|Goal-driven, multi-fidelity approaches for military vehicle system-level design|
|Applied Vehicle Technology|
design optimization, multidisciplinary analysis, Multifidelity analysis, multiphysics
Military requirements differ significantly now than they did just a decade ago. To meet the new and ever changing demands in a timely fashion the effort to develop new weapon systems needs to be significantly reduced. The need to bring new affordable and reliable weapons systems on line rapidly requires a quick, accurate and thorough assessment of the design space. AVT recently completed a several assessments of this need.
ET-054 explored the issue of affordable weapons systems and led to the formation of AVT-092, “Qualification by Analysis,” and AVT-093, “Integrated Tools and Processes for Affordable Weapons Systems.” AVT-093 focused on “the integration of tools and processes, not on the description of tools and processes.” AVT-093 also identified needs in multidisciplinary design optimization (MDO) that could be addressed using the integration of tools and processes in a distributed parallel computing environment that would enable a feedback of information from detail to preliminary and preliminary to conceptual design. AVT-092 recognized that these capabilities described in AVT-093 are necessary to achieve the objective of rapid design and qualification of new vehicles. Both teams recognized that there is a gap between the current technology and the desired end state of rapidly developing affordable weapons systems and developments in multidisciplinary technologies are key capabilities for closing that gap. Finally, AVT-237 focused on benchmarking the use and benefits of MDO for the development of military systems.
All these previous efforts have shown that there is benefit to including more engineering disciplines at higher levels of fidelity and coupled earlier in the development process. But, there is no mathematical framework to determine which disciplines, which level of coupling, or which level of fidelity is required to capture the physics most critical to a particular system’s design, or how to reach determinations in an optimal fashion with constrained computing resources. Currently, these decisions are based solely on experience and engineering judgement driven by flight regimes, system architectures, and or technology being included on the system. This approach works reasonably well for systems that are similar to previous designs, but can fail for unique and innovative system configurations or new technologies leading to inaccurate information being supplied to military leadership decision makers concerning system capability and technology assessment.
To identify the current state of the art associated with mathematically rigorous frameworks for adaptive selection of different sources of information from data/models for accurate prediction of the response of a system. Identify any algorithms that mathematize the dependence of goals on the disciplines modelled and the available modelling sources to create a “multi-source decision function” accounting for evaluation accuracy, analysis time, analysis cost, parametric representations of models, physical conditions, and configuration geometry and uncertainty in all these aspects. The algorithms should predict system behaviours critical to decisions being made by a variety of evaluators, including analysts, designers, or potentially other multi-physics algorithms. Identify algorithms that quantitatively assess: which disciplines/physics to couple (from all possible combinations); how tightly to couple the disciplines/physics; which theoretical model sources to select (from all available) from a given domain to predict the multi-physics response of the system for goal driven decisions.
(1) Mathematically rigorous information-theoretic frameworks for assessing variable-fidelity information across multiple length and time scales; compatibility of coupled multi-physics models with different fidelity;
(2) Developing efficient and accurate algorithms for utilizing high-fidelity methods (e.g., PDE solvers) to construct low-fidelity models (e.g., reduced-order models) while maintaining accurate representation of underlying physics; conversely, utilizing low-fidelity models to determine when high-fidelity is needed, what is the underlying functional dependence of "fidelity" to variables contained in the model;
(3) Designing with variable-fidelity possibly uncertain numerical simulations, where requirements to choose level of fidelity are needed to account for dynamic system modeling;
(4) Inclusion of data from tests and system operation as feedback into a robust design process featuring stable computations.
(5) Benchmark test problems by which different relevant methods and frameworks can be compared and assessed.