Summer Program on Transportation Statistics, Why Highway Crashes Have Recurring Patterns? - Roya Amjadi, Aug 14, 2017
1 like495 views
The presentation discusses recurring patterns in highway crashes, highlighting the influence of seasonal and time-based factors on crash frequency. Key objectives include the need for time series methodologies to enhance safety planning and proposing low-cost safety countermeasures targeting high-risk groups, particularly younger drivers. It emphasizes the importance of understanding crash data and contributing factors to develop effective prevention strategies.
Summer Program on Transportation Statistics, Why Highway Crashes Have Recurring Patterns? - Roya Amjadi, Aug 14, 2017
- 1. “What Governs Highway Crashes Recurring Patterns?” Roya Amjadi, FHWA, Turner-Fairbank Research Center SAMSI, Summer Program on Transportation August, 14, 2017 1
- 2. Overview • Purpose of Presentation – a crash course in highway safety • Highway Safety Concerns • FHWA and NHTSA roles in safety • Literature Review for Seasonal Adjustments • Seasonal and Recurring Patterns in Highway Crashes • Selected Crash Topics for Recurring Patterns • Summary • Discussion/Questions 2
- 3. Purpose of Presentation 1. Outreach and Communicate the recurring patterns of highway crashes. 2. Identify the need for using Time Series Statistical methodologies for highway planning, design, operation, and safety improvements. 3. Propose development of “highway crash specific” Time Series methodologies using crash data. 3
- 4. Evaluation of Low Cost Safety Countermeasures Is a 40 State Departments of Transportations study 4
- 5. Death Timing May Be Predictable Canadian Statistics’ Richard Trudeau Health Reports, Summer 1997, stated: • While death may happen at any time as a result of illness or accident, its timing is predictable to some extent. • Deaths due to specific causes tend to follow a yearly cycle. • Because seasonal upsurges of deaths from specific causes are predictable, preventive health and safety measures may reduce the toll. 5
- 6. Center for Disease Control Mortality Report, 2014 (2,626,418 total death) 1- Diseases of heart (heart disease) 614,348 2- Malignant neoplasms (cancer) 591,700 3- Chronic lower respiratory diseases 147,101 4- Accidents (unintentional injuries) 135,928 5- Cerebrovascular diseases (stroke) 133,103 6- Alzheimer’s disease 93,541 7- Diabetes mellitus (diabetes) 76,488 8- Influenza and pneumonia 55,227 9- Nephritis (kidney disease) 48,146 10- Vehicle Crashes (before 2007) 43,000 10- Intentional self-harm (suicide) 42,826 11- Septicemia (bacterial infection) 38,940 12- Chronic liver disease and cirrhosis 38,170 Vehicle Crashes 32,675 6
- 7. Why Seasonal Adjustment? Long Term Goal: 1. Eliminate cyclical/seasonal influences for development of reliable crash models. • Identify major concerns for highway crashes, and advance research for Safer design and operation. Short Term Goal: 1. Identify causes behind crash upsurges in certain time of day, day of week, month of year, or other time periods. • Determine “effective” low-cost safety countermeasures . 7
- 8. Highway Crash Major Contributing Factors: 1. Driver (i.e. age, gender, behavior, intoxication) 2. Vehicle (i.e. body type) 3. Road geometric design (i.e. number of lanes, speed limit, curve, grade, and median) 4. Roadside Design (i.e. structure, culvert, ditch, curb, sign, light pole, and fence) 5. Environment (i.e. weather, light, and trees) 6. Other major factors (i.e. economy, and TBD) 8
- 9. FHWA and NHTSA Roles In Highway Safety FHWA: • Works with State and responsible for: – the design, construction, and maintenance of the roads – Responsible for “roads” safety NHTSA: • Directs the highway safety and consumer programs for: – Driver – Vehicle 9
- 10. FATALITY ANALYSIS REPORTING SYSTEM (FARS) Crashes with Minimum One Fatality • Since 1994, highway crashes have caused: – Average 39,348 fatalities per year – Over 2.4 million nonfatal injuries per year • Cost over $200 billion to our national economy per year 10
- 11. Literature on Seasonal Adjustment Very Limited! • A. Karlsson and K. Willero, Time Series Analysis of Fatalities in the Traffic, Swedish National Road and Transport Research Institute, 2005 – used time series models to explain correlation between “observed” and “estimated” number of crash fatalities. • The predictions showed that Swedish “Zero Vision” will not be fulfilled without radical changes in highway safety and operation. • Predicted number of fatalities will stay at the same level as in recent years.
