Evaluation of Low-Cost Safety Improvements Pooled Fund Study 2017 Traffic Records Forum Dr. Carol Tan, FHWA, and Ms. Kim Eccles, VHB

Presentation Overview What are low cost safety strategies? Background and objectives Crash modification factors (CMFs) Example approach Recent results and findings Evaluations in progress Contacts and resources

Low-Cost Safety Strategies Generally infrastructure-based strategies that agencies apply as part of Highway Safety Improvement Program. These could be design strategies such as realigning a curve or an intersection (or pictured here, a road diet with restriping) or traffic control device improvements such as enhanced signing or markings. Around 100K or less.

Background and Objectives

Background on ELCSI-PFS Effort Need objective measures for effectiveness before investing Evaluation of Low-Cost Safety Improvements Pooled Fund Study FHWA established in 2005 to move tried and experimental to proven Initially 24 State members and 4 phases Goal: To develop reliable quantitative estimates of effectiveness in reducing crashes Effort continued in 2013 as the Development of Crash Modification Factors (DCMF) Program Evaluates higher cost improvement also Addressing methodological issues in CMFs and SPFs

40 States in the Pool – Coast to Coast Implement strategies Pool resources for evaluations Provide data on strategies (e.g. crash, roadway, volume) Provide direction to study Share best practices

CMFs and Better Decision-Making CMFs – Crash Modification Factors Estimates of the impact of strategies on crashes CMFs under 1.0 represent a reduction in crashes from applying strategy CMFs over 1.0 represent an increase in crashes Provide method to calculate comparisons between alternatives or projects Kim

CMFs and Better Decision-Making (Cont.) CMF usage in alternative countermeasure selection CMFs help answer the question – which pedestrian crossing strategy should I use to improve my specific midblock crossing? Each of the three strategies pictured has a different CMF.

CMFs and Better Decision-Making (Cont.) CMF usage in site selection prioritizing process CMFs help me answer the question – which of the three sites do I treat? I only have enough money to fund one project. Which one will give me the best b/c? I need to make the best use of my dollars.

CMFs: How do we know? Before/After Studies Cross-sectional Studies Compare safety performance before and after infrastructure change Numerous study designs often with some form of reference group Cross-sectional Studies Compares locations “with” to those “without” strategy Locations have to be very similar in all other regards New study design introduce techniques for this “matching” Most CMFs studies take one of two forms – Before/after or cross sectional.

CMFs: How do we know? There are a lot of sources of CMFs including this pooled fund study, NCHRP research, and research conducted by States or universities.

CMF Clearinghouse www.cmfclearinghouse.org FHWA Resource Quantifiable estimates of crash impact of strategies Up to date resource www.cmfclearinghouse.org

Example Approach to Strategy Evaluation

Edge Line Rumble Strips (ELRS) on Curves Pavement marking applied directly over rumble ELRS applications Milled rumble strips Audible vibratory pavement markings

ELRS on Curves Goals Provide better visibility for nighttime, wet pavement markings Alert motorists as they depart the travel lane Courtesy FHWA

Study Design – Study Questions What is the impact on crashes of applying this strategy? Total Fatal and injury Run-off-road Nighttime Nighttime run-off-road Do effects vary by type of treatment? Do effects vary by site characteristics?

Study Design – Methodology Empirical Bayes (EB) before-after Establish reference group Predict safety for after period assuming no treatment Observe actual safety for after period with treatment Compare the two Reference group Untreated sites adjacent to treated sites As similar to treated sites as possible Used to calibrate safety performance functions

Study Design – Initial Concerns Spillover or Migration Effects Spillover effect results in fewer crashes at adjacent sites after treatment Crash migration results in more crashes at adjacent sites after treatment Compared adjacent and non-adjacent reference sites

Volunteer States Florida Kentucky Mile-years of data 32 curves 6.41 miles Audible-vibratory Kentucky 229 curves 15.6 miles Milled stripEs Mile-years of data 210.36 Before 99.41 After

Analysis Findings Identify impact on total crashes and several subsets (e.g., fatal and injury, run-off-road, etc.) – aggregate models When possible, look at characteristics of sites – disaggregate models Benefit/cost analysis Total crash impact Service life Life cycle costs

