学术文献库

The rise in computational capability has increased reliance on simulations to inform aircraft design. However aircraft airworthiness testing for flight certification remains rooted in real-world experiments performed after manufacturing an aircraft prototype. Leveraging multi-fidelity modeling and uncertainty quantification, we present a framework creating a stochastic representation of the aircraft, uses it to simulate flight certification maneuvers, and determines the likelihood of successfully meeting the certification requirement. We focus on uncertainties associated with Computational Fluid Dynamics simulations solving the Reynolds-Averaged Navier-Stokes equations. The simulation predictions and associated uncertainties are combined with data from other analysis tools to create stochastic aerodynamics and controls databases. The databases describe the aircraft's behavior across its flight envelope and provide probability distributions for its predictions. Databases are generated for two aircraft configurations, the National Aeronautics and Space Administration (NASA) Common Research Model and the Generic T-tail Transport aircraft. Samples from the databases, representing different aircraft behavior, are created. Each sample is run through a flight simulation representing a real-world airworthiness test performed by the Federal Aviation Administration (FAA). These tests are agglomerated to create distributions of the performance metrics, quantifying the probability that the aircraft succeeds in performing the certification maneuver. Simulating flight certification testing before building a full-size aircraft prototype mitigates the enormous costs of expensive redesigns late in the aircraft design process. The calculation of the failure rates provides design suggestions to ensure the aircraft can meet the certification requirement with a prescribed success rate.