Using the Lattice Boltzmann Method to Estimate the Air Transport Properties of Polar Firn
Loading...
Authors
Toller, Justin
Issue Date
2023
Type
Thesis
Language
Keywords
Computational Fluid dynamics , firn , image processing , Lattice boltzmann method , micro-ct , paleoclimate
Alternative Title
Abstract
Polar firn is characterized by small, interconnected pores by which air from the overlying atmosphere can diffuse through the firn column and eventually become trapped within glacial ice. How well we characterize the pore space, and the ability of air to migrate through it, ultimately dictates how well we can reconstruct past climate records. Modern models use the proxy of firn density to determine the rate at which air can diffuse through the firn column. Including firn microstructure through the consideration of intrinsic permeability of gas transport through firn models has shown promise as a proxy to better capture transport mechanisms. The intrinsic permeability of firn is constrained by the physical geometry of the organization of the pore space. Recently, the coupled use of x-ray micro-computed tomography (Micro-CT) and computational fluid dynamics have been proven to be a viable alternative to direct measurement of intrinsic permeability in the laboratory. Lab measurements of intrinsic permeability are inherently complicated by the fact that performing these measurements may induce changes to the pore space due to sublimation, or mishandling. Micro-CT measurements enable non-intrusive reconstruction of the three-dimensional pore space that can be used to create computational models that numerically estimate the intrinsic permeability from the Micro-CT scans. Given the complex geometry of the pore space in glacial firn, the Lattice Boltzmann Method (LBM) has gained popularity for performing these simulations and if properly constrained, could become a standard in obtaining the important proxy of intrinsic permeability of firn cores. To this end, the following research illustrates the importance of image preprocessing parameter consideration and selection of an appropriate collision model when applying LBM to numerically estimate the intrinsic permeability of glacial firn from Micro-CT data. Specifically, I analyze a range of thresholding or binarization choices which take the grey-scaled (0-255) raw images and turn them into black and white pixel distributions for input into LBM software. In addition, I compare the single-relaxation time to the two-relaxation time collision operators, commonly applied in LBM simulations for firn conditions. I then compare the simulated results to lab measured values of intrinsic permeability for firn samples from NEEM, Greenland. The results highlight that the thresholding choice is extremely important to consider when simulating the intrinsic permeability. Furthermore, the single-relaxation time collision operator is shown to both introduce non-physical variability as a function of viscosity while also suffering significant stability issues. Comparatively, the two-relaxation time collision operator estimates are viscosity independent\textemdash as is required based on the definition of intrinsic permeability\textemdash while also being stable over a larger range of viscosities. Finally, I consider the current methods of obtaining the important parameter of open porosity of firn when modelling the difference in age between glacial ice and the air trapped within it.
Description
Citation
Publisher
License
Creative Commons Attribution 4.0 United States