USF Libraries
USF Digital Collections

Safety evaluation of right turns followed by U-turns as an alternative to direct left turns


Material Information

Safety evaluation of right turns followed by U-turns as an alternative to direct left turns conflict analysis
Physical Description:
Lu, John
Dissanayake, Sunanda
Castillo, Nelson
Williams, Kristine
Florida -- Dept. of Transportation
Center for Urban Transportation Research
Place of Publication:
Tampa, Fla
Publication Date:


Subjects / Keywords:
Automobile driving -- Steering   ( lcsh )
Traffic conflicts   ( lcsh )


Statement of Responsibility:
by John Lu, Sunanda Dissanayake, Nelson Castillo and Kristine Williams ; submitted to Florida Department of Transportation.
General Note:
"October 2001."
General Note:
"Volume II of three reports based on the project 'Methodology to quantify the effects of access management on roadway operations and safety'."

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 50873456
usfldc doi - C01-00088
usfldc handle - c1.88
System ID:

This item is only available as the following downloads:

Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nmm 22 Ka 4500
controlfield tag 008 s2001 flu|||| |||||| eng
datafield ind1 7 ind2 024
subfield code a C01-00088
2 doi
(OCoLC) 50873456
1 100
Lu, John.
0 245
Safety evaluation of right turns followed by U-turns as an alternative to direct left turns :
b conflict analysis /
c by John Lu, Sunanda Dissanayake, Nelson Castillo and Kristine Williams ; submitted to Florida Department of Transportation.
Tampa, Fla. :
Center for Urban Transportation Research,
"October 2001."
"Volume II of three reports based on the project 'Methodology to quantify the effects of access management on roadway operations and safety'."
Automobile driving
x Steering.
Traffic conflicts.
Dissanayake, Sunanda.
Castillo, Nelson.
Williams, Kristine.
Dept. of Transportation.
8 773
t Center for Urban Transportation Research Publications [USF].
4 856


SAFETY EVALUATION OF RIGHT TURNS FOLLOWED BY UTURNS AS AN ALTERNATIVE TO DIRECT LEFT TURNS -CONFLICT ANALYSIS (VOLUME II OF THREE REI'ORTS BASED ON THE PROJECT "METHODOLOGY TO QUANTIFY THE EFFECTS OF ACCESS MANAGEMENT O N ROADWAY OPERATIONS AND SAFETY") By John Lu, Ph.D., P .E. Sunanda Dissanayake, Ph.D. Nelson Castillo Department of Civil and Environmental Engineering and Kristine Williams AlCP Center for Urban Transpo rtation Research University of South Florida Submitted to: Florida Department of T ransportation Traffic Operations Office MS 36 605 Suwannee Street Tallahassee, FL 32399 October 200 I I


ABSTRACT Th is project eva l uate d the safety and operational impacts of two alternative l efttu rn treatmen ts from d riveways/sid e streets. The two trea tments were (I) direct left turns and (2) right turns followed by U-tum s Safety analyses of the alternatives were conducted using two major approaches: traffic c rash data analysis and conflict analysis. Findings related to the t raflic conflict analysis are documented in this report. Two other reports document the crash data and operational analyses. Ten si tes were selected for field data collection where each site experienced one or both of the left tum alternatives from the driveway or side street. Video cameras were set up o n scaffoldings to achieve enough viewing height and all the traffic movements at the selected sites were reco rde d. These videotapes were later rev i ewed and conflict data related to di rect l eft turns or right turns followed by Uturn mo vements were gathered together with correspon ding traffic volumes and conflict se v er iti es Nine different conflict types related t o the left tum moveme nts were considered. The average number of and conflicts per thousand involved vehicles were estimated using the c o llected data. The average number of hou rly confli ct s for direct left turns wa s 6 .35, whereas the corre s ponding v a l ue for right rums U-turns was 4.2. When the results we r e separated by t i me period, the differences were more significant during peak periods. The average number of conflicts per th ousand in v olved vehicles for direct l eft turns and right turns followed by U-turns were 30.2 and 18.7 resp ec ti vel y. A before and after comparison was also conducted at a site tha t u nderwen t a median closure thereby allowi ng only righ t turns followed by U-turos Results showed th a t the to t al average number of confl ict s per hour was reduced by almost 50% by replacing direct left turns w ith right turns followed by U-turns. Confl ict severity fo r the after period was also significantly lower than that for the before period. Several different approaches t o evaluating traffic conflicts resulted in the conclusion tha t the right turns followed by U -t urn mov eme nt was s afe r th an that of direct l eft tum move ment. II


ACKNOWLEDGEMENT The research team would like to thank the Florida Department of Transporta tion for funding this very useful research project with importa nt field applications. The team of panel members that consisted of Vergil St

TABLE O F CONTENTS Page Number ABSTRACT ................... .. .... . ... ........ . .. .... .. . .... . ............. . . ...... .... ... ii ACKNOWLEDGEMENT ................. . ... . ....... .......... ... . . .... ... . .... ........... i i i TABLE OF CONTENTS ......... ...... ........ .. ... .. ........................ ... .................. iv LIST OF TABLES ... ... .... ... ... . .. ....... ...... .... .. . .... .... ... ........ . ....... ......... vi LIST OF FIGURES .. ......... .. ... .. . .... .. ... .......... ....... . .. . .............. .... ....... vii CHAPTER I INTRODUCTION ... . ... ... .... .. .. ... .... .... .... .................. .. ... ... 1 1 1 Background ... ... .... ........... .. .......... ................................ . .... ... I l.2 Outline of the Report ..................... ......... .. .... .. .. ... .. ..... .............. 3 1.3 Sel ect i o n o f the Study S u bjec t ...... .... .. .... ........... ............................ 4 1.4 T h e Selected Researc h Subject. ......... .. .. .. ........................ ............ .. .4 1.5 P r oblem S tatement... . .......... ..... ........ ......... .... .. ........... .. ............ 9 1 .6 Research Objec tives ...... ....... .. . ....... .. .. . . ....... . ......... . ......... 1 1 1.7 Past Stud i es .. . ... ... .... ... . .. ... . .. .... ......... .... ... . ... . ... . . ............ 11 CHAPTE R 2. METHODOLOGY ...... .. .. .. ........... .. .... ... .. .. ............. ... ........ 19 2.1 Sit e Selection .............. ....... .. .... .. ... . . ........ . ... ....... . . .... .. .... .... 1 9 2.2 Sample Sizes . . ... ............ . .. ... ............ ..... .. ... . ........... ........... 20 2.3 Types of Conflicts Studied .... .. .. .. ...... ................................ .. .. ........ 21 2.4 Identifi ca tion of Traffic Conflicts ....... ... .. .. .......... .. .. .. .. ... .. . ........... 27 2.5 Data Reduction P rocedur e ................ .. .. .... ... ............................... .31 2.6 The Stud ent' s 1 Test ...... ........... .. .. ........ .. .................. .............. ... 3 2 2 7 One-wa y Analysis of Variance Test ...... .. ................ ...... .................. 33 CHAPTER 3. DATA COLLECTION .......................................................... 3 5 3 I Descript ion of Study .. .. ..... ................................. .. .. ....... 35 3.2 Data Collection Equipment ...................... .... .... ............. ............... 4 0 3.3 Field Procedure .... . ............. ...... ....... . . .... .. .... . .. .... . .......... .41 3 4 Data Reduction .......... ........... . . .... . .... .... . ..... ............. ... ....... .41 I\'


CHAPTER 4. DATA ANALYS I S ... .. ...... ............... .............................. .. .43 4 1 Descriptive Analysis ......... ... ... .... .......... .... ... ... ... ... .. ............. .. .43 4 .2 Conflict Rates ............... ........ .... ........ .. . .... .......... .......... .......... 54 4.3 Severity Analysis ...... .... .... .. ............. ....... .. . ........... ................ .56 4.4 Swn mary ......... ... .... ................... .......... ................ .............. ... 65 CHPTER 5. CASE STUDY: BEFORE AND AFTER COMPARISON ............. .. .... 66 5.1 Descriptive Analysis .. ........................ .... .. .................... ............ 66 5.2 Conflic t Rates ........... ...... ... .. ... .. . ... .. .. . .. .......... .... .. ........... 68 5.3 Severity Analysis ......... ... ... ... ....... ........ ... .. ... ............ . .. .......... 75 5.4 Summary ............. ... . ........... .... ........... ... . ........ ..................... 7 8 CHAPTER 6. SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS .. ...... .. 80 6.1 Summary ...................................... .. . ............. ........ ................. 80 6 2 Conclusi ons ........ .................. .. .. . .. .. ..... .. .. .......... .. .. .............. 81 6.3 Recommendat ions .... .. ... .. . ...... ............... .. . .. ............ ....... ... .. 82 REFERENCES ........... ....... .................... . .. ............ ............. ............. .. 83 v


Table Tabl e 1.1 Tabl e 1.2 Tabl e 2 1 Table 3.1 Table 4. 1 Table 4.2 Table 4.3 T able 4.4 Table 4.5 Table 4.6 Table4.7 Table4.8 Table 4.9 Table 4. 10 T able 5 1 Table 5.2 Table 5.3 Table 5.4 Table 5.5 Table 5.6 LIST OF TABLES Page N o. Vehicles Involved in Single and Two-Vehicle Crashes by Vehicle Maneu v er. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . 6 TTC and ROC Values ....... .. .... ... .. ................... . . .. .. ..... .. ..... I 6 Definitions of Different Conflict Rates .. ........................... . .. ..... 32 Dcscdption of the Selected Sites ........... .............. ...... : ................ 36 Verification of Sample Size for RTUT Movement .. . .. . . ................ .43 Verification of Sample S i ze for DLT Movement ........... . .. . . . ....... .44 Summary of th e Total Number of Observations .......................... ... .45 A v erage N umber of Analysis Used for Analysis ... ................. ......... .46 Average Da il y N umber of Conflicts ............. ... ............................. 47 Number of Conflicts per Thousand Involved vehicles ....................... .57 Time to Collision and Risk of Colli sion Scores ....... ....................... 57 Sev e rity Percen t iles by Movement and Con flict Type ....................... 61 ANOVA Test Results to Test Severity Differen c es Be tween RTUT and DLT Maneuvers Considering ROC Score o nly .................. 64 ANOVA T est Results to Test Severity Differences Between RTUT and DLT Maneuvers Considering Both ROC and TTC Scores .. ... 64 Number o f Conflicts per Hour for DLT Movemen t s during Befo r e Period .............. ........................................................ 68 N umber of Conflicts per Hour for RTUT Movemen t s during Before and After Periods .. .................... .... ......................... .. .. 69 Comparison of Total Number of Conflicts per Hour During Before and After Time Periods .............. .................................. 71 Average Number of Conflicts per Thousand Invo l ved Vehicles, Me t hod 1 ........................................................ .................. 72 Average N umb er of Confl i cts per Thousand Invo l ved Vehicles Method 2 ... ... ........... .. .......................... . .. ................... ...... 73 Results of the F Test to Check the Variances of Conflict Rates D uring Before and After Periods .................................. ............. 7 4 Vl


Table 5 7 Table 5 8 Table 5.9 Table 5.10 Table 5.11 Table 5.12 Results of t h e 1 test Comparing Differences Bet ween Conflict Rates During Before and After Periods .................................. .. .. 75 Percentiles of Sever ity for DLT and RTUT Maneuvers fo r Before Period ............... .. ... ....... . .. .. .. .. .. ................... ...... .. 75 Per cent i les of Severity for RTUT Maneuvers for After Period ........... .. 76 ANOV A Test Results to Test Severity Differences Between RTUT and DL T Movements Considering ROC Score ....................... 77 A NOVA Test Results to Test Severity Differences Between RTUT and DL T Movements by C o nsidering both ROC and TTC Scores, After I m provement.. .............................................. 78 ANOV A Test Res u lts to Test Severity Differences for RTUT Before and Afte r Improvement.. ......................... ... .. . .... 79 vu


Figure figure J.J F i gure 1.2 Figure 1.3 Figure 1.4 Figure 2.1 Figure 2 2 Figure 2.3 Figure 2.4 Figure 2.5 Figure 2.6 Figure 2.7 Figure 2. 8 F igure2.9 F igure 2.10 Figure 2. 1 1 Figure 3 1 Figu re 3.2 Figure 3.3 Figure 4.1 Figure 4 .2 Figure 4.3 Figure4.4 Figure 4.5 F igure 4.6 Figure 4 7 LIST OF FIGURES Page No. Contlict Points at Four -l eg Intersections ........................................ 5 Direct Lefttum Movement ... ..................................................... 7 Safety Conc erns Rela t ed t o Direct Lefttum Movements .. ................... 8 Description of Right Turn Followed by U -tu rn at a Median Openiog ...... 9 Conflict Type CJ .................................. .......... ............... ........ 23 Conflict Type C2.. .... ..... ...... ... . .............. .......... . ............... 23 Conflict Typ e C3 : ........... . ... . ........ ... ....... . ............. .............. 24 Confl ict Type C4 ............... .... ... ..... ......... ............... . ............ 24 Confl ic t Type C2U-T ................................. ............................ 25 Conflict Type CS ..... ............. ................ ......... .......... ... .. .. ... 25 Conflic t Type C6 .................................................................. 26 Conflict Type C7 ....... ... ......... .... ........................ . ................ 26 Conflic t Type C8 ................................................ .. ............... 27 The Form Used for Recording Traffic Conflicts .............................. 29 Flow Chart Describing Conflict Identification and Data Required by Observers ............................................................ 30 Location of Selected Sites ......... .................................. . .......... 35 General Layout of Sites I, 3-After Im provement and 8 ..................... 37 General Layout of Sites 2, 3-Before Improvement 4, 5, 6 and 7 ........... 37 Number of Daily Con flic ts by Type, Site I. ............ ......... ............ .48 Number of Daily Conflicts by Type Site 2 ................................... .48 Number of Daily Conflicts by Type, Site 4 .................................... 50 Number of Daily Conflicts by Type, Site 5 .................................... 50 Number of Daily Conflicts by Type, Site 6 .................................... 5 1 Number of Daily Conflicts by Type, S i te 7 ................................... 51 Number of Daily Conflicts by Type, Site 8 .................................... 52 viii


Figure 4.8 Average Number of Daily Conflicts by Type, RTU T Movement ....... . 53 Figure 4.9 Average Number of Daily Conflicts by Ty pe, DLT Movement ............ 53 Figure 4.10 Conflicts by Time Period, RTUT Movement. ..... ...................... . . 55 Figure 4 1 1 Conflicts by Time P eri od, DLT Move m ent ............ ...... ....... .. ....... .55 F igure 4.12 Comparison Average Hourly Conflicts Between RTUT and DL T Movements by Tim e P eriod ............................. .. . .... . . .. ... 56 Figure 4.13 Distribution of Severity, Conflict Type C l .......... .......... . ........... 58 Figure 4.14 Distribution of Severity, Conflict Type C2 ............ ........ .... .......... 58 Figure 4 1 5 Distribution of Severity, Conflic t Type C3 .......... .......................... 59 Figure 4 1 6 Distribution of S e verity Co n fl i ct Type C 4 ........... ............... ......... 59 Figure 4.17 Distrib ut ion of Severity Confl i ct Type C2U T ................ .. .. ........... 60 Figure 4.18 Distribution of Sever i ty Conflict Type CS ................ .. .......... ....... 60 Figure 4. J 9 Distribut i on of Severity, Conflict Type C8 ..................... .............. 61 Figure 4.20 Comparison of Overall Severity of C onflict s Considering ROC Scores ................................ .................. ...................... 63 Figure 4 .21 Comparis on of Overall Severity o f Conflicts Considering ROC and TTC Scores .. .......................................................... 63 Figure 5.1 Number of Daily Conflicts by Type Befo r e the Improvem e nt.. .......... . 67 Figure 5.2 Numbe r of D aily Confli cts by Type After the Improvement ............... 67 F igure 5.3 Conflicts per H our of Observation by Time PeriodRTUT Movements Before Improv ements ............... ........................ ....... 69 Figure 5.4 Co n flicts per Hour of Observation by Time Period DL T Movements Before Improvements .. .... ............. ................ . ....... 70 Fig u re 5.5 Conflicts per Hour ofObse. rvation by T ime Period RTUT Movements After Improvements ....................................... ......... 70 IX


