You are on page 1of 7

d-q Space Vector Analysis for Line-Starting Permanent Magnet Synchronous Motors

Akeshi Takahashi, Satoshi Kikuchi, Hiroyuki Mikami, Kazumasa Ide and Andreas Binder
Abstract -- In order to boost up the calculation precision of d-q space vector analysis, new approach for estimating parameters of line-starting permanent magnet motors is developed. Introducing leakage-flux and magnetizing-flux variations dependant on not only stator excitation but also rotor one, transient characteristics and steady-state performance are calculated. Thorough the comparison with finite element analysis, the validity of the proposed method is verified. Especially, it is notable that it can provide at short times accurate quasi-steady-state average torque and precise critical load torque for self-starting.

Index Termsline-starting, space vector analysis, leakage flux, magnetizing flux

I.

INTRODUCTION

URING the design stage of line-starting permanent magnet (PM) synchronous motors, it is indispensable to ensure the sufficient starting characteristics as well as the rated performance. From the viewpoint of both time consumption and cost for constructing the test machine, the analytical approach on a computer that can realize rapid characteristic estimation and least expense is of importance. Numerical analysis methods for transient-state characteristics can be divided into two main groups: one is to analyze the electromagnetic field with the finite element method (FEM), and the other is to solve the basic equations with direct- and quadrature-axis space vector. The former can provide the accurate results including harmonic components due to complicated motor geometry by combining the electrical-circuit model and kinetic system [1]. On the other hand, the latter can realize the short-time analysis in exchange for less accuracy, and thus boosting up its calculation precision has been an important challenge and is the main aim of this paper. It should be noted that there is also another noble method: reluctance network analysis, which has been widely studied and used due to its compatibility between accurate and short-time calculation [2], [3], although it is outside the scope of this paper. The d-q space vector analysis needs parameters, such as inductance and resistance, and hence numerous approaches for the estimation of these values have been investigated and reported. In the early 1980s, Honsinger accomplished the first work introducing the constant parameters that included saturation effect of iron core, although these parameters were not analyzed but measured with a test motor [4]. Afterwards, as the analysis approach with the FEM had been improved [5], [6], it became possible to calculate such constant parameters without any measurement, taking into account the space harmonics [7] and the magnetic saturation [8], [9]. However, parameters
A. Takahashi, S. Kikuchi, H. Mikami, and K. Ide are with Hitachi Research Laboratory, Hitachi, Ltd., 7-1-1, Omika-cho, Hitachi-shi, Japan (e-mail: akeshi.takahashi.hc@hitachi.com). A. Binder is with the Institute for Electrical Energy Conversion, Darmstadt University of Technology, Landgraf-Georg-Strasse 4, D-64283, Darmstadt, Germany.

variations dependent on current changes had not been considered until it was measured and introduced into the dq space vector equations by Consoli [10]. His work was superior in terms of representing the flux-linkage variations as the function of both d-axis and q-axis stator current. In 1990s, with the development of computer performance, one became able to achieve the widely changing parameters dependent on the current variations by using the finite element analysis (FEA). Rahmans paper first introduced not only the parameters variations on the direct and quadrature axes but also d-q cross-coupling effect, and finally predicted steady-state characteristics with high accuracy [11]. Afterwards, the availability of transient-state calculation was also studied and presented [12]-[14]. However, treating the transient state with the d-q space vector analysis, one must pay attention to the fact that stator-side and rotor-side excitation has more or less different flux paths. This means that the magnetizing flux, which would be inherently equivalent whether stator or rotor excitation, can be dependent on its flux source (see Figs. 2 and 3), and hence the interference of both excitation in the magnetizing and leakage flux must be considered. From this viewpoint, the former studies have deficit, only dealing with the magnetizing-flux excited by the stator current; in [12], magnetizing flux in air gap generated by the stator excitation was adopted and leakage flux was neglected; in [13] and [14], although the leakage reactance attributed to the rotor excitation was calculated only for each slip, neither leakage- nor magnetizing-flux maps related to the rotor current were treated. Therefore, more consideration about how far the flux variation can be affected by the excitation is kind of needed. It should be noted that combinations of stator-current and rotor-current input are so myriad that the perfect map for their whole variation is difficult to make. And also, such complicated works should be avoided from the viewpoint of the simple and short-time design of the d-q space vector analysis. In this paper, the leakage-flux and magnetizingflux behavior are investigated by the provisional FEA, and it is discussed how the magnetizing-flux linkage, which would be common between the stator and rotor sides, should be assumed. And then, the obtained parameters are used for the d-q space vector analysis. To verify the validity of the introduced method, the transient-state performances are calculated and compared with the FEA results. All analyses are performed for a two-pole prototype motor with PN = 5 kW, nN = 3000 min-1, VN = 200 V, Y-connection (see data in Table 1 and Fig. 1). Neither the skin effect nor the d-q cross-coupling effect is taken into account, which will be studied in a future report. II. BASIC THEORY OF CONVENTIONAL METHOD The transient-state equations for line-starting PM synchronous motors are expressed in per-unit values in the d-q-reference frame:

978-1-4673-0141-1/12/$26.00 2012 IEEE

134

TABLE I DATA OF PROTOTYPE MOTOR Outer diameter of stator Inner diameter of stator Axial length of iron core Number of poles Number of slots per pole and phase Stator slot type Stator slot height Stator teeth width Winding connection Number of rotor slots PM material PM remanent flux density PM relative permeability Rotor cage material Output power Rated speed 160 mm 90 mm 90 mm 2 5 semi-closed 13.5 mm 3.8 mm Y 22 Nd-Fe-B 1.20 T 1.04 aluminum 5 kW 3000 min-1

torque, respectively, is per-unit time: = N t, and J is starting time constant: J = N TJ , where N = 2 fN and TJ = 319.8 ms. Reference values for the per-unit system are the peak values of the rated phase voltage
2 UN,ph = 2 115.5 V and of the rated current 2 IN = 2 14.5 A, the rated frequency fN = 50 Hz. For simplicity, the symbol prime is omitted in the following notation, and the rotor-side parameters are basically represented only with the capital subscript. In (5) to (8), one can redefine the magnetizing-flux linkages dm and qm as the function of the currents:

x dm id + x dm i D + pd = dm (id + i D )
(10) (11) On the other hand, the leakage flux is expressed as

x qm iq + xqm iQ = qm iq + iQ

PM Rib Flux barrier

x D i D = xrb i D + D (i D ) xQ iQ = xrb iQ + Q iQ

x s iq = x sb iq + q i q

x s id = x sb id + d (id )

( )

(12) (13) (14) (15)

( )

Cage bar

Fig. 1. Rotor cross section.

d = (x dm + x s ) i d + x dm i D ' + pd q = (x qm + x s ) i q + x qm iQ Q ' = x qm i q J
'

d d m q d d q u q = rs i q + + m d d d D ' 0 = rD ' i D ' + d d Q ' ' ' 0 = rQ iQ + d u d = rs i d +

(1) (2)

(3)
(4) (5) (6) (7) (8) (9a) (9b)

D ' = x dm i d + x dm + x D ' i D ' + pd


qm

( + (x

+ xQ '

) ) i

'

d m = me m L d

me ( ) = iq ( ) d ( ) i d ( ) q ( )

The subscripts d and q represent direct- and quadratureaxis stator quantities, respectively, while D and Q represent direct- and quadrature-axis rotor cage quantities, and s represents stator. The primed values signify rotor quantities related to the stator winding data via a transformer ratio. i, u and are current, voltage and flux linkage space vector components, respectively, r is winding resistance, xm and x are main and leakage inductance, respectively, p is PM flux linkage of the stator winding, m is mechanical angular velocity, me and mL are electromagnetic torque and load

