Existing numerical models of the flow and inhaled drug delivery in the mouth and throat regions show poor agreement with the actual particle depositions patterns observed in experimental tests. The limitations of these models can be attributed mainly to the use of idealized geometries, and an inaccurate representation of the flow field often with little attention to resolving the turbulent flow. This effort aims to address these limitations by (a) developing a high fidelity, time-accurate simulation of the turbulent flow in the complex geometry of the airways and (b) improving the efficiency and scalability of this predictive tool which can be valuable in the optimization of inhaled drug delivery.

Our efforts use realistic models of mouth and throat geometries obtained from magnetic resonance imaging (MRI) scans. The scans are processed to generate 3D throat/lung models, which are imported into the flow-simulator directly. Direct numerical simulations, which capture all the scales in the turbulent flow, are carried out. The turbulence significantly affects particle trajectories and deposition, especially in the case of smaller particles.

Examples of the flow and particle deposition patterns in the mouth-throat geometry are shown in the videos below. The scan was obtained from MRI, provided by GlaxoSmithKline. Contours of the instantaneous velocity are shown at left, and particle transport and deposition are shown in the model at right.

Our work utilizes an immersed boundary approach to model the geometry. Therefore, structured grids are used despite the complexity of the mouth-throat geometry, which are immersed on the computational grid. Immersed-boundary methods greatly simplify the task of grid generation and eliminate the need of constructing new computational meshes for every new throat geometry. Our method is further optimized by applying the immersed boundary technique to curvilinear grids, which generally follow the airways and minimize unused cells.