Convective dispersion during steady flow in the conducting airways of the human lung

The adverse health effects of inhaled particulate matter from the environment depend on its dispersion, transport, and deposition in the human airways. Similarly, precise targeting of deposition sites by pulmonary drug delivery systems also relies on characterizing the dispersion and transport of therapeutic aerosols in the respiratory tract. A variety of mechanisms may contribute to convective dispersion in the lung; simple axial streaming, augmented dispersion, and steady streaming are investigated in this effort. Flow visualization of a bolus during inhalation and exhalation, and dispersion measurements were conducted during steady flow in a three-generational, anatomically accurate in vitro model of the conducting airways to support this goal. Control variables included Reynolds number, flow direction, generation, and branch. Experiments illustrate transport patterns in the lumen cross section and map their relation to dispersion metrics. These results indicate that simple axial streaming, rather than augmented dispersion, is the dominant steady convective dispersion mechanism in symmetric Weibel generations 7-13 during normal respiration. Experimental evidence supports the branching nature of the airways as a possible contributor to steady streaming in the lung.     

Oscillatory flow and gas transport through a symmetrical bifurcation

Axial gas transport due to the interaction between radial mixing and radially nonuniform axial velocities is responsible for gas transport in thick airways during High-frequency oscillatory ventilation (HFO). Because the airways can be characterized by a bifurcating tube network, the secondary flow in the curved portion of a bifurcating tube contributes to cross-stream mixing. In this study the oscillatory flow and concentration fields through a single symmetrical airway bifurcating tube model were numerically analyzed by solving three-dimensional Navier-Stokes and mass concentration equations with the SIMPLER algorithm. The simulation conditions were for a Womersley number, alpha = 9.1 and Reynolds numbers in the parent tube between 200 and 1000, corresponding to Dn2/alpha 4 in the curved portion between 2 and 80, where Dn is Dean number. For comparison with the results from the bifurcating tube, we calculated the velocity and concentration fields for fully developed oscillatory flow through a curved tube with a curvature rate of 1/10, which is identical to the curved portion of the bifurcating tube. For Dn2/alpha 4 < or = 10 in the curved portion of the bifurcating tube, the flow divider and area changes dominate the axial gas transport, because the effective diffusivity is greater than in either a straight or curved tube, in spite of low secondary velocities. However, for Dn2/alpha 4 > or = 20, the gas transport characteristics in a bifurcation are similar to a curved tube because of the significant effect of secondary flow.     

Comparison of steady and pulsatile flow in a double branching arterial model

Data are presented to compare fluid flow parameters for steady flow with those for time-varying flow in a simplified two branch model which simulates the region of the abdominal aorta near the celiac and superior mesenteric branches of the dog. Measurements in the model included laser doppler anemometry velocity profiles during steady flow, sinusoidal flow with a superimposed mean flow (referred to as simple oscillatory flow) and arterial pulsatile flow. Shear rate measurements were made by an electrochemical technique during steady flow. Flow visualization studies were done during steady and pulsatile flow. Fluid flow effects in the simplified model during steady flow showed many similarities to the results from previous steady flow studies in a canine aortic cast. Shear rates in the region of the proximal (first, or celiac) branch were independent of flow rates in the distal (second, or mesenteric) branch, but the shear pattern within the proximal branch changed significantly as flow in the proximal branch increased. Shear rates on the proximal flow divider (leading edge into the distal branch) depended primarily on the flow rate to the proximal branch, but not on flow to the distal branch. At certain daughter branch flow ratios (approximately 2:1, proximal to distal), flow separation was promoted at the outer wall of the second branch, but flow separation did not occur in the first branch. In contrast to the canine aortic case results, flow separation was never detected on the distal (mesenteric) flow divider of the simplified model. This observation reflects the subtle effects of geometry on flow since the mesenteric flow divider in the canine cast protrudes into the main flow whereas the distal flow divider in the simplified model does not. There were distinct differences in the flow phenomena between steady, simple oscillatory and arterial pulsatile flow. Peak shear rates during pulsatile flow were as much as 10--100 times greater than steady flow shear rates at comparable mean flow rates. Particularly noteworthy for the pulsatile flow with a Womersley parameter of sixteen were very blunt velocity profiles throughout systole, and the absence of flow separation or reversal in those regions of the model that exhibited flow separation during steady flow. The shape of the waveform influences the nature of the flow during time-varying flows. Future studies of fluid dynamics in model systems must consider the pulsatile nature of the flow if a true interpretation of arterial flow phenomena is to be made.     

