Aortic valve disease (AVD) is one of the most common cardiovascular diseases worldwide and is associated with high morbidity and mortality. Blood flow in the aorta is often assumed laminar, however aortic valve diseases and valve prostheses may alter flow in the aorta causing significant deviations from the normal aortic haemodynamics; which is already complex. This complexity combined with the pumping action of the heart creates conditions capable of producing laminar to turbulence transition, and relaminarisation all within a single cardiac cycle. There are limited studies considering turbulence effects in diseased and prosthetic valves in patient-specific settings and consequently, turbulence effects are not yet well understood.

The main aim of this thesis is to evaluate turbulence effects in various aortas with healthy, diseased and prosthetic aortic valves using scale-resolving simulations. Large-eddy simulation (LES) was identified as an appropriate numerical approach based on its capabilities in modelling laminar, transitional and turbulent flows. Magnetic resonance imaging (MRI) and 4D flow MRI were used to reconstruct aortic geometries and derive physiological boundary conditions; allowing simulations to be performed under patient-specific anatomical and flow conditions. The LES approach was evaluated in both idealised and patient-specific settings. Simulation results obtained on a well-established idealised medical device were in excellent agreement with prior interlaboratory experimental results. The LES methodology was then adapted for patient-specific applications and results were compared with in vivo measurements obtained with 4D flow MRI. The patient-specific LES model was able to reproduce aortic flow patterns and velocities to a good degree of accuracy.

After validation, the LES methodology was applied to various patient-specific aortas with healthy, diseased and prosthetic aortic valves. Turbulence characteristics were evaluated to understand the conditions under which flow transitions to turbulence; and to evaluate effects of turbulence on local haemodynamic indices. It was found that turbulence was produced in the ascending aorta and/or the proximal descending thoracic aorta, and turbulence production was dependent on both local anatomy and valve physiology. Disturbances were present in all cases, even in the presence of a healthy aortic valve. These findings indicate that turbulence effects should not be neglected in numerical models of the aorta. Finally, a comparative study into different computational approaches was performed on an aorta with valve stenosis to evaluate model capabilities and accuracy. While all simulations could accurately predict velocities and flow patterns throughout the aorta, there were differences in other haemodynamic parameters including wall shear stress and viscous energy loss. Compared to LES, lower resolution simulations were inadequate for prediction of turbulence-based parameters.