Motivation
There two kind of models applicable for aerosol modeling in nuclear industry.
A) Eulerian family, with two subclasses:
- Drift-flux / mixture model: dispersed phase is dilute and relative velocity relaxes quickly (the “equilibrium slip” case)
- Eulerian aerosol dynamics via Population Balance Models (PBM): can be considered a more complete extension of the models above but allows accounting for aerosol transport when the distribution evolves (coagulation, condensation, etc.). This one includes two type of models: (i) sectional (size-bin) models (robust, and used in nuclear severe-accident tools such as MELCOR, MAEROS, etc.) and (ii) moment methods (QMOM/DQMOM, etc.)
B) Lagrangian family, which track representative particles through the (Eulerian) flow field. These models actually solve for particle motion without approximations of an effective continuum particle field (such as the ones assumed for the models above). These models are more natural for very dilute aerosols, complex deposition rules, and when we care about distribution of outcomes (where particles land, residence time PDFs, etc.). For more dense aerosols, the concept of parcels / super-particles should be introduced.
The nuclear industry has long experimented with models of type A and a drift flux model is already under development in OpenPronghorn under issue #51 However, a more complete Lagrangian model is needed.
Design
The Lagrangian particle tracking model should:
- Model the behavior of individual particles in the MOOSE mesh, including forces of gravity/buoyancy, drag, lift, Brownian diffusion, thermophoresis, diffusiophoresis, and turbophoresis
- Leverage GPU acceleration via the torch implementation in MOOSE
Impact
This model is needed to simulate accidents involving aerosol dispersion with significantly higher accuracy than other methods available in the nuclear industry.
Motivation
There two kind of models applicable for aerosol modeling in nuclear industry.
A) Eulerian family, with two subclasses:
B) Lagrangian family, which track representative particles through the (Eulerian) flow field. These models actually solve for particle motion without approximations of an effective continuum particle field (such as the ones assumed for the models above). These models are more natural for very dilute aerosols, complex deposition rules, and when we care about distribution of outcomes (where particles land, residence time PDFs, etc.). For more dense aerosols, the concept of parcels / super-particles should be introduced.
The nuclear industry has long experimented with models of type A and a drift flux model is already under development in OpenPronghorn under issue #51 However, a more complete Lagrangian model is needed.
Design
The Lagrangian particle tracking model should:
Impact
This model is needed to simulate accidents involving aerosol dispersion with significantly higher accuracy than other methods available in the nuclear industry.