USA Banner

Official US Government Icon

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure Site Icon

Secure .gov websites use HTTPS
A lock ( ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.

U.S. Department of Transportation U.S. Department of Transportation Icon United States Department of Transportation United States Department of Transportation
FHWA Highway Safety Programs

SECTION 4: HFST Site Selection Process

SECTION 4: HFST Site Selection Process

As mentioned in section 3.2, the initial selection of two high-crash locations with high friction demand served Kentucky well in launching their HFST program. The significant after treatment crash reductions were realized immediately, which led to the selection of additional sites to more clearly prove the benefit. To develop an effective HFST program, however, KYTC needed a process to screen for the most appropriate locations to be treated. This section describes the evolution of KYTC's site selection screening processes.

4.1 Crash Database - KYOPS

KYTC uses the Kentucky Collision Analysis for the Public (KYOPS) database that was developed and is maintained by the Kentucky State Police. This Kentucky crash database is considered comprehensive and of high quality for the KYTC road network. It is also considered a key asset for the agency's safety program. The crash database and information are available to the public at www.crashinformationky.org. The Kentucky State Police has continuously improved KYOPS in terms of both data quality and detail since 2000. Thus, KYTC has a high level of confidence in the data from 2008 and forward. To reduce errors and increase quality, the majority of data is collected electronically. Another quality element of the data collection system is that crash locations are generated automatically by the electronic data collection system.

4.2 HFST Site Selection Screening Process I – The 30 Worst

In 2009, at the time the first two HFST installations were ongoing, KYTC developed its first HFST site selection screening process. Since HFST was in its infancy, the process was designed to find sections that had a high probability of realizing benefits (i.e., crash reductions) in a short time frame as a means of continuing to gain support from KYTC personnel. Thus it was a reactive type screening process.

The process included an analysis of 3 years of roadway departure crash data (total crashes and total wet weather crashes) on the two-lane, two-way (TLTW) rural State road system, curves,5 and ramps (section lengths varied as a horizontal curve database was the source). The definition of roadway departure crash includes the following crash types in the Kentucky crash data base: non-intersection, non-parking lot, non-private property, single vehicle, fixed object, non-fixed object, head on, overturn/rollover, ran off road left/right/straight, sideswipe/opposite direction.

KYTC identified the 30 highest crash sections (i.e., those that had greater than 50 percent wet/dry crash ratio) as priority candidate sites for HFST. Because the two 2009 HFST installation sections were identified within these 30, KYTC had greater confidence that the screening process was reasonable.

4.3 HFST Site Selection Screening Process II – 2010 RWDIP

In 2009/2010 KYTC collaborated with FHWA to develop the KYTC Roadway Departure Safety Implementation Plan (RwDIP) discussed in detail above in section 3.1. The HFST site selection screening process incorporated in the RwDIP included 4 years of crash data (2004-2008) on the rural State road system, including roadway departure crashes (defined in section 4.1) (total crashes and total wet weather crashes). The road system analyzed was segregated into 3,000 foot sections (curve database was not utilized for this analysis) and identified by county, route, and milepoint. The threshold value to be considered a candidate HFST site was 8 wet total crashes (8 was selected due to data showing high return on investment) and a minimum wet to total crash ratio of 0.35.

The RwDIP site selection process identified 227 candidate HFST sections. All of the "Worst 30" were included in the RwDIP process. All 227 candidate sections have been field assessed (see section 4.4) to determine if HFST is appropriate treatment, and of these, approximately 160 of the candidate sections were determined to be appropriate for HFST installation. HFST installations based on the RwDIP process started in 2010, and by Fall 2012 approximately 160 sections had been constructed or programmed for construction.

4.4 HFST Site Selection Screening Process II – Using the Empirical Bayes Methodology

KYTC is currently developing a more advanced screening process with assistance from the Kentucky Transportation Center (KTC) at the University of Kentucky.

A regression model incorporating Empirical Bayes methodology is currently being used to screen the network for candidate segments to be treated with HFST. The Empirical Bayes methodology is a data-driven, science-based approach and is described in detail in the Part C of the AASHTO Highway Safety Manual.

The Empirical Bayes methodology is used to more precisely estimate the number of crashes that would have occurred at an individual treated site in the after period had a treatment not been implemented. The effect of the safety treatment is approximated by comparing the sum of the predicted number of crashes in the after period for all treated sites with the number of crashes, corrected using Empirical Bayes, recorded after the safety treatment was installed. The advantage of the Empirical Bayes approach is that it correctly accounts for observed changes in crash frequencies before and after a treatment that may be due to regression-to-the-mean.

The steps below outline the process specific to this application for HFST.

  • In assigning crashes to specific segments, KTC used 0.3 mile lengths that did not overlap.
  • KTC developed one safety performance function (SPF)6 for all wet crashes and one for serious wet crashes (fatal, incapacitating and evident injuries). The SPFs use both segment length and traffic volume as variables.
  • To apply the model to homogeneous sections, an SPF was developed for each highway type (e.g. rural two-lane, urban four-lane divided). The form of the equation is

y = L × ea × Vb

where:

y = predicted crashes
L = segment length
V = average annual traffic volume, and
a and b are coefficients that describe the behavior between length, volume, and the predicted number of crashes.

4.5 HFST Field Site Assessment

After the site selection screening process has identified candidate sections, a field site assessment is conducted to determine if HFST is an appropriate treatment. One of greatest time and cost factors of implementing HFST, but considered highly critical to KYTC success to date, is that one of the three HQ KYTC traffic engineers led each of the field assessments. Items considered during the field assessment included but were not limited to:

  • Drainage
  • Superelevation
  • Driveways
  • Coordinate future planned work on road section by the district
  • Sight distances
  • Crash analysis
  • HFST treatment for one or both directions of road
  • HFST limits
  • Confirm ponding water condition is not present
  • Constructability
  • Existing pavement condition has expected life of greater than about 3 years.

The value of having a small pool of personnel to lead each field assessment is consistency in conducting the field assessment and improving future field assessments by incorporating lessons learned from previous assessments. The major challenge of having a small pool of personnel is the time and cost to conduct the field assessment.