DEVELOPMENT OF CRASH MODIFICATION FACTORS FOR FOUR LANE NATIONAL HIGHWAY

Abstract

In India rapid enlargement and expansion of the road system and the rise in count of motorized vehicles have resulted in a significant rise in levels of both freight movement and passenger. In tandem safety related issues have emerged. In recent years, the number of road accidents and fatalities have been growing significantly. The increasing rate of accidents on roads put a question mark on safety of Highways and to manage them is of vital importance. In India, more d than 1, 38,258 fatalities occurs on highway system every year (till May. 2013). An accident takes place every 1.5 minutes and in every 6.2 minutes a person in road accidents. The national highways constitute only 2% of the total road network but carry 40% of the traffic and accounts for 25% of road crashes, out of which 28% to injuries and 34% lead to fatalities. For management, it is essential to develop crash reduction tecimique for safety on road. Crash Modification Factor (CMF) is a technique that can be used as an effective safety tool. Crash modification factors (CMFs) are used to measure the safety impacts of modifications with respect to identified geometric features. A stepwise linear regression method known as Multivariate Adaptive Regression Splines (MARS) was applied to develop CMFs for various roadway parameters such as median width, shoulder width, Barrier Availability for total crashes taking a section of 55 Km of National Highway 58. This method is suitable for high dimensional inputs and serves as extension of linear model that can model linearities also and the impact of more than one variable on crashes is also considered. The methodology further gears crash predictions from the model developed to find changes in geometric design features. One year of crash data were used for the analysis. The results shows that MARS have high prediction proficiency and showed satisfactory goodness-of-fit statistics. Median width and Median opening were identified as the influential variables. As per the results if there is no median opening, there is a reduction in total crashes by 13% but as the number of openings per kilometer increases, crashes also increases by 15% and 32%. Similarly, the places where the median width is 1 .2m (e.g. flyovers), at that place there is 4% increase in crash or we can say probability of accident at such location is high, but as the median width increases to 11.5 (e.g. near Khatauli bypass), the crashes decreases by 7%.