4068
IEEE TRANSACTIONS ON MAGNETICS, VOL. 49, NO. 7, JULY 2013
Investigation and Countermeasures for Demagnetization in Line Start Permanent Magnet Synchronous Motors Jian-Xin Shen, Peng Li, Meng-Jia Jin, and Guang Yang Department of Electrical Engineering, Zhejiang University, Hangzhou 310027, China Line start permanent magnet synchronous motors (LSPMSM) may get demagnetized under some special conditions. The demagnetization is usually investigated with finite element analysis (FEA) by using the nominal properties of the magnets. However, the properties decay after the magnets have suffered the demagnetizing magnetic motive force (DMMF), causing spread of the demagnetization area. Therefore, it is proposed in this paper that iteration of FEA should be taken, considering the decaying of the magnets properties. Moreover, four rotor configurations are proposed, so as to enhance the anti-demagnetization capability. FEA with these four configurations proves that the comprehensive configuration with both dual squirrel cages and magnetic barriers on the rotor is rather effective to protect the magnets, but hardly deteriorates the motor operation performance under the rated condition. Index Terms—Demagnetization spreading, dual squirrel cages, line start permanent magnet synchronous motor, magnetic barrier, magnetic bridge, magnetic shielding.
I. INTRODUCTION
A
LINE START permanent magnet synchronous motor (LSPMSM) has a stator which is the same as that of the traditional synchronous motor, and a rotor which contains both squirrel cage and permanent magnets. Such motors can be directly powered with mains power supply [1]–[3], and have high efficiency and power factor over a wide load range, hence, are attractive to many industry applications. However, the LSPMSM risks getting demagnetized of the rotor magnets due to the following issues: 1) during start-up, the stator magnetic motive force (MMF), which is usually very large, rotates at a different speed from the rotor, hence, it can be opposite to the rotor excitation direction, making a significant demagnetizing magnetic motive force (DMMF) on the magnets [4]; 2) During abnormal operation, severe DMMF may exist and harm the magnets permanently [5]; and 3) under heavy load condition, the stator MMF is strong and is almost located at the q-axis, resulting in an inclined field inside the magnets and causing partial demagnetization [6]. The phenomenon of demagnetization has been widely studied [4], [7]–[11], usually using finite element analysis (FEA). However, in the existing literatures, only the initial properties of the magnets (such as the nominal remanence, coercivity, permeability, temperature coefficients, and even the nonlinear demagnetization curve) were set in the FEA. However, this is actually insufficient to predict the severity of demagnetization, as will be detailed in Section II. On the other hand, demagnetization can be restricted by, for example, optimal designs of the rotor geometry [7]–[9], and/or utility of dual layers of interior permanent magnets instead of a single layer of magnets [4]. In Section III, some other countermeasures against demagnetization will be proposed, the effectiveness of which will be verified with FEA. Manuscript received November 05, 2012; revised January 16, 2013; accepted January 28, 2013. Date of current version July 15, 2013. Corresponding author: J.-X. Shen (e-mail:
[email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TMAG.2013.2244582
