FSI simulations of mechanical aortic valves through LS-DYNA: comparative analysis of operator split Lagrangian-Eulerian and ICFD-based approaches

The aortic valve is located between the left ventricle and the aorta, and it fulfills the essential function of ensuring unidirectional blood flow from the ventricle to the aorta. The aortic valve is susceptible to diseases such as aortic stenosis and regurgitation, which result in left ventricle overload and potentially heart failure. A common treatment for aortic valve disease is aortic valve replacement, entailing the substitution of the native valve with a mechanical or biological prosthesis. Mechanical aortic valves ensure higher durability with respect to biological ones, but they induce fluid dynamic alterations that increase the risk for thrombosis, hemolysis, and device failure. The computational investigation of the fluid dynamics of mechanical aortic valves allows for examining the flow patterns determined by different device positionings and designs, thereby supporting the optimization of these aspects and the consequent reduction in risks associated with valve presence. To this purpose, numerous studies have relied on fluid-structure interaction (FSI) simulation, which allows for investigating the mutual interaction between blood flow and the prosthetic valve. The literature describes a wide range of FSI approaches, with simulation results partially influenced by the selected numerical method. In this context, the aim of this thesis is twofold: (i) to conduct FSI simulations to investigate the effect of three different rotation angles and two different valve designs on fluid dynamics and (ii) to investigate the effect of two different FSI algorithms on simulation results. Two geometric models resembling the Abbott Regent and the On-X aortic valves were generated and assembled with an idealized model of the aortic root. The fluid domain, valve housing and valve sewing cuff were discretized with three-dimensional elements, while valve leaflets were discretized with shell elements. Two cardiac cycles were simulated, imposing pressure boundary conditions at the inflow and outflow sections of blood domain. An operator split Lagrangian-Eulerian FSI approach implemented in Ansys LS-DYNA was adopted to compare three different valve rotation angles and the two valve models. Subsequently, the results obtained for the Regent valve in one of the three selected rotation angles were compared with those obtained simulating the same scenario utilizing a boundary-fitted FSI approach and LS-DYNA ICFD solver. Simulation results highlighted that FSI simulations successfully captured the main fluid dynamic patterns determined by bileaflet mechanical aortic valves, including the three-jets flow configuration. Velocity color maps and swirling strength contours demonstrated that fluid dynamic results are more affected by valve design than by its rotation angle. Furthermore, the comparison between the two different FSI approaches showed that both the operator split Lagrangian-Eulerian and the ICFD solver reproduce the same main fluid dynamic patterns. However, ICFD results show higher flow rate values and less synchronization in leaflets rotation. In conclusion, the operator split Lagrangian-Eulerian approach proved to be a flexible approach to investigate the effect of valve positioning and design on fluid dynamics. The ICFD solver features less flexibility in model definition and high sensitivity to mesh and time step settings, but it ensures convergence and accuracy of results at leaflets-blood interface.