Extension of the finite volume particle method for fluid-structure interaction in cardiac valve prostheses
MetadataShow full item record
This item's downloads: 30 (view details)
Cardiac valve prostheses are used to treat diseased native heart valves and failing prosthetic valves. Computational modelling methods have many applications in cardiovascular modelling, such as designing cardiac valve prostheses, modelling the implantation process, and determining patient-specific treatments. Computational fluid dynamics (CFD) methods based on particles are potentially advantageous for the modelling of cardiac valve prostheses, as they can handle the large deformations of the fluid domain that occur during the opening and closing of a cardiac valve. The aim of this thesis is to extend the meshless CFD method, the finite volume particle method (FVPM), to enable modelling of cardiac valve prostheses with fluid-structure interaction (FSI). In this work, an improved particle transport velocity correction was developed for FVPM, to maintain a uniform particle distribution throughout the large deformations of the fluid domain. In cardiovascular modelling, it is necessary to model the solid, as well as the fluid, due to the strong coupling between the two. FVPM was combined with a finite element solver, FEBio, to model FSI. The FSI method was enhanced with a novel method for the interaction between thin structures and CFD methods based on particles. This method allows for the modelling of thin structures, such heart valve leaflets, without requiring unnecessarily small particle sizes, which are computationally expensive. Each development was validated against theoretical and experimental data from the literature. Recent studies suggest that thrombosis in cardiac valve prosthesis occurs in regions of flow stasis. The FVPM-FEBio FSI model for thin structures developed in this work was applied to a study on the effect of valve tilt angle on flow stasis in an idealised 2-D transcatheter heart valve. Particle residence time (PRT) was used to evaluate flow stasis in the neo-sinus of the valve. The FVPM-FEBio PRT agreed with experimental data in the literature. Furthermore, the regions of flow stasis detected by the developed model corresponded to the location of thrombosis in clinical studies. No significant difference was found in PRT for the three tilt angles studied here. The work in this thesis shows that FVPM is suitable for modelling cardiac valve prosthesis. It has the ability to discretise fluid domains of complex geometry, and can handle large deformations of the fluid-solid interface that occur during valve opening and closing. The thin heart valve leaflet was modelled with a particle size that was independent of the structure thickness, and larger than the structure thickness, improving on computational cost as fewer fluid particles are required. PRT was predicted in a straightforward manner, as additional particles could be easily added into the FVPM domain. The developed method is also capable of modelling various implantation positions as shown by the valve tilt angles modelled here.