II. SPREAD OF PARTIAL DEMAGNETIZATION In the conventional investigation of demagnetization with FEA, the whole magnets are equally set with the nominal properties. The significance of demagnetization can then be investigated. However, such investigation is not adequate, because after the first impact of the DMMF, the magnets may be locally demagnetized, with both remanence (Br) and coercivity (Hc) being reduced. In other words, different parts of a magnet now have different properties, and the whole magnets do not have the original capability to withstand the further DMMF. Thus, the demagnetization spreads to more area owing to the further impacts of DMMF. Therefore, it is essential to set the decayed properties to the magnets and resume FEA iteratively, until the demagnetization does not spread any more. Fig. 1 shows the flow chart of iteration process. After each step of FEA, the field inside the magnets is examined in order to determine which area of the magnets is permanently demagnetized. For simplicity of determination, the criteria is set as: in a certain area of the magnets if the flux density along the original magnetizing direction is less than the knee point (for the magnets with a nonlinear second-quadrant demagnetization curve) or less than 0 (for the magnets with a linear second-quadrant demagnetization curve), this area is considered as being permanently demagnetized. Then, the properties of this area are reset as and for the simplicity of processing, but the permeability keeps unchanged. Moreover, the DMMF also changes when the magnets are partially demagnetized, therefore, it should be recalculated. Thus, after resetting the magnet properties and the DMMF, a further FEA is carried out to determine the new demagnetized area of magnets, until the result converges. By way of example, a 3-phase 380 V 15 kW LSPMSM with N35SH Nd-Fe-B magnets is studied. Its rotor configuration is shown in Fig. 2(a). In the first step of iteration, the nominal properties (i.e., the linear demagnetization curve with T kA/m) are set to the magnets, and then FEA and reveals that a small area (2.04 mm ) around the bottom of the V-shaped magnets is demagnetized. In the second step of iteration, the properties of this small area of magnets are reset, while those of the rest area remain unchanged, and the DMMF is recalculated with the decayed magnets. Then, through the second step of FEA, more area (3.52 mm ) of the magnets is found
0018-9464/$31.00 © 2013 IEEE
SHEN et al.: INVESTIGATION AND COUNTERMEASURES FOR DEMAGNETIZATION IN LINE START PERMANENT MAGNET SYNCHRONOUS MOTORS
Fig. 1. Iteration process to investigate spread of partial demagnetization.
Fig. 2. Rotor configurations: (a) Original for the studied 15 kW LSPMSM, (b) Configuration with magnetic barriers and bridges.
Fig. 3. Investigated rotor configurations for the 150 W LSPMSM: (a) Original, (b) With copper sheets, (c) With dual squirrel cages, (d) With magnetic barriers and bridges, (e) With magnetic barriers and dual cages.
being demagnetized. Thus, further iterations are made, until the demagnetization area does not spread any more (converging at 11.10 mm after eight iterations). III. PROTECTION AGAINST DEMAGNETIZATION Various protection methods against demagnetization have been proposed. In this section, a 3-phase 120 V 1500 rpm 150 W 0.96 Nm LSPMSM with DM3845 ferrite magnets is investigated. Ferrite magnets are much easier to be demagnetized than the Nd-Fe-B magnets, hence, special attention should be paid. Nominal properties of the DM3845 ferrite magnets include: T, kA/m, the knee point of the second-quadrant demagnetization curve is (0.03 T, kA/m), and, above the knee point the demagnetization curve is linear. The rotor configuration is shown in Fig. 3(a). There is an extra air gap beneath each magnet bar. Such an air gap should not exist in the practical motors. However, it is used hereby, so that the related demagnetization can be directly compared with the case when a piece of copper sheet is inserted beneath the magnet, which will be explained in Subsection A. Moreover,
4069
Fig. 4. Investigation of the 150 W LSPMSM with original rotor configuration: (a) Field distribution in magnets and flux lines in rotor core and magnets, (b) Normal-direction flux density (Bn) along outer surface of magnets.
