Bicyclist Health and Safety Issues on Four Alternative

Transportation Routes

Monroe County Alternative Transportation Plan Risk Assessment

Craig Harper · Zeynep AltinayCourtney Bonney ·Max Jie Cui

Outline

Health

• Health Effects• Exposure• Modeling• Risk Characterization

Safety

• Causes of Accidents• Predictive Modeling

•Intro and Problem Formulation•Current Routes and Alternatives•Hazard ID

•Uncertainty•Recommendations•Conclusion

Introduction

Problem Overview: When sharing roads with vehicles, bicyclists are exposed to health and safety risks from exhaust and accidents

Key Question of Project: How would risks to bicyclists change on 4 priority routes with alternatives considered by Monroe County?

Goal: Inform the Monroe County Council of the relative risks to both health and safety from current and alternative routes.

Target population

• Current study:– Adult male

• 18-30 years old• 70 kg weight• Cycling at a moderate

pace (13 mph)

– Asthmatic adult male• Future Studies:– Adult female– Elderly– Children

Hazard ID

• Criteria Pollutants– Sulfur Dioxide– Nitrogen Oxides– Particulate matter < 2.5 µm

• Hazardous Air Pollutants– VOCs (e.g. Benzene)

Pollutants emitted from vehicles Function of: fleet makeup, traffic volume, fuel composition, season

Dispersal of PollutantsFunction of: wind velocity, mixing height, season, buffer width

Inhalation of Pollutants Function of: inhalation rate (varies with population)

Pollutant DoseFunction of: absorption

Health Response (acute or chronic)

Endpoints: Risk to bicyclists from particular pollutants (mg/kg/day) over the course of 30 years

Accident Rate

Road Characteristics(type, lane width, shoulder, sidewalk, signage, bike lanes, etc)

Traffic and Bicycle Volume (vary spatially and temporally)

Confounding FactorsWeather, distractions

Driver and Bicyclists Error

Endpoints: Predicted number of accidents on a given route (accidents/year)

Bicyclists’ Health and Safety: A Conceptual Site Model

Bicyclists share roads with vehicles

ROUTES

Route 1State Road 46Commuter Route with a possible greenway option that would encourage recreational users

Vehicle Traffic Volume:Current: 10704-19071 Avg: 15000Alternative 1: 10700Alternative 2: 4900-13000

Route 2State Road 45Recreational Route from Lake Lemon into Bloomington

Vehicle Traffic Volume:Current: 3422-11491 Avg: 5225Typical Multiuse Volumes:

Route 33rd to Ivy TechCommuter Route to Ivy Tech

Vehicle Traffic Volume:Current: 102 – 42803 Avg: 17100Alternative 1:Alternative 2:

Route 4Fairfax RdRecreational Route from Clear Creek Trail head to Monroe Lake Beach and Four Winds Resort

Volume:Current: 49-6860 Avg: 2270

OptionsCurrent

ConditionsSigned Bike Route Urban Bike Lane Multi-Use Path

Hazardous shoulders for road

bikes

Safe for use by both vehicles and

bicycles

Greater predictability of

bicyclists and cars

Dangerous at intersections if

unmarked

Suitable for the most experienced

bicyclists

Not suitable for street

w/inexperienced bicyclists

Greater confidence for inexperienced

bicyclists

Creates a feeling of “green space” for

riders and pedestrians

More dangerous at blind hills and curves and stops

Suitable for low speeds and traffic

volumes

Speeds >40mph, w/curb and/or

gutters

Could cause inverse effect due to higher pedestrian volumes

Minimum 10.5’ Vehicle Travel

Lane (VTL)

