Development of Multidisciplinary, Multi-Fidelity Analysis and Integration of Aerospace Vehicles

ABSTRACT: Aurora Flight Sciences, in collaboration with the Massachusetts Institute of Technology (MIT), proposes to further develop and test in realistic settings an innovative method for representing and managing multidisciplinary design information from a wide range of analysis tools. The practical implication resulting from this novel approach is a mathematical framework to confidently determine the level of analysis fidelity required to maintain an acceptable level of design uncertainty at any point in the design process. To this end, key design parameters are modeled within a Bayesian context to represent the level of disciplinary uncertainty or confidence in each parameter during the design process. Tools and models are characterized in the same way using their inherent level of disciplinary uncertainty. Estimation Theory is used to systematically merge the results from multiple tools and models in a manner that is commensurate with their respective levels of fidelity. The entire history of information generated from all previous activities, regardless of their individual levels of fidelity, is preserved and used to synthesize updated design parameters. Therefore, the central element of the proposed innovation is that the results from higher fidelity analyses are used to complement, rather than supplant, the results from lower fidelity analyses. BENEFIT: The generic, far-reaching benefit of the proposed innovation is that it enables system managers to use lower-fidelity models and tools more confidently because the disciplinary uncertainty is explicitly and actively maintained and propagated throughout the design process. In addition, since lower fidelity information is preserved throughout the design process, our proposed approach potentially reduces the number of calls to high-fidelity models and tools. For Aurora, the proposed innovation is a key enabling technology for design-related activities as applied to unconventional vehicle concepts. To this end, the development of the proposed innovation would help Aurora maximize the utility of its collection of design tools. For the sponsor, the multi-fidelity management of a multidisciplinary design concept using Estimation Theory directly applies to the AFRL mission statement of discovering, developing and delivering a wide range of revolutionary technologies. More immediately, this project provides Aurora several commercialization opportunities. To this end, a part of the Phase II effort is envisioned to include the development of a generic software library containing all the core innovative algorithms and integration of this toolbox into Aurora’s internal multi-fidelity, multidisciplinary aircraft design optimization platform. This possibility is reinforced by the generic Bayesian framework at the core of the innovation, which enables its application over a variety of design processes beyond aircraft conceptual design.