMBRES / TANQUE VERDE RD EBlWB FIRST AVE. IINA RD. NB ** FIRST AVE. I ORANGE GROVE RD. NBlSB FIRST AVE. I RIVER RD. NBISB INA RD. ILA CANADA DR. EBMlB INA RD. /LA CHOLLA BLVD. EBJWB INA RD. IOLD FATHER RD. EB INA RD. I THORNYDALE AD. NB/SB/EB/WB KOLB RD. IVALENCIA AD. NBlSB/EBWe (P) LA CHOLLA BLVD. /ORANGE GROVE RD. NB/SB/EB/WB (P) MISSION RD. / VALENCIA RD. EBlWB TABLE C-8 CONT. NUMBER OF ACCIDENTS BEFORE AFTER DIFFERENCE INTERSECTION % CHANGE ORANGE GROVE RD. I SKYLINE DR. NB (P) 10.50 12.71 2.21 21.05 WB 3.90 6.67 2.77 71.03 SB ** 2.80 186.86 1.38 184.12 -1.42 -2.74 -50.71 -1.47 RIVER RD. / SWAN RD. SUNRISE DR. /SWAN RD. TOTAL NOTE: indicates the approaches included in the analysis (P) indicates approaches with protected only left-turns ** protectedJpermiteedleft-turns in the before period and protected left turns in the after period TABLE C-9 WILLCOXEN TEST BASED ON TOTAL ACCIDENTS PlMA COUNTY INTERSECTIONS INTERSECTION DIFFERENCE IN TOTAL INTERSECTIONACCIDENTS AJO WAY / PAL0 VERDE RD. NBISB* -8.03 ALVERNON WAY / IRVINGTON RD. SBIEB 6.01 ALVERNON WAY I VALENCIA RD. EB/WB -1.98 CAMPBELL AVE. / RIVER RD. NB/WB 2.56 CRAYCROFT AVE. / RIVER RD. SB/WB 3.13 CRAYCROFT AVE. / SUNRISE DR. NB(P) -1.22 WB -0.90 DODGE BLVD. / RIVER RD. DOS HOMBRES / TANQUE VERDE RD EBNB FIRST AVE. I INA RD. NB** FIRST AVE. IORANGE GROVE RD. NB/SB FIRST AVE. I RIVER RD. NBISB INA RD. / LA CANADA DR. EBNB INA RD. / LA CHOLLA BLVD. EBWB INA RD. / OLD FATHER RD. EB INA RD. / THORNYDALE RD. NB/SB/EB/WB KOLB RD. / VALENCIA RD. NB/SB/EB/WB(P) LA CHOLLA BLVD. / ORANGE GROVE RD. NBISBIEB/WB(P) MISSION RD. I VALENCIA RD. EB/WB ORANGE GROVE RD. / SKYLINE DR. NB(P) -3.91 -0.45 -0.92 13.39 5.52 2.92 6.29 -2.37 3.45 -6.89 2.21 RANK W/ SIGN -20 17 -6 INTERSECTION TABLE C-9 CONT. DIFFERENCE IN TOTAL INTERSECTION ACCIDENTS RIVER RD. / SWAN RD. SUNRISE DR. / SWAN RD. NOTE: * indicates the approaches included in the analysis P indicates approaches with protected only left-turns ** protectedlpermitted left-turns in the before period and protected left-turns in the after period. RANK W/ SIGN TABLE C-10 TOTAL INTERSECTION ACCIDENTS (CITY OF TUCSON ARTERIAUARTERIAL INTERSECTIONS) INTERSECTION ACCIDENTS PER YEAR BEFORE AFTER DIFFERENCE % CHANGE 1NTERSECTlON AJO WAY I MISSION RD. AJO WAY I INTERSTATE 19 AJO WAY I 12TH AVE. ALVERNON WAY / BROADWAY BLVD. ALVERNON WAY /22ND ST. BROADWAY BLVD. I CAMPBELL AVE. BROADWAY BLVD. / COUNTRY CLUB RD. BROADWAY BLVD. I CRAYCROFT RD. . 12.33 14.5 2.17 17.60 16.33 25.00 24.00 15.0 19.0 24.5 -1.33 -6.00 0.50 -8.14 -24.00 2.08 65.33 20.33 38.5 27.0 -26.83 6.67 -41.07 32.81 21.00 14.5 -6.50 -30.95 15.33 8.0 -7.33 -47.81 BROADWAY BLVD. I KOLB RD. BROADWAY BLVD. /SWAN RD. BROADWAY BLVD. / WILMOT RD. CAMPBELL AVE. / FORT LOWELL RD. 19.67 21.33 11.5 26.0 -8.17 4.67 -41.54 21.89 22.33 20.00 9.5 22.5 -12.83 2.50 -57.46 12.50 CAMPBELL AVE. / GRANT RD. CAMPBELL AVE. / SPEEDWAY BLVD. CONGRESS ST. / GRANADA AVE. CONGRESS ST. / INTERSTATE 10 24.67 34.00 5.67 28.00 29.5 38.0 7.0 18.0 4.83 4.00 1.33 -10.00 19.58 11.76 COUNTRY CLUB RD. / GRANT RD. COUNTRY CLUB RD. /SPEEDWAY BLVD. COUNTRY CLUB RD. / VALENCIA RD. CRAYCROFT RD. / GOLF LINKS RD. 17.67 22.5 4.83 27.33 6.67 2.33 9.0 4.0 2.33 1.67 34.93 71.67 10.33 20.5 10.17 98.45 CRAYCROFT RD. I22ND ST. FORT LOWELL RD. / ORACLE RD. GOLF LINKS RD. / KOLB RD. GOLF LINKS RD. / WILMOT RD. 42.00 41.5 -0.50 -1.19 3.33 17.67 19.00 20.5 22.0 18.5 17.17 4.33 -0.50 515.62 24.50 -2.63 GRANT RD. / ORACLE RD. GRANT RD. / STONE AVE. GRANT RD. / SWAN RD. GRANT (KOLB) R0.ITANQUE VERDE RD. 21.67 23.00 22.0 23.0 0.33 0.00 1.52 0.00 29.33 28.0 -4.53 19.33 22.0 -1.33 2.67 23.46 -35.71 13.81 TABLE C-10 CONT. INTERSECTION INTERSECTION ACCIDENTS PER YEAR BEFORE AFTER DIFFERENCE % CHANGE GRANT RD. IFIRST AVE. GRANT RD. / INTERSTATE 10 KOLB RD. /SPEEDWAY BLVD. KOLB RD. / 22ND ST. MAIN AVE. /SPEEDWAY BLVD. MlRACLE MILE / ORACLE RD. NOGALES HIGHWAY / VALENCIA AD. ORACLE RD. I PRINCE RD. ORACLE RD. 1RIVER RD. ORACLE RD. I WETMORE RD. SPEEDWAY BLVD. / STONE AVE. SPEEDWAY BLVD. 1 SWAN RD. SPEEDWAY BLVD. / WILMOT RD. SPEEDWAY BLVD. 1 INTERSTATE 10 ST. MARY'S AD. / INTERSTATE 10 SWAN RD. / 22ND