Journal of Molecular Liquids 209 (2015) 121–133

Contents lists available at ScienceDirect

Journal of Molecular Liquids

journal homepage: www.elsevier.com/locate/molliq

CFD simulation of airflow behavior and particle transport and deposition

in different breathing conditions through the realistic model of
human airways
M. Rahimi-Gorji ⁎, O. Pourmehran, M. Gorji-Bandpy, T.B. Gorji
Babol Noshirvani University of Technology, Mechanical Engineering Department, Babol, Mazandaran, Iran

a r t i c l e i n f o a b s t r a c t

Article history: In this work, the airflow behavior and particle transport and deposition in different breathing conditions such as
Received 17 March 2015 light breathing condition (15 L/min), normal breathing condition (30 L/min) and heavy breathing condition
Received in revised form 13 May 2015 (60 L/min) are investigated. The realistic geometry data was reconstructed from CT-scan images of the human
Accepted 15 May 2015
airways with 0.5 mm thickness of slices. The CT-scan images (DICOM files) are imported in the 3D-DOCTOR
Available online 28 May 2015
software and all slices were segmented. Then, the output has been imported in CATIA-V5 software. Finally,
Keywords:
face, volume, mesh and extension tubes at inlet and outlets were created and then imported into ANSYS FLUENT
Airflow behavior 15. The Lagrangian approach is used to evaluating the transport and deposition of inhaled micro-particles. The
Particle deposition presented results showed that for dp = 5 μm and 10μm, when flow rate ¼ 30 min L
and for dp = 1 μm, when
Breathing conditions flow rate ¼ 15 min L
, the particle deposition fraction have maximum amount. For flow rate ¼ 15 min L
and ¼
Realistic geometry L
30 min , the maximum deposition occurs in the zone number 1 and for flow rate ¼ 60 min L
occurs in the zone
CFD simulation number 4. Also, the maximum pressure distribution happens when flow rate ¼ 60 min L
which would be accept-
able. According to the results, the particles tended to go to the right branch and the minimum number of particles
crossed the zone numbers 6 and 11.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction
It is common for people to behold particles in the air and sense
air entering into human bodies through breathing. Human lungs are
one of the body's important and largest organs. Their function is gas
exchange, delivering oxygen and removing wasted carbon dioxide.
Air goes into the body through nostrils and mouth into trachea, which
is then divided into left and right primary bronchi. The bronchus branch
out into secondary bronchi and then each subdivides into tertiary
bronchi and so on, reducing diameter throughout the structure until
the respiratory bronchioles which scatter air to the alveoli. Today,
drug delivery to the specified location is concerned. In fact, human
health by controlling particles that enter human body is considered.
Aerosol drug therapy, which mainly delivers the drug location of inter-
est, is a quickly advancing field of research [1]. With the recent advances
in analysis of micro- and nanoparticles, drug delivery has indicated
great potential for pulmonary application, not only for local therapy
but for systemic therapy as well.
⁎ Corresponding author at: Babol Noshirvani University of Technology, P.O. Box 484,
Babol, Iran.
E-mail addresses: m69.rahimi@yahoo.com (M. Rahimi-Gorji), gorji@nit.ac.ir
(M. Gorji-Bandpy).
Inhaling through the mouth, therapeutic drugs have their site of
action in the large and/or small conducting airways. Inhalers designed
for the delivery of therapeutic factors maximize the respiratory deposi-
tion as inhalable aerosol particles. Although, particles may deposit in
another place in the respiratory tract, mainly in the upper respiratory
tract or peripheral airspaces resulting in undesirable side effects and
waste. Deposition of these particles in the lung periphery furthermore
is being futile in this region. The amount and location of particle deposi-
tion in the respiratory tract depends on both the particle size and the
human's flow patterns during inhalation [2,3].
Computational Fluid Dynamics (CFD) has been applied to character-
ize the fluid flow in human airway models. CFD has obtained significant
interest in both the medical and engineering community because of its
non-invasive character. It can predict the fluid flow characteristics when
one or multiple input flow variables are changed. In addition, it permits
investigation of different flow variables and fluid forces to a level of fine
detail.
Simple geometric configuration led to CFD simulations and gave a
first vision into particle deposition and flow patterns in the airways
[4–6]. Bala'sha'zy [7] investigated a wide range of submicron and
micron-size particles in bifurcating tubes numerically for computing
trajectories. The fluid flow and particle deposition patterns in an
asymmetric single bifurcation are simulated by Gatlin et al. [8].