Minimum 14’ VTL Minimum 10.5’ VTL plus 5’ marked

bike lane

Minimum 10.5’ VTL, 3-6’ buffer, and 5-10’ paved

trail

HEALTH

Exposure: Methods

• EPA ‘s Mobile 6.2 Emissions Modeling Software

• Estimates emissions (g/s or g/day)• Assumes average fleet makeup, traffic

volumes, seasonal variations, fuel composition, average speed

http://elseware.univ-pau.fr/MAINPAGEPUB/carpollu/pol1.gif

(Schnelle and Dey, 2000)

Dispersion Box ModelConcentration (C)

Where ,The emission rate per unit area

Assumptions of the box model:1. Concentrations are homogenous

within the box.2. Sources distribute uniformly.3. Emitted pollutants instantaneously

and uniformly mix.4. A wind of constant speed flows

across the cells cross-sectional area

uH

qLC

LWQq

Calculation of Intakes

Where:

• I ≡ intake (mg/kg bodyweight/day)• C ≡ chemical concentration (mg/s)• CR ≡ contact rate (m3/hr)• EFD ≡ exposure frequency and duration

– EFD = EF*ED• EF ≡ exposure frequency (days/year)• ED ≡ exposure duration (years)

• BW ≡ bodyweight; the average bodyweight over the exposure period (kg)• AT ≡ averaging time; time over which exposure is averaged (days)

ATBW

EFDCRCI

1

Combined Health Effects• Respiratory

– Inflammation– Reduced Lung Function (FEV1/FVC)– Increased Upper Respiratory

Infections• Bronchitis• Pneumonia

– Allergic Reactions– Exacerbation of COPD, Asthma,

and Emphysema

• Central Nervous System– Headaches/Dizziness/Vomiting– Brain damage

• asphyxiation

– Stroke– Coma (VOCs)

• Cardiovascular– Increased myocardial

ischemia• Pro-inflammatory mediators

– Atherosclerosis• Leukocyte and platelet

activation

– Arrhythmia– Increased risk of diabetes

and hypertension

• Cancer– Lung Cancer– Leukemia

• Premature Death

Non-Cancer

• Reference dose=Threshold Dose/U.F.• U.F.s depend on the type of study• Large RfCs indicate weaker pollutants

Pollutant EffectLOAEL (ug/m^3)

NOAEL (ug/m^3)

U.F. Total M.F. U.F. * M.F.

RfC (mg/m^3) Study

Benzene

Decreased lymphocyte count 17100 30 9 270 6.33E-02 Ward et al. 1985

PM2.5 11 10 3 30 3.67E-04Harvard Six Cities, ACS, Dockery et al.

SO2 FEV decline 1144 10 3 30 3.81E-02 WHO 2006

NOx

decline in pulmonary function 1880 10 3 30 6.27E-02

EPA 1993, Berglund M. et al. 1993

U.F. Modifying FactorThreshold Dose Reference ConcentrationPollutant and Effect

Cancer

• VOCs (Benzene as an example) and PM have the ability to cause cancer

• Risk measured as Unit Risk: risk per µg/m3 breathed

• Benzene – Leukemia (EPA, 1998)– Unit Risk = 7.8E-03 (mg/m3)-1

– Slope Factor = 2.73E-02 (mg/kg-day)

33 1020701

daymkgSFRisk

lyIntakeChronicDairSlopeFactoRisk

Modeling Uncertainties

• Calculated RfC from threshold doses – corrected for uncertainty (see table)

• Utilized @Risk to run 5000 iterations – 5 frequency durations ranging from 50-250 days

• Used @Risk to place uncertainty values around:– wind speed

– mixing height

– width of box

ATBW

EFDCRCI

1uH

qLC

LW

Qq

RfC

IHQ

Non-Cancer Output from @risk

• Calculated a HQ with nested uncertainties for the longest route in 4 seasons

• NOx: HQ>1

• All other pollutants HQ<1

• Relative Hazard Index, sum of the HQs, calculated for varying proposed alternatives

Results: Non-Cancer

EF (days) HI (current) HI (signage) HI (bike lane) HI (multiuse)

