Stator Inter-turn Fault Detection of Synchronous Machines Using Field Current Signature Analysis

Prabhakar Neti, Student Member, IEEE Subhasis Nandi, Senior Member, IEEE
Department of Electrical and Computer Engineering University of Victoria Victoria, BC, Canada pneti@ece.uvic.ca, snandi@ece.uvic.ca
Abstract Our recent observations suggested that harmonics in the field current are very promising to detect stator inter-turn faults in synchronous machines. So far, an increase in some of the even harmonics in the field current has been reported to detect such faults. However no explanation has been provided for the cause of these harmonics. Moreover, the even harmonics can significantly increase with supply unbalance as well as time harmonics, which can lead to a serious confusion. Hence, in this study, an in-depth investigation was conducted to determine the origin of various harmonic components in the field current and their feasibility to detect stator faults. It was found that, owing to structural asymmetries of the field winding, some of these components clearly increased with stator inter-turn fault. The findings are helpful to detect faults involving few turns without ambiguity, in spite of with supply unbalance and time harmonics. Both simulation and experimental results are presented in this paper. The diagnosis results have also been verified using a rotormounted search-coil, which can also be used to detect even a oneturn stator fault very effectively. Keywords- Inter-turn fault,Synchronous motor,WFA, MWFA

I.

INTRODUCTION

The loss of dielectric strength of the winding insulation in electric machines leads to inter-turn faults. The time elapsed in the growth of an incipient fault leading to a catastrophic failure of the machine varies from a third of a second to several minutes in the case of random-wound machines and can even be much longer in the case of form-wound machines [1]. Hence, an early detection of such faults involving few turns is very much desirable in order to protect them. Many researchers have analyzed the field current of synchronous machines against stator and rotor inter-turn faults [2-4]. Particularly, an increase of 8th harmonic in the field current of a 4-pole synchronous generator with a stator fault was highlighted in [2]. A detailed analysis has been presented on brushless synchronous generators using the 180 Hz positivesequence stator voltage as an indicator of stator fault, the 30, 90, and 150 Hz armature circulating currents as indicators of field winding deterioration, and the 120 Hz exciter field current as an indicator of rotating rectifier diode shorts [3]. As discussed later in this paper, some of the even harmonics in the field current (including the 8th harmonic) can increase with supply unbalance and time harmonics that can
Funding provided by: NSERC, CFI and UVic

lead to a serious confusion. Hence, in order to unambiguously detect stator inter-turn faults in salient-pole synchronous motors, some of the other frequency components in the field current have been monitored. It was observed that owing to the structural asymmetries of the field winding, a stator inter-turn fault could increase these components in the field current. The frequency components that showed a clear increase due to stator faults but were least affected by the supply unbalance and time harmonics have been highlighted. Internal asymmetry of the machine, that affected earlier diagnosis results adversely, has now been used to an advantage for the first time. Hence unambiguous detection of inter-turn faults was possible with high sensitivity. A detailed mathematical analysis of this phenomenon has also been presented. In practice, all the machines are expected to possess internal asymmetries due to manufacturing imperfections and nonhomogeneity of iron. The asymmetries can be in the magnetic path, stator winding, air-gap and/or rotor cage/winding. In the present study, simulation results have been obtained by introducing asymmetry in the field winding. Experimental results have been obtained by physically shorting different number of stator turns in a 2 kW motor. A comparative study of the characteristic frequency components has been put forward with different fault levels and supply unbalance. The influence of load and power factor of the machine on the test results is also discussed. The test results have also been verified by performing a harmonic analysis on the voltage induced in a rotor-mounted search-coil, which can be also be used for stator fault detection very effectively, even up to one turn. In order to protect the motor using the proposed scheme, only one signal needs to be monitored. Hence it is very easy to implement this scheme compared to the negative sequence technique [1] that requires at least six sensors (three for line voltages and three for line currents). Less number of sensors implies less number of channels of the data acquisition system, less data processing and therefore a very high-speed tripping of the faulty machine at a lower cost. II. DESCRIPTION OF MOTOR AND SELECTION OF FAULT

The salient-pole synchronous motor under consideration is a 208V, 1800 rpm, 60 Hz, 2 kW, 4-pole random-wound machine having 36 stator slots. The stator has a 3-phase double layer lap winding and the rotor has 20 damper bars. The field winding of the machine comprises of two series-connected coils, each placed on the diagonally opposite rotor poles. This