Gas dispersion in volume-cycled tube flow. I. Theory.

A mathematical theory is derived for the dispersion of a contaminant bolus introduced into a fully developed volume-cycled oscillatory pipe flow. The convection-diffusion equation is solved for a tracer gas bolus by expressing the local concentration field as a series expansion of derivatives of the area-averaged concentration. The local, as well as the area-averaged, concentration is determined for a uniform initial slug or Gaussian bolus. The effect of various flow parameters such as Womersley parameter, Schmidt number, and tidal volume is investigated. The overall dispersion is characterized by a time-averaged effective diffusion coefficient, which for long duration coincides with previous dispersion theories based on a constant linear axial concentration profile. The effective diffusion coefficient can be determined from the local time history of concentration, independent of the spatial location or the initial tracer bolus. Furthermore the local peaks of the concentration-time curve follow a decaying curve dictated by the time-averaged effective diffusion coefficient. Thus the theory is directly applicable for dispersion measurements in oscillatory tube flows, a basis for the pulmonary airways application, as shown by Gaver et al. (J. Appl. Physiol. 72: 321-331, 1992).     

Numerical investigation of the three-dimensional flow in a human lung model

The flow field at inspiration and expiration in the upper human airways consisting of the trachea down to the sixth generation of the bronchial tree is numerically simulated. The three-dimensional steady flow at a hydraulic diameter-based Reynolds number Re(D)=1250 is computed via a lattice-Boltzmann method (LBM). The simulation is validated by the experimental data based on particle-image velocimetry (PIV) measurements. The good agreement between numerical and experimental results is evidenced by comparing velocity contours and distributions in a defined reference plane. The results show the LBM to be an accurate tool to numerically predict flow structures in the human lung. Using an automatic Cartesian grid generator, the overall process time from meshing to a steady-state solution is <12h. Moreover, the numerical simulation allows a closer analysis of the secondary flow structures than in the experimental investigation. The three-dimensional streamline patterns reveal some insight on the air exchange mechanism at inspiration and expiration. At inspiration, the slower near-wall tracheal flow enters through the right principal bronchus into the right upper lobar bronchus. The bulk mass flux in the trachea is nearly evenly distributed over the left upper, center and lower lobar bronchi and the right center and lower bronchi. At expiration, the air from the right upper lobar bronchus enters the right center of the trachea and displaces the airflow from the lower and center right bronchi such that the tracheal positions of the streamlines at inspiration and expiration are switched. The flow in the left bronchi does not show this kind of switching. The findings emphasize the impact of the asymmetry of the lung geometry on the respiratory air exchange mechanism.     

Spatial and temporal variation of secondary flow during oscillatory flow in model human central airways

Axial and secondary velocity profiles were measured in a model human central airway to clarify the oscillatory flow structure during high-frequency oscillation. We used a rigid model of human airways consisting of asymmetrical bifurcations up to third generation. Velocities in each branch of the bifurcations were measured by two-color laser-Doppler velocimeter. The secondary velocity magnitudes and the deflection of axial velocity were dependent not only on the branching angle and curvature ratio of each bifurcation, but also strongly depended on the shape of the path generated by the cascade of branches. Secondary flow velocities were higher in the left bronchus than in the right bronchus. This spatial variation of secondary flow was well correlated with differing gas transport rates between the left and right main bronchus.     

Measurement of axial diffusivities in a model of the bronchial airways.

Values for the effective axial diffusivity D for laminar flow of a gas species in the bronchial airways have been obtained as a function of the mean axial gas velocity u by experiment measurements of benzene vapor dispersion in a five generation glass tube model of the bronchial tree. For both inspiration and expiration D is seen to be approximately a linear function of u over the range of Reynolds' numbers 30-2,000 corresponding to peak flows in bronchial generations 0-13 under resting breathing conditions. The diffusivity for expiration is seen to be approximately one-third that for inspiration due presumably to increased radial mixing at bifurcations during expiration. The effective diffusivities relative to the molecular diffusivity can be expressed by the formulas D/Dmol = 1 + 1.08 NPe for inspiration and D/Dmol = 1 + .37 N-Pe for expiration. These velocity dependent diffusivities help to explain the short transit times of gas boluses from mouth to alveoli and will aid in the analysis of airway gas mixing by mathematical transport equations.     

