CFD analysis of mucous effect in the nasal cavity

Abstract

This research aims to investigate the airflow patterns and particle deposition in a healthy human upper airways. A realistic 3-D computational model of the upper airways including the vestibule was developed using a series of CT scan images of a healthy human. Simulations of the airflow fields in the upper airway passages were performed by solving the Navier-Stokes and continuity equations for breath rate 20 L/min. The trajectory analysis approach was applied to study the particle transport and deposition for the model with and without mucous lining. The presented results revealed that the mucous layer can have significant impact on airflow analysis and there were noticeable differences in the amount of particle deposition in each models.

Keywords

Computational fluid dynamics (CFD)mucous effect nasal airway particle deposition

1. Introduction

Inhaled particle deposition in the airway channels has caused serious health problems. These particles consist of dust, microorganisms, photochemical smog and other irritants. From the toxicological point of view, all particles which are smaller than 10 $\mu$ m in diameter can be biologically active and could cause allergic reactions and even cancers in disposed individuals [1, 2]. On the other hands, for Inhaled drug delivery purposes it is essential to make sure particle deposition occurs at targeted areas. For instance, it is required to maximize the deposition of pharmaceutical aerosol which are larger than 20 $\mu$ m in nasal cavities and pharyngeal passage before entering the lungs. On many other cases particles smaller than 5 $\mu$ m [3] and sometimes in the sub-micrometer size range are used in order to deliver the drug to the lungs [4]. Therefore, estimating the fraction of particles that penetrates into the lung and predicting the distribution of deposited particles in different regions of the human upper airways is of considerable interest not only to determine what reaches the lung, but also to quantify the dose to upper airways for both the health risk and therapeutic effectiveness of inhaled particles. Particle deposition in the nasal airway has been the subject of many investigations and the effect of different factors has been studied both numerically and experimentally.

Computational fluid dynamics (CFD) method, which gained more popularity due to the advances in computer technologies, can provide an alternative to the experimental techniques for predicting particle deposition in nasal airways. Many researchers have studied airflow patterns and particle depositions in this field [5, 6]. Techniques have included computational fluid dynamics and particle image velocimetry (PIV) which resulted in better understanding of the physiological and pathological aspects of the nasal cavity [7]. Even though particles are carried by the breathing air through the respiratory system, their trajectories differ from the airflow streamlines due to their inertia and effects of various forces. The most important forces are gravity, drag, and the Brownian excitation. Particles are then deposited on the walls of the airway by sedimentation, impaction, and/or diffusion mechanisms

Particle trajectory analysis can give scientists essential information in the field of inhalation and aerosol therapy. Experimental studies into aerosol deposition have been investigated in the past by a number of researchers [8, 9], but a thorough search of such literatures yielded that this effect was ignored through the airflow and particle deposition studies and its details has not been discussed appropriately. Hence, the mucous layer thickness should be included as a structural feature in 3D analytical models and solved as a non-linear model. Mucous layer is a thin mucous film covered the nasal airway boundary and served as a protection mechanism which is critical for the capture and removal of aerosol particles. Mucous is mostly composed of around 95% water with the assumed thickness of 5 to 50 $\mu$ m [10, 11]. It also works as a barrier to the open-air particulate matters, preventing infection and tissue injury by the action of mucociliary clearance. According to reference [7] factors such as cilia, mucous lining and epithelial tissues have become new field of interest for research and study. It was found that even minor changes of the nasal cavity at particular locations might affect the airflow distribution and thus the particle deposition. This study discussed such effects in nasal airflow caused by the micrometer changes of the mucous layer thickness along the nasal passageway. Differences in maximum velocities resulted from the mucous layer and visualization of the nasal airflow were studied. Additionally, Comparison between the models with and without mucous layer was investigated as well. This paper can also provide new opportunities that will be crucial for further analysis and understanding of mucous effects and particle deposition in nasal cavity.

2. Methodology

2.1 Model outline and mesh

A three dimensional nasal computational model was developed based on the computed tomography (CT) scans of a healthy 39-year old female. The scans data was procured at the Universiti Sains Malaysia (USM), Medical Campus Hospital. The CT scan images were obtained from the axial, coronal and sagittal planes. The increment between each slice of the scan images is 0.8 mm. The scan images were segmented by defining threshold values ranging from $-$ 444 to 2037 HU. Segmentation was performed slice by slice on the scan images to ensure the accuracy of the selected region of interest. By using the image processing software MIMICs, the 2D scan images were converted into a 3D nasal cavity model. For non-mucous model, the threshold in MIMICS neglected the soft tissues along nasal cavity as the nasal cavity was assumed to be decongested and vice versa for the mucous layer case. For mucous layer case, its thickness was assumed to be 50 $\mu$ m and constant in along the nasal cavity. The properties of mucous layer was assumed to be similar along the nasal wall as it was a preliminary study to identify how does the micrometer thickness of mucous layer affecting the nasal deposition.

