Multiscale simulation of respiratory airflow using physiologically consistent geometry and boundary conditions in OpenFOAM

Abstract

In computed tomography (CT)-based computational fluid dynamics (CFD) simulations of the human respiratory system, no or few studies have incorporated both realistic upper and lower airways, along with extensions to CT-unresolved higher-generation airways. In this study, we present a CT-based, physiologically consistent CFD model of the human airway that integrates artificial airway extensions down to the transitional bronchioles within the OpenFOAM framework. The model includes a hybrid turbulence approach combining Reynolds-averaged Navier-Stokes (RANS) and large eddy simulation (LES), and a state-of-the-art airway mesh smooth learning (AMSL) technique for constructing accurate airway geometries. Physiologically consistent boundary conditions are applied using airflow data derived from one-dimensional network simulations. We investigate the impact of the hybrid RANS-LES model on airflow characteristics, pressure distribution, and particle deposition by comparing it with conventional turbulence models, including the wall-adapting local eddy-viscosity (WALE) model for LES and the k-ω SST model for RANS. The AMSL method is also evaluated against the traditional Taubin smoothing technique. Our results show that pressure does not monotonically decrease throughout the upper respiratory tract but exhibits a continual decrease in the lower tract, independent of airway generation. The hybrid RANS-LES model demonstrates flow patterns and particle deposition characteristics comparable to those of the LES model and proves an improved fidelity over traditional RANS models. Furthermore, the AMSL technique significantly influences airflow behavior and particle deposition, highlighting the importance of accurate geometry processing. In conclusion, the proposed physiologically consistent CFD model, implemented in the OpenFOAM framework, demonstrates strong potential for clinical and research applications by offering enhanced accuracy and reliability. The use of an integrated airway model, extending from the upper airways to artificially constructed distal airways, facilitates a better understanding of multiscale airflow dynamics in the lungs.

Keywords: Airway surface smoothing; Computational fluid dynamics; Computed tomography; Hybrid RANS-LES; Particle deposition.