Design and Analysis of a Bearingless Permanent-Magnet Motor for Axial Blood Pump Applications

Because of their high power density and compact size, permanent-magnet (PM) motors have been commonly used to drive rotary blood pumps (RBPs), which are focused on the treatment of end-stage heart failure or as the bridge to a heart transplant. In this paper, a bearingless PM motor has been proposed for axial blood pump applications. The finite-element method (FEM) is used to predict the electromagnetic characteristics of the designed motor with improved performance. Two topologies are investigated, namely the integral-slot and distributed-windings method and the fractional-slot and double-layer concentrated windings method. Both motors are analyzed and optimized. FEM reveals that, compared with the integral-slot motor, the fractional-slot motor offers significantly enhanced performance, including reduced cogging torque, improved back electromotive force (back EMF), and decreased magnetic flux leakage. Finally, hydraulic experiments have been conducted in a mock-circulation loop to validate the feasibility of the designed motor for an axial blood pump. The results show that the fractional-slot bearingless PM motor can drive the RBP to produce physiological blood flow with reasonable efficiency.


I. INTRODUCTION
In the past few decades, rotary blood pumps (RBPs) have greatly improved the treatment of end-stage heart failure, either as the bridge to heart recovery or to a heart transplant. The increasing use of RBPs is related to multiple positive attributes, including reduced size, improved durability, and enhanced survival rate compared with volume-displacement pulsatile pumps [1]- [4]. Biocompatibility and durability are the two main concerns for the long-term or permanent use of RBPs. The former refers to controlling blood trauma to a permissible level, and the latter refers to solving the problem of life-limiting wear on mechanical bearings in the rotor. Blood trauma includes hemolysis and thrombosis. Hemolysis is defined as the hemoglobin disassociation into plasma for the break of membranes of red cells, while thrombosis refers to breaks [5], [6]. In order to reduce blood trauma, many new The associate editor coordinating the review of this manuscript and approving it for publication was Victor Hugo Albuquerque . designs (e.g., streamlined design), as well as new structures for RBPs, have been put forward by researchers [1], [4], [7]. The bearing problem has been extensively investigated and non-contact bearing systems have been developed for longterm support of end-stage heart failure patients [7]- [11].
These non-contact bearing systems include an electromagnet / position-sensor bearing system, hydrodynamic bearing system, and a magnetically levitated bearing system. In an electromagnet system, such as HeartMate III and DuraHeart, the rotor is suspended using electronic position control and electromagnets. They offer the advantage of no life-limiting wear due to no contact between the bearing surfaces and low blood shear stress due to relatively large clearances between the rotating element and the housing. However, the complexity of the system increases instability and unreliability. For a hydrodynamic system, such as HeartWare HVAD, the rotor is suspended by fluid forces in thin blood films separating the rotor and pump housing based on the relative motion of surfaces. The hydrodynamic bearings are simple and reliable, but the load-blood film is prone to high shear stress, which theoretically contributes to high hemolysis. In a magnetically levitated bearing system, the PMs are generally used in combination with hydrodynamic or electromagnetic bearing elements for stabilization in one or more directions of movement. For the complexity of this type of motor, it is difficult to obtain miniaturization for some special RBPs with the required overall size.
The purpose of this paper is to propose a new bearingless PM motor for an implantable axial blood pump with the overall size required (delivering the blood directly from the left ventricle to the aorta), in which both biocompatibility and durability are considered to provide long-term circulatory support for end-stage heart failure patients. The designed motor offers promising mechanical integrity and a compacted structure, which is described in Section II. In Section III, the finite element method (FEM) is used to analyze the electromagnetic characteristics and to reduce the cogging torque of the designed motor. In Section IV, hydraulic experiments are conducted to validate the feasibility of the motor in the axial blood pump.