where xsb and xrb represent the overhang leakage flux of the stator and the rotor, respectively, and d , q , D and Q represent the slot leakage flux. The reason to separate the overhang leakage flux from the slot leakage flux is that the former comes from the conventional analytical formula, while the latter can be determined directly from the provisional FEA. In what follows, for example, the dm which adopts the id input in the provisional FEA is expressed as dm(id), while the d with the id input is expressed as d(id), and the others are subject to the similar manner. The flux curves used for the space vector analysis comprise fundamental space fluxes. Fig. 2 shows physically what the d, D, dm , d and D represent. Fig. 2 also depicts the difference in flux linkage due to the only stator excitation and the only rotor excitation, where the overhang leakage flux xsb id is neglected. The magnetizing flux dm(id) and dm(iD) shown in (a) and (b), respectively, which would be inherently equivalent, can flow more or less different flux paths, and hence can be non-identical. Fig. 3 illustrates the flux line chart of the dm(id) and dm(iD) generation, with the following conditions: a) id = -14.6 pu, b) iD = -14.6 pu. Obviously, the leakage flux paths in (a) are different from those in (b); at the center of pole, the leakage flux in (a) occurs over eight teeth, while that in (b) occurs over ten teeth. This leads to the difference in the magnetic resistance and hence in the total flux generated by the same magnetomotive force. Fig. 2 also implies that the slot leakage flux d might be dependent on not only id but also iD, because the flux generated by the iD input causes the interference in the main flux path and hence the leakage flux path in the stator. Therefore, the d should be expressed as the function of id and iD. However, combinations of stator-current and rotorcurrent input are so myriad that the perfect map for their whole variation is difficult to make. And also, such
*

135

Stator Gap PM

d (id)

Stator Gap PM

D (iD)

Flux linkageydhd ,(p.u.) (pu) yd, dm

1.5 1 0.5 0 -0.5 -1 -1.5 -2 -15 -10

d(id) d (i d ) dh(id) dm (i d )

d (id)
Rotor

D (iD)
Rotor

dm (id)

dm (iD)

(a) with id input (iD = 0) Fig. 2. Schematic of flux linkage.

(b) with iD input (id = 0)

-5

0 5 i d (p.u.) (pu)

10

15

(a) d (id) and dm (id) curves


0.05

Leakage d (p.u.) (pu) Flux d

0 -0.05 -0.1 -0.15 -0.2 -0.25 -15 -10 -5 0 5 i d (p.u.) (pu) 10 15

(a) dm(id)

Fig. 3.

(b) dm(iD) Flux line chart (a): id = -14.6 pu, b): iD = -14.6 pu).

Fig. 4.

(b) d (id) curve FE Analysis results of direct-axis flux linkage with id input.

complicated works should be avoided from the viewpoint of the simple and short-time design of the d-q space vector analysis. In the next chapter, it is discussed how the magnetizing and leakage flux should be formulated. III. PROPOSED METHOD In the proposed method, the provisional FEA determines the parameters d , q , D , Q , dm and qm according to the two postulates described in the following terms A and B.

q (i q ) = q (i q ) Q (i q )
= q i q qm i q

D (i D ) = D (i D ) d (i D ) = D (i D ) dm (i D ) Q (iQ ) = Q (iQ ) q (iQ )


= Q iQ qm iQ

( )

( )

(17) (18) . (19)

( )

( )

A.

Leakage Flux

It is first assumed that the leakage flux d is only dependent on id but not influenced by any other current input. In the same way, q is only dependent on iq, D on iD, and Q on iQ. This postulate is derived from the original equations (5) to (8). For example, setting iD in (5) and (7) at zero, and assigning (12) to (5), one can obtain

d (i d ) = d (i d ) D (i d ) = d (i d ) dm (i d )
*

(16)

where the overhang leakage flux xsb id is neglected because the provisional FEA is 2-D field solutions. Although in reality the slot leakage flux d might be dependent on not only id but also iD, there is no means for justifying the d behavior in the case of the coupled inputs of id and iD. The other leakage flux q (iq), D (iD), and Q (iQ) are subject to the same manner:

