A new methodology for the real-time limited-stop bus service design problem

Providing limited-stop bus services can improve the efficiency of bus systems. This paper proposes a new two-stage strategy for providing real-time limited-stop bus services for a corridor and the corresponding model is developed. In the first (tactical planning) stage, given the maximum number of different limited-stop services, an operator determines a set of limited-stop services based on historical bus travel times and passenger arrival rates. In the second (operational) stage, an operator selects one service from the set of limited-stop services obtained in the first stage for each limited-stop vehicle based on (short-term) predictive travel times and passenger arrival rates. Prediction errors are considered in the second stage. An enhanced artificial bee colony algorithm is developed to solve the first-stage model and the Monte Carlo Simulation method is adopted to solve the second-stage model. Numerical results are presented to illustrate the effectiveness and efficiency of the strategy and the effect of prediction errors.