2360
1-4244-0365-0/06/$20.00 (c) 2006 IEEE

has been done for the ease of construction. Each coil side occupies each of the inter-polar gaps. The cross-section of rotor of synchronous motor with two-coil field winding is shown in Fig. 1. A search-coil is also mounted on one of the pole faces of the rotor. Also, several taps have been brought out from the stator winding (phase-a) to physically create the inter-turn faults. Ideally, though the flux pattern produced by the two-coil field winding will be similar to that produced by a standard four-coil field winding (one coil on each pole), their turns functions are different as given later in (9) and (10). However, due to inevitable manufacturing and magnetic asymmetries, the turns function of any actual field winding will be more like the one given in (11). This will be referred to as the asymmetric field winding. MECHANISM OF INDUCTION OF EVEN HARMONICS IN FIELD CURRENT UNDER SUPPLY UNBALANCE With unbalanced 3-phase voltages, a balanced 3-phase stator winding produces a reverse rotating MMF wave given as III.

Br can induce a particular frequency component in a winding if and only if the pole-pair of Br corresponding to that frequency matches with one of the pole-pairs of the winding [5]. From (4), the 120 Hz component is produced with n = 1 and f = 60 Hz. The pole-pair associated with 120 Hz component is n*p = 2 for the chosen motor. Hence, the 120Hz component can be induced in the two-coil, four-coil and asymmetric field windings, since all of these windings can produce the pole-pair number 2 as seen in (9)-(11). For n = 5, 360 Hz component can be produced. Similarly, due to supply unbalance, other even harmonics (including 8th harmonic) can be induced in the two-coil, four coil and asymmetric windings whose turns functions are given in (9)-(11). Clearly, since the 8th harmonics was used for fault detection in [2]; ambiguous detection is very likely. IV. MECHANISM OF INDUCTION OF VARIOUS HARMONIC COMPONENTS IN FIELD CURRENT WITH STATOR FAULT In the case of a synchronous motor with stator inter-turn fault, induction of various harmonic components in the field winding can be proved as follows. The stator has a shorted loop (can thus be treated as a single-phase winding) carrying current at supply frequency that generates two counter-rotating MMF waves [6], which can be given as

Fs = As cos [np + t ]

(1)

n = (6k1) other than fundamental, is the order of space harmonics produced by a balanced 3-phase stator winding, k = 1, 2, 3... etc., = angular position with respect to the stator, p = no. of fundamental pole-pairs = 2, = 2f and f = supply frequency. Considering only the DC part of the specific permeance function (P0), the flux density produced by this MMF, with respect to stator, can be given as

Fsf = A f cos [k t ]

(5)

Where, k = 1, 2, 3 ... etc. Considering the DC part of the specific permeance function (P0), the flux density produced by this MMF, with respect to stator, can be given as

B s = As P0 cos (np + t )
or with respect to rotor,

(2)

B sf = A f P0 cos (k t )
or with respect to rotor,

(6)

B r = As P0 cos (np '+ np s t + t + 1 )

(3)

B rf = A f P0 cos (k '+ k s t t + 2 )
Substituting s = /p in (7), we can have

(7)

Here, = angular position with respect to the rotor and s = synchronous speed in radian/second. Substituting s = /p in (3), the flux density with respect to rotor, can be given as

Brf = A f P0 cos [k '+{( k / p ) 1} t + 2 ]

(8)

B r = As P0 cos {np '+ ( n + 1) t + 1 }

(4)

Figure 1. Rotor cross-section of synchronous motor with two-coil field winding.

The components that showed promise from experiments are 60, 90, 150 and 210 Hz. With p =2 and k = 4, from (8), the 60 Hz component can be produced. Since k = 4, the 60Hz component can be induced in the symmetric winding of two-coil structure and the asymmetric windings of both structures, since the asymmetric winding can produce all k = 1, 2, 3...pole-pairs, as given in (11). However its magnitude is expected to be much higher for the two-coil structure since the pole pair number 4 arises out of the symmetric structure itself. Similarly, with k = 1, 3, 5 and 7, the 90Hz and 150Hz components can be induced in the asymmetric field winding. These components cannot be induced in the ideal two and four coil field windings, since these windings produce pole pair numbers 2, 4, 6etc. (9) and 2, 6, 10...etc. (10) respectively. Using a similar argument, it can be shown that all other even harmonics (120, 240, 360, 480 Hz, ... etc.) can also be induced in the ideal two-coil, four-coil as well as asymmetric field winding due to an inter-turn fault in the stator winding. Table I shows for different values of k and p = 2 the various harmonics that can be induced in the asymmetric field winding with 60 Hz utility supply.

