A CFD model for the prediction of haemolysis in micro axial left ventricular assist devices

This paper describes the development and application of a computational model based upon Computational Fluid Dynamics (CFD) software simulation technology to predict haemolysis in micro Left Ventricular Assist Devices (mu LVAD). A CFD model, capturing the full three dimensional geometry of the device, together with an explicit representation of the rotating machinery based upon a rotating reference frame, is solved transiently. Mixed meshes with the order of a million elements are required to resolve the flow adequately and so to enable solutions in a reasonable time (e.g., 3 h) the model is solved on a high performance parallel cluster. Haemolysis is a measure of damage occurring in the blood and is conceived as accumulating as it passes through parts of the device where it encounters high shear forces. As such, the haemolysis model is based upon tracking the behaviour of particles released at the inlet throughout the flow domain and calculating the damage accumulated by each individual particle as it traverses the device. In order to ensure the model predictions of haemolysis are noise free from a statistically significant perspective then it is demonstrated that the number of particles to be tracked must exceed 20000 in any simulation experiment. Comparisons with experimental data from a companion paper demonstrate the effectiveness of the CFD simulation embedding the haemolysis model. (C) 2012 Elsevier Inc. All rights reserved.