An Open Toolbox for Integrated Optimization of Public Transport

Abstract

Public transport networks are a vital factor in the development of urban and peri-urban mobility. Their design is the main contributor to their possible success. Determining line planning, timetabling and vehicle scheduling is a complex set of tasks, which are interconnected in many ways. The development is typically tackled by hand incorporating multiple objectives in a time-consuming way. Within each task, decisions require multi-objective optimisation. These planning steps are usually carried out sequentially to simplify matters. However, research has shown that integrated approaches lead to superior solutions.On the other hand, integrated approaches increase computational complexity. We report on a toolbox which can seamlessly integrate the three tasks (line planning, timetabling, vehicle scheduling). A general framework formalises all relevant information and ensures the exchange and interoperability between intermediate solutions of the three separated problems. Then, we formulate the three problems to iterate between the planning steps within an approach inspired by the eigenmodel structure [1]. We refine existing approaches to minimise costs and travel time jointly. Thanks to the extensions and framework we may shrink the size of the problems without loss of optimality. The eigenmodel can significantly reduce complexity, while the extended formulations for line planning, timetabling and vehicle scheduling allows consideration of all relevant limitations.We report a study in the region of Bern, Switzerland. The solution computed for a network of regional buses significantly decreases the demanded effort of the planners. The added value lies in the improved quality of the solution and the shorter response time, enough to support interactive optimisation.