I. INTRODUCTION 1.1 Background As the na tion's roadway system becO!JleS more congested and the number of vehicular crashes increases, the importance o f access managem e nt is increasing. The management of access has been identified as one of the most critical elements in roadway planning and de s ign (/ ). Access management has been defined as the process of managing access to land development whi l e simultaneously preserving the safety and efficienc y of the surrounding roadway system (2). It helps achiev e the necessary balance between traffic movement and property access by careful control of the location, type, and design of driveways and s treet iotersections. This is accomplished by classifying highways with respect to the level of access and mobility they are expected to provide, and theo, identifying and applying the most effective techniques to preserve that function. The impacts of poten tial techniques on traffic performance and safety are important considerations when deciding which technique to implement. Access management deals with the control and regulation o f the spacing and design of medians, median openings, driveways freeway interchanges and traffic sig n als. Typical access management measures cover the type and design of medians and median openings; the location and spacing of intersect.ions; the spacing and design o f interchanges; and location spacing, and design of driveways and street connections. The location design and operation of driveways play a significant role in access management. AASHTO Green Book "A Policy on the Geometric Design of Highways and Streets", indicates that "Driveways are, in fact, at-gr-dde intersection s and should be designed consistent with the intended use. The number of crashes is disproportionately higher at driveways than at other intersections; thus their design and locat i on merit special consideration." (3) In the "Access Management, Location and D esign Participant Notebook" the potential access management techniques are categorized into six groups (4). These categories are related to traffic operational actions, which serve to minimize the fre quency and severity of traffic conflicts. T he six categories are: l


I) Limit nwnber of conflict points: These techniques directly reduce the frequency of either basic conflicts or encroachment conflicts, or reduce the area of conflict at some or all driveways on the highway by limiting or preven t ing certain types of maneuv ers 2) Separate conflict areas: T hese techniques either reduce the number of driveways or directly i ncrease th e spacing between driveways and intersections. They indirectl y reduce the frequency of conflicts by separat ing turning vehicles at adjacent access points and by increasing the decision-processing time for the through driver between successive conflicts with drivewa y vehicles at successive driveways. 3) Remove turning vehicles fwm through traffic Janes: These techniques directly reduce both the frequency and severity of conflicts by provid ing separate paths and storage areas for turning vehicles. 4) Reduce the number of turning movements: The provision of cross-circulation between adjacent properties and the provision of service roads allows inter-site movemen t without reemry to the abutting major roadway. The elimination of short distance slow movements reduces the n um ber of conflicts along the major roadway 5) Impro ve driveway operations: These techni ques allow drive rs to maneuver from and to the major roadway more efficiently and safely. 6) Improve roadway operations: These techniques are primarily of a policy nature, which are intended to preserve th e functional integrity of th e roadway. Ditlerent standards are commonly app lied depending on the category of the road I n general, the benefits of access management measurements can be summarized as: improved safery, improved traffic flow and fuel economy increased capacity, and reduced delay and vehicle emissions. Improved safety is one of tbe most important benefits of proper access management. The safery bene fits of access management techniques have been attributed to reduct io n in traffic conflict points improved access design, and larger driver response time to potential conflicts. Various research efforts have evaluated the impacts of access management on roadway safety. The "Access Management, L ocation and Design, Participant Notebook suggests that effective access management can reduce crashes by as m uch as 50%, increase capacity by 23-45%, and reduce traveltime and delay as much as 40-60% (4). In a study 2


of the statistical relationship between vehicular crashes and highway access, conducted for the Minnesota Department of Transportation, the results from two approaches, a comparison of crash rates using a random sample of roadways from the State's highway system and a before-and-after comparison of crashes, suggested a strong and statis t ically sound re l ationship between level of access and crash rates (5) It showed that crash rates reduced with improvements to median opening spacing in both rural and urban roadway categories B01meson and McCoy concluded that crash rates on facilities with non traversable medians are lower than that of facilities with center two-way left turn l anes (TWLTL) (6). These studies provide important infonnat i on on various access management methods and techniques. However, q u estions still remain surrounding the effects of specific access management treatments on roadway safety and operations. Some of t h ese concerns relate to the safety impac t s of U-tum movements at median openings, the effec t of medians on intersection capacity the safety i mpacts of continuous right-tum lanes and the effect of medians on side street operations. Other questions relate to median and driveway design practices such as right -in right-out only designs, and appropriate driveway channelization measures. Some of t he se questions remain u nexplored e i ther because quantification of some treatments is difficult or because not enough data are availab l e for the evaluation of alternative treatments. There fore more research is needed in order to evaluate the traffic operational and safety impacts of these techniques. 1.2 Outline of the R e port This report on the confl ict analysis of direc t left turns versus U -tums consists of six chapters. Chapter I provides an overview of the research project including a summary of the pas t studies in this subject area. Chapter 2 describes the methodology followed in the research including a detailed description o f the types of conflicts and basic concep t s such as minimum required sample sizes, confl ict rates, and statistical tests Chap ter 3 presents the field data collection procedure in which the details of the sites selected for data collection, equipment used, field procedure, and data reduction procedure is provided. Chapter 4 contains the analysis of the data collec t ed for both direct left tum and right tum 3


followed by U-tum movements which includes number of conflicts, conflic t rates, and severity of conflicts. Chapter 5 presents the safety impac t s of converting a full m edian opening to directional media n opening based on a conflict analysis using before and after comparison. The final chapter Chapter 6 provides th e summary conclusions and recommendations of this study. 1.3 Selection of the Study Subject With the intention of identifying the t eclmique that most needed evaluation, a number of previous studies regarding access management techn i ques were reviewed. Current state and national literature reviewed included but not limited to Transportation Research Board (TRB) publications, proceedings of the TRB Natio n al Access Management Conferences, reports from the National Cooperative Highway Research Program (NCHRP), publication s by AASHTO, Institute of Transportation Engineers (ITE) recommended practices and the ASCE Journal of Transportation Engineering. Jn addition current rules, regulations, standards, and practices in Florida were reviewed. Based on the literature review the project team's experiences, and FDOT review, the subject selected for analysis was the right tum followed by U-tum as an alternative to a direct left tum from a driveway or side street. The main reasons for selecting this subject were: I) L ittle documentation of quantified results and conclusions regarding this subject are available although the impact of U-tums on safety and operations has been identified as one of the important issues in access management. 2) lt is feasible to q uantify the safety and operational impacts of these alternatives. Both crash data and po ten tial sites for case studies are available. 3) The results of the traffic operational and safety analysis can assist agencies like FOOT with decisions relative to insta lling medians or closing median openings. 1.4 The Selected Research Subject The safety impacts of right turns followed by U-tums as an alternative to direct l eft turns from driveways or side stree ts could be explained in relation to their effect on the number 4


of conflict points. A conflict poi n t is defmed as t h e point at which two traffic movements intersect each o ther For exampl e Figure 1.1 illustrates that th ere are 3 2 co n flict points at a four-leg intersection with a full median ope ning, whereas a directional median opening experiences only 8 conflict points 32 Conflict Points ) ( 8 Confl icts Figure 1.1 Conflict P oints at Fourleg Intersec t ion s Also, it can be seen that left-turning movemen t s either from a cross-street or left -tum bay generate most of the conflict points. Thus left tum maneuvers are of special consideration whe n eva l uating eithe r signalized or uns ignaliz ed intersections. Thi s is s u pported by th e fact tbat approximately te n and e l even percent of the total n umber of 5


crashes that occurred in 1998 and 1999, respectively, involved driv ers turning to the left, as shown in Table 1 1 De spite several research studies co nducted regarding l eft tu m movements, problems with left turn movements s t ill rema in. However little research has been performed on the effect of l eft tu'!' maneuvers from driveways or side streets. Crash rates at driveways are higher than that of other intersections {3) Furthermore, more tha n two thirds of all driveway-related crashes i nvolve left turning veh i cles (7) Table 1.1 Vehicles In volved in Single and Two -V ehicle Crashes by Vehicle Maneuver 1998 1999 Vehicle Maneuver Number % Number % (Thou sands) (Thousands) Going Straight 5,783 57.50 5,13 1 51.60 Turning Left 1,030 1 0.20 1 ,071 10.80 Stopped in Tr affic Lane 991 9.80 1,096 11.00 Tuming Right 365 3.60 360 3.60 Slowed in Traffic Lane 4 62 4 .60 495 5 .00 Merging/Changing 314 3.10 375 3.80 N egotiating Cune 139 1.40 374 3 .80 Backing Up 133 1.30 235 2.40 Pas sing O ther Vehicle 134 1.30 142 1.40 Starting in Traffic Lane 130 1.30 264 2 .7 0 Leaving Parking Space 55 0 50 66 0.70 Maki ng U Tum 56 0.60 54 0.50 Entering Parking Space 23 0 .20 25 0 .3 0 Disabled Ill Traffic 12 0.10 15 0.20 Other Maneuver 4 3 9 4 .40 241 2 40 Total * 10,065 100 9,942 100 .00 . So u rce : Nattonal H tg hway T raffic Safety Admm t stratton (2001). For a better un derstand ing of the safety problems re late d to l eft-t u rning movements from a driveway or a side street, a brief description is provid e d here F igure 1 2 illu strates a driver who wants to turn to the left from a driveway that is located be tween two signalized intersections According to the behavior of the driver there are three scenarios that might occur. First, if the drivers that are not aggressive will wait for a suitable gap 6


before crossing the arterial which in most cases would be selected after the p l atoo n of the upstream signal clears The second scenario occurs when a driver exposed to a long delay while waiting for an acceptable gap decides to turn righ t and make a U rum at the next median opening. The third scenario occurs when a driver experiences a long delay in turning left, in which case the driver behavior becomes impatient and aggressive, tending to accept shorter gaps than are advisable. In such situations, the driver is willing to take "the risk of crossing the stree t which could result in a conflict or even a collision. This is due to the fact that overly short gaps can cause high-speed differentials and turbu l ence in the through traffic stream because the vehicles in the through traffic stream may be required to decelerate or attempt to change lanes (8). Once the driver proceeds with the direct left turn movement, be/she must be aware of other vehicles interacting at the same time. Full median openings allow several movements at the same time therefore traffic conflictS can be created not only because of tluough traffic vehicles, but also because of vehicles turning into and out of the same driveway. If other vehicles do not interfere with the egress of the vehicle then the driver crosses the arterial and enters into the median storage area There again the driver must find a suitable gap in order to merge into the inner lane and accelerate to the mean speed of the oncoming through traffic. --------------------tCII -------------------------Figure 1.2 Direct Left -tum Movement. 7


The ma jor prob l ems that could be encoun t ered during the direct left tum movement arc illustrated in Figure 1.3. First (a), a vehicle departing from a driveway at a v e ry low speed creates a speed d iffer ential p r oblem with oncomi n g through traffic. Second, several cars may some t imes be observed in a median storage area making it possible to find l eft turning vehicles waiting for a gap with p art of the ve h icle encroaching on the inner Jan e of oncoming throug h traffic, as shov,n in Figure 1.3 (b). Another prob lem i s that drivers waiting for a gap on the storage area may have s ight obstructions. Sight obstructions cou l d be due to the left tumin volume or due to other v ehi c les located in the storage area, as shown i n Figure 1 3 (c). The lack of sigh t dis t ance is dangerous because drivers cannot clearly see approaching vehicles so that they may merge onto the arter i a l by t r i al and e rror" or by waiti n g until the sig h t is completely clear as shown i n Figure 1.3 (d). _________ fXl ___ mo_ _ __ ____ JXI ---------------------------------e-------& ---------_____ ..,.. .... __ -------& a !I b !i --------_ __ _ _ __ J11 ______ Limited Sight ---------------------------------------------------------------------c & !a & d Figure 1.3 Sa fet y Concerns Re l ated to Direct Left Tum Movements. 8


To overcome the safety problems generated by left tum vehicles, different treatments have been proposed, such as median U-turns, jug handles and recently by NCHRP Report 420, "Jmpacrs of Access Managemenr Techniques" the right tum followed by U tum as alternatives for direct left turns from a driveway (9). The lat!er consists of vehicles turning to the right and then changing lan es until they get to the next m edi .an opening, where they can mak e a U-turn. This alternative is attractive since it docs not require a major finan c i al investment such as roadway reconstruction or right of way acquisition. Another advantage is that the effect of U-tum movements on left tum lanes at signalized imersections could be avoided when proper intcnn ediate tum lanes are provided. Preliminary results from a study performed by the North Carolina State University suggested that the reduction in sat uration !low of left turn lanes is I 0 percent when U turns are between 65 and 85 percent of the volume in the left-tum lane and the corresponding reduction is 20 percent when U-tums exceed 85% (10). The procedure followed by vehicles turning to the right and then making a U-tum at a median opening is illustrated in Figure 1.4. It shows a driver waiting at a driveway until t here is a sui table gap for turning to the right and merge onto the arterial. Onc e the driver enters the arterial, available gaps are evaluated so that he/she can change lanes until the median opening is reached. If there is proper coordination with the s ig nal located ups tream of the driveway, it is sometimes, possib le for the arriving vehicle to find an adequate gap to mak e the U-tum without waiting. Otherwise, the driver has to wait at the U-tum bay. 1.5 Problem Statement Current regulations of the Florida Department o f Transportation (FOOT) require all highways with a design speed greater than 40 mph to have restrictive medians. FOOT also designs and constructs some new unsignalized median openings as directional openings. In addition, FOOT may close an existing unsignalized median opening or convert it to left-tum or U-tum only. Hence, the purpose of this project is to evaluate the safety and operational impacts ofU-turns on th e State highway system. 9


--------------------------Oil FIXI Figure 1.4 Description of Right Tum Followed by U-tum a t a Median Opening. Because right-tum followed by U-Tum movements require drivers to weave a certain distance and then evaluate available gaps for making a U-tum maneuver some drivers view this procedure as "Wlsafe". A survey conducted by TEl Engineers and P la nners found that o ne of th e most common complaints was that U-tum movements would result in safety problems (1 J ). On the other hand, the same survey found that 57 percent of the respondents did not feel inconvenienced by U-tums. Therefore it is ne c essary to examine the relative safety of right tum!U-tums compared to di rec t left-turns. Although, theoretical reduction of conflict points from 32 to 8 suggests a decrease of approximately 70 percent, th is value does not really quantify tb.e reduction in the actual number of con flicts or its effe ct on traffic safety. Moreover, there is little information on the safety and operational effects of median closures and few gu id elines for practitioners relative to closing median openings The National Highway Institute Course No. 15255 has general descriptions sugges ting when to implement a conflict reduction technique (4) For example installing a non-traversable median is recommended when undivided highways have a pattern of left-tum cras hes. However more detailed and quantified results regarding the safety implications of left turn alternatives should be available to transportation engineers and planners to assist them in the decisio n making process. 10


