- Email: [email protected]
Investigation of the influence of different cutting procedures on the global and local magnetic properties of non-oriented electrical steel
Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
Contents lists available at ScienceDirect
Journal of Magnetism and Magnetic Materials journal homepage: www.elsevier.com/locate/jmmm
Investigation of the influence of different cutting procedures on the global and local magnetic properties of non-oriented electrical steel H. Naumoski a,n, B. Riedmüller b, A. Minkow b, U. Herr b a b
Daimler AG, R&D, 89081 Ulm, Germany Institute of Micro- and Nanomaterials, University Ulm, 89069 Ulm, Germany
art ic l e i nf o
a b s t r a c t
Article history: Received 10 February 2015 Received in revised form 17 April 2015 Accepted 12 May 2015 Available online 14 May 2015
The process of manufacturing iron cores for electric machines out of electrical steel sheets can strongly affect the magnetic properties of the material. In order to better understand the influence of cutting on the iron losses, a characterization of the magnetization behavior near the cutting edge is needed. The local magnetic properties of the material are modified by the cutting process which leads to an increase in the iron losses measured for 5 mm wide ring core samples by nearly 160% at low inductions. We present investigations on the effect of cutting by observation of the magnetic domain structure of 0.35 mm thick non-oriented electrical steel. By using the magneto-optical Kerr-effect on a ring samples the local magnetic properties of the material after processing are characterized in the form of domain wall displacements under an applied external ac-field. The influence of various cutting techniques on the magnetic properties was studied before and after stress relief annealing. This method allows a quantitative analysis of the influence of different cutting techniques on the micro-magnetic properties of nonoriented electrical steel for rotating machines. & 2015 Elsevier B.V. All rights reserved.
Keywords: Magneto-optical Kerr-effect Influence of cutting Magnetic properties Non-oriented electrical steel
1. Introduction For stator and rotor cores in rotating machines non-oriented electrical steel is used for flux conduction. In order to shape these cores electrical steel sheets have to be cut in laminations. Various cutting operations like punching, laser cutting and spark erosion can be used, which induce stresses inside the material close to the cutting edge. Due to these stresses the magnetic properties of the material are deteriorated which can be determined by an integral measurement of the magnetization curve of the bulk material. Previous works tried to identify the local magnetic properties near the cutting edges by using needle probes [1–3]. Local magnetic properties after punching have also been investigated with spatially resolved measurements [4] or using a single-sheet-tester with repeatedly cut sheets to determine the influence of the cutting length in relation to the sample volume [5,6]. The effect of punching and shearing was investigated by imaging the domain structure by using the magneto-optical Kerr effect [7]. In other works [8,9] the authors observed additionally the domain wall movement by applying an ac-field to a processed lamination, but without particularizing the discreet manner how the field is applied into the sample. n
Corresponding author. Fax: þ49 711 305 2166359. E-mail address: [email protected] (H. Naumoski).
http://dx.doi.org/10.1016/j.jmmm.2015.05.031 0304-8853/& 2015 Elsevier B.V. All rights reserved.
In this paper the authors will present fundamental investigations on the effect of different cutting procedures using microscopic analysis and integral measurements of the magnetic properties. In order to analyze the local magnetic properties of the cut specimen near the edges a method was developed using the magneto-optical Kerr-effect (MOKE) [10]. The aim is to visualize the local influence of cutting on the magnetization behavior in form of magnetic contrast. Several cutting techniques have been compared with respect to their magnetic contrast before and after stress relief annealing. The results of the microscopic investigations are compared with integral measurements of electrical loss. The investigations were made on a fully processed commercial electrical steel grade (as used for stator and rotor cores of electrical machines in electrical and hybrid vehicles), and will therefore be useful for optimization of the manufacturing process and reduction of loss in electrical machines.
2. Experimental procedure Fully processed non-oriented electrical steel with a thickness of 0.35 mm was used for this study. The specimens were taken from neighboring parts in the center of a mother coil to guarantee constant material properties. The most important metallurgic parameters for the magnetic properties are as follows: average grain size of 96 mm and chemical composition with 2.8% silicon,
H. Naumoski et al. / Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
0.4% aluminum and 0.2% manganese content. A CO2 laser was used to cut the material at a power of 1500 W, a speed of 20 m/min and a N2-gas stream of 10 bar. The punching machine was operated at 100 cuts/min with a lift of 44 mm using sharpened blades and a clearance of 0.02 mm. Reference samples showing magnetic properties not affected by heat or mechanical deformation have been produced by electrical discharge wire cutting (spark erosion). The spark erosion was performed using a brass wire at a diameter of 0.25 mm placed in a water quench. To recover the original material properties after the cutting process, stress relief annealing was performed under nitrogen atmosphere at 820 °C for three hours with a cooling phase of two hours. To get a first impression of the effect of cutting the microstructure of the edge profiles of the cut samples was examined [11]. The integral magnetic properties, in particular core losses at 50 Hz and 400 Hz, hysteresis and magnetization curve, were determined using field-metric measurements on ring core samples as specified in DIN 50460 [12]. The cut laminations with an outer diameter of D¼ 55 mm and an inner diameter of d¼ 45 mm were stacked insulated from each other to a height of 5 mm. The magnetic field exciting the ring at the observed area was calculated according to Eq. (1), with n¼ 100 for the number of windings, I for the applied current and lm for the magnetic path length. To calculate the average magnetic field over the cross section of a ring the magnetic path length is calculated using Eq. (2).
