Share this article on:

The MVAD Pump: Motor Stator Core Loss Characterization

Mesa, Kelly J.; Ferreira, Antonio; Castillo, Samir; Reyes, Carlos; Wolman, Justin; Casas, Fernando

doi: 10.1097/MAT.0000000000000180
Adult Circulatory Support

Investigation of the miniature ventricular assist device (MVAD) pump motor stator core loss behavior was conducted. During operation, the ferromagnetic core in the pump’s motor is magnetized by alternating magnetic fields, which, in turn, create intrinsic energy losses in the core material; these losses are known as core losses. A core loss fixture and a method to characterize the magnetic behavior of the MVAD pump stator over a range of frequencies were developed. The MVAD pump motor design features a three phase brushless DC stator with ferromagnetic laminations and copper wire windings arranged in a six slot configuration. The stator’s magnetic behavior is important because its core magnetic losses impact pump system efficiency. A system to measure the core loss of MVAD pump stators was developed using a custom core loss fixture consisting of 16 copper wire turns wound in a closed loop geometry bundle; the stator under test was then placed within this bundle. The instrumentation consisted of a signal generator, a power amplifier, and a power analyzer. Power analyzer parameters of current, voltage, and power were collected for several runs with a sinusoidal frequency sweep of 0 to 50 kHz; data were collected for the fixture with and without stators. The magnetic losses inherent to the fixture were characterized independently as a baseline presenting a flat frequency response. The core loss power measurements of individual stators yielded a characteristic bandpass frequency response morphology with a peak core loss found around 2.3 to 2.5 kHz. In conclusion, this method could be used to describe the transfer function of the stator’s core magnetic behavior. It also has the potential to be used for future motor evaluation and for investigation of core loss performance variability between different stators during manufacturing operations. CAUTION: Investigational device. Limited by United States law to investigational use.

From Advanced Product Development, HeartWare Inc., Miami Lakes, Florida.

Submitted for consideration March 2014; accepted for publication in revised form November 2014.

Disclosure: The authors have no conflicts of interests to report.

Correspondence: Fernando Casas, PhD, Advanced Product Development, HeartWare, Inc., 14420 NW 60th Av., Miami Lakes, FL 33014. Email: fcasas@heartwareinc.com.

Recent advancements in mechanical circulatory support technologies continue to focus on miniaturization and ease of implantation of ventricular assist devices (VADs).1,2 The MVAD pump3 is a miniature axial flow VAD capable of providing full continuous output flow support. The main components of the MVAD pump include the pump’s inlet, flow tube, impeller, stator, housing, and volute outlet (Figure 1). The single-piece machined and magnetized MVAD pump impeller also serves as the motor rotor. The pump is implanted in the apex of the heart with the outflow graft anastomosed to the aorta completing the blood flow path through the implanted system. The MVAD pump operation is based on wide bladed rotor technology supported by a contactless passive hybrid hydraulic magnetic suspension system. The axial support of the pump’s impeller is created from the impeller-stator magnetic attraction, and the radial support arises from the hydrodynamic thrust bearings. The MVAD pump motor design features a three phase four pole brushless DC configuration, a pump motor stator core with ferromagnetic laminations, and a six slot configuration with copper wire windings (Figure 1).

During operation, the ferromagnetic core in the motor is magnetized by alternating or rotating magnetic fields creating intrinsic energy losses in the core material. These losses are known as core losses. From basic motor science, it is known that when an alternating current is applied to a ferromagnetic core it induces an electromotive force (EMF) not only in the coils but also in the core itself; this EMF will create eddy currents in the core. Furthermore, eddy currents are undesirable because they decrease the magnetic flux and create direct power losses. The primary way to decrease power losses induced by eddy currents is by constructing the motor core out of thin metal sheets or laminations rather than from a solid piece.4 In addition to eddy currents, the alternating current applied to the core will generate hysteresis losses. Hysteresis losses contribute to the total core loss.5 In rotating machines, hysteresis losses are caused by the changes in the magnetization vector with every rotation of the motor.