In the FEA, varying the d-axis stator current id and keeping the rotor current iD set at 0 pu, d(id) and dm (id) curves can be achieved as shown in Fig. 4. Calculating the difference of these two curves, d (id) can be obtained. As it is clear from Fig. 4(b), d (id) is not linear function of id, but the saturated curve. Normally, the leakage flux is represented by constant inductance, and hence is proportional to the current input. However, one has to pay attention to the fact that a strict linear characteristic of the leakage flux is only based on the linear property of the magnetic steel sheet. In other words, any saturation of iron core expropriates the linearity of main-flux and leakageflux variation. This is because the saturation in the main flux path increases a total magnetic resistance, and hence the total flux generated under the constant magnetomotive force is decreased, leading to the nonlinearity of the leakage flux. More detailed explanation with the magnetic circuit described in chapter 4 can help to understand this phenomenon. It should be noted that the minus value of d at id = 0 originates in the fact that the rotor flux linkage D (id) comprises whole flux generated by PMs while the stator flux linkage d (id) does not include the leakage flux in the rotor.

136

Varying the d-axis rotor current iD and keeping the stator current id set at 0 pu, D (iD) and dm (iD) curves can be achieved as shown in Fig. 5. Calculating the difference of these two curves, D (iD) can be obtained. In this case D (iD) is approximately the linear function of iD, as is well known. But it should be noted that the D variation is not strictly linear. The detailed behavior of the D can also be explained by the magnetic circuit as described in chapter 4. In the same way, the q (iq) is not the linear function of iq but the saturated curve, while the Q (iQ) is approximately the linear function of iQ.

Flux linkage D ,(p.u.) (pu) dm

1.5 1 0.5 0 -0.5 -1 -1.5 -2 -15 -10

D (iD ) D(iD) dh(iD) dm (iD )

-5

0 (pu) i D (p.u.)

10

15

Leakage D (p.u.) (pu) Flux D

Magnetizing Flux This term investigates magnetizing flux behavior, and introduces the second postulate for dm and qm. According to (5), (7) and (10), the FE analysis with a single id input yields D (id) = dm (id), while a single iD input yields d (iD) = dm (iD). Fig. 6 shows the FEA results of dm(id) and dm(iD). dm denotes the difference between dm(id) and dm(iD) :
dm = dm (i d ) dm (i D )

B.

(a) D (iD) and dm (iD) curves


0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 -15 -10 -5 0 5 (pu) i D (p.u.) 10 15

(20)

Flux linkage (p.u.) dm (pu)

Basically, dm(id) and dm(iD) should be identical and hence dm should be constantly zero, but actual curves of them are not exactly identical because of the difference in the local flux paths. Aiming at the simple treatment of dm, it is secondly assumed that the magnetizing flux dm is expressed by the only one function dm (i d + i D ) dependent on the sum of id and iD:

Fig. 5.

(b) D (iD) curve FE Analysis results of direct-axis flux linkage with iD input.

2 1.5 1 0.5 0 -0.5 -1 -1.5 -2 -15 -10

dm (i d dh(id)) dm (i D dh(iD))

dm

id + i D dm (id ) + dm (i D ) = 2 2

(21)

where id = iD. This postulate arises one question: how widely the dm can cover the actual dm generated by myriad combinations of id and iD. According to (5) and (10), the magnetizing flux generated by the simultaneous inputs of id and iD can be separated from the total flux linkage d (id, iD):

-5

0 5 (pu) i d , i D (p.u.)

10

15

(a) dm (id) and dm (iD) curves


0.2

Deviation dm (pu) dm (p.u.)

dm (i d + i D ) = d ( id , iD ) d (id )

(22)

where the assumption defined in term A is still valid that d(id) is only dependent on id but not influenced by any other current input. Fig. 7 shows FEA results of the dm calculated by (21) and the dm by (22). Although some deviations are recognized, it is possible to substitute the dm for dm regardless of the combinations of id and iD. By the way, according to (7) and (10), the magnetizing flux dm is expressed in another way:

0.15

0.1 0.05 0 -15 -10 -5 0 5 (pu) i d , i D (p.u.) 10 15

dm (id + i D ) = D ( id , iD ) D (iD )

(23)

Fig. 6.

