Session 3 Highway Safety Manual General Overview Joe Santos, PE, FDOT, State Safety Office November 6, 2013
Workshop Series Wed. Oct. 30 Wed. Nov. 6 Wed. Nov. 13 Wed. Nov. 20 Wed. Dec 4 Wed. Dec. 11 Wed. Dec. 18 Wed. Jan. 8 Wed. Jan. 15 Wed. Jan. 22 Wed. Jan. 29 Highway Safety Evaluation Highway Safety Manual Application and Science of Crash Reduction Factors Requirements for HSIP Applications Safety Funding Categories/Requirements/Conditions Is Your Project Feasible? What s Next and How Do We Move Forward? B/C Calculations plus NPV Calculations New WP Guidelines 2014 Safety Projects & The Local Agency Program (LAP) Development of the Safety/LAP Project Schedule for Funding Purposes Safety/LAP Project Development Key to Successful Safety Programs Today s Presentation Highway Safety Manual
Road safety management is in transition. The transition is from action based on experience, intuition, judgment, and tradition, to action based on empirical evidence, science, and technology Ezra Hauer (2005) 1 3
Learning Outcomes HSM: Purpose Audience Structure Advantages Resources 1 4
How Important is Safety? Programming and prioritization System planning Program administration Policy development Project development Operations and maintenance Public affairs Interagency coordination 1 5
Safety Trade Offs? Mobility Connectivity Safety Costs Environment Right of Way Balance the array of issues 1 6
What is Safety? The HSM uses crashes as a measure of safety 1 7
What does the HSM do? Quantify Safety Measure Performance Informed Decisions 1 8
HSM Vision Akin to HCM 1 Definitive; represents quantitative state of theart information 2 Widely accepted within professional practice of transportation engineering 3 Science based; updated regularly to reflect research 1 9
Why do we need it? Limited Resources, Science, and Technology Legislatively Mandated Priorities 1 10
Where can the HSM be used? JURISDICTION State County Region Local Facility Type Two Lane Rural Highways Rural Multi Lane Highways Urban Suburban Arterials Freeways (new!)
Who is the target user? Transportation Planners Traffic Engineers Safety Engineers Designers Capital Programmers 1 12
When can the HSM be used? 1. System Management 2. Project Development 1 13
How can the HSM be used? Identifying sites with potential for crash reduction Identifying crash patterns and treatments Conducting economic appraisals Evaluating the crash reduction benefits of implemented treatments Estimating the crash reduction effects of design decisions 1 14
Highway Safety has Two Dimensions Nominal Safety Substantive Safety Standards Compliance Expected or Actual Crash Frequency and Severity
Nominal Safety vs. Substantive Safety Nominally Safe Streets and Highways + Application of Highway Safety Research and Results + Performance Monitoring Substantively Safe Streets and Highways 1 25
Crash vs Prediction and Reduction 17 1 17
The HSM and Other Documents and 1 18
Advance Safety Knowledge Descriptive Predictive 1 19
HSM Content and Structure 1 20
Vol. 1 - Part A Introduction Human Factors Fundamentals Vol. 3 - Part D Crash Modification Factors (CMFs) Vol. 1 - Part B Roadway Safety Management Process Vol. 2 - Part C Predictive Method 1 21
The HSM provides better methods to improve the bottom line Better safety analysis tools to support decision making More safety cost effective investments More lives saved and injuries avoided per dollar invested
Fundamentals KEY TERMS AND DEFINITIONS
What is Safety? The HSM uses crashes as a measure of safety
Crashes
Random Events
Rare Events Relative Proportion of Crash Events Part A: Figure 3 2; Page 3 6
PREDICTING CRASHES
Crash Frequency N x = Number of Crashes Number of Years
Crash Estimation Historical Crash Records vs. Predicted Crash Frequency How did a facility perform? vs. How should a facility perform?