2361

TABLE I. k Freq. (Hz) 1 30 90 2 0 120

HARMONIC COMPONENTS IN FIELD CURRENT 3 30 150 4 60 180 5 90 210 6 120 240 7 150 270 8 180 300 9 210 330 10 240 360

1000 Turns 500 0

V. MODELING OF SYNCHRONOUS MOTOR WITH STATOR INTER-TURN FAULT AND ASYMMETRIC FIELD WINDING In order to carry out the simulation studies, the salient-pole synchronous motor has been modeled including the damper bars. Following are the four different machine models obtained for each of the two-coil and the four-coil field windings, (a) healthy machine with symmetric field winding, (b) healthy machine with asymmetric field winding, (c) faulty machine with symmetric field winding and (d) faulty machine with asymmetric field winding. Thus, totally eight different models were obtained. The asymmetric field winding has been modeled by choosing different number of turns in one of the coils. The models of faulty machine can accommodate different number of shorted turns in the stator winding. The Fourier series expansions of the turns functions of the symmetric two-coil and standard field windings are expressed in (9) and (10) respectively. Fourier series expansion given in (11) represents the general form of asymmetric field winding. The coefficients will be different for the two-coil and standard structures. d and q are the angles subtended by the rotor polar and inter-polar arcs at the center respectively. The turns functions of symmetric and asymmetric field windings of both two-coil and standard structures are shown in Fig. 2 and Fig. 3 respectively.

1000 Turns 500 0

2 3 4 Angle (Mechanical Radian)

Figure 2. Turns functions of the two-coil field winding, symmetric (top) and asymmetric (bottom).

1000 Turns 500 0

nrf 1 ( , ) = aorf 1
a orf 1 = N rf d

+ amrf 1 cos [ pm( )] m = 1,2,3,...

2 N rf

1000 Turns 500 0

(9)

, N rf = 1260 , a mrf 1 =

sin (m d ) (10)

2 3 4 Angle (Mechanical Radian)

nrf 2 ( , ) = a orf 2
N rf 2

+ a qrf 2 cos [ pq( )] q = 1,3,5,...

2 N rf

Figure 3. Turns functions of the standard field winding, symmetric (top) and asymmetric (bottom).

0.8

1 2

a orf 2 =

, N rf = 1260, a qrf 2 =

sin (q d ) (11)
Inductance (H)

0.4

nrfa ( , ) = aorfa +

awrfa cos [w( )] w = 1,2,3,...

After computing various magnetizing and mutual inductances of the machine using winding function approach (WFA) and modified winding function approach (MWFA) [710], the dynamic models of the synchronous motor have been obtained using the coupled circuit approach [9]. The mutual inductances between stator phase-a and the field windings with both of symmetric and asymmetric cases are shown in Fig. 4 (for two-coil structure) and Fig. 5 (for standard structure).

-0.4

-0.8

2 3 4 Angle (Mechanical Radian)

Figure 4. Mutual inductance between the stator phase-a and the two-coil field winding, symmetric (1) and asymmetric (2).

2362

0.8

1 2

and the line current of the experimental motor (with two-coil structure) are obtained as shown in Fig. 9. It is interesting to
-50 60 Hz 30 Hz 180 Hz 120 Hz 300 Hz 240 Hz 90 Hz 150 Hz 270 Hz 330 Hz 210 Hz

0.4 Inductance (H)

PSD (dB)

-100 0 50 100 150 200 250 300 350

-0.4 PSD (dB)

-50

120 Hz 240 Hz 30 Hz 330 Hz 90 Hz 150 Hz 210 Hz 270 Hz 60 Hz 180 Hz 300 Hz

-0.8

2 3 4 Angle (Mechanical Radian)

-100 0 50 100 150 200 Frequency (Hz) 250 300 350

Figure 5. Mutual inductance between the stator phase-a and the standard field winding, symmetric (1) and asymmetric (2).

Figure 6. PSD of field current of the simulated motor with asymmetric twocoil (top) and standard field winding (bottom) with four-turn fault. -50 PSD (dB) -60 -70 -80 HB -50 PSD (dB) -60 -70 -80 -90 HB HU T1 T2 T3 T4 60Hz 120Hz 480Hz 90Hz 150Hz

