A High-Order and Multivariate Interpolation Method For Adapting Reduced-Order Models to Continuous Parameter Changes

This paper presents an interpolation method for adapting projection-based ReducedOrder Models (ROMs) to new sets of physical and/or modeling parameters. The new method is based on the Grassmann manifold and its tangent space at a point and is applicable to many structural, fluid, and multidisciplinary engineering problems. It is illustrated here with the computational mechanics-based reduced-order modeling of a complete fighter aircraft configuration and the adaptation in near real-time of the resulting parameterized ROM to new flight conditions. Good correlations are obtained with results from direct and therefore more expensive ROM reconstructions and from computationally intensive high-fidelity simulations. These results demonstrate the potential of the proposed ROM adaptation methodology for the near real-time analysis of large-scale computational models using ROM databases. This in turn contributes to strenghtening the potential of computational mechanics for design and optimization.