Spouted Bed

Eruption! The moment of spout breakthrough

Often used for drying and mixing, spouted beds are fluidized by high- speed gas jets to drive particle mixing. At right, MFIX-Exa is used to simulate a spouted bed experiment recently studied at NETL. The bed consists of 2.2M HDPE particles which are fluidized by single-celled jet. The jet velocity is ramped linearly from zero, exceeding 100m/s at breakthrough. The simulation was carried out on NETL's Joule2 HPC using 1600 CPUs. The animation was The animation was rendered with Blender.

Animation courtesy of Justin Weber & William D. Fullmer, NETL, Morgantown WV.

Shameless Self-Promotion

Soft-sphere molecular dynamics simulation with heavily designed initial conditions

A fun MFIX-Exa simulation that shows the evolution of 100k elastic particles in a fully periodic domain. Most of the domain is initially static, except for a collection of particles which are given a thermal (Maxwellian) speed in specific regions. Shock waves of momentum and concentration emanate from the thermal regions as the granular gas evolves to a homogeneous state. Animation shows particles colored by velocity magnitude rendered by Ovito and played back in reverse.

Animation courtesy William D. Fullmer, NETL, Morgantown WV.

Dilute Riser Flow

Falling clusters in the near-wall region

Circulating fluidized beds (CFBs) span a wide range of operating conditions throughout their subsystems. The riser section is typically characterized by dilute transport. Here, a 1.8 m section (~9M particles) of a pilot-scale CFB riser at PSRI is modeled, just 1/10th of the physical system's length. As high speed (15 m/s) gas flow drives the 650 micron HDPE particles vertically, denser than average (1%) regions can form at (or migrate to) the wall and fall against the mean flow. This phenomena can be observed on the right in the video while a streamer climbs slowly on the left. We note that the domain is too short, as the streamer has interacted with its periodic image. Future work will simulate the full 18 m length of the riser. Simulation by MFIX-Exa develop (git hash ef171c9d) using 24 GPUs on NETL's Joule2 HPC. The animation rendered with Blender.

Animation courtesy of Justin Weber & William D. Fullmer, NETL, Morgantown WV.

Boyce Single Bubble

Qualitative Benchmarking: Single bubble injection

Recent experiments by C.M. Boyce and coworkers considers the rapid, high-speed injection of a single bubble into an incipiently fluidized bed. Here, the 50 m/s jet is turned on for 150 ms causing a single large bubble to form, rise through the 3 mm particle bed and erupt a the surface. Ovito animation shows particles in the center 10 mm thick slice of the 190 mm diameter bed, colored by their vertical velocity ranging from -0.3 m/s (black) to 1.0 m/s (white). Simulation models 260K particles with MFIX-Exa 19.08 using 32 CPU cores.

Animation courtesy William D. Fullmer, NETL, Morgantown WV.

Granular Rayleigh-Taylor

Qualitative Benchmarking: Gravitational granular flow instability

The "granular" variant of the classical RT instability superimposes an assembly of 1.12M random loose packed particles over air. The instability finger/bubble pattern which reaches a (statistical) equilibrium between coarsening via finger merging and refinement via bubble splitting. Black and white ovito animation shows particle location. Simulation performed with MFIX-Exa 19.08 238 16 CPU cores.

Animation courtesy of William D. Fullmer, NETL, Morgantown WV.

Mehrdad's Bed

Qualitative Benchmarking: Ordered bubbling via periodic fluidization

Typical random bubbling patterns under uniform fluidization can be driven to ordered states under periodic fluidization. At certain system sizes and conditions, a regular left-right alternating single bubble pattern can be achieved. The animation, rendered in Blender, shows time 10s to 15s played back at 1/4x speed. Simulation considers 166k particles run with MFIX-Exa 19.08 using 32 CPU cores.

Animation courtesy of Justin Weber & William D. Fullmer, NETL, Morgantown WV.

Clustering in the HCS

Qualitative Benchmarking: Dissipation induced clustering instability

A homogeneous cooling system (HCS) of size 256 x 256 x 8 particle diameters containing 50k particles (concentration ~5%) with an initial thermal Reynolds number of 20, restitution coefficient of 0.8 and particle-fluid density ratio of 1000. The system is significantly above the critical length scale for the onset of clustering causing the originally homogeneous state to develop spatial inhomogeneities. Animation shows particle velocity magnitude ranging from 0.0116 m/s (white) to 0 (black). Simulation run with MFIX-Exa 19.08 using 16 CPU cores.

Animation courtesy of William D. Fullmer, NETL, Morgantown WV.