The functionality and the efficiency of smart cities are highly dependent on the effectiveness of critical infrastructures, such as electrical grids, gas networks, water supply, telecommunication, and transportations. Those lifeline systems are tightly interconnected one to another due to geographical, physical, logical and cyber interdependencies, causing the well-known ‘domino effect’. Besides interdependency, age affects the performance of critical infrastructures and their robustness against threats posed by natural hazards, environmental conditions, and terrorism. This paper presents a mathematical approach for the maintenance problem applied to several and interconnected infrastructures in a smart ecosystem. While in literature, the problem is faced from each single infrastructure point of view, in this paper the researchers analyse all the infrastructures in the same district, considering also their interdependencies. The development of a management system for maintenance depends primarily on the availability of information on the actual state and on the reliable risk assessment methods for forecast possible future behaviours of the considered infrastructures. CISIApro is an agent-based simulator designed to assess the short-term effects of failures and cyber-attacks. Therefore, CISIApro is able to evaluate the current risk of interconnected infrastructures and to forecast the future state of every single system. The CISIApro outcome is exploited by a decision support system for the maintenance scheduling problem which suggests to operators the ranking to repair each infrastructure based on a priority list. Due to its complexity, maintenance scheduling is a multi-objective optimization problem, where several aspects must be considered. The solving method is a heuristic approach exploiting the Pareto optimum called ELECTRE II. The proposed tool provides an ordered list of maintenance works for the different infrastructures, that are suggested to different operators. The case study is a small district where several infrastructures coexist demonstrating the feasibility and effectiveness of the proposed approach.

The paper has been accepted at the conference.