50 0.945 0.385 0.348 0.230

100 1.891 0.771 0.697 0.460

150 2.837 1.155 1.046 0.691

200 3.783 1.541 1.395 0.921

250 4.728 1.926 1.744 1.152

Using the Mean

Results: Cancer

BenzeneIntake = 0.14 µg/m3

Risk Level Concentration

E-4 (1 in 10,000) 13.0 to 45.0 µg/m3

E-5 (1 in 100,000) 1.3 to 4.5 µg/m3

E-6 (1 in 1,000,000) 0.13 to 0.45 µg/m3

Data gaps/uncertainty

• Mobile 6.2– default traffic volume assumption– no account of road dust– exposure from ingestion– mixing height assumptions– interactive effects of pollutants

SAFETY

3 June 2008, US-Mexico Border

Bicycle Accident Rates: Contributing Factors

• Road Characteristics

• Traffic and Bicycle Volume

• Confounding Factors

• Driver and Bicyclist Error

Predictive Modeling of Accidents:Data Sources

Bloomington Police Department

Bike and Pedestrian accident of 2008

DOT (Department of Transportation)

Average Traffic flow data of a day

Griffy Weather Station Precipitation data of 2008

Self-collection On-site risk factors

Assumption: Bicycle/pedestrian volume

Months of CyclingMichael Steinhoff and Julie Harpring. (2008). Transportation and Sustainability

on the Indiana University, Bloomington Campus.

For the pedestrian, we assume the flow is relatively the same in spring, fall and winter. We dropped the number by 70% for summer because most of the

students will go home. We also assumed that the behavior of residents remains constant throughout the year.

Variables for 2 Types of Model

Model 1 Variables Model 2 Variables

Bicycle Flow Lane Width

Pedestrian Flow Bike Lane Width

Weekday/Weekend Traffic Flow

Precipitation Intersections

Bicycle flow2 Crosswalk, Curb

Pedestrian flow2 Commercial vs. Residential

Model Type I – Bicycle

abm = Hourly Bike flow adjusted by month; t=7.07

aps = Hourly pedestrian flow adjusted by season; t=5.5

week = (Weekend=1, weekday=0); t=4.4

Y = − 0.00308 + 0.70576abm – 0.00513aps – 0.25012week + 0.00014143B2 + ut

R2= 16.36% F=17.5 P=0.0001

Y = Number of accident(s) on each day of 2008

B2=Abm2 ; t=0.88

Model Type II

Y = 0.61697 +0.00005965TF +0.06912LW2 +0.19403BLW -0.84127Int -0.28712Curb -0.21508SD+ 0.34976 CR

R2=96.63% F =36.81 P=0.0001

TF = Average Traffic Flow per day (2008) t=7.39

LW = Lane Width t=24.99

Intersection (INT) = (Yes=1, No=0) t=-3.55

Y = # of Accidents on each selected road in 2008

Curb (CB) = (Yes=1, No=0) t=1.88

CR = (Commercial =1,Residential=0) t=2.27

Sidewalk (SD) = (Yes=1, No=0) t=1.45

BLW = Bike Lane Width t=3.87

Limitations

Data is very limited in this

area.

Cannot account for human behavior

Mixed-Poisson Distribution

Model

Specification Error

Next steps to improve accident modeling

• Collect more data of risk characteristics on our primary routes (accidents!)

• Adjust the model by adopting Mixed Poisson Distribution and take human behavior into consideration

• Improve the assumptions by getting more official data

Conclusions

• Little evidence of serious risk due to air pollutants on current routes

• Cannot make predictions of accidents on rural routes based on our model

• Cannot make generalizations about effects of multi-use path with our model

• Traffic calming measures (reduction of volume) seems to be more effective at reducing accidents than adding bike lanes

Further Considerations

• Value of increasing perceived safety• Produce a map of county bike routes with

safety rating based on road characteristics to inform bicyclists of options