Being a particle physics laboratory, the European Organization for Nuclear Research (CERN) plans, constructs, and maintains installations emitting ionizing radiation during operation. Activation of present material is a consequence. Hence, fire scenarios for certain CERN installations must take into account the presence of radioactive material. Releases of gaseous, liquid, or solid combustion products, e.g., attached to aerosols, are taken so far into account by a worst case approach. Scenarios taking place in underground installations assume hence a smoke transport coefficient of 100% of release toward the surface level, independent of the local geometry. For a radioactive inventory identified in a certain fire load, this results in a conservative release.

To overcome this conservative worst case approach, a computational fluid dynamics model based on FM Global’s fireFoam 2.2.x is proposed. Its Lagrangian library was modified in order to provide aerosol release and deposition information based on more detailed interaction data between Lagrangian particles and their surrounding geometry. Results are shown for a CERN-typical large-scale experimental cavern placed 100 m below surface level. A simple diffusion burner is modeled inside the cavern to create a thermal plume emerging from a 1.5-MW fire over 14 min. Lagrangian particles are used to model aerosols with an aerodynamic diameter of 1, 10, and 100 μm, injected into the emerging thermal plume. Results for particle release and deposition vary according to aerodynamic diameter. In the present case, maximums of ~32% and 39% are found for 1- and 10-μm particles, respectively, being released to the surface level.