As shown in Fig. 1, the reconstruction of the upper airway section was made in CATIA (Dassault Systems, SA) to smoothen and create the final model. Then the computational domain was exported to Gambit 2.3.16 (Fluent Inc., Lebanon, USA) to generate a hybrid mesh with 7,900,000 elements, which consisted of a total of 5 layers of prism mesh near the wall boundary. Moreover, for the residual plot, it was considered as acceptable only when the residual values in all equations had converged to an acceptable value i.e., between three to four orders of magnitude (10 ${}^{-3}$ to 10 ${}^{-4}$ ). All these steps were carried out with professional assistance of ear, nose and throat (ENT) experts.

Figure 1.

Cross-sectional view of the nasal cavity showing the hybrid mesh [5].

2.2 Flow characteristic and boundary condition

The numerical simulation was performed using the commercial CFD solver COMSOL Multiphysics. The simulation was carried out on an Intel (R) Core (TM) i5-5200U CPU @ 2.20 GHz, 2 cores, which basically required approximately 3 days of runtime for entire simulation. For the boundary condition, the wall for the nasal model was assumed to be rigid with no slip condition. The mass inflow rate defined at nostril and no pressure outflow condition was considered at the nasopharynx. The gravity factor was ignored during the simulation. Inspiratory flow rate of 20 L/min was used in the simulation. Comparison and validation were carried out using the analysis of the mucous and non-mucous models with previous researches [5, 8, 12, 13, 14, 15]. All computational analysis are validated and therefore the research continued with evaluation of mucous layer effect on the particle deposition.

The air flow simulation was based on the Navier-Stokes equations for the continuity Eq. (1):

$\displaystyle\frac{{\partial u}_{i}}{{\partial x}_{i}}=0$ (1)

And conservation of momentum, Eq. (2):

$\displaystyle u_{j}\frac{{\partial u}_{i}}{{\partial x}_{j}}=-\frac{1}{\rho}% \frac{\partial p}{{\partial x}_{i}}+\frac{\partial}{{\partial x}_{j}}\left(% \left(v+v_{t}\right)\frac{{\partial u}_{i}}{{\partial x}_{i}}\right)$ (2)

where $u_{i}$ is the air velocity in three Cartesian coordinate direction, i.e., $i=$ 1, 2 and 3, $p$ is the pressure, $\rho$ is the fluid density, $v$ is the air kinematic viscosity and $v_{t}$ is the kinematic turbulent (eddy) viscosity. The eddy viscosity is computed with the use of a turbulence model. The air flow was assumed to be incompressible and steady. The “Turbulent Flow, SST” RANS model, is used to simulate the flow velocity fields. This model has been found to provide very good calculations of wall bounded flows even with highly separated regions. The SST combines the advantages of both the $k$ - $\varepsilon$ and $k$ - $\omega$ methods. The $k$ - $\varepsilon$ turbulence model has a near-wall treatment allowing accumulation of nodes towards the wall without any special non-linear damping function, whereas the $k$ - $\omega$ turbulence model is more accurate for flows involving strong streamlines curvature. The constant parameters for turbulence model in this study are: $a_{1}$ is 0.31, $\beta_{1}$ is 0.075, $\beta_{2}$ is 0.0828, $\sigma_{\omega 1}$ is 0.5, $\sigma_{\omega 2}$ is 0.856, $\sigma_{k1}$ is 0.85, $\sigma_{k2}$ is 1, $K_{v}$ is 0.41, and B is 5.2. In this study, there are about 90000 tetrahedral type mesh and around 50000 triangular type mesh.