http://dx.doi.org/10.1016/j.molliq.2015.05.031
0167-7322/© 2015 Elsevier B.V. All rights reserved.
122 M. Rahimi-Gorji et al. / Journal of Molecular Liquids 209 (2015) 121–133

volume fraction (PVF). For PVF b 10−6, particle motion is affected by

Nomenclature
continuous phase properties while basically there is no feedback from
the dispersed phase. This class is known as one-way coupling. For
x x coordinate
10−6 b PVF b 10−3, feedback of the dispersed phase on the properties
y y coordinate
of the continuous phase fluid dynamics must also be taken into account,
z z coordinate
this class is called two-way coupling. When PVF N 10−3, a dense flow is
ui mean velocity in tensor notation (m/s)
characterized that this class is known as four-way coupling and it
ug fluid (air) velocity (m/s)
should be considered particle–particle interactions [31,32].
up particle velocity (m/s)
In this work, the realistic geometry data was reconstructed from CT-
Rep particle Reynolds number
scan images of the human airways. It can be illation from the mentioned
ρ fluid density (kg/m3)
references that there is almost no literature available in which realistic
ρp particle density (kg/m3)
human respiratory tract model was applied for the study of the effects
dp particle diameter (m)
of the different breathing conditions on airflow behavior and particle
CD Drag coefficient
transport and deposition. The main aim of this research is investigating
DE deposition efficiency
the airflow behavior and particle deposition fraction in different breathing
p mean static pressure (pa)
conditions such as light breathing condition (15 L/min), normal breathing
condition (30 L/min) and heavy breathing condition (60 L/min). Accord-
Greek symbols
ing to the results, by realizing the effects of different breathing conditions
υ kinetic molecular viscosity (μ/ρ)
on particle deposition fraction, a deep understanding of air-particle
υT kinetic eddy viscosity (μT/ρ)
dynamics in the tracheobronchial airway is obtained and it can be very
τij Reynolds stress tensor
useful for medication and drug delivery experts and committee.
μ kinetic viscosity
α1, α2, α3 drag constants

Although, Farag et al. [9,10], utilized a similar geometric configura-

tion, experimentally behold flow separation in the bifurcation re-
gion under similar flow conditions. Thus bifurcation models that
mentioned previously may not prepare a realistic formation of de-
position in respiratory airways. Developed models have been either
symmetric [11,12] or asymmetric [13]. However, these general
models do not include the irregularities that every lung could
have. Some researchers [14–16] have concentrated on patient-
specific lung geometries and their results are more relevant to the
clinical results. Longest et al. lately demonstrated a CFD-based sto-
chastic individual path modeling moves towards to guess the deliv-
ery of pharmaceutical aerosols all over the tracheobronchial airways
[17,18]. Some new characteristics of the airflow and particle deposi-
tion in complex, multi-generation healthy and diseased airways
were reported such as Ref [19].
An important part of this research is considering a realistic geometry
model of the human airways. Most previously proposed models in-
volved a simplified geometry model such as Wieble model [20] or
Horsfield model [21] etc. This missing information related to the
three-dimensional (3D) disposition of branches had to be detected to
construct a realistic model of the bronchial tree. Recent studies have
used airway models that were reconstructed from magnetic resonance
(MR) or computed tomography (CT) imaging data [22,23]. Fleming
et al. [24] have applied 3-D radionuclide imaging with CT scanning of
the airways subject to better report the deposition locations of aerosols
in the respiratory system. Golshahi et al. [25] obtained interesting re-
sults with measurements of deposition of micrometer-sized particles
(0.5–5.3 mm) in replicas of older children (4–14 years old) to develop
correlations including geometric dimensions of airways and breathing
patterns as a predictive tool for future potential improvements in envi-
ronmental and pharmaceutical standards.
There are fundamentally two various approaches in the analysis of
the phenomenon of particles dispersed in the air flow in the respiratory
system such as Euler–Lagrange method and Euler–Euler method [26]. In
the Euler–Lagrange method, a particle trajectory is computed by solving
equations of the motion for each particle in the Lagrangian approach
[27,28]. On the other side, in the Euler–Euler method, a particle con-
centration distribution of the bearer fluid is calculated [29,30]. General-
ly, interactions between particles can be classified, based on particle Fig. 1. Flowchart of research process.
 M. Rahimi-Gorji et al. / Journal of Molecular Liquids 209 (2015) 121–133 123

