The methodology for predicting the expected level of traffic safety in the different types of level intersections. Thesis Università degli studi di Trieste.
20130531 ST [electronic version only]
Trieste, Università degli studi di Trieste, 2009, VI + 76 p. , 53 ref.
|Samenvatting||Traffic accidents cause high material and human losses, what is reflected in society. The logical result is the need for efficient road safety in designing a new and the existing road system. Slovenia is, like other new EU Member States, aware of its tasks for improving traffic safety. In accordance with very clear demands of European transport policy about road safety – that is an EU recommendation of halving the number of road accident victims in the European Union by 2010 – Slovenia has also put into its national program a decision to halve the number of dead casualties on Slovenian roads. Unfortunately, the current situation in the field of road safety in Slovenia is - despite the highly ambitious plans - still not satisfactory. It has to be admitted that traffic safety in Slovenia has been improved during the last years but we still have not achieved the objectives of reducing road accidents, injured participants or dead casualties. One of the "steps" to achieved desired level is also to improve existent road infrastructure. Road infrastructure improvement supposed to be applied to "black spots" first. For safety management it is well known, that we have three main motives for safety management: economic effectiveness, professional and institutional responsibility, and fairness. Survey among 25 EU states about estimating the most effective short, medium, and long term measures - both at national level and at EU level - shown, that measures related to infrastructure safety management (such as high-risk site management - black spot management), road safety audit and road safety inspection, are generally recognized as a high priority. While high-risk site management is a short-term measure, other infrastructure safety management measures make their impact in the medium to long term. One of the possible approaches to identification of "safety problems" (safety deficiencies: accident frequency, accident rate, accident severity) is also use of accident prediction models (APM). With those models and use of various criteria we can detect not only "black spots" but also "larger targets". A large number of "statistical predictive safety models" are described in the literature. Many attempts were done to use those models to establish a relationship between various traffic parameters and the number of accidents at road sections or road intersections. For statistical safety prediction models it is often suggested that accident occurrences are discrete, sporadic and random in nature. Thereby it is suggested to use Poisson regression models. The variation in accident occurrence is also considered to be due in part to the systematic variation in identified traffic measures such is traffic flow rates, measures of speed and intersection design parameters. Discrete, Poisson or negative binomial distributions are usefully applied to estimate the number of accidents that occur at road sections / intersections over a particular period of time. For accident prediction modeling the "generalized linear modeling" approach has been found to be particularly useful. This approach accounts for the fact that the dependent variable (e.g. number of accidents) does not need to be normally distributed (as is often the approach to describe the relationship between accident frequency and traffic flows on major and minor roads at intersections). For the purpose of this work I collect different types of data for 60 level intersections on state road network in Slovenia. I divide those intersections into four different groups: 3-leg without left turn lanes on major road, 3-leg with left turn lanes on major road, 4-leg without left turn lanes on major road and 4-leg with left turn lanes on major road. All intersections were in rural area - outside urban area - with limited influence (or no influence at all) of pedestrians or / and cyclists. For observed intersections I collect different types of needed data: data about traffic accident for last 5 years (from 2002 - 2006), data about traffic (AADT on major and minor road), data about geometrical elements of the intersections, alignment of legs and other needed data (lane / shoulder width, speed limit, lighting, present of left / right turn lanes, type of terrain etc.) After study of relevant literature I consider two different types of APM, which seemed to be correct and useful for my research work. I evaluate those different types of APM and made calculation for data which I obtained. At the end I made the correlation between those models with use of an empirical Bayesin method for calculating adjusted accident frequency. (Author/publisher)|
|Full-text||Beschikbaar Niet beschikbaar, klik om contact op te nemen voor een digitalisatie verzoek|
|Suggestie?||Neem contact op met de SWOV bibliotheek voor uw opmerkingen|