Part C: Predictive Method Estimate expected average crash frequency given a specific geometric design, traffic volume, length, and a given time period
Expected Average Crash Frequency Crash Frequency = Short Term Measure
DATA NEEDS AND CONSTRAINTS
Data Needs 1. Facility data 2. Traffic volume data 3. Crash data
Observed Crash Data Limitations 1. Quality and accuracy 2. Reporting thresholds 3. Severity indeterminacy 4. Jurisdiction differences 5. Regression to the mean
Limitation: Randomness and Change 1. Regression to the mean 2. Natural crash frequency variation 3. Roadway characteristic variations
Regression to the Mean 7 6 5 4 3 2 Roll Value 1 0 Roll Value Cumluative Average 0 20 40 60 80 100 120 140 160 Roll Number
Crashes Regress to the Mean as Well Short Term Average Crash Observed Crash Freq uency Short Term Average Crash Frequency Short Term Average Crash Expected Average Crash Frequency Years Adapted from Part A, Figure3 4, Page 3 11
Roadway and Environment Variation Consider Site Changes Over Time: Land use Traffic volume Weather Traffic control Geometric design
Observed Crash Frequency Site Selected for Treatment due to Short Term Trend Before Average Crashes RTM Reduction Expected Average Crash Frequency (Without Treatment) RTM and RTM BIAS Years AFTER Perceived Effectiveness of Treatment Actual Reduction due to Treatment Adapted from Part A, Figure 3 5, Page 3 12
RTM Bias Regression to the Mean (RTM) If we do not account for RTM, we cannot say the crash difference is due to the treatment.
PREDICTIVE METHOD OVERVIEW
Simplified Predictive Method
Elements of Predictive Method 1. Safety Performance Functions 2. Crash Modification Factors 3. Calibration Factor (if available) 4. Empirical Bayes, (if crash history is known)
Predicted Average Crashes Predicted Crash Frequency N predicted = SPF x (CMF 1 x CMF 2 x.) x C Where: SPF = Safety Performance Function CMF = Crash Modification Factors C = Local Calibration Factor (Eq. 3-3, pg. 3-17)
http://www.dot.state.fl.us/safety/transsafeng/highwa ysafetymanual.htm
Fundamental Element #1 Safety Performance Functions PREDICTIVE METHOD
Safety Performance Function (SPF) Product of Statistical Modeling Process Mathematical expression Used to estimate average crash frequency for base condition
SPFs in the HSM Rural two lane 1 segment SPF 3 intersection SPFs Rural multilane 2 segment SPFs 3 intersection SPFs Urban and suburban arterials 5 segment SPFs 4 intersection SPFs 18 SPFs in First Edition HSM
Rural Multilane Highway Three Leg Stop Controlled Intersections SPF N spf int = exp[ a + b ln(aadt maj ) + c ln(aadt min )] (Eq. 11-11, pg. 11-21) (with Table 11-7)q
SPF Base Condition Each SPF has its own base conditions Facility Type Feature Condition Rural Multi Lane Undivided Segment Lane Width 12 Feet Shoulder Width & Type 6 Feet, Paved Side Slopes 1:7 or Flatter Lighting None Automated Speed Enforcement None What if my project is different?
SPF Base Conditions Rural Two Lane, Two Way Roadway
SPF Base Conditions Rural Two Lane, Two Way Roadway
Fundamental Element #2 Crash Modification Factors PREDICTIVE METHOD
CRASH MODIFICATION FACTOR (CMF) Crash Modification Factors represent the relative change in crash frequency due to a change in one specific condition, when all other conditions and characteristics remain constant
Crash Modification Factors (CMF) CMFs are the ratio of the crash frequency at a site under two different conditions CMF = Therefore Crashes for Condition a = CMF Crashes for Condition b Where Crash Frequency with Site Condition a Crash Frequency with Site Condition b Condition a is a change from base or existing condition Condition b is the base or existing condition (Eq. 3 5, pg. 3 19) CMFs may serve as an estimate of the effect of a particular geometric design or traffic control feature or the effectiveness of a particular treatment or condition.