Fig. 2. The overarching theme of research.

Fig. 3. CT-scan images of present study. Red circles indicate locations of central airway stents.
124 M. Rahimi-Gorji et al. / Journal of Molecular Liquids 209 (2015) 121–133

Fig. 4. (a) Three dimensional reconstructed geometry from CT-scan images. (b) Realistic 3D airway model (extended inlet and outlets with CATIA-V5 software). (c) Divided zones of
geometry.

Fig. 6. Comparison of the particle deposition fraction between present work's results and
Fig. 5. 3D volume mesh generation. Nowak et al.'s work [12].
 M. Rahimi-Gorji et al. / Journal of Molecular Liquids 209 (2015) 121–133 125

Fig. 8. Total deposition fraction of different zones for different flow rate with dp = 5 μm.

2. Methodology

2.1. Three dimensional reconstruction of human respiratory system from

CT-scan images

The geometry of the human airway was obtained from computed

tomography (CT) scan of a healthy 63 years old, non-smoking male.
The CT-scan was started from the inlet of trachea and extended to sec-
ond generation. The slice thicknesses of CT-scan images were 0.5 mm
that the total 373 slices were considered for construction from the
inlet of trachea to second generation (G0–G2). The CT-scan images
(DICOM files) are imported in the 3D-DOCTOR software which is a
strong software in field of image processing. Then, all slices segmented
and exported the constructed model to .STL (Standard Tessellation
Language) format. The output is prepared for importing in CATIA-V5
software for converting the .STL format to CATPART format, then has
been converted to IGES format. Finally, face, volume, mesh and exten-
sion tubes at inlet and outlets were created using ANSYS-Workbench
15 and a mesh file was produced, which was then read into ANSYS
FLUENT 15. Figs. 1 and 2 show a flowchart for CFD simulations of
human airways and the overarching theme of research respectively.

Fig. 7. 3D view bar for deposition fraction of different zones for three particle diameter
with (a) flow rate ¼ 15 min
L
, (b) flow rate ¼ 30 minL
, and flow rate ¼ 60 min
L
.

Fig. 9. Total deposition fraction of different flow rate with different particle diameter.
126 M. Rahimi-Gorji et al. / Journal of Molecular Liquids 209 (2015) 121–133

