Web platform architecture for CFD simulations and real-time analysis on HPC resources

The availability and continued growth of high performance computing (HPC) is opening new avenues for complex physics based software simulations. The usage of high performance computing is particularly important in high-fidelity large-eddy simulation of multi-physics engineering problems such as the development of more e cient and less polluting advanced energy technologies. Large-eddy simulation is a branch of computational fluid dynamics (CFD). While high performance computing based large-eddy simulation is common amongst researchers, its adoption in commercial industries is still hindered by inherent complexities in utilizing the associated software tools. The proposed solution herein is a web-based platform with user-interface tools that support large-eddy simulation in high performance computing environments. Development through Phases I and II will build a web platform that accomplishes several overall goals: provides immediate value to large-eddy simulation software usability through stand-alone user-interface tools creates a collaborative web-based simulation framework that engages computational experts, designers and decision-makers enables greater knowledge and insight generation from high-fidelity large-eddy simulation data The web-platform user interface tools are targeted at a commercially marketed large-eddy simulation code. This code was developed from turbulence modeling methodologies resulting from the Department of Energys Predictive Science Academic Alliance Program" at Stanford University. During Phase I the web based platform and three user interface tool prototypes were deployed. The platform demonstrated the capability to connect to computing resources that run large-eddy simulation software. The user interface prototypes then simplified tasks associated with simulation setup, interactive analysis and data management. Commercial user feedback on the tools was gathered and assimilated into objectives for Phase II. This second phase will focus on refining the front facing user experience of the web tools while also enhancing their underlying functionality. For example, one innovative prototype from phase I allowed engineers to interactively inspect simulation data much like a radiologist inspects planar magnetic resonance imaging (MRI) output. Phase II will reduce the setup steps associated with this tool and add additional analysis probes to the maps like pan and zoom interface. Probes for data correlations and time animation will require additional back-end features from the large-eddy simulation code. Additional research and development will explore modularization of the web platform to allow a plug-in like capability for new user interface tools. With respect to energy technologies, large-eddy simulation is poised to impact the design of cleaner and more e cient gas turbines. The web platform will ensure that industry users can e ciently derive knowledge from simulation to create design insights.