Identification of Promising Sites on Secondary Highways Using Inventory Data

1
Identification of Promising Sites on Secondary
Highways Using Inventory Data

• Joseph E. Hummer, Ph.D., P.E., NC State Univ.
• Carl Hultgren, P.E., Kimley-Horn Associates
• Asad J. Khattak, Ph.D., UNC-Chapel Hill

2
Sponsors

• Southeastern Transportation Center
• USDOT
• NC State University
• UNC-Chapel Hill
• Also thanks to Tennessee DOT and University of
  Tennessee

3
Motivation

• Rural, two-lane highways have high collision and
  fatality rates
• Agencies should identify and treat fixable
  (promising) sites
• How can they do that with poor or no collision
  data?

4
Problems with Collision Data

• Lower reporting rates in rural areas
• Officer has other priorities
• Location reported incorrectly
• Other data missing, incomplete or inaccurate
• Poor database management

5
Our Idea Only Use Inventory Data

• In many states inventory data are comprehensive,
  accurate and up-to-date
• Technology to collect and maintain inventory data
  is improving more rapidly than for collision data
• Proactive
• This paper reports development and testing of
  inventory method

6
Sites With Promise

• Foundation of our inventory method
• Developed by Hauer (1996)
• A site does not need to be unduly hazardous for
  there to be an opportunity to reduce accidents
  cheaply. Nor is it necessary to have accidents
  in a cluster for there to be a genuine need for
  remedial action.

7
7-Step Inventory Method

• Step 1 - Select population of highway sites
• For our application, two counties in Tennessee
• Good inventory data
• Good collision data (for comparison)
• Varied terrain
• Rural two-lane roads
• ADT below 5,000

8
Step 2 - Divide sites by type

• 28 bridge, 343 horizontal curve, and 57 segment
  sites in our application
• Good models available
• Most needed data available

9
Step 3 Predict collision frequency

• Use best available collision prediction model
• Bridgesused Turner model, 1984
• Curves--used Zegeer, et al. model, 1991
• Segmentsused Bared and Vogt model, 1997
• Should calibrate models before use

10
Step 4 Estimate CM cost

• 13 common countermeasures (CMs) considered for
  this application
• CM costs from North Carolina DOT

11
Step 5 Estimate collision reductions

• Estimate collision reduction potential of each
  possible CM at each site
• Use collision reduction factors (CRFs) on
  collision frequencies from Step 3
• For this application, used average of CRFs from
  California DOT and FHWA

12
Steps 6 and 7

• Step 6 - Determine cost-effectiveness of each CM
  at each site
• Combine results of Steps 4 and 5
• spent / collision saved
• Step 7 - Identify the most promising sites

13
Comparing Methods

• Did our application of the inventory method pick
  truly promising sites?
• Compare results to standard method based on
  collision data
• Tennessee DOT version of rate quality control
  method
• Same set of sites

14
Types of Sites Identified
15
Expert Opinion

• Photos and facts about 18 identified sites
• Sent to 35 highway safety professionals in 8
  Southeastern states
• DOT Traffic Engineers
• DOT Design Engineers
• Highway Patrol Officers
• FHWA Division Engineers
• 24 experts responded (69)

16
Two Questions for Experts

• How hazardous is each site?
• Scale of 1 to 7
• Which CM do you recommend for each site?
• Choose from a list of 19 CMs
• Choices included do nothing and other

17
Question 1 - Hazard Ratings
18
Question 1 - Hazard Ratings
19
Question 2 - Cost-effectiveness
20
Question 2 - Cost-effectiveness
21
Conclusion

• Agencies with poor or no collision data on
  secondary highways should consider inventory
  method to identify promising sites
• Inventory data often good and improving
• Models and CRFs available and improving
• Method simple to apply
• In our application inventory method performed as
  well as collision method according to experts