The Particle Tracing for Fluid Flow (FPT) module was used to describe the particle trajectories and to derive the collection efficiency by counting the number of particles at the outlet. The particle trajectories are computed using Newtonian formulation with the Stokes drag law,

$\displaystyle\frac{d}{dt}\left(m_{p}V\right)=\frac{1}{\tau_{p}}m_{p}(u-v)$ (3)

where $V$ is the particle velocity and $\tau_{p}$ is the particle velocity response time:

$\displaystyle\tau_{p}=\frac{\rho_{pd^{2}p}}{18\mu}$ (4)

where $\rho_{p}$ is the density of the particles and $d_{p}$ is the particle dimeter. The density of particles normalized according to fluid velocity at the inlet. This means that there are more particles released where the inlet velocity magnitude is highest and fewer particles released where the magnitude is low. The particles are subjected to drag forces, coming from the results of “Turbulent Flow” module and they released uniformly across the inlet. Particles are assumed to be freeze at all flow walls.

Figure 2.

Velocity distribution for 20 L/min. (a) Non-mucous (b) Mucous models.

Figure 3.

Pressure distribution for 20 L/min. (a) Non-mucous (b) Mucous models.

3. Results and discussion

3.1 Model validation

The developed model of nasal cavity in this study was previously validated in terms of the pressure drop [16]. For this reason, the methodology will not be repeated here again. Moreover, For the Mucous effect, the results shows a good agreement with earlier work of Lee et al. [10]. Air flow characteristics and its behavior inside the nasal cavity is equally important for both nasal physiology determination and associated diseases diagnosis. The breathing rates can be ranged from 5 to 12 L/min for light condition activities and from 12 to 40 L/min for heavy work activities such as physical exercise [17]. In this paper, the turbulent breath rate of 20 L/min was considered.

Figure 4.

Particle trajectories for the model with mucous thickness of 50 $\mu$ m.

Figure 5.

Deposition rate of 20 $\mu$ m particle with inhalation rate of 20 L/min.

3.2 Velocity and pressure distribution

Figure 2 illustrates the velocity slices for flow rate of 20 L/min for mucous and non-mucous models. By having the mucous layer with thickness of 50 $\mu$ m, there is an increase in the velocity magnitude, particularly in vestibule and nasal valve regions. The middle plane sections are the areas where the models experienced the lowest velocity magnitude. As can be seen, the nasal model morphology is one the influential factors on the airflow patterns and presence of mucous layer can cause different behavior of velocity profile in nasal cavity. The airflow acceleration and deceleration is obvious in the model due to change in cross sectional area of the air path way.

Figure 3 displays the pressure contours for the models with and without mucous layer. According to Fig. 3, in the model without mucous layer, the pressure magnitude is higher in comparison with the mucous included model which is understandable due to higher velocity in this model. In both cases, the pressure contours are stronger in the nostril and vestibule regions. These high pressure zones can also be observed in the area where the direction of the airflow changes inside the cavity which are more significant in the left planes. Accordingly, the pressure value decreases until it exits the nasopharynx.

3.3 Particle trajectory and deposition rate

Equations (3) and (4) have been used to calculate particle trajectory which were inserted through nostril inlet when flow velocity is known. The articles deposit on the nasal wall when the distance between the particle center and the surface is less than or equal to the particle radius the particle is assumed to be deposited on the wall. The deposition rate in each region, identified by means of particle accumulator in that specific region. Figure 4 shows the particle trajectory in the nasal model with mucous thickness of 50 $\mu$ m. The diameter of particles is 20 $\mu$ m with inlet mass flow rate of 20 L/min.

According to Fig. 5, for flow rate of 20 L/min and particle size 20 $\mu$ m, most particles were deposited at the vestibule region compared to the pharynx region. This is due to the higher particle inertia and deviation from the flow streamlines at the vestibule region. The obtained results were analyzed and compared with the previous study which was done without the mucous layer inclusion [9]. The values are differ around 3 percent between two cases (mucous and non-mucous) and its higher for the case with mucous layer specifically in the vestibule and middle plane regions which is 32% and 25% respectively. The reason for the higher deposition rate in the vestibule region is due to the higher particle inertia and deviation from the flow streamlines at this region. The lowest deposition rate, on the other hand, is for the nasal valve area which is also experienced the lowest difference between two models.

4. Conclusions

A CFD study of airflow and particle deposition in the human upper airway has been the subject of many researchers. However, no numerical study of the particle deposition in an upper airway model with mucous layer has been presented so far in the literatures. Addition of mucous layer in the nasal model disturbed the airflow and thus changed the maximum velocity especially at the middle plane as all the turbinates were involved in determining the maximum velocities. The morphology of the upper airway was found to affect significantly the airflow pattern and the deposition fraction of the microparticles. There was a noticeable difference between models with mucous and without mucous layer. Findings in this study proved that mucous effect cannot be ignored in the numerical simulations and it is required for more precise and in-depth investigations. In this paper, particle transport and deposition in the human nasal model section is limited to one micro-particle size (20 $\mu$ m) and one breath rate (20 L/min). But, it is known that different particle diameters and inhalation rates have a different effects on the deposition behavior. Therefore, further research can be included with take into accounts such parameters of interest.

