MODITIC - report on source term modelling

The European Defence Agency (EDA) project B-1097-ESM4-GP “MOdelling the DIspersion of Toxic Industrial Chemicals in urban environments” (MODITIC) has in the period from 2012 to 2016 studied the release and transport of neutral and non-neutral chemicals in complex urban environments, in order to enhance the understanding of the dominating physical processes involved, and to support improvements in modelling techniques. WP3000 concerns agent characterization and source modelling, this includes modelling the flashing, expansion, evaporation and rainout processes. Four studies have been conducted within WP3000 by four research institutes: • DGA tested Reynolds-averaged Navier-Stokes models for simulating the release of a pure gas or two-phase ammonia jet impinging an obstacle. The small time step required makes this method inappropriate for large geometrical domains. In addition, unphysical results for some parameters were obtained. • FFI used large eddy simulations approaches for the same two cases. The demands for the time step size and the grid resolution is even stricter in this case, precluding simulation of the entire experimental duration, as well as for large domains. Furthermore, in order to calculate the flashing of the liquid jet and the evaporation of the liquid droplets, a two-way coupling between the gas phase solver and the Lagrangian liquid droplet solver is needed, but the model used does not include such a two-way coupling at the present time. • FOI tested an approach where the conditions in the jet are calculated by an analytical approach and given as input to a subsequent computational fluid dynamics (CFD) simulation of the dispersion of gas and droplets. They validated the approach with field tests of the release and dispersion of liquified ammonia, and the results are encouraging. • Finally, INERIS used a large eddy simulations approach where the inlet conditions is specified from real (experimental) energy spectrum profiles. The vapour source term is specified according to an analytical calculation. The simulations are compared with field experiments of the release of ammonia, and the results are promising regarding the complexity for describing both near and far field dispersion. None of the institutes involved in MODITIC have a CFD capability that can fully model all the physical processes associated with a two-phase release of ammonia. The CFD methodologies produced somewhat unphysical results. In addition, the computer resources needed for a full CFD simulation makes such models inappropriate for large domains. On the other hand, a decoupled approach, where the rapid processes (flashing and expansion of the jet) are calculated with an empirical or analytical approach, and the slow processes (gas dispersion and air entrainment) are simulated with CFD, yields encouraging results. It is recommended to use the decoupled approach when modelling such demanding processes.