Abstract
Although the advancement and understanding of blood-contacting devices (ventricular assist devices, artificial hearts and mechanical heart valves) and diseases (arterial and aortic stenosis and abdominal aortic aneurysms) has been significant in recent years, yet blood damage remains an issue. Fluid dynamic stresses can contribute to the damage of the blood cells, resulting in haemolysis and thrombosis in extreme cases. Along with generating local flow disturbances, the Reynolds number (Re) can lie within the transitional and turbulent flow regimes, normally uncommon for blood flow. The complexity of the flow fields present in these devices and diseases are amplified by blood’s rich rheology of shear-thinning and viscoelastic properties, largely influenced by the interaction of red blood cells (RBCs) and plasma proteins. This thesis aims to understand the influence of blood’s shear-thinning and viscoelastic properties on transitional flow fields through a combination of high-fidelity computational fluid dynamics (CFD) simulations and experimental rheometry.Shear-thinning blood rheology was first modelled using Large Eddy Simulation (LES) over a backward facing step (BFS) geometry showing that rheology influenced recirculation regions at both laminar and transitional Reynolds numbers. Turbulence statistical analysis demonstrated that shear-thinning blood transitioned at later Re, while for Newtonian blood rheology, this occurred at earlier Re, consistent with existing experimental observations in the literature. Using a spatially averaged definition of Re explained some differences and resulted in a transition closer to the Newtonian model.
Direct numerical simulation (DNS) using the Phan-Thien-Tanner (PTT) viscoelastic model showed that turbulent structures at the wall were influenced by elasticity components with considerable differences found when compared to Newtonian blood rheology. Possible drag reduction features were distinguishable in the flow fields, which could explain the experimental delays in the blood. Studying the BFS geometry but with DNS, and the PTT model, resulted in numerical instabilities at transitional Re due to the decoupling of polymer stresses and velocity fields.
Experimental measurements of blood analogues and equine blood using rotational rheometry identified a primary and secondary flow regime; defined by a change in relative torque against a modified Reynolds number, ˜ R. As haematocrit increased in equine blood, the transition to secondary flow occurred at lower ˜ R, even when accounting for shear-thinning effects. An indication that particle concentration and viscoelasticity of RBCs influenced the development of secondary flow. The flow field was then explored numerically for Newtonian and shear-thinning blood rheologies. Turbulent flow was established at higher ˜ R for cone-plate flow, while for parallel plate, turbulent flow occurred at lower ˜ R, with no differences between rheologies. An eddy forming at the periphery of the geometries grew in radial and azimuthal directions with turbulent flow defined by the break up of this eddy into multiple vortices, with small differences between rheologies.
The results gathered in this thesis have paved the way for further investigations into the impact of blood rheology in secondary, transitional and turbulent flows. This work has performed a comprehensive numerical investigation of non-Newtonian blood flow over a backward facing step, an important consideration for understanding flow separation in blood-contacting devices. The study of a non-linear viscoelastic blood model at transitional and turbulent Re has not been studied in depth before, and the results highlighted differences compared to Newtonian blood rheology; it should be investigated further. Additionally, the differences found when studying secondary flows of equine blood at various haematocrits demonstrate that something is missing in current blood models that can underestimate the secondary flow field in numerical simulations. While the differences might not be significant but should be considered in devices where these flows are present; blood-contacting devices.
Looking to the future, high fidelity turbulence modelling is required to model the smallest scales. Still, the complexity of implementing this, along with high fidelity rheological models such as the PTT model, is excessive. While the viscoelastic properties of blood may not be as dominant as the shear-thinning properties; the models presented show viscoelastic blood at turbulent and transitional Re has clear differences in the mean and instantaneous flow field. Furthermore, how we define Re needs to be established for fluids exhibiting shear-thinning and viscoelasticity properties to identify the exact differences at transition. Nevertheless, accurate modelling (turbulence and rheological) is required to understand the mechanisms for flow-induced blood damage at the smallest scales.
| Date of Award | 20 Sept 2022 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Katharine Fraser (Supervisor), Andrew Cookson (Supervisor) & Richie Gill (Supervisor) |
Cite this
- Standard