A fully implantable, axial flow blood pump has been developed in our hospital. Both in vitro and in vivo tests showed that the hemolysis and thrombus characteristics of the pump were in an acceptable but not in an ideal range. Computational fluid dynamics (CFD) and in vitro test results showed that the pump worked at off-design point with a low hydraulic efficiency; CFD analysis also showed regions of reverse flow in the diffuser, which not only decreases the pump's hydrodynamic efficiency, but also increases its overall potential for blood trauma and thrombosis. To make a blood pump atraumatic and nonthrombogenic, several methods were taken to reach a final model of the optimized blood pump using CFD, which decreased the rotational speed from 9,000 to 8,000 rpm, and the design flow rate from 11 to 6 L/min. More significantly, the flow separation and recirculation in the diffuser region were eliminated, which mitigated the traumatic and thrombus effect on blood. The acceptable results of the numerical simulations encourage additional in vitro and in vivo studies.

The gaps between the blades and the shroud (or hub) of an axial blood pump affect the hydraulics, efficiency, and hemolytic performance. These gaps are critical parameters when a blood pump is manufactured. To evaluate the influence of blade gaps on axial blood pump performance, the flow characteristics inside an axial blood pump with different radial blade gaps were numerically simulated and analyzed with special attention paid to the hydraulic characteristics, gap flow, hydraulic efficiency, and hemolysis index (HI). In vitro hydraulic testing and particle image velocimetry testing were conducted to verify the numerical results. The simulation results showed that the efficiency and pressure rise decreased when the gap increased. The efficiency of the axial blood pump at design point decreased from 37.1% to 27.1% and the pressure rise decreased from 127.4 to 71.2 mm Hg when the gap increased from 0.1 to 0.3 mm. Return and vortex flows were present in the outlet guide vane channels when the gap was larger than 0.2 mm. The HI of the blood pump with a 0.1 mm gap was 1.5-fold greater than that with a 0.3 mm gap. The results illustrated poor hydraulic characteristics when the gap was larger than 0.15 mm and rapidly deteriorated hemolysis when the gap was larger than 0.1 mm. The numerical and experimental results demonstrated that the pressure rise, pump efficiency, and scalar shear stress decreased when the gap increased. The HI did not strictly decrease with gap increases. The preliminary results encourage the improvement of axial blood pump designs.

This paper presents a numerical investigation of the potential aerodynamic benefits of using endwall contouring in a fairly aggressive duct with six struts based on the platform for endwall design optimization. The platform is constructed by integrating adaptive genetic algorithm (AGA), design of experiments (DOE), response surface methodology (RSM) based on the artificial neural network (ANN), and a 3D Navier-Stokes solver. The visual analysis method based on DOE is used to define the design space and analyze the impact of the design parameters on the target function (response). Optimization of the axisymmetric and the non-axisymmetric endwall contouring in an S-shaped duct is performed and evaluated to minimize the total pressure loss. The optimal ducts are found to reduce the hub corner separation and suppress the migration of the low momentum fluid. The non-axisymmetric endwall contouring is shown to remove the separation completely and reduce the net duct loss by 32.7%.

A throughflow model based on the time-marching finite volume approach is described in this paper. The governing equations are derived by circumferentially averaging the three-dimensional Navier-Stokes equations neglecting the circumferentially non-uniform and viscous terms. An inviscid blade force model similar to the Large-particle method is derived. The viscous blade force has been modeled by the distributed loss model. The convective fluxes of the governing equation are discretized with the Edward's low-diffusion flux-splitting (LDFSS) scheme. And a point-iterative Symmetric Gauss-Seidel (SGS) scheme is used in the temporal discretization. The throughflow model has been applied to the NASA Rotor 67 and a high-load transonic fan stage ATS-2. The reasonable good agreements with the experiments and the 3D viscous computations show the potential of the method.

