The axial blood pump with a magnetically suspended impeller is superior to other artificial blood pumps because of its small size. In this article, the distributions of velocity, path line, pressure, and shear stress in the straightener, the rotor, and the diffuser of the axial blood pump, as well as the gap zone were obtained using the commercial software, Fluent (version 6.2). The main focus was on the flow field of the blood pump. The numerical results showed that the axial blood pump could produce 5.14 L/min of blood at 100 mm Hg through the outlet when rotating at 11,000 rpm. However, there was a leakage flow of 1.06 L/min in the gap between the rotor cylinder and the pump housing, and thus the overall flow rate the impeller could generate was 6.2 L/min. The numerical results showed that 75% of the scalar shear stresses (SSs) were less than 250 Pa, and 10% were higher than 500 Pa within the whole pump. The high SS region appeared around the blade tip where a large variation of velocity direction and magnitude was found, which might be due to the steep angle variation at the blade tip. Because the exposure time of the blood cell at the high SS region within the pump was relatively short, it might not cause serious damage to the blood cells, but the improvement of blade profile should be considered in the future design of the axial pump.

The centrifugal blood pump with a magnetically suspended impeller has shown its superiority as compared to other artificial heart pumps. However, there is still insufficient understanding of fluid mechanics related issues in the clearance gap. The design nature of the pump requires sufficient washout in the clearance between the impeller and the stationary pump housing inner surface. In this study, numerical simulations were carried out to investigate the flow fields in the gap of the Kyoto-NTN centrifugal blood pump. The flow patterns in the gap region of the blood pump were presented and regions of high and low velocity were identified. It was found that the radial velocity of the blood in the gap was closely related to the pressure distribution at the exit of the impeller, both the highest pressure gradient and the highest radial velocity in the gap occurred at an angular position of 170 degrees . The mass flow rate in the gap was estimated to be 25.2% of the pump outflow, which is close to the measurement results of a five times enlarged test pump. The wall shear stresses on the gap surface were found to be over 21 Pa and below 300 Pa, which is correspondingly higher than the threshold of thrombi formation and is lower than the shearing threshold of red blood cells. Comparison of the 1 : 1 simulation model with the measurement results on a five times enlarged test pump indicates that there are some differences in the resulting radial velocity distributions in the gap and thus the washout mechanism. Two symmetrical high washout regions at both the cutwater and splitter plate were observed in the simulation instead of a single washout region at the splitter plate found in the experimental study. This may be due to the scaling effect of the enlarged test pump; also the medium used in the experiment is different from the simulation.

Implantable ventricular assist devices have been regarded as a promising instrument in the clinical treatment of patients with severe heart failures. In this article, a three-dimensional model of the Kyoto-NTN magnetically suspended centrifugal blood pump was generated and a computational fluid dynamics solution of the inner flow field of the pump including the static pressure distributions, velocity profiles, and the shear stress distributions of the blood was presented. The results revealed that reverse flow generally occurred in the impeller blade channels during the operation of the pump, due to the imbalance of the flow and the pressure gradient generated in the blade channels. The flow pattern at the exit of the blade channels was varying with its angular positions in the pump. The reverse flow at the exit of the impeller blade channels was found to be closely related with the static pressure distribution in the volute passage. Higher pressure in the volute caused severe backflow from the volute into the blade channels. To clarify the effects of a moving impeller on the blood, shear stresses of the blood in the pump were investigated according to the simulation results. The studies indicated that at the beginning of the splitter plate and the cutwater, the highest shear stress exceeded 700 Pa. At other regions such as the inlet and outlet of the impeller blade channels and some regions in the volute passage, shear stresses were found to be about 200 Pa. These areas are believed to have a high possibility of rendering blood trauma.

Generally, there are two types of impeller design used in the axial flow blood pumps. For the first type, which can be found in most of the axial flow blood pumps, the magnet is embedded inside the impeller hub or blades. For the second type, the magnet is embedded inside the cylindrical impeller shroud, and this design has not only increased the rotating stability of the impeller but has also avoided the flow interaction between the impeller blade tip and the pump casing. Although the axial flow blood pumps with either impeller design have been studied individually, the comparisons between these two designs have not been conducted in the literature. Therefore, in this study, two axial flow blood pumps with and without impeller shrouds were numerically simulated with computational fluid dynamics and compared with each other in terms of hydraulic and hematologic performances. For the ease of comparison, these two models have the same inner components, which include a three-blade straightener, a two-blade impeller, and a three-blade diffuser. The simulation results showed that the model with impeller shroud had a lower static pressure head with a lower hydraulic efficiency than its counterpart. It was also found that the blood had a high possibility to deposit on the impeller shroud inner surface, which greatly enhanced the possibility of thrombus formation. The blood damage indices in both models were around 1%, which was much lower than the 13.1% of the axial flow blood pump of Yano et al. with the corresponding experimental hemolysis of 0.033 g/100 L.