Discrete Phase Models¶
A discrete phase model should be set when the flow of a discrete phase (particles) with a continuum is modelled. This chapter shows how a simulation with a discrete phase model should be set up on the SimScale platform.
Creation of a discrete phase model¶
Definition of a discrete phase model is mandatory when performing a Discrete phase model simulation. In the tree, navigate to “Advanced Concepts” and add a discrete phase model. A DPM four-way coupling is supported that takes into account full continuum-particle and particle-particle interactions into account.
Apart from general settings, such as maximal Courant number, integration and interpolation schemes, particle density and packing, the model includes several sub-models. Each model is listed and described in the following:
- Particle forces
Forces acting on particles are categorized in five groups:
- Drag: several drag models are provided. Furthermore, it is possible to define drag for non-spherical particles, or no drag at all.
- Lift: two well-known Saffman-Mei and Tomiyama lift models are available.
- Pressure gradient
- Virtual mass
- Injection models
Allows different methods of introducing the discrete phase to the computational domain:
Cone nozzle injection: allows definition of an cone nozzle as an injector. Injection position, direction could be set.
Injection from boundary: this type allows the discrete phase to be injection from a boundary, e.g. an inlet. Injection velocity and the name of the boundary, i.e. Patch name should be provided by the user.
Placement on boundary: this type allows particles to be placed on a boundary. Injection velocity is automatically calculated using the velocity field at the boundary. The name of the boundary, i.e. Patch name should be provided by the user.
Upload initial positions: this type allows a CSV file to be uploaded which specifies the initial distribution of particles in the domain. An example of the input CSV file is shown here:
1.60 2.53 0.05 0.00 1.76 0.01 2.32 5.20 0.18 0.70 0.25 0.73
Each line corresponds to one particle. The first column specifies position of particle in X-coordinate, the second in Y, and the third in Z-coordinate.
Upload injector specifications: this type a allows uploading a CSV file that defines several injection points. An example of the input CSV file is shown here:
1.04 1.31 3.01 6.54 2.11 1.01 0.50 1.12 0.73 2.12 1.20 0.18 2.53 3.05 2.32 7.66 8.44 0.21
Each line corresponds to one injection source. Each injection source is defined by its position (the first three columns), injection velocity (columns 4, 5, and 6), particle diameter, particle density, and injector mass flow rate.
Additionally, each model allows particle size distribution, mass flow rate, injection interval, and several other particle properties to be defined at the time of injection.
- Dispersion model
Particle dispersion model could be defined. The following choices are available:
- Gradient dispersion: dispersion takes place in the opposite direction of turbulent kinetic energy gradient with a Gaussian distribution.
- None: no dispersion
- Stochastic dispersion: dispersion takes place in random directions with a Gaussian distribution.
- Patch interaction model
Defines interaction of domain boundaries with the discrete phase. The following models are available:
Local: interaction with each boundary is defined separately. Two choices are available:
- Escape: allows particles to leave the domain from this boundary (e.g. outflow). A comma-separated list of faces should be provided by the user.
- Stick: Freezes each particle that touches this boundary. A comma-separated list of faces should be provided by the user.
All other boundaries which are not of type Escape or Stick will be wall with rebound factor \(1.0\).
None: no specific interaction is defined.
Reound: all boundaries are wall with the user-supplied rebound factor.
Wall: define all boundaries as:
- Rebound: elasticity and resitution coefficients should be provided
- Escape: allows particles leaving the domain
- Stick: freezes particles
- Packing model
Defines inter-particle stress models to guarantee stability of dense particle flows. The following models are available:
- Explicit: based on the current velocity field, this model limits velocity of particles that are moving towards regions of close pack.
- Implicit: in this model, particle volume fraction is solved over time on the Eulerian mesh and applied to the lagragian particle to contorl packing.
- None: no packing model is applied.
- Damping model
Damping is requried to enhance the stability of the simulation when collision happens. The following models are available:
- Relaxation: relaxes velocity of particles based on the the local average over a time-scale. Several models are available to compute the time-scale.
- None: no damping takes place.
- Isotropic model
allows modelling of particle scattering as a result of collisions. The following models are available:
- Stochastic: particle velocities are rendomized using a Gaussian distribution. Momentum and energy are conserved in this process.
- None: no scattering takes place.
In this example, a discrete phase model simulation for a flow calibration rig is shown. Please refer to the project library to find more example projects with complete discrete phase model setup.