2.2. Construction of the computational model Momentum equations:

Fig. 3 depicts four slices of CT-scan images of present study which

" !#
red circles indicate locations of central airway stents. The reconstructed ∂ui ∂u 1 ∂p ∂ ∂ui ∂u j
model of the human airway is displayed in Fig. (4-a) which including þ uj i ¼ − þ ðυ þ υT Þ þ ð2Þ
∂t ∂x j ρ ∂xi ∂x j ∂x j ∂xi
from the trachea down to the G2. To prevent any reverse flow, the
inlet and outlets were extended upstream and downstream respective-
ly. Fig. (4-b) and (c) show the realistic 3D model for the total volume of ui and uj (i, j = 1, 2, 3) are the velocity components in x, y and z
the airway (extended inlet and outlets with CATIA-V5 software) and directions.
divided zones of geometry, respectively. As realistic human respiratory
tract are irregular in shape, an unstructured tri/tetrahedral hybrid In this study, different flow rate are assumed for inspiratory pattern.
volume mesh was created inside the airway model that is depicted In fact, three conditions of breathing are considered such light breathing
in Fig. 5. The cell quantity in the computational model is about condition (15 L/min), normal breathing condition (30 L/min) and heavy
1,800,000 that obtained after the grid independency test for four breathing condition (60 L/min). Boundary conditions for fluid flow in
different grid sizes (about 790,000, 12,500,000, 1,800,000, and this study are mass flow rate for inlet, pressure outlet for outlets and
2,100,000 cells). It was observed that the computed results become no slip condition for walls.
almost independent of the grid size beyond 1,800,000 cells. After The problem was calculated steady using k–ω SST turbulence model.
meshing, the CFD software package Fluent (ANSYS-FLUENT 15) was ap- For the pressure–velocity coupling, the SIMPLEC algorithm was used.
plied to solve the flow governing equations with finite volume method Different terms in the transport equations were discretized using the
(FVM) on an arbitrarily shaped flow area with suitable boundary condi- second-order upwind numerical scheme. The governing equations of
tions. The steady-state solution of the flow field was assumed to have k–ω turbulence model are written as follows:
converged when the residuals decreased to less than 10− 6. Air was Turbulent kinetic energy (k) equation:
considered incompressible and Newtonian fluid with constant density
ρ, viscosity μ and fluid static pressure p. The governing equations for
the airflow in the respiratory tract are given as follows [33]:
" #
Continuity equation: ∂k ∂k ∂u ∂ ∂k
þ uj ¼ τi j i þ ðυ þ σ k υT Þ ð3Þ
∂t ∂x j ∂x j ∂x j ∂x j

∂ui
¼ 0; ð1Þ
∂xi

Fig. 10. Cross sectional views of the velocity magnitude contour of airways for dp = 5 μm
L L L L L L
with (a) 15 min , (b) 30 min and (c) 60 min. Fig. 11. Contours of pressure distribution for (a) 15 min, (b) 30 min and (c) 60 min.
 M. Rahimi-Gorji et al. / Journal of Molecular Liquids 209 (2015) 121–133 127

Pseudo-vorticity (x) equation: on the convective/radioactive heat and mass transfer from the particle,
" # using the local continuous phase conditions as the particle moves
∂ω ∂ω ω ∂u ∂ ∂ω through the flow. Numerous applications using DPM model are particle
þ uj ¼ α τ i j i −βω2 þ ðυ þ σ ω υT Þ : ð4Þ
∂t ∂x j k ∂x j ∂x j ∂x j separation and classification, spray drying, aerosol dispersion, bubble
sparging of liquids, liquid fuel and coal combustion. The particle trajec-
tory was computed through the equation of the balance of forces acting
Where the turbulent viscosity is given as υT = Cμfμk/ω, and the func- on that particle. The equation describing the particle velocity, in the
tion fμ is defined as fμ = exp[−3.4/(1 + RT/50)2] with RT = ρk/(μω). The Lagrange formulation, for a Cartesian coordinate system has the form
other coefficients in the above equations are:
 
 ∂up   g x ρp −ρ
C μ ¼ 0:09; α ¼ 0:555; β ¼ 0:8333; β ¼ 1; σ k ¼ σ ω ¼ 0:5: ¼ F D u−up þ ; ð5Þ
∂t ρp

