Evaluation of Alternative Methods for Identifying High Collision Concentration Locations Raghavan Srinivasan 1 Craig Lyon 2 Bhagwant Persaud 2 Carol Martell.

Slides:



Advertisements
Similar presentations
Complete Street Analysis of a Road Diet Orange Grove Boulevard Pasadena, CA Aaron Elias Engineering Associate Kittelson & Associates Bill Cisco Senior.
Advertisements

Interactive Highway Safety Design Model (IHSDM)
FUTURE CMF RESEARCH AND CHALLENGES Traffic Records Forum October 27, 2014 Daniel Carter, UNC HSRC.
HSM Implementation Tools Safety Analyst
Spring Before-After Studies Recap: we need to define the notation that will be used for performing the two tasks at hand. Let: be the expected number.
SafetyAnalyst Michael S. Griffith FHWA July 2003.
Analysis and Multi-Level Modeling of Truck Freight Demand Huili Wang, Kitae Jang, Ching-Yao Chan California PATH, University of California at Berkeley.
Session #4, Forrest Council, Slides. The Potential Impacts of a Towaway Reporting Threshold on Driver/User and Roadway Safety Programs Forrest M. Council.
Protection Values for VOR-Defined ATS Routes
Spring  Crash modification factors (CMFs) are becoming increasing popular: ◦ Simple multiplication factor ◦ Used for estimating safety improvement.
Roadway Safety For Local Agencies Doug Bish Traffic Services Engineer Oregon Department of Transportation.
Investigation of Varied Time Intervals in Crash Hotspot Identification Authors: Wen Cheng, Ph.D., P.E., Fernando Gonzalez, EIT, & Xudong Jia; California.
Enhanced Safety Prediction Methodology and Analysis Tool for Freeways and Interchanges James A. Bonneson August 2012 NCHRP Project
Case Study 4 New York State Alternate Route 7. Key Issues to Explore: Capacity of the mainline sections of NYS-7 Adequacy of the weaving sections Performance.
Spring  Types of studies ◦ Naïve before-after studies ◦ Before-after studies with control group ◦ Empirical Bayes approach (control group) ◦ Full.
Spring INTRODUCTION There exists a lot of methods used for identifying high risk locations or sites that experience more crashes than one would.
Incorporating Safety into the Highway Design Process.
Incorporating Temporal Effect into Crash Safety Performance Functions Wen Cheng, Ph.D., P.E., PTOE Civil Engineering Department Cal Poly Pomona.
Session 10 Training Opportunities Brief Overview of Related Courses in USA / Canada Geni Bahar, P.E. NAVIGATS Inc.
Statistics for Managers Using Microsoft Excel, 4e © 2004 Prentice-Hall, Inc. Chap 12-1 Chapter 12 Simple Linear Regression Statistics for Managers Using.
Mapping Cyclist Activity and Injury Risk in a Network Combining Smartphone GPS Data and Bicycle Counts PhD Candidate: Jillian Strauss Supervisors: Luis.
Center for Risk Management of Engineering Systems University of Virginia, Charlottesville 26 Schedule.
University of Maryland Department of Civil & Environmental Engineering By G.L. Chang, M.L. Franz, Y. Liu, Y. Lu & R. Tao BACKGROUND SYSTEM DESIGN DATA.
Role of SPFs in the Interactive Highway Safety Design Model (IHSDM) Mike Dimaiuta LENDIS Corporation.
The Empirical Bayes Method for Safety Estimation Doug Harwood MRIGlobal Kansas City, MO.
Network Screening 1 Module 3 Safety Analysis in a Data-limited, Local Agency Environment: July 22, Boise, Idaho.
1 Validation and Implication of Segmentation on Empirical Bayes for Highway Safety Studies Reginald R. Souleyrette, Robert P. Haas and T. H. Maze Iowa.
2-1 LOW COST SAFETY IMPROVEMENTS The Tools – Identification of High Crash Locations – Session #2.
1 CEE 763 Fall 2011 Topic 1 – Fundamentals CEE 763.
Safety management software for state and local highway agencies: –Improves identification and programming of site- specific highway safety improvements.
Jason J. Siwula, PE – Safety Engineer DOES 24+0=22+2? AN INTRO TO HSM METHODS.
Role of SPFs in SafetyAnalyst Ray Krammes Federal Highway Administration.
Chapter 4 Linear Regression 1. Introduction Managerial decisions are often based on the relationship between two or more variables. For example, after.