Better information and methods for evaluating safety and operational impacts of median changes are also important when interacting with the general public. Gwynn noted that a frequent public c omp laint was that the agencies do not have adequate supporting reasons for implementing median c banges (II). The general public has a "show me" attitude that is difficult to satisfy unless there is sufficient documentatio n, especially quantifiable doc umentation of the basis for the median changes. 1.6 Research Objectives The goal o f this stu dy was to quantify the safety impact of vehicles making right turns followed by U-turns, and compare it to that of direct left tum movements from drive ways. The safety evaluation was divided into two parts. The first part consisted of a cross comparison of crash rates for several Florida sites The second part consisted of conducting a traffic conflict study to evaluate and compare the implications of these mane uvers The latter is described in this report. Accordingly, the objectives of this study were: I) To estimate the average number of traffi c conflicts for bo th direct left turn and right tum followed by U-tu m maneuvers, 2) To estimate the average conflict ra tes for each of the two left turn ing a lt ernatives from driveways and to compare these values, 3) To evaluate the severity of conflicts generated by direct left tum and right tum followed by U -tum maneuvers and to compare the severities. 4) To compare the safety effects of replacing direct left turns with right tum followed by U-tum movements by conducting a case study using befor e and after analysis. 1. 7 Past Studies 1.7.1 General There were no previous reports or articles regarding the safety evaluation of left tum alternatives from driveways or side streets by using a traffic conflict study. However there have been sevei:al past studies tha t used traffic conflicts at signalized and II


unsignalized intersections as a tool for evaluating the relatio nship between crashes and conflicts, the relationship between conflicts and volumes and ranking of hazardous intersections. Some of them are briefly mentioned here. Traffic conflic t studies began with Perkins and General Motors in 1967 when they wer e analyzing the involvement of OM cars in crashes. In their study, they defined typical incidents that could be related to a specific crash type at intersections A traffic confl ict was defined as an evasive action taken by a driver in order to avoid a collision (I 2). The procedure followed in that research was known as the Traffic Conflict Technique (13. 14) Sev e ral studies followed th e General Motor's study b u t it was only in 19 77 that an accepted definition of a traffi c c onflict was given. This defmition was A conflict is an observable situation in which two or more road users approach each other in space and time t o such an extent that there is a risk of collision if their movements remain unchanged" (15). A more general definition was found i n Parker et al, which defined a traffic conflict as "an ev ent involving two or more road users in which the action of one user causes the other to make an evasive maneuver to avoid a collision" (1 3, I 4). Traffic conflict studies hav e been used for various purposes as stated by Ha uer: "Traffic co nflic ts technique is applicable to a variety of situations. It can he used to measure t he effectiveness of devices, layouts, design, and procedures (16). 1.7.2 Conflicts and Crashes Traditio nally most traffic safety problems are detected by collecting and analyzing the history of crashes at a particular site. However, several problems have been documented with analyzing sites based upon crash records (17. 18, I 9). The first problem is with the reliability of police reported crash records because the number of crashes tends to be under-rep orted. Second, crash databases are subject to human error during the input of data. Therefore code errors, reporting errors and even interpretation errors may appear (20) Third, the waiting time for obtaining an adequate sample size may be too long (19). This waiting time could have serious impli cations because several collisions may occur before any correction could be made to treat or correct a deficiency (21. 22). Other concerns are the randomness of crashes and the human factors associated with them 12


Crashes that involve human factors may not provide enough details on geometric conditions that might have caused the crash {12). Alte rnat ive ly, traffic conflict studies have been accepted for several reasons. F i rst, d ata can be collected within a short period of time so that an enginee r does not have t o wait for the occurrence of several crashe s to improve the conditions of a site (1 3, 14, 21. 23, 24) Second the effectiveness of a treatment can be evaluated within a s hort period and if this fa i ls to correct the problem then the countenneaswe can be changed again in a very short tim e (13, 14). Third traffic conflict s include human factors because the behavior of drivers can directly be observed in the field (1 2). Fourth, traffic conflict studies can be used with or without crash data since each type of conflict is associated with a particular type of crash (13 14). Finally, traffic conflict data provides information about tr affic volumes, routine conflicts, moderate conflicts erratic maneuvers, severe con flicts or nca r-miss crashes, and other minor crashes while crash data can only give information on property damage and injury severity (19). Although the traffic conflict techn ique has been used widely it has als o created controversy especially when determining the relationship between conflicts and crashes. Some r esearchers have found a significant relationship between conflicts and crashes. Migletz et aJ found that crashes and conflicts were significant at the l 0 percent level of significance and Salman and Almaita found that the relationsh ip was linear and statistically significant (18, 25). In contrast, other researchers have found that the relationship is not very strong. Engel fowtd that the correlation between conflicts and crashes was poor, although he found that by separat ing types of conflicts and th eir correspo nding crashes the correlation could be inc reased ( 26). Brown concluded that the relationship betw een tota l crashes and conflicts was poor, but if data were segregated by types of conflicts and crashes the relationship wou ld have been significant (12). Although former research is contradictory Glauz et al stated, "the proper use of conflicts is to estimate an expected rate of crashes as opposed to predicting the actual number that might occur in a particular year'' (26). Henc e, traffic conflicts can be used as a surrogate of crashes. 13


1.7 3 Conflicts and Traffic Volume The relationship between traffic volumes and conflicts has also been controversial. In a study conducted by Migletz et al., different t ypes of volumes were defined acco r ding t o the type of conflict being studied (26)'. For opposing left tum conflicts the volume was defined as the square root of th e product of the l eft tum volume and opposing through vo l u m e stunme d over two approaches at unsignalized intersections Through cross-traffic conflicts were related to the through cross traffic volume, which was defined as the square root of the product of through cross traffic from right (or lefi) volwne w ith th e through vol ume summed over the four approaches at both signalized and unsignalized intersections. Same direction conflicts were related to the same directio n volume which wa s defined as sum of the volumes of all the ap p roach e s. Salman and Almai t a worked on the det ermination of the re l ationship between conflic ts and v olumes at three-leg unsignalized intersections in which the number of confl i cts per hour was compared with two types of volwnes {18) One volume was defined as the summation of all th e volumes entering the in t ersec t ion, and the other o ne was defined as the square root of th e p r oduct of th e volwnes that generated conflicts. Linear regre s s i on models were used for correlating same direction l eft tum, right turn, and s am e direct i on conflicts with volumes. It was found that the correlation factor between conflicts and the square roo t of t h e product betwee n vol umes was l arger than that of summation of volumes Katamine studied diff e r en t volume defi n itions for co n flicts at unsignalizcd intersections using data collected at 1 5 four leg unsignalized intersectio n s (23) Traffic volumes cons i s ted of t hr o u gh traffic, right tum, l eft tum, and U-tum vo l umes These vol u mes were then combined into thirteen diff e re n t definitio n s and related w i th eleven types of c onflicts It was found that the total entering volume, which corresponds to the summat i o n of all the volumes at the intersection, was sign ific ant ly correlated to most of the conflict types, although u sing the to t al volwne m a y not ex p lain why different conflicts occur at the same intersection. 1 4


Sayed et at. deve lope d a traffic conflict simulation for unsignaliz.ed intersections by using a microscopic model i n which only da t a corresponding to veh i cles involved in severe conflicts were recorded for further e va l ua t ion (20). It also i n volved the c o ncept of gap acceptance in orde r to evaluate the behavior of different types of d r ivers. This model randomly generat e d vehicles w it h differ e n t characteris t ics and the driver assigned to that vehicle behaved according to a mini m um gap acceptan c e capabi lit ies based on the sex and age of the drive r. A conflict was recorded if the dis t ance between the involved vehicles divided by the mean speed was less than 1.5 seconds The sim ulat ion software used data from unsignalized in t ersections and produced a set of curves for differ ent types of roads Sayed stated, i f o nly low traffic vo l umes are considered (vol u mes less than the warrants for a tratlic signal) conflicts may appear to be proportional to t he square root of the conflicting volumes. The large increase in simu l ated conflicts at h igh tratlic volumes is probably the resul t of allowing drivers i n the mode l to accept shorter gaps as their delay increases (20) 1.7.4 Confli c t Severity In the past s t udies, sev e ri t y of traffic confl i cts have been measured and evaluated by using both objective and subjective methods. Objective methods rely on physica l propert i es such as time, distan c e, and speed w h ile s u bjective mea.'mres rely on human observers to record the perceived risk at the moment i n w h ich the conflict occurred. Among the objective methods of measurement, Hayward developed the concept of Time to Collision (TIC) to differentiate severe near misses from near misses that were no t severe by usi n g phys i cal properties such as dis t ance and time {28). TIC is defined as the time required f or two vehicles t o collide if they con t inue a t their present speeds on t he same pat h s Hayward defined the minimum TTC as the perception and reaction time of drivers b ecause fa iling to rea c t t o a n event wou ld result in a collision. Near misses were recorded by usi ng a video camera and used for determi n ing TIC values for rear-end, lane change, cutoff, broadside, and rig h t of way cases. I t was determined that the mean TIC value was 1.46 seconds and that near misses occ u rred when the TIC was less than 1 0 second. 1 5


Other objective mea s ureme n t methods have a l so been used for evaluating the severity o f confl i c t s Allen et al. m ade a complete description of other objective mea sure ments i nclu ding the co n cept o f Encroachme n t Time (ET) (29). ET was defined as the time during which a left turning vehicle infriogcs upon the path of an oncoming through traffic ve hi cle. Another method was using th e gap t i me (GT) conc ept, whi ch correspo n ds t o the time at which the offe nded vehicle was expected t o arrive at th e conflict point m inus the time at which the off ending vehicl e cleared the path. Anot her method was "zonal braking technique", in whi ch zones where confl ic ts with a pr edetem 1 ined 11'C values could be observed. Once a conflict occurs in a given zone, its corresponding TTC was assoc ia ted to the risk of the maneuver assuming that th e speed did not differ grea tly. A method based on the Deceleration Rate (DR} has also been used, although DR greatly differs among drivers Another method was the Post Encroachment Time (PET}, which was the difference betwee n the t ime at which the offending vehicle clears the path o f an oncoming vehicle, and the time at wh ich the offended veh ic le arrives at the potentia l poin t of collision. Sayed et a! co n si dered both object ive and subjec tive methods for evaluating the risk o f conflicts because if only objective methods were used, they could overestimate the risk factor (20) Thus, it was recommended t o combine these two for ob t aining a reasonable risk value. For obj ective measurements th e study defined thr ee "ranking zones in which TIC woul d vary from 0 to I second I to 1.5 seconds, and more than 1.5 seconds For the subj e ctive e v aluation observers assigned a val ue known as Risk Of Collision (ROC) based on the sever i ty of me risk perceived while collect i ng data in the field. The TTC and ROC values had equal weigh t for low, modera te and high risk conflicts, as shov,n in Tab le 1.2 Tab le 1.2 TIC and ROC Score Values. TIC and ROC Scores TIC (sec onds ) ROC I 1.51 2.00 Low Risk 2 1.00-1.50 Medium Risk 3 0.00-0.99 High Risk 16


Katamine evaluated the relationship between conflic t s and secondary conflicts at 15 unsignalized intersections based upon a severity grading system created by Glauz and Migletz (23 27). The first grade corresponded to a conflict that occurs when a vehicle appl.ies the b r akes in a precautionar y way when another v ehicle merges onto the arterial and tries to move into another lane The second grade occurred when drivers have enough time to apply brakes or make evasive maneuvers. A third grade conflict occurs when the conflicted vehic l e experiences a rapid deceleration changes lanes or stops in order to avoid a collision and where t he dri ver has no t im e for a controlled maneuver. The most severe grade occurs when a near miss occurs. The study found that thr o ugh cross-trallic conflic t s w i th the traffic f rom right and the left produced a high percen t age of the conflicts with severity grade 4 as well as secondary conflic t s Salman and Almaita al s o used a severity grading system for eval u ating the safety of three-leg unsignalized intersections (18). The purpose of that study was to rat\k the intersections according t o the amount of risk. Each conflict was assigned a risk factor and the intersections wit h unusually high number of conflict values were determined using a risk index The risk index was cal culated by using following equations : where, n R.l j = .I: R.lij t=l R.lij = K; x (1.v )ij wi K i = _.!,_ n I wi i = l R.l; =total risk index for intersectionj. R .l ; ; = risk index for conflict type i at intersection j. K ; = relative weight for conflict type i. W; =weighting factor for conflict type i (l.V .);j =indicator value tor conflict type i at intersectionj. n = number of conflicts 17


The weighting factor was based on a subjective scale, whi ch ranged from I to 3: I b ein g the score for the least severe conflict, and 3, the score for the most severe conflict. The ind ic a tor valu e corresponded to the number of conflicts per thousand entering veh i cles on t he site. After the risk index w a s determined for each intersection this value was incorporated into the conflict matrix so that the most dangerous intersections could be found In addition to using various methods to determine the severity of confl icts, resear ch ha s also been conducted to evalua t e more accurate methods. Shinar evalua ted th e re l i a bi lit y of subjective evaluations ( 3 0). The methodology co n s i sted of filming conflicts at a three leg intersect ion and evaluating those conflicts acco r di ng to three types of m easureme n ts The first measurement was objecti vely determine d by using the de c e l erat ion of the vehicle as an i ndicator The other two measurements were subj e ctive in w h ich risk sc ores ran gi ng from 0 to I 00 were g i ven and the time to collision of the maneuver was estimated and assigned a score. The study found that people seemed to ha ve an interna l c oncep t of what constitutes a conflic t or a near crash and were consistent in their eva l uation of vehicle movements based on that concept. Kruysse performed two experiments to determine the information used by observers when assessing the severity of conflicts (31). The research used TTC as the objective measurement and compared its results to the risk perceived by subjects. Results showed subjects were reliable for observing traffic confl i cts if they were ta ugh t to see initial s t ages of conflicts The study also found that the commencement of a maneuver had more informa t ion regarding the evaluation of the severity than the other stages of the conflict. Thus, if one part of the conflict observed was considered as dangerous, the confli c t would be judged dangerous, even if the rest of the information contained by the conflicts indicated it was not dangerous. Also, it could happe n that the commenceme n t of the conflict was not dangerous but the evasive action was, so that the conflict was considered dangerous. 18


2. METHODOLOGY This chapter describes the methodology used in collecting, classifying, and analyzing traffic conflic t s including the criteria used for se l ection of sites, esti m ation of sample sizes defin i tions of the types of conflicts evaluated i dentification o f traffic conflicts, and data reduction. 2.1 Site Selection Accord ing to the object i ves of the study, two sets of sites were selected for direct left turns (DL T) and right turn s followed by U-turns (RTUT). The first type correspond s to those sites where the driveway vehicles in need of making a left turn could either turn to the left directly or tum t o the right and make a U turn The second type corresponds to those sites where only turning to the right was allowed, so that, only RTUT movemen t s would be expected but not DL T movements. Each site was selected based upon the following requirements: I) The arterial or major road mus t have three or more lanes in each direction. 2) Traffic volume on the driveway should be relatively high so that adeq u ate t urni n g vehic les could be s tudied. 3) The mi n imum distance between the driveway and upstream signal should be at least 200ft, which is the median value o f the distance travel e d during driver perception-reaction time and the impact distance due to a r i ght tuming vehicle (3). 4) The downstream signal was located at an appropriate distance away from the driveway in order to avoid the effec.ts of possi ble spill backs. 5) Posted speed on the major road is equal to or greater than 4 5 MPH. 2.2 Sample Sizes The samp l e size fo r a traffic conflic t study depends on th e types of conflicts that are being studied, traffic volumes on both the arterial and driveway, and the lev e l of significance required for the study. Traffic Conflict Technique for Safoty and Operation: 19


Engineer's Guide provides information about the number of hours of data collect ion recommended for obu.ining the minim um sample size for both signalized and unsignalized intersections wi th three or fou r leg approaches low and high speeds, and for two and four lane roads {1) Also it explains the procedure for calculating th e number of observation hours needed based on the level of significance, error of the hourly mean hou rly variance estimated from previous conflict studies and the hourly mean number o f conflicts of a specific t ype, as follows : where, "=(oo ,.!....),a ; p J'. n = number of hours of observation needed t = statist ic from the normal distribution related to the selected level of significance a., p = error of the hourly mean, cr/ =hourly variance of co nflicts estima te d from previous studies and Y = hourly mean number of conflicts of a specific type. Tbis procedure requires previous knowledge of the m e an and variance of the conflict numbers, cr/ and Y, for calculating the samp l e size. Hence application of this method was not possible since the hourly variance and mean numbers of conflicts due to the two maneuvers evaluated in this study w ere unknov.n. Another procedure used in calculating th e sample size is described in the Manual of Transportation EngineeringSwdies, Chapter 12 (14) This method estimates the number of approaching vehicles that must be counted for a specific level of significance and error. Then, the number of vehicles calculated is compared to that of the field study so that it can b e corroborated that the estimated conflict ra te is between the desired errors. The eq uatio n used to calculate the sample size with this method is: where, 20


n = number of vehicles to be counted, p = expected proportion of vehicles observed that are involved m a conflict z = sta t istic that is based on the leve l of significance desired, D = perm ined lev el of absolute error of sample size. For example if one wants to find the sample size, with a l evel of sign ificance of 5 percent and with an absolute e rror of 5 percent, the number of vehicles wo u ld be n= 0.50 (I -0.50)*( 1 .961.05)A2= 384 approach vehicles In the example presen ted above, the expected proportion of vehicles observed and involved i n conflictS was unknown. Thus, a conservative estimate of the samp l e s iz e could be obtained b) assuming p as 0.5 This method was followed to calculate the sample size of this study. 2.3 Types of Conflicts Studied No t all th e conflict types as defined in previous studies are important for the maneuv ers evaluated in this study (I 2). Therefore it was necessary to evaluate the types of conflictS that were significant to the scope of the study. This resulted in the selection and evaluation of nine relevant types of conflictS, which are briefly described as follows. I) Right -Turn Out of the Driveway (CI). occ urs when a vehic le (offending vehicle) waiting at a driveway, turns to the ri ght and gets onto the major roa d, placing another vehicle (conflict ing vehicle) on the majorroad with increased potential of a rear e nd or sideswipe collision. See Figure 2 .1. 2) SlowVehicle, Same-Direction Conflict (C2). occurs when a right turning vehicle is already on the major road and begins to accelerate while on the path of a major road vehicle thus, the major road vehicle is encountered with increased potential of a rear end collision This type of conflict is shown in Figure 2.2. 3) Lane Change Conflic-t (C3j, occurs w hen a vehicle from a driveway that turned to the right changes from one lane to another (weaving) until it reaches the U-tum bay 2!