VI. SIMULATION RESULTS A thorough simulation study was carried out using the eight different models of the machine as mentioned in the earlier section. The results were obtained at full-load, 0.88 lagging power factor and rated field current of the machine. The spectra for motor models with asymmetric two-coil and the four-coil field windings and four-turn fault are shown in Fig. 6. Since the coefficients of the cosine functions in (11) are different for the two structures, the frequency components in the two have dissimilar magnitudes. However no components other than those predicted by Table I show up, proving the theory presented in section IV conclusively. A detailed comparative study of the various frequency components in the field current has also been presented, as shown in Figs. 7 and 8. The condition of the motor is shown on x-axis of the plots. HB and HU denote the healthy machine fed by balanced and unbalanced supplies respectively whereas T1 to T4 represent 1 to 4 turn short in the stator winding. The plots have been normalized with respect to the DC component of the field current. As predicted in the preceding sections, the 120 and 480 Hz components showed up with supply unbalance as well as with turn faults. This was observed in the case of symmetric and asymmetric field windings of both structures. Hence the changes in the even harmonics may not clearly distinguish the faulty and unbalanced supply conditions. In the case of both symmetric and asymmetric two-coil structures of field winding, the 60 Hz components were sensitive to the inter-turn faults as predicted. However, this component did not show any increase in the symmetric standard field winding as predicted and did not increase significantly even in the case of asymmetric winding for the chosen amount of asymmetry. Most importantly, in the case of asymmetric windings of both structures, the 90 and the 150Hz components showed a clear increase only with stator faults as predicted. Also, in the case of healthy machine with both symmetric and asymmetric field windings, these components were absent with supply unbalance as shown in Figs. 7 and 8. The utility supply has many time harmonics, which can also induce some harmonic components in the field current. In order to investigate their influence, first the spectra of the line voltage

HU

T1

T2

T3

T4

Condition of the motor Figure 7. The frequency components in the field current of simulated motor with two-coil structure, top (symmetric) and bottom (asymmetric).
-50 PSD (dB) -60 -70 -80 HB -50 PSD (dB) 60Hz 120Hz 480Hz 90Hz 150Hz

HU

T1

T2

T3

T4

-70

-90 HB

HU

T1

T2

T3

T4

Condition of the motor Figure 8. The frequency components in the field current of simulated motor with standard structure, top (symmetric) and bottom (asymmetric).

2363

0 60Hz 180Hz PSD (dB) -50 120Hz 240Hz 480Hz 300Hz 420Hz

-100 0 0 100 60Hz 30Hz 200 180Hz 240Hz 300 300Hz 400 500

PSD (dB)

-50

120Hz 90Hz 150Hz

420Hz 480Hz 360Hz

-100

100

200 300 Frequency (Hz)

400

500

Figure 9. PSD of line voltage (top) and line current (bottom) of the experimental motor with two-coil field winding. -40 PSD (dB) -60 -80 360Hz 180Hz 240Hz 300Hz

60Hz 120Hz

420Hz 480Hz

-100 0 -40 PSD (dB) -60 -80 100 60Hz 120Hz 180Hz 240Hz 300Hz 200 300 400 360Hz 420Hz 500

the sensitivity of the current sensor is vitally important in detecting the harmonics in question. Otherwise they may get lost in the noise floor. A significant increase in the 120 and 480 Hz was observed in the field current of the healthy machine with supply unbalance (Vab=199.9V, Vbc=208.1V and Vca=206.9V) as shown in Fig. 11. The variation of the 120 and 480 Hz under supply unbalance and turn fault condition of the motor at full-load (0.8 lag) is shown in Fig. 12. An even larger increase of these harmonics under supply unbalance compared to faulty conditions clearly indicates the ambiguity. The 150 Hz components in the excitation current showed the most prominent increase with the faults involving few turns. This was observed at all the three power factors under no-load, half-load and full-load conditions as shown in Fig.13. The supply unbalance seems to have minimal influence on this harmonic. Further, the harmonic analysis of voltage induced in the rotor-mounted search-coil showed an even more encouraging increase of these frequency components with stator inter-turn fault. The rotor search-coil can accept a frequency of any polepair number independent of the asymmetry in the machine. The 90 Hz in the rotor search-coil voltage was found to be a very promising indicator of even one-turn fault at all operating conditions and was least affected by supply unbalance as shown in Fig. 14. The magnitudes of different harmonics have been normalized with respect to the DC and the 30 Hz components of the field current and the rotor search-coil voltage respectively.
-20 PSD (dB) 120Hz 30Hz 60Hz 90Hz 180Hz

480Hz

-40 -60 -80 -100 0 -20 -40 PSD (dB) -60 -80 -100 0 -20 -40 PSD (dB) -60 -80

240Hz 300Hz

360Hz 480Hz

-100 0 100 200 300 Frequency (Hz) 400 500

Figure 10. PSD of field current of the simulated motor with asymmetric twocoil (top) and standard field winding (bottom) with utility voltage data.