HSM: Another Tool for Safety Management in Wyoming 1 Excellence in Transportation.
9-1 Using SafetyAnalyst Module 4 Countermeasure Evaluation.
1 Element 1: The Systemic Safety Project Selection Process Element 1: 4-Step Project Selection Process.
The Highway Safety Manual: A New Tool for Safety Analysis John Zegeer, PE Kittelson & Associates, Inc. HSM Production Team Kittelson & Associates, Inc.
An Examination of “Fault,” “Unsafe Driving Acts” and “Total Harm” in Car-Truck Collisions Forrest Council (HSRC) David Harkey (HSRC) Daniel Nabors (BMI)
University of Minnesota Intersection Decision Support Research - Results of Crash Analysis University of Minnesota Intersection Decision Support Research.
1 Tobit Analysis of Vehicle Accident Rates on Interstate Highways Panagiotis Ch. Anastasopoulos, Andrew Tarko, and Fred Mannering.
Putting Together a Safety Program Kevin J. Haas, P.E.—Traffic Investigations Engineer Oregon Department of Transportation Traffic—Roadway Section (Salem,
CE 552 Week 9 Crash statistical approaches Identification of problem areas - High crash locations.
July 29 and 30, 2009 SPF Development in Illinois Yanfeng Ouyang Department of Civil & Environmental Engineering University of Illinois at Urbana-Champaign.
Calibrating Highway Safety Manual Equations for Application in Florida Dr. Siva Srinivasan, Phillip Haas, Nagendra Dhakar, and Ryan Hormel (UF) Doug Harwood.
Calibration of SPFs in the HSM, IHSDM, and SafetyAnalyst Doug Harwood Midwest Research Institute.
NCHRP Crash Reduction Factors for Traffic Engineering and ITS Improvements UNC HSRC VHB Ryerson University (Bhagwant and Craig)
Continuous Risk Profile: A Simple Method for Identifying Sites for Safety Investigation. Koohong Chung, Ph.D. California Department of Transportation Highway.
Fall  Crashes are “independent” and “random” events (probabilistic events)  Estimate a relationship between crashes and covariates (or explanatory.
HSIS Annual Meeting, 10/2006 NCHRP 17-30: Traffic Safety Evaluation of Nighttime and Daytime Work Zones Raghavan Srinivasan Forrest Council.
Impact of Intersection Angle on Safety HSIS Annual Liaison Meeting David Harkey, Bo Lan, Daniel Carter, Raghavan Srinivasan, Anusha Patel Nujjetty May.
Cable Median Barrier with Inside Shoulder Rumble Strips on Divided Roads Raghavan Srinivasan, Bo Lan, & Daniel Carter, UNC Highway Safety Research Center.
1 THE HIGHWAY SAFETY MANUAL Michael S. Griffith Federal Highway Administration July 26 th, 2004.
Session 2 History How did SPF come into being and why is it here to stay? Geni Bahar, P.E. NAVIGATS Inc.
Role of Safety Performance Functions in the Highway Safety Manual July 29, 2009.
Evaluating the performance of three different network screening methods for detecting high collision concentration locations using empirical data Prepared.
Rural Intersection Decision Support - Crash Analysis Rural Intersection Decision Support - Crash Analysis Presented at Pooled Fund Meeting April 19, 2004.
FHWA: Revision of Thirteen Controlling Criteria for Design; Notice for Request and Comment. Comments Due: December 7, 2015 Jeremy Fletcher, P.E., P.S.M.
HSM Applications to Suburban/Urban Multilane Intersections Prediction of Crash Frequency for Suburban/Urban Multilane Intersections - Session #9.
LOW COST SAFETY IMPROVEMENTS Practitioner Workshop The Tools – Identification of High Crash Locations – Session #2.
1 The Highway Safety Manual Predictive Methods. 2 New Highway Safety Manual of 2010 ►Methodology is like that for assessing and assuring the adequacy.
Low Cost Safety Improvements Pooled Fund Study Analytical Basics Dr. Bhagwant Persaud.
Segment Crash Analysis
Artificial Realistic Data (ARD)
Applied Technology and Traffic Analysis Program(ATTAP) MIDCAP & MUID
Before-After Studies Part I
26th CARSP Conference, Halifax, June 5-8, 2016
Network Screening & Diagnosis
Kittelson & Associates, Inc.
Clark County, WA Safety Management Program
Contributing Factors for Focus Crash Types and Facility Types Raghavan Srinivasan University of North Carolina Highway Safety Research Center (UNC HSRC)
Presentation transcript:

Evaluation of Alternative Methods for Identifying High Collision Concentration Locations Raghavan Srinivasan 1 Craig Lyon 2 Bhagwant Persaud 2 Carol Martell 1 Jongdae Baek 1 Meeting with CALTRANS, December 7 and 8, UNC Highway Safety Research Center 2 Persaud and Lyon, Inc. 1

Background States and local agencies have limited budgets Need an efficient process to identify sites with potential for safety improvement Conventional methods have problems: –Potential bias due to regression to the mean Sites with a randomly high count may be incorrectly identified –Use of crash rates implicitly assume crash frequency and traffic volume are linearly related (shown to be incorrect quite often) Meeting with CALTRANS, December 7 and 8,

Objectives Evaluate various methods using data from California Identify methods that are optimal for identifying locations for further review Assess the recently released software product (SafetyAnalyst) for network screening Meeting with CALTRANS, December 7 and 8,

Overview of Methods Table C method Level of Service (LOSS) method Methods based on the empirical Bayes (EB) procedure Continuous Risk Profile (CRP) method Screening based on high proportions of specific crash types Meeting with CALTRANS, December 7 and 8,

Table C Method Compare observed accident frequency with critical crash frequency –Average accident rate (base rate + AADT factor) –Computed for different rate groups (i.e., type of facilities) –For roadway segments, use sliding window of 0.2 miles with 0.02 mile increments –Intersections (250 feet radius) –Time periods: 3, 6, and 12 months –Select windows where observed frequency is greater than critical crash frequency Meeting with CALTRANS, December 7 and 8,

Table C Method, contd. Does not account for potential bias due to Regression to the Mean (RTM) –Can be a major issue since a maximum of 1 year of data is used Uses crash rates –Incorrectly assumes that crash frequency is proportional to traffic volume No ranking of sites Meeting with CALTRANS, December 7 and 8,

Level of Service (LOSS) Method Develop Safety Performance Functions (SPFs) –Equations that relate expected crash frequency to site characteristics –Predict average accident frequencies Compare crash frequency with average frequencies from SPFs Accounts for non-linear relationship between crash frequency and volume Does not account for RTM No automatic ranking of sites Meeting with CALTRANS, December 7 and 8,

Methods based on the EB procedure Expected crashes for a site (E) = w*(SPF prediction) +(1-w)*(Observed frequency) Expected Excess for a site = E – SPF prediction Implementation –Sliding Window –Peak Search Meeting with CALTRANS, December 7 and 8,

Sliding Window Fixed window that moves in defined increments A window can include portions from multiple segments Each segment is characterized by the maximum value calculated at any window position in or overlapping the segment Meeting with CALTRANS, December 7 and 8,

Peak Search Incrementally growing window lengths are selected No windows span multiple segments Window starts at left boundary of a segment and moves incrementally until it reaches the end Test of significance based on the coefficient of variation Meeting with CALTRANS, December 7 and 8,

Methods based on the EB procedure, contd. Regression to the Mean (RTM) is accounted for Accounts for non-linear relationship between traffic volume and crash counts Windows of varying length in the peak search method allow a more detailed assessment of crash risk Sites are ranked based on EB Expected or EB Expected Excess Meeting with CALTRANS, December 7 and 8,