A fully implantable axial left ventricular assist device LAP31 was developed for Chinese or other heart failure patients who need partial support. Based on the 5-Lpm total cardiac blood output of Chinese without heart failure disease, the design point of LAP31 was set to a flow rate of 3 Lpm with 100-mmHg pressure head. To achieve the required pressure head and good hemolytic performance, a structure that includes a spindly rotor hub and a diffuser with splitter and cantilevered main blades was developed. Computational fluid dynamics (CFD) was used to analyze the hydraulic and hemodynamic performance of LAP31. Then in vitro hydraulics experiments were conducted. The numerical simulation results show that LAP31 could generate a 1 to 8 Lpm flow rate with a 60.9 to 182.7 mmHg pressure head when the pump was rotating between 9,000 and 12,000 rpm. The average scalar shear stress of the blood pump was 21.7 Pa, and the average exposure time was 71.0 milliseconds. The mean hemolysis index of LAP31 obtained using Heuser's hemolysis model and Giersiepen's model was 0.220% and 3.89 × 10-5% respectively. After adding the splitter blades, the flow separation at the suction surface of the diffuser was reduced. The cantilever structure reduced the tangential velocity from 6.1 to 4.7-1.4 m/s within the blade gap by changing the blade gap from shroud to hub. Subsequently, the blood damage caused by shear stress was reduced. In conclusion, the hydraulic and hemolytic characteristics of the LAP31 are acceptable for partial support.

The ventricular assist pumps do not always function at the design point; instead, these pumps may operate at unfavorable off-design points. For example, the axial ventricular assist pump FW-2, in which the design point is 5 L/min flow rate against 100 mm Hg pressure increase at 8,000 rpm, sometimes works at off-design flow rates of 1 to 4 L/min. The hemolytic performance of the FW-2 at both the design point and at off-design points was estimated numerically and tested in vitro. Flow characteristics in the pump were numerically simulated and analyzed with special attention paid to the scalar sheer stress and exposure time. An in vitro hemolysis test was conducted to verify the numerical results. The simulation results showed that the scalar shear stress in the rotor region at the 1 L/min off-design point was 70% greater than at the 5 L/min design point. The hemolysis index at the 1 L/min off-design point was 3.6 times greater than at the 5 L/min design point. The in vitro results showed that the normalized index of hemolysis increased from 0.017 g/100 L at the 5 L/min design point to 0.162 g/100 L at the 1 L/min off-design point. The hemolysis comparison between the different blood pump flow rates will be helpful for future pump design point selection and will guide the usage of ventricular assist pumps. The hemolytic performance of the blood pump at the working point in the clinic should receive more focus.

This paper presents comparative numerical studies to investigate the effects of blade sweep on inlet flow in axial compressor cascades. A series of swept and straight cascades was modeled in order to obtain a general understanding of the inlet flow field that is induced by sweep. A computational fluid dynamics (CFD) package was used to simulate the cascades and obtain the required three-dimensional (3D) flow parameters. A circumferentially averaged method was introduced which provided the circumferential fluctuation (CF) terms in the momentum equation. A program for data reduction was conducted to obtain a circumferentially averaged flow field. The influences of the inlet flow fields of the cascades were studied and spanwise distributions of each term in the momentum equation were analyzed. The results indicate that blade sweep does affect inlet radial equilibrium. The characteristic of radial fluid transfer is changed and thus influencing the axial velocity distributions. The inlet flow field varies mainly due to the combined effect of the radial pressure gradient and the CF component. The axial velocity varies consistently with the incidence variation induced by the sweep, as observed in the previous literature. In addition, factors that might influence the radial equilibrium such as blade camber angles, solidity and the effect of the distance from the leading edge are also taken into consideration and comparatively analyzed. ª 2015 Production and hosting by Elsevier Ltd. on behalf of CSAA & BUAA.

Performance of the pump in vitro and in vivo was considered sufficient for an LVAD. Further design improvements are being undertaken in terms of hemolysis and thrombosis to improve the biocompatibility of the pump.