Introduction to CPFD Software and the Barracuda Virtual Reactor
Fluid-Particle flows
Examples
Fluidized Beds
Ore Reactors
Cyclones
Risers
Deep Beds
What is different than single phase flows?
Particles are NOT fluids:
Particles are discrete entities (cannot be subdivided like a fluid)
Particles have a size distribution
Particle cannot completely fill a space
Particles occupy a physical volume (and displace fluid)
Particles CAN support a shear stress (while fluids CANNOT support a shear stress)
Other considerations:
Coupling between particles and fluids
Wall treatments
Boundary treatment
Thermal, chemistry…
Modeling approaches
To accurately simulate fluid-particle flows, one must model the effects of:
Drag and coupling
Dilute flows (<1% by volume)
Dense flows (up to close pack)
Many particle sizes or size distribution
Multiple types of particles (size, density, composition)
Particle interactions (walls, other particles)
Heat transfer
Chemical reactions
Gas-phase (homogeneous)
Gas + particles (heterogeneous)
Changing particle composition
There are different approaches to modeling particle flows
The CPFD? modeling approach
Overview
What about the large number of particles?
Lesser number of computational particles
The particle field is resolved by using a reasonable number of computational particles
Each computational particle represents one or more actual particle(s) with identical physical properties
The physics are computed on the individual particle (e.g. drag based on size, chemistry, etc.)
All changes experienced by the computational particle are applied to all actual particles represented by that computational particle (proper fluid displacement)
Many CPFD calculations utilize between 500,000 and 5,000,000 computational particles
What about particle contact and collision?
Modeled, rather than directly computed
Collision detection can be prohibitive with millions of computational particles
Rather than computing which particle a given computational particle will impact, the CPFD method is more concerned with the question “is a collision likely to occur?”
The collisions are then subjected to various models
Enduring contact at close-pack handled via a non-linear stress tensor
BGK-type collisional damping
What about inter-phase coupling?
Inter-phase interpolation operators, and tight, bi-directional coupling
Different particles experience different motion, even though both are in the same cell
All sub-grid particle motion is coupled back to the fluid phase momentum equation