Predicting the behavior of fire in confined enclosures with and without mitigation using water-based firefighting systems is a topic of great interest for both civil and military fields. Although notable advances have been made in this area from prescriptive or analytic approaches, these conventional approaches fail in reproducing fire behavior in polydisperse amorphous massively multi-compartmented structures (e.g. naval vessels, high-rise buildings, warehouses, or nuclear plants). The reasons are manifold: the lower relevance of the physical models used the difficulty in accounting for the role of some influential factors on fire behavior, and/or the requirement of prohibitive amounts of memory and computational resources. The present project aims at modeling and achieving super-real time simulations (faster than real time) of fire propagation and its limitation using water systems in such complex structures. The approach we propose to describe large-scale fire behavior is based on the purely stochastic small-world network model developed by Watts et Strogatz in the end of the nineties to account for both short-range and long-range connections, between adjacent and remote compartments. To finely describe the dynamics of fire, the original network model will be extended by incorporating a weighting procedure on sites, the weights being determined from the characteristic times of heat transfer from compartment to compartment: wall conduction, hot smoke flow through corridors, openings and ventilation pipes, cable trays. This determination will be made from specific lab-scale experiments -by exposing both homogeneous and composite wall panels to incident radiant heat flux representative of fire conditions- and from numerical simulations of fire behavior and its interaction with the water spray using a deterministic macroscopic two-phase model. Complementary experiments will be conducted in a dedicated fire box in order to validate the macroscopic model, but also to collect basic data on the level of thermal stress the structure may undergo. Concept validation consists of two steps. First, we will use results obtained from experiments conducted in the submarine part of the USS SHADWELL. These experiments involved fire propagation in a few compartments and the operation of a water spray system. Second, the partners will have to define an amorphous polydisperse network composed of a large number of compartments. A sensitivity analysis will be carried out in order to identify and classify the most influential parameters of the fire propagation model, and to study the response of the system to the variations of these parameters. The capability of the model to elaborate fire risk mapping and to evaluate network properties will be demonstrated by means of a statistical study. Network properties include percolation threshold, super-nodes (highly-connected nodes) and critical channels along which the fire will propagate preferentially. Evaluating network properties should help in developing better strategies to reduce fire consequences. The present project is a preliminary phase of a long-term project, intended to provide designers and managers with a decision making tool allowing them to reduce the vulnerability of buildings and infrastructure and to allocate fire-fighting resources to maximize public and firefighter safety, reduce environmental impacts, and lower fire-fighting costs.