2.3. Discrete phase model (DPM) where FD(u − up) is the drag force per unit particle mass (acceleration
due to drag) and
The discrete phase model (DPM) can be included in the ANSYS
18μ C D Rer
FLUENT model by defining the initial position, velocity, size and temper- FD ¼ 2 24
: ð6Þ
ature of individual particles. These initial conditions along with the ρp dp
physical properties of the discrete phase are utilized to initiate trajecto-
ry and heat/mass transfer calculations. The trajectory and heat/mass Here, u is the fluid velocity, up is the particle velocity, μ is the dynamic
transfer calculations are based on the force balance on the particle and viscosity of the fluid, ρ is the fluid density, ρp is the density of the particle

L L L
Fig. 12. Streamlines of airflow in airways for (a) 15 min, (b) 30 min and (c) 60 min .
128 M. Rahimi-Gorji et al. / Journal of Molecular Liquids 209 (2015) 121–133

and dp is the particle diameter. Rer is the relative Reynolds number, which iteration for each particle, the information about position, time, and
is defined as three components of velocity as well as the speed with which the parti-
  cles cross the control volume boundaries was obtained. Particles of spe-
ρdp up −u cific sizes were injected in a uniform distribution at the inlet face of the
Rer ¼ : ð7Þ
μ computation domain, and tracked through the geometry until they met
one of three fates: (1) trapping on a surface by collision, (2) escape from
The drag coefficient, CD is obtained using the following formula: the domain through one of the outlet faces, or (3) continued suspension
in the flow. The fate of the particles were then recorded and summa-
a2 a3 rized as a particle history file. One-way coupling is assumed between
C D ¼ a1 þ þ ; ð8Þ
Rer Re2r the air and particle flow fields and the interaction between particles is
also neglected because the particle flow is dilute [31,32]. In this work,
where ai are constants that employ to smooth spherical particles over the particle characteristics include:
several ranges of Re given by Morsi and Alexander [34].
Eq. (3) is the particle trajectory equation. It has been solved by the
kg
same commercial software. ANSYS-FLUENT 15 guesses the trajectory • Particle density: ρp ¼ 1000 m3 , spherical particles.

• Mass flow rate: m¼ 4:2097  10−11 kg −9 kg



of discrete phase particles by integrating the force balance on each s , 5:2622  10 s and 4:2097 
−8 kg
particle, which is written in a Lagrangian reference frame. After each 10 s.

  
Fig. 13. Particles situation at t = 0.01 s after start injection with dp = 5 μm for (a) m¼ 15 min
L
, (b) m¼ 30 min
L
and (c) m¼ 60 min
L
.
 M. Rahimi-Gorji et al. / Journal of Molecular Liquids 209 (2015) 121–133 129

• Particle diameter (dp): 1, 5, and 10 μm. conditions such as light breathing condition (15 L
min), normal breath-
• Velocity: the same as inlet velocity. ing condition (30 L L
min)
and heavy breathing condition (60 min ). As
early mentioned, one of the most important features of this research
kg is using realistic geometry that reconstructed from CT-scan
Also, the air properties are assumed constant (ρair ¼ 1:2 m3 andμ air ¼
images. Fig. 6 shows the comparison of the particle deposition frac-
1:8  10−5 m
Ns
2) and the flow regime is consider steady and turbulent. The tion between present work and Nowak et al. [12] work results. They
deposition fraction (DF) is defined as studied particle deposition in a geometry based on the Weibel lung
model. In their steady simulations, the used flow rate was 28:3 min L
. It
Number of deposited particles
DFð%Þ ¼  100: ð9Þ is obvious that the results of present work have good agreement
Number of injected particles
with Nowak et al.'s results and little difference between these are
due to the different flow rate (30 and 28.3 L/min).
3. Results and discussion Fig. 7 depicts the deposition fraction in different zones for flow
rate ¼ 15 min
L L
(7-a), 30 min L
(7-b) and 60 min (7-c). As can be seen in this
The main aim of this work is investigating the airflow behavior and figure, independent of the flow rate, the maximum deposition happens
particle transport and deposition fraction in different breathing for dp = 10 μm. For flow rate ¼ 15 L
min
L
and 30 min , the maximum

