A key problem in traffic safety is to forecast the impact of a proposed strategy or new technology on safety performance, as characterized by the functional relationship between accident counts and traffic volume. This paper illustrates a practical, measurement-based approach to the safety performance prediction problem as applied to a prototype roadside warning system near Munich, Germany, which is designed to reduce the danger of secondary accidents following incidents on highways. Upon detection of an incident, a warning is transmitted quickly to drivers by illumination of a selected range of individually addressable, blinking roadside light posts with variable blink frequency and color. At the present implementation stage, vehicles are recipients, but not yet sources, of warning information, so that activation proceeds either manually or by means of a separate automatic incident detection system. As a basis for decision support and optimization, a series of experiments was performed measuring the precise effects of system activation on individual driver behavior and traffic characteristics. However, due to design constraints, the experimental incident was carried out under low-risk conditions: a truck was parked on the road shoulder in a mock breakdown. Nonetheless, judicious application of probability and statistical considerations, failure mode analysis, behavior modeling, and traffic flow modeling permits one to extrapolate measurements to a hypothetical high-risk incident and to draw quantitative inferences on the potential of the warning system for accident prevention. The paper reports the identification of a collective safety process of speed matching resulting from the activation of this warning system. Several lines of quantitative analysis, including Monte Carlo simulation techniques applied to the experimental measurements, support the conclusion that the proposed warning system will be effective in reducing secondary accidents (assuming primary incident detection and activation of the system are sufficiently fast): The warning increases alertness and far-sighted driving, resulting in a collectively induced increase in the viscosity of the traffic flow. This viscosity increase in turn reduces otherwise dangerous speed gradients. It is explicitly shown that this collective effect substantially lowers the probability of secondary accidents. The measurement-based analysis and simulation procedures reported here may also be useful for forecasting the potential safety benefits of related systems.

  • Availability:
  • Corporate Authors:

    PTRC Education and Research Services Limited

    Glenthorne House, Hammersmith Grove
    London W6OL9,   England 
  • Authors:
    • KATES, R
    • KELLER, H
    • Lerner, G
  • Conference:
  • Publication Date: 2000


  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 00797314
  • Record Type: Publication
  • Report/Paper Numbers: VTI konferens 13A, part 7
  • Files: TRIS, ATRI
  • Created Date: Aug 2 2000 12:00AM