This maneuver may place through-traffic vehicles wirb increased potential of rear -end and sideswipe collisions. See Figure 2.3 4) U-/urn Conflict (C4), occurs when a vehicle making a U-tum places vehicles coming from rbe opposite direction with increased potential of a sideswipe or angle accident. This type of conflict is illustrated in Figure 2.4. 5 ) Slow UTurn Vehicle. Same Direction Conflicl (C2 U1), occurs when a vehicle that completed the U -tum maneuver and accelerates; placing an oncoming major-road vehicle with increased potential of a rear-end c()ltision This type of conflict is similar to conflict type C2, but it was exclusively designated for vehicles making U-tum. In this type of conflict the speed differential involved could be even more dangerous than that of conflict type C2 because U -tum maneuvers are usually made at a very low speed making the stop distance greater This type of conflict is graphically illustrated in Figure 2.5. 6) Left-Tum Out of Driveway: Conflict From Righi (C5), occurs when a vehicle on rbe driveway turns to the left and places a major-road vehicle with the right-of-way with increased potential of sideswipe and right-angle collision, as shown in Figure 2.6. 7) Direct-Left Turn ond Left-Turn in From Right Conflict (C6) occurs when a left turning vehicle from rbe driveway places a veh icl e turning into the same driveway with increased potential of sideswipe or angle collision, as shown in Figure 2.7 8) Direct-Left Turn and Left-Tum in From -Left Conflict (C7) occurs when a left turning vehicle from the driveway places a vehicle turning into the opposite driveway with increased potential of sideswipe or angle collisions, as shown in Figure 2.8 9) Left-Tum Out of Driveway: Conflict From Left (C8), occurs when a left turning vehicle located on the median storage area places an oncoming major-road vehicle wirb i ncreased potential of a rear-end or sideswipe collision, as shown in Figure 2.9. 22


-------------------------------------------------------I I Figure 2 1 Confli ct Typ e C I -----------------------------------------------------l I ( liD fill _,;"" -------------Fig ure 2 2 Confli ct Type C2. 23


' ' .,->:: ' . .. ---------------I figure 2.3 Conflict Type C3 ., ----'-----------------9 ---'. . / ---------------------M ---------------------figure 2 .4 C on fli ct Type C4 2 4


m -------------------+----' -------------------i ----- ' / / --------------------------------------------------Figure 2.5 Conflict Type C2U T ----------------------------------------------------Figure 2.6 ., ----------' ' I !a ll!i Conflict T ype C5 25 -------------------


----------------------' ----------. -.---------- Fig ure 2. 7 Con flict Typ e C6 __ ) __ -----------------1 --------ft--------____________ !1 ______ ___ t F igure 2.8 Confl i ct Type C7 26


-------------------------------------------------------------------I I ( F i gure 2 9 Conflic t Type C 8 2. 4 Iden t ifi cation o f T r affic Confli cts This study cons idered evasive mane u vers such as applying brakes, swervmg, or no t iceably decelerati n g in order to avoid a collision as conflicts. Brake appl ications have been used i n several studies as indica t ors of the occurrence of a conflict, even t ho ugh this might pose some proble m s in some c ases (18 23, 24, 28, 33 34, 35). For example, brake applications cou ld be made as a precautio n ary action which cou l d be m i sunderstood as a conflict, they may not provide infonnation regarding the severity of the conflict, in some cases collisions may occur when vehic les a c celerate to avoid a collision instead of applying brakes, and brake li ght s could have mechanical failures (29) Swerving is also used as an indi c ator of the occurrenc e of a conflic t although it may not sometimes be clear whether swerving was because of a co n flict or not. Using percept io n of deceleration o f a vehicle i s useful for detecting conflicts when there are no brake ligh t indications because of a mechanical fail u re. Howe ver, by using this ind icat ion, conflicts may become too subjective Even though these problems are fo und, brake applications, swerving and noticeable deceleration could be used as indicators of the occurrence of traffic conflicts. 27


Several researchers have noted tha t observers are the most important element when conducting a traffic conflict srudy because their reliability bas a serious impact on the validity of the data (/4, 29). T herefore, training and educating observers was one of the most important factors considered in the initia l stages of the project. In order to train the observers, the Traffic Conflict Techniques For Saftty and Operations: Observers Guide and the Center for Urban Transportation Research's "Training Tool Kit were used (14). With the intensive training provided and the experience gained through the research the judgements of the. observers were very reliable Having trained the observers, the next step was preparing the data collection forms necessary to extract the conflict data from the videotapes recorded at the selected sites. After review i ng several traffic conflict studies it was decided that the standard forms used in generdl cases did not meet the needs of this research. Therefore, a special form was designed as shown in Figure 2.1 0 Column I was used for recording the time at which the conflict occurred, columns 2 to I 0 were used for record in g primary and secondary conflict types, column II was used for recording the distance between the offending and conflicting vehicle once the conflicted vehicle app lied the brakes column 12 was used to record the risk of collision score given by the observer, and the l ast co l wrut was used for indicating the existence of a special event. T he flow chart shown in Figure 2.11 describes the process followed by observers to record the conflicts. When a vehicle departed from the driveway the time was recorded and the observer tracked the vehicle in order t o see if the movement was a DL T RllJT, or neither of them. I f the maneuver was either a RllJT or DLT, it was observed for conflicts. If a conflict occurred, the time of occurrence, conflict type, risk of coll is ion perceived during the evasive action, and distance were recorded The Risk Of Collision (ROC) was evaluated by assigning a score according to the level of severity of the evasive action perceived. The score values ranged from I to 3, with 1 being a low-risk conflict and 3 a high-risk conflict. 28


1::! S ltt: oau Cotlec:uon Date: ws Date ot o.ra Analysis : sa t---1....--Dbservu: me c: 111re oro ' ype s n. -r---_,:J ) _, r .. .J'\ ..-....,--: 1:1 1:2 C3 C4 C$ e C7 Ct C Z U T t 911111 "' "' f tCOIII I SC sec:ol'l41 t y Conmct Figure 2.10 Th e Form Used for Recording Traffic Conflicts. 2 9


Vehic l e i Departs Type of Movement Is this DLT? Next Vehicl e i Traffic Conflic t Traffic Conflict YES YES Time Conflict Type R isk Of Collision Distance Time Conflic t Type Risk Of Collision Distance Figure 2 .11 Flow Chart Describing Confl i ct Identification and Data Required by Observers 30


2.5 Dat a Reduct i on Proced u re Once traffic conflicts were recorded and coun t ed s t eps were taken to verify the accuracy of the data before performing analyses First, data must be reviewed so that any noticeable error or missing record could be corrected promptly. For example, a record may indicate two different types of conflicts therefore, information on conflicts must be reviewed and only one tYPe of conflict must be selected. It was al so important t o verify that data was collected a t the same time in all the approaches and to disregard those periods of time in which there were missing values Determining daily confl ict c oun t s required that the number of conflicts collected in a site was adjusted to one day Traditionally, conflict studies have cons i dered eleven h ours as o ne day, starting at 7 :00 AM and ending at 6:00 PM (13. 17, 18, 23, 25, 27). Parker and Zegeer recommend adjusting data for those periods in which data were not collected (13. 14). The me t hod calcu l ates the number of conflicts based on the observed conflicts before an d after the period in w h ich data was not collected The adjusted number of conflicts for non observed periods was calculated by using t h e following equation: c1 + c2 (TTNOP) ANOC = X ..;.__--'2 RP where, ANOC = adjusted non-{)bserved period c o nflicts, Cl = number of conflicts occurred before the non-observed period, C2 = number of co n flicts occurred after the non-observed period, TTNOP = t otal time of non-obs e rved period, RP of recording period After th e adjusted no n -{)bserved p e riod conflicts were estimated, the daily numb e rs of conflicts were obtained by adding all the conflicts irrespective of whether they came from observations or from adjusted values for non-observed period s When the daily numbers of conflicts were obtained several types of c o nflict rates cou l d be calculated. Table 2 I presents the two types ofrates used in this study. T he first one is th e r atio between the 31


number of conflicts and the number of hours of observation Conflicts per ho u r could be used to show the conflicts that might be found during d ifferent time periods such as peak and non peak periods or the total average. Tbe second conflict rat e is the ratio bet ween the number of conflicts and traffic vo l ume This rate is defined as the number of conflicts per thousand involved vehic l es by maneuver type Table 2.1 where Definition of Different Conflict Rates. Ra t e De f inition Conflicts per Hou r CR1= Number of co nflicts Number of hours Conflicts per Thousand Involved Vehicles CR1 = conflict rate I. CR2 = conflict rat e 2. N umber of conflicts CR2 = v1 v2 V 1 = traffic volume on arterial, according to conflict type xiOOO V2 = vol u me ofRTUTIDLT maneuver, according to conflict type 2 6 T h e S tudent 's t Test The Student's t test could be used to compare mean values of t wo samp les that do not follow normal distribut i on. This t est was used in this study to determine if the differences between conflic t rates of RTUT movements were significantly different from that ofDLT movements. In thi s test the null hypothes i s (Ho) is that the difference between the means of the sam p l es is equal to zero The 1 statistic calculates the probability of the difference of the two m eans; say x1 and x2 having a valu e greater than or equal to the observed value. However, the 1 te s t assumes that the two samp les belong to the same population and t herefore, the variances must not be significant l y different. This condition has to be verified by u sing the Fiest before the 1 test is performed (36). 32


The variance ratio test, known as the F test, compares the difference between the variances of two samples The F test adopts the null hypothesis (Ho) that the two samples belong t o the same population. The F sta ti stic is c a lculated as: where, S1 2 and S/ are the variances of the samples and S1 2 S22 If the calculated F val ue exceeds the crit i ca l F value, the null hypothesis is rejected and the variances are considered to be signilicantly different. Once the difference between variances is verified to be not significant, the 1 value can be calculated as: where, x1 and X l = observations from the two samples, n, and n, = number of observations in the two samples, x and x, =mean values of the observations from two samples I f the calculated 1 value is greate r than the critical/ valu e a t a given l eve l of sign i ficance, the null hypo thes is is rejected and it could be concluded that the diffe r ence between the two conflict rates is significant. 2.7 One-way Analysis of Variance (ANOVA) Test The Analysis of Variance i s a method used for detennining whether two sets of data are significantly differen t from each other. The anal ysis of variance splits the variance of all the elements into variance between samples and variance within samples These are calculated as the sum of the squares of deviations divided by the corresponding degrees of freedom, and compared by the F test (36) The ANOV A test was used for detcnnining if the severity of both RTUT and DLT movements was significantly different from each other. The null hypothesis (Ho) assumed that the severity of conflicts generated by RTUT 33


movements was not different from that of conflicts generated by .PLT movements I f the calculated F va l ue was greater than the critical F value then the n ull hypothesis was rejected and th e difference would be significant. 34


3. DATA COLLECTION This chap ter describes the procedure followed during data collection and reduction and also include s the detai l s of the selected sites equipment used. procedure followed in the field, and data reduction process. 3.1 Description of Stu dy Locations Ten sites around the Tampa Bay area were selected for data collection based on the criteria described previously Figure 3.1 shows the location of these sites. Table 3.1 presents a description of the geometric characteristics of the selected sites. Figure 3 2 illustrates the general l ayout and lo cation of cameras for Sites I 3-After improvement and 4. Figure 3 2 illustrates the general layout of Sites 2, 3-Before improvement, 4 5 6, and 7. Figure 3.1 Location of Selected Sites. 35


Table 3.1 Descriptions of the Selected Sites. Intersection Site Number of Lanes tO Arterial Driveway Fowler 1\ ve. & St. I 3 I Fowler Ave. & 191 St. 2 4 2 US 1 9 & Ave. 3 3&4 I Bruce B Downs & Medical Center 4 3 I Hillsborough Ave. & Golden St. 5 3 2 US 19 & Enterprise Center 6 3 I US 19 &: !nnisbrook 7 3 2 Fowler A vc:. & S2n d Sr. 8 3 I Gunn Highway & Hangert 9 2 2 Bnce B Downs & Pebble Cre<:k 10 2 I NOTE: Distan c e A: Distance from driveway to upstream signal. Distance B : Distance from driveway co Ut\lm bay. Distance C: Distan ce from U t u m bay to downstream -Signal 37 Maneuver allowed Posted Speed Distances (ft) by Median DLT RTUT (mph) A B c No Yes 50 950 800 700 Yes Yes 45 700 570 1350 Yes Yes 55 600 420 1620 Yes Yes 45 870 970 1160 Yes Yes 45 850 300 750 1 Yes Yes 55 1700 550 4750. Yes Yes 55 528 0 6()0 5808 No Yes 50 1200 580 530 Yes Yes 45 2120 590 2238 No Yes 45 1000 850 850


. ------------------............... __ /. ...... ........ .__ ----------- ------,.J-'-:_ __________ ---I Figure 3 2 General Layout of Sites I, 3 -After Improvement and 8. . =............... ... -______ ? _ .... -_.,.,. c=;;::::, (/=::--, m . [5]c ::::.: ..:"1-.:::: Figur e 3.3 General Layout of Sites 2, 3-Before Improvement, 4, 5, 6 and 7. 37


Site I is located in Tampa, on Fow ler Avenue and 46'h Street. Fowler Avenue is a major divided arterial oriented in the east-west direction with three lanes in each direction. 46!h Street is a local street with one lane in each direction. The median on Fowler Ave is raised and at 461 Street it becomes a direct i onal median opening. Because of this drivers that come from 4611' St. who want' to take Fowler Avenue to the west have to make RTUT. The posted speed limit is 50 mph Site 2 is located on Fowler Avenue and 17lh Street, beside the University Square Mall. At tlus point, Fowler Avenue still is a s ix-lane road divided with a raised median However, there is a full median operting located across the driveway. The driveway has two lanes for egress of vehicles wb. ich provides the options of DLT and R TU T movements to drivers. The posted speed l i mit is 45 mph. Site 3 is located on US J 9 and I J 6'h Avenue in Pinellas County US 19 is a major arterial oriented in the north-south direction, with three and four lanes on the southbound and northbound, respectively. T he northbound and southbound lanes are separated by a raised median The posted speed limit in this segment is 55 mph. This site is of special interest because the median was improved within the data collection time period. Initially a full median opening allowed vehicles to make DLT maneuvers, whereas after the improvement only right turns were permitted. Therefore, a before and after study of this site could be conducted to provide insightful information about DL T and RTUT maneuvers, without the concern of having different site characteristics. Site 4 is located on Bruce B. 00\\'IIS Blvd. and 131" Avenue. This site is located at the entrance to the parking Jot of the Veteran's Hospital Bruce B. Downs is an arterial with three lanes in each direction at the selected location. It is oriented in the north-south direction and has a posted speed limit of 45 mph The driveway has one lane for egress. S i te 5 is located on Hillsborough Avenue at the driveway for Verizon, a major telecommunications employer The driveway permits ingress to and egress from Veri:z.on's offices. Hillsborough Avenue is an arterial oriented in the east-west direction with three lanes in each direction Eastbound and westbound lanes are divided by a raised 38


median. This site presents a full median open ing located across the driveway The driveway has two l anes for egress where one lane is used for left turns and the other lane is for right turns. The posted speed limit in this segment is 45 mph. Site 6 is located oo US 19 and Emerprise Center ill the Palm Harbor a r ea of Pinellas County The driveway is adjacent to an office building. US 19 is a major arterial with three lanes in each direction, separated by a raised median. The driveway has only one lane for egress. On this site drivers have t h e option to make either DLT or RTU T maneuvers. T h e posted speed limit is 55 mph Site 7 is located on U S 19 and lnnisbrook in the Tarpon Springs area of Pinellas County In this case, the driveway provides access and egress for residents living in nearby complexes. ln this segment, US 19 still has three lanes in each direction and is divided with a raised median. Site 8 is located on Fowler Avenue and 52"" Street in the Temple Terrace area Fowler A venue is a major divided art erial oriented in th e east-west direction, with three lanes in each direction 52"" Street is a local street with one lane in each direction serving residents and customers of surrounding residential areas and stores, including a supermarket. The median is raised and restricts DLT maneuvers with a directional median opening. The posted speed limit is 50 mph Site 9 is located on Gunn Highway and Hangert Street. Gunn Highway is a divided arterial oriented in the northwest direction, with two lan es in each direction. Hangert St. provides access and egress t o resi d ents liv ing in nearby complexes. In this segment the posted speed limit is 45 mph. Site 10 is located on Bruce B. Downs Blvd. and Pebble Creek in the New Tampa area. Bruce B Downs Blvd. is a divided arterial oriented in the northwest direct ion, with two Janes in each direction The driveway has one lane for the egress of vehicles and the median restricts DL T movements. The posted speed limit is 45 mph. 39


