HPC Benchmarks

Presented below is a performance analysis of FLOW-3D/MP up to 128 cores for some of the typical applications of the software, namely water & environmental, metal casting, microfluidics, and aerospace.

The Performance Metric used for the analysis is defined as the total number of times a given simulation can be run in a single day (24 hours). Larger bars indicate better performance.

The benchmarks were run on nodes with dual Intel Xeon E5-2670 processors (total of 16 cores per node) connected by Infiniband interconnects of upto 40 Gbps.

Aerospace

Sloshing in an Aircraft Fuel Tank

Aerospace sloshing HPC performance benchmarkIn this simulation, the fuel sloshing in an F-16 aircraft fuel tank at various flight conditions was studied.
Mesh: 0.7 million cells
Physical models: Free-surface tracking, non-inertial reference frame, gravity, electric potential, and RNG turbulence model with dynamically computed maximum turbulent mixing length
Numerical Models: Implicit advection, implicit GMRES, split Lagrangian VOF

Fuel Sloshing during a Satellite Launch

Aerospace sloshing HPC benchmarkIn this simulation, fuel sloshing during a satellite launch was studied over various phases of the mission that included: engine ignition, pitch disturbances, engine shutdown, zero-gravity coasting, and settling thrusters followed by a final engine ignition.
Mesh: 1.0 million cells
Physical models: Free-surface tracking, adiabatic bubbles, non-inertial reference frame with angular velocity, surface tension, and laminar viscous flow
Numerical Models: Implicit GMRES

Water & Environmental

Hydraulic Jump

Hydraulic jump HPC benchmarkIn this simulation, the hydraulic jump and the overall flow over a spillway were studied.
Mesh: 14.6 million cells
Physical Models: Free-surface tracking, gravity, air entrainment, and RNG turbulence model with dynamically computed maximum turbulent mixing length
Numerical Models: Implicit GMRES

 

Flow over a Spillway

Flow over a spillway HPC benchmarkIn this simulation, the flow over a spillway was studied in order to optimize the spillway geometry at various flow conditions.
Mesh: 3.7 million cells
Physical models: Free-surface tracking, gravity, viscosity, and RNG turbulence model with dynamically computed maximum turbulent mixing length
Numerical Models: Implicit GMRES

 

Flow over a Fish Ladder

Flow over a fish ladder HPC benchmark In this simulation, the flow over a fish ladder was studied as part of a design analysis.
Mesh: 2.3 million cells
Physical models: Free-surface tracking, gravity, viscosity, fluid sources model, and RNG Turbulence model with dynamically computed maximum turbulent mixing length
Numerical Models: Implicit GMRES

 

Metal Casting

High Pressure Die Casting – Piston Shot

HPDC HPC benchmarkIn this simulation, the high pressure die casting of an aluminum A380 part using a piston shot was studied.
Mesh: 2.2 million cells
Physical Models: Air entrainment, cavitation, defect tracking, gravity, heat transfer, general moving object, and RNG turbulence model with dynamically computed maximum turbulent mixing length
Numerical Models: Implicit advection and implicit GMRES

 

Thermal Stress Evolution

Thermal stress evolution HPC benchmarkIn this simulation, the development of thermal stresses during the solidification process caused by non-uniform cooling of a V6 engine block composed of aluminum A380 was studied.
Mesh: 0.5 million elements (finite element solver)
Physical Models: Free-surface tracking, laminar viscosity, and surface tension
Numerical Models: Finite-element solver for Fluid Structure Interaction and implicit GMRES

 

Engine Block Casting

Engine block casting HPC benchmarkIn this simulation, the gravity casting of an engine block was studied.
Mesh: 3.6 million cells
Physical models: Free-surface tracking, gravity, heat transfer, solidification, and viscous laminar flow
Numerical Models: Implicit GMRES

 

High Pressure Die Casting of an Engine Block

High pressure die casting HPC benchmarkIn this simulation, the high pressure die casting process of a 3.5L Duratec engine block was studied.
Mesh: 2.0 million cells
Physical models: Free-surface tracking, air entrainment, adiabatic bubbles, defect tracking, gravity, heat transfer, solidification, scalar tracking, and viscous laminar flow
Numerical Models: Implicit GMRES

 

Gravity Casting

Gravity casting HPC benchmarkIn this simulation, the gravity casting of a large part using red brass 844 was studied to investigate the formation of defects during the process.
Mesh: 3.0 million cells
Physical models: Gravity, heat transfer, free-surface defect tracking and scalar transport.
Numerical Models: Implicit SOR

 

Microfluidics

Ink Drop in a Printer Nozzle

Microfludics HPC benchmarkIn this simulation, the formation and discharge of an ink drop in a printer nozzle was studied.
Mesh: 2.0 million cells
Physical models: Free-surface tracking, laminar viscosity, and surface tension
Numerical Models: Implicit GMRES

 

Best Case Scenario

Lid Driven Cavity

Lid driven cavity HPC benchmarkThe standard lid driven cavity problem was simulated to demonstrate the scaling potential of FLOW-3D/MP. This is a fully filled, perfectly load balanced simulation which is used to validate standard CFD codes.
Mesh: 10.0 million cells
Physical models: Viscosity and RNG turbulence.
Numerical Models: Implicit GMRES

 

Download a PDF of these benchmarks >

For more information, please contact Flow Science at sales@flow3d.com or 1-505-982-0088.