H=
n* I lm
lm = π
(1)
D−d In
() D d
(2)
In order to analyze the micro-magnetic properties of the electrical steel near the cutting edge a method was developed using the magneto-optical Kerr-effect (MOKE). The domain observations were made using a Zeiss microscope with high stage stability equipped with a Hamamatsu image processing system. As specimens one lamination of the stacked rings used for the integral magnetic measurement was taken. These rings were embedded and primed. In order to get a smooth surface necessary for the Kerr-microscopy a final oxide polishing with colloidal silica was used to finish of the surface. For the introduction of a magnetic field the ring was cut out and coiled up with 100 windings, as shown in Fig. 1(a). To determine the magnetic flux and field distributions inside the ring the setup was simulated using FEM. The whole flux goes through the ring because of the high permeability of the steel and no flux concentration could be observed.
127
Fig. 1(b) shows the domain structure of the punched specimen near the cutting edge in the demagnetized state obtained with the Kerr microscope from the observed surface, marked in (a). The domains can directly be observed even near the cutting edge (as seen in the box). This technique allows observing the domain wall movement in real time while applying an external magnetic field to the specimen. Under the conditions used here, the main contribution to the contrast originates from the longitudinal Kerr effect. The aim of this investigation is to image the domains that have changed during an applied ac-field with specific field amplitude. In this way, regions of the sample may be identified where the magnetization reversal process has already started at the applied magnetic field. The difference image method [13,14] was used for this purpose. In this method a reference image is recorded by averaging 32 images while applying an external 50 Hz ac-field with given amplitude. After turning off the ac-field a demagnetized domain state is visible, from which the reference image is subtracted in the next step. The result is an image of the domains in regions where the contrast changed during ac-field application. The regions in which the contrast did not change show no contrast after subtraction of the reference image and appear gray. Fig. 2 shows a schematic representation of the subtraction procedure [10]. The investigations were made on several places close to the edges of the specimens; the figures show representative domain images. As a reference, domain structures inside the bulk of the material were obtained by imaging regions in the middle of the rings which are far away from the cutting edges to minimize the influence of the cutting process. Fig. 3 shows representative domain patterns of such unaffected regions.
3. Results and discussion 3.1. Examinations before annealing First the microstructure of the cut specimen was examined. Optical micrographs of the cross sections of the cutting edges made after polishing and grain boundary etching with nital are shown in Fig. 4. The direction of cutting is marked, the edge is on the outer diameter. As expected, spark erosion (Fig. 4a) leaves no visible traces of deformation on the surface, the cut edge is straight due to the procedure. The punched edge (Fig. 4b) shows the typical burr (circle) and a zone of plastic deformation (arrow) on the other side [4]. Laser cutting with a CO2-laser (Fig. 4c) leaves a smooth edge and no traces of plastic deformation are visible within the material. The observed profile of the laser cut edge can
Fig. 1. (a) Primed and winded specimen used for the MOKE investigations and (b) domain pattern at the punched edge.
128
H. Naumoski et al. / Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
Fig. 2. Schematic picture showing the origin of contrast formation in the difference image method. As an example, the contrast of a moving domain wall is shown [10].
be explained by the fact that the laser beam hits the ring surface on the upper left corner of the cross section (circle). Since a protective N2 gas stream is applied during the laser cutting (which flows from left to right in Fig. 4c), the molten material is blown away to the right which leads to a rounding of the upper right edge of the cross section. Fig. 5 shows the dc-magnetization curves, Fig. 6 the dc-hysteresis curves for the cut ring core samples. Table 1 lists the maximum permeabilities, according to (3) and the coercivities obtained from the dc-magnetic measurements.
μr = max (dB/d (μ 0 H ))
(3)
The magnetization curve of the spark eroded sample shows the highest permeability and reaches the highest induction. Also it has the lowest coercivity and is taken as the reference. The magnetization curve for the punched sample shows shearing. The magnetic induction does not reach the value of the spark eroded one. Even for the high field amplitude of 1600 A/m a reduction of the polarization of about 5% remains. The permeability is significantly reduced. The most deteriorated properties are observed in the laser cut sample. It is well known that the rapid heating that occurs when cutting with a laser affects a large zone near the sample’s edge and changes the metallurgic properties of the steel as well as the magnetic ones. The sample shows a much lower permeability. The polarization is lower but crosses the one of the punched sample at 1000 A/m. Similar results have been reported by [11]. The difference to the polarization of the reference sample remains about 2% even at high applied field. The dc-hysteresis measurements approve the results obtained by the magnetization
measurement. The cutting procedure causes a shear in the hysteresis and the coercivity is raised, too. The laser-cut sample shows larger shearing than the punched one. Also the coercivity is increased by more than 100% compared to the value of the eroded ring, where the punched ring shows an increase of 17%. The magnetization behavior directly influences the losses. Fig. 7 shows the realtive additional ac losses of the cut material at (a) 50 Hz and (b) 400 Hz normalized to the reference (spark eroded sample) as a function of the induction.
Padd, rel = Pcut /Pref −1
(4)
Table 2 shows the absolute power losses measured at 50 Hz; 1.5 T and 400 Hz; 1.0 T. The losses are increased by both cutting techniques and also at both frequencies. In particular laser cutting has a large influence on the losses. Punching shows a nearly constant increase of about 20% in the field range below 1.2 T compared to the eroded ring (Fig. 7a). Above 1.4 T the difference reduces to about 15%. In contrast, laser cutting shows much higher losses for low inductions: for 0.2 T there is an increase of 159%, and at 1.0 T and 1.5 T the losses still increase by 75% and 35% compared to the spark eroded reference, respectively. The measurement at 400 Hz shows a similar trend for the additional losses. For punching the relative additional losses are slightly lower than at 50 Hz. The laser-cut specimen shows also lower relative additional losses at 400 Hz, especially at inductions below 1.4 T. Although the absolute losses increase with increasing frequency the relative additional losses decrease. These observations can be explained by the fact that at 400 Hz the eddy current losses dominate, which are not significantly affected by the cutting procedure.
Fig. 3. Difference image Kerr micrographs obtained in the middle of a specimen representing the magnetic structure not affected by the cutting procedure at a magnetc acfield amplitude of 1600 A/m.
