Reduced-Order Modeling for Rapid Noise Prediction

Master's Thesis

Research Motivation

The Aviation Environmental Design Tool (AEDT) is simulation software for modeling and assessing the environmental ramifications (e.g., fuel consumption, emission, and noise) of aircraft operations. In practice, the application of the AEDT could range from a single aircraft operation over an airport to thousands of aircraft operations at multiple airports. As a consequence, the AEDT is typically required to run repeatedly for aviation environmental research, which may create considerable computational costs. To enhance the computational efficiency of noise evaluation in the AEDT, this research devised a rapid approximation of the AEDT via Reduced-Order Modeling (ROM) as the technique is effective in approximating large-dimensional data.

Key Idea

This research aims to develop a reduced-order AEDT noise model with Proper Orthogonal Decomposition (POD) and ordinary Kriging (or called Gaussian Process). In particular, POD and ordinary Kriging were leveraged for orthonormal basis extraction and basis coefficient prediction, respectively. For demonstration, a reduced-order AEDT model was developed by associating two AEDT outputs---departure and approach noise---with five atmospheric parameters---elevation, temperature, pressure, relative humidity, and headwind---for single departure and arrival flights.

SEL noise surfaces and contours obtained by the AEDT for single departure and arrival flights: (a) departure noise; and (b) approach noise

Atmospheric parameters and their ranges determined based on the monthly average weather of 150 major airports in the United States

Results

The constructed reduced-order AEDT noise model showed remarkable approximation capability, even in the worst case shown below, due to basis coefficients accurately estimated by the Kriging models. More importantly, the approximate AEDT noise model required about 0.0042 sec. in contrast to the original AEDT noise model, necessitating about 7 sec. to complete the noise simulation of single departure and arrival flights.

SEL noise contours for the worst verification case: (a) departure noise; (b) approach noise

Because the reduced-order AEDT noise model runs instantaneously, it is expected to greatly benefit from many aviation environment impact research scenarios. For illustration, consider two use cases of AEDT noise simulation: local- and national-level studies, each of which represents an airport operating 1,000 departure/arrival flights and 50 airports with each operating 1,000 departure/arrival flights, respectively. Considering upfront computational costs for the offline process of ROM, the proposed approach is found hardly helpful for a local-level study, but is clearly beneficial for a national-level study.

Computational time for two use cases by the original and reduced-order AEDT noise model

Publication

• S. Kim, D. Lim, and K. Lee, "Reduced-Order Modeling Applied to the Aviation Environmental Design Tool for Rapid Noise Prediction," Journal of Aerospace Engineering, Vol. 31, No. 5, 2018. doi.org/10.1061/(ASCE)AS.1943-5525.0000860