(b) dm FE Analysis results of difference between dm (id) and dm (iD).

Fig. 8 shows the FEA results of the dm calculated by (21) and the dm by (23). Although some deviations become bigger than that in Fig. 6, the magnetizing flux defined in (21) is to be introduced in the proposed method. In the same way, it is assumed that the magnetizing flux qm is expressed by the only one function qm i q + iQ

dependent on the sum of iq and iQ:

qm

i q + iQ qm (i q ) + qm (iQ ) = 2 2

(24)

where iq = iQ.

137

2 1.5 + i D ) (pu) 1 0.5 0 -0.5 -1 -1.5 -2 -15 -10 -5 0 i d + i D (pu) 5 iD = -14.6 pu iD = -9.8 pu iD = -4.9 pu iD = 0 pu iD iD = 4.9 pu iD = 9.8 pu iD = 14.6 pu dm dm (ave.) 10 15

Stator

Air gap

Rotor

R Rs R r

Fr Fp

dm (i

1
Fs

R s

Rp Rr

Fig. 7.

FE analysis results of direct-axis magnetizing-flux linkage dm calculated by (21) and the dm calculated by (22).
2 1.5 dm (i d + i D ) (pu) 1 0.5 0 -0.5 -1 -1.5 -2 -15 -10 -5 0 i d + i D (pu) 5 id = -14.6 pu id = -9.8 pu id = -4.9 pu idid = 0 pu id = 4.9 pu id = 9.8 pu id = 14.6 pu dm (ave.) dm 10 15

Fig. 9. Simple magnetic circuit for the line-starting PM motor (R: magnetic resistance, F: magnetomotive force, : flux; subscripts s: stator, r: rotor, : air gap, p: permanent magnet, : leakage).

TABLE II RELATIONSHIP BETWEEN K AND C = 2K/(1.2K + 0.1). k C 0.1 0.91 0.2 1.18 0.3 1.30 0.5 1.43 1.0 1.54 1.5 1.58

l d (id ) =

1 (1.9 Fs 0.09 F p ) . R s

(26)

Fig. 8.

FE analysis results of direct-axis magnetizing-flux linkage dm calculated by (21) and the dm calculated by (23).

However, in reality, the magnetic saturation of the iron core makes nonnegligible the magnetic resistance such as R s. Assuming R s = kR s , where k is nonlinear coefficient, and applying the same conditions as the above linear case to the other magnetic resistance, one can obtain
nl nl nl d (id ) = d (id ) D (id ) = 1nl + 3nl

IV. THEORETICAL PROOF OF PROPOSED MODEL This chapter theoretically proves the nonlinear property of the leakage flux and the inconsistent property of the magnetizing flux described in chapter 3.

1 R s

2k 0.2 Fs + Fp 1.2k + 0.1 1.2k + 0.1

(27)

where superscript nl means the nonlinear case. When k = 0.1 or R s = 0.1R s , nld (id) is
nl d (id ) =

A.

Theoretical Leakage Flux Fig. 9 shows the simple magnetic circuit for the linestarting PM motor. R represents magnetic resistance, F represents magnetomotive force, and represents flux. The subscripts s is stator, r is rotor, is air gap, p is permanent magnet, and is leakage. Applying the Kirchhoffs second law for the whole closed loops and solving the equations, one can calculate . Assuming the linear case, i.e., infinite permeability of iron core, the magnetic resistance R s and R r can be neglected. And more, setting the stator-side magnetomotive force Fs at an arbitrary value except for zero and the rotor-side Fr at zero, 1 l represents ld (id), while 3 l represents lD (id), and hence ld (id) is expressed as
Rp 1 = + R R + R R +R R R s r p p r R Fs Fp R r R + R R p + R pR r

1 R s

(0.91Fs 0.91F p )

(28)

