Software for Automated Generation of Reduced Thermal Models for Spacecraft Thermal Control
Thermal analysis is increasingly used in the engineering of spacecrafts at every stage, including design, test, and ground-operation simulation. Currently used high-fidelity modeling and simulation tools at NASA are computationally prohibitive and not fully compatible with integrated analysis of spacecrafts. We propose to develop and demonstrate an innovative Model Order Reduction (MOR) software to automatically generate nonlinear reduced thermal models for spacecraft analysis. The underlying principle of our approach is to project the original full models onto a characteristic, low-dimensional subspace, yielding reduced models with markedly low computational orders.During Phase 1, key technology elements were developed and proof-of-concept was successfully demonstrated. A MOR engine encapsulating carefully selected nonlinear MOR algorithms, a reduced model solver and a verification module along with facile data exchange interfaces, were developed in an integrated software environment. By way of whole-satellite (LISA) case studies, critical evidence was established that reduced thermal models enable unprecedented speedup (10?500X) and accuracy (<0.3%) for spacecraft analysis and design. In Phase 2, MOR engines will be optimized for enhanced computational performance. Robust constituent linear algorithms and domain-wise projection spaces will be developed to improve simulation stability and accuracy. New MOR capabilities to address variable-dependent models and parameterized MOR to accommodate design perturbations will be investigated. Our MOR software will be extensively verified and demonstrated for complex spacecraft thermal analysis. An application programming interface (API) will be developed in close collaboration with leading NASA vendors (C & R Technologies) to facilitate technology insertion and transition.
Small Business Information at Submission:
CFD Research Corporation
AL Huntsville, AL 35805-1944
Number of Employees: