1. Traffic & Safety Academy Slides from Traffic and Safety Academy
CE 635
Fall 2016
Slides from
Traffic and Safety Academy
Eric Green, MSCE, PE, PhD Candidate
Kentucky Transportation Center

2. Traffic & Safety Academy
Fall 2016
Introduction
Safety Culture and Needs

3. Learning Objectives Introduce Highway Safety Manual
Present new safety approaches
Demonstrate crash data access, limitations, and interpretation
Illustrate Highway Safety Manual applications

4. Traffic & Safety Academy
Fall 2016
Why Are We Here?
Safety culture
Radical change in how safety is quantified
Aid in accessing and understanding crash data

5. How is Highway Safety Used?
Project level decisions
Design elements
Intersection improvements
Roadside improvements
System level decisions
Project selection
Project priorities

6. Current Approach Use design criteria Crash rates
Adherence to criteria does not guarantee safety
Crash rates
Volume effect not linear
Geometry and cross section effects
Crashes are random events

7. What about these alternatives? Is one “safer” than the others?

8. Would You Fly or Drive?
Would you get on a plane if a full 747 crashed every week?
33,561 fatalities in the US in 2012
Although this doesn’t account for exposure… Odds:
-1 in 11 million
-1 in 5000

9. Highway Safety Manual Systematic crash evaluation -Crash frequency
-Prediction of crash frequency
Ability to quantify effects of design choices
-Geometry
-Cross sectional elements

10. Traditional Approaches
Buildup
CRC
Crash rates
Crash frequency
Newspaper

11. Alphabet Soup – (1/2)
MMUCC – Model Minimum Uniform Crash Criteria – provides consistency of crash data state-to-state
MIRE – Model Inventory of Roadway Elements – provides consistency for roadway data
MUTCD – Manual on Uniform Traffic Control Devices

12. Alphabet Soup -(2/2) KYTC – Kentucky Transportation Cabinet (KY’s DOT)
KTC – Kentucky Transportation Center – University of Kentucky research center
FHWA – Federal Highway Administration
NHTSA – National Highway Traffic Safety Administration

13. KABCO (K) Fatal (A) Incapacitating Injury
(B) Non Incapacitating Injury
(C) Possible Injury
(O) Property Damage Only
Kentucky recently modified these categories to confirm with MMUCC
(A) Suspected Serious Injury
(B) Suspected Minor Injury

14. Kentucky’s Traffic Records Coordinating Committee
Kentucky Traffic Records Advisory Committee (KTRAC)
Mission: Reduce the number of fatalities and injuries and the severity of injuries related to road trauma
Interagency and intergovernmental
Voluntary membership

15. GPS study overview Looked at location of a random sample of crashes
Assessed accuracy by reading narrative
Plotted crashes by milepoint and latitude/longitude

17. Paper Reports Submitted manually
GPS from Magellan, Google® Maps, 911 system
County, Route and Mile point (CRMP) using reference system

19. Traffic & Safety Academy
Fall 2016
eCrash KYOPS
MapIt
Added October 1st 2007
CRMP data linked to GPS
Between/Intersect streets
Searchable
RT Unique:
“034-US ”

20. Crash Data Data limitations -Quality and accuracy
-Reporting thresholds
-Severity determination
-Differences among jurisdictions

22. Crash Data plotted by GPS Showing County Check

23. Conclusions/Recommendations
92% of all crashes were accurate compared to around 50% in the previous study
Most of this improvement can be attributed to the implementation of the MapIt system in eCrash
A large majority of incorrectly located were largely due to a lack of reference points
The MapIt system currently requires user to know where they are on a map
Identified errors should be investigated by KYOPS

24. Obvious Patterns Jefferson County 054-WK-9001 -10 -Route
could be based on frequency/
paper
054-WK
-Route
re-designation

25. Traffic & Safety Academy
Fall 2016
Other Accuracy Issues
Non-descriptive codes
-Unknown
-Inattentive/distracted
Properly coded crashes
-Intersection and ramp indicators
Roadway Geometry
-Usually matched with HIS
-Current study investigating this issue
Unhelpful codes are often used as a default. Simple yes/no fields are often misused or ignored such as ramp and intersections. These should be automatic.

26. Improvements to KYOPS
Kentucky’s Open Portal Solution (KYOPS) is used to input and view crash data (as well as other police databases)
Aerial photos
Integrated GPS receivers
Training
Integration with KYTC’s HIS roadway data
MMUCC compliance (Model Minimum Uniform Crash Criteria)

27. Traffic & Safety Academy
Fall 2016
Summary
Kentucky crash data is very good compared to other states
Know the limitations
Use latitude/longitude or roadway
ID/derived milepoint to identify location
Be aware of paper reports (1.3% as of 2012)
Crash reports are secondary to public safety
Narratives can help decode non-descriptive codes
Codes of unknown or inattentive

28. Critical Rate Analysis
The old way of quantifying safety

29. Critical Rate Analysis
Traditional way to quantify safety
Comparison of a section of road to an average section of road of a similar type
Comparison of the average rate to that of the critical rate (incorporates a 99.5% confidence)

30. What are Buildup and CRC?
Means to rate and rank roadway based on their collision rates
Contain collisions from the CRASH database that:
Have an identified route number
Have an identified mile point

31. Critical Numbers
Critical numbers were calculated:

32. Using the Buildup – Critical #
Consult Table 6 for the Critical Number to Use Based on the Roadway Type

33. Critical Crash Rates
Cc= critical crash rate Ca= average statewide crash rate for type of intersection K= constant related to level of statistical significance selected (a probability of was used, K = 2.576) M= exposure (M is in terms of million vehicles (MV)

