Highway Safety Manual at the Project Level - PowerPoint Presentation

Highway Safety Manual at the Project Level Partnering Conference 2016 Eric Green, MSCE, PE Kentucky Transportation Center Mike Vaughn Kentucky Transportation Cabinet

Would you expect these alternatives to perform the same over a 30-yr project life? Should we know the differences in safety performance for these alternatives before investing millions of taxpayer dollars? CAN we know the differences in safety performance for these alternatives before investing millions of taxpayer dollars?

Incorporating Safety at the Project Level “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…”

Approaches for Considering Safety Examined in reference to compliance with standards, warrants, guidelines, and traditional design procedures The expected performance of a roadway in terms of crash frequency and severity Source: AASHTO Source: AASHTO

Nominal vs. Substantive Safety 5 Nominal Safety Substantive Safety

YES!! Should we know the differences in safety performance for these alternatives before investing millions of taxpayer dollars? CAN we know the differences in safety performance for these alternatives before investing millions of taxpayer dollars?

Highway Safety Manual Project Level Eric Green

Decision Making Process Environment Costs Society Safety Capacity Right of Way

Basic Safety Questions? How can I predict safety performance of geometric design features?How can I quantitatively calculate safety effects?How can I justify decisions based on safety implications?

Highway Safety Manual Purpose Resource for professionalsSafety knowledgeDecision making tools

A Safe Roadway? List what are the top three items that make a roadway safe

Safe Roadway?

Safe Roadway?

Crash Prediction Uses Program levelSegment prioritizationProject levelProject needs assessment Communication with the publicDocumentationTort Defense 1- 14

HSM Contents Introduction, Human Factors, Fundamentals Safety Management Process Predictive Methods Crash Modification Factors

Some More Questions What is the number of crashes for a given scenario?2-lane rural road with 5,000 AADT 4-lane rural divided arterial with 20,000 AADT What will be the safety effect if weWiden shoulderAdd a median Add a left-turn lane

How HSM Works Safety Performance Function(SPF): regression equationCrash Modification Factor (CMF): adjustment factorsCalibration Factor (C): local adjustment SPF CM F C Predicted Crashes

Part C: Predictive Methods (1/3) Safety Performance Functions (SPFs) Regression models Predict expected crash frequency Developed from data for a number of similar sites Developed for specific site types and “base conditions” SPF

SPF Development Average Daily Traffic (ADT) Crashes/Year SPF

Part C: Predictive Methods (2/3) Crash Modification Factors (CMFs) Adjusts SPF predicted value Accounts for differences between base and site specific conditions CM F

Part C: Predictive Methods (3/3) Local calibration factor (Cr) Adjusts SPF from HSM to local conditionsAddresses local variations C

HSM Prediction Models Prediction models forHomogeneous highway segments Intersections Segments based on homogeneousGeometryAADT ranges

Segments and Intersections

HSM Process Segment Facility Geometry, AADT Define Data Limits, Type, Time, Geometry, AADT Predict Crashes SPF Adjust Crashes CMF, C

Cross Sectional Elements

Safety and Operational Effects Lane Width Shoulder Width SideslopeClear Zone Crashes Operations Head-on Capacity Wider is “better” Wider means “faster” Run-off-Road Capacity Wider is “better” Functionality Run-off-road (severity) Maintenance Flatter is better Flatter is better Run-off-road Horizontal sight (frequency and severity) distance

HSM Crash Prediction Method Total predicted crashes: expected number of crashes Int.: Expected crash frequency for all intersections Seg .: Expected crash frequency for all roadway segments Int. Seg . Total Predicted Crashes

Prediction Model N predicted i = Nspf i x (CMF1 … CMF x) CiNpredicted i = Expected crash frequency for an individual element N spf i = Expected crash frequency for base conditions for an individual elementCMFi = Crash Modification Factors for individual design elementsCi = Calibration factor

Segment SPFs 2-lane RuralNspf rs = (AADTn) (L) (365) (10-6) e- 0.312 Multilane, Rural Undivided Nspf ru = e(a + b Ln AADT + Ln L) Multilane, Rural DividedNspf rd = e(a + b Ln AADT + Ln L)

Base Conditions: 2-Lane Rural Lane Width: 12 feetShoulder Width: 6 feet Shoulder Type: PavedRoadside Hazard Rating: 3Driveway Density: <5 driveways/miGrade: <3%(abs) Horizontal Curvature: None Vertical Curvature: None Centerline rumble strips: NoneTWLTL, climbing, or passing lanes: None Lighting: NoneAutomated Enforcement: None

Base Conditions: Multilane Roads AllLane width 12 feetShoulder type PavedNo lighting or automated speed enforcement Undivided Shoulder width 6 feet Sideslopes 1V:7H or flatterDivided Shoulder width 8 feetMedian 30 feet

