Numerical modelling of fluid-structure interactions for fluid-induced instability in the upper airway

摘要

This study is concerned with fluid-structure interactions (FSI) involved in the human upper airway, in particular, those associated with snoring and obstructive sleep apnoea/hypopnoea syndrome (OSAHS). Further examining this area of interest, the goal of the current research is to contribute further understanding and enhance development of computational modelling, for retroglossal obstruction and palatal snoring. To that end, the investigation was divided into three major parts. Firstly, extending previous laminar, 2-D reduced Navier-Stokes model, an idealised 3-D computational model was constructed for studying retroglossal obstruction. A full Navier-Stokes solver in an Arbitrary Lagrangian-Eulerian (ALE) framework was coupled to a linear thin shell, where both laminar and turbulent flow was investigated. Numerical results showed increase flow-induced tongue replica deflection under turbulent conditions and demonstrated cross-flow pressures that may encourage side wall collapse. In the second part of the thesis, palatal snoring was further examined and its potential to detect retroglossal obstruction was proposed. In order to investigate this, flow-induced instability of a cantilever plate in an obstructed channel was modeled and a relationship between critical velocity and obstruction depth was established. Correlating the critical velocity with typical breathing flow curve, a time difference between palatal snoring episodes or onset of palatal snoring, may represent a key variable for non-invasive measurement of retroglossal obstruction severity. Further enhancement of the 2-D computational model by including contact was proposed using an immersed boundary method (IBM). This may represent a more complete model of palatal snoring by modelling pre- and post-contact response of unstable cantilever plate, which showed potential to capture more complicated palatal snoring signals. Finally, the third part of this thesis examined flow-induced instability of a soft palate in a 3-D realistic upper airway. To model this, a full 3-D Navier-Stokes solver under an ALE framework was coupled to a non-linear soft palate model. Appropriate soft palate properties were applied giving palatal snoring frequency within range of clinically measured values. Palatal flutter was observed at high flow rates, demonstrating irreversible transfer of flow energy to soft palate. This computational model may perhaps be exploited for future investigation of more accurate palatal snoring, necessary for developing non-invasive snoring signals for measurement or diagnosis of retroglossal obstruction.