Analysis of right-turn merging conflicts on channelized right-turn - PowerPoint PPT Presentation

Analysis of right-turn merging conflicts on channelized right-turn lanes at signalized intersections Presenter: Tim De Ceunynck (tim.deceunynck@uhasselt.be) Authors: Tim De Ceunynck, Stijn Daniels, Elke Hermans, Annelies Geussens, Tom Brijs, & Geert Wets Hasselt University – Transportation Research Institute (IMOB), Belgium Aliaksei Laureshyn Lund University - Traffic & Roads, dpt. of Technology and Society, Faculty of Engineering LTH, Sweden

Background � Channelized Right Turn Lanes (CRTL) often applied to improve mobility � Road designers believe they improve road safety � However, little research about safety performance of right turn lanes (focus is usually on left turn lanes) � Especially few guidelines about most appropriate type of traffic control at CRTL � More research in this field needed (Al-Kaisy & Roefaro, 2010) De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Main research question? � Differences in safety performance between yield-controlled CRTL and signal-controlled CRTL? De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Background � Features yield-controlled CRTL (YCRTL): � Avoids unnecessary waiting � Road user needs to select own gap � Features signal-controlled CRTL (SCRTL): � Most interactions are separated in time � Road user does not need to select a gap De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Methodology � Semi-automated traffic conflict observation � Look at serious conflicts as surrogate for analysis of crash data � Steps: 1)Collect video data 2)Manual pre-selection of dangerous event 3)Software-assisted calculation of vehicle trajectories � conflict indicators are measured in a more objective and reliable way than manual observation (Laureshyn, 2010) De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Methodology 1. Collect video footage � 2 intersections on ring road around Hasselt � Very similar in lay-out and purpose in network; comparable AADT (but differences in volumes per direction) � Take volumes per direction into account when quantifying number of conflicts De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Methodology 1. Collect video footage (ct’d) SCRTL = 2 lanes; YCRTL = 1 lane � � in line with design guidelines (SCRTL) (YCRTL) De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Methodology 2. Manual pre-selection of dangerous events 1 full week of video for each location is � viewed All merging events that are considered � dangerous are selected (with large safety margin) 3. Conflict severity measurement Software-assisted estimation of road users’ � trajectories Automated calculation of time to collision � (TTC) Calculation of TTCmin, Post Encroachment � Time (PET) and Time-to-Accident (TA) De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Results � 249 situations pre-selected and measured � SCRTL: 126 � YCRTL: 123 � Only severe situations included in further analysis � Selection criteria: � TA = serious (as defined by STCT) � SCRTL: 4 � YCRTL: 10 � TTCmin ≤ 2 sec � SCRTL: 19 � YCRTL: 13 � PET ≤ 1 sec (only in case of yield violation!) � SCRTL: 22 � YCRTL: 29 De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Research question 1 � QUESTION: Which type of CRTL leads to most dangerous situations? � Simply compare the number of dangerous situations at each of both locations… …no? De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Results � At which type do most serious conflicts occur? � Correct for exposure: � # right turning vehicles per week at SCRTL: 8 708 � # right turning vehicles per week at YCRTL: 33 852 � Χ ²-test and adjusted standardized residuals applied � TA value: � No significant difference in number of conflicts between both types SCRTL YCRTL Serious conflicts 4 10 No serious conflicts 8.704 33.842 p-value 0,504 De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Results � TTCmin and PET: � Proportion of conflicts is higher at SCRTL if we correct for the number of right turning road users Absolute numbers Adjusted standardized residuals SCRTL YCRTL SCRTL YCRTL PET ≤ 1sec 22 29 4,0 -4,0 No PET ≤ 1sec 8686 33823 -4,0 4,0 p-value <0,001** Absolute numbers Adjusted standardized residuals SCRTL YCRTL SCRTL YCRTL TTCmin ≤ 2sec 19 14 5,3 -5,3 No TTCmin ≤ 2sec 8689 33838 -5,3 5,3 p-value <0,001** De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Results � 2 out of 3 indicators indicate that relatively more right turning vehicles are involved in conflicts at the SCRTL � Conclusion: SCRTL is more dangerous than YCRTL… … no? De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Research question 2 � QUESTION: With which type of conflicting traffic stream do most dangerous situations occur at each of both types of locations � EASY: Simply compare number of dangerous situations from each type that occur at both locations… … no? De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Results Serious TA Adj. st. residuals conflicts SCRTL YCRTL SCRTL YCRTL Straight 0 8 -2,7 2,7 through Left turn 0 2 -1,0 1,0 U-turn 4 0 3,7 -3,7 N (total) 4 10 p-value <0,001** PET close Adj. st. residuals TTC min close Adj. st. residuals encounters encounters SCRTL YCRTL SCRTL YCRTL SCRTL YCRTL SCRTL YCRTL Straight 0 6 -2,3 2,3 Straight 1 7 - 3,1 3,1 through through Left turn 1 21 -4,8 4,8 Left turn 0 6 -3,3 3,3 U-turn 21 2 6,3 -6,3 U-turn 18 0 5,3 -5,3 N (total) 22 29 N (total) 19 13 p-value <0,001** p-value <0,001** De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Results � Correct for exposure: � Exposure to number of vehicles from the conflicting direction is more relevant here SCRTL YCRTL Straight through 35 604 63 937 Left turn 8 594 15 802 U-turn 1 365 168 De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Results Serious TA Adj. st. conflicts residuals SCRTL YCRTL SCRTL YCRTL Straight Conflicts 0 8 -2,1 2,1 through No conflicts 35.604 63.929 2,1 -2,1 TTC min close Adj. st. p-value 0,057* encounters residuals Left turn Conflicts 0 2 SCRTL YCRTL SCRTL YCRTL No conflicts 8.594 15.802 Straight Conflicts 1 7 through No conflicts 35.603 63.930 p-value 0,544 p-value 0,730 U-turn Conflicts 4 0 No conflicts 1.365 168 Left turn Conflicts 0 6 -1,8 1,8 No conflicts 8.594 15.798 1,8 -1,8 p-value 1,000 p-value 0,097* U-turn Conflicts 18 0 No conflicts 1.351 168 p-value 0,248 PET close Adj. st. encounters residuals SCRTL YCRTL SCRTL YCRTL Straight Conflicts 0 6 -1,8 1,8 through No conflicts 35.604 63.931 1,8 -1,8 p-value 0,095* Left turn Conflicts 1 21 -3,0 3,0 No conflicts 8.593 15.783 3,0 -3,0 p-value 0,003** U-turn Conflicts 21 2 No conflicts 1.348 166 p-value 0,770

