A COMPARATIVE EVALUATION OF SBES and SST k-ω TRANSITION TURBULENCE MODELS IN CARDIOVASCULAR COMPUTATIONAL FLUID DYNAMICS: PATIENT-SPECIFIC LEFT VENTRICULAR ASSIST DEVICE FLOWS AND FDA'S BENCHMARK NOZZLE

Left ventricular assist devices (LVADs) are mechanical pumps that partially or fully support the function of the left ventricle. However, around 10% of patients are estimated to experience a stroke within a year of implantation. Computational fluid dynamics (CFD) enables detailed analysis of blood flow and associated hemodynamic parameters, offering insights into stroke risk. Yet, in LVAD-supported aortas, it remains unclear whether turbulence modeling is necessary to accurately capture hemodynamics. To address this, we first assessed the performance of the Stress-Blended Eddy Simulation (SBES) model in the FDA’s benchmark nozzle at a Reynolds number of 2000, and then compared it against SST Transition and laminar models in three patient-specific LVAD–aorta geometries using Ansys Fluent. For this purpose, patient-specific data and CT-derived geometries were provided by Erasmus MC. Results from the FDA nozzle indicated that SBES overestimated jet breakdown location, although axial velocities and shear rates remained consistent with experimental data at certain locations, the values at jet breakdown region differ greatly, however longer simulations improve agreement. Across the patient specific models, all three CFD approaches produced similar peak wall shear stress and velocity distributions even in cases with high inlet flow suggesting predominantly laminar flow. These results indicate that, while SBES can capture small-scale flow instabilities, simpler laminar or transition models may suffice for routine LVAD–aorta simulations, thereby reducing computational cost.