High Fidelity Dynamic Analysis of Launch Vehicles on Single-Image Supercomputers

Abstract

** Understanding the transonic buffet condition during launch is important for the mission of any launch vehicle to be successful. Accurate computational modeling of this flight condition requires software based on high fidelity solution modules that s imulate the occurring transient fluid/structure interaction using CFD/CSD methods. The necessary software to model these types of atmospheric launch conditions was cost effectively generated by extending the capabilities of the current NASA software HiMA P. HiMAP is an efficient super modular process to simulate the aeroelastic behavior of aerospace vehicles using high -fidelity flow equations such as the Euler or Navier -Stokes equations. In this project, a dynamic transient structural sub -domain analysis capability was created. This Computational Structural Dynamics (CSD) software module uses 9 -degree -of freedom triangular finite elements and a consistent mass matrix for dynamic transient analysis. A post -processing routine was added to calculate the str ess distributions in the structural model. The CSD module was designed in such a way that it could be plugged directly into the CFD code for accurate fluid -structure interaction analysis. The CFD grid set -up as provided for flow analysis in HiMAP and the e xisting fluid -structure interface in the CFD code were used as is. The CSD module was formulated for parallel implementation, using the MPI library. All aspects of the code were and continue to be tested in stand -alone and integrated mode, both in series a nd in parallel. Since the resulting CFD/CSD based transient aeroelastic simulation is very time consuming, it is most cost -effectively performed using closely coupled cluster computers. In this research, computational resources built on commodity PC and/o r work - station clusters were utilized. Specifically, the resulting new code was implemented on the Linux based High Performance Computer Boomer at the OU Supercomputing Center for Education and Research (OSCER) of the University of Oklahoma - an Aspen Lin ux cluster with 270 Pentium4 XeonDP processors - and will also be implemented on a single -image Linux processor cluster within Columbia at NASA Ames Research Center. The performance of the new aeroelastic code is presently being evaluated by performing hig h-fidelity computations of the fluid -structure interaction of four simple examples: bodies of revolution in a hemisphere -cylinder, a cone -cylinder, a blunted cone -cylinder, and an ogive -cylinder configuration, which all represent typical vertical launch ve hicle models.