VALIDATION OF CFD RESULTS USED FOR HEMOLYSIS PREDICTION

Abstract

Computational fluid dynamics (CFD) analyses have played important roles in developing heart valves, blood pumps, and other mechanical cardiovascular support devices. Increasingly, investigators are attempting to predict hemolysis directly from their CFD simulations of the devices they are developing. This requires a hemolysis prediction model usually derived from a power-law equation of shear stress and exposure time, and a shear field estimate based on various assumptions underlying the CFD analyses. However, most of these investigators either have no experimental data at all to validate their CFD simulations or only have global performance measurements for comparison. Our experience with one of our blood pumps has shown that for the same pump system, slight changes of the meshing parameters will result in pressure rise predictions that are all confirmed by experimental performance measurements. However, the hemolysis prediction varies by 36%. With another blood pump we developed, two simulations using the same meshing parameters but two different turbulence models again accurately predicted the pressure rise, but the hemolysis prediction varied by an order of magnitude. With either pump, different simulations resulted in different pressure contours in local regions, even though gross performance was the same. Our results indicate that it is inadequate to validate CFD simulations with pressure-flow-power data. It is necessary to validate CFD simulations with detailed flow field measurements, before a hemolysis prediction can be made with useful accuracy.