Safety Effectiveness of Pedestrian Crossing Treatments

Transcription

1 Portland State University PDXScholar TREC Friday Seminar Series Transportation Research and Education Center (TREC) Safety Effectiveness of Pedestrian Crossing Treatments Christopher Monsere Portland State University, Let us know how access to this document benefits you. Follow this and additional works at: Part of the Civil Engineering Commons, and the Transportation Engineering Commons Recommended Citation Monsere, Christopher, "Safety Effectiveness of Pedestrian Crossing Treatments" (2017). TREC Friday Seminar Series This Book is brought to you for free and open access. It has been accepted for inclusion in TREC Friday Seminar Series by an authorized administrator of PDXScholar. For more information, please contact

2 Safety Effectiveness of Pedestrian Crossing Enhancements Friday Transportation Seminar October 13, 2017 Chris Monsere, Portland State University 1

3 Total Crash Fatal Crash Background: Oregon Pedestrian Crashes Total Fatal

4 3

5 Research Objectives To estimate the effectiveness of pedestrian crossing enhancements (PCE)s at midblock locations on multimodal safety in Oregon design contexts to derive crash modification factors (CMFs) for Oregon. 4

6 5

7 6

8 7

9 Background Zeeger studied 1,000 marked and 1,000 unmarked crosswalks at unsignalized intersections and mid-block locations in 30 U.S. cities On multilane facilities with ADT higher than 12,000, a marked crosswalk alone without any other enhancements was associated with a statistically significant higher pedestrian crash rate than the pedestrian crash rates of unmarked crosswalks. 8

10 Countermeasure Name High-Visibility School (Yellow) High CMF Value Low CMF Value Highest Star Rating CROSSWALK MARKING Highest Star Rating CMF Value Total Number of CMF Studies Available High-Visibility Crosswalk RAISED MEDIANS Raised Median with Marked Crosswalk Raised Median with Unmarked Crosswalk Raised Pedestrian Crosswalks RAISED PEDESTRIAN CROSSINGS SIGNAL-RELATED TREATMENTS Leading Pedestrian Interval Scramble Phase (Barnes Dance) Pedestrian Hybrid Beacon Vehicle/Pedestrian Crashes, 2 Angle, Head on, Left Turn, Rear End, Rear to Rear, Right Turn, Sideswipe Crashes, 3 Rear-End Crashes, 4 Head-on Crashes, 5 All Crashes, 6 Vehicle/Bicycle, Vehicle/Pedestrian Crashes 9

11 New CMFs Safety Estimates CMFs from NCHRP 841 (2017) PHBs = RRFBs = Pedestrian refuge islands = Advanced YIELD or STOP signs and markings =

12 Data Collection Summary Collect detailed data on 3 types of crossings (n=191) High-visibility Flashing Amber RRFB Note: Crossing only included if installation date could be determined. 11

13 Crossings Mapped (n=191) 12

14 Before and After

15 Crossing Type By Install Year 14

16 15

17 Exposure Motor vehicle AADT per year (factored based on nearby count stations) Pedestrian No systematic counts Explored pedestrian estimation models Explored land-use characterization Neighborhood Concept (C-F) Walk Score Indicator data Presence and distance to bus stop, major shopping center, school, hospital, signal 16

18 Summary of Supplemental Data (1) Data Element Number of Obs. 17 Mean Standar d Deviatio n Max Number of Lanes Posted Speed (mph) Number of Bike Lanes Number of Sidewalks Distance to Bus Stop Shelter (ft) ,697 0 Distance to School (ft) 142 1, , ,913 0 Distance to Signal (N,W) (ft) 119 1, , , Distance to Signal (S,E) (ft) 105 1, , , Number of Light Poles Number of Curb Ramps Number of Curb Extension Number of Ped Advance Sign Assemblies Number of School Min

19 Summary of Supplemental Data (2) Stop Here for Ped Sign Stop on Red Sign Yield Pavement Marking Raised Pedestrian Pedestrian Refuge Raised Median Two Way Left Turn Lane Overhead Signs School Signs Pedestrian Signs School Major Shopping Center Bus Stop Shelter Bus Stop at Crossing One-Way Yes No 0% 20% 40% 60% 80% 100% 18

20 Pedestrian Activity Level Very low Low Medium Low Medium Medium - High High Isolated/Rural Presence of any one - bus stop, school, shopping center or hospital within ¼ mile (1320 ft) Presence of any two - bus stop, school, shopping center or hospital within ¼ mile Presence of any three bus stop, school, shopping center or hospital within ¼ mile Presence of all four bus stop, school, shopping center, and 19 hospital within ¼ mile Major Urban Center/Urban near Major City Presence of any one - bus stop, school, shopping center or hospital within ¼ mile Presence of any two - bus stop, school, shopping center or hospital within ¼ mile Presence of any three bus stop, school, shopping center or hospital within ¼ mile Presence of all four bus stop, school, shopping center or hospital within ¼ mile

21 Merging and linking crash data Buffer 500 ft. initially, then reduced to 300 ft. for first screen of crashes. 20

22 Distance from Crosswalk (ft) Ped crash distance from crosswalk Crash ID 21

23 Distance from Crosswalk (ft) Rear-end distance from crosswalk ft either direction final filter Crash ID 22

24 Analysis Summary Pedestrian Crash Data By Crash Data Element By Selected Supplemental Variables Safety Effectiveness Simple Before/After Comparison Group Cross-Section Empirical Bayes Rear-End Rear-End Crash Data By Crash Data Element By Selected Supplemental Variables Safety Effectiveness Simple Before/After Comparison Group Cross-Section Empirical Bayes 23

