Modeling and simulation of car accidents at a signalized intersection using cellular automata

In this paper, we propose a two-lane cellular automata model that explains the relationship between traffic-related parameters and likelihood of accidents P a c at a signalized intersection. It is found that, the risk of collision rises as well as the lane-changing probability P c h g augments, besides, the accidents and inflow α show a nonlinear relationship. Moreover, P a c exhibits three different phases (I, II and III) depending on α . Likewise, the system exhibits a second (first) order transition from phase I to phase II when P c h g >0 ( P c h g = 0 ). Nevertheless, the transition from phase II to phase III is of first (second) order when P c h g > 0 ( P c h g = 0 ). In addition, the result analysis shows that the distribution of accidents according to the intersection sites is not equiprobable. Furthermore, when the traffic arriving strength is not very high, the green light length of one road should be increased to restrain P a c and enhance the road safety. Finally, the model results illustrated that the traffic at the intersection is more dangerous adopting asymmetric lane-changing rules than symmetric ones.