Recent Results and Findings

Recent Results and Findings Over 40 evaluations conducted to date Edge Line Rumble Strips on Rural, Two-Lane Horizontal Curves Intersection Multi-strategy Signalized Intersection Multi-strategy Stop Controlled Corner Clearance at Signalized Intersections Profiled Thermoplastic Pavement Markings Cable Median Barriers with Shoulder Rumble Strips Horizontal Curve Realignment Flashing Yellow Arrow Right in/Right Out at Stop Controlled Intersections Red Light Indicator Lights

Edge Line Rumble Strips on Rural, Two-Lane Horizontal Curves Kentucky Ohio Statistic Total Injury ROR Nighttime Nighttime ROR Estimate of CMF 0.75* 0.64* 0.74* 0.63* 0.75 Standard error of estimate of CMF 0.09 0.14 0.11 0.19 Statistic Total Injury ROR Nighttime Nighttime ROR Estimate of CMF 0.79* 0.78* 0.75* 0.71* Standard error of estimate of CMF 0.04 0.07 0.05

ELRS on Curves: Economic Analysis Milled Rumble StripEs (KY) $2,500 per mile for rumble strip 12-15 year service life Stripe cost not added Audible Vibratory Pavement Markings (FL) $3,800 per mile for 6” audible vibratory marking Unknown service life

ELRS on Curves: Economic Analysis B/C Ratios Milled Rumble StripEs 121.6 Audible Vibratory Pavement Markings 5.8 assuming 2-year service life 27.12 assuming 10-year service life

Intersection Multi-Strategy Applications SCDOT statewide intersection improvement program Initiated in 2009 8,300 intersections screened 2,200+ intersections selected 2% of all State-maintained intersections 50% of intersection crashes and fatalities Focused on low-cost improvements Signals, signs, and markings Signalized Intersections (158 intersections) Stop-controlled intersections (918) Kim to do next series

Multi-Strategy: Signalized Intersections Signal improvements Replace signal heads (some intersections but not all) Replace pedestrian signal heads, pushbuttons, and signs Install backplates with retroreflective borders Restripe stop bars and crosswalks Install overhead signs R10‐12 R3‐5L R3‐5R

Multiple Strategies at Signalized Intersections

Multiple Strategies at Signalized Intersections Disaggregate results – compared across characteristics Appear to be more effective (total crashes) Urban intersections vs rural (0.95 vs. 0.98) Three legged vs four-legged (0.86 versus 0.97) 2 lane/2 lane vs. 4 lane/2 lane (0.78 versus 0.98)

Multiple Strategies at Signalized Intersections Economic Analysis Assumptions and Estimates 7 year service life Pavement marking and signing costs approximately $7,000/intersection Signal head replacement, ped improvements ranged from $10,000-18,000/intersection Estimated cost of AVG crash $95,186 in 2015 dollars Use the final rev tech brief in this file to finish this slide: \\ralnc\projects\38110.01 DCMF_TaskA\tech\Task A4\SC-Multiple Strategies\Tech Briefs\FHWA Review - Signalized\FHWA Final Review 07-19-2017

Multiple Strategies at Signalized Intersections (Cont.) Economic Analysis Results The treatments saved 50.6 crashes/year AVG reduction of 0.6 crashes/site/year across the 84 sites The treatments resulted in reduction of 26.5 fatal and injury crashes, 0.3 crashes less per site per year Various intersection treatments can be cost effective in reducing crashes.

Multi-Strategy: Stop Controlled Sign improvements Double 48” x 48” STOP and YIELD signs Double 36” x 36” intersection warning signs Double 48” x 48” advanced traffic control signs Fluorescent yellow sheeting Retroreflective sign posts Advance street names

Multi-Strategy: Stop Controlled Marking improvements Stop bars (placed within 4 – 10 ft of edge of through lane) Yield bars for yield conditions Dashed white edgeline through intersection (major road) Marked turn lanes Lane arrows Word “ONLY” * As needed, remark: Stop bars Crosswalks Arrows Words

Multiple Strategies at Stop-Controlled Intersections

Multiple Strategies at Stop-Controlled Intersections – Economic Analysis Construction cost - $5,900 7 year service life Average cost for a stop-controlled intersection crash was $132,071 in 2017 Service Life Lower Bound Average B/C Upper Bound 3 years 7.1 12.4 17.5 7 years 14.5 25.5 35.9

