safety management software for state and local highway agencies: –improves identification and...
TRANSCRIPT
• Safety management software for state and local highway agencies:– Improves identification and programming of site-
specific highway safety improvements– Incorporates state-of-the-art safety management
approaches with computerized analytical tools
http://www.safetyanalyst.org/
Who is DevelopingSafetyAnalyst?
• Federal Highway Administration
• Technical Working Group
•19 participating (pooled-fund) States
•1 local highway agency
•2 MPOs
Participating States
SafetyAnalyst Modules
Module 1 – Network Screening
Module 2 – Diagnosis and Countermeasure Selection
Module 3 – Economic Appraisal and Priority Ranking
Module 4 – Countermeasure Evaluation
Status of Development
• Module 1 - Network Screening– Beta testing of interim version underway
• Module 2 - Diagnosis and Countermeasure Selection– Interim version available for testing Dec 2006
• Module 3 - Economic Appraisal and Priority Ranking– Interim version available for testing Dec 2006
• Module 4 – Countermeasure Evaluation– Beta testing of interim version underway
Advantages of SafetyAnalyst over Existing Techniques
• Integrates/automates all parts of safety management process
• Applies state-of-the-art analytical procedures
• Strong cost-effectiveness component
…Enables engineers to make more informed decisions more efficiently
Module 1:Network Screening
Module 1 - Objectives
• Screen entire roadway network, or portion of network, and to identify sites with potential for safety improvement
• Rank sites with potential for safety improvement
• Select sites for further investigation within Module 2 - Diagnosis and Countermeasure Selection
Types of Network Screening
• Basic network screening– With Peak Searching on roadway segments– With Sliding Window on roadway segments
• High proportion of specific accident type• Sudden increase in mean accident frequency• Steady increase in mean accident frequency• Corridors with promise
Basic Network Screening• Uses Empirical Bayes methodology
– Combine observed and predicted accidents• Expected accident values expressed as:
– Expected accident frequency– Excess accident frequency
• Two screening approaches for roadway segments– Peak searching– Sliding window
Basic Network Screening:(with Peak Searching on Roadway Segments)
• For roadway segments, individual sites are divided into windows of size 0.1 mi– Accident frequencies are calculated for each window within
a site• Windows are flagged when:
– Expected value greater than user-specified limit– Expected value is statistically reliable
• If no windows are flagged, incrementally increase window size by 0.1 mi and test again
• More than one window pertaining to a site can be flagged
• Rank order site based upon expected or excess accident frequencies
0.0 mi 0.1 mi 0.2 mi 0.3 mi 0.4 mi 0.5 mi 0.6 mi 0.67 mi
Win # 1
Win # 2
Win # 3
Win # 4
Win # 5
Win # 7
0.03 mi
0.07 mi
Roadway Segment
Win # 6
Note:Window length = 0.1 miExpected accidents = (acc/mi)Limiting Value = 5 acc/miCVLimit = 0.5
Peak Searching Concepts
0.0 mi 0.1 mi 0.2 mi 0.3 mi 0.4 mi 0.5 mi 0.6 mi 0.67 mi
Win # 1
Win # 2
Win # 3
Win # 4
Win # 5
0.03 mi
Roadway Segment
Win # 6
0.07 mi
Note:Window length = 0.2 miExpected accidents = (acc/mi)Limiting Value = 5 acc/miCVLimit = 0.5
Peak Searching Concepts
0.0 mi 0.1 mi 0.2 mi 0.3 mi 0.4 mi 0.5 mi 0.6 mi 0.67 mi
Win # 1
Win # 2
Win # 3
Win # 4
Win # 5
Roadway Segment
Note:Window length = 0.3 miExpected accidents = (acc/mi)Limiting Value = 5 acc/miCVLimit = 0.5
Peak Searching Concepts
Peak Searching Concepts
0.0 mi 0.1 mi 0.2 mi 0.3 mi 0.4 mi 0.5 mi 0.6 mi 0.67 mi
Win # 1
Win # 2
Win # 3
Win # 4
Roadway Segment
Note:Window length = 0.4 miExpected accidents = (acc/mi)Limiting Value = 5 acc/miCVLimit = 0.5
Peak Searching Concepts
0.0 mi 0.1 mi 0.2 mi 0.3 mi 0.4 mi 0.5 mi 0.6 mi 0.67 mi
Win # 1
Win # 2
Win # 3
Roadway Segment
Note:Window length = 0.5 miExpected accidents = (acc/mi)Limiting Value = 5 acc/miCVLimit = 0.5
Peak Searching Concepts
0.0 mi 0.1 mi 0.2 mi 0.3 mi 0.4 mi 0.5 mi 0.6 mi 0.67 mi
Win # 1
Win # 2
Roadway Segment
Note:Window length = 0.6 miExpected accidents = (acc/mi)Limiting Value = 5 acc/miCVLimit = 0.5
Peak Searching Concepts
0.0 mi 0.1 mi 0.2 mi 0.3 mi 0.4 mi 0.5 mi 0.6 mi 0.67 mi
Win # 1
Roadway Segment
Note:Window length = Segment lengthExpected accidents = (acc/mi)Limiting Value = 5 acc/miCVLimit = 0.5
Site No. 1
MP 1.0 MP 2.6
First Sliding WindowW = 0.3 mi
1.1 mi 1.2 mi 1.3 mi 1.4 mi 1.5 mi
Sliding Window Concepts
Site No. 1
MP 1.0 MP 2.6
First Sliding WindowW = 0.3 mi
1.1 mi 1.2 mi 1.3 mi 1.4 mi 1.5 mi
Second Sliding WindowW = 0.3 mi
Sliding window is moved incrementallyby 0.1 mi along the roadway segment.