ST. 13.33 6.5 -6.83 5.33 17.67 41.33 17.5 16.0 35.0 12.17 -1.67 -6.33 -51.24 228.33 -9.45 -1 5.32 11.33 11.33 17.0 21.5 24.67 18.5 5.67 10.17 -6.17 0.33 50.04 89.76 -25.01 4.95 6.67 7.0 15.00 20.67 22.67 24.33 14.5 9.5 5.5 20.0 -0.50 -11.17 -17.17 -4.33 3.33 -54.04 -75.74 -17.80 33.00 12.5 21.33 12.0 -20.50 -9.33 -62.12 -43.74 953.5 -61.49 -6.06 VALENCIA RD. I 12TH AVE. WETMORE RD./ FIRST AVE. WlLMOT RD. I 5TH ST. WILMOT RD. / 22ND ST. INTERSTATE 10 / 22ND ST. 5TH AVE. I INTERSTATE 10 TOTAL 1014.99 TABLE C-11 WILLCOXEN TEST BASED ON TOTAL INTERSECTION ACCIDENTS (CITY OF TUCSON ARTERlAL/ARTERlAL INTERSECTIONS) INTERSECTIONS AJO WAY / MISSION RD. AJO WAY I INTERSTATE 19 AJO WAY / 12TH AVE. ALVERNON WAY / BROADWAY BLVD. ALVERNON WAY / 22ND ST. BROADWAY BLVD. I CAMPBELL AVE. BROADWAY BLVD. I COUNTRY CLUB RD. BROADWAY BLVD. / CRAYCROFT RD. BROADWAY BLVD. / KOLB RD. BROADWAY BLVD. /SWAN RD. BROADWAY BLVD. / WILMOT RD. CAMPBELL AVE. / FORT LOWELL RD. CAMPBELL AVE. / GRANT RD. CAMPBELL AVE. /SPEEDWAY BLVD. CONGRESS ST. / GRANADA AVE. CONGRESS ST. / INTERSTATE 10 COUNTRY CLUB RD. / GRANT RD. COUNTRY CLUB RD. /SPEEDWAY BLVD. COUNTRY CLUB RD. / VALENCIA RD CRAYCROFT RD. / GOLF LINKS RD CRAYCROFT RD. /22ND ST. FORT LOWELL RD. / ORACLE RD. GOLF LINKS RD. / KOLB RD. GOLF LINKS RD. / WILMOT RD. GRANT RD. / ORACLE RD. GRANT RD. / STONE AVE. GRANT RD. / SWAN RD. GRANT (KOLB) RD.iTANQUE VERDE RD. GRANT RD. / FIRST AVE. GRANT RD. 1 INTERSTATE 10 DIFFERENCE IN TOTAL INTERSECTION ACCIDENTS 2.17 -1.33 -6.00 0.5 RANK W/ SIGN 19 -11.5 -32 5.5 INTERSECTIONS TABLE C-11 CONT. DIFFERENCE IN TOTAL iNTERSECTlON ACCIDENTS KOLB RD. ISPEEDWAY BLVD. KOLB RD. /22ND ST. MAIN AVE. 1 SPEEDWAY BLVD. MIRACLE MILE / ORACLE RD. NOGALES HIGHWAY / VALENCIA RD. ORACLE RD. I PRINCE RD. ORACLE RD. / RIVER RD. ORACLE RD. / WETMORE RD. SPEEDWAY BLVD. / STONE AVE. SPEEDWAY BLVD. i SWAN RD. SPEEDWAY BLVD. / WILMOT RD. SPEEDWAY BLVD. / INTERSTATE 10 ST. MARY'S RD. / INTERSTATE 10 SWAN RD. / 22ND ST. VALENCIA RD. / 12TH AVE. WETMORE RD. 1 FIRST AVE. WILMOT RD. / 5TH ST. WILMOT RD. / 22ND ST. INTERSTATE 10 / 22ND ST. 5TH AVE. / INTERSTATE 10 RANK W/ SIGN TABLE C-12 TOTAL INTERSECTION ACCIDENTS PER YEAR (CITY OF TUCSON ARTERIAUCOLLECTOR INTERSECTIONS) INTERSECTION ALVERNON WAY / 29TH ST. AUTO MALL DR. I ORACLE RD. BROADWAY BLVD. / COLUMBUS BLVD. BROADWAY BLVD. I RANDOLPH WAY BROADWAY BLVD. I ROSEMONT BLVD. CHERRY AVE. / 22ND ST. COLUMBUS BLVD. / 22ND ST. GRANT RD. / WILMOT RD. LIMBERLOST RD. / FIRST AVE. ORACLE RD. I ROGER RD. SANTA CtARA AVE. / VALENCIA RD. TUCSON BLVD. / VALENCIA RD. TOTAL INTERSECTIONACCIDENTS PER YEAR BEFORE AFTER DIFFERENCE % CHANGE 30.67 9.33 12.67 8.33 8.67 14.00 20.33 10.00 7.33 8.67 6.67 10.00 146.67 1 8.50 3.00 10.00 8.50 1 1.50 17.00 8.50 14.50 6.50 7.00 3.50 12.00 1 20.50 -12.17 -6.33 -2.67 0.17 2.83 3.00 -11.83 4.50 -0.83 -1.67 -3.17 2.00 -26.17 -39.68 -67.85 -21.07 2.04 32.64 21.43 -58.19 45.00 -1 1.32 -1 9.26 -47.53 20.00 -1 7.84 TABLE C-13 WILCOXEN TEST BASED ON TOTAL INTERSECTIONACCIDENTS (CITY OF TUCSON ARTERIALJCOLLECTOR INTERSECTIONS) DIFFERENCE IN INTERSECTIONS ACCIDENTS PER YEAR ALVERNON WAY / 29TH ST. -12.17 AUTO MALL DR. / ORACLE RD. -6.33 -2.67 BROADWAY BLVD. I COLUMBUS BLVD. BROADWAY BLVD. / RANDOLPH WAY 0.17 BROADWAY BLVD. / ROSEMONT BLVD. 2.83 CHERRY AVE. / 22ND ST. 3.00 COLUMBUS BLVD. I22ND ST. -1 1.83 GRANT RD. 1 WILMOT RD. 4.50 LIMBERLOST RD. / FIRST AVE. -0.83 ORACLE RD. / ROGER RD. -1.67 SANTA CLARA AVE. / VALENCIA RD. -3.17 TUCSON BLVD. / VALENCIA RD. 2.00 RANK W/ SIGN -12 -10 -5 1 6 7 -11 9 -2 -3 -8 - 4 APPENDIX D PHOENIX AREA INTERSECTION ANALYSIS D-1 SUMMARY WORKSHEEI'S FOR INTERSECTION DELAY D-2CITY OF PHOENIX 3RD CAR ACTUATION SI'UDY APPENDIX D-1 SUMMARY WORKSHEETS FOR INTERSECTION DELAY STUDY INTERSECTION: SURVEY DATE: Glendate 8 51st Ave Leading 3rd C a r Delay 01/10/90 Northbound Southbound Eastbwnd Westbound NB SB EB VB Intersection Totals Time Period L t Th/Rt Lt Th/Rt L t Th/Rt L t Th/Rt Total Total Total Total L t Th/Rt Total ----------- ------------------- ----------- = c f t l D = = = = = = = f l = = E = = = = t = C = I = L ~-------------I------------=-------------------=-----------=-------------------=----------- 60 101 71 98 90 69 41 77 33 255 250 171 132 54 2376 261 345 235 808 310 4:45-5:OOpn 17 5:OO-5:15 pn 12 16 5:15-5:3Opn 5:SO-5i45 pa 9 622 507 608 639 49 53 W 96 135 56 23 510 579 331 246 639 519 624 648 150 151 168 137 1666 2430 606 714 387 269 233 290 212 125 1488 1434 1179 1094 1721 1724 1391 1219 1043 1976 860 5195 6055 326 340 212 165 606 -------------------.