How are CMFs Developed? 2-lane, 2-way Rural Roadway Segment, 4,000 ADT 11 Lane Width Average Crashes 12 lanes = 200 (base condition) Average Crashes 11 lanes = 210 210 CMF 1r = = 1.05 200 CMF 1r = 1.05
CMF Example Automated Speed Enforcement 2-lane Rural Highway Total Crashes CMF 12r = 0.93 (Volume 2, Page 10 31) Predicted Average Crashes = CMF* (base condition crashes) = 0.93* base condition crash frequency
CMFs Modify the Base Condition N predicted = N SPF x (CMF 1x CMF 2x CMF yx ) C x (Eq. 3 3, pg. 3 17) Where N SPFx CMF yx C x = Base condition crash frequency from SPF for site type x = Crash modification factors = Calibration factor for site type x
Apply CMF if: Differences from base conditions: Geometrics Roadside Conditions Lighting, etc.
Sample CMFs Rural Two Lane, Two Way Roads CMF Description Reference CMF 1r Lane Width Eq. 10 11, Table 10 8 CMF 2r Shoulder Width & Type Eq. 10 12, Table 10 9, 10 10 CMF 3r Horizontal Curves Eq. 10 13 CMF 4r Superelevation Variation Eq. 10 14, 10 15, 10 16 CMF 5r Grade Table 10 11 CMF 6r Driveway Density Eq. 10 17 CMF 7r Centerline Rumble Strips See text CMF 8r Passing Lanes See text CMF 9r Two Way Left Turn Lanes Eq. 10 18, 10 19 CMF 10r Roadside Design Eq. 10 20 CMF 11r Lighting Eq. 10 21, Table 10 12 CMF 12r Automated Speed Enforcement See text
Fundamental Element #3 Calibration Factors PREDICTIVE METHOD
Calibration Adjust predictive models to local conditions Provide method to address local variations Climate Driver populations Animal populations Crash reporting thresholds Crash reporting system procedures
Calibration Factor Observed Crashes all sites C = Predicted Crashes all sites (Eq. A 1, pg. A 7, V.2)
FDOT Segment Calibration Factors Facility Type KABC Severity Rural Two Lane Undivided (R2U) 1.00 Rural Four Lane Divided (R4D) 0.68 Urban Two Lane Undivided Arterial (U2U) 1.02 Urban Two Lane Arterial with Center Two Way Left Turn Lane (U32LT) 1.04 Urban Four lane Undivided (U4U) 0.73 Urban Four Lane Arterial Divided 1.63 Urban Four Lane Arterial with Center Two Way Left Turn Lane (U52LT) 0.70 FDOT Safety Website
FDOT Intersection Calibration Factors Facility Type KABC Severity Rural Two Lane: Three Leg Stop Controlled (R2 3ST) 1.00 Rural Two Lane: Four Leg Stop Controlled (R2 4ST) 1.00 Rural Two Lane: Four Leg Signalized (R2 4SG) 1.00 Rural Multi Lane: Four Leg Signalized (RM 4SG) 1.00 Urban Arterial: Three Leg Signalized (U 3SG) 1.00 Urban Arterial: Four Leg Signalized (U 4SG) 1.00 FDOT Safety Website
Fundamental Element #4 Empirical Bayes Method PREDICTIVE METHOD
Now, let s talk about the EB Method Simplified Predictive Method
EB Method Concept AADT Crash Frequency
EB Method Equations N expected = w N predicted + (1 w) N observed (Eq. 3 9, pg. 3 24) Weighted Adjustment w = 1 1 + k ( N predicted ) all study years (Eq. 3 10, pg. 3 24) Overdispersion Parameter (from SPF)
Elements of EB Method Only when predicted and observed crash frequencies are available Apply weighting factor to adjust for variability of SPF (overdispersion) Produces Expected average crash frequency Past & Future conditions
Further HSM Resources National http://www.highwaysafetymanual.org/pages/default.aspx http://www.highwaysafetymanual.org/pages/training.aspx Florida http://www.dot.state.fl.us/safety/11a SafetyEngineering/TransSafEng/HighwaySafetyManual.shtm 1 72
Learning Outcomes HSM: Purpose Audience Structure Advantages Resources 1 77
Questions? Joe Santos, PE, FDOT, State Safety Office Joseph.Santos@dot.state.fl.us 850.414.4097
COST