H. Naumoski et al. / Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
129
Fig. 4. Optical micrographs of the cross sections of the cutting edges of the investigated electrical steel by various techniques: (a) spark eroded, (b) punched and (c) laser cut.
Fig. 5. Measurement of the magnetization curves of the ring cores for spark erosion, punching and laser cutting before annealing.
Fig. 7. Relative additional losses at (a) 50 Hz and (b) 400 Hz of the ring cores for punching and laser cutting before annealing normalized to the spark eroded one. Fig. 6. Measurement of the dc-hysteresis of the ring cores for spark erosion, punching and laser cutting before annealing.
Table 1 Maximum permeabilities obtained from dc-magnetization measurements and coercivities obtained from dc-hysteresis measurements before annealing.
Permeability Coercivity
Table 2 Absolute power losses in W/kg of the 5 mm wide ring core specimens measured at 50 Hz; 1.5 T and 400 Hz; 1.0 T before annealing.
Eroded
Punched
Laser-cut
Pfe (W/kg)
Eroded
Punched
Laser-cut
8700 29 A/m
5800 35 A/m
1500 70 A/m
50 Hz; 1.5 T 400 Hz; 1.0 T
2.75 21.8
3.15 24.9
3.70 32.4
130
H. Naumoski et al. / Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
Fig. 8. Magnetization process obtained with the difference image approach near the cutting edges spark erosion, punching and laser-cutting (rows) of the specimens for two ac-field amplitudes 130 A/m and 1600 A/m (columns) before annealing.
Fig. 8 shows the difference domain images at the cut edges on applying a low and a high ac-field amplitude, calculated according to Eqs. (1) and (2), of 130 A/m and 1600 A/m respectively. The cut edge is at the top of the micrographs. For the spark eroded sample domain contrast can be observed over the whole imaging region of about 400 mm by applying a low field amplitude of 130 A/m. Only in a narrow region of under 100 mm directly at the cut edge the contrast is missing, indicating a degradation of the magnetic properties. By raising the field
amplitude to 1600 A/m these regions also show full domain contrast which means that they also contribute to the flux conduction under these conditions. The structure of the domain pattern at the cut edge looks similar to the reference pattern obtained, shown in Fig. 3. The punched specimen shows a different behavior. For a field amplitude of 130 A/m hardly any domain contrast can be found in a region of about 300–400 mm from the cutting edge. For a field amplitude of 1600 A/m this region of poor contrast shrinks to
H. Naumoski et al. / Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
131
Fig. 9. Optical micrographs of the cross sections of the cutting edges of the investigated electrical steel after annealing: (a) spark eroded, (b) punched and (c) laser cut.
Table 3 Maximum permeabilities obtained from dc-magnetization measurements and coercivities obtained from dc-hysteresis measurements after annealing.
Permeability Coercivity
Table 4 Absolute power losses in W/kg of the 5 mm wide ring core specimens measured at 50 Hz; 1.5 T and 400 Hz; 1.0 T after annealing.
Eroded
Punched
Laser-cut
Pfe (W/kg)
Eroded
Punched
Laser-cut
8800 29 A/m
8600 30 A/m
6900 34 A/m
50 Hz; 1.5 T 400 Hz; 1.0 T
2.75 21.6
2.78 21.8
2.86 22.8
Fig. 10. Magnetic polarization of the ring cores for spark erosion, punching and laser cutting after annealing.
Fig. 11. Magnetic polarization of the ring cores for spark erosion, punching and laser cutting after annealing.
about 100–150 mm, which is similar to the average grain size of our samples. The structure of the domains is also modified. A change of magnetic contrast is observed compared to unaffected material
Fig. 12. Relative additional losses of the ring cores for punching and laser cutting after annealing at f ¼50 Hz normalized to the spark eroded, annealed one.
shown in Fig. 3. A comparison with the integral magnetic measurements (Fig. 5) suggests that the lower permeability and the lower polarization at 1600 A/m of the punched specimen are due to the deterioration of the magnetic properties in the region close
132
H. Naumoski et al. / Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
Fig. 13. Magnetization process obtained with the difference image approach near the cutting edges spark erosion, punching and laser-cutting (rows) of the specimens for two ac-field amplitudes 130 A/m and 1600 A/m (columns) after annealing.
to the cut edge. For the ring used with a width of 5 mm a loss in polarization of 5% means that a zone of 250 mm width does not contribute to the polarization. Since the ring has two edges this means that each edge is associated with a magnetically dead zone of 125 mm. This result correlates well with the MOKE investigations where a degraded zone of about 100–150 mm is observed. For the laser cut sample we also observe little contrast in a region of 300 mm away from the cut edge for the low field amplitude of 130 A/m. At the field amplitude of 1600 A/m some regions close to the edge (see upper right-hand corner) still do not show magnetic contrast, whereas. Further away from the edge contrast is found at
this field amplitude. The domain patterns show wide and long stripe domains that look distorted. These results correlate with the integral magnetic properties obtained with the ring cores. For low field strengths there is nearly no domain wall movement near the edge and further inside the material which leads to a lower macroscopic permeability. The MOKE observations also shed light on the origin of the differences between the magnetization curves of the different samples (Fig. 5). Since the laser-cut sample does not show a large zone without contrast at high field amplitude, the global polarization of the sample gets larger than the polarization of the punched one at fields above 1000 A/m.
