The CFD simulations performed were similar to the previous nasal ECG (12 (
link)) and oral EEG (18 (
link)) studies and employed the Fluent 12 (ANSYS Inc.) software supplemented with user routines. In brief, a low Reynolds number (LRN) k-ω turbulence model was used to simulate the flow, which can be laminar, transitional, or turbulent at different locations in the flow field. This turbulence model has previously been used for the successful prediction of aerosol transport and deposition in upper airway models (26 (
link),28 (
link),29 (
link)). To evaluate the variable temperature and relative humidity fields, the coupled governing equations of heat and mass transport, reported in detail by Longest and Xi (30 ) and Longest
et al.(31 ), were used. Lagrangian transport equations coupled with user-defined functions were employed to estimate the particle trajectories, growth, and deposition. User-defined functions were implemented to improve estimation of near-wall conditions and to simulate the aerosol evaporation and condensation in the complex three-dimensional temperature and humidity fields (12 (
link)). User-defined functions were also used to account for anisotropic near-wall turbulent dispersion, Brownian diffusion of the initially submicrometer aerosols, and near-wall interpolation of fluid velocities (12 (
link)). Our previous studies have demonstrated that this combination of a commercial code with multiple user-defined functions is capable of accurately capturing particle deposition arising from inertia, sedimentation, diffusion, and turbulent dispersion (19 , 31 ). The Kelvin effect, influences of excipient and drug hygroscopicity, and the effect of droplet temperature on surface vapor pressure were considered in the droplet size change calculations based on previous studies (12 (
link)). The influence of the droplet on the carrier phase was neglected and a one-way coupled approach was implemented in the model. The details of solving the above equations using realistic boundary conditions can be found in Longest
et al. (12 (
link)) and Tian
et al. (18 (
link)).
For simulating droplet trajectories and aerosol size increase, initially monodisperse size distributions were implemented. Condensational growth of the aerosols then led to a polydisperse aerosol size distribution within the airway models. With the control experiments, an initial droplet size of 3.6 μm was found to match the experimentally measured aerosol size exiting the nasal cannula and entering the NMT model. The aerosol size exiting the Aeroneb Lab nebulizer could not be implemented directly due to high depositional losses in the neonatal T-connector, which reduced the MMAD of the aerosol and was not included in the CFD model. For the EEG and ECG simulations, the experimentally measured size exiting the mixer tubing (900 nm) was implemented at the model inlet. In all droplet simulations, 9000 initial particles were injected into the steady state flow stream and increasing this number had a negligible effect on deposition fractions.
The computational mesh was constructed using the ANSYS ICEM 10 package (Ansys Inc., Canonsburg, PA) and consisted of tetrahedral and hexahedral elements. Hexahedral control volumes were used primarily in the connective tubing, cannula geometry, and tracheal region, whereas tetrahedral elements with a thin layer of near-wall wedge control volumes were used to resolve the nasal passages through the pharynx. Grid density testing produced convergent results in terms of regional deposition for a control volume cell count of approximately 1.3 million for the EEG NMT geometry and 1.5 million for the ECG NMT model.
Golshahi L., Tian G., Azimi M., Son Y.J., Walenga R., Longest P.W, & Hindle M. (2013). The Use of Condensational Growth Methods for Efficient Drug Delivery to the Lungs during Noninvasive Ventilation High Flow Therapy. Pharmaceutical research, 30(11), 2917-2930.