Optimization Model And Algorithm For Bus Lane Network Design In Multi-modal Urban Traffic System

In order to connect all the scattered bus lanes as a network, building bus lanes is becoming more and more popular in big cities. But, the non-bus traffic flow will take more time travelling along the road, when the bus lanes were set up without consideration. Thus, the bus lane network should be organized systematically. Nowadays, many problems of bus lanes have been studied, such as bus lane setting forms and the lane setting conditions of traffic flow et cetera. In this paper, based on multi-mode traffic flow assignment modal, the optimization of bus lane network design will be studied. Followings are the main works of this paper.1. Impedance function with bus lane is proposed, based on the multi-mode traffic, and combined with the characters of the traffic flow. Then, a numeric test has been taken for this function. And, it shows that the impedance of cars will increase after setting the bus lane and the attraction of travelling by buses will grow up.2. Based on the introduced impedance function, the traffic flow is assigned to the urban road network, which contains bus lanes. The results show that, when the urban traffic is in serious condition, there will more passengers taking buses rather than cars when the bus lanes were set. But these cannot be clearly seen, when there is little traffic. Beyond that, there will be more bus riders with decreased total cost, when the bus lanes were set correctly.3. A bi-level programming modal for the bus lane network optimization is proposed, by introducing a decision variable to the impedance function. The upper modal is to minimizing the total cost. And the VI (variational inequalities) modal for the traffic flow assignment, as a restriction of the upper one, is the lower. The two modals communicate via a variable of bus lane setting. Then, the method of exhaustion and algorithm of branch and bound are introduced. The first is able to find the global optimal solutions for the modal, but it cost too much time to compute, if the network is huge. Whereas the later algorithm will cost little, but the optimal solutions are local, which is determined by the initial solution.