Modeling Of Human Respiratory Flow

Open Access
Dogan, Gulkiz
Graduate Program:
Aerospace Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
Committee Members:
  • Robert Francis Kunz, Thesis Advisor
  • CFD
  • pulsatile flow
  • pipe flow
  • CFD solver
  • human respiratory flow
  • lung geometry
  • nphase
  • medical image
This work represents a contribution to and advancement of several components of a semi-automated subject-specific end-to-end medical image through CFD (computational fluid dynamics) analysis capability for the human respiratory system. Specifically, this thesis presents geometric segmentation, manipulation and grid generation algorithms for the human lung, development of quasi-one-dimensional (Q1D) geometric and flow modeling approaches for the unresolved “convective regime,” verification and validation of pulsatile pressure-forced boundary conditions, a script-based multidisciplinary simulation framework for respiration, and a number of representative CFD simulations of respiration. In the area of geometric modeling, four specific contributions are made: i) semi-automated processing of medical image data to derive upper airways and lobe geometries for in-vivo subjects, ii) creation of partitioning and truncation algorithms, with application to the conducting airways of a rubber cast model of a dead subject, iii) automated unstructured 3D gridding of the trachea through generation 5-8, and, iv) interfacing this upper bronchi and lobe geometry with a volume filling algorithm for the sub-resolved bronchi. For unsteady pressure-forced flows, as in the respiration simulations pursued here, the specification of well posed and accurate streamwise boundary conditions are of concern, especially where inflow and outflow from pressure boundaries arise. The adequacy of the boundary condition approaches taken in this work are demonstrated by comparison of unsteady 2D/3D CFD simulations with known analytical solutions to the incompressible Navier-Stokes equations for non-dimensional frequencies and Reynolds numbers of relevance to respiration. These contributions are summarized in detail and employed in concert with other elements of a respiratory simulation framework under development at Penn State University.