Methodology of micro-PIV investigation of blood flow in channels with micro-structures

Abstract

As the medical devices industry advances in the developing and production of mechanisms and equipment for fighting disorders of the cardiopulmonary system, new engineering problems arise. Such devices are extra corporeal membrane oxygenators (ECMO), as well as external ventricular assist devices (VADs). Since, their work includes blood handling, it is important to assess how the blood is affected, while flowing through such devices. Particle image velocimetry (PIV) is an approach to observe fluid flows in specific conditions and obtain important knowledge regarding the mechanical properties of the flow, e.g. velocity fields, shear strain/stress, etc. With the help of PIV it is possible to obtain some of the information regarding the blood flow through assistive medical devices. Moreover, predictions made with computational fluid dynamics (CFD) could be validated. In this work a methodology for derivation of reliable data from micro-PIV experiment is established in series of experiments with simple rectangular channel. The best configuration of the system was chosen through tests with different settings, such as different interrogation windows, working fluids, particles and concentrations of particles. The results were compared to CFD simulations. After the best conditions of the setup were found, a channel was produced to replicate the cross-section of ECMO-membrane. This channel was used to observe the transverse flow of transparent fluid, mixture of xanthan gum, sucrose and water, with viscosity similar to this of the blood. The velocity fields in between the acrylic rods, substituting the fibres, were observed at 4 different flow rates. The mean absolute percentage error (MAPE) between the experiments and respective CFD simulations was estimated between 12 and 17 %. The experimental results were further used to quantify the hemolysis between two fibres. Furthermore the blood damage was compared to theoretically estimated Sherwood number for 6 different velocities. Eventually, an optimal velocity in between the fibres was proposed, in the meaning of best mass transport on the cost of minimum blood damage. As a last experiment, a channel with real fibres, attached parallel to the flow, was prepared. The velocity profile in the middle was compared to velocity profile from a CFD simulation of a channel with the same geometry, where the fibres are simulated as rigid bodies. The resulting MAPE was 4%. Additionally, concepts for improvement of the methodology were included.