  
Fig. 14. Particles situation at t = 0.03 s after start injection with dp = 5 μm for (a) m¼ 15 min
L
, (b) m¼ 30 min
L
and (c) m¼ 60 min
L
.
130 M. Rahimi-Gorji et al. / Journal of Molecular Liquids 209 (2015) 121–133

deposition occurs in the zone number 1 and for flow rate ¼ 60 min L
, it oc-
curs in the zone number 4. For various flow rate such as 15 , 30 and
L Fig. 10 shows the cross sectional views of the velocity magnitude
60 min , the maximum deposition fraction of them are 6.5 %, 26.4 % and
7.94 % respectively (dp = 10 μm). The location of minimum deposition
was not precisely predictable and didn't follow a specific pattern. It is
obvious, for flow rate ¼ 60 min L
and dp = 1μm, deposition fraction is
zero due to large value of flow velocity and the small value of particle
diameter. Also according to this figure, when dp = 1 μm, the maximum
deposition fraction is for flow rate ¼ 15 min
L
in location of zone number 4
and when dp = 5μm and 10μm, the maximum deposition fraction is for
flow rate ¼ 30 minL
in location of zone number 1. Fig. 8 depicts the total
deposition fraction of different zones for different flow rate with dp =
5 μm that clarified the illustrations of Fig. 7. Next, Fig. 9 demonstrates
the total deposition fraction of different flow rate with different particle
diameters by graph. As can be seen in this figure, all of the explanations
of early figures (Figs. 7 and 8) could be corrected and acceptable. In fact,
this figure illustrated that for dp = 5 μm and 10 μm, when flow rate ¼
L
30 min and for dp = 1 μm, when flow rate ¼ 15 minL
, the particle deposi-
tion fraction is maximum.
contour of airways for flow rate ¼ 15 min L
(10-a), 30 minL
(10-b) and
L
60 min (10-c). Because of the realistic geometry, cross sectional of differ-
ent parts of the airways aren't constant and as a result, some parts are
narrower. For this reason, the velocity magnitude in smaller cross-
sectional area is more than other locations. Contours of pressure distri-
bution for different flow rate are displayed in Fig. 11. According to these
contours, the maximum pressure distribution occurs when flow rate ¼
L
60 min that it is logical. Fig. 12 shows the streamlines of airflow in air-
ways. The presented figure depicted that in some places of airways, vor-
tices have been formed that causing more turbulence intensity in the
flow. This phenomena happens due to the realistic human respiratory
tracts are irregular in shape. Based on the observation that emerged
from the figure, it can be said in first bifurcation, for all three cases
(flow rate = 15, 30 and 60 min L
) more fluid flows to the right branch.

  
Fig. 15. Particles situation at t = 0.05 s after start injection with dp = 5 μm for (a) m¼ 15 min
L
, (b) m¼ 30 min
L
and (c) m¼ 60 min
L
.
 M. Rahimi-Gorji et al. / Journal of Molecular Liquids 209 (2015) 121–133 131