50

100

150 120Hz

200

250 240Hz

300

350

400 360Hz

450

500

note that the line current has more harmonics than the voltage, such as 30, 90,150 Hz. etc. It is therefore very clear that these additional harmonics are due to causes internal to the machine. Some of these causes are eccentricity, saturation, structural asymmetry etc. All these frequency components can induce various harmonics in the field current of an actual machine, as can be seen from Figs. 11 and 15 in the experimental section. However, with the kind of asymmetry considered for simulation, only supply voltage harmonics are reflected back in the field current (Fig.10) and not the 90 and 150 Hz components. Fig. 10 has been obtained using actual voltage data collected from utility supply. VII. EXPERIMENTAL RESULTS A motor that has a two-coil field winding was rigorously tested in the laboratory with different fault levels at various loads and power factors. In order to ensure unambiguous detection of the fault, the influence of supply unbalance on the test results was also carefully observed. It was also noticed that

30Hz

60Hz 90Hz 180Hz

300Hz

480Hz

50

100

150 120Hz

200

250

300

350

400 360Hz

450

500

60Hz 30Hz 90Hz

240Hz 300Hz

150Hz 180Hz

480Hz

-100 0 50 100 150 200 250 300 Frequency (Hz) 350 400 450 500

Figure 11. Experimental PSD of field current under full-load, 0.8 lagging pf condition with balanced supply (top), with unbalanced supply (middle) and with 4-turn fault (bottom) for two-coil field winding.

2364

-24 PSD (dB)

120 Hz

-28

-32 HB -60 PSD (dB) -64

HU

T1 480 Hz

T2

T3

T4

-68 HB

HU

T1

T2

T3

T4

Condition of the motor Figure 12. The 120Hz (top) and the 480Hz (bottom) components in the field current under full-load, 0.8 lagging pf condition (Experimental).
-60 PSD (dB) -70 0.8 Lag UPF 0.8 Lead

-80 HB -50 PSD (dB) -60 -70 -80 HB -60 PSD (dB) -70 -80 -90 HB HU T1 T2 T3 T4

brackets below each of the average value. Low values of the standard deviations imply less fluctuation of these harmonics under the given conditions of the motor. Most importantly, the 150 Hz components in the field current and the 90 Hz components in the rotor search-coil stood out prominently. One set of data was also collected for the motor with standard field winding at full load (0.8 lag). The corresponding plots under balanced supply, unbalanced supply and 4-turn fault are shown in Fig. 15. Once again the 90, 150 and 210 Hz components clearly increased with fault and showed no increase under supply unbalance. Thus, pursuit of similar harmonic components to detect stator faults in any salient-pole synchronous machine holds distinct promise. It was observed that the magnitudes of these harmonics seem to be dependent on the operating point of the machine. A probable reason may be explained by looking closely at Table I. For example, the 90Hz component can be caused by both k = 1 & 5. Similarly, the 150Hz component can be caused by both k = 3 & 7. Due to the variations in the load and power factor, the space distribution of these components may change resulting in variation of detection. Other factors such as, saturation, other time harmonics; eccentricity related low frequency components might also cause change in them. Tables similar to Table I can evaluate the influence of other time harmonics. Tables XI and XII show the influence of the 120 and 180 Hz time harmonics (predominant ones in supply voltage, as seen in Fig. 8 (top)), respectively on the field harmonics. A possible way to circumvent this problem is to have a look-up table describing the best harmonic to be selected in the field current or the rotor search coil voltage under a given operating condition.
10 PSD (dB) 0 0.8 Lag UPF 0.8 Lead

HU

T1

T2

T3

T4

-10 -20 HB 10

HU

T1

T2

T3

T4

HU

T1

T2

T3

T4

Condition of the motor Figure 13. The 150Hz component in the field current under no-load (top), half-load (middle) and full-load (bottom), 0.8 lagging pf condition (Experimental).

PSD (dB)

-10 -20 HB 10 HU T1 T2 T3 T4

In order to ensure dependability of these harmonics, five sets of data have been collected under all conditions of the motor, healthy motor with balanced supply (HB), healthy motor with unbalanced supply (HU) and with 1 to 4 turn shorts (T1 to T4). This was repeated under no-load, half-load and full-load conditions at three power factors (0.8 lag, 0.8 lead and unity power factor (UPF)). The average values of the 90, 150 and 210 Hz components in the field current as well as in the rotor search-coil voltage are shown in Table II to Table X. The corresponding sample standard deviation has been shown in

PSD (dB)

-10 -20 HB

HU

T1

T2

T3

T4

Condition of the motor Figure 14. The 90Hz component in the rotor search-coil under no-load (top), half-load (middle) and full-load (bottom), 0.8 lagging pf condition (Experimental).

