Improving resilience in Critical Infrastructures through learning from past events

Modern societies are increasingly dependent on the proper functioning of Critical Infrastructures (CIs). CIs produce and distribute essential goods or services, as for power transmission systems, water treatment and distribution infrastructures, transportation systems, communication networks, nuclear power plants, and information technologies. Being resilient, where resilience denotes the capacity of a system to recover from challenges or disruptive events, becomes a key property for CIs, which are constantly exposed to threats that can undermine safety, security, and business continuity. Nowadays, a variety of approaches exists in the context of CIs’ resilience research. This dissertation starts with a systematic review based on PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) on the approaches that have a complete qualitative dimension, or that can be used as entry points for semi-quantitative analyses. The review identifies four principal dimensions of resilience referred to CIs (i.e., techno-centric, organizational, community, and urban) and discusses the related qualitative or semi-quantitative methods. The scope of the thesis emphasizes the organizational dimension, as a socio-technical construct. Accordingly, the following research question has been posed: how can learning improve resilience in an organization? Firstly, the benefits of learning in a particular CI, i.e. the supply chain in reverse logistics related to the small arms utilized by Italian Armed Forces, have been studied. Following the theory of Learning From Incidents, the theoretical model helped to elaborate a centralized information management system for the Supply Chain Management of small arms within a Business Intelligence (BI) framework, which can be the basis for an effective decision-making process, capable of increasing the systemic resilience of the supply chain itself. Secondly, the research question has been extended to another extremely topical context, i.e. the Emergency Management (EM), exploring the crisis induced learning where single-loop and double-loop learning cycles can be established regarding the behavioral perspective. Specifically, the former refers to the correction of practices within organizational plans without changing core beliefs and fundamental rules of the organization, while the latter aims at resolving incompatible organizational behavior by restructuring the norms themselves together with the associated practices or assumptions. Consequently, with the aim of ensuring high EM systems resilience, and effective single-loop and double-loop crisis induced learning at organizational level, the study examined learning opportunities that emerge through the exploration of adaptive practices necessary to face the complexity of a socio-technical work domain as the EM of Covid-19 outbreaks on Oil & Gas platforms. Both qualitative and quantitative approaches have been adopted to analyze the resilience of this specific socio-technical system. On this consciousness, with the intention to explore systems theoretic possibilities to model the EM system, the Functional Resonance Analysis Method (FRAM) has been proposed as a qualitative method for developing a systematic understanding of adaptive practices, modelling planning and resilient behaviors and ultimately supporting crisis induced learning. After the FRAM analysis, the same EM system has also been studied adopting a Bayesian Network (BN) to quantify resilience potentials of an EM procedure resulting from the adaptive practices and lessons learned by an EM organization. While the study of CIs is still an open and challenging topic, this dissertation provides methodologies and running examples on how systemic approaches may support data-driven learning to ultimately improve organizational resilience. These results, possibly extended with future research drivers, are expected to support decision-makers in their tactical and operational endeavors.