Oscillatory flow in a cone-and-plate bioreactor

Motivated by biometric applications, we analyze oscillatory flow in a cone-and-plate geometry. The cone is rotated in a simple harmonic way on a stationary plate. Based on assuming that the angle between the cone and plate is small, we describe the flow analytically by a perturbation method in terms of two small parameters, the Womersley number and the Reynolds number, which account for the influences of the local acceleration and centripetal force, respectively. Working equations for the shear stresses induced both by laminar primary and secondary flows on the plate surface are presented.     

Pressure drop and flow rate measurements in a human aortic bifurcation cast for steady and pulsatile flow

Pressure drop and flow rate measurements in a rigid cast of a human aortic bifurcation under both steady and physiological pulsatile flow conditions are reported. Integral momentum and mechanical energy balances are used to calculate impedance, spatially averaged wall shear stress and viscous dissipation rate from the data. In the daughter branches, steady flow impedance is within 30% of the Poiseuille flow prediction, while pulsatile flow impedance is within a factor of 2 of fully developed, oscillatory, straight tube flow theory (Womersley theory). Estimates of wall shear stress are in accord with measurements obtained from velocity profiles. Mean pressure drop and viscous dissipation rate are elevated in pulsatile flow relative to steady flow at the mean flow rate, and the exponents of their Reynolds number dependence are in accord with available theory.     

A mechanical approach to the longitudinal dispersion of gas flowing in human airways

The aim of this work is to contribute to elucidating the mechanism underlying gas mixing in the human pulmonary airways. For this purpose, a particular attempt is made to analyse the fluid mechanical aspects of gaseous dispersion using bolus response methods. The experiments were performed on five normal subjects by injection of 10 cm3 bolus of He, Ar and SF6 into the latter part of the inspired airstream, in such a way that the whole bolus entered the inspiratory flow and was recovered during the following expiration. The results, presented in a logarithmic plot of dimensionless variance (dispersion of the output bolus) against the Peclet number, show that gaseous dispersion is only slightly dependent on the nature of the tracer gas but is strongly related to flow velocity. This is in agreement with the theory of turbulent or disturbed dispersion; however, it seems that Taylor laminar dispersion does not play a significant role in the airways.     

Pulsatile blood flow and oxygen transport past a circular cylinder

The fundamental study of blood flow past a circular cylinder filled with an oxygen source is investigated as a building block for an artificial lung. The Casson constitutive equation is used to describe the shear-thinning and yield stress properties of blood. The presence of hemoglobin is also considered. Far from the cylinder, a pulsatile blood flow in the x direction is prescribed, represented by a time periodic (sinusoidal) component superimposed on a steady velocity. The dimensionless parameters of interest for the characterization of the flow and transport are the steady Reynolds number (Re), Womersley parameter (alpha), pulsation amplitude (A), and the Schmidt number (Sc). The Hill equation is used to describe the saturation curve of hemoglobin with oxygen. Two different feed-gas mixtures were considered: pure O(2) and air. The flow and concentration fields were computed for Re=5, 10, and 40, 0< or =A< or =0.75, alpha=0.25, 0.4, and Schmidt number, Sc=1000. The Casson fluid properties result in reduced recirculations (when present) downstream of the cylinder as compared to a Newtonian fluid. These vortices oscillate in size and strength as A and alpha are varied. Hemoglobin enhances mass transport and is especially important for an air feed which is dominated by oxyhemoglobin dispersion near the cylinder. For a pure O(2) feed, oxygen transport in the plasma dominates near the cylinder. Maximum oxygen transport is achieved by operating near steady flow (small A) for both feed-gas mixtures. The time averaged Sherwood number, Sh, is found to be largely influenced by the steady Reynolds number, increasing as Re increases and decreasing with A. Little change is observed with varying alpha for the ranges investigated. The effect of pulsatility on Sh is greater at larger Re. Increasing Re aids transport, but yields a higher cylinder drag force and shear stresses on the cylinder surface which are potentially undesirable.     

