We’re two years into CEEC, and we’re starting to get a real look at how our efforts will help push computational fluid dynamics forward. This summer, we started conducting our first smaller waLBerla and Vistle simulations with limited data sets to see how Vistle visualization tools could improve the analysis experience of fluid-particle simulations, such as those needed in our Lighthouse Case (LHC) concerning localized erosion of offshore wind-turbine foundations. In particular, we were excited to see how the virtual reality (VR) capabilities of the Cave at the Higher-Performance Computing Center Stuttgart, our partner institution, when combined with Vistle, could help scientists identify and understand erosion in coupled fluid-particle flows.
Our next goal is to scale up simulations, allowing us to “swim” through particle systems in the Cave. Given the scale and complexity of our LHC simulations, this immersive experience could enable faster and more accurate pinpointing of fluidization—where particles begin to behave like fluids, a precursor to piping erosion, the critical phenomenon for our LHC.
To more easily identify fluidization, we’ve implemented a Vistle module for the LHC, which we’re in the process of optimizing for graphic processing unit (GPU) performance, a vital step to running this module at exascale. This module creates connecting lines in the flow visualization that indicate which particles overlap, where the thickness of the line indicates how much they overlap (the thicker the line, the greater the overlap). The connecting lines represent the contact forces between the particles. This contact force network can then be used to analyze the simulation: If, for example, long chains form, this indicates an unstable area that can lead to piping erosion.
As we look ahead, waLBerla and Vistle developers aim to achieve even more refined results in the coming years as well as apply similar tactics to more of the CEEC LHCs. In the meantime, if you want to learn more about Vistle and how it might enhance your own work, make sure to check out our Vistle webinar on January 29th!
