Parametric Uncertainty Quantification of Aviation Environmental Design Tool

摘要

Aviation Environmental Design Tool (AEDT), developed by the US Federal Aviation Administration, is a comprehensive computer model that estimates environmental impacts from aviation including fuel burn, emissions, and noise. In this study, a parametric uncertainty and sensitivity analysis are performed to identify the main contributors to AEDT output uncertainties and to gain better insights on the areas of future AEDT improvements. The study is performed in four steps. First, the sources of the uncertainties that may impact the key environmental metrics are characterized. The AEDT input parameters are grouped into categories of aircraft terminal area performance, aircraft cruise performance, noise characteristics, engine emissions, and airport weather. Their uncertainty bounds, probability distributions, and the correlations are defined. Second, a sensitivity analysis quantifies the impacts of independent changes of the input parameters to the key environmental metrics. Utilizing artificial neural network based surrogate models, Monte Carlo Simulations are performed to obtain the probability distributions of the environmental metrics. Applying Copulas theory, joint probability distributions of input parameters are modeled to capture the impact of correlations among the input parameters. Finally, Global Sensitivity Analyses provide Total Sensitivity Indices that quantify the significance of input probability distributions to variance in output distributions.