II. PROPOSED MOTOR FOR AN AXIAL BLOOD PUMP A. BEARINGLESS PM MOTOR
The axial blood pump using the proposed bearingless PM motor is shown in Fig. 1. This pump consists of an inlet guide vane, impeller, outlet guide vane, motor, and housing. The outer diameter is no more than 33 mm. Its rotor incorporates the impeller without mechanical bearings and its stator is inset with the windings and armature teeth. When the motor windings are excited by an electric current, a rotating magnetic field is generated. The rotor magnets and the impeller begin to rotate and then impart energy to the blood to increase arterial blood flow and pressure.
There is no bearing devised for the durability consideration. Therefore, there is no mechanical friction and heat generation when there is no contact between the stator and the rotor in the blood pump. It is expected that the risk of thrombosis, which is always forming along the bearing in the blood pump, will be reduced greatly and the problem of mechanical wear will be solved.
The proposed bearingless motor with a 12/4 slot is shown in Fig. 2. The motor consists of a stator iron core, stator windings, four-pole rotor magnets distributed in a cylindrical rotor core, and a double-layer sleeve. The PMs are magnetized radially and inserted between the sleeve and the iron rotor core. The inner layer of the sleeve is made of iron to protect the PMs and the outer layer is a polytetrafluoroethylene film with a thickness of no more than 0.5 mm. When the rotor rotates, a thin blood film forms between the polytetrafluoroethylene layer and the inner surface of the stator. It works as a hydrodynamic bearing to suspend the rotor without complex control and prevents friction between the rotor and the stator in the case of motor instability. On the other side, the special attribute of the polytetrafluoroethylene film helps to reduce the friction even at startup and shutdown. As for the high shear stress of hydrodynamic bearings, which contributes to high-level hemolysis [1], our experimental data reveal that the hemolysis level of the PM axial blood pump is within the permissible level. This is achieved by optimizing the impeller design and pump structure, which is presented in [12]. One reason is that hemolysis is not only related to high shear stress, but also to the residence time of blood cells in the highshear-stress areas. For the high rotating speed of the axial blood pump, the residence time is relatively short, and thus, hemolysis remains at a low level. Another reason is that the optimized pump design could affect the flow patterns that are directly related to hemolysis.

B. REDUCTION OF COGGING TORQUE
High cogging torque always results in torque ripple, and thus decreases the stability of the motor [14], [15], [16]. Cogging torque is related to many motor parameters. In this section, two topologies are designed to analyze the influence of slots and windings on cogging torque [17], [18]. The topology with 12/4 integral slot and distributed windings has been shown in Fig. 2, and the new one with 6/4 fractional slot and concentrated windings is shown in Fig. 3. Their parameters are listed in Table 1. Model 1 represents the integral-slot motor and mode 2 represents the fractional-slot motor.

III. FINITE-ELEMENT ANALYSIS
FEM is used to investigate electromagnetic characteristics including the magnetic field distribution, back EMF, cogging VOLUME 8, 2020  torque, and magnetic flux density [19], [20], [21]. The motor can be modeled as an axisymmetric model in the 2-D field simulator. Fig. 4 shows the magnetic field distributions of both motors at no load. It can be seen from Fig. 4A that the magnetic flux leakage in the domain near the stator tooth top in the integralslot motor is attributed to the structure of the magnetic pole and slot. When the PM flux passes the gas gap, the stator armature, and then back to the PM, forming a circle, the poles reach the top of the stator armature teeth through the gas gap. Thus, flux leakage appears near the stator tooth. Meanwhile, no evident magnetic flux leakage is found in the fractionalslot and concentrated-windings motor in Fig. 4B. Its magnetic flux density is up to 1.39 T in the stator yoke, which is no more than the armature saturated flux density of 1.5 T. Fig. 5 compares the magnetic flux densities of both motors in a half circle on the side of the rotor sleeve, and both averages reach 0.43 T. The fractional-slot model is nearly trapezoidal without any evident fluctuations. Fig. 6 compares the back-EMF waveforms of the motors, and both amplitudes reach 10 V. Compared with the saddle shape in the model with an integral slot and distributed windings, the back-EMF of the fractional-slot motor is relatively smooth, without evident fluctuations. The proposed brushless  DC motor is driven by a square wave current. When the width of the flat top portion of back EMF is not enough, torque ripples, which eventually causes vibration and noise. The width of flat top in the back-EMF increases by nearly 30% in the fractional-slot motor will reduce torque ripple and make the motor more stable [15], [16].
The cogging torque of both motors is compared in Fig. 7. The maximum cogging torque of the fractional-slot model reaches 4 mN · qm, much lower than the maximum value of 28 mN · qm of the integral-slot motor. The fluctuation of the cogging torque in the fractional-slot motor decreases greatly, with an average of 10% that of the integral-slot motor. Cogging torque contributes to torque ripple in the motor, which causes mechanical vibrations of the motor. In the integral-slot motor, there are twelve slots and four poles. Therefore, three slots correspond to one pole, or six slots correspond to one pole-pair. The total cogging torque is the sum of cogging torque produced by each slot. In the integral-slot motor, the cogging torque waveform from every slot has the   same amplitude and phase with each other. However, in the fractional-slot motor, there are six slots and four poles. So the slots cannot distribute symmetrically in the magnet field, and thus the phase of cogging torque produced in each pole will vary. For cogging torque is vector, the sum of cogging torque produce in each slot decreases compared to the integral-slot motor. Fig. 8 shows the fast Fourier transform (FFT) analysis of the cogging torque of the two models. The cogging torque of both models is much higher at the 3rd, 6th, and 9th harmonic orders than at others. Compared with the integral-slot model, the fractional-slot model has significantly reduced cogging torque, which verifies the decrease of cogging torque in the fractional-slot model.

