Specification and modelling of the infrastructure systems
This task is dedicated to the definition and specification of the hypothetical case-study of interdependent infrastructure systems. Based on previous works that have designed hypothetical infrastructure systems for demonstration purposes, the following system of systems will be proposed:

• an electric power transmission network that is assumed to go from the power generating facility to high/medium voltage transformation stations, which are in turn supposed to deliver electricity to the built areas in the vicinity.
• a road network that is assumed to ensure the accessibility of the power generating facility and the built areas: the vulnerable elements will mostly be the bridges.
• a set of spatially extended built areas dependent on utility systems and transportation accessibility and industrial facilities located inside the Grenoble basin
• an industrial site located outside the Grenoble basin comprising several buildings, which is typical of power generating facilities.

This set of infrastructure systems will be arbitrarily around the Grenoble basin area, for which many data on the soil properties and topography are available.

Simulation of loss scenarios
A set of Monte Carlo simulations will be performed in order to quantify the performance losses of the systems. While such an approach has become quite common over the last few years, the main difference here consists in the use of meta-models obtained by physics-based simulations (WP2) for the generation of scenario-based ground-motion maps. We build the correlation matrix of the residuals of the ground-motion simulations, thus leaving the possibility to reproduce the correlation structure of the intra-event variability that is usually used with ground-motion prediction equations. Indirect impacts, such as utility losses or accessibility losses for the built areas, will then be quantified thanks to the sampled physical damage of the exposed components and to a subsequent systemic analysis (e.g. connectivity or serviceability analysis).

Integration of the probabilistic risk analysis chain into a decision support system
This task aims to exploit the previous Monte Carlo simulations in order to build a simplified Bayesian Network that will represent and quantify the logical relations between the various variables in the risk analysis chain (e.g. event magnitude, spatially-distributed ground-motion parameters at the vulnerable sites, damage states of the elements at risk, system performance indicators, etc.). This hybrid formulation, based on Monte Carlo simulations that are used to build approximate probabilistic relations between the variables, will prevent some of the computational issues that are inherent to Bayesian Networks. As a result, this simplified Bayesian Network will lay the ground for an integrated decision support system, in which risk managers and end-users will have the opportunity to set some assumptions and observer how the remaining variables are updated, through a Bayesian inference.