Understanding the flow dynamics and particle deposition in the respiratory system is important in determining the efficiency of aerosolised drug delivery and the toxicity of airborne pollutants. The large geometric variation across subjects has a pronounced effect on the flow and particle deposition, and therefore motivates the need for patient-specific predictions. In the treatment of respiratory diseases, the ultimate goal is to develop capabilities for personalised medicine in order to provide optimal patient-specific drug delivery. Numerical models offer a cost-effective and non-invasive approach, compared to in vitro and in vivo testing. However, accurate and efficient simulations pose a challenge due to the complexity of the airway geometries and the complexity of the flow in the airways. 

Accurate solution of the flow field and deposition patterns motivate the use of direct numerical simulations (DNS) in order to resolve the flow. Due to the high grid resolution requirements, it is desirable to adopt an efficient computational strategy. We employ an immersed boundary method developed for curvilinear coordinates, which allows the use of structured grids to model the complex patient-specific airways, and can accommodate the inter-subject geometric variations on the same grid.  A Lagrangian particle tracking scheme is adopted to model the transport and deposition of aerosol particles, and we apply a non-rigid registration method to construct dynamic airway models conforming to the patient’s breathing. 

geometric variation