In second generation in right branch, for flow rate = 15 and 30 L

min
fluid flows to the left branch of this bifurcation, but for flow rate ¼ 60
more fluid flows to the right branch of this (zone number 11). In second
generation in left branch, for flow rate = 15 and30 min L
more fluid flows to
the right side of this bifurcation (zone number 7), but for flow rate ¼
L
60 min , more fluid flows to the left side of this (zone number 6).
Figs. 13, 14, 15 and 16 depict the particle situation at different time
after injecting. According to these figures, the particles tended to go to
the right branch. As can be seen in Fig. 16, the minimum number of
particles crossed the zone numbers 6 and 11. Fig. 17 demonstrates the
particle situation when t = 0.4 s and dp = 5 μm for flow rate ¼ 15 min L
(17-a), flow rate ¼ 30 minL
(17-b) and flow rate ¼ 60 min L
(17-c). In fact,
the total period of particle injection is 0.805 s, but this figure shows
the particle position at 0.4 s after the start of injection. As can be seen
more L
and 60 min , the maximum deposition fraction occurs in location of
L
min,
zone number 1. In these two cases, the total trapped particles were
about 8.5 % and 5.2 %, respectively. In fact, this figure confirmed the pre-
vious explanations and figures about the particle deposition.
4. Conclusion
Investigation of the airflow behavior and particle transport and de-
position fraction in three breathing condition such as light breathing
condition (15 L/min), normal breathing condition (30 L/min) and
heavy breathing condition (60 L/min) has been performed. The realistic
geometry data was reconstructed from a CT-scan images of the human
airways and imported into CFD simulation software. As can be seen in
figures, independent of the flow rate, the maximum deposition occurs
for dp = 10 μm. In fact, According to the results, for dp = 5 μm and


in this figure, when flow rate ¼ 15 min L

, the maximum deposition fraction 10 μm, when m¼ 30 min L
and for dp = 1 μm, when flow rate ¼ 15 min L
,
occurs in locations of zone numbers 1 and 3. In this case, the particles the particle deposition fraction have maximum amount. For flow rate ¼
that trapped to walls were about 4.3 %. Also, when flow rate ¼ 30 min L
15 L L
min and 30 min, the maximum deposition occurs in the zone number 1

  
Fig. 16. Particles situation at t = 0.8 second after start injection with dp = 5 μm for (a) m¼ 15 min
L
, (b) m¼ 30 min
L
and (c) m¼ 60 min
L
.
132 M. Rahimi-Gorji et al. / Journal of Molecular Liquids 209 (2015) 121–133

  
Fig. 17. Particles situation when t = 0.4 s with dp = 5 μm for (a) m¼ 15 min
L
, (b) m¼ 30 min
L
and (c) m¼ 60 min
L
.

