Computational fluid dynamics (CFD) assessments in blood pumps (BPs) typically rely on constant boundary conditions, despite the dynamic nature of the cardiovascular system. Consensus on CFD methodologies for simulating BPs under realistic conditions is lacking, and qualitative validation against in vitro data, particularly regarding the dynamic pressure head-flow rate (HQ) hysteresis curve, remains absent. This study aims to validate a CFD framework capable of capturing HQ hysteresis. Time-varying boundary conditions were derived from a hybrid in vitro mock circulation. Computational fluid dynamics parameters, including boundary conditions (pressure versus mass flow), time step size (4°-72° per step), rotation modeling (frozen rotor versus sliding mesh), and turbulence modeling (none versus k - ω SST), were iteratively refined. Results were validated by assessing the overlap of simulated and measured HQ hysteresis using the Jaccard Index (JI). Dynamic HQ hysteresis was captured only with mass flow boundary conditions, not with pressure boundary conditions (JI = 0.62 vs. 0.37). Time step size, rotation modeling (except for frozen rotor without averaging), and turbulence modeling had minimal effect on HQ hysteresis but significantly influenced flow field resolution and computational efficiency. Critical parameters emerged in boundary conditions and motion modeling, whereas others involved trade-offs between flow field accuracy and computational cost.
Klocker et al. (Tue,) studied this question.