H. Naumoski et al. / Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
3.2. Examinations after annealing Optical micrographs of cross sections of the annealed samples are shown in Fig. 9. No significant changes on the cut edges are visible. Only the punched and annealed edge shows regeneration of small grains at area of the burr (circle). Grain growth is not expected to have a significant effect here since the initial grain size of about 100 mm. Table 3 lists the maximum permeabilities and obtained coercivities after annealing. After the annealing all samples show improved magnetic properties when compared to the as-cut samples. The punched specimen reaches similar permeability and coercivity values after annealing as the spark eroded sample. The properties of the laser cut sample also improved after annealing, but the values are still lower than for the other samples. As shown in Fig. 10 the polarization at 130 A/m does not reach the value of the reference sample. However, all samples reach the polarization of the reference at field amplitudes above 1000 A/m. Fig. 11 shows the dc-hysteresis measurement, that confirms the results of the dc-magnetization. While the punched specimen has recovered, the laser-cut still shows a shear in the hysteresis and an increase in coercivity. The measurement of the losses also confirms the result of the dc-measurements. Table 4 lists the absolute losses measured at 50 Hz/1.5 T and 400 Hz/1 T, respectively. Fig. 12 shows the relative additional losses according to (4). While both punched and laser-cut samples show a reduction in the losses compared to the as-cut state, the laser cut sample still shows significantly higher losses than the spark eroded reference. Fig. 13 shows the difference domain images at the cut edges after annealing for field amplitudes of 130 A/m and 1600 A/m respectively. For the spark eroded sample only slight changes compared to the as-cut state can be observed. For the punched specimen we observe magnetic contrast all over the sample for both low and high field amplitude, comparable to the spark eroded and annealed specimen. The domain patterns at the edge are quite comparable to the reference image taken inside the material, showing no residual influence of the cutting process. This result corresponds well to the integral magnetic measurements of the punched and eroded ring cores after the annealing which show similar magnetic properties. However, the laser cut sample still shows impeded domain wall movement after the annealing for low field amplitude directly at the cut edge (box on the left hand side). This indicates that while most parts of the material have recovered, the magnetization processes are still affected by the cutting treatment in some regions close to the cut edge. This behavior can be linked to the integral magnetic properties which still show a lower permeability and higher losses after the annealing compared to the other cutting techniques. This behavior could be linked to a melting or heat affected zone (HAZ) in the immediate proximity of the cut edge due to the laser-cutting procedure. This was also observed by [15], where this zone was determined at about 50 mm which correlated well with the results obtained by the MOKE analysis in Fig. 13.
4. Conclusion It was shown that after cutting by different techniques the magnetic properties of electrical steel are affected. The micro-
133
magnetic behavior in the region near the cut edge has been studied using the magneto-optical Kerr-effect. As reference for unaffected magnetic properties specimens cut by spark erosion were used. The obtained micro-graphs show zones at the cut edges without magnetic contrast for excitations with two ac magnetic field amplitudes of 130 A/m and 1600 A/m. The global magnetic properties obtained with field-metric measurements correlate well with the micro-magnetic observations from the MOKE method. The results show that industrial punching has a negative effect on the magnetic properties. The maximum permeability decreases by 35% and the losses increase by 20% for low and high frequencies over a wide range of induction for the used geometry. This can be correlated with micro-magnetic observations using the Kerr effect: domain wall displacement is impeded and the domain structure is changed over the first two to three grain rows near the edge. The effect of laser cutting is even more pronounced. The permeability decreases to 17% of its initial value and the losses for small inductions increase by over 150%. The micro-magnetic investigations show that these changes correlate quantitatively with the area fraction of the regions affected by the cutting procedure. Stress relief annealing recovers the magnetic properties, as shown by the integral measurements and by domain observation. However, the laser cut sample still shows a degraded region of heat affection (HAZ) that results in slightly higher losses than that of the spark eroded reference sample. The method demonstrated here allows quantitative determination of the extension of the magnetically affected zone for different cutting procedures. In future, it may be used to identify cutting parameters which minimize the size of the affected zone. This would allow a reduction of the losses, or a reduction of the annealing temperature and time without intolerable deterioration of the magnetic properties.
References [1] R. Rygal, A.J. Moses, N. Derebasi, J. Schneider, A. Schoppa, J. Magn. Magn. Mater. 215–216 (2000) 687. [2] G. Crevecoeur, L. Dupre, L. Vandenbossche, R. Van de Walle, IEEE Trans. Magn. 44 (6) (2008). [3] G. Loises, A.J. Moses, IEEE Trans. Magn. 37 (4) (2001) 2755. [4] H.M.S. Harstick, M. Ritter, W. Riehemann, IEEE Trans. Magn. 50 (4) (2014). [5] A. Schoppa, J. Schneider, J.-O. Roth, J. Magn. Magn. Mater 215–216 (2000) 100. [6] A. Schoppa, J. Schneider, C.-D. Wuppermann, J. Magn. Magn. Mater. 215–216 (2000) 74. [7] K. Senda, M. Ishida, Y. Nakasu, M. Yagi, J. Magn. Magn. Mater. 304 (2006) 513. [8] M. Takezewa, Y. Wada, J. Yamasaki, T. Honda, C. Kaido, IEEE Trans. Magn. 39 (5) (2003) 3208. [9] M. Takezewa, K. Kitajima, Y. Morimoto, J. Yamasaki, C. Kaido, IEEE Trans. Magn. 42 (10) (2006) 2790. [10] H. Naumoski, L. Vandenbossche, A. Maucher, S. Jacobs, U. Herr, X. Chassang, in: Proceedings of the IEEE Conference on EDPC, 2014. [11] M. Emura, F.J.G. Landgraf, W. Ross, J.R. Barreta, J. Magn. Magn. Mater. 254–255 (2003) 358. [12] DIN 50460, 1988. [13] D.A. Herman Jr., B.E. Argyle, IEEE Trans. Magn. 22 (5) (1986) 772. [14] A. Huber, R. Schäfer, Magnetic Domains, Springer, Berlin, 1998. [15] A. Belhadj, P. Baudouin, F. Breaban, A. Deffontaine, M. Dewulf, Y. Houbaert, J. Magn. Magn. Mater. 256 (2003) 20.
