Game-Theoretic Approaches for Service Contracting in Rail Infrastructure Maintenance

Project members: Z. Su (Zhou), prof.dr.ir. B. De Schutter (Bart), S. Baldi
Keywords:Transportation and infrastructure, Railway networks, reverse Stackelberg Game
Sponsored by:NWO, ProRail

Since the privatization of Dutch railways and the institutional separation between infrastructure management and train operations, performance of maintenance on rail infrastructure is jointly determined by multiple parties, namely, ProRail as the infrastructure manager, NS as the train operator, and their respective service contractors. This might result in misalignment of objectives among stakeholders, who make their decisions independently, and therefore the optimal performance of service activities at the whole system level cannot be guaranteed. New approaches for service contracting are required in order to allow the railway manager, ProRail, to induce its contractors and transportation operators to align their own objectives with ProRail's objective regarding adequate, efficient and effective rail infrastructure maintenance.

A single-leader-multi-follower reverse Stackelberg game setting will be applied to this problem. The infrastructure manager will be modeled as the leader, who is at the position to design the game by setting the rules of information exchange and by establishing the structure of the incentives to which other players must respond, thus trying to induce the followers to contribute to the optimization of the leader's performance, while maximizing their own utilities. The key issue is to develop methods to design such an incentive, which can be expressed in various policies (rewards, penalties, conditions, etc.).

Efficient numerical solutions to reverse Stackelberg game will also be investigated in this project, especially for games with a dynamic setting or incomplete information. Important properties of the equilibria, like stability, optimality will also be addressed, as well as the robustness of the solution. The game-theoretic methodology for optimal contracting developed in this project can further be extended to other applications that fit in a hierarchical structure (e.g. roadway maintenance, network charging, etc.).

© Copyright Delft Center for Systems and Control, Delft University of Technology, 2017.