Results � When we look at the proportions of dangerous situations of each intersection individually, we see that � At the SCRTL � U-turn conflicts are overrepresented for every indicator � Conflicts with straight through traffic are less common than expected � Inconclusive about conflicts with left turning vehicles � At the YCRTL � No significant difference in proportion of conflicts for TA values � According to PET and TTCmin: � Overrepresentation of left turn conflicts � Underrepresentation of straight through conflicts compared to � contradicts with previous analysis that showed more straight through conflicts at YCRTL…?

Results � HIDDEN Serious TA conflicts SCRTL Adj. st. residuals Conflicts No conflicts Conflicts No conflicts Straight 0 35.604 -3,8 3,8 through Left turn 0 8.954 -1,0 1,0 U-turn 4 1.365 11,4 -11,4 p-value <0,001** YCRTL Conflicts No conflicts Straight 8 63.929 through Left turn 2 15.802 U-turn 0 168 p-value 1,000 De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Results � HIDDEN TTCmin close encounters SCRTL Adj. st. residuals Conflicts No conflicts Conflicts No conflicts Straight 1 35.603 -7,7 7,7 through Left turn 0 8.954 -2,1 2,1 U-turn 18 1.365 23,4 -23,4 p-value <0,001** YCRTL Adj. st. residuals Conflicts No conflicts Conflicts No conflicts Straight 7 63.930 -2,4 2,4 through Left turn 6 15.798 2,4 -2,4 U-turn 0 168 -0,2 -0,2 p-value 0,055* De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets

Results � HIDDEN PET close encounters SCRTL Adj. st. residuals Conflicts No conflicts Conflicts No conflicts Straight 0 35.604 -8,9 8,9 through Left turn 1 8.953 -1,7 1,7 U-turn 21 1.348 25,4 -25,4 p-value <0,001** YCRTL Adj. st. residuals Conflicts No conflicts Conflicts No conflicts Straight 6 63.931 -8,0 8,0 through Left turn 21 15.783 7,1 -7,1 U-turn 2 166 7,9 -7,9 p-value <0,001* De Ceunynck, Daniels, Laureshyn, Hermans, Geussens, Brijs, & Wets