A microscopic lane changing process model for multilane traffic

Abstract

In previous simulations lane-changing behavior is usually assumed as an instantaneous action. However, in real traffic, lane changing is a continuing process which can seriously affect the following cars. In this paper, a microscopic lane-changing process (LCP) model is clearly described. A new idea of simplifying the lane-changing process to the car-following framework is presented by controlling fictitious cars. To verify the model, the results of flow, lane-changing frequency, and single-car velocity are extracted from experimental observations and are compared with corresponding simulation. It is found that the LCP model agrees well with actual traffic flow and lane-changing behaviors may induce a 12%–18% reduction of traffic flow. The results also reflect that most of the drivers on the two roads in a city are conservative but not aggressive to change lanes. Investigation of lane-changing frequency shows that the largest lane-changing frequency occurs at a medium density range from 15vehskm−1lane−1 to 35vehskm−1lane−1. It also implies that the lane-changing process might strengthen velocity variation at medium density and weaken velocity variation at high density. It is hoped that the idea of this study may be helpful to promote the modeling and simulation study of traffic flow.