Comparison of (26) with (28) clarifies that the nonlinear case yields less leakage flux, and that the leakage flux is dependent on the magnetic resistance R s or the iron core saturation. If the R s increases due to more saturation, the leakage flux will further decrease: for example, k = 0.5 or R s = 0.5R s , which can be caused by huge magnetomotive force Fs, leads to the significant saturation of the leakage flux
nl d (id ) =

1 (0.29 Fs 1.43Fp ) . R s

(29)

l l l d (id ) = d (id ) D (id ) = 1l + 3l

Applying the same conditions as the above, and setting the Fs at zero and the Fr at an arbitrary value except for zero, D (iD) in the nonlinear case can be expressed as
D (iD ) = D (iD ) d (iD ) = 3 + 1 =
1 2k ( Fr + Fp ) R s 1.2k + 0.1
=C

(25) where superscript l means the linear case. Since all the magnetic resistance and Fp in (25) are constant, ld (id) is the linear function of magnetomotive force Fs or current. Applying rough approximation of R = 0.01R s , R p = 0.1R s and R r = R s to (25), ld (id) can be calculated as

(30) Table 2 represents the relationship between k and the coefficient C = 2k/(1.2k + 0.1). As is clear from (30) and Table 2, the D is nonlinear function of Fr. However, one

138

B.

Theoretical Magnetizing Flux Setting the Fs at an arbitrary value except for zero and the Fr at zero, 3 is equal to D (id) or dm (id):
R R r Fs + (R s + R s ) (R s + R + R r ) R s2 Fp = s det R

dm (id ) = 3

Torque (pu)

has to pay attention to the fact that the increasing rate of C is much more moderate than that of k; for example, comparing k = 1.0 with k = 0.1, the k increases 10 times while the C increases only 1.7 times, which results in the apparent linear characteristic of D .

12 10 8 6 4 2 0 -2 -4 -6 -8 0 0.2

d-q space vector analysis FEA

0.4 Time (s)

0.6

(a) Torque versus time


1.2 1 Rotation (pu) 0.8 0.6 0.4 0.2 0 -0.2 0 0.2 Time (s) 0.4 0.6 d-q space vector analysis FEA

(31) where
detR = (R s + R s ) (R s + R + R r ) (R r + R p + R r ) (R s + R s ) R r2 (R r + R p + R r ) R s2

(32) Inversely, setting the Fr at an arbitrary value except for zero and the Fs at zero, 1 is equal to d (iD) or dm (iD):

dm (iD ) = 1
= R s R r Fr + R s R r F p det R
U- phase current (pu)

(33)

(b) Rotation versus time


15 10 5 0 -5 -10 -15 0 0.2 Time (s) 0.4 0.6 d-q space vector analysis FEA

Comparison of (31) with (33) clarifies that dm (id) and dm (iD) are generated by the different flux paths. V. VALIDITY OF PROPOSED METHOD

The obtained parameters in chapter 3 are used for the d-q space vector analysis. In order to verify the validity of the proposed method, the transient-state performances are calculated and compared with the FEA results. Fig. 10 shows the results of the d-q space vector analysis during start up, with the following conditions; u = 1.00 pu, fs = 1.00 pu, J = 100.48 pu, mL = 0 pu. For comparison, the FEA time-stepping results are also represented. Due to the switching on of the stator voltage, the 50Hz starting current and the DC component occur in the stator winding, and results in the 50Hz-pulsating starting torque. During the start-up, all results exhibit a good agreement. There are some errors between the d-q space-vectoranalysis results and the FEA results, which is mainly because the d-q space vector analysis takes into account neither the d-q cross-coupling effect nor the influence of the current displacement. Also, it may be another reason that the provisional FEA determines the parameters according to the two postulates described in chapter 3. Fig. 11 shows the quasi-steady state characteristics. In order to verify the advantages of the proposed method, the results calculated with the conventional method are also presented. The conventional method only considers the magnetizing flux in the air gap generated by the stator excitation, while the leakage flux is neglected, leading to the big deviations as shown in Fig. 11(a). On the other hand, the proposed method exhibits better agreement with the FEA results, as shown in Fig. 11(b). Even in the proposed method, the peak values around slip = 1 do not agree with the FEA results. The reason of the errors is that the d-q cross-coupling and harmonic effects contribute to pulsating-torque generation. Table 3 represents the detailed