Although core loss measurement techniques5,6 and data are widely available for DC motors of several configurations, the published literature to date is very limited to nonexistent in the area of mechanical circulatory support. In the most recent third-generation VAD systems, such as the MVAD pump, the motor rotor doubles as the pump’s impeller; in this configuration, the stator is further away from the rotor when compared with a traditional DC motor. The narrow air gap, a critical design parameter in conventional motors, is replaced by a larger gap filled with circulating blood. Due to this design requirement, the motor stator specifications are now not only connected to electrical motor performance but include hematological constraints such as hemolysis potential and red blood cell pump transit time. Smith et al.7 emphasized the need for modified impeller geometries, usually with reduced impeller diameter when compared with most commercial pumps. The larger clearances are intended to avoid areas of high fluid stress that might induce stagnation, separation, or recirculation in the flow path. Neethu et al.8 presented an optimization design approach that maximizes motor efficiency and torque for an axial flow VAD. However, heating caused by eddy currents remains an issue to be addressed by the designers. In addition, prototyping tests of their design are still pending to verify the suggested approach.

The complexities of the magnetic interactions occurring in the stator while the VAD is operating propagate to the behavior of the core losses. Core losses are not only an important motor design requirement but, in the mechanical circulatory support application, they need to be characterized to ensure minimal impact in pump performance as a consequence of manufacturing variability. Inherently, there are several sources of core loss variability: core material, thickness tolerance, winding number of turns, geometry, and assembly processes. Core loss variability can impact the pump by creating an increase in pump power consumption; it can also affect motor control parameters including derived parameters such as flow estimates, as well as mechanical parameters, such as the impeller’s axial suspension. Therefore, a method to measure motor stator core loss is necessary to ensure consistent pump performance. In this application, the total stator core loss encompasses the eddy current losses and the hysteresis losses created by the external oscillating magnetic field.

The goal of the study was to develop a system to measure the core loss of MVAD pump stators to 1) understand and characterize the magnetic properties of the stator cores and 2) develop an automated system that could be used during manufacturing operations in particular at the receiving/inspection (RI) manufacturing stage.

Back to Top | Article Outline

Methods

The core loss measurement system was based on the diagram presented in Figures 2 and 3. The individual elements’ powers in the series circuit were measured and compared for a range of frequencies of interest. In particular, the ratio of fixture power to total input power provided information on fixture and stator core effects. In this application, the system had to be sensitive enough to be able to measure small differences in the performance of the stator cores. The peak of the measured fixture power was selected as a potential indicator of these differences among a population of stators. The range of the frequency sweep was chosen so that effects caused by the six step trapezoidal motor commutation strategy along with the pulse width modulation switching frequency could be captured.

Back to Top | Article Outline

System Architecture

A core loss measurement system including a custom-made fixture, a signal generator, a power amplifier, and a power analyzer was developed and tested. The fixture was built with 16 copper wire turns wound in a closed loop geometry bundle. This arrangement also mechanically held in place the stator under test. A magnetic circuit was completed once the fixture was powered. The sinusoidal excitation signal was provided by an Agilent Signal 33220A generator (Agilent, Santa Clara, California) and swept from 0 to 50 kHz. The signal was amplified using a high-fidelity AE Techron 7224 (AE Techron, Elkhart, Indiana) power amplifier and the amplifier’s gain adjusted to maintain a constant amplitude of 2 V at the amplifier’s output terminals (VI(t) on Figure 3). A Yokogawa WT1600 digital power meter (Yokogawa, Newnan, Georgia) was used to capture electrical parameters at various circuit locations.

During individual test runs, the power amplifier drove the circuit using a reference sinusoidal waveform produced by the signal generator. The power analyzer measured a set of parameters including root mean square (RMS) voltages, RMS currents, and active and reactive powers. The complete core loss test included a comprehensive frequency sweep. The system was fully automated via LabVIEW (National Instruments, Austin, TX).