the stator outer diameter and inner diameter are 130 and 80 mm, respectively, the air gap length is 0.5 mm, and the stack length is 90 mm. The starting procedure of the LSPMSM is simulated with an FEA software, Ansoft Maxwell, thus the maximum current amplitude is obtained. Then, to simplify the FEA model, the motor rotor is stalled, while 3-phase sinusoidal currents with the obtained amplitude are fed into the motor armature windings. In this way, the most severe demagnetization can be emulated. Fig. 4(a) shows the worst situation of demagnetization, indicating that the field inside the magnets is weak, whilst the flux lines hardly go through the magnets. Furthermore, it is found that the field inside the magnets is actually opposite to the original magnetizing direction. By way of example, Fig. 4(b) shows the normal-direction flux density (Bn) along the outer surface of the magnets which faces the motor air gap, from the outer corner to the inner corner (note, the magnet width is 17.8 mm). Here, the negative flux density means that the field is opposite to the original magnetizing direction. Clearly, the magnets have been fully (100%) demagnetized. Countermeasures should be taken to protect the magnets, according to the 3 issues stated in Section I. Clearly, the DMMF due to issues-(1) and (2) behaves as a varying (e.g., rotating, fluctuating or pulsing) field in the rotor, hence, can be processed with magnetic shielding and/or a magnetic bypass; while the DMMF due to issue-(3) is almost constant in the rotor during steady-state operation, hence, it can be dealt with by a magnetic bypass or a magnetic barrier. In the following subsections, the effectiveness of various protection methods, which utilize the magnetic shielding, bypass and/or barrier, is investigated, with comparison to the situation in Fig. 4. The investigation is carried out with four new motors which have the same stator and the same ratings as the aforementioned 150 W LSPMSM, but use modified rotor configurations. A. Magnetic Shielding With Copper Sheets A piece of copper sheet is inserted beneath each magnet bar, as shown in Fig. 3(b). Its thickness is the same as that of the extra air gap in Fig. 3(a). When the DMMF behaves as a varying field, eddy current is induced inside the copper sheet and further generates a field against the DMMF. Thus, the DMMF is weakened by the eddy current inside the copper sheet, and has less demagnetizing impact on the magnets. Therefore, the copper sheet performs as the magnetic shielding, protecting the magnets from the varying DMMF. Fig. 5(a) shows that the flux density in some area of the magnets is less than the knee point (0.03 T). And, the field in other area is even opposite to the original magnetizing
4070
IEEE TRANSACTIONS ON MAGNETICS, VOL. 49, NO. 7, JULY 2013
that of the copper sheets. As a result, the magnets are less demagnetized, see Figs. 5(b) and 6(b), and the total demagnetized area is reduced to 63.3%. Certainly, the magnets are not sufficiently protected, either. However, the protection of dual squirrel cages can be comprehensively employed together with other methods. On the hand, in Fig. 3(c) the second cage is placed beneath the magnets. If it is located above the magnets, the space may be not enough, and more importantly, the magnets-excited air gap field may be distorted. Therefore, it is often avoided to place the second cage above the magnets. C. Magnetic Barriers and Bridges Fig. 5. Field distribution in magnets and flux lines in rotor core and magnets of the 150 W LSPMSM with modified rotor configuration: (a) With copper sheets, (b) With dual squirrel cages, (c) With magnetic barriers and bridges, (d) With magnetic barriers and dual cages.
Fig. 6. Normal-direction flux density (Bn) along outer surface of magnets of the 150 W LSPMSM with modified rotor configuration: (a) With copper sheets, (b) With dual squirrel cages, (c) With magnetic barriers and bridges, (d) With magnetic barriers and dual cages.