Contents lists available at ScienceDirect
Journal of Magnetism and Magnetic Materials journal homepage: www.elsevier.com/locate/jmmm
Investigation of the influence of different cutting procedures on the global and local magnetic properties of non-oriented electrical steel H. Naumoski a,n, B. Riedmüller b, A. Minkow b, U. Herr b a b
Daimler AG, R&D, 89081 Ulm, Germany Institute of Micro- and Nanomaterials, University Ulm, 89069 Ulm, Germany
art ic l e i nf o
a b s t r a c t
Article history: Received 10 February 2015 Received in revised form 17 April 2015 Accepted 12 May 2015 Available online 14 May 2015
The process of manufacturing iron cores for electric machines out of electrical steel sheets can strongly affect the magnetic properties of the material. In order to better understand the influence of cutting on the iron losses, a characterization of the magnetization behavior near the cutting edge is needed. The local magnetic properties of the material are modified by the cutting process which leads to an increase in the iron losses measured for 5 mm wide ring core samples by nearly 160% at low inductions. We present investigations on the effect of cutting by observation of the magnetic domain structure of 0.35 mm thick non-oriented electrical steel. By using the magneto-optical Kerr-effect on a ring samples the local magnetic properties of the material after processing are characterized in the form of domain wall displacements under an applied external ac-field. The influence of various cutting techniques on the magnetic properties was studied before and after stress relief annealing. This method allows a quantitative analysis of the influence of different cutting techniques on the micro-magnetic properties of nonoriented electrical steel for rotating machines. & 2015 Elsevier B.V. All rights reserved.
Keywords: Magneto-optical Kerr-effect Influence of cutting Magnetic properties Non-oriented electrical steel
1. Introduction For stator and rotor cores in rotating machines non-oriented electrical steel is used for flux conduction. In order to shape these cores electrical steel sheets have to be cut in laminations. Various cutting operations like punching, laser cutting and spark erosion can be used, which induce stresses inside the material close to the cutting edge. Due to these stresses the magnetic properties of the material are deteriorated which can be determined by an integral measurement of the magnetization curve of the bulk material. Previous works tried to identify the local magnetic properties near the cutting edges by using needle probes [1–3]. Local magnetic properties after punching have also been investigated with spatially resolved measurements [4] or using a single-sheet-tester with repeatedly cut sheets to determine the influence of the cutting length in relation to the sample volume [5,6]. The effect of punching and shearing was investigated by imaging the domain structure by using the magneto-optical Kerr effect [7]. In other works [8,9] the authors observed additionally the domain wall movement by applying an ac-field to a processed lamination, but without particularizing the discreet manner how the field is applied into the sample. n
Corresponding author. Fax: þ49 711 305 2166359. E-mail address: [email protected] (H. Naumoski).
http://dx.doi.org/10.1016/j.jmmm.2015.05.031 0304-8853/& 2015 Elsevier B.V. All rights reserved.
In this paper the authors will present fundamental investigations on the effect of different cutting procedures using microscopic analysis and integral measurements of the magnetic properties. In order to analyze the local magnetic properties of the cut specimen near the edges a method was developed using the magneto-optical Kerr-effect (MOKE) [10]. The aim is to visualize the local influence of cutting on the magnetization behavior in form of magnetic contrast. Several cutting techniques have been compared with respect to their magnetic contrast before and after stress relief annealing. The results of the microscopic investigations are compared with integral measurements of electrical loss. The investigations were made on a fully processed commercial electrical steel grade (as used for stator and rotor cores of electrical machines in electrical and hybrid vehicles), and will therefore be useful for optimization of the manufacturing process and reduction of loss in electrical machines.
2. Experimental procedure Fully processed non-oriented electrical steel with a thickness of 0.35 mm was used for this study. The specimens were taken from neighboring parts in the center of a mother coil to guarantee constant material properties. The most important metallurgic parameters for the magnetic properties are as follows: average grain size of 96 mm and chemical composition with 2.8% silicon,
H. Naumoski et al. / Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
0.4% aluminum and 0.2% manganese content. A CO2 laser was used to cut the material at a power of 1500 W, a speed of 20 m/min and a N2-gas stream of 10 bar. The punching machine was operated at 100 cuts/min with a lift of 44 mm using sharpened blades and a clearance of 0.02 mm. Reference samples showing magnetic properties not affected by heat or mechanical deformation have been produced by electrical discharge wire cutting (spark erosion). The spark erosion was performed using a brass wire at a diameter of 0.25 mm placed in a water quench. To recover the original material properties after the cutting process, stress relief annealing was performed under nitrogen atmosphere at 820 °C for three hours with a cooling phase of two hours. To get a first impression of the effect of cutting the microstructure of the edge profiles of the cut samples was examined [11]. The integral magnetic properties, in particular core losses at 50 Hz and 400 Hz, hysteresis and magnetization curve, were determined using field-metric measurements on ring core samples as specified in DIN 50460 [12]. The cut laminations with an outer diameter of D¼ 55 mm and an inner diameter of d¼ 45 mm were stacked insulated from each other to a height of 5 mm. The magnetic field exciting the ring at the observed area was calculated according to Eq. (1), with n¼ 100 for the number of windings, I for the applied current and lm for the magnetic path length. To calculate the average magnetic field over the cross section of a ring the magnetic path length is calculated using Eq. (2).
H=
n* I lm
lm = π
(1)
D−d In
() D d
(2)
In order to analyze the micro-magnetic properties of the electrical steel near the cutting edge a method was developed using the magneto-optical Kerr-effect (MOKE). The domain observations were made using a Zeiss microscope with high stage stability equipped with a Hamamatsu image processing system. As specimens one lamination of the stacked rings used for the integral magnetic measurement was taken. These rings were embedded and primed. In order to get a smooth surface necessary for the Kerr-microscopy a final oxide polishing with colloidal silica was used to finish of the surface. For the introduction of a magnetic field the ring was cut out and coiled up with 100 windings, as shown in Fig. 1(a). To determine the magnetic flux and field distributions inside the ring the setup was simulated using FEM. The whole flux goes through the ring because of the high permeability of the steel and no flux concentration could be observed.