3.2 Data Collection Equipment At the beginning of the study, manual traffic conflict observers were cons id ered as the data collection methodology, as several pas t conflict s tud i es had utilized that meth od. Howeve r two reasons supported the ide a of using video cameras inste a d First, i t would be ext remely di ffi c u lt to record traffic c o n fli ct s and t r affic vo lu mes at the same ti m e beca use the obsener had to track and identify the man euver that each vehicle made when it depa rted from the d riv eway. Second onsite observa t ions canno t easily be verified which is o f specia l concern when the o bserver has to reg ister several types of conflicts ( 11). I n contrast using video cameras would allow the o b server t o revi e w all the con flic t s that might have o ccurred a t t h e sit e and if a dou b t persists the videotape can be replayed until the matter i s clarified and a mor e i nforma tive d e cision can be made. Also te cb.nological advances on video cameras a llo w a precise registration of the time the co nflic t occurred (21) Nevertheless, using v i deo came ras for gathering confl ict dat a requires severa l hours reducing data from videos Three cameras we re used for recording conflicts. Two of the cameras, Sony DCRTRV 320 Digital g TM and Sony DCR-VX700 were dig ita l cameras so that mov ement s on the aneria l wou ld be captured with a very high quali ty. The other vid e o camera, a Sony CCDTRV68 was used to record m ovemen ts on the U-tum bay when an extra camera was needed Altho ugh the cameras had high resolution, an important factor had to be resolved before beginni n g data colle ction. If the cameras were set up close to the ground level it would hav e been impossible to collect all the traffi c conflicts since the vehicles traveling in the inner lan e would be covered by other vehicl es trave l ing on the middle o r outer lanes. A solu t ion for this problem was t o install the camera at an adequate height that is at least 15 ft above the ground, by usi n g s caffolding Th e advantages of using scaffolding wer e that it p r ovided a stable and sa fe means of pro v iding e no ugh height to install cameras and it was not as expensi ve as the other equipment, such as buc ket and scissors lifters. The scaffolding and cameras were posi tio ned and installed a c cordi ng to the needs of each si t e as previously shown i n Figures 3.1 and 3.2 40


Traffic volumes were also recorded while collecting data. An automatic data recorder ADR-1000 from Peek TrafficT" was used to obtain the traffic volume on the arteriaL Right-tum, direct left-turn, left-tum in and right tum plus U Tum maneuver volumes were obtained from videos. 3.3. Field Procedure Data were collected during weekdays under normal traffic conditions, good weather and dry pavement conditions. Weekdays in this study were considered as from Monday through Thursday, where normal conditions are expected to prevaiL Also, data were collected during peak and non-peak periods. The morning peak period between 7:00AM and 9:00 AM and afternoon peak period between 4:00 PM and 6:00PM were considered in this study. However, not all the data were collected during the same time period for aU sites because every site has different land uses, which changes the d riveway volume pattern For example Site 2 was located near a mall, which means that there were few vehicles between 7 :00 AM and 12:00 PM. Therefore, data was collected between 2:00 PM and 6:00PM. Two other important factors were considered in the data collection and reduction process. First, all cameras and counters were synchronized before the start of the data collection. The purpose of this was to eliminate tedious data matching such as vehicle's color and model, especially when 3 or 4 cameras were involved simult aneously. Second, when installing the cameras on the scaffolding, it was very important to set them at a good angle and zoom in order to accurately detect brake lights or any other evasive maneuvers 3.4. Dala Reduclion Data was reduced from videotapes by following the procedure explained in the flow chart described in the methodology section. Each ve. hicle from the driveway under investigation was tracked down and departure time from the driveway arrival time either at the left-turn storage area or at the U-tum bay and deparrure time from the U-tum bay or median opening were recorded Then, the videos from the cameras located on the opposite bound side were reviewed at that specific time, which acce l era ted the reduction 41


process without affecting the quality Another advantage of doing this was that t his informatio n could later b e used to determ i ne each maneuver vo l u m e. It is important to high l igh t t hat the data reduction process was very time consuming and the approximate l y three to four hours of time was requ i red to reduce one hour of video recor de d i n the fie ld. Based on the information reco rd ed on the form designed for that purpose a database was created i n order to analyze the data collected from videos in a fast and efficient way T h e software used was Microsoft Access because of its u ser-friendly characteristics and it s capability t o export information t o spreadsheets su ch as Exce l. The database consisted of three t ables t hat contained the informa t ion recorded in the form used for red u cing data from videos departure t ime of vehicles l eaving t he driveway, and information of the traffic volume on the arterial. Several queries were designed in such a way that contlicts, arterial vo l ume, and maneuver volume would be assigned to the respec t ive 15 mi nu te interva l of the day For example int erval I corresponds to the tim e period between 7:00 AM and 7: I 5 AM, and so on. After the queries were run, their results were exported to an E x cel spreadsheet so that further data analysis could be perfonned 42


4. DATAANALYSlS Th is chapter describes the number of conflicts and conflict rates for both DL T and RTUT movements and a safety comparison between the two alterna t ives. Conflict severity related to these two maneuvers are also presented and compared 4.1. Descriptive Analysis It was first necessary to veri!)' if the sample size of t he data collected was satisfactory. and the r efore minimum sample sire was estimated us ing the equation described previously with a confidence l evel of 95 percent. Moreover the percentage of vehicles associated with each movement and conflict type was calculated i n order to evaluate the proponion of vehicles involved in a conflict, as presented in Tabl e s 4.1 and 4 2. The sample siz.e requirement was no t met tor the last two sites whic h were four-lane roads. T herefore the conflict data analys es presented in this chapter were based on the data collected at the s ites with six or more Janes only. Table 4.1 Verification of Sample Size for RTUT Movement Conflict Average Number RTUT PRruT D Sample Size Type of Conflicts Vehicles Satisfied (I) (2) (3) (4) = (2) /(3 (5) (Yes/No) Cl 57.4 1372 004 62 Yes C2 64.9 1 372 0.05 69 Yes C3 68.3 1372 0.05 73 Yes C4 73.4 1372 0.05 78 Yes C2UT 35 8 1372 0,03 39 Yes PnruT"' Percentage of RTUT vehicles involv e d in a conflict. n =Nu mb er of vehicles estimated for sample size 43


Table 4.2 Verification of Sample Size fo r D L T Movement Contlict Average Number DLT maneuver Type of Conflicts Veh. icles Sample Size n Satisfted (1) (2) (3) {4)=(2) 1 (3 (5) cs 180.8 1430 0.13 170 Yes C6 106.6 1430 0,07 106 Yes C7 13. 9 1430 O.DI 15 Ye s C8 78.6 1 430 0.05 80 Yes PoLT-Percentage o f DLT vehicles involved in a conflict. n = N umber o f vehicles estimated for sample s i ze. A dcscdptive analysis was performed to evaluate the information gathered i n the field. T able 4 3 presents a summary of the total of J 654 conflicts collected in the field at eight sites and used in the analyses. I f data from both direct ions at a given time period was not available such da t a were disregarded in the analysis Table 4.4 present s a summary of t he average numb e r of conflicts by conflict type for each of the s i tes with three or more l anes i n each direction Also, the average daily number of conflic t s for each site and conflict type were obtained based o n the average number of conflicts and these values are given in Table 4.5, and Figures 4 1 to 4 7 graphically i ll ustrate the individual data at each site except for site 3, f o r all conflict types Since site 3 experienced a median closure making it an ideal site for conduc t ing a before and after type of eva l uation analysis r esults related to that site are documented separat e ly in Chapter 5 44


Table 4.3 Summary of the Total Number of Obse rv atio n s. Site Conflicts Conflict Type Total C1 C 2 C3 C 4 cs C6 C7 C8 C2UT I No. 6 4 22 23 IS N/A N / A N/A N/A 28 152 (%) 42.1 1 4.5 15. 1 9.9 ---18 4 100 2 No. 4 9 3 75 221 5 36 356 (%) 1.1 2 5 0.8 0 8 21.1 62.1 1.4 I 0. J -100 3 No. 15 1 7 15 6 150 71 18 7 4 5 371 Before ( % ) 4 4 6 4 1.6 40.4 19.1 4 9 19.9 1.3 100 3 No. 4 0 3 6 89 1 18 N/A NIA NIA N/A 44 327 After (%) 12.2 ll 27.2 36.1 13. 5 100 4 No. J 9 2 2 37 18 1 2 8 1 (%) 1.2 1l.l 2 5 2.5 45.7 22.2 14.8 -100 5 No. 2 II 2 I 39 22 3 17 97 (%) 2.1 11.3 2 1 1 40.2 22.7 3.1 17 5 -100 6 No. I 1 2 3 9 24 I 2 II 2 65 (%) 1. 5 18.5 4.6 13.8 36. 9 1.5 3. 1 1 6.9 3.1 100 7 No. 2 15 5 41 14 I 21 99 (%) 2 1 5 2 5.1 41.4 14.1 I 21.2 -100 8 No. 26 22 22 18 N/A NIA NIA N/A 18 106 (%) 24.5 20.8 20.8 17 -17 100 TOTAL 155 153 164 172 366 347 29 171 97 1654 45


Table 4.4 Average Number of Col)fticts Used for Analysis Site Descript ion Contlict Type Total Cl C2 C3 C4 C5 C6 C7 C8 C2UT I No. 21.2 7.6 7.1 4.5 N/A N/A N/A N/A 8.3 48.7 (%) 43.5 15.6 14.6 9 2 -17 100 2 No. 0.6 1.3 0 5 0.4 10.3 32.3 0.7 5 3 -51.4 (%) 1. 2 2.5 I 0.8 20 6 2 8 1.4 10 3 100 3 No. 8.5 9.3 8.7 4 93.5 44. 5 I 0.8 41.9 2.5 223.7 Before (%) 3.8 4.2 3 9 1.8 41.8 19 9 4.8 18. 7 1.1 100 0 No. 16.6 1 5 .5 37.4 53.4 N/A N/A N/A N/A 17.4 140.3 After (%) 11.8 I I 26.7 38.1 ---12.4 100 4 No. 0.5 6.5 1.5 1.5 26 14 9 59 (%) 0 8 II 2.5 2 5 44. 1 23.7 15.3 -100 5 No. 0.5 3.7 0.5 0 25 15 7.3 0 9 5.4 33.6 (%) 1. 5 II 1.5 0.7 44.7 21.8 2.7 16. 1 -100 6 No. 0.5 6 2 4.5 13 0.5 I 5 5 I 34 (%) 1.5 17.6 5.9 13.2 382 1.5 2.9 16.2 2 9 100 7 No. I 7 5 2.5 23 8 0.5 11.5 54 (%) 1.9 13.9 4.6 42.6 14.8 0.9 21.3 -100 8 No. 8 7 6 8 2 4.8 NIA N/A N/A N/A 6.6 35 .2 (%) 22.7 2 1.6 23.3 13.6 18 8 100 TOTAL 57.4 65 68.4 73.3 180.8 106 6 13.9 78.6 35.8 679. 9 4 6


Table 4 .5 Ave r ag e D a ily Number of Conflic t s Conflict Type Tota l Site Cl C2 C3 C4 cs C6 C7 C8 C2UT I 27. 2 10.6 7. 8 9.8 N / A N/A N/A NIA 13.5 68 9 2 0.7 1.4 0 6 0 4 11.3 67.5 0 8 1 6 5 99.2 3 8.5 1 3 3 10 7 4.0 117. 44.5 1 4 8 43.8 2.5 259.6 3 16.6 1 5.5 37.4 53 3 N/A N/A N/A N / A 1 7.4 1 4 0.3 4 0 5 7.5 2.5 3 0 34.0 17. 5 -1 0 0 75 5 0.5 3 7 0.5 0 3 45.0 7 3 0 9 5 4 6 3. 6 6 0.5 6 0 2.0 4 .5 22. 0 0.5 1.0 5 5 1.0 43 7 1.0 8 0 3 5 29.5 1 2 0 0.5 15 5 -70 8 26. 1 9.6 19 2 4 8 N/A N/ A N/A N/A 15 6 75.3 At Site 1 approximate l y 3 4 p erc en t of a ll the confl i c ts were rel a ted t o U-turn mane u ve r s (i.e. Co nflic t t ypes C2U-T and C4). Of these 42 p er cent corres p o n ded t o the U turn mane u v e r itself, and 5 8 pe rcen t t o conflict type C2UT. Conflict t ypes Cl, C2, and C3 accounted for 39, I S, and 1 2 percen t of the to tal conflicts, respect i vely as shown i n Fig ure 4 .1. At Site 2 the average numb e r of daily conflicts was considera b ly higher for D L T movemen t s (97%) than that of RTUT mov em ents (3%), as show n in Figure 4.2 It i s interest in g to n ote that 70 percent of the c onflicts asso c iated w ith DL T m ovements were of type C6, which i s expla i ned by a h igh vo l ume of left -turnin g vehic l es into the driv e way. The hig h left-tum in vo l ume i s caused by the l ocation of the driveway, w hi ch was located nearby a mall Also, it is i n teresting t ha t the daily average n u mber of conflict type C8 was 46 percen t h ighe r than tha t of co n flict type C5. This may be the result of the lack of sight suffered by dri ve r s waiting i n the med i an storage area The lack o f s ight was d ue to high left turni n g vehic l es into t h e m all and t h e involvement of differ en t ve h icles o n the median 47


70 j i 60 so I O :;; 4 0 1 2 = z 3 0 1.2: a 20 1 0 10 I > < 27.2 ______ ,_ _ _______ _ ___ .. ______________________ --------------. -... .. --.. ---_______ ., _ _____________ ., _ __________ ____ _____ _ --.. .. ------...... -.. ----------------------9.8 13 .5 1 0 6 7.8 0 Cl C2 C4 cs C6 C7 C8 C2U T ConfllqType Figure 4 1 Number of Daily Conflicts by Type Site I 70 60 5 0 4 0 ------.. ---.---3 0 20 _____ ______________ .... ___ ....... ) ..... 1 0 0 7 1 4 0 6 0.4 0 8 C l <::2 C3 C4 cs C 6 C7 <::8 C2U T C o nfl i ct Type F i g ure 4 2 Number of Daily Confl icts by Type, Site 2. 48


