摘要: |
Understanding the relationships between roadside accidents and roadside design is imperative to developing cost-effective, road-related countermeasures to improve roadside safety. Much of what is known today about the relationships remains to be qualitative in nature. Recent studies have suggested that new, cost-effective analysis approaches and data collection efforts are essential if a mote quantitative basis of such relationships is to be developed. Historically, models used in previous studies to develop the relationships have been categorized as using either an accident based approach or an encroachment-based approach. The former has a solid statistical ground, but has been criticized as being overly empirical and lacking engineering basis. The latter, on the other hand, has analytical and engineering strengths, but has been described as being foil of subjective assumptions and lack of sufficient supporting data. In addition, these two approaches seem to have been treated as two competing, disconnected approaches, and very few attempts have been made by roadside safety researchers to combine the strengths of both. The purpose of this study was to look for ways to combine the strengths of both approaches. The specific objectives were (1) to present the encroachment-based approach in a more systematic and coherent way so that its limitations and strengths can be better understood from both the statistical and engineering standpoints, and (2) to apply the analytical and engineering strengths of the encroachment-based approach to the formulation of mean functions in accident-based models. To demonstrate the strength of mean functions so obtained, accident-based models were developed using such mean functions for guardrail and utility pole accidents. Furthermore, to show how the accident-based model can be useful to the encroachment based model, the developed accident models were used to estimate the roadside encroachment rate-a basic input parameter that is required by the encroachment-based model and is expensive and technically difficult to collect. The estimated rates were found to be consistent with those obtained in earlier encroachment-based studies. This is an indication that estimating basic encroachment parameters using accident-based models can be a viable approach to reducing encroachment data collection cost. In addition, unlike estimating encroachment parameters from the field-collected encroachment data, the use of accident-based models to estimate encroachment parameters does not require the development of a procedure to distinguish between controlled and uncontrolled encroachments, which can be subjective and technically difficult to do in practice. This paper concludes with a discussion on future research. |