TL;DR
This study addresses inconsistencies in computational fluid dynamics (CFD) modeling of the human sinonasal cavity by systematically investigating mesh density's impact on airflow and particle deposition. They found that approximately 4 million tetrahedral elements are needed for airflow grid independence, and hybrid tetrahedral-prism meshes significantly improve particle deposition accuracy compared to purely tetrahedral meshes.
What Problem Does This Paper Address?
Accurately simulating airflow and particle deposition in the complex geometry of the human sinonasal cavity using CFD is crucial for understanding drug delivery and respiratory health. However, the literature shows significant ambiguity regarding the appropriate mesh density and type required for reliable, "grid-independent" results. This lack of standardization stems from variations in nasal geometry (e.g., unilateral vs. bilateral, presence of sinuses) and the diverse CFD variables used to assess mesh convergence. These inconsistencies lead to unreliable simulation outcomes, particularly for particle transport, which is highly sensitive to near-wall flow dynamics. This paper aims to provide clear guidelines for mesh refinement in sinonasal CFD modeling.
Methods
The researchers employed a two-stage mesh refinement strategy.
Stage 1: Airflow Simulation (Tetrahedral Meshes) They generated nine different global mesh densities using unstructured tetrahedral elements, ranging from 0.25 million to 8 million cells. Mesh quality was ensured by maintaining a minimum tetrahedral element aspect ratio. Grid convergence for airflow was assessed by monitoring volumetric airflow rate, transnasal pressure drop, and airflow velocity profiles under physiological, pressure-driven conditions.
Stage 2: Particle Transport Simulation (Hybrid Tetrahedral-Prism Meshes) Building upon the airflow-converged mesh, they introduced prism layers near the airway walls. This hybrid mesh approach is standard for capturing near-wall phenomena. They systematically varied the number of prism layers (1-6) and their thickness to determine mesh independence for particle deposition. Simulations involved releasing particles of varying sizes (1-20 μm) and tracking their trajectories using the Discrete Phase Model in Fluent.
Datasets
The study utilized a 3D reconstruction of a complete sinonasal cavity derived from computed tomography (CT) scans of a single subject who had undergone functional endoscopic sinus surgery (FESS). This post-surgical geometry was chosen because FESS restores nasal patency to a near-normal state, and its complexity allows for a thorough investigation of mesh refinement. The authors also compared their simulation results to experimental data from Kelly et al. (2004) for validation.
Key Insights and Ideas
- Hybrid Meshes are Crucial for Particle Deposition: The most significant finding is that purely tetrahedral meshes tend to overpredict particle deposition. Hybrid meshes, incorporating prism layers near the walls, are essential for accurately capturing near-wall particle behavior and achieving asymptotic deposition results.
- Airflow and Particle Transport Have Different Convergence Criteria: Mesh independence for airflow is achieved at a lower cell count (around 4 million tetrahedral elements) than what is required for accurate particle deposition, which necessitates finer near-wall resolution via prism layers.
- Specific Regions Require More Refinement: While overall convergence is achieved, certain regions within the sinonasal cavity, like the left maxillary sinus, showed less convergence and may benefit from localized mesh refinement for highly accurate local deposition predictions.
- Standardization is Key: The study highlights the need for standardized approaches in CFD sinonasal modeling, including consistent reporting of mesh refinement variables and potentially standardized geometries, to improve the reliability and comparability of research findings.
Key Results
- Airflow Grid Independence: Approximately 4 million unstructured tetrahedral elements were found to be sufficient for achieving grid independence in airflow simulations, with relative changes in outlet volume flow rate and transnasal pressure drop being minimal (<0.2% and <0.1%, respectively) beyond this density.
- Particle Deposition Convergence: Hybrid meshes with three prism layers (total thickness of 0.3 mm) demonstrated asymptotic behavior for sinonasal particle deposition.
- Impact of Prism Layers: Inclusion of boundary prism layers reduced deposition fractions compared to tetrahedral-only meshes, indicating that tetrahedral-only meshes overestimate deposition.
- Validation: The simulation results from the hybrid tetrahedral-prism mesh model showed good agreement with experimental data from Kelly et al. (2004) when deposition fractions were plotted against the deposition parameter (d²ΔP).
Limitations and Future Work
- Single Subject Geometry: The study used a 3D model from a single subject, which, while post-surgical, may not represent the full spectrum of sinonasal anatomies. Future work could explore the influence of anatomical variations (e.g., septal deviation, turbinate hypertrophy) on mesh sensitivity.
- Laminar Flow Assumption: The simulations were conducted under laminar flow conditions. While justified for resting breathing rates, turbulent flow can occur during more vigorous breathing. Future studies could investigate mesh refinement for turbulent sinonasal airflow and particle transport.
- Generalizability: The authors caution that generalizations about specific mesh densities (e.g., 4 million tetrahedral elements) should be made cautiously, as the optimal mesh might depend on the specific geometry and the CFD solver used.
- Wall Roughness: The study acknowledges that wall roughness can influence particle deposition, but its effect on mesh density analysis remains poorly understood and warrants further investigation.
Summary updated