Conflict types observed at Site 3 are explained in more detail in Chapter 5. Site 4 bad a higher percentage of DL T movements rather than RTUT movements. This explains why DLT accounted for 82 percent of all the conflicts, as s hown in F igur e 4.3. Approximately 55, 29, and 16 percent of the conflicts corresponded to DL T conflict types CS, C6, and C8, respectively The high percentage of conflict type C5 may be due to the narrow width and length of the median storage area. It was frequent l y observed that left turning vehicles encroached on through -t raffic lanes. Figure 4.4 shows the number o f daily conflicts at Site 5 Conflicts related to DLT movements were 92 percent of the total conflicts counted. Conflict types C5 C6, C7, and C8 accounted for 77, 12 2 and 9 percent of all the conflicts respectively. Data further shows that the number of conflicts of type C6 was 36 percent higher than that of conflict type C8. At Site 6, 67 percent of the total conflicts corresponded to DL T movements whereas 33 percent corresponded to RTUT movements a s shown in Figure 4.5. Among the conflicts related to DLT movements conflict types C5 C6, C7, and C8 accounted for 76 2 3 and 19 percent respective l y. With regard to conflicts associated with RTUT maneuvers 43, 14, 32, 7, and 4 percent corresponded to conflict types C2, C3, C4, C2U-T and Cl; respectively A possible reason for the high number of conflict type C2 can be attributed to vehicles traveling at high speeds on the through-traffic lanes. These values also show that 39 percent of the conflicts were associated with U-tum maneu v ers. At Site 7, DLT conflicts accounted for 82 percent of the total while RTUT conflicts only accounted for 18 percent. The roost common type of conflict among those related to D L T maneuvers is C$, with 52 percent, followed by conflict type C8, with 27 percent, and conflict types C6 and C7 As for conflicts related t o RTUT conflicts, the most common conflict type was C2, which consisted of 68 percent of all the conflicts followed by conflict types C3 and Cl. The high percentage of conflict types C2 and C3 may be explained by the high speed of through traffic vehicles. See Figure 4 6 49


I 0 I 70 60 s o , E 40 -------------------i4 6 ----------.-. --, z 30 .!< Q 20 !1, 10 05 0 .1---Cl 7.5 C2 17 5 C3 C4 C5 C6 C7 Conflict Type F igure 4.3 Number of Daily Conflicts by Type, Site 4. 70 10.0 C8 C2U-T 1 .;; 60 -----------------------------------"' c i so ... .... -.... M ... ---.. --.. -4s. o -.. -.. -.. - o 40 ---------------1 0 .. .... j,,;-.. ................ .. 0.5 o.s 0 3 i C1 C2 CJ C 1 CS C6 C 7 C8 C2U-T l__ _____________________ c_ooru __ Figure 4.4 Number of Daily Conflicts by T ype, Site 5 50


70 l 1'2 60 ---------------------.......... .... .. 8 so _____ ,. ____ ...................... .......... ____________ 0 j 4 8 20 1 0 < Figu re 4.6 5.5 o.s z.o 0.5 1 0 1.0 C l C2 C3 C6 G 7 C$ C2UT Conflicl Type Num b er of Daily Confli cts by Type, Site 6. 8.0 1-.0-------35-------C l C2 CJ C4 29.5 cs ----1s.s -1 2 0 0.5 C6 C7 C8 C2U-T Number of Daily Conflicts by Type, Site 7 51


At Site 8, it was found that 27 percent of the conflicts were related to the U-tum maneuver (i.e. conflict type s C4 and C2UT) while the others accounted for 73, as shown in Figure 32. When comparing conflict types C4 and C2U-T, it was found that the number of conflict type C4 was 69 percent lower than that of conflict type C2 U T. Figure 4.7 also shows that the most frequent conflict types on this site were conflict type Cl and C3, with 35 and 25 percent respectively. This may be caused by a lim i ted weav ing distance f 70 60 c g u so 0 -4j) .i:: z 30 .?;-26.1 ;; c 2 0 :il. 1 0 > < tS:rr 0 C l C2 C3 C4 C6 C7 CS C2U -T Conflict Figure 4 .7 Number of Daily Conflicts b y Type, Site 8 When data was pooled by maneuver type, results previous l y presented vary. Figure 4.8 shows the average number of daily conflicts for RTUT movements. An average of 94 .8 daily conflicts can be observed in one day, with 24 percent corresponding t o conflict types Cl and C4, foll owed closely by conflict type C3 wit h 23 percent, then by conflict type C2U-T with 16 percent and finally con.flict type C2 with 13 percent. Thi s data indicates that on average 40 percent of the conflicts correspond to the U -tum maneu v er, 24 percent to conflict type CJ 23 percent to conflict type C3, and finally conflict type C2 with 13 percent. It is important to note that conflict type Cl occurred most of the time ei t her on the middle or inner Jane, contrary to what was expected at the beg i nni n g of the study. 52


j so I 40 e < I 8 0 lO 2 E I z 20 l\ < 10 A a 0 Figure 4.8 -----------. --.. .. 23. 3 Cl C o nnict Type 22. 6 C4 C2l.IT Average Number of Daily Conflicts by Type, RTUT Movement. Figure 4.9 shows the average number of daily conflicts created by DL T movements. An average of 87. 2 daily conflicts can be counted in one day; from which 50 percent correspond to conflic t t ype C5, 29 percen t to conflict type C6 1 8 percent to confl icts type C8 and 3 percent t o conflict type C7 These results suggest that left tum in volume has an important impact on DLT movements from a driveway Also, as it was expected from the beginning most of the conflicts are due to the left tum itself. so 43. 2 .1! "' 0 40 30 E z ll. 20 __ ________ _____ J6J_ < "' 0 10 0 cs C6 C7 ca Confliel l)pe Figure 4 9 Average Number of Daily Conflicts by Type, DL T Movement 53


A comparison of the average number of conflicts shown in Figures 4 .8 and 4.9 suggests that the t otal number of conflicts due to DLT movements are 8 percent lower than that of RTUT movements. This res u lt may be misleading because different factors could affect the oc<:urrence of traffic conflicts For example, the volume of RTUT movements is different from that of DLT movements. Hence the use of conflict rates seems to be a more appropriate means to compare the effect of the maneuvers compared in this study. The values presented i n Figures 4.8 and 4.9 only depict w ha t type of conflicts and how many of them would be found by maneuver type. 4.2. Conflict Rates Two types of rates were calculated for the purpose of safety analysis. First the number of conflicts per hour of observat i on was used at each site and then the average value is presented Second the number of conflicts per thousand involved vehicles was calculated and values are presented 4.2.1. Conflicts per hour Figure 4.10 presents the average hourly rate of RTUT conflicts by time period. I nterest ing ly, the number o f conflicts per hour of observation during non-peak hour periods was 10 percent higher than that of peak periods. In addition by ana l yzing the difference s among conflict t ypes, different trends co uld be observed For example conflicts due to Utums were reduced by 65 percent during peak hours, which might be explained by drivers being more careful when selecting an ad equate gap for making the maneuver. As of t he other types of conflicts, C3 decreased by 19 percent, and C2 was red uced by 76 percent whereas conflict type Cl increased by 76 p ercent. On the other band, DLT movements created more conflicts during peak hours, as shov.'II in Figure 4.11 The hou rly conflict rate o f DL T movements was 23 percent higher than that of RTUT conflicts. Confl ict type C5 incr eas ed by J 0 percent, C6 by 2 percent, C7 by 62 percent, and C8 by I 05 percent. It is interesting that conflict type C6 slightly increased during peak hours which may indicate that the effect of left-tum in movements is contingent on vehicles making DL T movements 54


Figure 4.12 shows the comparison of the total average number of conflicts per hour of the two maneuvers In general, the DLT conflict rate was 51 percent higher than the conflict rate of RTUT maneuvers. By desegregating data into peak and non peak periods the c onflict rates of DLT conflicts were 29 and 75 percent higher than those of RTUT maneuver s during non peak and peak hours, respectively 10 9 8 7 u "' 6 .:::! <:: 5 c c 4 v ----ro. -.. .. -.... .. -... .. -O n 0 .29 3 f! 2 > < I I I 0 NON PEAK PEAK TOTAl AVERAGE l Figure 4 .10 Conflicts by Time Period, RTUT Moveme nt. 10 9 8 0 X 7 u e.. 6 5 c c 8 4 -ti I .. 3 u 2 > < I ,. ____ _ ----------------------------------r----r-------------0 NONPEAK TOTAL AVERAGE Figure 4.11 Conflicts by T ime Perio

4.2.2. Conflicts per Thousand Involved V chicles Based on the results of previous studies, the square root of the product of the volumes involved in conflicts was considered the best for calculating the conflict rate The tot al number of conflicts, through traffic vehicles, maneuvering vehicles, and conflict rates were obtai ned for each site Tab l e 4.6 presents the number of conflicts per thousand involved ve hicles at eac h site. The values given in Table 4.6 indicate that except for site 6 the other sites had l ower conflict rates for RTUT movements. Moreover, Tab le 4.6 indicates that th e average conflict rate for RTUT was 38 percent lower than that ofDLT movements. 1 0 9 8 0 7 ::c Q. 6 s <= c 4 0 v 3 2 l > < 0 . . --. ---------------------------T.or ----s:69-------------6 .3> -4.41 ------4 26 __ ...... ...--NON PEAK PEAK TOTAL AVERAGE Figure 4.12 Comparison Average Hour ly Conflicts Between RTUT and DLT Movements by T ime Period. 4.3. Severity Analysis The sever ity of conflicts was analyzed by con s ide ring two approaches A subjective score which was based on the Risk of Collision (ROC) of the maneuver and an objective score which was based on the concept of Tim e to Collision (TTC). Each conflict received ROC and TTC scores in order to evaluate the severity of the conflicts for both RTUT and DLT maneuvers. These scores are given in Tab le 4.7. 56


Table 4.6 Number of Conflicts per Thousand Involved Vehicles. Site DLT RTUT I N/A 19.61 2 39.36 10.12 3 Before 49.11 19.02 3 After NIA 36.80 4 28.46 I 1.69 5 26.90 12.10 6 11.33 17.74 7 26. 12 12.4 1 8 N / A 28.90 Average 30.21 18.71 T able 4.7 Time To Collision (TIC) and Risk Of Collision (ROC) Scores TTC and ROC Scores TIC ROC I 1.51 < TIC seconds Low Risk 2 1.00 -1.50 seconds Medium Risk 3 0.00 -0.99 seconds High Risk 'The severit y of traffic conflicts was determined by the sum of both ROC and TIC scores. Therefore the overal l severity score ranged from 2 to 6. However it was not possib l e to define a TTC value for conflict types C6 and C7 because these maneuvers do not occupy the same path and speed data was not available for calculat ing the TIC value. Therefore conflicts types C6 and C7 were not assigned TIC scores. Hence, two approaches were selected for comparing t h e severity between movements. First, all the co nflicts generated by RTUT and DLT movements were compared based on the ROC scores, and second a comparison o f those conflic t s with both TIC and ROC scores was performed for the two movements. The frequency and cumulative frequency of the severity for each conflict type with ROC and TIC scores were calculated and illustrate d in Figures 4.13 through 4.19. 57


120 40 20 i 1 00% 80% 40% 20% 0 ffi 2 l I S 6 Sever ity IEEiJFrequency Cumulative%; Figure 4 .13 Distr i bution of Severi t y, Contlict Type C I. 120 100% I!Kl 80% 80 >. fi 60 40 600/o 40% 20 20% 0 0% 2 3 4 5 6 Severity l liiiiiFrequency -Cumulative% I F igure 4 .14 Distribution of Severity, Conflic t Type C2. 58


60 I 50 I I 40 I > u c 0 J O i "' I 20 10 0 ...... ................... ___ --------------------.. -----------------2 3 4 5 6Mote Severity I Frequency 0/o j 100'/o 90% 80% 70% 60% 50% 40% 30% 200.4 \0'1 0'/o Figure 4.15 Distribution of Severity, Conflict Type C3 100 100% 90 ____ ____ ............ 90'/o 80 .......... ------------80'/o 70 -----___ _________ ........ 70'/o 1;-60 60% c 50% v 50 = ---------------"' 1! "-40 -----.............. --40% 30 30% 20 ----20% 10 10'/o 0 0% 6 J-Frequency Figure 4 .16 Distribu t ion of Severity Conflict Type C4. 59


70 100% 60 80% 50 >.. 0 40 "' 0" 30 "' 1 .::: 20 I 10 60% 40% 20% 0 0% 2 3 4 5 6 Severicy i EEJFrequency j Fig ure 4 17 Distribution of Seve r ity, Confli ct Type C2U-T. 160 140 120 "' 100 0 l 30 60 "" 40 20 0 2 3 4 5 6 Frequency -Cumulative % l 1000/o 90% 70% 60% SOOA 40% 30% 20% 10% 0% Figure 4.18 Distribution of Severity, Conflic t Type CS. 60


80 10 60 I gso I 40 1t 30 20 10 0 2 3 4 5 6 Se.veri I!!!EJFrequency -Cumulative% 1 70"/o 6GOA,. 50% 40% 30 % 20"/o I 0"/o 0% Figure 4 .19 Dis t ribution of Severity Conflict Type C 8 Based o n the se figures, t h e IS'h, 50th, 85th, and 95th percentile values were calculated for all co n flicts with both ROC and TIC scores. These v a lues are g i ven in Table 4 .8. Table 4.8 S everi t y Pe r centi le s by Move ment and Conflict T ype. Movement Co nflic t Severity RTUT T ype 15t h 50th 85 th 95'" C l 2.0 2.0 3 0 4.0 C2 2 0 2 0 4 .0 5.0 C 3 2.0 3.0 4.0 5.0 C 4 2.0 2 0 4.0 4 .0 C2U-T 2 0 2.0 3.2 4.0 DLT cs 2.0 3 0 4 0 6 0 C8 2 0 3.0 5.0 6.0 Table 4.8 suggests that fifty percent of th e t i me co n flic t type C 3 could b e cons i dered riskier than other confl i cts i nvo l ved in a RTUT maneuve r However eighty five percent of th e time conflict t ype C I and C2UT are considered t o have a medium severity 61


whereas the other conflict types have a severity score of 4 w h ic h might be interpreted as medium. Regarding the sev e ri ty of conflicts associated with DLT movements the results are a lso interesti n g Fifty percent of the time the severity of conflict typ es C5 and C8 will be considered as l ow-medium severity whereas e igh ty five percent of the time conflict type C8 v.ill be riskier than conflict type C5. This may be due to the lack of visibility for drivers waiting for a gap, and that they may get onto the major road by "trial and error". When comparing the severity scores ofRTUT and DLT movements DLT conflicts seem t o have a higher severity. figures 4.20 and 4.2 1 gmphically illus trate the average severity values for both DLT and RTUT movements. Figure 47 illustmtes the overall severity of conflicts based on ROC scores while Figure 48 illustmtes the overall severity of conflicts when both TIC and ROC scores are analyzed. Both Figure 47 and Figure 48 indicated that conflicts gene ra ted by RTUT movements have a lower severity t han confli cts generated by DLT movements. To determine if the sev erit y ofDLT conflicts was significantly different from the severity of RTUT movements, Analysis of Variance ( ANOV A) tests were conducted. ANOV A tests were calculated by using Microsoft Excel s Analysis of Data Tool. The null and alternative hypothes i s, Ho and Ha, for the AN OVA tests were stated as: Ho: Severity ofRTUT movements is not different from that ofDLT movements. Ha: Severity of RTUT movements is different from that ofDL T movements. A 0.05 level of significan ce (a) was selected for the ANOVA test and results are presented in Tables 4.9 and 4 10 Table 14 presents the results when RTUT and DLT maneuvers are compared based on the RO C score, while Table 15 the results of the test when both ROC and TTC scores are considered. 62


3.0 .-----------------, 2.5 2.0 1.5 1.45 1.38 1.0 +---D L T RTUT Maneuver Figure 4.20 Co mpar ison of Overa ll Severity of Conflicts Considering ROC Scores. 6 0 5 5 8 (/} 5.0 (.) 0 4 5 ..: "0 4.0 "' 3.5 3 .0 2 5 ,. -----"3J)7" --------------. --------2 72 < 2.0 DLT R-rur Mane\IVer Figure 4.2 1 Comparison of Overall Seve ri ty of Conflicts Considering RO C and TIC Scores. 63