127
Fig. 1(b) shows the domain structure of the punched specimen near the cutting edge in the demagnetized state obtained with the Kerr microscope from the observed surface, marked in (a). The domains can directly be observed even near the cutting edge (as seen in the box). This technique allows observing the domain wall movement in real time while applying an external magnetic field to the specimen. Under the conditions used here, the main contribution to the contrast originates from the longitudinal Kerr effect. The aim of this investigation is to image the domains that have changed during an applied ac-field with specific field amplitude. In this way, regions of the sample may be identified where the magnetization reversal process has already started at the applied magnetic field. The difference image method [13,14] was used for this purpose. In this method a reference image is recorded by averaging 32 images while applying an external 50 Hz ac-field with given amplitude. After turning off the ac-field a demagnetized domain state is visible, from which the reference image is subtracted in the next step. The result is an image of the domains in regions where the contrast changed during ac-field application. The regions in which the contrast did not change show no contrast after subtraction of the reference image and appear gray. Fig. 2 shows a schematic representation of the subtraction procedure [10]. The investigations were made on several places close to the edges of the specimens; the figures show representative domain images. As a reference, domain structures inside the bulk of the material were obtained by imaging regions in the middle of the rings which are far away from the cutting edges to minimize the influence of the cutting process. Fig. 3 shows representative domain patterns of such unaffected regions.
3. Results and discussion 3.1. Examinations before annealing First the microstructure of the cut specimen was examined. Optical micrographs of the cross sections of the cutting edges made after polishing and grain boundary etching with nital are shown in Fig. 4. The direction of cutting is marked, the edge is on the outer diameter. As expected, spark erosion (Fig. 4a) leaves no visible traces of deformation on the surface, the cut edge is straight due to the procedure. The punched edge (Fig. 4b) shows the typical burr (circle) and a zone of plastic deformation (arrow) on the other side [4]. Laser cutting with a CO2-laser (Fig. 4c) leaves a smooth edge and no traces of plastic deformation are visible within the material. The observed profile of the laser cut edge can
Fig. 1. (a) Primed and winded specimen used for the MOKE investigations and (b) domain pattern at the punched edge.
128
H. Naumoski et al. / Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
Fig. 2. Schematic picture showing the origin of contrast formation in the difference image method. As an example, the contrast of a moving domain wall is shown [10].
be explained by the fact that the laser beam hits the ring surface on the upper left corner of the cross section (circle). Since a protective N2 gas stream is applied during the laser cutting (which flows from left to right in Fig. 4c), the molten material is blown away to the right which leads to a rounding of the upper right edge of the cross section. Fig. 5 shows the dc-magnetization curves, Fig. 6 the dc-hysteresis curves for the cut ring core samples. Table 1 lists the maximum permeabilities, according to (3) and the coercivities obtained from the dc-magnetic measurements.
μr = max (dB/d (μ 0 H ))
(3)
The magnetization curve of the spark eroded sample shows the highest permeability and reaches the highest induction. Also it has the lowest coercivity and is taken as the reference. The magnetization curve for the punched sample shows shearing. The magnetic induction does not reach the value of the spark eroded one. Even for the high field amplitude of 1600 A/m a reduction of the polarization of about 5% remains. The permeability is significantly reduced. The most deteriorated properties are observed in the laser cut sample. It is well known that the rapid heating that occurs when cutting with a laser affects a large zone near the sample’s edge and changes the metallurgic properties of the steel as well as the magnetic ones. The sample shows a much lower permeability. The polarization is lower but crosses the one of the punched sample at 1000 A/m. Similar results have been reported by [11]. The difference to the polarization of the reference sample remains about 2% even at high applied field. The dc-hysteresis measurements approve the results obtained by the magnetization
measurement. The cutting procedure causes a shear in the hysteresis and the coercivity is raised, too. The laser-cut sample shows larger shearing than the punched one. Also the coercivity is increased by more than 100% compared to the value of the eroded ring, where the punched ring shows an increase of 17%. The magnetization behavior directly influences the losses. Fig. 7 shows the realtive additional ac losses of the cut material at (a) 50 Hz and (b) 400 Hz normalized to the reference (spark eroded sample) as a function of the induction.
Padd, rel = Pcut /Pref −1
(4)
Table 2 shows the absolute power losses measured at 50 Hz; 1.5 T and 400 Hz; 1.0 T. The losses are increased by both cutting techniques and also at both frequencies. In particular laser cutting has a large influence on the losses. Punching shows a nearly constant increase of about 20% in the field range below 1.2 T compared to the eroded ring (Fig. 7a). Above 1.4 T the difference reduces to about 15%. In contrast, laser cutting shows much higher losses for low inductions: for 0.2 T there is an increase of 159%, and at 1.0 T and 1.5 T the losses still increase by 75% and 35% compared to the spark eroded reference, respectively. The measurement at 400 Hz shows a similar trend for the additional losses. For punching the relative additional losses are slightly lower than at 50 Hz. The laser-cut specimen shows also lower relative additional losses at 400 Hz, especially at inductions below 1.4 T. Although the absolute losses increase with increasing frequency the relative additional losses decrease. These observations can be explained by the fact that at 400 Hz the eddy current losses dominate, which are not significantly affected by the cutting procedure.
Fig. 3. Difference image Kerr micrographs obtained in the middle of a specimen representing the magnetic structure not affected by the cutting procedure at a magnetc acfield amplitude of 1600 A/m.