Continuous Risk Profile (CRP) Compute cumulative count of accidents Subtract cumulative expected number of accidents from cumulative count leading to a cumulative excess Compute moving average of cumulative excess using a sliding window If moving average is negative, a value of zero is assigned; positive values remain the same Plot revised moving average with distance Meeting with CALTRANS, December 7 and 8,

CRP, contd. Does not account for RTM because only observed crashes are used Does not account for traffic volume Uses constant window size – same limitation as other sliding window approaches No procedure for ranking of sites Meeting with CALTRANS, December 7 and 8,

Screening Based on High Proportions Ranks locations that have a proportion of a specific accident type higher than a threshold value Could reveal existence of accident patterns susceptible to correction Uses only accident data RTM is not accounted for Does not account for traffic volume Meeting with CALTRANS, December 7 and 8,

Methods Selected for Evaluation Expected Crashes based on EB procedure Expected Excess Crashes based on EB procedure LOSS method Table C method High proportion method (as an illustration for intersections) Meeting with CALTRANS, December 7 and 8,

Evaluation Approach How does each method rank high those locations that are more likely to have large number of crashes in the future? How does each method rank high the locations that are identified as needing improvement (among the ones that were selected for review by Table C)? Compare the characteristics of top ranked locations in each method Meeting with CALTRANS, December 7 and 8,

Data California roadway, intersection, and crash files obtained from the Highway Safety Information System (HSIS) for HSIS staff linked the location of ramps and intersections to roadway segment file –Determine roadway segments within intersections Crashes within 250 feet of the center of intersections were assigned to intersections –Interchange location is not available: 0.3 miles on either side of a ramp was used as a proxy for interchange influence area Meeting with CALTRANS, December 7 and 8,

Data Preparation, contd. Caltrans provided excel file called ‘TableCInv6yr’ –Sites selected from Table C for investigation –Sites identified as ‘improvement recommended’ Roadway segments –Rural two lane roads –Urban freeways Intersections –Rural 4 leg minor stop controlled intersections –Rural 4 leg signalized Meeting with CALTRANS, December 7 and 8,

Safety Performance Functions Two types of SPFs developed with HSIS data –Type 1 SPF (similar to default SPFs in SafetyAnalyst) AADT is the only term for segments Major and Minor road AADT are the only terms for intersections –Type 2 SPF Other site characteristics in addition to AADT –Roadway segments (shoulder width, median width, etc.) –Intersections (presence of right turn or left turn lanes, etc.) For freeways, separate SPFs developed for within and outside interchange influence area SPFs developed for many crash types Meeting with CALTRANS, December 7 and 8,

Conclusions and Recommendations Methods based on EB procedure generally identify sites with higher AADTs and higher expected crashes. Top ranked sites based on the EB methods have more crashes in the future compared to the Table C method. When data on investigated sites was used, Table C did quite well (as expected). However, other methods did well in many cases. Future research could consider using cost effectiveness data. Meeting with CALTRANS, December 7 and 8,

Conclusions and Recommendations, contd If sufficient samples are available, directly estimating SPFs from CA data are better. –This study estimated SPFs for two intersection types and two roadway types –Future research could develop SPFs for other facility types Methods from EB procedure work better with longer segments. With expanded lengths, an entire segment could be flagged instead of just the window. Meeting with CALTRANS, December 7 and 8,

Conclusions and Recommendations, contd Future research could examine the possibility of coming up with weights that reflect potential cost benefit values. The ‘proportions’ method could be used in combination with EB methods as a diagnostic tool. Future research could investigate if combination of methods would provide better results. Meeting with CALTRANS, December 7 and 8,

Conclusions and Recommendations, contd Develop a database of interchange locations along with a link to the ramps from each interchange. Compile data on lengths of individual ramps. Obtain more recent and accurate minor road AADT data. Develop a database with detailed information about treatments implemented. Meeting with CALTRANS, December 7 and 8,