and for flow rate ¼ 60 L

min occurs in the zone number 4. Also, the max- Appendix A. Supplementary data
imum pressure distribution happens when flow rate ¼ 60 min L
that it is
acceptable. The presented results showed that, the particles tended to Supplementary data to this article can be found online at http://dx.
go to the right branch and the minimum number of particles crossed doi.org/10.1016/j.molliq.2015.05.031.
the zone numbers 6 and 11. Also, at 0.4 s after injecting for the states
of flow rate ¼ 15 min
L L
, 30 min L
and 60 min with dp = 5 μm, the particles References
that trapped to the walls were about 4.3%, 8.5% and 5.2%, respectively. [1] F.Q. Hu, L. Wei, Z. Zhou, Y.L. Ran, Z. Li, M.Y. Gao, Adv. Mater. 18 (2006) 2553–2556.
[2] N. Durcan, C. Murphy, S.A. Cryan, Mol. Pharm. 5 (2008) 559–566.
[3] T.R. Flotte, B.L. Laube, Chest 120 (2001) 124S–131S.
Conflict of interest [4] I. Balashazy, W. Hofmann, J. Aerosol Sci. 26 (1995) 273–292.
[5] I. Balashazy, W. Hofmann, J. Aerosol Sci. 24 (1993) 745–772.
The authors declare that they have no conflict of interest. [6] Z. Zhang, C. Kleinstreuer, C.S. Kim, J. Aerosol Sci. 33 (2002) 257–281.
[7] I. Bala'sha'zy, J. Comput. Phys. 110 (1994) 11–22.
[8] B. Gatlin, C. Cuicchi, J. Hammersley, D. Olson, R. Reddy, G. Burnside, ASME, vol.
Acknowledgments FEDSM971997. 1–6 (Vancouver, British Columbia, Canada).
[9] A. Farag, J. Hammersley, D. Olson, T. Ng, ASME, vol. FEDSM981998. 1–8 (Washing-
ton, DC).
The authors would like to thank Mr Hosseinzadeh, expert at Sari [10] A. Farag, T. Ng, J. Hammersley, J. Olson, ASME, vol. FEDSM981998. 1–8 (Washington,
Parto Mazand Medical Imaging Center, for providing us with the CT- DC).
scan images that were used to generate the airways model. Also, we [11] A. Li, G. Ahmadi, Aerosol Sci. Technol. 23 (1995) 201–223.
[12] N. Nowak, P.P. Kakade, A.V. Annapragada, Ann. Biomed. Eng. 31 (2003) 374–390.
thank the reviewers for their useful and beneficial comments that im- [13] C. Van Eartbruggen, C. Hirsch, M. Paiva, J. Appl. Physiol. 98 (2005) 970–980.
proved this paper. [14] C. Kleinstreuer, Z. Zhang, L. Zhang, J. Respir. Physiol. Neurobiol. 163 (2008) 128–138.
 M. Rahimi-Gorji et al. / Journal of Molecular Liquids 209 (2015) 121–133
[15] C. Kleinstreuer, Z. Zhang, Annu. Rev. Fluid Mech. 42 (2010) 301–334.
[16] G. Tian, P.W. Longest, G. Su, M. Hindle, Ann. Biomed. Eng. 39 (2011) 1136–1153.
[17] G. Tian, P.W. Longest, G. Su, R.L. Walenga, M. Hindle, J. Aerosol Sci. 42 (2011)
781–799.
[18] P.W. Longest, S. Winchurkar, T. Martonen, J. Aerosol Sci. 37 (2006) 1234–1257.
[19] A. Farkas, I. Balashazy, J. Aerosol Sci. 38 (2007) 865–884.
[20] E.R Weibel, New York: Academic (1963).
[21] K. Horsfield, D. Gladys, D.E. Olson, G.F. Finlay, G. Cumming, J. Appl. Phys. 31 (1971)
207–217.
[22] T. Gemci, V. Ponyavin, Y. Chen, H. Chen, R. Collins, J. Biomech. 41 (2008) 2047–2054.
[23] T. Sun, J. Wang, X. Li, P. Lv, F. Liu, Y. Luo, Q. Gao, H. Zhu, X. Guo, Comput. Methods
Prog. Biomed. 111 (2013) 519–524.
[24] J. Fleming, J. Conway, C. Majoral, L. Tossici-Bolt, I. Katz, G. Caillibotte, et al., J. Aerosol
Med. Pulm. Drug Deliv. 24 (2011) 49–60.
133
[25] L. Golshahi, M.L. Noga, R.B. Thompson, W.H. Finlay, J. Aerosol Sci. 42 (2011)
474–488.
[26] C. Kleinstreuer, Z. Zhang, J.F. Donohue, Annu. Rev. Biomed. Eng. 10 (2008) 195–220.
[27] H. Takano, N. Nishida, M. Itoh, N. Hyo, Y. Majima, J. Aerosol Med. 19 (2006)
314–328.
[28] E.A. Matida, M. Ilie, W.H. Finlay, Aerosol Sci. Technol. 42 (2008) 10–17.
[29] Z. Zhang, C. Kleinstreuer, C.S. Kim, Ann. Biomed. Eng. 36 (2008) 2095–2110.
[30] J. Xi, P.W. Longest, Effects of oral airway geometry characteristics on the difusional
deposition of inhaled nanoparticles, J. Biomech. Eng. ASME 130 (2008).
[31] S. Elghobashi, Appl. Sci. Res. 52 (1994) 309–329.
[32] C. Crowe, T. Troutt, J. Chung, Annu. Rev. Fluid Mech. 28 (1996) 11–43.
[33] B.R. Munson, D.F. Young, T.H. Okiishi, 5th ed., John wiley & Son, Inc. (2006).
[34] S.A. Morsi, A.J. Alexander, J. Fluid Mech. 55 (1972) 193–208.