Sliding Window Concepts
Site No. 23 Site No. 24
MP 35.4 MP 36.2 MP 36.7
X = 3.1
Note:Window length = 0.3 miIncrement length = 0.1 miExpected accidents = (acc/mi)Limiting Value = 5 acc/mi
X = 4.3
X = 5.6
X = 7.8
X = 9.8
X = 7.5
X = 6.0
X = 4.5
X = 3.5
X = 3.0
X = 4.0
X = 5.1
X = 5.5
Site No. 23 Site No. 24
MP 35.4 MP 36.2 MP 36.7
Note:Window length = 0.3 miIncrement length = 0.1 miExpected accidents = (acc/mi)Limiting Value = 5 acc/mi
X = 5.6
X = 7.8
X = 9.8
X = 7.5
X = 6.0
X = 5.1
X = 5.5
Site No. 23 Site No. 24
MP 35.4 MP 36.2 MP 36.7
Note:Window length = 0.3 miIncrement length = 0.1 miExpected accidents = (acc/mi)Limiting Value = 5 acc/mi
X = 5.6
X = 7.8
X = 9.8
X = 7.5
X = 6.0
X = 5.1
X = 5.5
High Proportions of Specific Accident Type
• Objective:– Identify sites having higher than expected
proportions of specific target accidents– Rank sites based on difference observed proportion
and expected proportion of target accident
• Methodology– Calculate observed proportion (TOT only)– Calculate the probability that observed proportion is
greater than limiting proportion (i.e., avg for site & accident type)
– Site flagged when probability is greater than some user-specified significance level
• Roadway segments– Similar to sliding window approach– Longer windows are needed to reduce
variance (e.g., 1.0 mi)– More than one window pertaining to a site
can be flagged• Site ranked based upon maximum difference
between observed proportion and expected proportion
High Proportions of Specific Accident Type (cont.)
Sudden Increase in Mean Accident Frequency
• Screening for safety deterioration– Calculate differences in mean yearly
accident frequencies– For the time period with the largest
difference:• If the percentage increase is greater than a
user-specified limiting value• Then perform test of significance
• Based on observed accidents• Based on total accidents• Flagged sites are not rank ordered
Steady Increase in Mean Accident Frequency
• Screening for safety deterioration– Fit regression model to data of accident
frequency versus year• If value of slope is greater than a user-
specified limiting slope• Then perform test of significance
• Based on observed accidents
• Based on total accidents
• Flagged sites are not rank ordered
Screening for Corridors with Promise
• Analysis of extended corridors (e.g., 10 mi or more)– Roadway segments, intersections, and ramps
grouped together
• Rank order corridors based upon:– Accidents/mi/yr– Accidents/million veh-mi/yr
• Based on observed accidents
Demonstration of Module 1: Network Screening
SafetyAnalyst Modules
Module 1 – Network Screening
Module 2 – Diagnosis and Countermeasure Selection
Module 3 – Economic Appraisal and Priority Ranking
Module 4 – Countermeasure Evaluation
Module 2 – Diagnosis and Countermeasure Selection
• Display collision diagram (links to third-party software)
• Identify collision patterns
• Conduct diagnostic investigations
• Suggest countermeasures that address identified collision patterns
• Select appropriate countermeasures
Module 3 – Economic Appraisal and Priority Ranking
• Perform economic analysis of alternative countermeasures for a specific site
• Perform economic analysis of improvements across selected sites
• Select mix of sites and countermeasures to get maximum benefits within a specified budget
• Develop priority ranking of alternative improvements
Module 4:Countermeasure Evaluation Tool
Objective
• Determine safety effectiveness (percent reduction in crashes) for specific implemented countermeasures
• Conduct before-after evaluation of crash frequencies using the Empirical Bayes (EB) approach
• Conduct before-after evaluation of shifts in crash severity or crash type proportion
Why the Evaluation Tool?
• The goal of SafetyAnalyst is to help highway agencies determine how funds can be spent in the most cost-effective manner to improve safety.
• The results of Module 4 can be used to update the accident modification factors (AMFs) that are used within Module 3 for economic appraisal and priority ranking of countermeasures to be implemented at sites.
When to Use the Evaluation Tool
• A countermeasure has been implemented at a number of sites
• The agency wants to assess how effectively the countermeasure performed:– Did it improve the safety performance at a
site?– Did it reduce a specific target accident type?
What Information Is Needed
• Locations of improved sites• Countermeasure(s) to evaluate for each site
– Year of implementation– Countermeasure name(s)
• Site characteristics• ADTs (before and after improvement)• Yearly accident counts (before and after)• Safety Performance Functions (SPFs)
Types of Analyses Conducted
• The safety effectiveness of countermeasures is quantified through the use of before-after statistical evaluations.
• Two types of before-after evaluations can be conducted:– Percent change in accident frequencies, due to the
implemented countermeasure, is evaluated by an Empirical Bayes (EB) technique.
– Shift in proportion of specific collision types is evaluated using the Wilcoxon signed rank test
Demonstration of Module 4: Countermeasure Evaluation Tool