Quality-Driven Deformation of Unstructured Meshes for Navier-Stokes Solutions of Geometrics with Moving Surfaces

This research addresses the problem of deforming, unstructured solution domains associated with aircraft simulations involving moving boundaries. Current techniques are either not robust in the presence of the highly-stretched elements associated with viscous CFD simulations, are prohibitively expensive in terms of computational resources and parallel scalability, or are so narrowly-focused on a particular solution domain that their general usefulness is lost. A new two-region mesh deformation technique is developed that caters to viscous meshes. The method is then incorporated into a new method for handling the mesh movement needed to accurately model moving control surfaces typically used in maneuvering flight. An objective quality measure applicable to unstructured, viscous meshes is formulated using a solver-focused approach, and available control parameters for the deformation technique are tuned using the quality measure to provide desired mesh characteristics for a known surface motion.;The deformation method is shown to maintain a valid mesh for deformations of an AGARD 445.6 wing corresponding to tip deflections of about 38% of the root chord and tip rotations of about 29 degrees. The method requires approximately 5% of the time required by other techniques to deform similar viscous meshes. The technique for modeling control surfaces results in substantial improvements in matching a realistic target surface profile for control deflections up to 60 degrees. Application of the technique to a NACA 0015 wing with aileron shows minimal degradation in surface smoothness, and valid viscous meshes are maintained for control deflections to at least 30 degrees. The new quality metrics are shown to correctly identify poor element configurations in a viscous mesh, and the globally-weighted measure is seen to be sensitive to mesh deformation as well as changes in the control parameters. The global measure is successfully incorporated into a genetic optimization process to determine a set of valid deformation parameters for a prescribed control deflection range. Coupled flow solutions demonstrate successful parallel operation of the technique on up to 256 processors and show that the cost of the method as a percentage of overall solution time step is roughly 1/3 of that reported in the literature for other techniques.