Simulation of the effects of mechanical nonhomogeneities on expiratory flow from human lungs

A computational model for expiration from lungs with mechanical nonhomogeneities was used to investigate the effect of such nonhomogeneities on the distribution of expiratory flow and the development of alveolar pressure differences between regions. The nonhomogeneities used were a modest constriction of the peripheral airways and a 50% difference in compliance between regions. The model contains only two mechanically different regions but allows these to be as grossly distributed as left lung-right lung or to be distributed as a set of identical pairs of parallel nonhomogeneous regions with flows from each merging in a specified bronchial generation. The site of flow merging had no effect on the flow-volume curve but had a significant effect on the development of alveolar pressure differences (delta PA). With the peripheral constriction, greater values of delta PA developed when flows were merged peripherally rather than centrally. The opposite was true in the case of a compliance nonhomogeneity. The delta PA values were smaller at submaximal flows. Plots of delta PA vs. lung volume were similar to those obtained experimentally. These results were interpreted in terms of the expression used for the fluid mechanics of the merging flows. delta PA was greater when the viscosity of the expired gas was increased or when its density was reduced. Partial forced expirations were shown to indicate the presence of mechanical nonhomogeneity.     

Effects of curved inlet tubes on air flow and particle deposition in bifurcating lung models

In vivo bifurcating airways are complex and the airway segments leading to the bifurcations are not always straight, but curved to various degrees. How do such curved inlet tubes influence the motion as well as local deposition and hence the biological responses of inhaled particulate matter in lung airways? In this paper steady laminar dilute suspension flows of micron-particles are simulated in realistic double bifurcations with curved inlet tubes, i.e., 0 degrees < or =theta< or =90 degrees, using a commercial finite-volume code with user-enhanced programs. The resulting air-flow patterns as well as particle transport and wall depositions were analyzed for different flow inlet conditions, i.e., uniform and parabolic velocity profiles, and geometric configurations. The curved inlet segments have quite pronounced effects on air-flow, particle motion and wall deposition in the downstream bifurcating airways. In contrast to straight double bifurcations, those with bent parent tubes also exhibit irregular variations in particle deposition efficiencies as a function of Stokes number and Reynolds number. There are fewer particles deposited at mildly curved inlet segments, but the particle deposition efficiencies at the downstream sequential bifurcations vary much when compared to those with straight inlets. Under certain flow conditions in sharply curved lung airways, relatively high, localized particle depositions may take place. The findings provide necessary information for toxicologic or therapeutic impact assessments and for global lung dosimetry models of inhaled particulate matter.     

Model of forced expiratory flows and airway geometry in infants.

Early measurements of autopsied lungs from infants, children, and adults suggested that the ratio of peripheral to central airway resistance was higher in infants than older children and adults. Recent measurements of forced expiration suggest that infants have high flows relative to lung volume. We employed a computational model of forced expiratory flow along with physiological and anatomic data to evaluate whether the infant lung is a uniformly scaled-down version of the adult lung. First, we uniformly scaled an existing computational model of adult forced expiration to estimate forced expiratory flows (FEF) and density dependence for an 18-mo-old infant. The values obtained for FEF and density dependence were significantly lower than those reported for healthy 18-mo-old infants. Next, we modified the model for the infant lung to reproduce standard indexes of expiratory flow [forced expiratory volume in 0.5 s (FEV(0.5)), FEFs after exhalation of 50 and 75% forced vital capacity, FEF between 25 and 75% expired volume] for this age group. The airway sizes obtained for the infant lung model that produced accurate physiological measurements were similar to anatomic data available for this age and larger than those in the scaled model. Our findings indicate that the airways in the infant lung model differ from those in the scaled model, i.e., middle and peripheral airway sizes are larger than result from uniform downscaling of the adult lung model. We show that the infant lung model can be made to reproduce individual flow-volume curves by adjusting lumen area generation by generation.     

Some flow visualization and laser-Doppler-velocity measurements in a true-to-scale elastic model of a human aortic arch--a new model technique.