H. Naumoski et al. / Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
129
Fig. 4. Optical micrographs of the cross sections of the cutting edges of the investigated electrical steel by various techniques: (a) spark eroded, (b) punched and (c) laser cut.
Fig. 5. Measurement of the magnetization curves of the ring cores for spark erosion, punching and laser cutting before annealing.
Fig. 7. Relative additional losses at (a) 50 Hz and (b) 400 Hz of the ring cores for punching and laser cutting before annealing normalized to the spark eroded one. Fig. 6. Measurement of the dc-hysteresis of the ring cores for spark erosion, punching and laser cutting before annealing.
Table 1 Maximum permeabilities obtained from dc-magnetization measurements and coercivities obtained from dc-hysteresis measurements before annealing.
Permeability Coercivity
Table 2 Absolute power losses in W/kg of the 5 mm wide ring core specimens measured at 50 Hz; 1.5 T and 400 Hz; 1.0 T before annealing.
Eroded
Punched
Laser-cut
Pfe (W/kg)
Eroded
Punched
Laser-cut
8700 29 A/m
5800 35 A/m
1500 70 A/m
50 Hz; 1.5 T 400 Hz; 1.0 T
2.75 21.8
3.15 24.9
3.70 32.4
130
H. Naumoski et al. / Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
Fig. 8. Magnetization process obtained with the difference image approach near the cutting edges spark erosion, punching and laser-cutting (rows) of the specimens for two ac-field amplitudes 130 A/m and 1600 A/m (columns) before annealing.
Fig. 8 shows the difference domain images at the cut edges on applying a low and a high ac-field amplitude, calculated according to Eqs. (1) and (2), of 130 A/m and 1600 A/m respectively. The cut edge is at the top of the micrographs. For the spark eroded sample domain contrast can be observed over the whole imaging region of about 400 mm by applying a low field amplitude of 130 A/m. Only in a narrow region of under 100 mm directly at the cut edge the contrast is missing, indicating a degradation of the magnetic properties. By raising the field
amplitude to 1600 A/m these regions also show full domain contrast which means that they also contribute to the flux conduction under these conditions. The structure of the domain pattern at the cut edge looks similar to the reference pattern obtained, shown in Fig. 3. The punched specimen shows a different behavior. For a field amplitude of 130 A/m hardly any domain contrast can be found in a region of about 300–400 mm from the cutting edge. For a field amplitude of 1600 A/m this region of poor contrast shrinks to
H. Naumoski et al. / Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
131
Fig. 9. Optical micrographs of the cross sections of the cutting edges of the investigated electrical steel after annealing: (a) spark eroded, (b) punched and (c) laser cut.
Table 3 Maximum permeabilities obtained from dc-magnetization measurements and coercivities obtained from dc-hysteresis measurements after annealing.
Permeability Coercivity
Table 4 Absolute power losses in W/kg of the 5 mm wide ring core specimens measured at 50 Hz; 1.5 T and 400 Hz; 1.0 T after annealing.
Eroded
Punched
Laser-cut
Pfe (W/kg)
Eroded
Punched
Laser-cut
8800 29 A/m
8600 30 A/m
6900 34 A/m
50 Hz; 1.5 T 400 Hz; 1.0 T
2.75 21.6
2.78 21.8
2.86 22.8
Fig. 10. Magnetic polarization of the ring cores for spark erosion, punching and laser cutting after annealing.
Fig. 11. Magnetic polarization of the ring cores for spark erosion, punching and laser cutting after annealing.
about 100–150 mm, which is similar to the average grain size of our samples. The structure of the domains is also modified. A change of magnetic contrast is observed compared to unaffected material
Fig. 12. Relative additional losses of the ring cores for punching and laser cutting after annealing at f ¼50 Hz normalized to the spark eroded, annealed one.
shown in Fig. 3. A comparison with the integral magnetic measurements (Fig. 5) suggests that the lower permeability and the lower polarization at 1600 A/m of the punched specimen are due to the deterioration of the magnetic properties in the region close
132
H. Naumoski et al. / Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
Fig. 13. Magnetization process obtained with the difference image approach near the cutting edges spark erosion, punching and laser-cutting (rows) of the specimens for two ac-field amplitudes 130 A/m and 1600 A/m (columns) after annealing.
to the cut edge. For the ring used with a width of 5 mm a loss in polarization of 5% means that a zone of 250 mm width does not contribute to the polarization. Since the ring has two edges this means that each edge is associated with a magnetically dead zone of 125 mm. This result correlates well with the MOKE investigations where a degraded zone of about 100–150 mm is observed. For the laser cut sample we also observe little contrast in a region of 300 mm away from the cut edge for the low field amplitude of 130 A/m. At the field amplitude of 1600 A/m some regions close to the edge (see upper right-hand corner) still do not show magnetic contrast, whereas. Further away from the edge contrast is found at
this field amplitude. The domain patterns show wide and long stripe domains that look distorted. These results correlate with the integral magnetic properties obtained with the ring cores. For low field strengths there is nearly no domain wall movement near the edge and further inside the material which leads to a lower macroscopic permeability. The MOKE observations also shed light on the origin of the differences between the magnetization curves of the different samples (Fig. 5). Since the laser-cut sample does not show a large zone without contrast at high field amplitude, the global polarization of the sample gets larger than the polarization of the punched one at fields above 1000 A/m.