-----.------------------------------ TOTAL Glendale L 51st Ave Leading 3 r d Car 01/10/90 Volune INTERSECTION: SURVEY DATE: Northbound Southbound E a s t b d Uestbwnd NB SB EB UB Intersection Totals Time Period L t Th/Rt L t Th/Rt tt Th/Rt L t Th/Rt Total Total Total Total L t Th/Rt Total =3115==D=L===l==I=S=LII=====S====I==S======-------------------ISlSElSSI====I=====EZ============================= 4:45-5:OO pn 5:OO-5:15 pm 5:15-530 pn 5:SO-5:45pn 13 15 13 13 304 286 314 303 30 39 42 41 162 155 132 166 41 43 23 26 177 200 155 162 44 50 42 27 218 261 240 220 317 301 327 316 192 194 174 207 218 243 178 188 262 311 282 247 128 147 120 107 861 902 841 851 989 1049 961 958 3455 3957 ------.-----*--------------.--------*.----------------------.---------------------------.- TOTAL 54 1207 152 615133 694163 939 1261 767 827 1102 502 Glendale 6 51st Ave Leading 3rd Car 01/10/90 Delay Per Vehicle (sec/veh) INTERSECTiON: SURVEY DATE: Northbourd Southbovnd E a s t W Uestbwnd NB SB EB VB Intersection Totals Tim Period L t Th/Rt L t Th/Rt L t ThfRt L t Th/Rt Totat Total Total Total L t Th/Rt Total clta=at='c==C==D.=========================5===================s===~============s~===================== 4:45-5:00 pn 20 5:OO-5:15 pn 12 5:15-530 pn 18 5:30-5:45 pn 10 31 27 29 32 25 20 35 22 9 TOTAL 30 26 8 26 31 8 27 7 19 9 22 19 17 12 33 41 20 13 35 33 21 17 30 26 29 31 12 12 14 10 22 21 18 13 35 34 21 16 17 29 27 29 12 19 27 25.70 27 30 27 18 26 24 21 19 26 25 22 19 ----.----*-------------------------------------------------15 27 22.55 22.95 IWTERSECTIOW: SURVEY DATE: Glendale 6 51st Ave Lagging Detoy 04/2/90 Northbound S o u t h M Eastbwnd Time Perlod C t Th/Rt L t Th/Rt L t Th/Rt Vestbound WB Sf3 EB VB Intersection l o t a t s Lt Th/Rt Total Total Total Total L t Th/Rt l o t a t ....................................................................................................... 4:30-4:45 pn 4:45-5:OOpn 5:OO-5:15pn 5:15-5:30 pn 26 23 6 33 359 431 626 285 158 133 278 161 503 396 437 373 155 146 198 146 161 537 252 649 198 860 149 1035 384 315 525 304 385 454 632 316 661 529 715 534 539 698 461 901 723 1058 450 1184 500 554 680 489 1783 1791 2448 1995 2283 2545 3128 2484 ---------------------------------------------.-----88 TOTAL 1699 730 1709 645 Clendale B S l s t Ave INTERSECTION: SURVEY DATE: 04/02/90 1528 760 3081 1787 2439 2173 3841 Lagging Volune Uorthboond Swthbound Eastbocsd Th/Rt L t Time Perlod l t Th/Rt L t Th/Rt Uestbovd US SB 14 10 9 11 TOTAL 44 IWTERSECTIOW: SURVEY DATE: 270 281 362 293 ---- ----- - --- --- --- - --- ------- 44 58 53 43 1206 198 166 198 153 172 42 36 35 37 689 150 Clendale & 51st Ave 04/02/90 EB UB 233 284 191 -291 230 371 250 304 210 256 206 215 235 223 215 703 185 904 1250 887 853 1089 180 143 Th/Rt Total ---------========= 45 52 42 46 193 187 Intersection Totals Th/Rt Total Total Totat Total L t Lt s=a=lll==ID==t='====a====I==II=C-----=---=-------==--------- 4:45-5:OOpn 5:OO-5:ISpn 5:15-5:SOpn 5:30-5:h5 pn 2223 8017 10240 180 278 243 272 296 862 857 925 858 1007 1013 l W 995 577 3502 407P 145 156 139 137 lagglng Delay Per Vehicte (sec/veb) Northbcund Southbound Eastbovd Tim Perlod L t Th/Rt l t Th/Rt L t Th/Rt Uesttwwnd UB SB €0 CIS Intersection Totals Lt Th/Rt Total Total Total Total L t Th/Rt Total 8=I==S=LZ=1D=llfllP=f==X=EI===I====I==::DILI==I::===I=====ft=====E==I=9I=5::=====I====E=L================ 4:45-5:OO pn 5:OO-5:15pn 5:15-5:3Opn S:30-5:45pn TOTAL 28 35 10 45 20 23 26 14 54 34 79 56 45 30 43 33 55 61 85 59 30 25 44 32 30 21 55 37 65 33 54 71 49 35 51 56 62 20 23 26 16 47 31 52 37 34 31 50 30 38 56 58 60 62 51 21 41 38 53 57.7934.34 A 52 53 73 54 31 31 40 35 U 35 44 37 -.--------------*----.-----------------------------.