MODEL FIDELITY REFINEMENT FOR SYSTEM-LEVEL DESIGN OF A REUSABLE LAUNCH VEHICLE

This paper extends research in multidisciplinary reliability-based design optimization by using model uncertainty quantification to integrate multiple levels of design. The bi-level design of a reusable launch vehicle with a component fuel tank is studied from this perspective. The system design includes a conceptual, low-fidelity weights analysis with significant model uncertainty, while the tank design provides a more rigorous analysis for the weight of one component with significantly less uncertainty. Evaluating the uncertainties of the disciplinary analysis provides two important benefits. The first is as a metric for selecting appropriate disciplinary model(s) to integrate into the system-level analysis for a desired level of reliability. The second benefit is as the basis for iterating between design levels. Rather than fully integrating the system and component analyses, this method uses model uncertainty to leverage the component design to refine the system analysis. Thus the communication between design levels is less stringent than for fully-integrated reliability based optimization.