High-performance Computing and Data-driven Modeling of Aircraft Contrails

点击翻译按钮获取中文摘要

This project aims to model the early-phase formation of condensation trails ("contrails"), which are ice clouds generated by water exhaust from aircraft engines. Although the contrails initially appear to be linear, they can spread under favorable atmospheric conditions and form cirrus clouds that can persist for hours and eventually become almost indistinguishable with natural cirrus. Contrails and contrail-cirrus are indeed the most uncertain aviation contributions to the Earth-radiation budget (that is, the balance of energy entering, reflected, absorbed, and emitted by the Earth). Modeling this complex multi-physics problem is challenging because the different physical processes interact at different time and spatial scales. This project addresses this challenge using high-resolution numerical simulations that rely extensively on high-performance computing and advanced visualization techniques to help identify, capture and model contrail features. The primary focus of the modeling is on the early phase of contrail evolution where the presence of large-scale motions generated by the aircraft wake vortices and the small-scale perturbations induced by the jet and wake turbulence interact. The contrail parameterization results will support the integration of emissions into global atmospheric models. The sensitivity of contrail properties to the initial particle emissions may suggest potential mitigation strategies. Underrepresented minority students from the University of Illinois at Chicago, a Hispanic Serving Institution and a Minority Serving Institution, will be engaged in the computational research.The specific goals of the research are: (1) to carry out the first fully three-dimensional spatial large-eddy simulations (LES) of contrail formation that include the full aircraft geometry and to develop an accurate data-set of contrail evolution in the jet and vortex regime; (2) to identify the three-dimensional contrail features and fit parameters by journaling the simulation workflow using advanced visualization techniques; (3) to reduce the large dimensionality of the generated data-set and provide a general and accurate model of contrail structure at the end of the jet regime using Artificial Neural Networks based on high-fidelity data training; and (4) to reconstruct contrail global properties over the full time evolution using statistically inspired methods such as Polynomial-Chaos expansions for parameterization into global models. The techniques developed through this work will enable contrail parameterizations that are consistent with the physical assumptions and the conservation equations used in global atmospheric models. These techniques will further handle large uncertainties in the microphysical and optical/radiative properties of ice particles. The techniques will also handle large variability in atmospheric conditions that determine the background and boundary conditions for LES of contrails. A visual analysis framework will enable detection and analysis of flow features, as well as multi-scale and multi-run analysis of multiple simulations and parameters.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.