During the first part of the program, the objective was to study the different mechanisms involved in the propagation of hot gases and smoke from a burning room to adjacent rooms, via all sorts of communicating elements (open doors, leakages around firebreak doors, air vents, piping, etc.) and under the action of ventilation.
Separate effect and global tests
To achieve this, two types of fire tests were carried out: tests specific to the study of a particular physical phenomenon (called "separate effects"), such as the flow through a doorway, and so-called global tests which consist in fires in a configuration that is representative of nuclear scenarios. The three first campaigns were dedicated to the study of separate effects: PRISME Source, for the characterisation of fire sources in open atmosphere and in vitiated environment, PRISME Door, to study the spread of heat and smoke through open doors, and PRISME Leak, to study the propagation of heat and smoke through passages (openings, leakage through a firebreak door, ducting crossing the fire compartment). Finally, the last campaign, PRISME Integral, consisted in carrying out tests according to different accident scenarios and real fire sources (cables and electrical cabinet).
The fire source characterisation tests were carried out in an open environment under the SATURN, a calorimeter used for the study of fire sources in an open atmosphere which is located on the IRSN Cadarache site (Bouches du Rhône). They consisted in burning pans of hydrocarbons, electrical cabinets or cables while especially determining the power of the fire.
Photo of the SATURN hood © IRSN
The other tests were carried on the large-scale facility, DIVA, also located on the Cadarache site, which is composed of three rooms and a hallway which are confined and ventilated like nuclear rooms. The fires are studied there during their complete development until their extinction due to lack of combustible material or depletion of oxidant.
Five experimental campaigns were carried out between 2006 and mid-2011 including over 35 large-scale experiments. The PRISME Source campaign used a single ventilated room. It reproduced the effect of under-oxygenation, caused by the vitiation of atmosphere, on the way in which the power of a hydrocarbon fire evolves depending on the pan size, the type of fuel and the flow rate of the ventilation. The following experimental campaigns covered the flow of the smoke and hot gases through open doors between two or three rooms, through air vents or leaks in firebreak doors. Finally the so-called "global" tests were carried out, in which all the rooms in the experimental installation were used, with complex solid sources and the activation of fire extinguishing systems (especially sprinklers) or fire dampers on the ventilation.
Effect of ventilation on the power of the fire
The Prisme program made it possible to better understand the effects of ventilation on the power of a fire which breaks out in a confined and ventilated room, and especially on the duration of the fire. Depending on the rate of renewal of the ventilation, the fire can extinguish quickly because of the decrease of the concentration of oxygen in the burning room. But the PRISME tests have shown that a balance can establish itself between the air coming from the ventilation ducts and the power of the source: in this case the combustible material burns more slowly until fire extinction. For example, the same fire source can last 2.5 times longer than in the open air in a room ventilated using an hourly renewal rate of 4.7. The major contribution of the PRISME tests is a better understanding of the effect of under-oxygenation on the evolution of a fire's power.
Correlative and analytical pyrolysis models were developed and validated. These models make it possible to improve the estimations of changes over time of a fire in a confined and ventilated environment using input data obtained from an open atmosphere.
The PRISME program also made it possible to quantify the effect on the spread of smoke of "mixed" convection, which combines the forced convection created by ventilation, and natural convection caused by the high temperature of the smokes. The mechanical ventilation of the burning room can significantly modify the flow of gases that develops naturally through an open door between two rooms. Depending on the ventilation network, mechanical ventilation contributes to unbalancing the flows entering and leaving the burning room, and to changing the position of the neutral zone (height at which the speed of flows is null) at the door.
New models for software
The volume of data collected during this program made it possible to assess the capacity of software to simulate different fire scenarios. The new models, especially for pyrolysis, were built into the software from the different partners (the Sylvia and Isis software for the IRSN) and validated thanks to the experimental data obtained during this program.
Further, inter-software comparisons were organised by the IRSN in the context of a group attached to the PRISME program: the different partners compared the results of fire simulations with the experimental data. This group especially tested and analysed several "metrics" (i.e. mathematical norm) indicators used during the validation process to objectively assess the differences between the experimental data and the simulation results. Amongst those recommended by the standards, two metrics were revealed to be complementary to assess the capacity of software to simulate a fire. The first metric assesses the relative differences between the numerical and experimental results for local values, as extrema. The second metric is the standardised Euclidean distance which is used to assess the difference between the numerical and experimental results on the duration of the fire. This metric behaves as a global error measurement.
A sensitivity test was also carried out using six different calculation software programs, including the IRSN's Sylvia. The influence of six input parameters (power of the fire, radiative fraction of the flame, thermal properties of the walls, etc.) was tested to calculate nine scales of interest (the temperature of the gases and walls, the concentration of oxygen, the heat fluxes, etc.). For all the software, the results show that the most influential input parameter is the power of the fire, which demonstrates the need to continue the efforts to improve its modelling.
Simplified diagram of the DIVA installation © IRSN