2365

TABLE II. HB 90 Hz 150 Hz 210 Hz 90 Hz 150 Hz 210 Hz -68.26 (0.76) -76.55 (0.49) -85.18 (1.46) -13.5 (0.09) -23.48 (0.35) -26.94 (0.52)

HARMONICS AT N O-LOAD (0.8 LAGGING PF) CONDITION

TABLE VI. HB 90 Hz 150 Hz 210 Hz 90 Hz 150 Hz 210 Hz -74.30 (0.28) -76.71 (0.61) -90.53 (1.79) -17.39 (0.01) -25.97 (0.05) -26.54 (0.04)

HARMONICS AT H ALF- LOAD (0.8 LEADING PF) CONDITION

HU -68.78 (0.81) -76.78 (1.45) -82.6 (1.29) -13.72 (0.2) -22.86 (0.54) -21.42 (0.31)

T1 T2 Field Current -68.7 -68.71 (0.22) (0.49) -76.12 -72.17 (0.94) (1.36) -84.53 -78.96 (0.83) (2.89) Rotor Search-coil -11.99 -7.46 (0.16) (0.89) -24.74 -15.49 (0.11) (2.09) -26.92 -21.5 (0.18) (1.3)

T3 -67.89 (0.44) -65.59 (0.56) -72.18 (0.29) -2.3 (0.49) -5.33 (0.8) -13.75 (0.58)

T4 -68.31 (0.51) -61.72 (0.28) -68.47 (0.46) 1.46 (0.25) 0.55 (0.35) -8.78 (0.35)

HU -75.26 (1.01) -77.09 (1.04) -82.51 (1.86) -17.54 (0.04) -42.29 (0.59) -21.15 (0.18)

T1 T2 Field Current -74.28 -71.35 (0.81) (0.21) -78.38 -74.34 (1.03) (0.64) -87.53 -82.15 (1.08) (1.91) Rotor Search-coil -15.01 -8.71 (0.19) (0.16) -26.32 -14.81 (0.26) (0.40) -24.72 -19.59 (0.54) (0.45)

T3 -68.79 (0.54) -68.28 (0.36) -76.80 (0.27) -3.42 (0.39) -6.59 (0.44) -13.55 (0.42)

T4 -66.00 (0.10) -63.99 (0.38) -72.89 (0.41) 1.21 (0.10) -0.69 (0.25) -9.00 (0.35)

TABLE III. HB 90 Hz 150 Hz 210 Hz 90 Hz 150 Hz 210 Hz -69.21 (0.86) -79.07 (0.88) -86.14 (2.37) -15.06 (0.05) -19.43 (0.2) -26.77 (0.37)

HARMONICS AT N O-LOAD (0.8 LEADING PF) CONDITION

TABLE VII. HB 90 Hz 150 Hz 210 Hz 90 Hz 150 Hz 210 Hz -75.68 (0.81) -74.94 (0.61) -85.88 (0.97) -20.10 (0.11) -25.85 (0.29) -30.90 (0.44)

HARMONICS AT H ALF - LOAD (UPF) CONDITION

HU -69.98 (0.74) -77.17 (0.71) -87.25 (2.27) -15.58 (0.06) -21 (0.23) -18.96 (0.24)

T1 T2 Field Current -68.66 -69.31 (0.07) (0.36) -82.26 -77.4 (2.26) (0.64) -86.22 -81.35 (2.65) (0.86) Rotor Search-coil -12.17 -6.36 (0.15) (0.14) -22.06 -13.03 (0.33) (0.4) -26.11 -19.74 (0.5) (0.42)

T3 -69.09 (0.28) -70.31 (0.42) -76.16 (0.98) -1.72 (0.06) -4.47 (0.1) -13.19 (0.11)

T4 -70.11 (0.29) -66.93 (0.28) -73.37 (0.59) 1.5 (0.01) 0.62 (0.02) -8.99 (0.08)

HU -76.34 (0.88) -75.54 (0.89) -81.64 (1.25) -18.57 (0.09) -28.87 (0.33) -23.04 (0.21)

T1 T2 Field Current -74.98 -70.28 (0.24) (0.52) -76.41 -70.11 (0.71) (0.25) -81.69 -78.31 (2.14) (1.44) Rotor Search-coil -16.03 -8.42 (0.46) (0.08) -28.91 -13.93 (0.30) (0.15) -25.40 -18.63 (0.16) (0.13)

T3 -65.09 (0.15) -62.53 (0.21) -71.22 (0.51) -2.21 (0.10) -4.77 (0.21) -12.13 (0.36)

T4 -62.51 (0.71) -58.60 (0.56) -67.06 (0.21) 1.85 (0.46) 0.44 (0.55) -7.22 (0.21)

TABLE IV. HB 90 Hz 150 Hz 210 Hz 90 Hz 150 Hz 210 Hz -67.86 (0.49) -79.45 (1.73) -86.53 (1.96) -13.67 (0.05) -21.16 (0.17) -26.18 (0.22)

