Three-Dimensional Computational Fluid Dynamics Simulations of Ozone Uptake in the Respiratory Tract

Taylor, Adekemi Bisola

Three-Dimensional Computational Fluid Dynamics Simulations of Ozone Uptake in the Respiratory Tract

Open Access

Author:: Taylor, Adekemi Bisola
Graduate Program:: Chemical Engineering
Degree:: Doctor of Philosophy
Document Type:: Dissertation
Date of Defense:: October 09, 2006
Committee Members:: James S Ultman, Committee Chair/Co-Chair
Ali Borhan, Committee Chair/Co-Chair
Abdellaziz Ben Jebria, Committee Member
John Michael Cimbala, Committee Member
Keywords:: airway bifurcation
computational fluid dynamics
mass transport
lower respiratory tract
dosimetry modeling
ozone
Abstract:: Ozone (O3), a highly reactive gas, is the major component of photochemical smog and causes a reproducible heterogeneous pattern of lung injury. We hypothesize that this spatial distribution of lung injury mirrors an analogous distribution of O3 dose delivered to different tissue sites in the respiratory tract. The purposes of this study were: 1) to predict the local uptake of O3 using three-dimensional computational fluid dynamics (CFD); and 2) to investigate the effects of flow rate, airway structure and chemical reaction in the respiratory tract lining fluid (RTLF) on O3 uptake. These objectives were achieved by using the commercial CFD package FLUENT (ANSYS Inc., Lebanon, NH) to numerically solve the three-dimensional continuity, Navier-Stokes, and convection-diffusion equations in single idealized airway bifurcations as well as an anatomically accurate airway geometry. To investigate the effect of flow rate on O3 uptake, steady inspiratory and expiratory flow simulations in an idealized symmetric bifurcation with a branching angle of 90° were performed at Reynolds numbers based on the parent branch radius (Re) ranging from 100 to 500. The reaction of O3 in the RTLF was assumed to be so rapid that O3 concentration was negligible along the entire surface of the bifurcation wall. The total rate of O3 uptake was found to increase with increasing flow rate during both inspiration and expiration. Hot spots of O3 flux appeared at the carina of the bifurcation for virtually all inspiratory and expiratory Re considered in the simulations. At the lowest expiratory Re, however, the location of the maximum flux was shifted to the outer wall of the daughter branch. For expiratory flow, additional hot spots of flux were found on the parent branch wall just downstream of the branching region. To investigate the effects of airway structure, steady inspiratory and expiratory flow simulations were conducted in single idealized airway bifurcations at Re = 300, also assuming an infinitely fast reaction in the RTLF. The effect of branching angle was studied in three idealized symmetric single airway bifurcations with branching angles of 45°, 90°, and 135°. The average dimensionless flux < N > (dimensionless uptake rate normalized by dimensionless surface area) increased with increasing branching angle during both inspiratory and expiratory flows. The effect of the mode of branching was studied by comparing simulations in an idealized symmetric 90° bifurcation to those conducted in an idealized monopodial 90° bifurcation. The overall < N > in the symmetric 90° bifurcation was higher than in the monopodial bifurcation during inspiratory and expiratory flow. During inspiratory flow, the minor daughter branch of the monopodial bifurcation had the highest < N > of all the regions in both geometries. Before incorporating a finite RTLF reaction into the CFD model, two RTLF reaction models were compared, both modeling the interaction between O3 and endogenous substrates in the RTLF as quasi-steady lateral diffusion with homogeneous chemical reaction. The first reaction model assumed a reaction rate that was pseudo-first order with respect to of O3 concentration, while the second assumed a reaction rate that was first order with respect to both the O3 3 and substrate concentrations and second order overall. The of O3 concentration profiles within the RTLF as well as the flux of of O3 into the RTLF and tissue were virtually identical in the two models. Because of its greater simplicity, the pseudo-first order reaction model was selected for incorporation into the CFD simulations investigating the effects of RTLF reaction-diffusion parameters on O3 uptake distribution. To study the effect of RTLF thickness on O3 uptake into the RTLF and underlying tissue, the pseudo-first order reaction rate constant was characterized by defining a dimensionless Damkohler number, based on the parent branch radius (DaR). Steady inspiratory flow simulations at Re=300 were carried out at DaR = 1.7x107 and DaR = 1.7x1010 on idealized symmetric 90° bifurcations with RTLF thicknesses ranging from 0.014% to 0.282% of the parent branch radius. The rate of uptake in the RTLF was generally insensitive to the thickness of the RTLF layer, except at the lower DaR at which the RTLF uptake rate increased markedly as the ELF thickness was reduced below about 0.07% of the airway radius. At the lower DaR, the tissue uptake rates were more sensitive to RTLF thickness than were the RTLF uptake rates, but at the higher DaR, no O3 reached the tissue, regardless of RTLF thickness. The effect of the reaction rate constant, characterized by the Damkohler number based on the RTLF thickness (Da), was investigated by simulating flow and O3 uptake in the idealized 90° symmetric and 90° monopodial bifurcations at Re of 10, 100, and also at Re = 1000 in the symmetric bifurcation, for Da ranging from 1 to 106. An increase in led to an increased rate of O3 uptake into the RTLF but a reduced uptake rate in the underlying tissue. The sensitivity of both uptake rates to flow rate became more pronounced as the reaction rate constant was increased. In the final study, two steady inspiratory flow simulations at Re = 152 in an anatomically accurate geometry representative of the first three generations of the airways of a rhesus monkey beginning at the trachea was constructed. Steady inspiratory flow simulations were conducted, one assuming an instantaneous reaction at the bifurcation walls and the other assuming a pseudo first order RTLF reaction at an intermediate Da of 1000. In both simulations, hot spots of flux were located at all three carinas of the geometry, similar to what was found in the idealized single airway bifurcations.

Tools