-----37.66 IMTERSECTION: SURVEY DATE: Northern 8 S l s t Ave 01/09/90 Leading 3 r d Car Delay Northbound Southbound Eestbarnd Westboud NB SB €6 VB Intersection Totals Th/Rt L t fh/Rt L t Th/Rt L t Th/Rt Total Total Total Total l t Th/Rt Total Time Period L t ~=~==IO~=~~====I=C=P=~===~=E==~~============C===~~~=~=~=~=~==========E=Z==~==E~=========I====I==D====== 5:OO-5.:15 pn 5:15-5:30 pn 5:30-5:45 pn 5:45-6:OOpn TOTAL 73 66 70 164 106 186 82 180 85 190 92 718 f53 379 270 5918 286 718 365 2120 45 1504 43 1172 82 2005 43 1237 213 INTERSECTION: SURVEY DATE: 77 Northern 01/09/90 L 51st Ave 42 45 39 24 573 1056 684 499 1549 1215 2087 1280 150 2812 6131 241 257 246 260 824 835 464 615 1101 362 523 723 1004 2485 2962 2959 3165 3248 2196 3229 3408 3520 2425 1014 11568 12582 270 243 272 229 Leading 3 r d Car Volm Northboud Southbound E a s t b a d Vestbound WB SB EB V8 Intersection Totals Time Period L t Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Total Total Total Total L t Th/Rt Total aa=n====mmxa=sm=~=c======cz===s============aemman=aanna=aa====a==%=a=a=======a====a====a=============z= 5:OO-5:15pn 5:15-5:30pn 5:30-5:45 pn 5:45-6:OO pn 49 46 60 51 320 294 280 281 38 35 39 45 161 151 157 143 51 44 38 46 272 244 243 236 38 264 24 29 14 290 283 274 369 340 340 332 199 186 196 188 323 288 281 282 302 314 312 288 176 149 166 156 1017 979 963 934 1193 1128 1129 1090 -------.-----------------------------------*-*-------*-----*---------*------.-------------- TOTAL 206 1115 157 612 179 Northern lL 51st Ave 01/09/90 INTERSECTION: SURVEY DATE: 995 105 1111 1381 769 1174 1216 647 3893 4540 Leading 3rd Car Delay Per Vehicle (sec/veh) Northband Southborad Eastboud Vestband NB SB €0 UB Intersection Totals Time Period t t Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Total Total Totat Total L t Th/Rt Total - - - - - - - = ~ = = = = = a ' a I = = a = = = = e-----------=-----------------------------==---.............................. =CDI====Z=======lf====IE=L:===IIEEI------ 5:OO-5:15 p 5:15-5:30 pn 5:30-5:45 pn 5:45-6:OOpn 14 14 21 13 71 30 60 31 107, 66 25 23 15 18 17 20 31 28 34 30 40 46 23 17 17 28 20 26 33 55 36 27 63 54 92 58 18 21 19 21 38 43 25 19 31 53 35 27 44 48 51 35 23 24 25 22 41 45 47 33 -*---------------------*----------------------------------------------------.--*----------- TOTAL 16 76 ' 2 7 18 31 32 21 38 67 20 32 37 23.51 44.57 41.57 INTERSECTION: SURVEY DATE: Northern B 51st Ave. 04/09/90 Lagging Delay Northbound Southbound Eastbcund Uestbound WB SB EB VB I n t e r s e c t i o n T o t a l s Time Period L t Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt T o t a l Total T o t a l Total L t Th/Rt T o t a l --------- -------- ~ = ~ ~ = = = = = = ~ = = ~ l = l = = ~ = = = = ~ ~ -----=---~ ~ ~----=--------301-------------= = = E E = =--------------r---------== ~ ~ = = 5:OO-5:15 pa 5:15-5:30 pn 5:30-5:45 pn 5:45-6:OOpn 205 69 156 1258 1020 926 60 392 135 157 148 104 TOTAL 510 3596 544 414 340 911 134 251 241 1010 71 296 242 627 28 311 176 522 61 458 1463 57s 1109 1092 1082 298 452 1251 1251 869 698 592 646 1120 359 814 558 574 401 3061 2856 2941 1523 3855 3414 3515 1924 1816 4069 2717 2347 10361 12708 549 408 444 415 ------------.---*-------.----------.-------..-------------------------------.------*------1272 999 3070 Northern L Slst Ave INTERSECTION: SURVEY DATE: 294 2423 4106 Lagging Volm 04/09/90 Northbound Southboud Eastborrnd Uestbound NB SB EB WB I n t e r s e c t i o n T o t a l s Time Period L t Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt T o t a l Total Total Total L t Th/Rt T o t a l ~~======1~====a=~=r1====~a=~~==t~a~t=~=~===z~a~u=~=~~======~======~===e:======1==2::r:====~1=======~====== 5:OO-5:15 pn 5:15-5:30 pm 5:30-5:45pn 5:45-6:00 pm 61 49 61 39 340 342 333 317 34 37 32 39 173 158 164 149 46 41 46 39 280 281 214 210 29 22 28 23 210 1332 142 644 172 985 102 312 286 326 279 401 391 394 356 207 195 196 188 1203 1542 786 341 308 354 302 170 149 167 140 1105 1067 1037 955 1275 1216 1204 10% 1157 1305 626 4164 4790 326 322 260 249 ---------------.-------------------.-------.