Numerical uncertainty quantification of the hemodynamic parameters of a cerebral aneurysm using the Grid Convergence Index (GCI) method

Resumo

A cerebral aneurysm consists of a pathological dilation of an intracranial artery, and its rupture is associated with a lethality rate of up to 50%.  Computational Fluid Dynamics (CFD) has been widely used to investigate rupture risk factors by analyzing hemodynamic parameters of blood flow, especially the wall shear stress (WSS) and its time variation, quantified by the oscillatory shear index (OSI).  A critical aspect of employing CFD is carefully managing discretization errors intrinsic to the computational mesh.  Quantifying these errors and ensuring reliable results necessitates mesh convergence verification, which evaluates the solution's sensitivity to mesh refinement.  A major challenge in CFD simulations of aneurysms is finding a mesh refinement level that balances result accuracy with computational cost.  While most literature on the topic includes mesh refinement analysis, the uncertainty of numerical results, crucial for proper results interpretation, is rarely reported.  A widely used approach in the literature to estimate discretization error and numerical uncertainty is the Grid Convergence Index (GCI).  This study applies the GCI methodology to quantify numerical uncertainty from discretization in aneurysm blood flow simulations.  The flow was solved using the Finite Volume Method implemented in the free, open-source software package OpenFOAM (Open Source Field Operation and Manipulation).  We evaluated  uncertainty  by using three meshes generated  with the OpenFOAM package.  Numerical uncertainty was estimated for average WSS and OSI on the aneurysm sac surface to assess the rupture risk of a patient-specific aneurysm.