Table 4.9 ANOV A Test Results to Test Severity Differences Between RTUT and DLT Ma neuvers Considering ROC Score Only. SUNIMARY Groups Count Sum Average Variance RTUT Sever ity 738 1016 1.376694 0.286675 DL T Seve rit y 902 1311 I 453437 0 40571 ANOVA Source of Variation ss df MS F P -value Fcrit Betw een Groups 2 390546 I 2.390546 6.788409 0.009258 3.847134 Within Groups 576 8235 1638 0.352151 To tal 579.214 1639 Tab l e 4.10 AN OVA Test Results to T est Severity Differ e nces B etween RTUT and DLT Maneuve rs Considering Both ROC and TTC Scores SUMMARY Groups Count Sum Aver a ge Variance RTU T Sev erity 738 200 5 2 716802 0.906 122 DLT Severity 529 1623 3.068053 1.29460 2 ANOVA Source of Variation ss df MS F P -value Fcrit Between Groups 38 01631 I 38.01631 35 58679 3.1E-09 3.848811 Within Groups 1351.362 1265 1.06 827 T otal 1389 378 1 266 64


The result of the AN OVA test on the severity of RT UT and DLT movements based on the ROC score indicated that the F value (6.79) at the 5% level of significance was larger than the critical F value (3.84). Thus, the null hypothesis of equal level of severity is The ANOVA test on the severity of those conflicts with both TIC and ROC scores i ndicated similar resu.lts. The F value (35 .59) at the 5 percent level of significa n ce was much larger than the critical F value (3.84) thus, the null hypothesis was rejected too. These results indicate that the .level of severity of DL T conflicts compared to that of RTUT conflicts is higher and significant at a confidence level of 95 percent 4.4. Summary Results indicated that the RTUT technique could reduce both the number of conflicts per hour and the number of conflicts per thousand involv e d vehicles. Furthermore an analysis of the severity of the contlicts generated by both RTUT and DLT movements indicated that the overall severity of conflicts generated by RTUT movementS was lower than that of conflicts generated by D L T movements 65


S. CASE STUDY: BEFORE AND AFTER COMPARISON This chapter presents the resu lts obtained by analyzing tbe conflict data at a site wbere the full median opening was to a directional median opening. A before and after comparison of average number of conflicts conflict rates, and severity le v els of conflicts at the site are presented. 5.1. Descriptive Analysis The site located on U.S. 1 9 an d 116lh Avenue was the subject of a geome tric improvement related to le ft-tum movements from the side street. Initially, the median was a full median opening that allowed vehicles from the local street (116'h) to either tum left directly or to tum right and the n make a U-tum at a median opening loc a ted 400 feet from the street in order to travel northbound. The median opening was approximately 120 feet in length which allowed three or more vehicles to wait in the med ian storage area w hile im pedin g the movement of other vehicles. The median opening was closed and converted into a directional median opening, so tha t vehicles departing from the sid e street could only tum to the right. A detailed anal ysis of the conflicts observed a t this site was conducted because this site allows the possibility to examine and evaluate the implications of changing a full median opening to a directional median opening in the frame of a before and after study without the influence of the site variables. A to tal of 371 conflicts were recorded during the before period and a total of 327 confli cts during the after period After the average number of conflicts was calculated the number of daily conflicts was calculated. Figures 5.1 and 5.2 present the daily number of conflicts for before and after time periods. As it was expected the number of confl icts due to RTUT maneuvers would increase since more vehicles perform such maneuvers. However, the total average numbers of daily conflicts during before and after time periods were 238 and 140 conflicts respectively, which was a reduction o f 41 percent. This suggests that even if the number of conflicts due to RTUT movements has increased the application of RTUT instead of DLT maneuver ind eed reduced the total number of conflicts. Although the comparison of Figures 5.1 and 5.2 provide important information 66


regarding the reduction in the daily number of conflicts it is much more meaningful to use conflict rate s for eval u ating the changes because they can take i nto account the fac tors such as corresponding tra.ffic volumes th at generate conflicts, as well as its effect during different time periods. 140 120 ------------------) 17 s I()() c 6 80 B a 0 40 ;< 20 .&.5_-JQ.l_ -----4.0 2 5 0 C l C2 CJ C4 cs C6 C7 CS C2UT Figure 5.1 Number of Daily Contlicts by Typ e Before the Improvement. 140 I 120 i 3 *

5 2. Con flict Rates 5 2. 1. Conflic t s per H o u r Table 5.1 presents conflict rates corresponding to the vehicles turning to the left during the before period. As expected the largest number of conflicts was due to the left t u m movement itself (conflict type CS) followed by conflicts due to the interaction between left tum in and left tum out movements (conflict type C6) conflic t types C8 and C7 These fmdings suggested that the influence of left tum in vehicles on vehicles coming out of the driveway was important Number of conflic t s per hour during peak and non peak periods were 23.92 and 19.00 respective l y whereas the mean total contlict rate was 21.46 The data o n DL T conflicts are not available for the after period as that movement was prohibited. Table 5.1 Number of Conflicts per Hour for DLT Movements during Before Period Time Period Conflict Type Total C5 C6 C7 C8 (confliccs/hour) Before Non Peak Rate 10.60 4 .60 1.00 2.80 19.00 Peak Conflict Rate 10.13 5.38 1.46 6 .96 23.92 Total Confltct 1 0 36 4.99 1.23 4.88 21.46 Rat e Table 5.2 and Figures 5 3 through 5.5 present conflict rates for each type of conflict by t ime period and th e total for the before and after periods. Values given in T able 5.2 indicate that conflicts caused spedfically by the U-tum maneuver drastically increased after the impwvement due to t h e increase in the RTUT vo l ume. Total average number of RTUT conflicts per hour during before and after time periods were 3.72 and 12.60 respectively. However i t was interesting to n ote that the conflict rate during peak hours was about 9 percent lower than that of non-peak hours for the after period. This may be because U-tum conflicts (C4), and merging conflicts (CI) may decrease due t o drivers 68


being more careful in selecting gaps when the thro ugh traffic volume on the arterial is high. Tab l e 5.2 Before After Figure 5.3 Number of Conflicts per Hou r for RTUT Movements during Before and After Periods Time Period Conflict T ype Total NonPeak Rate Peak Conflic t Rate Total Conflict Rate Non-Peak Rate Peak Conflict Rate Total Co nflict Rate i g 5 J 4 Iii) 3 l'

Figure 5.4 Figure 5.5 30 = .:! 25 -"' > .0 0 20 "-0 :> 0 J: 15 1 0 -<:: = 0 u 5 0 -------------.... ---.. ----------------------------2 :su -_ ) ,Q9 ... 4.60 10.60 696 1.46 5 .38 1 0.13 4 .88 !.23 4.99 --l, NON PEAK P EA.!( TOTALAVERAGE i CS 1!!JC6 oC7[;!C8] J Conflicts per Hour of Observation by Time Per iod DLT Movements Before Improvements. 30 .2 25 ;; > .0 0 20 .... 0 15 :> 0 J: r---, 1.48 ;<>.. 10 5.35 .. --3 39 I -" ;;:: 8 5 0 1.26 1.67..,__ NONPEAK PEAK TOTAL AVERAGE Conflicts per Hour of Observation by Time Period-RTUT Movem e nts After Improvements. 70


The comparison between the total number of conflicts per hour during before and after time periods and the reduction in the number of conflicts given in Table 5.3 suggested an importan t fact. The number of conflic ts per hour wa s re d uced by 41.3 percent when the DLT technique was prohibited by forcing such drivers to make RTUT Table 5.3 Comparison o f Total Number of Conflicts per Hour During Before and After Time Periods Number of Conflicts/Hour N umber of Conflicts/Hour Reduction % During \be Before Period D uring the After Period Due to Due to Total Due to Due to Total {Before-After) x 100 DLT RTUT DLT RTUT Before Peak Peri od 23 .92 3.20 27.12 0 13.14 13.14 51.5% Non Pea k Period 19.00 4.25 23 .25 0 12.06 12 .06 48 .1% Total Average 2146 3.72 25.18 0 12.6 0 12.60 49.9% 5.2.2. Conflicts per Thousand Involved Vehicles The average number of conflicts and traffic volwnes during different t i mes of the day were calculated for bolb before and after periods which were \be basis for the development of conflict rates. The conflict rate was calculated by dividing the average number of conflicts by \be square root of \be product of the volwnes involved since it had a better correlation based on findings of severa l previous studies (17, 26 36). T wo different melbods were used for ca lc ulating lbese val ues. The first method involved the calculation of the confl ict rate for every conflict type for each hour of the day. Then, for e ach type of conflict \be number of conflicts was divided by \be square root of \be product of the major road and the maneuver volume. After the rate was calculated for every conflict type, the rate for a determined hour was calculated as \be summation o f the conflict rate for the types of conflicts involved. In other words, the hour ly rate for conflicts associated with DL T movements was calculated as the summation of the rates 71


of conflicts types C5, C6 C7, and C8, whereas the rate for conflicts related to RTUT m ove men ts co rresponded to the summation of conflicts types C1, C2 C3, C4 and C2U T T h en, an average value of all the hours was calculated and taken as the final oonflict rate. Table 5.4 presents the conflict rate for each and every hour and the average value for before and after time periods estimated by using this method The seoond method consisted o f obtaining the total average number o f confl i cts that occurred during an eleven-hour day for each conflict Then, these values were d i vided by the square root o f the product of the total volume on the arterial and the total volume of the ma n e u ver in the day T able 5 5 presents the values calculat ed by using this method Since values obtained by using both methods were v ery si m ilar t o each o ther it was assumed that both methods could be applied for calculat i ng the number of conflicts per thous and involved vehicles. Table 5.4 Average N u mbe r of Conflicts per Thousand Involved Vehicle s, Method I Conflicts Per Thousand Involved Vehicles Time Before After D LT RTUT Both RTUT Types (3) 7:00 8:00 10 3.45 15.26 118.71 26.82 8:00 9:00 71.73 26.21 97.94 27.85 9:00-10:00 40.80 13. 88 5468 37.08 l 0:00 11: 00 35 1 1 20.20 55.31 53 63 11:00 -12:00 32.38 18.62 51.00 61.96 12:0013:00 51.70 21.94 7 3 .64 27.48 13 :00-1 4:00 57.05 5.38 62 .43 32.82 14 :00 15:00 43 90 11.72 55.62 44.86 15:00-16:00 37.85 34.37 72.22 34.49 16:0017:00 51.35 26.19 77.54 4 6.28 17:00 18:00 51.91 1 3 68 65.59 28.04 AVERAG E 52.47 18.86 7 1.33 38.30 72


Table 5.5 Average Number of Conflicts per Thousand I nvolve d Vehicles Method 2 Conflic t ConJlicts Per Thousand Involved Vehicles Type Before After DLT RnJT Both RTUT Types Cl N / A 4.15 4 .15 4.35 C2 N /A 6.51 6 5 1 4 07 C3 N/A 5.20 5 .20 9.82 C4 N / A 1.95 1.95 14.00 C5 26.15 N!A 26.15 N/A C6 9.90 NIA 9.90 N/A C7 3.30 N/A 3.30 N/A C8 9.76 N I A 9 .76 N/A C2UT N/A 1.22 1.22 4.57 TOTAL 49.11 19.02 68.13 36.80 To determine i f the conflict rat es of before and after periods are significantly different a / -tes t was conducted using the values in Table 5 4. This statistical test could be performed eve n if the underlying distr i bution does not follow the Normal Distribution Nevertheless, if a / -test is going to be conduc ted, it mus t first be verified whether the variances of the samples arc s igni ficantly different as the type of t-test depends on the variances of the two populations. This was evaluated by apply ing tbe variance rat io tes t, known as the F test. T he n u ll and alternative hypotheses in this test were: variances of the two samples belong to the same population Ha : N u ll hypothesis is not true. A level of significance selected for this test was 5 percent. T able 5.6 presents the r es u lts of the F test, based on the values given in Tabl e 5.4 for before and after time periods. Results s hown in Table 5.6 indicate that at 95 percent confidence level the F stat is tic for the test was 3.050 whereas the critical F statistic was 2.978 The p-value for the test was 0 046 indicating tha t the null hypothesis cmmo t be accepted at a 5 percent level of significan c e or the difference between variances was significm1t. Therefore the /test was 73


conducted to compare the two mean conflict rates considering unequal variances If the ca l cu l ated t value at a given level of significance is greater than the critical t, the null hypothesis is rejec ted and it could be concluded that the difference is significant, otherwise the differenc e is not significant. The nu ll and alternativ e hypothesis for the 1 test were as fo llows: H o : The conflict rates during b efore and after periods are equal. H : The null hypothesis is not true T able 5 6 Results of the F Test to Check th e Variances of Conflict Rates During Before and After Periods F-Test Two Sample for Variances Parameter Before After Mean 71.33 38. 30 Variance 430.38 1 41.09 Observations 11 1 1 F 3 050 P(F <=f) one-tail 0.046 F Critical one-tail 2.978 Table 5 7 presents the results of the t-te s t for comparing the mean conflict rates during before and a fter time periods, assuming unequa l variances. Since tbe value of the calculated t statistic (4.58) was much great e r than the critical 1 statistic with 5 p e rcen t level of significance ( 1.75), the null hypothesis was rejected and the differences between the conflict rates could be cons i dered significant l y different at the 95 percent confidence level. Therefore prohibition of DL T through the median closure where sucb movements were replaced by RnJT reduced the number of conflicts as well as confl ict rates thereby improving safety 74


Table 5.7 Results of the /-test Comparing Differences Between Conflict Rates During Before and After Periods !-Test: Two-Sample Assumi n g Unequal Variances Parameter Before After Mean 71.33 38.30 Variance 430.38 141.09 Observations II 11 t Stat 4.583 P(T <-t) one-tail 0.0001 t Cri tica l one-tail 1.745 5.3. Severity Analysis As previously mentioned, the severity of each conflict was obtained and analyzed usin g two approaches for before and after periods Table 5 8 and 5.9 present the 15'", 50'", 85th, and 95th percentile severity levels calculated for all conflicts by using the index that used both ROC a n d TIC scores for before and after periods. T hey suggest that 50% of the time conflicts related to RTUT maneuvers w ere more severe than thos e relate d to DL T maneuvers. However a different outcome can be expected 85% of the time because the severity of DLT conflic t s seems t o be hig her than that of RTU T conflicts Table 5.8 Percentiles of Severity for DL T and RTUT Maneuvers for Before Period Movement Conflict Percentile RTUT Type 15th 50t h 85"' 95th Cl 2.0 2 0 4 0 4.0 C2 2.0 2.0 3.0 4.2 C3 2.0 30 4 0 4.0 C4 2.8 3.5 4.3 4.8 C2U T 4.0 4.0 4.0 4.0 DLT C5 2.0 3.0 4.0 5.5 C8 2.0 3 .0 5.0 6.0 75


Tab le 5 .9 Percen t iles of Severity for RTUT Maneu vers for After Period Movem en t Conflict Percentile RTU T T y p e 15th 50th 85th 95th Cl :?. 0 2.0 3.0 4.0 C2 2 0 2.0 2.8 4.0 C3 2 0 3 .0 4.0 5 .0 C4 2.0 3 0 4 0 4 .0 C2UT 2.0 2.0 3 .0 4 0 Tab l e 5 9 presents severity values for the conflicts that occurred after t he impr oveme n t w as done. In general it seems that the sev erity of RTUT c on flic ts is the same as that o f th e before period Neverthe le ss, the severity of conflict t ype s C3 (lane change confl ic t ) increased to a medium-high level and the severi t y of U -tums decreased to a mediwn level. Analysis o f Variance (ANOVA) tests we re conducted i n or der to determine if there w as a significant difference in the seve rity of conflicts before and after the i m p rovemen t The n u ll a nd alternat i ve hypothesis, Ho an d Ha, for the ANO V A tests wer e : Ho: Severity l eve l s ofRTUT m oveme nts is si m ilar to that of DLT movements Ha: Severity levels ofRTUT movements is different f rom that ofDLT move ments A 0.05 level of sig n ifi cance was sel ected for the ANO V A test, and th e resu lt s are p r e se nte d in Tabl e 5.10. The result of the ANOVA tes t on the severity ofRTUT and DLT movemen t s based on the ROC sco re i ndicates that the F v al ue (2 5.8 3) was larger than the critical F value (3.85) a t the 5 percent lev e l of sign i ficance Thus the d iffer e n ce is h ighly significant and null hypothes i s of equal le v el of seve rity i s reje cted. In other words the severity levels of DLT and RTUT are differe nt when cons i dering the ROC score and therefore by replacing DLT by RTUT the serio u sness of con fli cts cou ld be red u ced according t o the findings at this site. 76