direction. For example, Fig. 6(a) shows that the normal-direction flux density (Bn) along the outer surface of the magnets is always less than 0.03 T. However, the field distribution inside the magnets is unequal, therefore, as confirmed by detailed field investigation, there is some area inside the magnets where Bn is higher than 0.03 T. Hence, the magnets are not fully demagnetized. The total demagnetized area is reduced from 100% in the original rotor to 86.3% in this modified rotor. Here, the copper sheet has the almost same permeability as the extra air gap in Fig. 3(a), hence, the reduction of the demagnetized area is solely resulted in by the magnetic shielding of the copper sheets. Moreover, little difference is observed from FEA if the copper sheets are placed above the magnets. Nevertheless, no matter where the copper sheets are placed, they are not effective enough to protect the magnets. B. Magnetic Shielding With Dual Squirrel Cages The squirrel cage contains a current if there is a varying DMMF in the rotor. This current also generates a field against the DMMF and protects the magnets. Here, a second cage is designed, Fig. 3(c), while its shielding function is stronger than
From Fig. 4(a), it is seen that flux lines go through the rotor teeth between the adjacent poles (i.e., those located at the top left and bottom right corners of the figure), causing inclined field in the magnets and consequently demagnetizing the magnets [6]. If a magnetic barrier (i.e., an air or aluminum window) is designed in the tooth to obscure the flux, as shown in Fig. 2(b), the inclined field will be significantly reduced. However, for small size motors, the rotor tooth is narrow, allowing little area to design the air/aluminum window. Therefore, it is proposed here to remove the tooth and fill in the space with an aluminum bar of the squirrel cage, as shown in Fig. 3(d). Moreover, a magnetic bridge at the bottom of the V-shaped slot, as shown in Figs. 2(b) and 3(d), not only enhances the rotor mechanical strength, but also provides a bypass for the DMMF, thus, decreases the risk of demagnetization of the magnets. Fig. 5(c) presents the field distribution in the magnets. Moreover, the field direction in the magnets is examined. For example, Bn is given in Fig. 6(c). It is found that the total demagnetized area of the magnets is reduced to 27.4%. D. Combination of Magnetic Barriers and Dual Cages From the preceding subsections, it is seen that the dual squirrel cages can weaken the impact of the varying DMMF, while the magnetic barriers can obscure the inclined flux in the magnets, both helping protect the magnets from being demagnetized. Therefore, these two configurations are combined, as shown in Fig. 3(e). Consequently, the magnets are hardly demagnetized. Fig. 5(d) shows that the field in most area of the magnets is sufficiently strong and is much higher than the knee point on the second-quadrant demagnetization curve of the magnets. Moreover, Fig. 6(d) gives the results of Bn along the outer surface of the magnets. Actually, the field direction inside the magnets almost aligns with the original magnetizing direction. In this case, the total demagnetized area is reduced to 9.3% only. IV. INVESTIGATION OF START PERFORMANCE The FEA in Section III, in which the rotor is stalled and the sinusoidal armature currents have the largest amplitude constantly, has maximized the demagnetization. However, the actual demagnetization is usually less severe. The actual start performance of the 150 W LSPMSM with all the rotor configurations shown in Fig. 3 is investigated with iterative FEA. The flux density inside the magnets, in the aspects of
SHEN et al.: INVESTIGATION AND COUNTERMEASURES FOR DEMAGNETIZATION IN LINE START PERMANENT MAGNET SYNCHRONOUS MOTORS
4071
TABLE II COMPARISON OF RATED OPERATION PERFORMANCE
Fig. 7. Field distribution in magnets and flux lines in rotor core and magnets of the 150 W LSPMSM during practical start-up: (a) With original rotor, (b) With rotor of magnetic barriers and dual cages.
TABLE I DEMAGNETIZED AREA OF MAGNETS DURING START-UP
both magnitude and direction, is checked carefully so as to determine the demagnetized area in the most severe case. Table I gives the demagnetized area of the magnets. Under the practical start condition, the configuration with the magnetic barriers and bridges becomes less effective if compared with the case stated in Section III, while the dual cages configuration becomes more effective, but the comprehensive configuration with both dual cages and magnetic barriers always performs the best. Therefore, different rotor configurations have their specific advantages under various work condition. Moreover, Fig. 7 compares the field distribution inside the magnets in the original rotor and the rotor with dual cages and magnetic barriers, respectively. The latter field is generally higher than the knee point on the demagnetization curve of the ferrite magnets. V. INVESTIGATION OF OPERATION PERFORMANCE Besides the influence on magnets protection, the influence of the four proposed rotor configurations on the normal operation performance should also be considered. FEA shows that the air gap field distribution is slightly changed by the magnetic barriers, or by the dual cages if the second cage is placed above the magnets [note, in Fig. 3(c) and (e) the second cage is actually placed beneath the magnets], and the field amplitude is slightly reduced by the copper sheets in Fig. 3(b) while the reduction percentage depends on the thickness of the copper sheets and magnet bars. Such changes can all be reflected in the motor normal operation performance, which is also investigated with FEA. Table II details the comparison of the original rotor configuration and the proposed configuration with both dual cages and magnetic barriers, in the aspects of root-mean-square (RMS) values and total harmonics distortion (THD) of the back electromotive force (EMF) and armature current, and the energy losses in different motor parts, all under the rated operation condition and assuming that the magnets are not demagnetized. It is seen that the proposed rotor configuration hardly deteriorates the operation performance.