- 12. NHTSA Fatality Analysis Reporting System (FARS) FARS is a database for crashes with Minimum One Fatality. • Used data for 2010 to 2015 (6 years.) – Isolated data for: 1. Vehicle “Number 1” AND 2. Person Type “Driver” for vehicle number 1. Extracted: I. Non-Intersection cases= 125,680 crashes II. Intersection cases= 1,948 crashes Only used Non-Intersection data 12
- 13. Intersections • Where two or more roads meet with different traffic volumes. • May be stop controlled or traffic signal controlled. • By design have different road geometrics, speed (if not stopped), lighting, and pedestrian volume. • Traffic flow is not all straight forward, and vehicles may turn in different directions. • Have many accesses, or access control to business or other properties located around intersection that may flow dynamics. • Other different characteristics from straight or curved segment of the road. 13
- 14. Recurring Highway Crash FARS Data, 2010-2015 Selected crash highlights: 1. Hourly, weekly, and monthly crash reoccurring patterns 2. Driver Age 3. Cause of Crash (first harmful events) a. Motor Vehicle in Transport b. Pedestrian c. Curb 4. Vehicle (Motorcycle) 5. Vehicle Speed 14
- 15. All NON_INTERSECTION FIRST HARMFULL EVENT 15 Event Frequency Percent Number Total 125,680 100 1 Motor Vehicle In-Transport 33,097 26.3 2 Pedestrian 17,755 14.1 3 Rollover/Overturn 14,839 11.8 4 Tree (Standing Only) 13,123 10.4 5 Ditch 5,738 4.6 6 Embankment 5,463 4.3 7 Curb 4,127 3.3 8 Utility Pole/Light Support 3,644 2.9 9 Guardrail Face 3,643 2.9 10 Culvert 2,403 1.9 11 Pedalcyclist 2,385 1.9 12 Fence 2,166 1.7 13 Traffic Sign Support 1,798 1.4 14 Mail Box 1,629 1.3 15 Parked Motor Vehicle 1,548 1.2 16 Other Post, Other Pole or Other Supports 1,329 1.1 17 Other Fixed Object 1,323 1.1 18 Concrete Traffic Barrier 1,101 0.9
- 16. All “Non-Intersection” Crashes It has recurring patterns for 24 hours of the day 16
- 19. All “Non-Intersection” Seasonal Crashes Crash patterns are different per day and night. 19
- 20. Driver Age Category Pattern 20
- 21. 21 Age category crash percentages for; 15-19 through 50-54 are higher than their population percentages.
- 22. Driver Age category 18-29 Crashes have 24 hour, weekly, and monthly patterns. 22 - 42% of crashes occurred from 7 pm-3 am. - 55% of crashes occurred on Friday, Saturday, and Sunday.
- 23. SPEED RELELATED CRASH Sample= 61,327 Odds of Having a Fatal Crash for Age 18-29 Compared to Age 40-50 Yes, Racing 5.73 Yes, Exceeded Speed Limit 1.68 Yes 1.39 Yes, Specifics Unknown 1.38 Yes, Too Fast for Conditions 1.31 Unknown 0.94 No 0.82 23 0% 2% 4% 6% 8% 10% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 CRASH PERCENTAGE PER HOUR OF DAY 2010-2015 Age 18-29 Age 40-50 FIRST HARMFULL EVENT Odds Curb 1.57 Utility Pole/Light Support 1.46 Tree (Standing Only) 1.26 Mail Box 1.21 Fence 1.16 Other Fixed Object 1.15 Embankment 1.09 Concrete Traffic Barrier 1.09 Guardrail End 1.07 Traffic Sign Support 1.06 Parked Motor Vehicle 1.04 Other Post, Pole or Supports 1.03
- 24. Driver Age category 18-29 Vehicle Occupant Number and Safety 24 NUMBER OF OCCUPANTS Sample= 61,327 Age 18-29 Cashes (40,169) Age 40-50 Crashes (21,158) Odds of Having a Fatal Crash Age 18-29 Compared to Age 40-50 1 24,960 15,949 0.82 2 8,731 3,410 1.35 3 3,418 929 1.94 4 1,768 438 2.13 5 804 198 2.14 6 253 100 1.33 7 106 50 1.12 8 55 23 1.26 9 23 20 0.61 10 12 9 0.70
- 25. Motor Vehicle in Transport Crashes 25 Have a 24 hour recurring pattern. - 36% of crashes occurred between 12 pm to 5 pm (5 hour peak.) - Crashes were 2.12 times more likely to occur in period of 12 pm-5 pm compared to crashes that occurred in rest of the hours.