------------- TOTAL IWTERSECTIOEI: Northern L 51st Ave SURVEY DATE: 04/09/90 Lagging Delay Per Vehicte (sec/veh) N o r t h b o u d Southbound E a s t k Time Period L t Th/Rt L t Th/Rt L t Uestbound Th/Rt L t NB SB EB WE Intersection Totals Th/Rt T o t a l Total T o t a l Total L t Th/Rt Total C~=l~=~l=~PSf=f~=~=L=ll=~~~===~~==~~LEI=DI=IK~=~==~=~I=~===E=~==D====I~===t====fL========i:====== 5:OO-5:15pn 5:15-5:30pn 5:30-5:45pn 5:45-6:00 pn 50 27 23 56 45 42 19 60 64 69 40 36 111 24 88 27 79 31 68 49 54 44 37 69 48 15 40 22 30 50 16 55 43 41 19 40 31 34 33 58 58 50 42 26 31 47 18 72 56 52 43 41 40 43 24 45 42 44 26 TOTAL 36 40 57 30 47 43 30 40 35 53 31 56.24 37.32 39.80 58 ---------------------.------------------------------87 INTERSECTIW: SURVEY DATE: Olive b S l s t Ave. 01/11/90 LEADlNG 3RD CAR Delay Northbound Southbound Eastbwnd Uestboud RE SB EB UB Intersectton Totals Time Period L t Th/Rt l t Th/Rt L t Th/Rt L t Th/Rt Total Total Total Total Lt Th/Rt Total --- ------------------- ------ 11:~~1=~=an~f=t====ii~=s~1=e1a~a~=~u~~=~~=r~===~=~oe=======t====se==t=e==---=-------------------~~------ - 5:OO-5:lSpn 5:15-5:30pn 5:30-5:45pn 5:45-6:OO pn 64 T8 87 84 477 734 531 520 102 107 120 102 198 180 212 117 81 58 82 44' 207 185 186 154 52 40 88 41 670 681 492 325 541 812 618 604 300 287 332 219 732 221 2168 2575 1138 722 721 500 566 299 283 377 271 1552 1780 1421 1116 1851 2063 1798 1387 997 2389 1230 5869 7099 288 243 268 198 ----------.--..-.-.---.-*---------.--.---..------------------ TOTAL 313 2262 431 IIITERSECTIOW: SURVEY DATE: 707 265 Olive & 5 l s t Ave. 01/11/90 LEMINC 3RD CAR Vole Northbound Southbound Eastbound U e s t W UB SB EB VB Intersection Totals Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Totat Total Total Total L t Th/Rt Total Time Period L t ~ ~ ~ ~ I ~ P I = ~ ~ ~ = I - P I ~ ~ = ~ ~ ~ = = ~ ~ ~ ~ ~ ~ ~ = ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ I ~ ~ I L ~ I I I I P I ~ I I = I = = ~ = = = = = I = ~ ~ ~ = I 5:OO-5:lS pn 5:15-5:30pn 5:30-5:45pn 5:45-6:OOpn 67 58 59 44 349 329 323 317 35 44 51 45 197 189 171 178 36 30 29 28 228 1318 175 735 123 170 100 162 176 31 39 40 35 320 416 387 382 361 232 233 222 223 206 220 191 204 351 406 385 332 169 171 179 152 1036 1075 1001 968 1205 1246 1180 1120 1329 1546 910 821 1474 671 4080 4751 367 U S 297 --------------------------------------------------..----- TOTAL Olive & 51et Ave. 01/11/90 INTERSECTIW: SURVEY DATE: 698 145 LEADING 3RD CAR Delay Per Vehicle (sec/veh) Northbound S o u t h b a d Eastbourj Uestboud N8 SB E0 YB Intersection Totals Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Total Total Total Total L t Th/Rt Total Time Period L t =Ol=Zlll==l===EI==========EE=DI==SE=DI==5II==I=D==ID==5=====B=Z=Z===========================I=======1========= 5:OO-5:15pn 515-5:30pn 5:30-5:45 pn 5:45-6:OOpa 14 20 22 29 21 33 25. 25 44 36 35 34 15 14 19 10 34 29 42 24 18 15 17 13 25 15 33 18 31 28 21 16 20 31 24 25 19 18 22 15 21 17 21 15 31 27 23 17 TOTAL 21 26 ' 3 7 14 32 16 23 24 25 19 18 24 27 25 32 27 22 25 21 17 23 25 23 19 --------*----------------------.-----------------------------27.50 21.58 22.41 INTERSECTION: SURVEY DATE: Olive i% 51st Ave. 04/13/90 LAGGING Delay Northbound Southbound Eastbound Uestboud NB SB EB UB Intersection Totals Time Period L t Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Total Total Total Total L t Th/Rt Total L~S~~D===~~=II~~==~I=====E=~E=BE=~=~===~:~~I=====~=P=I===I===L======D=====I~=I=L:====::E================ 5:05-5:20pl159 5:20-5:35 $in 93 5 3 5 - 5 5 0 pn 77 5:50-6:OSpn 67 342 269 389 181 515 118 251 108 419 105 349 63 262 80 256 81 359 239 1197 293 168 1067 297 111 496 325 59 364 501 482 592 318 688 530 380 364 464 1436 3% 1235 317 607 406 423 505 386 315 2317 2098 1570 11% 3089 2603 1956 1511 1286 329 1274 577 3124 1893 1962 1603 3701 1978 7181 9159 772 .-.-.----.-----------------------------*---------.