SPF Application Example 2-Lane road with AADT = 3,500 vpd and 2 miles longN spf rs = (AADTn) (L) (365) (10-6) e - 0.312 =

Deviating from Base Conditions Use of CMFs Npredicted i = Nspf i x ( CMF 1 … CMF x)NOTE: for 2-lane rural roads a calibration factor (Cr) is also used 2- 33

CMF Notes Values from tablesSimilar to HCMSome require adjustment Computed for specific crash types Need to ensure crash distributionThe effect of application of multiple CMFs is multiplicative, not additive

2-Lane Rural Segment CMFs

CMF Example Calculation (1/5) CMF1r = (CMF ra – 1.0)pra + 1.0 Eq. 10-11

Crash Distribution

CMF Example Calculation (2/5) AADT 3,500 vpd; Lane width 11 feet CMF1r = (CMFra - 1.0) pra + 1.0 =

CMF Example Calculation (3/5) CMF2r = (CMFwra CMFtra– 1.0)pra + 1.0

CMF Example Calculation (4/5)

CMF Example Calculation (5/5) AADT 3,500 vpd; 4 foot turf shoulder CMF2r = (CMFwra CMF tra – 1.0)p ra + 1.0 =

Application of CMFs Npredicted i = Nspf i x (CMF1 … CMFx ) AADT 3,500, Length 2 miles, 11-foot lanes, and 4-foot turf shoulder N predicted rs = (1.87)(1.03)(1.12) = 2.15 crashes per year

Crash Modification Factors All crashes or some types/severityPart CFrom SPF development Part DIndependent researchCountermeasuresNumber of CMFs at once

Project Level Design element safety predictionUse of adjusted SPFsUse of CMFs Cost-benefit justification Estimate crash costsCompare with construction costs

Example #1

Example (1/5) AADT 3,500, 2 mile length, 11-foot lanes and 4-foot turf shoulder Improvement options12-foot lanesPave shoulder

Example – Option 1 (2/5) New CMF 1r = (CMF ra – 1.0) pra + 1.0 =

Example – Option 2 (3/ 5) CMF2r = ( CMF wra CMFtra– 1.0) pra + 1.0 =

Example (4/5) Anticipated crash reductions Option 1 3%Option 2 4%Crash predictionsN o = (1.87)( 1.03)(1.12) = 2.15N1= (1.87)(1.00) (1.12) = 2.09 N2= (1.87)(1.03)(1.08) = 2.09Which one to use?

Example (5/5) Estimated costs Option 1 $100,000/mile 10 yearsOption 2 $250,000/mile 20 yearsCost per crash reducedC1 = 100000/[(2.15-2.09)(10)] = $166,478 C2 = 250000/[(2.15-2.09)(20)] = $196,857

Cost-Benefit Analysis Treatment mayAffect severityNot affect number of crashesNeed to estimate crash savings over project lifetime Cost-Benefit can assist

FHWA Costs per Crash Fatal $4,008,900 Injury A $216,000 Injury B $79,000Injury C $44,900 PDO $7,400 Source: HSM 2010

Average Crash Cost Estimation Crash Severity Distribution Cost ($) Average Cost ($) Fatal 0.02 4,008,900 80,178 Injury A 0.05 216,000 10,800 Injury B 0.10 79,000 7,900 Injury C 0.15 44,900 6,735 PDO 0.68 7,400 5,032 Total 1.00 110,645 Option 1 monetary gains: (0.06)(110645) = $6,646 per year

Example #2

Example (1/3) Existing conditionsAADT 18,000; 2-lane rural road; 10-foot lanes, 2-foot shoulderOption 1 12-foot lane; 8-foot shoulder; 2-lane roadOption 212-foot lane; 8-foot shoulder; 4-lane divided

Example (2/3) Condition Crashes Existing 5.4 Option 1 2.9 Option 2 2.4 Costs Option 1 $7.2 mil/mile Option 2 $21.5 mil/mile

Example (3/3) Cross Section Crashes per Year Cost (millions) Miles 2 Lane, 10 ft L, 2 ft S 5.4 -- -- 2 Lane, 12 ft L, 8 ft S 2.9 $7.2 69.4 4 Lane, 12 ft L, 8 ft S 2.4 $21.5 23.3 Available budget $500 m to improve 2 lane roads Miles to improve w/$500 m

Cautions and Caveats It’s a modelData quality Randomness CMF effectMultiplicativeNumber of CMFsUnder review and evaluationCrash type and severityJudgment Understand strengths and weaknesses

Michael Vaughn, PE Mike.Vaughn@ky.gov (502) 782-4923 Highway Safety Improvement Program Division of Traffic OperationsKentucky Transportation Cabinet  Eric Green, PEEric.green@uky.edu (859) 257-2680 Research EngineerTraffic and SafetyKentucky Transportation Center