Hussein Ezz ElDin - Academia.edu (original) (raw)
Uploads
Papers by Hussein Ezz ElDin
Flows of dense particle suspensions are of great interest to engineering, science and medicine. I... more Flows of dense particle suspensions are of great interest to engineering, science and medicine. Immerse boundary (IB) methods are commonly employed in simulations of such systems, but are usually confined to spherical particle suspensions. Extension to rigid particles of arbitrary shape introduces significant additional complexities on the IB tracking algorithms, as well as the rigid body dynamics. This increases the cost in the fluid-structure interaction (FSI) schemes employed. In this work we present a computational scheme targeted to the above problem, applicable to computations involving millions of particles on leadership high performance computing platforms. The fluid equations are discretized by standard, central, finite-differences on a staggered mesh and the equations of motion for each particle are employed on the Eulerian reference using Euler angles or quaternion variables. A Lagrangian forcing IB method is employed, using the Lagrangian particle framework of FLASH. Fluid and particle equations of motion are strongly coupled using a partitioned scheme. We present the details of the parallel implementation as well as scaling tests and results on the sedimentation of arbitrary shaped particles.
Flows of dense particle suspensions are of great interest to engineering, science and medicine. I... more Flows of dense particle suspensions are of great interest to engineering, science and medicine. Immerse boundary (IB) methods are commonly employed in simulations of such systems, but are usually confined to spherical particle suspensions. Extension to rigid particles of arbitrary shape introduces significant additional complexities on the IB tracking algorithms, as well as the rigid body dynamics. This increases the cost in the fluid-structure interaction (FSI) schemes employed. In this work we present a computational scheme targeted to the above problem, applicable to computations involving millions of particles on leadership high performance computing platforms. The fluid equations are discretized by standard, central, finite-differences on a staggered mesh and the equations of motion for each particle are employed on the Eulerian reference using Euler angles or quaternion variables. A Lagrangian forcing IB method is employed, using the Lagrangian particle framework of FLASH. Fluid and particle equations of motion are strongly coupled using a partitioned scheme. We present the details of the parallel implementation as well as scaling tests and results on the sedimentation of arbitrary shaped particles.