---- TOTAL '396 1497 676 O l i v e & 51st Ave. 04/13/90 INTERSECTION: SURVEY DATE: LAGGING Volune Northbound Southbound Eastb0u-d Westbound NB SB EB VB Intersection Totals Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Total Total Total Totat L t Th/Rt Total Time Period L t ~~CEIL~I~DDEL=LX==II.i~===li=CE~~====I:~~~L==iX=~=~~=IDfSIEI~?5I¶DL~=I¶==II5ECI~i:S=~1=5I=~I=III==e===I~ 5:05-5:20pn 5:20-5:35 pn 5:35-5:5Opn 550-6:05pn TOTAL 50 48 40 33 179 183 149 149 34 25 27 24 214 1226 171 660 110 60 61 56 37 INTERSECTICN: SURVEY DATE: 351 370 261 244 176 175 147 163 51 41 43 25 345 363 354 286 411 431 317 281 229 231 189 182 210 200 174 187 3% 404 397 311 1% 175 166 119 1051 1091 911 842 1246 1266 1077 961 661 160 1348 1440 831 771 1508 655 3895 4550 LAGGING Delay Per Vehicle [sec/veh) Olive 6 51st Ave. 04/13/90 Northbound Southbwnd Eastbound Uestbotmd N8 SB EB VB Intersection Totals Tim Period L t Th/llt L t Th/Rt L t Th/Rt L t Th/Rt Total Total Total Total L t Th/Rt Total --------------- I~~DElt~~=f~tL=.iIDD==IILSC=I==~E===¶C=~L=.i=~=t=====I¶P==O=L:=C=C:D~====~=====~==~=======--------------- 5:OO-5:15 pn 5:15-5:30pn 5:30-5:45pn 5:45-6:OOpn 40 23 21 27 15 16 30 15 81 57 44 49 35 29 26 26 46 38 44 51 31 25 30 30 70 61 39 35 52 44 21 19 18 17 28 17 45 34 30 30 33 27 33 33 54 46 23 20 59 43 35 40 33 29 26 21 37 31 27 24 ---*-----------------------.----------------------------*-------------------------------- TOTAL 28 18 ' 5 9 29 45 29 54 35 20 35 31 37 45.30 27.65 30.19 Peorfa & 51st Ave. 01/04/90 INTERSECTION: SURVEY DATE: Leading 3rd Car Northbound Southboud Eastbound Westbound N0 SB EB VB Intersection Totals Time Period L t Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Total Total Total Total L t Th/Rt Total t = = = = = = ~ ~ = ~ ~ ~ ~ a ~ 1 ~ 5 = ~ = ~ 1 ~ = = ~ = ~ ~ n = = ~ ~ = = ~ t = = c ~ n = ~ = ~ ~ ~ ~ n ~ = = a = ~ ~ n = ~ s = = = = 1 = = ~ ~ ~ = = 1 f 4:30-4:45pa 4:45-5:OOpa 5:OO-5:15 pn 5:15-5:30pn 46 58 122 84 228 76 226 118 297 62 350 76 210 211 240 248 61 69 53 54 185 369 185 213 135 145 266 247 274 284 419 434 648 602 578 645 2 329 302 324 783 747 fl42 892 246 433 236 267 318 390 501 461 1271 1408 1298 1456 1589 1798 1799 1917 --------.----------------*-------------------.-------------------*-------------.----------- TOTAL 310 1101 332 909 237 1670 5433 7103 Lesdtng 3rd Car Yolw Peoria L 51st Ave. 01/04/90 IUTERSECTION: SURVEYDATE: 950 791 2473 1411 1241 1187 3264 Northbound Southboud Eastbound Vestbamd NB SB EB WE Intersection Totals Time Period L t Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt t o t a l Total Total Total L t Th/Rt Total ~ = n n ~ a ~ ~ ~ = = n ~ ~ x n a ~ ~ ~ 1 ~ ~ ~ ~ a ~ ~ ~ = ~ ~ n ~ a ~ = = = ~ a ~ ~ n = ~ ~ n = a = ~ = c = c = = ~ ~ = = ~ c = = ~ = ~ ~ n = ~ = ~ 4:30-4:45pn 4:45-5:OOpn 5:OO-5:15pn 5:15-5:30pn 218 252 214 2?7 27 32 61 62 28 44 47 44 156 205 209 207 20 28 27 20 159 240 269 252 28 45 45 38 271 335 3W 407 245 284 275 339 299 580 435 445 17P 268 296 272 184 249 256 251 103 149 180 166 804 1032 1082 1143 907 1181 1262 1307 -----------------------*---------------------.------------------------.-------------*------ TOTAL 182 INTERSECTION: 777 961 163 Peoria L 51st Ave. 01/04/90 SURVEY DATE: 920 95 156 1403 1143 940 1015 1559 596 4061 4657 Leading 3rd Car Delay Per Vehicle (sec/wh) Northboud Southbamd Eastbound Vestbovld NB SB EB WE Intersection Totals Time Period L t Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Total Total Totat Totat L t Th/Rt Total ~ = I = ~ 4:30-4:45 pa 4:45-5:OO pa 5:OO-5:15pn 5:15-5:3Opn ~ 26 27 30 20 ~ ~ 16 13 21 19 P ~ 41 40 20 26 ~ ~ 20 15 17 18 L ~ 46 37 29 41 = ~ 17 23 10 13 ¶ ~ ~ C 72 M 58 27 22 24 88 98 ~ ~ 17 15 23 19 ~ ~ z 20 18 19 ~ ~ 21 25 12 15 ~ ~ ~ 39 29 29 30 I t 46 39 42 42 ~ P S 24 20 18 19 ~ C ~ 26 23 21 22 ---------------------------------------.----------------------------------------------*---- TOTAL 26 17 ' 3 1 18 37 15 76 26 19 20 18 31 42.03 20.07 22.88 ~ D X ~ ~ ~ INTERSECTION: SURVEY DATE: Peoria & 5 l s t Ave. 04/26/90 Lagging Oelay Y o r t h b o u d Southbound Eastbomd Uestboud tJB SB €0 V8 I n t e r s e c t i o n Totals Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Total Totat Total Total L t Th/Rt Total Time Period Ct Rff==C=C=DIIIC115SI=EII===E=I¶=====LfIDS==II==E=I=I==f==I==EE==I=If== 4:30-4:45pn217 4:45-5:0Opa162 5:OO-5:15 pm 104 5:15-5:30pn148 289 343 401 524 319 406 150 255 631 1557 1130 407 3% 297 378 259 129 181 113 506 505 505 672 139 956 140 754 140 1157 140 816 950 744 858 977 726 801 447 633 -- - ----------------- --- 1209 1095 873 894 1039 1297 1090 9S6 934 837 575 656 2602 2236 2713 2695 35% 3073 3288 3351 3002 10246 13248 -------------------.