Parallel Adaptive High-Order CFD Simulations Characterizing SOFIA Cavity Acoustics
Speaker: Michael Barad, Applied Modeling and Simulation Branch, NASA Ames Research Center
Presentation
Abstract
This work presents one-of-a-kind MPI-parallel computational fluid dynamics simulations for the Stratospheric Observatory for Infrared Astronomy (SOFIA). SOFIA is an airborne, 2.5-meter infrared telescope mounted in an open cavity in the aft of a Boeing 747SP. Higher-order accurate simulations using NASA’s Launch Ascent and Vehicle Aerodynamics (LAVA) CFD solver are presented. These simulations focus on how the unsteady flow field inside and over the cavity interferes with the optical path and mounting of the telescope. A temporally fourth-order accurate Runge-Kutta, and spatially fifth-order accurate WENO-5Z scheme was used to perform implicit large eddy simulations. An immersed boundary method provides automated gridding for complex geometries and natural coupling to a block-structured Cartesian adaptive mesh refinement framework. Strong scaling studies using NASA's Pleiades supercomputer with up to 32K cores and 4 billion cells shows excellent scaling. Dynamic load balancing based on execution time on individual AMR blocks addresses irregularities caused by the highly complex geometry. Limits to scaling beyond 32K cores are identified, and targeted code optimizations are discussed.
Biography
Michael Barad is a research scientist in the Applied Modeling and Simulation Branch at NASA Ames Research Center. Mike received his M.S. at UC Berkeley in 1997 focusing on computational environmental fluid mechanics. He received his Ph.D. at UC Davis in 2006, where his dissertation was "An Adaptive Cartesian Grid Projection Method for Environmental Flows". During his Ph.D. he was a DOE Computational Science Graduate Fellow, and through this he conducted his research within the Applied Numerical Algorithms Group at Lawrence Berkeley National Laboratory. From 2006-2009 he was at Stanford's Environmental Fluid Mechanics Laboratory as a NSF Mathematical Sciences Postdoctoral Research Fellow. There he continued to focus on adaptive mesh refinement (AMR) algorithms, and applied them to simulating multiscale nonlinear oceanic internal gravity waves. While at Stanford he also taught several graduate level courses. Mike then came to NASA Ames in 2009 and works within the LAVA group where his focus is applications and numerical tools. In LAVA, he has developed a high-order high-performance immersed boundary tool for solving unsteady Navier-Stokes equations using adaptively refined Cartesian grids, requiring only a water-tight surface triangulation. Using the LAVA solver he has worked on numerous NASA projects, including: the 21st century launch complex at KSC; SLS buffet analysis; NESC Battery simulations; Landing Gear Noise; and many others. He has published in the Journal of Fluid Mechanics, Journal of Computational Physics, and others. Mike received the NASA Early Career Achievement Medal in 2014.