HARMONICS AT N O-LOAD (UPF) CONDITION

TABLE VIII. T4 HB 90 Hz 150 Hz 210 Hz 90 Hz 150 Hz 210 Hz -70.6 (0.53) -79.67 (2.28) -81.28 (2.12) -10.29 (0.12) -24.45 (0.12) -24.95 (0.3)

HARMONICS AT FULL- LOAD (0.8 LAGGING PF) CONDITION

HU -68.7 (0.33) -75.68 (0.79) -85.34 (2.1) -14.96 (0.07) -21.83 (0.23) -20.02 (0.41)

T1 T2 Field Current -67.95 -68.45 (0.29) (0.18) -79.44 -77.38 (1.26) (0.86) -86.07 -81.64 (1.96) (1.55) Rotor Search-coil -11.81 -6.9 (0.08) (0.49) -22.28 -15.1 (0.19) (1.15) -25.89 -20.98 (0.17) (0.45)

T3 -68.91 (0.39) -71.31 (0.81) -76.59 (1.58) -2.15 (0.31) -5.64 (0.56) -14.23 (0.49)

HU -71.48 (0.79) -81.68 (1.65) -77.43 (1.13) -11.22 (0.07) -27.52 (0.21) -21.58 (0.47)

-70.14 (0.36) -67.17 (0.54) -73.65 (1.46) 1.47 (0.15) 0.21 (0.31) -9.24 (0.47)

T1 T2 Field Current -71.26 -73.83 (0.43) (0.57) -77.19 -71.18 (1.97) (1.14) -78.51 -77.32 (0.75) (1.41) Rotor Search-coil -8.69 -5.01 (0.05) (0.03) -36.2 -17.56 (0.32) (0.15) -21.51 -16.12 (0.15) (0.1)

T3 71.18 (0.58) 65.66 (0.40) 75.78 (0.61) -0.21 (0.08) -7.69 (0.2) -1.67 (0.42)

T4 -67.11 (0.62) -62.63 (0.42) -75.08 (0.69) 2.66 (0.04) -3.23 (0.11) -9.39 (0.23)

TABLE V. HB 90 Hz 150 Hz 210 Hz 90 Hz 150 Hz 210 Hz -80.28 (2.20) -72.46 (1.24) -81.45 (0.83) -19.45 (0.22) -23.96 (0.28) -35.66 (0.55)

HARMONICS AT H ALF - LOAD (0.8 LAGGING PF) CONDITION

TABLE IX. HB 90 Hz 150 Hz 210 Hz 90 Hz 150 Hz 210 Hz -81.05 (1.7) -85.27 (1.51) -89.32 (1.25) -18.1 (0.05) -26.16 (0.27) -25.66 (0.18)

HARMONICS AT FULL- LOAD (0.8 LEADING PF) CONDITION

HU -78.75 (1.31) -72.03 (0.91) -77.62 (1.66) -17.58 (0.03) -24.20 (0.10) -22.93 (0.08)

T1 T2 Field Current -77.35 -69.52 (1.90) (0.80) -75.15 -66.05 (1.07) (0.64) -79.26 -73.86 (2.29) (1.94) Rotor Search-coil -15.58 -8.29 (0.14) (0.51) -32.04 -13.88 (0.25) (0.63) -28.33 -19.76 (0.71) (0.31)

T3 -63.53 (0.22) -58.25 (0.39) -67.14 (0.57) -2.46 (0.17) -4.21 (0.19) -12.01 (0.21)

T4 -59.65 (0.28) -54.55 (0.28) -63.06 (0.29) 2.44 (0.14) 1.05 (0.16) -6.60 (0.10)

HU -82.8 (1.21) -83.98 (1.56) -84.47 (1.44) -18.63 (0.05) -28.7 (0.51) -20.68 (0.19)

T1 T2 Field Current -78.55 -78.19 (0.59) (0.75) -84.45 -83.38 (1.31) (0.94) -89.37 -86.58 (0.93) (0.76) Rotor Search-coil -15.15 -9.58 (0.05) (0.11) -24.35 -17.45 (0.22) (0.11) -23.41 -19.25 (0.18) (0.07)

T3 -75.87 (0.19) -80.06 (1.01) -85.88 (1.03) -3.72 (0.12) -9.68 (0.16) -14.34 (0.24)

T4 -73.88 (0.34) -76.24 (0.66) -83.84 (1.13) 0.37 (0.23) -4.55 (0.59) -10.67 (0.37)