Back to Top | Article Outline

Fixture Modeling and Parameter Identification

Physically, the fixture was built using an inductor element coupled to a resistor. Based on this premise, the fixture modeling was developed as two different R-L configurations as shown in Figure 3. The resistive element (Rf) in the model represents the resistance of the wires, whereas the inductive element (Lf) represents the inductance resulting from the geometry of the wire bundle that constitutes the fixture. The power analyzer allowed for the measurement of active and reactive power components of the given load, facilitating the estimation and validation of the R-L model’s parameters. The measured power components without a stator in the fixture were used to estimate the parameters in both series and parallel R-L configurations. The fixture’s model parameters were identified from the acquired input-output data set using Matlab (The MathWorks, Natick, Massachusetts).

The first model of the fixture’s impedance (Zf) consisted of having a resistance (Rf) and inductance (Lf) in series; Zf was represented as

where ω is the angular frequency. The second model included the same two Rf and Lf parameters arranged in parallel, with Zf represented as

Back to Top | Article Outline

Stator Core Loss Characterization

Following the characterization study on the performance of the fixture, tests were conducted to investigate the core loss characteristics of the MVAD pump stator. A representative set of 30 stators was randomly obtained from stock. These stators had completed the RI process and were all approved to be conformant to the drawing’s specifications. The system was calibrated after instrumentation warm-up and before each run so that the voltage measured at the output of the amplifier (input to the fixture circuit), VI(t) in Figure 3, with the fixture in place but without a stator core, was 2 ± 0.001 V at 1 kHz. The calibration ensured consistent measurements throughout the test series and mitigated potential variability introduced by cabling or any other physical variables. Each stator was installed in the fixture and run over a frequency range spanning from 0 to 50 kHz at a resolution of 10 Hz. The system’s data set was automatically captured to file via LabVIEW and included frequency, input and output voltages, fixture current, and input and output powers.

Back to Top | Article Outline

Results

Fixture Modeling

A parameter estimation study was conducted on both series and parallel fixture models (Figure 3). The mean square error (MSE) figure of merit was used to rank the model’s fit to the experimental data. The MSE for the series R-L model was 1.461 × 10−6, whereas the MSE for the parallel R-L model was 0.508. Based on the MSE and similarity to the physical and electrical configuration of the fixture, the series R-L model of the circuit was selected as the best fit to the fixture’s behavior. For the series R-L model, the values of the estimated parameters Rf (resistance of the fixture) and Lf (inductance of the fixture) were 0.509 ± 0.01 Ω and 32.79 ± 0.001 μH, respectively. The parameters for the parallel configuration Rf and Lf were 1.015 ± 0.01 Ω and 94.22 ± 0.001 μH, respectively.

The impedance behavior of the fixture from 0 to 5 kHz compared with the theoretical model is presented in Figure 4. These results provided confirmation that the series R-L model is capable of adequately reproducing the behavior of the physical core loss fixture in the frequency range of interest.

The fixture was further characterized by analyzing the power transfer function (output-to-input power ratio) versus frequency. The frequency response of the power transfer function was flat from 10 Hz to 5 kHz. Analysis of the fixture’s response indicated that the fixture did not interfere with the stator core loss measurements.

The core loss fixture behaved as a high pass filter, as expected for a series R-L circuit model. As frequency increases, the voltage at the fixture’s terminals tends to follow the input voltage. However, current decreases because the inductor’s impedance increases with frequency. Thus, the fixture’s power also decreases with increasing frequency as shown in Figure 5 (trace B).

