Publikationstyp: Konferenz: Sonstiges
Art der Begutachtung: Peer review (Abstract)
Titel: KaleidoSim : massive simultaneous cloud computing for multiphysics simulations
Autor/-in: Boiger, Gernot Kurt
Sharman, Darren
Drew, Dominic
et. al: No
Tagungsband: Multiphysics 2021
Seite(n): 29
Angaben zur Konferenz: International Conference of Multiphysics, Online, 9-10 December 2021
Erscheinungsdatum: Dez-2021
Verlag / Hrsg. Institution: International Society of Multiphysics
ISSN: 2409-1669
Sprache: Englisch
Schlagwörter: Multiphysics; Cloud; Cloud computing; MSCC; Simulation; CFD
Fachgebiet (DDC): 530: Physik
Zusammenfassung: The cloud-computing software KaleidoSim has been developed. While the current web-platform focuses on compatibility with the open source CFD toolbox OpenFOAM, KaleidoSim enables MSCC Massive Simultaneous Cloud Computing functionality for any Multiphysics simulation software. Thereby the idea is to provide an easy-to-use, web-browser-based software, which allows the user to run any Multiphysics simulation in the cloud within minutes. Rather than just replacing local hardware, KaleidoSim has been specifically designed to tackle the issue of horizontal scaling in the cloud. While vertical scaling (parallelization) can well be conducted according to a standard selection of cloud-computers with up to 224 cores each, KaleidoSim’s key functionality is about the orchestration of multiple cloud-machines running simultaneously. KaleidoSim has been tested for workflows involving the seamless coordination of up to 500 individual, simultaneous simulation runs. Features include: i) The Self-Compile Option where OpenFOAM-based, self-composed software can be directly uploaded, compiled and run in the cloud; ii) The Katana File Downloader, which enables the selective download of any computed cloud data; iii) The In-Cloud Post-Processing Option for Paraview-based, automated post-processing via Python-trace utilities; iv) The Kaleidoscope Feature enabling the automated creation of large parameter studies; v) The API Application Programming Interface, for command-line-based access of the entire spectrum of KaleidoSim-functionality and for easy integration into third-party software. These capabilities allow the conduction of parameter studies, optimization runs and generally the creation of vast amounts of simulation data with unprecedented speed. This means that Kaleidosim effectively democratized vast computational resources and enables the integration of numerical studies of significantly higher data-intensity, investigative depth and thus quality, into every-day Multiphysics investigations. Application examples of several thermo-fluiddynamic scenarios can be demonstrated in which work-flow-speed-up-factors of no less than 50-100 can be achieved.
Volltext Version: Publizierte Version
Lizenz (gemäss Verlagsvertrag): Lizenz gemäss Verlagsvertrag
Departement: School of Engineering
Organisationseinheit: Institute of Computational Physics (ICP)
Publiziert im Rahmen des ZHAW-Projekts: Experimentaldesign für Simulationsläufe mittels Massivem Simultanem Cloud Computing (MSCC)
Enthalten in den Sammlungen:Publikationen School of Engineering

Dateien zu dieser Ressource:
Es gibt keine Dateien zu dieser Ressource.
Zur Langanzeige

Alle Ressourcen in diesem Repository sind urheberrechtlich geschützt, soweit nicht anderweitig angezeigt.