| |
By Madhusuden Agrawal, André Bakker, and Mike
Prinkey, Fluent Inc.
View the pdf of this article
Particle tracking models in most commercial CFD software, such as the
discrete phase model (DPM) in FLUENT, assume that particles are point
masses that do not interact. Large particles immersed in the fluid flow
cannot be modeled using this type of approach. The modeling of large (macroscopic)
particles requires special treatment to take into account effects such
as the blockage of fluid volume, the proper evolution of the drag force
and torque experienced by the particles, particle- particle as well as
particle-wall collisions, and friction dynamics.
Macroscopic particles continuously injected through a filter element
To account for these effects, a macroscopic particle model (MPM) has
been developed for FLUENT 6 using user-defined functions (UDFs) and a
customized graphical user interface (GUI). In the MPM approach, particles
are treated in a Lagrangian frame of reference. Each particle is assumed
to span several computational cells. A solid body velocity that describes
the particle motion (translational and rotational) is patched in these
cells. The volume fraction of the particle is also taken into account.
By patching the rigid body motion of the particle, momentum is effectively
added to the fluid. The integral of the momentum change, linear as well
angular, gives the drag force and torque for each particle. These are
used to compute the new positions and velocities of the particles at the
next time-step. Additional forces, such as body forces, can also be included
in the model. To detect a particle- wall collision, the model identifies
the boundary faces (wall surfaces) the particle intersected during the
previous time-step, if any. If a collision with a stationary wall is detected,
the incoming particle velocity is projected onto the normal and tangential
components of the reflected particle velocity, applying the coefficient
of restitution and friction factor, as appropriate. In the same way, the
model detects particle-particle collisions, and applies the principle
of conservation of momentum to obtain the final velocities of both particles.
The particle-wall collision algorithm also takes into account rotating
or moving walls, so it can be used with both the sliding and deforming
mesh models in FLUENT. The MPM UDF has been parallelized and works well
with the FLUENT parallel solver.
A spinning heavy ball dropped in water disrupts the liquid surface
Balls of different mass continuously injected into a rotating paddle-type
mixer
Pathlines show the air flow (including recirculation) generated by the
cue ball as it rolls towards the rack on a pool table (top) and just after
it strikes and disperses the balls (bottom)
The customized GUI is used to define all
user inputs for the macroscopic particle model.
The initial particle properties (positions,
velocities, mass, radius) for each particle stream
can be entered in the panel or read from a
formatted ASCII file. MPM postprocessing
tools have been coupled with FLUENT’s DPM
visualization tools, which allow particles to be
displayed as shaded spheres with a defined
radius. Transient particle data can also be saved
in a Fieldview data file format.
The macroscopic particle model has many
industrial applications, especially in the pharmaceutical,
chemical, material handling, and
sports industries. Several validations have been
performed. The tests have shown that a large
number of particles (up to 1000) can easily
be handled without the need for excessive computation
time.
|
|
|
