The railway network design, line planning and capacity problem: An adaptive large neighborhood search metaheuristic

In this chapter, we propose a model for the Railway Network Design and Line Planning (RNDLP) problem, integrating the two classical first stages in the Railway Planning Process. The network design problem incorporates costs relative to the network construction, proposing a set of candidate lines. The line planning problem is in charge of determining optimal frequencies and consequently train operations, taking into account rolling stock, personnel and fleet acquisition costs. Both problems are intertwined because the line design influences the selection of frequencies and the corresponding fleet size. We consider the existence of an alternative transportation mode for each origin-destination pair in the network. In this way, the rapid railway mode competes against the alternative mode for a given certain demand, represented by a global origin-destination matrix. Passengers choose their transportation mode according to their own utility. Since the problem is computationally intractable for realistic size scenarios, we develop an Adaptive Large Neighborhood Search (ALNS) algorithm, which can handle the RNDLP problem. As illustration, the ALNS performance is demonstrated in an artificial instance using estimated data from literature.