International audience; Fire hazard analyses and probabilistic fire safety analyses have demonstrated that fires may cause significant damages in nuclear power plants (NPPs). Fire modelling is nowadays applied by licensees or technical safety organisations to assess fire consequences in NPPs. Thereby, one important aspect is the availability of verified and validated fire models for such fire scenarios.Several members of the Organization for Economic Co-operation and Development (OECD) Nuclear Energy Agency (NEA) expressed their interest in participating in a joint international research project on the topic of fire events to be carried out under the auspices of the NEA. The PRISME (French acronym for “Fire Propagation in Elementary Multi-Room Scenarios”) Project was realized from 2006 to 2010, by the Institut de Radioprotection et de Sûreté Nucléaire (IRSN, France) in its facilities specially designed for large-scale fire tests in confined environments. In the continuity of the PRISME project, PRISME 2 was launched in July 2011 ending in 2016. The main experimental results of these two projects have been summarized in OECD/NEA reports [1][2]. In parallel to the experimental campaigns, PRISME partners evaluated the capabilities of various fire simulation codes for modelling fire scenarios based on the PRISME results. Both PRISME 1 and PRISME 2 Projects highlighted the strong interaction between the fire dynamics and the mechanical ventilation. Indeed, the analysis of the tests greatly contributed to enhance the knowledge of under-ventilated fires including realistic and complex fires. An improvement in the validation process of different fire models was also noticed during these two experimental programs. From these experimental findings and modelling considerations, some grey zones have been highlighted and allowed to define the outlines of the PRISME 3 Project. Various recommendations have been provided for addressing some further phenomena not studied in the past Projects. These phenomena are smoke stratification and spread, fire propagation between electrical cabinets, and electrical cable tray fires in confined and ventilated conditions for new configurations. The ongoing PRISME 3 Project aims at addressing the above mentioned three phenomena and at providing answers to various issues of interest for nuclear safety. A total of eight countries have joined the PRISME 3 Project: Belgium (Bel V and Tractebel-ENGIE), Finland (Technical Research Centre VTT), France (IRSN as Operating Agent and Électricité de France – EDF), Germany (Gesellschaft für Anlagen- und Reaktorsicherheit – GRS), Japan (Nuclear Regulation Authority – NRA and Central Research Institute of Electric Power Industry – CRIEPI), Korea (Korea Institute for Nuclear Safety – KINS and Korea Atomic Energy Research Institute – KAERI), United Kingdom (Office for Nuclear Regulation, – ONR) and the United States of America (United States Nuclear Regulatory Commission – U.S. NRC).The objective of the first campaign, named S3 for Smoke Stratification and Spread, is to study new configurations of interest for smoke propagation in a mechanically ventilated multi-room facility with simple fire sources. This choice is relevant for a complete validation of fire models on smoke propagation. The first topic of interest is to combine vertical and horizontal smoke propagation coupled with a mechanical ventilation system. This experimental configuration allows highlighting multiple interaction mechanisms of propagation during the fire scenario. The second topic of interest concerns the issue of multiple fire sources, simulating for example a seismically induced fire incident and its consequences on smoke propagation. The fire scenario involves two fire sources ignited simultaneously and located in two adjacent rooms or in two rooms separated by another one. The distance between fires is a key parameter determining different combustion regimes with or without interaction. The third topic of interest is smoke propagation induced by an elevated fire source. This configuration leads to a complex situation for the fire dynamics, since it evolves in a hot and vitiated environment. For the S3 campaign, six fire tests have been defined in the multi-compartments facility of IRSN, named DIVA, which is composed of a mechanical ventilation system. For this campaign, three or four rooms, of volume 120 m3 and 170 m3, have been implemented.The second campaign, named ECFS for Electrical Cabinet Fire Spread, aims to better understand the fire spread from an open-door cabinet to other adjacent or opposite cabinets, connected via cable trays. Four tests in the IRSN DIVA facility have been defined including the two configurations of interest: the adjacent one and the opposite one. For each configuration two tests involving HFR (halogenated flame retardant) or HFFR (halogen free flame retardant) insulated cables will be conducted. Two cable trays have been positioned above the cabinets and inside a false floor to diversify the potential paths of the fire propagation. Furthermore, for the adjacent configuration, the fire propagation through the walls and an air gap potentially separating the cabinets is also considered. In addition to these confined fire tests, four additional tests have been specified in open atmosphere. These tests are needed for characterizing the fire source or the fire spread from one electrical cabinet to another one in a reference configuration. The comparison of the two configurations will highlight the effect of the confinement on the fire spread.The purpose of the third campaign, named CFP for Cable Fire Propagation, is twofold: in a first step, the effect of the compartment geometry will be studied by conducting three cable tray fires in a corridor. The fire dynamics on a long cable tray will then be compared to those obtained in previous PRISME Project for shorter cable tray configurations. In a last step, additional scenarios involving the effects of under-ventilated conditions and of the cable tray configuration on cable fires will be investigated. In addition, an assessment of cable fire models used within simple or complex fire numerical tools will be conducted for these specific configurations. The cable type and the air renewal rate of the compartment will be considered. Consequences of such fire scenarios on the facility will be investigated through time sequences of gas pressure, temperature and concentrations inside the facility and in the ventilation network. The campaign is composed of six tests in the DIVA facility and two tests under the SATURNE calorimeter of IRSN which is composed of an extraction hood in an open domain on 20,000 m3. The main advances of PRISME 3 will make it possible, in a first step, to increase the predictive ability of models on smoke propagation problems for substantially complex situations. With regard to electric fires, especially electric cable fires, PRISME projects will also provide a fairly complete database and can therefore be used for model improvement. All the contributions of the project will allow the community to position itself on the scenarios of interest to study in the years to come.