(c) U-phase current versus time Fig. 10. Computation of FEA and d-q space vector analysis results at no-load starting (u = 1.00 pu, fs = 1.00 pu, mL = 0 pu).

data of the quasi-steady state characteristics. The deviation between the conventional method and the FEA becomes more than 20 %, while the proposed method offers less error within 10 %. The calculation accuracy of the average torque in the quasi-steady state results in the precision of critical load torque for the self-starting. It is 1.25 pu in the FEA (starting-success up to this value) whereas 1.52 pu in the proposed method. The deviation is 22%, which comes from the calculation errors shown in Table 3. In the proposed method, the critical load torque is 1.31 pu, which exhibits a good agreement with the FEA. These results indicate that the proposed method enables one to estimate accurate slip-versus-torque curves and starting capability at short times. VI. CONCLUSION In order to boost up the calculation accuracy of the d-q space vector analysis, the leakage-flux and magnetizingflux behavior were investigated and the obtained parameters were used for analysis program. First through the provisional investigation, it was found

139

12

Electromagnetic torque (pu)

Maximum torque 8 Average torque 4 0 -4 1 0.8 Minimum torque 0.6 0.4 Slip (pu) 0.2 0

TABLE III DETAILED DATA OF QUASI STEADY-STATE CHARACTERISTICS.


s =1 Ave. torque Deviation (pu) (%) 2.89 100 4.11 142 3.05 105 s = 0.5 Ave. torque Deviation (pu) (%) 2.24 100 3.00 134 2.44 109 s = 0.1 Ave. torque Deviation (pu) (%) 0.73 100 0.87 120 0.76 105

FEA Conventional Proposed

[7] [8] [9]

(a) Conventional method


12

Maximum torque 8 Average torque 4

[10] [11]

0 -4 1 0.8 Minimum torque 0.6 0.4 Slip (pu) 0.2 0

[12]

(b) Proposed method Fig. 11. Quasi steady-state characteristics (solid lines: d-q space vector analysis, dotted line: FEA).

[13] [14]

that - leakage-flux curve of stator excitation is saturated due to the magnetic resistance in the stator core, while that of rotor excitation is nearly proportional to current input, - the magnetizing flux, which would be inherently equivalent whether stator or rotor excitation, is not always identical but dependent on its flux source. Second, in order to verify the validity of the proposed method, simulation results were compared with the FEA results. It was found that - the quasi-steady state characteristics and the critical load torque for self-starting exhibited the good agreement with the FEA results, which indicted that the method enables one to estimate accurate starting characteristics at short times. VII.
[1] [2] [3] [4] [5]

A. Ishizaki and Y. Yamamoto, Asynchronous performance prediction of ac permanent magnet motor, IEEE Trans. Energy Conv., vol. EC-1, no. 3, pp. 101-108, Sep. 1986. M. A. Rahman and A. M. Osheiba, Performance of large line-start permanent magnet synchronous motors, IEEE Trans. Energy Conv., vol. 5, pp. 211217, Mar. 1990. S. M. Osheba and F. M. Abdel-Kader, Performance analysis of permanent magnet synchronous motors part:II operation from variable source and transient characteristics, IEEE Trans. Energy Conv., vol. 6, no. 1, pp. 8389, Mar. 1991. A. Consoli and A. Abela, Transient performance of permanent magnet AC motor drives, IEEE Trans. Ind. Appl., vol. IA-22, no.1, pp 32-41, 1986. M. A. Rahman and P. Zhou, Determination of saturated parameters of PM motors using loading magnetic fields, IEEE Trans. Magn., vol. 27, no. 5, pp. 39473950, Sep. 1991. I. Iglesias, L. Garcia and J. Tapplrit, A d-q model for the selfcomtated synchronous machine considering the effects of magnetic saturation, IEEE Trans. Energy Conv., vol. 7, no. 4, pp. 768776, Dec. 1992. P. Zhou, M. A. Rahman and M. A. Jabber, Field circuit analysis of permanent magnet synchronous motors, IEEE Trans. Magn., vol. 30, no. 4, pp. 13501359, July 1994. M. A. Rahman and P. Zhou, Field-based analysis for permanent magnet motors, IEEE Trans. Magn., vol. 30, no. 5, pp. 36643667, Sep. 1994.