IV. EXPERIMENTAL RESULTS
In order to validate the proposed motor for axial bloodpump applications, an experimental model has been built   in this work. As shown in Fig. 9, the experimental mock circulation loop consists of an artificial atrium, HP/M1205A physiological monitor, axial blood pump using the proposed motor, Transonic T110 flow meter, and control system. The control system can regulate the rotating speed of the motor, which includes a drive, TMS320F2812 control chip, and filter. A water-glycerin mixture of 30% volume glycerin is used as the hydraulic test fluid with a viscosity of 0.00036 Pa · s (similar to human blood at 37 • C). Fig. 10 shows the characteristic curves of the axial blood pump at different rotating speeds based on the hydraulic experiments in the mock circulation loop. In the characteristic curves, H represents the differential pressure and Q represents the flow rate of the pump. The blood pump hydraulic performance reveals that the pump driven by both motors can produce physiological blood flow and pressure to assist a child patient or an adult patient with mild heart failure at a rotating speed of more than 7500 rpm [22]. The blood pump driven by the integral-slot motor or by the fractional-slot motor can produce a differential pressure of 51 mmHg or 57 mmHg, respectively, with the same flow rate of 2 L/min and rotating speed of 9500 rpm. Compared with the integral-slot motor, the average differential pressure of the blood pump with fractional-slot motor increases by nearly 7.2% at 9500 rpm and with a flow of 2 L/min.
Compared with the integral-slot motor, the average differential pressure of the fractional-slot motor blood pump increases by nearly 7.2% at 9500 rpm. However, the surface material of the non-contact bearing and stability at startup and shutdown of the motor are among the important issues to be solved in future work.

V. CONCLUSION
In this paper, a bearingless PM motor has been proposed for an axial blood pump with compact size. FEM is used to predict motor performance. Hydraulic experiments have been conducted to validate the feasibility of the motor for axial blood-pump applications. Both the numerical and experimental results verify that the fractional-slot topology can reduce the size of the pump, significantly decrease the cogging torque, and enhance the reliability. The proposed motor is feasible for the axial blood pump, offering the merits of a compact structure, high stability.
YINGFEI ZHU received the B.Sc. degree in electrical engineering from Nanhang Jincheng College, Zhenjiang, China, in 2017. She was with Jiangsu Unversity. Her research interest includes modeling and analysis of artificial hearts.
HAO WANG received the B.Sc. and M.Sc. degrees in electrical engineering from Jiangsu University, Zhenjiang, China, in 2004 and 2007, respectively, where he is currently pursuing the Ph.D. degree in fluid machinery engineering. Since 2004, he has been with the School of Electrical and Information Engineering, Jiangsu University, where he is also a Lecturer. His areas of interest includes motor drive and control. He has authored and co-authored over 20 technical articles in these areas.
DONG ZHAO received the B.Sc. degree in safety engineering from Jiangsu University, Zhenjiang, China, in 2000. He is currently pursuing the M.S. degree in mechanical engineering with Texas Technology University, Lubbock, TX, USA. His research interests include mock circulation loops and research of tricuspid regurgitation. VOLUME 8, 2020