Table 5.10 ANOV A Test Resul ts to Test Severity Differences Between RTUT and DL T Movements Cons i dering ROC Score SUMMARY Groups Count Sum Average Variance DLT 313 497 1.587 86 0.44819 RTUT 327 441 I .34862 0.26459 ANOVA Source (I( Variation ss df MS F P-value Fcril Between Groups 9 .153 I 9.153 25 8286 4 9E -07 3 .8 5607 Within Groups 226.091 638 0.35437 Total 235.244 639 Table 5 .11 presents results of the ANOVA test on the severity of DLT and RTUT conflicts by using both TIC and ROC scores. The F value for the test (89. 09) at the 5 percent leve l of significance was much lar ger than the critical F value (3.85) and also, the p statistic was much lower than the selec ted 0 05 level. Thus the null hypothesis of equal sever ity l evels was rejected and the difference could be considered to be signifi can t. These results i n d i cate that the level of severity of RTUT conflicts as compared to tha t of DL T coni l icts was lower when considering both ROC and TfC scores, which was significant at a confidence level of95 percent. According to the s e findings it could be seen that irrespective of the type of index used in evaluating the severity, whether i t is ROC or ROC and TIC, severiti e s of DLT conflicts were m uch higher than those of RTUT conflicts at the site where the before and after study was conducted. This is an i n dication that the se ri ousness of traffic con flicts occurring at a site could be reduced by replacing DLT with RTUT. 77


Table 5 .11 ANOV A Test Results t o Test Sev erity Differences Between RTUT and DL T Movements by Considering both ROC and TTC Scores, After Improvem en t SUMMARY Gro u ps Count Sum Average Variance DLT 224 408 1.82143 0 43434 RTUT 327 441 1.34862 0.26459 ANOVA Source of Variat ion ss df MS F Pvalue F crit Bet ween Groups 29. 7172 I 29.7172 89.0961 l.IE-19 3.85845 With in Groups 183.114 549 0.33354 Total 212 .831 550 I t was also interesting t o test if there was a signific ant difference in the severity ofRTUT conflicts before and after the improvement. This was b e cause i t could be expecte d that after the geo m etric change was done more vehicles would make a RTUT movement making it possible to change the severities. An ANOV A test was performed a t the 5 percent leve l of significance to test whether there was a s i gnificant difference and results are given in Table 5.12. For this case, the F statis t ic for the test was 0 .7 9 and the critical F statistic was 3.86 Also the p-value for the tes t was 0.37 which is much larger than the 0 05 l evel selected for t he tes t. Therefore it could be concluded that the differences in the severity levels of RTU T conflicts before and after the geometric change were not significant at the 95 percent confidence leve l. 5.4. S ummary A before and afte r comparison was perfonned to evaluate the safety effects of closing a full median opening, thereby f or c ing drivers from t he side street to make right turns. Results indicated that the n umber of conflicts and the conflict rates were significant l y r e duced by imp l ementing the RTUT technique. Moreover the severity of the conflicts 78


generated by RTUT movements was low er than that of conflicts generated by D L T movements. Table 5 1 2 ANOVA Test Res ul ts t o Tes t Severity Differen ce s for RTUT Conflicts B efore and After Improv ement SUMMARY G r oups Co unt Sum Averag e Variance RTUT Be f ore Improvement 58 82 1.41379 0. 2 468 2 RTUT After I mprovement 327 441 1.34862 0.26459 ANOVA Source of Variation ss df MS F P-value F c rit Between Groups 0.20922 I 0.20922 0 7 987 1 0.37204 3 86585 Within Groups 100.326 383 0.26 195 Tota l 100.535 384 7 9


6. SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 6.1 Summary Transportation engineers and planners hav e used access management to imp rove operational and safety conditions of the road sys tem. One of the objectives of access management is to reduce the n umber of conflict points In particular, access management act ions often seek to m in imi ze direct left tum (DLT) movements from driveways as they generate many conflict points and increase th e potentia l for traffi c crashes. Medians are used t o rep lace DL T movements wit h right tum movements followed by U-tums (RTUT). This report is one of three reports that evaluated the safety and traffic operational effects of direct left turns vers us right tums followed by U-tums from driveways or side streets This res earc h focused on evaluating the safety of RTUT and DLT movements using traffic conflicts. A tota l of ten sites were selected for conducting the investigation. However, two sites on f our -lane ro a d s were disregarded because they did not m eet the sample size required for evaluating these maneuvers. The average number of daily traffic conflicts was calculated for each site and evaluated by conflict type. Nine types of conflicts were selected for this study, five of t hem corre spo nded to con flicts generated by RTU T movements while the rest corresponded to DLT movement s I t was found th at the most commo n conflict type generated by DLT movements was the left-tum maneuver itself (Conflict type C5) followed by c onflic ts between left-tum and left-t urn in move men t s (Confl ic t type C6), then by conflicts of left tunl veh ic l e s departing from the median storage area, and finally, the conflicts betwee n DL T and l eft turning vehicles into a driveway located on the opposi te s ide of the driveway studied. These values suggested tha t left tum in volume had an impo rtant impact on vehicles making a DLT m o vem e nt from a driveway The confli ct distribution for RTUT movements was s ha red almost equally by a ll the conflict types. Most of the conflicts occurr ed because of the ri ght tum maneuver (Conflict type C I), followed v ery closely by ch anging lanes (Conflict typ e C3), then by conflicts 80


created b y slow U-turn vehicles on the m ajor road (Co nflict type C2U T) and finally a right turning vehicle slow on the same direction A comparison of the total average number of conflicts per h ou r of the two maneuvers showed that RTUT c onfl icts had a conflict rate 3 4 percent lower than that of DL T movements. When the data were analyzed by time period, that is peak and non-peak h ours, the RT U T conflicts wer e slig htl y higher during non-peak hours ; whereas DLT conflicts where higher during peak hours. The n u mber of conflicts related to DL T movements was 29 percent higher than t hat ofRTU T movements w hile during non peak hours it was 75 percent h igher. In addition a comparison of the number of conflicts per thousa nd invo lved vehicles was performed. Results indicated tha t RTUT movements generated 38 percent fewer conflicts per thousand vehicles than DLT movements. The severity of conflicts was analyzed by means of subjective and objective m ethods. Two types of analyses showed that the severity of conflicts caused by RTUT movements w as lower than that ofDLT conflicts at the 95 percent confidence level. The before and after comparison showed that there was a significant difference betwee n the conflic.t rates of RTUT and DLT maneuvers. Additionally, the analysis of the severity of RTUT conflicts and DLT conflicts showed a significant difference at the 95 percent co nfidenc e level. On the other hand, a comparison of the severity of conflicts caused by RTUT maneuvers before and after the improvement showed that the difference between them was not significant. 6.2 Conclusions The analysis of RTUT and DLT from driveways using traffic conflicts resulted in several conclusions. These are presented in the following paragraphs. A comparison of the numb e r of conflicts per hour of RTUT and DLT movements suggested tbat RTUT movements indeed generate f ewer conflicts per hour. In addition when the effect of traffic volume was accounted into the comparison by means of a conflict rate, resu lts also showed the effe ctiv eness of RfUT movements in reducing the number of conflicts on six-lane roads. 81


Not only did RTUT movements reduce the number of conflicts, but also the severity of them. Findings of this study indicate d that RTUT conflicts had an overall severity significantly lower t han that ofDLT movemen ts. F inally based on the b efore and after comparison, rep l acing the DLT with RTUT is an appropriate method tor improving t he safety of road I!Scrs. 6 .3 Rec ommendat ion s It would be useful to continue t h is research to evalua te t he RT UT tech nique on four la ne roads. Also more relationships between conflic t s gen erat e d b y RTIJT movements aod other geometric characteristics should be studied such as weaving length and median width This study is limi ted t o loc a tions where the U-tum is made at median openings. It c ould be extended to evaluate the U-tums at Sib'Tlalized intersections and to address the effect of signal timing on RTUT movements. Be tte r signal timing may allow more vehicles to tum right and make a U -tum while preservi n g the operational and safety characteris tics of the road sys t em. Also it is advisable to check the crash history of the site where the case stud y was p er fonned. In a ddit io n another traJlic con fl ic.t study should be performed again at that site, as t his would h elp to evaluate if there has been a significant change in the number of traffic conflicts g ene r ated by vehi cles making RTUT movements. 82


REFERENCES I Institute of Transportation Engineers (1999). Traffic Engineering Handbook. 511' Edition pp. 306-347 2 Koepke, F., and Lev i nson H. (1992) NCHRP Report 348: Access Management Guidelines for Activity Centers. Transportation Research Board, National Research Co uncil, Washington D.C 3 Ame rican Association of State Highway and Transportatio n O f ficials (198 4). A Policy on the Geometric Design of Highways and Streets. pp 841. 4 National Highway Ins titute, NHI Course No. 133078 (2000) Access Management, Location and Design Participant Notebook. 5 Pres t on, H., K e lmer, D., Newton R. and Albrecht, C (1998) Statistical Relationship Between Vehicular Crashes and Highway Access Final Report 1998-27, Minnesota Department of Transportation Office of Research Serv i ces, Minneapolis Minnesota 6 Bonneson, J.A. and McCoy, P T (March 1998). Median Treatment S e l e ction for Existing Arterial Stree ts, /nsfitute a/Transportation Engineers Joumal, pp. 26-34 7 Levinson, H. S., (I 996). Restr i c t ive Medians and Two Way L eft Tum Lanes: Some Observations, Pro c eedings of the Second National Conference on Access Management. 8 Stover, V.G. (1994) I ssues Re l a t i n g t o Weaving on At-grade Roadways and the Right tum Followed by U-tum Maneuver: A Discussion Paper'', Center for Urban Transporta t ion Resea r ch. 9 Gluck, J., Levinson, H.S. and Stover, V .G. (I 999) Impacts of Access Management Techniques National Cooperative Highway Research Program Report 420, Transportation Research Board, Na t ional Research Council Washingt on, DC. 83


10 Adams, J.C. and Hummer, J.E. (1993). "Effects ofU-Tums on L eft-Tum Saturation F low Rates", Transportation Research Record J 398, T ransportation Research Board National Research Counci l Wa s hington, D.C pp. 90-100. II Gwyru1, D W. Jr., ( 1995). "Public I nformation Meetings for Access Managemen t Projects", bnp://www t .gov/reales t ate / am d5pu h .pdf. Ac c ess e d February 5,2001. 12 Brown, G.R. (1994). "Traffic Conflicts for Road Safety S tu dies" Canad i an Jo u rnal of Civil Enginee ring, Vol. 21, No I, pp.l-15. 1 3 Pa r ker, M.R., and Zegeer, C.V ( 1 989). "Traffic Conflict Tec h niq u e for Safety and Operations-Engineer's Guide" FHWA-IP-88-26 Fede r al Highway Administration U .S Department of T ranspo rt ation W a s h ington D.C. 14 P a r ker M R and Zegeer, C .V. ( 1 989) "Traffic Conflict Technique for Safety and Operations-Observer s Man u a l FHW A IP-88-27, Feder a l H i ghway Ad mi nistration, U. S. Department ofTransp onation Washington, D.C. 1 5 Amundsen, F., and Hyden, C. G. ( 1 977) Proceedings of Firs t Workshop on Traffic Conflicts Oslo, I n stitute ofTransport E conomics 1977. 16 Hauer, E (1978). "Design Consid e rations on T r affic Conflict Sutveys", Transportation Research Record 667, T rans p ortation Research Board, Natio n al Research Council, Washington, D C., pp 57-66. 17 Katamine N.M., and Hamarneh J.M. (1998) "Use of the Traffic Conflict T echn i que to Identify Hazardo u s I n tersections". Australian Ro a d Research Board, Road and T r ansportation Research, Vol. 7, No.3 pp. 1735. 18 Salman, N K and A lMaita, K J. (1995). "Safety Evaluation a t the Three leg, Unsignalized I ntersections b y Traffic Co n flict Technique" Transportation Research Record 1485, T r ansportation Research Board, Na ti onal Research Cou n cil, Wash i ngton, D C., pp. 1 77-185. 84


1 9 Zegeer, C.V., Deen, R C (1978). "Traffic Conflicts as a Diagnostic Tool in Highway Safety" Transportation Research Record 667, Transportation Research Board, National Research Council, Washington D.C. pp. 4 8 55. 20 Sayed, T ; Brown, G; and Navin F.' (1994) "Si mulation of Traffic Conflicts a t U nsig n alized Intersections with TSC-Si m Accident Analysis and Prevention, Vol.26, N o.5 1994, pp 593 -6 07. 21 Chin, H.C., and Quek S. T. (1997). "Measurement of Traffic Conflicts", Acciden t Ana ly sis and Prevention, Vo1.26, No.3 pp. l 69 185. 22 Say e d, T., a n d Zein, S. (1999). "Tr affic Conflict Standards for Intersections", Transportation Planning and Technology. Vol. 22, No.4, pp 309-323 23 Katamine, N.M. (2000). ''Nature and Frequency of Secondary Conflicts at Unsignalized Interse ct ions" Journal of Transportation Eng i neering Vol. 12 6 No. 2 pp. 129-132. 24 Torbic, D Borkowski, J E l eft e r i adou, L and McFadden, J (1998). "Relationships Between Traffic Operations and S a fety at Signalized I n t ersections P r oceedings of the Third In t ernational Symposium on Highway Capacity, Copenhagen Derunark. 25 Migletz, D J Glauz, W. D. and Bauer K.M (1985) "Relationships Between Traffic Conflicts and Acc i dents", Report No. FHW A/R.D-84 / 042 Federal Highway Adminis t ra t ion, U.S. Department of T ransportation, Washington, D.C. 26 Engel, U. (1985). "To W h a t Ext ent Do Conflict Studies Replace Accident Ana l ysis: Valida t ion of Conflict Studie s -An In t ernational Review. I n Organisme National De Securite Routiere. Proceedi n gs "Evalu ation 8 5", Paris, France Vo1.2, pp. 324-343. 27 Glauz, W. D Bauer, B. M., and Migletz, D. J. (1985). "Expected Traffic Conflict Rates and Thei r U se in Predicting Accidents" Transpor t ation Research Record 1026, Transportat ion Research Board, Nation al Research Counci l Washingto n D.C., pp. l-12. 85


28 Hayward, J.C. (1972). "Near-Miss Determination Through Use of a Scale of Danger". Highway Research Board Record No. 384, Washington D C., Highway Research Board, pp 24-33. 29 Allen, B.L., Shin, T., and Cooper, P.J. (1978). "Analysis of Traffic Conflicts and Collis i ons", Transportation Research Record 667, Transportation Research Board, National Research Council, Washington, D .C., pp. 67-73. 30 S hinar, D. (1984). "The Traffic Conflict Technique: A Subjective Vs. Objective Approach" Journal of Saf ety Research, Voll5, pp. 1 53-157. 31 Kmysse, H.W. (1991). "The Subject iv e E valuation of Traffic conflicts Based on a Interna l Concept of Dangerousness", Accident Analysis and Prevention, VoL 23, No I pp. 53-65. 32 Robertson, H.D., Hummer, J. E., and Nelson D C. (1994) "Manual o f Transportation Engineering Studies, Chapter 12" Institute of Transp ortation Engineers. Prentice Hall, NJ. 33 Tarrall M. B., and Dixon, K K. (I 998). "Conflict Analysis for Double LeftTum L anes W ith Protected Plus-Permitted Signal Phasing" Transportation Research Record 1635, Transportation Research Board, National Research Council, Washington, D.C ., pp. 1-19. 34 Sayed, T., and Rodriguez, F. (1999). "Accident Prediction Models for Urban Unsignalized Intersections in British Columbia" Transportation Research Record /665, TranspQrtation Research Board, National Research Council Washington, D.C., pp. 93-99 35 Crowe E C (1989). "Traffic Conflict Values for Three-leg, Unsignalized Intersections", Transportation Research Record 1287, Transportation Res ea rch Board National Research Council Washington, D.C., pp. 185-194. 36 Canavos G C. (I 984) Applied Probability and Statistical Methodr, Litt le Brown and Company Boston, Toronto 86


37 Sayed, T. (1997). "Estimating the Safety of Unsignali z ed Intersect ions Using Tra ffic Conflicts", Proceedings of Third In ternational Symposium on In tersec t ions Without T raffic Signals P ortland Oregon, pp. 230 235 87