H. Naumoski et al. / Journal of Magnetism and Magnetic Materials 392 (2015) 126–133
3.2. Examinations after annealing Optical micrographs of cross sections of the annealed samples are shown in Fig. 9. No significant changes on the cut edges are visible. Only the punched and annealed edge shows regeneration of small grains at area of the burr (circle). Grain growth is not expected to have a significant effect here since the initial grain size of about 100 mm. Table 3 lists the maximum permeabilities and obtained coercivities after annealing. After the annealing all samples show improved magnetic properties when compared to the as-cut samples. The punched specimen reaches similar permeability and coercivity values after annealing as the spark eroded sample. The properties of the laser cut sample also improved after annealing, but the values are still lower than for the other samples. As shown in Fig. 10 the polarization at 130 A/m does not reach the value of the reference sample. However, all samples reach the polarization of the reference at field amplitudes above 1000 A/m. Fig. 11 shows the dc-hysteresis measurement, that confirms the results of the dc-magnetization. While the punched specimen has recovered, the laser-cut still shows a shear in the hysteresis and an increase in coercivity. The measurement of the losses also confirms the result of the dc-measurements. Table 4 lists the absolute losses measured at 50 Hz/1.5 T and 400 Hz/1 T, respectively. Fig. 12 shows the relative additional losses according to (4). While both punched and laser-cut samples show a reduction in the losses compared to the as-cut state, the laser cut sample still shows significantly higher losses than the spark eroded reference. Fig. 13 shows the difference domain images at the cut edges after annealing for field amplitudes of 130 A/m and 1600 A/m respectively. For the spark eroded sample only slight changes compared to the as-cut state can be observed. For the punched specimen we observe magnetic contrast all over the sample for both low and high field amplitude, comparable to the spark eroded and annealed specimen. The domain patterns at the edge are quite comparable to the reference image taken inside the material, showing no residual influence of the cutting process. This result corresponds well to the integral magnetic measurements of the punched and eroded ring cores after the annealing which show similar magnetic properties. However, the laser cut sample still shows impeded domain wall movement after the annealing for low field amplitude directly at the cut edge (box on the left hand side). This indicates that while most parts of the material have recovered, the magnetization processes are still affected by the cutting treatment in some regions close to the cut edge. This behavior can be linked to the integral magnetic properties which still show a lower permeability and higher losses after the annealing compared to the other cutting techniques. This behavior could be linked to a melting or heat affected zone (HAZ) in the immediate proximity of the cut edge due to the laser-cutting procedure. This was also observed by [15], where this zone was determined at about 50 mm which correlated well with the results obtained by the MOKE analysis in Fig. 13.
4. Conclusion It was shown that after cutting by different techniques the magnetic properties of electrical steel are affected. The micro-
133
magnetic behavior in the region near the cut edge has been studied using the magneto-optical Kerr-effect. As reference for unaffected magnetic properties specimens cut by spark erosion were used. The obtained micro-graphs show zones at the cut edges without magnetic contrast for excitations with two ac magnetic field amplitudes of 130 A/m and 1600 A/m. The global magnetic properties obtained with field-metric measurements correlate well with the micro-magnetic observations from the MOKE method. The results show that industrial punching has a negative effect on the magnetic properties. The maximum permeability decreases by 35% and the losses increase by 20% for low and high frequencies over a wide range of induction for the used geometry. This can be correlated with micro-magnetic observations using the Kerr effect: domain wall displacement is impeded and the domain structure is changed over the first two to three grain rows near the edge. The effect of laser cutting is even more pronounced. The permeability decreases to 17% of its initial value and the losses for small inductions increase by over 150%. The micro-magnetic investigations show that these changes correlate quantitatively with the area fraction of the regions affected by the cutting procedure. Stress relief annealing recovers the magnetic properties, as shown by the integral measurements and by domain observation. However, the laser cut sample still shows a degraded region of heat affection (HAZ) that results in slightly higher losses than that of the spark eroded reference sample. The method demonstrated here allows quantitative determination of the extension of the magnetically affected zone for different cutting procedures. In future, it may be used to identify cutting parameters which minimize the size of the affected zone. This would allow a reduction of the losses, or a reduction of the annealing temperature and time without intolerable deterioration of the magnetic properties.
References [1] R. Rygal, A.J. Moses, N. Derebasi, J. Schneider, A. Schoppa, J. Magn. Magn. Mater. 215–216 (2000) 687. [2] G. Crevecoeur, L. Dupre, L. Vandenbossche, R. Van de Walle, IEEE Trans. Magn. 44 (6) (2008). [3] G. Loises, A.J. Moses, IEEE Trans. Magn. 37 (4) (2001) 2755. [4] H.M.S. Harstick, M. Ritter, W. Riehemann, IEEE Trans. Magn. 50 (4) (2014). [5] A. Schoppa, J. Schneider, J.-O. Roth, J. Magn. Magn. Mater 215–216 (2000) 100. [6] A. Schoppa, J. Schneider, C.-D. Wuppermann, J. Magn. Magn. Mater. 215–216 (2000) 74. [7] K. Senda, M. Ishida, Y. Nakasu, M. Yagi, J. Magn. Magn. Mater. 304 (2006) 513. [8] M. Takezewa, Y. Wada, J. Yamasaki, T. Honda, C. Kaido, IEEE Trans. Magn. 39 (5) (2003) 3208. [9] M. Takezewa, K. Kitajima, Y. Morimoto, J. Yamasaki, C. Kaido, IEEE Trans. Magn. 42 (10) (2006) 2790. [10] H. Naumoski, L. Vandenbossche, A. Maucher, S. Jacobs, U. Herr, X. Chassang, in: Proceedings of the IEEE Conference on EDPC, 2014. [11] M. Emura, F.J.G. Landgraf, W. Ross, J.R. Barreta, J. Magn. Magn. Mater. 254–255 (2003) 358. [12] DIN 50460, 1988. [13] D.A. Herman Jr., B.E. Argyle, IEEE Trans. Magn. 22 (5) (1986) 772. [14] A. Huber, R. Schäfer, Magnetic Domains, Springer, Berlin, 1998. [15] A. Belhadj, P. Baudouin, F. Breaban, A. Deffontaine, M. Dewulf, Y. Houbaert, J. Magn. Magn. Mater. 256 (2003) 20.