VI. CONCLUSION Demagnetization of an LSPMSM can be investigated with FEA. However, iteration should be taken during FEA, by considering the decaying of the magnets properties, so that the spread of demagnetization and the actual severity can be predicted. On the other hand, four rotor configurations are proposed to protect the magnets from demagnetization, and their effectiveness is comparatively studied. It is shown that the configuration with both dual cages and magnetic barriers performs the best to protect the magnets, and hardly deteriorates the normal operation performance. ACKNOWLEDGMENT This work was supported by the China 973 Program (2013CB035604), the Natural Science Foundation (51077116), and the 863 Program (2011AA11A101). REFERENCES [1] T. J. E. Miller, “Synchronization of line-start permanent-magnet AC motors,” IEEE Trans. Power Apparat. Syst., vol. PAS-103, no. 7, pp. 1822–1828, Jul. 1984. [2] E. Richter and T. W. Neumann, “Line start permanent magnet motors with different materials,” IEEE Trans. Magn., vol. 20, no. 5, pp. 1762–1764, Sep. 1984. [3] M. K. Andrew and I. M. Catherine, “The design of high-efficiency linestart motors,” IEEE Trans. Ind. Appl., vol. 36, no. 6, pp. 1555–1562, Nov. 2000. [4] T. H. Kim and J. P. Hong, “A study on the irreversible magnet demagnetization in single-phase line-start permanent magnet motor,” J. Appl. Phys., vol. 105, no. 07, p. 07F108, Feb. 2009. [5] A. Abbas, H. A. Yousef, and O. A. Sebakhy, “FE parameters sensitivity analysis of an industrial LS interior PM synchronous motor,” in Proc. 2008 IEEE PES General Meeting, Pittsburgh, USA, 2008, pp. 1–6. [6] S. Ruoho and A. Arkkio, “Partial demagnetization of permanent magnets in electrical machines caused by an inclined field,” IEEE Trans. Magn., vol. 44, no. 7, pp. 1773–1778, Jul. 2008. [7] G. H. Kang, J. Hur, H. Nam, J. P. Hong, and G. T. Kim, “Analysis of irreversible magnet demagnetization in line-start motors based on the finite-element method,” IEEE Trans. Magn., vol. 39, no. 3, pp. 1488–1491, May 2003. [8] K. C. Kim, S. B. Lim, D. H. Koo, and J. Lee, “The shape design of permanent magnet for permanent magnet synchronous motor considering partial demagnetization,” IEEE Trans. Magn., vol. 42, no. 10, pp. 3485–3487, Oct. 2006. [9] K. Y. Hwang, B. Y. Yang, S. H. Rhyu, B. T. Kim, and D. K. Kim, “Optimal rotor design for reducing the partial demagnetization effect and cogging torque in spoke type PM motor,” J. Appl. Phys., vol. 105, no. 07, p. 07F123, Apr. 2009. [10] Y. Zhilichev, “Analysis of permanent magnet demagnetization accounting for minor B-H curves,” IEEE Trans. Magn., vol. 44, no. 11, pp. 4285–4288, Nov. 2008. [11] M. Rosu, J. Saitz, and A. Arkkio, “Hysteresis model for finite-element analysis of permanent-magnet demagnetization in a large synchronous motor under a fault condition,” IEEE Trans. Magn., vol. 41, no. 6, pp. 2118–2123, Jun. 2005.