- 26. Pedestrian Crashes 26 - 61% of crashes occurred between 5 pm-12 am (8 hour peak) - Crashes were 3.16 times more likely to occur in period of 5 pm-12 am compared to crashes that occurred in rest of the hours.
- 27. Curb Crashes 27 - 66% of crashes occurred from 7:00 pm to 4:00 am. - Crashes were 3.3 times more likely to occur in period of 7 pm-4 am compared to crashes that occurred in rest of the hours.
- 28. Motorcycle crashes 28 Have 24 hour, weekly, and monthly patterns. - 359% more crashes occurred between 12 pm-2 am compared to other hours. - 56% more crashes occurred in Fri-Sun compared to Mon-Thu. - 71% more crashes occurred in May-Sept compared to Oct-April.
- 29. Speed Related Crashes Low Hanging Fruit 33% of non-intersection related crashes were speed related Newton’s Second Law applies for constant mass F= m dv/dt = ma • Compared to non-speed related crashes, speed related involved: 1.43 times more fatalities 1.42 times more sever injuries 2.00 times more crashes on curves 29
- 30. 30 0.0% 0.2% 0.4% 0.6% 0.8% 1.0% 1.2% 1.4% 1 3 5 7 9 11131517192123 1 3 5 7 9 11131517192123 1 3 5 7 9 11131517192123 1 3 5 7 9 11131517192123 1 3 5 7 9 11131517192123 1 3 5 7 9 11131517192123 PERCENTOFFATALCRASHES Speed and Non-Speed Related Crash Patterns 24 Hours (6 years) Non-Intersection Crashes with Min. 1 fatality Speed = 38,656 (33%) Non-Speed = 80,197 (67%) Period: 2010-2015 Non-Speed Related Speed Related Linear (Speed Related)
- 31. Speed Related Crash SAFETY IMPROVEMENTS? 1. Low Cost (short term) In segment with high percent of crashes: • Lower speed limit and advisory speed • Install variable speed limit • Remove/relocate fixed objects • Increase pavement friction • Enforce speed at night • Public education/communication • Nighttime visibility (markers) 2. Design (long term) • Increase curve radius and super elevation • Implement “clear zone” 3. Other (policy, vehicle design) 31
- 32. Selected Crashes with Minimum One Fatality • Possible low cost solutions for: – Driver age 18-29 – Night time – Speed – Weekends – Pedestrian – Curb – Other What governs crashes recurring patterns? 32
- 33. Opportunities to Learn From History! • If highway crashes look chaotic and difficult to predict, that means we have not recognized their patterns yet. • “Those who fail to learn from history are doomed to repeat it” Sir Winston Churchill Highway crashes are predictable and preventable Propose development of “highway crash specific” Time Series methodologies for outstanding safety concerns using crash data. 33
- 34. THANK YOU! • Questions? Email: Roya.Amjadi@Dot.Gov Phone: 202-493-3383 34
- 35. CRASH DATA RESOURCES • Despite the limited statistical methodologies available, USDOT agencies such as the FHWA, NHTSA, and FMCSA have succeeded in developing and maintaining comprehensive and reliable data resources (i.e., Highway Safety Information System, Fatality Analysis Reporting System, and Motor Carriers Management Information System) that have historical crash and roadway data for multiple years. • The Transportation Research Board’s second Strategic Highway Research Program (SHRP 2) (2006-2015) has collected an unprecedented amount of real -time data on driver behavior and the driving context. • Most State DOTs collect and manage their own crash highway data for roads, traffic, operation, planning, construction, work zone, maintenance, pavement, bridges, structures, roadside objects, and anything within the right-of-way, including weather and pavement temperature. 35
- 36. Other Available Data for Highway Safety Research • FHWA’s Highway Performance Management System (other highway data) • Bureau of Economic Analysis • U.S. Censes Bureau (i.e. population, education, age) • Bureau of Labor Statistics (i.e. employment) • Navy’s Astronomical Information Center (i.e. hour of day) • National Oceanic and Atmospheric Administration (NOAA) National Centers for Environmental Information (i.e. temperature) • Department of Motor Vehicles (DMV) • American Automobile Associations (AAA) • Insurance companies • Others 36
- 37. FARS Links • FARS data at; https://www- fars.nhtsa.dot.gov/Main/index.aspx • FARS data user manual at: https://crashstats.nhtsa.dot.gov/Api/Public/ ViewPublication/812315 37