-----.-----------------.--------------.-- TOTAL INTERSECTIOM: SURVEY DATE: 1477 682 3529 559 3683 2188 2607 4211 Peoria L 51st Ave. 04/26/90 4242 Lagging Volw Morthbound Southboud Eastbound Uestbwnd W8 SB €0 UE I n t e r s e c t i o n T o t a l s Tfme Period L t Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Total Total T o t a l Total L t Th/Rt Total ~ ~ ~ C 4:30-4:45pn 4:45-5:OOpn 5:OO-5:15pn 5:15-5:3Opn ~ I ~ 41 57 51 48 232 200 178 225 197 835 l = ~ 35 43 37 44 E E 44 55 52 48 336 352 398 3% C I 258 307 298 305 P I 37 32 32 30 = I E L ~ ~ I I ~ D I 1 ~ ~ 273 257 229 273 371 395 435 439 302 362 350 353 306 285 278 321 157 187 172 170 1095 1112 1120 1216 1252 1299 1292 1386 1059 1032 1640 1367 1190 686 4543 5229 269 253 246 291 ~ ~ 1 ~ ~ D --------------.-------------------.------.--------------. TOTAL INTERSECTION: SURMY DATE: 159 1481 199 1168 131 Peoria 6 51st Ave. 04/26/90 Lagging Delay Per Vehicle (sec/veh) Northbound S o u t h b o d E a s t b o u d Yestboud UB SB €0 VB I n t e r s e c t i o n Totals T i m Period L t Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Total T o t a l T o t a l T o t a l L t Th/Rt T o t a l ~ = ~ C I 430-4:45pn 4:45-5:OOpn 5:OO-5:15pn 5 : l S - 5 3 0 pa ~ 79 43 31 46 S L 19 26 34 35 S = 137 142 61 87 1 1 18 17 11 14 I 1 88 35 52 35 ~ I 55 36 43 48 ~ 56 66 66 70 D ~ 53 45 71 42 ~ O 28 29 33 37 ~ ~ 29 30 15 22 ~ I ~ 60 36 45 46 B 54 47 70 45 ~ I I ~ 89 67 50 58 ~ 36 30 ~ C S 33 42 35 58 36 33.83 58.00 36 -.------------------**--.----.---.----------------------------------*---------------------- TOTAL 48 28 , 1 0 7 15 51 45 64 52 32 24 46' 5365.64 ~ ~ ~ I ~ ~ I INTERSECTION: SURVEY DATE: Time Period Duntap & 35th Ave. 05/03/90 Leading 3rd Car Delay Northbound Southbound Eastbound Vestbound NB SB EB UB Intersection Totals Lt Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Totat Total Totai Total L t Th/Rt Total ------ =S==llf==DI=Lt========.I=~s=E==Ia==e=fI===II====::It===I===I=EZ~II===------====I==::=I====Is======E======== 4:15-4:3Opn 4:30-4:45p 4:45-5:OOpn 5:OO-5:15 pn TOTAL 116 1088 105 120 981 119 112 1461 83 157 2182 70 110 141 135 609 189 127 A 154 116 221 81 252 103 312 1204 565 1101 641 1 5 A 600 2339 520 722 573 679 585 681 722 703 404 1942 496 2303 411 2813 519 3643 2799 3224 4162 1329 2491 1830 10701 12531 237 295 356 441 2346 --------------------.------..-----.-------------------.-. 505 5712 3?7 INTERSECTION: 2117 575 Dunlap & 35th Ave. 05/03/90 SURVEY DATE: Time Period 415 603 490 754 373 2118 6217 2494 Leading 3rd Car Volune Northbound Southbound E a s t b a n d Uestboud NO SB EB VB Intersection Totals Lt Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Total Total Total Totat L t Th/Rt Total I~~~I=~L=III=~~~~===I=OIL~=~~IL:~==II=I=~I:O~~==II~I===I==IEI==========~==========L========~=~===~~=L= 4:15-4:fOpn 450-4:45pn 4:45-5:OOpm 5:OO-5:15 pn TOTAL 53 54 50 49 38 40 45 23 253 258 250 330 48 55 51 69 230 255 239 257 41 50 55 52 335 410 393 413 436 478 420 490 291 298 295 353 278 310 290 326 376 460 448 465 180 199 201 193 1201 1347 1252 1441 1381 1546 1453 1634 773 5241 6014 .-.---.------.---*--------.-----------.-----------------206 1618 146 1091 223 Dunlap & 35th Ave. 05/03/90 INTERSECTIOM: SURVEY DATE: Time Period 383 424 370 441 981 198 1551 1824 1237 1204 1749 Leeding 3 r d Car Delay Per Vehicle (sec/veh) Northboud Southbound Eastbound Uestboud NB SB EB VB Intersection Totals Lt Th/Rt L t Th/Rt L t Th/Rt tt Th/Rt Total Total Total Total L t Th/Rt Total =X==lSSC==l=lSlIOff===S=1SII1~X===il=ID5====D=ttI==I=L===f==f=I~=E======I===DD==============IDtt==3===D== 4:15-4:3Opn 4:30-4:45pn 4:45-5:OOpn 5:OO-5:15pn 33 33 41 45 28 46 25 35 48 43 35 59 74 TOTAL 37 53' 39 34 29 28 34 38 40 41 8 9 14 15 27 35 22 30 14 21 24 22 41 35 56 29 39 12 28 20 72 27 36 29 29 13 14 18 20 15 22 24 23 51 30 17 21 35.51 30.63 34 37 31 40 24 26 34 38 25 27 33 38 31.25 INTERSECTION: SURVEY DATE: Ounlap & 35th Ave. 05/16/90 Leading 1 s t Car Delay Northbound Southbound Eastbound Vestbourd NB SB EB UB Intersection Totals Time Period L t Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Total Total Total Total L t T h / ~ t Total =====us========~==~==crz===============================z=========:===================================== 4:15-4:3Opn 55 944 4:30-4:45 pn 69 472 4:45-5:00 pn 86 1561 5:OO-5:15pn113 956 47 57 58 91 741 55 749 90 757 121 764 96 80 52 201 51 166 MI 278 107 448 999 541 717 1647 740 1067 638 788 806 815 855 135 291 287 374 500 689 2213 2060 3201 2736 2422 2327 3526 3143 1208 10210 11118 209 267 325 407 777 8.47 -----------..