Electromagnetic torque (pu)

VIII.

BIOGRAPHIES

Akeshi Takahashi (M08) received the M. Eng. degree from Hokkaido University, Sapporo, Japan, in 2004, and Dr.-Ing. (Ph.D.) degree from Darmstadt University of Technology, Darmstadt, Germany, in 2010. Since 2004, he has been with Hitachi Research Laboratory, Hitachi Ltd., where he is engaged in rotating machine research and development. He was a Visiting Researcher in Darmstadt University of Technology from 2007 to 2008. Satoshi Kikuchi graduated Miyagiken Technical High School, Sendai, Japan, in 1988. Currently, he is with Hitachi Research Laboratory, where he is involved in rotating machine research and development as a Senior Researcher. He has been with Hitachi Ltd. since 1988. Hiroyuki Mikami (M95) received the M. Eng. degree from Ibaraki University, Hitachi, Japan, in 1990, and Ph. D. degree from Tohoku University, Sendai, Japan, in 2008. Currently, he is a Manager with Hitachi Research Laboratory, where he is involved in rotating machine research and development. He has been with Hitachi Ltd. since 1990. Kazumasa Ide (M94) received the M. Eng. and Ph. D. degrees from Tohoku University, Sendai, Japan, in 1988 and 1994, respectively. Currently, he is a Manager with Hitachi Research Laboratory, where he is involved in electric power conversion system. He has been with Hitachi Ltd. since 1988. Andreas Binder (M97SM04) received the Dipl.-Ing. (diploma) and Dr. Tech. (Ph.D.) degrees in electrical engineering from the University of Technology, Vienna, Austria, in 1981 and 1988, respectively. From 1981 to 1983, he was with ELIN-Union AG, Vienna, where he worked on the design of synchronous generators. From 1983 to 1989, he was with the Department of Electrical Machines and Drives, Technical University, Vienna. After this, he joined Siemens AG, first in Bad Neustadt, Germany, then in Erlangen, Germany. His main tasks included the development of dc and inverter-fed ac drives. Since October 1997, he has been the Head of the Institute of Electrical Energy Conversion, Darmstadt University of Technology, Darmstadt, Germany, where he is also a Full Professor. Dr. Binder was the recipient of the Power Engineering Society (ETG) Literature Award in 1997.

REFERENCES

[6]

K. Kurihara and M. A. Rahman, Steady-state performance analysis of permanent magnet synchronous motors including space harmonics, IEEE Trans. Magn., vol. 30, pp. 13061315, May 1994. V. Ostovic, Computation of saturated permanent magnet ac motor performance by means of magnetic circuits, IEEE Trans. Ind. Appl., vol. IA-23, no. 5, pp. 836-841, Sept./Oct. 1987. V. Ostovic, A simplified approach to the magnetic equivalent circuit modeling of electric machines, IEEE Trans. Ind. Appl., vol. 24, no. 2, pp. 308-316, Mar./Apr. 1988. V. B. Honsinger, Permanent magnet machine: asynchronous operation, IEEE Trans. Power App. Syst., vol. PAS-99, pp. 1503 1509, July 1980. K. Miyashita, S. Yamashita, S. Tanabe, T. Shimozu and H. Sento, Development of a high speed 2-pole permanent magnet synchronous motor, IEEE Trans. Power App. Syst., vol. PAS-99, pp. 2175-81, 1980. V. B. Honsinger, The fields and parameters of interior type of ac permanent magnet machines, IEEE Trans. Power App. Syst., vol. PAS-101, no. 4, pp. 867-876, Apr. 1982.

140
Powered by TCPDF (www.tcpdf.org)

You might also like