25 Pedestrian Crashes (%) Pedestrian Crash Distribution, By Severity 100 Before After FATAL INJURY A INJURY B INJURY C Crash Severity 24

26 Rear-End Crashes (%) Rear-End Crash Distribution, By Severity Before After FATAL INJURY A INJURY B INJURY C PDO Crash Severity 25

27 Observation-Years By Crossing Type Number of Lanes FLASH HI VIS RRFB Total

28 Pedestrian Crash Data: By Number of Lanes Number of Lanes Flashing Amber Hi-Vis RRFB Total Total

29 Risk Ratio = Percent of Crashes / Percent of Observation-Years By Number of Lanes By Posted Speed Flash 1.49 Flash 2.53 HI-VIS 2.60 HI-VIS 2.34 RRFB 1.32 RRFB

30 Safety Evaluation Methods Simple Before- After Comparis on Group Crosssectional Analysis Empirical Bayes Analysis Pedestria n X X no no Rear-End X X X X 29

31 Simple Before After (Ped, by PCE) Parameter RRFB ( ) RRFB ( ) FLASH ( ) HI-VIS ( ) HI-VIS ( ) Number of crosswalks Crashes in the after period (λ) Crashes in the before period w/o treatment (π) Estimated change in total number of crashes (δ) CMF=Index of effectiveness (θ) Standard deviation (θ) CMF (+/- 1 std. dev) CMF (95% C.I.) 0.42 to to to to to to to to to to 2.39

32 Cross-Sectional Model Number of pedestrian crashes = log AADT Variable Coefficient Log(AADT) 1.82** Constant Model Parameter Estimates Observations 124 Log likelihood Akaike Inf. Criteria Presence of treatment was not significant in estimating pedestrian crashes 31

33 Simple Before After (Rear, by PCE) Parameter RRFB FLASH HI-VIS Number of crosswalks Crashes in the after period (λ) Crashes in the before period without treatment (π) Estimated change in total number of crashes (δ) CMF=Index of effectiveness (θ) Standard deviation (θ) CMF (+/- 1 std. dev) 1.12 to to to 2.37 CMF (95% C.I.) 0.94 to to to

34 Cross Section Analysis (Rear-End) Number of rear end crashes = log AADT BusStop Treatment Variable Coefficient Log (AADT) 1.14*** Presence of Bus Stops 0.66** Treatment (RRFB or FLASH) 0.56*** Constant *** Model Parameter Estimates Observations 124 Log likelihood Akaike Inf. Criteria CMF estimated at

35 SPF for Rear-End Crashes, EB Analysis Variable Coefficie Standard Error nt Ln(AADT) 1.706*** Constant *** Observations 85 Log likelihood Akaike Inf Criteria theta Parameter Estimate N expected, T, B N expected, T, A Var (N expected, T, A) 2.49 CMF 0.93 Variance 0.05 SE (CMF) % CI % CI

36 Summary: CMFs for RRFB Pedestrian Parameter Simple Before-After Comparison Group Crosssectional Analysis Empirical Bayes Analysis CMF Standard Error Rear-end Parameter Simple Before-After Comparison Group Crosssectional Analysis Empirical Bayes Analysis CMF Standard Error

37 Challenges Challenges Not able to estimate SPF for pedestrian crashes No reliable way to estimate pedestrian activity Small number of crashes Short after duration of RRFB installs No consistent logging of installation dates and minor modifications Next Steps Looking at options for estimating ped volumes Details of median installations with RRFBs 36

38 Acknowledgements Research funded by Oregon DOT, SPR 778 See full report on ODOT website Miguel Figliozzi, co- PI Sirisha Kothuri Ali Razmpa Dan Hazel ODOT TAC Members 37

39 Countermeasure Name and Description Install enhanced RRFB pedestrian crossing at mid-block crossing location. Crash Type Pedestrian Rear-end Crash Severity All (KABCO) Time of Day All hours Crash Modification Factor Measures of Precision for the CMF (standard error/deviation) Prior Conditions Previously unmarked or at a location with prior high-visibility markings. The data set pooled these locations in the estimation of CMFs. Roadway Class Principal arterial, minor arterial, major collector, minor collector Road Division Type Undivided State Oregon Area Type Rural; Urban; Suburban Number of Through Lanes Two to five lanes (includes TWLTL) Speed Limit 20 mph to 45 mph Traffic Volume Range Average = 13,000 Traffic Control No control Intersection Type Roadway to pedestrian crossing (i.e., mid-block crossing). Years of Data Nine Four Type of Methodology Simple Before-After EB Before-After Sites for inclusion in the study were identified from a list of enhanced crossing locations from state Site Selection Criteria and local inventories. Sites were excluded primarily due to undetermined installation date of treatment. Sample Size Used (Crashes) 26 before, eight after 18 before, 26 after Sample Size Used (Sites) Biases Documentation Sites likely selected for pedestrian crash experience. Regression-to-the-mean bias present and not accounted for in simple beforeafter analysis. Changes in pedestrian volume also not accounted for in method. Sites not likely selected based on rear-end crash history. EB analysis approach includes adjustment for traffic volumes. Changes in pedestrian volume also not accounted for in method. 38