Back to Top | Article Outline

Stator Characterization

Performance of the stators was evaluated by comparing the total power used by the fixture when the stator was in place. Figure 5 (trace A) shows a snapshot of 10 representative stators and their respective average and standard deviation power performance from 0 to 50 kHz. The power variability across stators is depicted in Figure 5 (trace A). Figure 5 (trace C) shows the calculated stator power core losses. A characteristic response curve with a peak at around 2.5 kHz was further investigated. Figure 6 shows a close-up of the performance of four stators in a range of 2–3 kHz. The peak of the fixture power was used to differentiate the stators. Figure 6 depicts in detail the power used by four stators of the 30 tested during the study.

Results from a sample population of 30 stators demonstrated that the peak power consumption of all the stators under test occurred within a frequency range of 2.3–2.5 kHz. Stators 2 and 3 represented the maximum and minimum powers of the stator population, respectively. The measured power range between stators 2 and 3 was approximately 0.005 W corresponding to a span of approximately 1.5%.

Back to Top | Article Outline

Discussion

As described earlier, detailed motor performance data in mechanical circulatory support devices have not been widely reported in the literature. In particular, for contactless devices such as the MVAD pump where the pump impeller also serves as the free floating motor rotor, core loss behavior remains an important, albeit complex, field of study. The study described in this article explored a combined modeling and experimental approach to develop a system capable of capturing the core losses of the MVAD pump stator including the small differences within a stator population caused by manufacturing variability. Results from the modeling study confirmed that a series R-L model was capable of reproducing the electrical characteristics of the core loss fixture. The model could be used to explore the expected performance system variations with different circuit topologies. Furthermore, the modeling results also provided the initial building blocks for a more comprehensive model where more complex behaviors, such as the ones produced with motor commutation signals, could be explored. Results from the stator core loss study presented an initial indication of manufacturing variability among the stator population. In addition, the observed small differences in core loss between stators not only revealed a tightly distributed stator population but also confirmed that the core loss measurement system had the capability to discern small manufacturing differences expressed in the peak of the fixture’s power measurement. The study was focused to evaluate a single electrical configuration topology and a single fixture design.

Back to Top | Article Outline

Conclusion

A core loss measurement system for the MVAD pump was developed and tested. The system was used to characterize the core losses in the stator and to investigate differences within a population of stators. Results from the study demonstrated that the system was capable of measuring differences between stators when using the peak power as an indication of core loss. The system could be further used to better characterize and model the stator’s core magnetic behavior, and it has the potential to be used for future motor evaluation and motor design optimization. Overall, the results from the study support the potential use of the core loss system as a tool to be implemented during manufacturing operations with the final goal of assuring consistent pump performance.

Back to Top | Article Outline

References

1. LaRose JA, Tamez D, Ashenuga M, Reyes C. Design concepts and principle of operation of the HeartWare ventricular assist system. ASAIO J. 2010;56:285–289
2. Slaughter MS, Giridharan GA, Tamez D, et al. Transapical miniaturized ventricular assist device: Design and initial testing. J Thorac Cardiovasc Surg. 2011;142:668–674
3. Slaughter MS, Sobieski MA 2nd, Tamez D, et al. HeartWare miniature axial-flow ventricular assist device: Design and initial feasibility test. Tex Heart Inst J. 2009;36:12–16
4. Cullity BD, Graham CD Introduction to Magnetic Materials. 20092nd ed. Hoboken, New Jersey Wiley
5. Roshen WA. A practical, accurate and very general core loss model for nonsinusoidal waveforms. IEEE Trans Power Electron. 2007;22:30–40
6. Dlala E. Comparison of models for estimating magnetic core losses in electrical machines using the finite-element method. IEEE Trans Magnetics. 2009;45:716–725
7. Smith WA, Allaire P, Antaki J, et al. Collected nondimensional performance of rotary dynamic blood pumps. ASAIO J. 2004;50:25–32
8. Neethu S, Shinoy KS, Shajilal AS. FEA-aided design, optimization and development of an axial flux motor for implantable ventricular assist device. World Acad Sci, Eng Technol. 2011;5:70–74

MVAD; core loss; stator

Copyright © 2015 by the American Society for Artificial Internal Organs