This presentation was made at CAASE18, The Conference on Advancing Analysis & Simulation in Engineering. CAASE18 brought together the leading visionaries, developers, and practitioners of CAE-related technologies in an open forum, to share experiences, discuss relevant trends, discover common themes, and explore future issues.
High-performance computing workflows, or the coupling of multiple software tools and computing resources into a desired dataflow, are plagued with ad-hoc approaches, limited availability of expertise, and complexity connecting to diverse computing resources. The benefits of automating such practices within an organization, however, are large but often difficult to identify a clear ROI to justify setup and resource costs.
This presentation explores an open-source framework from Argonne National Laboratory used to facilitate the automation of an organization’s HPC workflow processes. These workflows are then made available through Parallel Works, a SaaS offering for building, running and sharing parallel computing workflows, transparently connecting to R-Systems bare metal HPC infrastructure.
We will demonstrate the application and ROI of workflow automation with the rLoop team, a complex decentralized organization designing and building one of the first Hyperloop pods. The Hyperloop is a proposed mode of passenger and/or freight transportation that combines the convenience of a train with the speed of an airplane via a levitating pod that may travel free of air resistance or friction conveying people or objects at high speed in a sealed tube or system of tubes. rLoop’s Hyperloop pod design presents a range of interesting multiphysics challenges and simulation activities, including structural analysis, compressible and incompressible fluid dynamics, composite, electromechanical, and electronics design among others.
This presentation will focus on demonstrating the value derived from rLoop’s aerodynamic designs using incompressible solvers in OpenFOAM, a leading open-source fluid dynamics software that can be nicely parallelized both horizontally and vertically. Two ROI aspects will be demonstrated including the benefit of large-scale design exploration over the OpenFOAM automated pipeline, as well as the democratization of these encapsulated aerodynamics workflows throughout the decentralized rLoop organization. We will showcase and quantify benefits to the rLoop organization including a more productive use of their major computing software and hardware resources, rigorous interrogation of a model’s parameter space leading to improved design, and the knowledge transfer of these computing practices from technical experts to engineers and analysts.
|Date||5th June 2018|