--------------------.-------------------.------- TOTAL 323 3931 253 3011 362 Dunlap B 35th Ave. 05/16/90 IUTERSECTION: SURVEY OATE: n 5 270 2543 4254 3264 1087 2813 leading 1st Car Volune Northbound Southbound Eastbound Uestboud NB SB €0 VB Intersection Totals Th/Rt L t Th/Rt tt Th/Rt L t Th/Rt Totnl Total Total Totat L t Th/Rt Total Time Perlod L t t ~ D I T 4:15-4:30pn 4:30-4:45pn 4:45-5:OOpn 5:OO-5:ISpn U ~ 32 35 53 62 I D ~ 371 362 416 377 D ~ 26 24 30 32 O I I 249 263 252 324 ~ 30 51 C D ~ 235 267 253 248 = U 45 40 52 63 I ~ 353 ~ ~ I 403 397 469 439 3 ~ ~ 275 287 282 336 I ~ ~ t ~ = ~ ~ ~ D ~ ~ C 1 398 423 432 444 133 150 201 211 1208 1275 1301 1330 1341 1425 1502 1541 182 1526 112 1088 201 1003 200 1497 1708 1200 1204 1697 695 5114 5809 66 54 383 380 381 265 318 319 302 ~ ~ ~ X ~ f t ~ Z : -----------------.------------------------------------------*- TOTAL INTERSECTION: SURVEY DATE: Dunlap i 35th Ave. 05/16/90 leading 1st Car Delay Per Vehicle (sec/veh) Northbound S o u t h b o d Eastbound Westbound HB SB EB VB Intersection Totals Th/Rt Total Time Period L t Th/Rt L t Th/Rt L t Th/Rt L t Th/Rt Total Total Total Total L t ----- -- ....................... ==1=1=====*=6EL=15=r=---------=-----=--I-------------------------S------- 4:15-4:30pn 4:30-4:45pn 4:45-5:OOpm 5:OO-5:15 pn 26 30 24 27 TOTAL 27 38 20 56. 38 ---------- ---===E----------C=I=3.==E==6=I=LLI== 27 36 29 43 45 43 45 35 28 26 28 27 5 11 10 17 17 19 17 25 19 25 28 29 37 20 53 36 43 42 43 36 8 14 13 19 19 24 27 29 '34 42 27 11 20 25 37 41 14 25 26.0729.95 24 27 24 29 27 24 37 31 27 24 35 31 ---------1---------------..---------.---------.---------------*----------------39 29.48 .Uorthccn f 43rd 05/1/90 IUTERSECT1OH: SURVEY DATE: Pcrtod ~ c a d i n g3rd C o r Avc. Oelay Worthbard Southboud ~t ~ h / ~~tt Th/Rt E-%tbolrd Th/Rt u~stbomd Lt ThfRt Ct W E SfJ €0 VB l n t c r s e c t i ~l o t a t s Total Total Total Total L t ~ h f ~r ott a ( --..---- er--:~=~=~==~=====e=~e=e~PIeE;ir=CDI=CiX==E=L==LEIIE=IS=i:==EE=====EE==IE9===C===~============E======f=------- 4:30-4:45pa 4:45-5:OO pn 5:00-5:15 pa 5:lS-5:'45 pa 99 112 142 413 1192 1031 2766 113 122 88 433 5202 407 60 64 243 135 243 184 186 117 242 I91 479 802 SO5 4'36 116 % 155 987 512 1102 1252 BC6 1143 2181 2928 805 736 2242 267 5116 583s 356 257 331 268 1103 919 1202 747 942 647 2336 514 399 SS8 592 21z 3251 2& 5587 3165 6129 1212 2978 5583 2043 13M5 15608 665 26M mo ~~I-~~~_-.____~-~~~~~~~~----.----.-.--------- TOTAL INrERSECTIOU: Northern & 43rd Aw. SURVEY DATE: OS/l/PO Worthboud Th/Rt Time Period L t Southbounj Lt fh/Rt Leading 3rd Car Votrme Eastbad Lt Th/Rt Vestbound Lt Th/Rt NB SB €8 WE Intersection Totats T o t a l Total Totat Total L t ThfRt Total ~ S P M - e - 5 = ~ l f = I = E = E - ~ = - = l L L i ~ - - L I ~ : f = C I = = ~ I E E = I = f = f = = ~ - 4:30-4:65 pp 4:45-5:OO pn 5:OO-5:lSpn 5:lS-5:45 pm 53 51 58 66 34 37 46 29 SO 353 397 408 414 - - - E - - - ----------__-------J ---- 475 690 496 603 219 211 239 226 333 293 339 304 390 443 437 4 a 181 187 192 207 1236 1250 1319 1390 1417 1437 1511 1597 Leading 3rd b r Delay Per Vehlcle 8 0 5000 4000 3000 ,000 1000 0 AM MID TIME PERIOD LEADING PM LAGGING * Figure F-I. Scottsdale Travel Time Study Cost/Hour * APPENDIX G DALLAS AREA SPECIAL PHASING SEQUENCE TYPICAL F'HASING PHASE A PHASE 0 PHASE C KI -.- Legend d 0--- ) r[ PEASE D novemen:, pro:ected green Kovenent, permitted green no\encn:. yellow chsnqc F i g u r e 1. T y p i c a l P h a s i n g . Source: Reference 14 SFEClkL PHASING PHASE B Figure 2 . Special Phesinc. Source: Reference 14