Flow studies were done in an elastic true-to-scale silicone rubber model of an aortic arch to study further hemodynamic influences on atherosclerosis. The model was prepared from a cast of a young woman. A revised model technique was used. The model had a compliance similar to that of the human aortic arch. Velocity measurements were done in the model with a two component laser-Doppler-anemometer in steady and pulsatile flow using a calcium chloride solution with a viscosity of eta = 3.18 mPas and density of rho = 1.28 kg/m3 at 20 degrees C. The time average Reynolds numbers over a whole cycle in the ascending aorta was Re = 1350. The Womersley parameter for pulsatile flow was a = 20. The pulse wave velocity in the ascending aorta was about c = 5.4 m/sec. The secondary flow behavior was discussed for steady and pulsatile flow. Reverse flows were found, especially along the inner radius of the aortic arch in the descending aorta in steady and pulsatile flow and also in small areas of the ascending aorta and at the branches of the aortic arch. The formation of atherosclerotic plaques at preferred local flow regions is discussed.     

Numerical simulations of pulsatile blood flow using a new constitutive model

In the present paper we use a new constitutive equation for whole human blood [R.G. Owens, A new microstructure-based constitutive model for human blood, J. Non-Newtonian Fluid Mech. (2006), to appear] to investigate the steady, oscillatory and pulsatile flow of blood in a straight, rigid walled tube at modest Womersley numbers. Comparisons are made with the experimental results of Thurston [Elastic effects in pulsatile blood flow, Microvasc. Res. 9 (1975), 145-157] for the pressure drop per unit length against volume flow rate and oscillatory flow rate amplitude. Agreement in all cases is very good. In the presentation of the numerical and experimental results we discuss the microstructural changes in the blood that account for its rheological behaviour in this simple class of flows. In this context, the concept of an apparent complex viscosity proves to be useful.     

Bolus contaminant dispersion in oscillating flow in curved tubes.

The investigation of longitudinal dispersion of tracer substances in unsteady flows has biomechanical application in the study of heat and mass transport within the bronchial airways during normal, abnormal, and artificial pulmonary ventilation. To model the effects of airway curvature on intrapulmonary gas transport, we have measured local gas dispersion in axially uniform helical tubes of slight pitch during volume-cycled oscillatory flow. Following a small argon bolus injection into the flow field, the time-averaged effective diffusion coefficient (Deff/Dmol) for axial transport of the contaminant was evaluated from the time-dependent local argon concentration measured with a mass spectrometer. The value of (Deff/Dmol) is extracted from the curve of concentration versus time by two techniques yielding identical results. Experiments were conducted in two helical coiled tubes (delta = 0.031, lambda = 0.022 or delta = 0.085, lambda = 0.060) over a range of 2 < alpha < 15, 3 < A < 15, where delta is the ratio of tube radius to radius of curvature, lambda is the ratio of pitch height to radius of curvature, alpha is the Womersley parameter or dimensionless frequency, and A is the stroke amplitude or dimensionless tidal volume. Experimental results show that, when compared to transport in straight tubes, the effective diffusivity markedly increases in the presence of axial curvature. Results also compare favorably to mathematical predictions of bolus dispersion in a curved tube over the ranges of frequency and tidal volume studied.     

Convective exchange between the nose and the atmosphere

It is generally accepted that there is little rebreathing of gas exhaled through the nose. A detailed physical model system has been used to quantify and identify the mechanisms responsible for this phenomenon. By the use of a cast of the upper respiratory tract and oscillating flows with a Reynolds number of 500 and nondimensional frequency of 1.6, corresponding to quiet tidal breathing through the nose, dye dilution measurements indicated an efficiency of tidal exchange of 0.95. Flow visualization studies performed to trace the expiratory flow, as well as the streamlines during steady inspiratory flow, support the hypothesis that the high efficiency of exchange is due to radical differences in the velocity fields between inspiratory and expiratory phases of this oscillatory flow. These findings confirm that convective gas exchange between the nose and the atmosphere is highly efficient; however, the underlying mechanism responsible for this exchange also maximizes the exposure of the respiratory system to aerosols contained in the ambient atmosphere.