Open Access
Porzio, David Philip
Graduate Program:
Master of Science
Document Type:
Master Thesis
Date of Defense:
Committee Members:
  • Robert Francis Kunz, Thesis Advisor
  • computational fluid dynamics
  • cfd
  • multiscale
  • human respiration
  • macroscale
  • breathing
  • microscale
This research contributes several elements to an existing and evolving multi-scale model of human respiration. This model takes patient’s medical scans (CT) and extracts the airways into a surface geometry file and associated skeletonized data file. A semi-automated geometry algorithm truncates the airways and gives them generational assignments along with planar pressure boundaries at their desired outlets. A volume mesh is then generated from the surface mesh that can be used for respiration simulation. The planar pressure boundaries interact with a second mesh comprised of automatically generated Quasi-1D pipes representing the lower airways which saves considerable amounts of computational time. This research improved on several of these modeling components. New algorithms were developed, coded, and verified for truncation and generational assignment that are dependent on a separate attribute algorithm, which works in conjunction to replace the previous geometry tool. Also, new automated software tools were developed to create idealized CAD models and meshes for computational simulation to calculate the losses in sub-grid scale bifurcation regions that connect the Quasi-1D pipes in the lower airways. The truncation algorithms were shown to be more robust than their predecessor when dealing with very complex medical imagery; likewise, the associated generational assignment algorithm was shown to have increased accuracy when compared to the previous version. However, mixed success was obtained from the development of loss models for the sub-grid scale bifurcation regions. Computational simulations gave consistent values for the loss associated with these geometries and demonstrated that the flow and subsequent loss in the lower airways is dominated by viscous effects. Unfortunately as these losses were orders of magnitude smaller than the head loss in such flows, the inherent slight variations from mesh to mesh that arises when changing a feature for comparison are enough to obscure the trends from being properly resolved.