References

Heyder

, Deposition of inhaled particles in the human respiratory tract and consequences for regional targeting in respiratory drug delivery, Proc Am Thorac Soc 1(4) (2004), 315–320.

Cheng

Y.S.

, Aerosol depoition in the extrathoracic region, Aerosol Sci Technol 37(8) (Aug 2003), 659–671.

Suman

J.D.

Laube

B.L.

and Dalby

, Validity of in vitro tests on aqueous spray pumps as surrogates for nasal deposition, absorption, and biologic response, J Aerosol Med 19(4) (2006), 510–521.

Worth Longest

and Hindle

, CFD simulations of enhanced condensational growth (ECG) applied to respiratory drug delivery with comparisons to in vitro data, J Aerosol Sci 41(8) (2010), 805–820.

Zubair

Riazuddin

V.N.

Abdullah

M.Z.

Rushdan

Shuaib

I.L.

and Ahmad

K.A.

, Computational fluid dynamics study of pull and plug flow boundary condition on nasal airflow, Biomed Eng Appl Basis Commun 25(4) (2013), 1350044.

Zubair

Riazuddin

V.N.

Abdullah

M.Z.

Ismail

Shuaib

I.L.

and Ahmad

K.A.

, Computational fluid dynamics study of the effect of posture on airflow characteristics inside the nasal cavity, Asian Biomed 7(6) (2013), 835–840.

Zubair

Ahmad

K.A.

Abdullah

M.Z.

and Sufian

S.F.

, Characteristic airflow patterns during inspiration and expiration: Experimental and numerical investigation, J Med Biol Eng 35(3) (2015), 387–394.

Ahmadi

Zubair

Ahmad

K.A.

and Nazira

, Study on nasal deposition of micro particles and its relationship to airflow structure, Int J Fluid Heat Transf 1(1) (2016), 1–11.

Riazuddin

V.N.

Zubair

Ahmadi

Tamagawa

Rashid

N.H.A.

Mazlan

and Ahmad

K.A.

, Computational fluid dynamics study of airflow and microparticle deposition in a constricted pharyngeal section representing obstructive sleep apnea disease, J Med Imaging Heal Informatics 6(6) (2016), 1507–1512.

10.

Lee

C.F.

Ahmad

K.A.

Ismail

and Hamid

S.A.

, Numerical study of mucous layer effects on nasal airflow 24 (February 2012), 27–28.

11.

Lee

C.F.

Ahmad

K.A.

Ismail

and Hamid

S.A.

, Mucous layer effects towards nasal airflow 24 (February 2012), 27–28.

12.

Zubair

Ahmad

K.A.

and Riazuddin

V.N.

, A review on the impact of aircraft cabin Air quality and cabin pressure on human wellbeing, in: Applied Mechanics and Materials 629 (2014), 388–394.

13.

Zubair

Abdullah

M.Z.

Ismail

Shuaib

I.L.

Hamid

S.A.

and Ahmad

K.A.

, Review: A critical overview of limitations of cfd modeling in nasal airflow, J Med Biol Eng 32(2) (2012), 77–84.

14.

Zubair

Ahmad

K.A.

Riazuddin

V.N.

Abdullah

M.Z.

Rushdan

Shuaib

I.L.

and Khan

, Numerical study on the effect of gender on the airflow characteristics inside the nasal cavity, Int J Adv Thermofluid Res 1(1) (2015), 2–16.

15.

Zubair

Riazuddin

V.N.

Abdullah

M.Z.

Ismail

Shuaib

I.L.

and Ahmad

K.A.

, Computational fluid dynamics study of the effect of posture on airflow characteristics inside the nasal cavity, Asian Biomed 7(6) (2013), 835–840.

16.

Riazuddin

V.N.

Zubair

Abdullah

M.Z.

Ismail

Shuaib

I.L.

Hamid

S.A.

and Ahmad

K.A.

, Numerical study of inspiratory and expiratory flow in a human nasal cavity, J Med Biol Eng 31(3) (2011), 201–206.

17.

Wen

Inthavong

and Wang

, Numerical simulations for detailed airflow dynamics in a human nasal cavity, Respir Physiol Neurobiol 161(2) (2008), 125–135.