Non-Newtonian and Turbulent Perspectives in Computational Cerebral Hemodynamics

Intracranial aneurysms can cause hemorrhagic strokes, with potentially fatal consequences for the patient if they rupture. Patients exhibiting increased rupture risks may require endovascular interventions to prevent cerebral hemorrhage. Clinical imaging, from which morphological information is extracted, is critical in assessing risks and to guide interventions. In this work, we use reconstructed vessels from clinical imaging to provide additional information using computational hemodynamic analyses of the cerebral vasculature networks. It is not well understood what level of modelling is required to accurately resolve cerebral hemodynamics. In our work, we conduct full convergence analyses of several geometries with different blood rheology and turbulence models. Unresolved DNS, RANS simulations and LES are conducted. We show that overall grid convergence is difficult to achieve, but achievable in regions of interest. Our results demonstrate that in regions of increased stress, blood's viscosity may significantly decrease due to its shear thinning properties. This is exacerbated for vessels with large lateral aneurysms. Also, turbulence modelling is demonstrated to be required when feeding vessels connect to terminal aneurysms. Based on this framework, we will conduct an extensive investigation of the effects of blood rheology and turbulence modelling in 100+ patient-specific cases to provide better guidelines and inform future modelling efforts in cerebral hemodynamics.