34. Actual Crash Rates

35. Critical Rate Factor Ratio of Actual Rate over Critical Rate
CRFs over 1 are more “dangerous” than the average section
Typically sorted descending
Prioritization lists can be generated from the top X percentage

36. CRF Limitations
Collisions only with county, route, mile point other collisions excluded
Cookie cutter lengths to find sections - it doesn’t find the best section
Collision reports are not correct all the time – read them carefully

37. CRF Limitations Sites with low AADTs tend to move to the top
No accounting for sites with zero crashes
Crash rates may be misleading

38. Traffic & Safety Academy
Fall 2016
Utility Pole Impacts

39. Traffic & Safety Academy
Fall 2016
Analysis
Objectives
Identify “Hot Spot” Pole Impact Locations
Methods
State Crash Data ( )
Location determined by:
County, Roadway
Direction of Travel
Mile Post (1/10 mile)

40. Occupants in Utility Pole Impacts State Crash Data 2003-2005
Traffic & Safety Academy
Fall 2016
Occupants in Utility Pole Impacts State Crash Data

41. Traffic & Safety Academy
Fall 2016
Interventions
Move poles away from roadway
Bury the cables
Protective guardrail

42. Change in Occupant Exposure
Traffic & Safety Academy
Fall 2016
Change in Occupant Exposure
No Interventions Were Put in Place Between

43. Traffic & Safety Academy
Fall 2016
Regression To The Mean
Sir Francis Galton, 1887
Sweet pea seed size in successive generations
Federal Highway Administration
U.S. Department of Transportation
Mean

44. Traffic & Safety Academy
Fall 2016
Federal Highway Administration
U.S. Department of Transportation
Regression to the Mean
It is a statistical phenomenon resulting from repeated observations of the same subject occurring with random error around a “True Mean” - Barnett, Van der Pols, and Dobson(2005)
Values are observed with random error around a true mean
Smaller Seeds
Mean
Larger Seeds

45. Traffic & Safety Academy
Mason
City
Waterloo
Cedar Rapids
Quad Cities
Des Moines
Council Bluffs
Iowa
Ames
Sioux
Dubuque
Fort
Dodge
Ottumwa
Marshalltown
Spencer
Clinton
1 yr
of data
Crash Density – 1 Year Average
Annual Fatal and Major Injury Crashes Per Mile
Traffic & Safety Academy
Fall 2016
Sample - DRAFT
Sample - DRAFT

46. Traffic & Safety Academy
Mason
City
Waterloo
Cedar Rapids
Quad Cities
Des Moines
Council Bluffs
Iowa
Ames
Sioux
Dubuque
Fort
Dodge
Ottumwa
Marshalltown
Spencer
Clinton
3 yrs
of data
Crash Density – 3 Year Average
Annual Fatal and Major Injury Crashes Per Mile
Traffic & Safety Academy
Fall 2016
Sample - DRAFT
Sample - DRAFT

47. Traffic & Safety Academy
Mason
City
Waterloo
Cedar Rapids
Quad Cities
Des Moines
Council Bluffs
Iowa
Ames
Sioux
Dubuque
Fort
Dodge
Ottumwa
Marshalltown
Spencer
Clinton
5 yrs
of data
Crash Density – 5 Year Average
Annual Fatal and Major Injury Crashes Per Mile
Traffic & Safety Academy
Fall 2016
Sample - DRAFT
Sample - DRAFT

48. Traffic & Safety Academy
Mason
City
Waterloo
Cedar Rapids
Quad Cities
Des Moines
Council Bluffs
Iowa
Ames
Sioux
Dubuque
Fort
Dodge
Ottumwa
Marshalltown
Spencer
Clinton
10 yrs
of data
Crash Density – 10 Year Average
Annual Fatal and Major Injury Crashes Per Mile
Traffic & Safety Academy
Fall 2016
Sample - DRAFT
Sample - DRAFT

49. Crash Frequencies
Crash analysis using crash frequencies also suffer the limitations such as regression to the mean
A countermeasure can reduce the severity of a crash which may not lower the crash count
Naïve before and after can be misleading

51. Traffic & Safety Academy
Fall 2016

52. How to address RTM More data More years of data Empirical Bayes
Some sites have small sample sizes
More years of data
Too long and you may be introducing unrealistic variables
Empirical Bayes

53. Kentucky Roadway AADT = 15,000 Crashes (5 years) = 50 Crash Rate = 179
Length = 1.0 mile
𝐶𝑟𝑎𝑠ℎ 𝑅𝑎𝑡𝑒= 𝐶𝑟𝑎𝑠ℎ𝑒𝑠∗100 𝑀 𝐿𝑒𝑛𝑔𝑡ℎ∗365∗𝑌𝑒𝑎𝑟𝑠∗𝐴𝐴𝐷𝑇

54. Kentucky Roadway AADT = 30,000 AADT = 15,000 Crashes (5 years) = 85
Casinos save lives!!!
AADT = 15,000
Crashes (5 years) = 50
Crash Rate = 179
Length = 1.0 mile
AADT = 30,000
Crashes (5 years) = 85
Crash Rate = 155
Length = 1.0 mile
𝐶𝑟𝑎𝑠ℎ 𝑅𝑎𝑡𝑒= 𝐶𝑟𝑎𝑠ℎ𝑒𝑠∗100 𝑀 𝐿𝑒𝑛𝑔𝑡ℎ∗365∗𝑌𝑒𝑎𝑟𝑠∗𝐴𝐴𝐷𝑇

55. What Did We Learn?
Crashes and AADT are not necessarily linearly related
Crashes are random events
Before and after analysis can be misleading
Location data can be incorrect
Crash codes can be incorrect