• White Facebook Icon
  • White Twitter Icon
  • White YouTube Icon

Numeric Systems GmbH

Lichtenbergstraße 8

D-85748 Garching (Munich)

Germany

info@numeric.systems

 

Welcome to the new CFD Performance Dimension of

Fast and Resilent

Utilizing GPUs (Graphics Processing Unit), pacefish® achieves much higher performance compared to CPUs (Central Processing Unit, Universal Processor) massively reducing the costs of a physical fluid flow simulation.

Using other CFD tools the user has to choose between short turn-around times and reliable simulation results. By drastically reducing the cost per simulation, pacefish® users get both:


Short turnaround time for transient fluid flow simulations and resilent physical simulation results.

 

pacefish® Highlights

Transient Lattice-Boltzmann Simulation

In order to determine the fluid flow behavior, pacefish® applies the Lattice-Boltzmann Method (LBM).

Transient modeling considers time-dependent effects like Vortex streets, dynamic vortex shedding as well as pressure and velocity fluctuations.

Robust Geometry Import

The integrated robust pre-processor imports complex geometries stored in the widely used STL data format tolerating minor defects.

Usually, preceding geometry cleanup is not required anymore.

Geometry-Adapted Local Grid Refinement

Due to local grid refinement simulation cells are used where needed, e.g. in the near-wall or in the wake region.

Based on the object geometry the integrated preprocessor automatically generates areas of local grid refinement.

Flexible Result Data Export

The result data is typically exported using the EnSight Gold and the CSV data formats enabling further processing using numerous tools like ParaView or Excel.

Physical Turbulence Modeling

In addition to the Smagorinsky LES model, pacefish® offers hybrid uRANS-LES (SST-DDES and SST-IDDES) as well as uRANS (K-Omega-SST) turbulence models.

By the particularly accurate modeling of the turbulent boundary layer the latter models achieve significantly more reliable predictions.

Multi-GPU Acceleration

For further performance increase pacefish® uses up to 16 GPUs operating at almost ideal scaling.

This allows simulations beyond 1 billion cells. Models consisting of more than 100 million cells achieve turnaround times of just a few hours.

The pacefish® Technology

Efficient use of modern graphics processors (GPUs) enables the exceptional pacefish® performance. GPUs are massive parallel processors consisting of more than 5000 cores, natively used for graphics rendering. GPUs have limited capabilities, but can handle information much faster than CPUs.

The GPU hardware architecture induces special software development challenges. With the very first source code line this requirements were taken into account such that pacefish® effectively exploits the benefits of the GPU architecture and neutralizes its limitations.

By using one single GPU pacefish® performance is already 20 times faster in comparison to a multi-core CPU. Utilizing up to 16 GPUs in parallel, pacefish® achieves impressive performance at almost ideal scaling.

Resilent forecast accuracy is the reward for intensive research, development and testing, in particular of wall and turbulence modeling for LBMs. Our proprietary know-how built up in over ten years overcomes important elementary weaknesses of the LBM and allows pacefish® to simulate models of industrial complexity.

 

Application Areas

Automotive Aerodynamics

Analyzing and optimizing the complex aerodynamics of vehicles.

Pedestrian Comfort

Analyzing the transient wind behaviour in urban canyons on changes in building design and placement.

Wind Loading

Determining and optimizing dynamic wind loads on the facade and structure of a building.

 

Usage options

Direct by Numeric Systems as classical rich-client application using own hardware.

By SimScale as web- and cloud-based Application in a browser.