Corner Clearance at Signalized Intersections Presented poster at TRB (Jan 2017) Poster at Urban Streets Symposium (May 2017)

Corner Clearance: Data Collection Similar to RIRO efforts (HSIS, GIS, field visit) Initially 222 sites from California Supplemented with 53 sites from Charlotte, NC

Corner Clearance Aggregate Results: Total Crashes Number of Corners with Driveways within 50 ft CMF S.E. 1 receiving corner 1.33 0.11 2 receiving corners 1.76 0.30 Bold = statistically significant at 95-percent level Italics = statistically significant at 90-percent level

Corner Clearance Aggregate Results: Total Crashes Number of Corners with Driveways within 50 ft CMF S.E. 1 approach corner 0.82 0.08 2 approach corners 0.67 0.13 Bold = statistically significant at 95-percent level Italics = statistically significant at 90-percent level

Number of Access Points Removed Economic Analysis Treatment Remove access points within 50 ft on receiving corners $81 per linear ft of curb and sidewalk 100 ft of curb and sidewalk = $8,100 per access point Assumptions Considered total crashes only No construction of new access points No effect for leaving existing driveways on approach corners Number of Access Points Removed Lower B/C Average B/C Upper B/C 1 94.6 172.0 237.3 2 165.9 301.7 416.3

Profiled Thermoplastic Pavement Markings SC and FL provided data CMF of 0.908 for nighttime wet-road, not statistically significant at 95% confidence but consistent A B/C of 3.65 suggests treatment can be cost effective

Cable Median Barriers with Shoulder Rumble Strips

Cable Median Barriers with Shoulder Rumble Strips Conducted an EB before/after analysis Illinois Kentucky Missouri States differed in before conditions Over 400 miles of installed strategy, several years of data

Cable Median Barriers in Combination with Rumble Strips Kentucky, Illinois, and Missouri Crash Type CMF SE of CMF Total 1.247 0.034 Injury and fatal (KABC) 0.745 0.040 Injury and fatal (KAB) 0.783 0.073 Cross-median 0.119 0.053 This study evaluated safety effectiveness of cable median barriers in combination with rumble strips on the inside shoulder of divided roads. This strategy is intended to reduce the frequency of cross-median crashes, which tend to be very severe. Geometric, traffic, and crash data were obtained for divided roads in Illinois, Kentucky, and Missouri. To account for potential selection bias and regression-to-the-mean, an empirical Bayes before–after analysis was conducted, using reference groups of untreated roads with characteristics similar to those of the treated sites. The analysis also controlled for changes in traffic volumes over time and time trends in crash counts unrelated to the treatment. In Illinois and Kentucky, cable median barriers were introduced many years after the inside shoulder rumble strips were installed; therefore, the evaluation determined the safety effect of implementing cable barrier along sections that already had rumble strips. Conversely, in Missouri, the inside shoulder rumble strips and cable barrier were implemented about the same time. Hence, the evaluation in Missouri determined the combined safety effect of inside shoulder rumble strips and cable barriers. The combined Illinois and Kentucky results indicate about a 27-percent increase in total crashes; a 24-percent decrease in fatal, incapacitating, non-incapacitating, and possible injury crashes; a 22-percent decrease in in fatal, incapacitating, and non-incapacitating injury crashes; and a 48-percent decrease in head-on plus opposite-direction sideswipe crashes (used as a proxy for cross-median crashes). The results from Missouri for total and injury and fatal crashes were very similar to the combined Illinois and Kentucky results. However, the reduction in cross-median crashes in Missouri was much more dramatic—showing a 96-percent reduction (based on cross-median indicator only) and an 88-percent reduction (based on cross-median indicator plus head-on). The economic analysis for benefit-cost ratios shows that this strategy is cost beneficial.

Horizontal Curve Realignment Curves – 33% of fatal, single vehicle crashes Most involve striking a fixed object or overturning Signing and striping are lower cost but have shorter service life Source: FHWA

Horizontal Curve Realignment Conducted a before-after empirical Bayes evaluation of horizontal curve realignments California – 36 realigned curves North Carolina – 11 realigned curves Ohio – 15 realigned curves Reference curves identified from nearby curves on same route as treated curves Small but meaningful sample

Horizontal Curve Realignment: Data Sources HSIS (Highway Safety Information System) Road characteristics Traffic volume Crash data Project files Project specifics Location verification Map measurement Before and after curve radius Presence of nearby curves

Horizontal Curve Realignment California North Carolina Ohio Crash Type CMF S.E. Total 0.3 0.1 Injury and Fatal Run-Off-Road plus Fixed Object 0.2 Dark 0.6 Wet-Road

Horizontal Curve Realignment – Economic Analysis Cost for realignment -$251,558 30 year service life, could be longer Benefit to cost: 3.17 to 1

Flashing Yellow Arrow NV, NC, OK, OR

Flashing Yellow Arrow Category Before Phasing After Phasing Legs Crash Type CMF 1 Traditional PPLT FYA PPLT on one road 3 Total 0.849* KABC 0.791* 2 4 0.889* 0.801* FYA PPLT on both roads 0.818* 0.782* Permissive or Traditional PPLT FYA permissive on one road 0.997 0.808* 5 Permissive 0.915 0.787* 6 At least one protected phase FYA PPLT without TOD 1.051 1.011 7 FYA PPLT with TOD 0.974 1.089

Flashing Yellow Arrow Cost $6,500 for 4-leg intersection, $1,625 per approach in Oklahoma Cost $6,000 per approach in Illinois 10 year service life Very large benefit costs, depending on treatment category – 69:1 for category 2

Turning Restrictions at Stop-Controlled Intersections

Data Collection HSIS data GIS data Traffic data: AADT and design speed Geometric data: # of lanes and median type Crash data GIS data Access type and location Traffic control Turning restrictions Enriched and verified using aerial imagery

Aggregate Results: RIRO Restriction Expected crash reduction Crash Modification Factor

Aggregate Results: Downstream Intersections (Signalized) Expected crash increase Crash Modification Factor Expected crash reduction

Aggregate Results: Downstream Intersections (Stop-controlled) Crash Modification Factor Expected crash increase Expected crash reduction

Disaggregate Analysis Disaggregated by: Traffic Volumes Number of mainline lanes Design speed on mainline No differences detected

Economic Analysis Impact is highly dependent on local conditions Hypothetical stop-controlled intersection Downstream signal Treatment cost 4-ft wide median @ $6 per sq-ft ($24 per linear-ft) Average distance = 1,210 ft (between signals) Installation cost = $26,500 B/C ratio 9.6 (range: 5.4 – 13.5)

Considerations RIRO benefit Crash migration Potential crash reduction at treated locations More cost-beneficial when treated locations have higher safety risk compared to downstream intersections Crash migration Potential for crash increases at downstream intersections Type of downstream traffic control Signalized: smaller percent increase, not statistically significant Stop-controlled: larger percent increase, more statistically significant Disaggregate analysis No differences detected

Red Light Indicator Lights (RLIL) RLIL illuminates simultaneously with red signal phase Allows enforcement to enforce downstream

RLIL - Results Statistic Total Fatal & injury Right-angle Left-turn Rear-end Disobey signal Night Estimate of CMF 0.939* 0.856* 0.905* 0.600* 1.016 0.713* 0.892* Standard error of estimate of CMF 0.022 0.027 0.042 0.041 0.033 0.048 0.034 Presented poster at TRB (Jan 2017)

RLIL – Economic Analysis Costs $50 to $300 per bulb Installation - $3,000 per intersection 5 to 10 year service life, no maintenance costs Benefit to Cost Ratio 92:1 for four-leg signalized intersection

On-Going Evaluations

Other Strategies Under Evaluation Now Pedestrian strategies Prohibition of permissive left turns to improve pedestrian safety Leading pedestrian intervals Study locations Chicago New York Toronto Charlotte

Other Strategies Under Evaluation Now Highway Friction Surface Treatments (HFST) Provide guidance on the use of HFST to reduce roadway departure crashes. Materials and specifications Appropriate locations and conditions

Resources and Contacts

ELCSI-PFS https://www.fhwa.dot.gov/research/ tfhrc/projects/safety/comprehensiv e/elcsi/index.cfm Google or Bing “ELCSI-PFS” (It’s easier!) Website has full reports and tech briefs for all evaluations Links to related content

Questions? Roya Amjadi, Office of Safety R&D, FHWA project manager Roya.Amjadi@dot.gov Carol H. Tan, Office of Safety R&D, FHWA Carol.Tan@dot.gov Kim Eccles (contractor to FHWA) keccles@vhb.com