2366

TABLE X. HB 90 Hz 150 Hz 210 Hz 90 Hz 150 Hz 210 Hz -85.16 (2.19) -80.7 (1.01) -85.7 (2.6) -18.63 (0.03) -28.75 (0.04) -27.18 (0.16)

HARMONICS AT FULL- LOAD (UPF) CONDITION

HU -89.15 (0.46) -84.82 (1.37) -83.34 (1.86) -16.98 (0.25) -34.98 (0.67) -22.39 (0.41)

T1 T2 Field Current -84.72 -85.84 (2.35) (1.04) -80.01 -75.67 (1.28) (0.9) -84.31 -81.6 (1.86) (0.81) Rotor Search-coil -15.7 -9.54 (0.16) (0.16) -31.6 -17.83 (0.29) (0.26) -24.72 -18.83 (0.42) (0.46)

T3 -74.83 (0.58) -70.66 (0.47) -78.56 (1.31) -2.99 (0.25) -8.13 (0.36) -12.87 (0.48)

T4 -69.66 (0.45) -66.86 (0.48) -75.26 (0.59) 1.45 (0.12) -2.44 (0.25) -8.73 (0.34)

TABLE XI.
k

HARMONIC COMPONENTS IN FIELD CURRENT DUE TO 120 HZ TIME HARMONIC IN STATOR VOLTAGE AND CURRENT 2 60 180 3 30 210 4 0 240 5 30 270 6 60 300 7 90 330 8 120 360 9 150 390 10 180 420

1 90 150

Freq. (Hz)

TABLE XII. HARMONIC COMPONENTS IN FIELD CURRENT CURRENT DUE TO 180 HZ TIME HARMONIC IN STATOR VOLTAGE AND CURRENT
k

1 150 210

2 120 240

3 90 270

4 60 300

5 30 330

6 0 360

7 30 390

8 60 420 360Hz

9 90 450

10 120 480

Freq. (Hz) -20 PSD (dB)

VIII. CONCLUSIONS New results to detect stator inter-turn faults in a salient pole synchronous motor have been obtained by analyzing the field current of the machine. The scheme is capable of detecting the less severe state of the fault unambiguously, particularly with supply unbalance and time harmonics. The internal asymmetry of the machine that has adversely affected the earlier fault diagnosis results has, for the first time, been shown as an advantage. A detailed theoretical proof for the induction of various frequency components has been furnished. Simulation study was carried out on four different cases involving two types of field windings. Detailed experimental results have been presented showing the influence of load and power factor of the machine on the diagnosis results. The 150 Hz component in the field current can be a very good indicator of fault, and is able to detect up to three-turn fault unambiguously. Further research will be directed towards improving the level of sensitivity in field current harmonics. Analyzing the voltage induced in the rotor search-coil also validates the diagnosis results. The 90 Hz component in the rotor search-coil appears to be an extremely promising indicator of even one-turn fault without any correction factor. The low values of standard deviations indicate less fluctuation of these harmonics at any given condition and hence imply more dependability. The scheme can be very easily implemented to ensure rapid tripping of the faulty machine at a cheaper cost since only one signal has to be analyzed. REFERENCES
G. B. Kliman, W. J. Premerlani, R. A. Koegl, and D. Hoeweler, A new approach to on-line turn fault detection in ac motors, IEEE-IAS Annual Meeting Conference, vol. 1, pp. 687-693, Oct. 1996. [2] J. Penman and H. Jiang, The detection of stator and rotor winding short circuits in synchronous generators by analysing excitation current harmonics, International Conference on Opportunities and Advances in International Electric Power Generation, Conf. Publ. No. 419, pp.137142, March 1996. [3] J. Sottile, F. C. Trutt and A. W. Leedy, Condition monitoring of brushless three-phase synchronous generators with stator windings or rotor circuit deterioration, IEEE-IAS Annual Meeting Conference, vol. 3, pp. 1587-1594, Sept./Oct. 2001. [4] S. Wan, H. Li, Y. Li and Y. Wang, The diagnosis method of generator rotor winding inter-turn short circuit fault based on excitation current harmonics, IEEE-PEDS International Conference, vol. 2, pp. 16691673, Nov. 2003. [5] G. Kron, Equivalent Circuits of Electric Machinery, New York: Wiley, 1951. [6] G. B. Kliman, R. A. Koegl, J. Stein, R. D. Endicott and M. W. Madden, Noninvasive detection of broken bars in operating induction motors, IEEE Trans. Energy Conv,, vol. 3, no.4, pp. 873-879, Dec. 1988. [7] X. Luo, Y. Liao, H.A. Toliyat, A. El-Antably and T.A. Lipo, Multiple coupled circuit modeling of induction machines, IEEE Trans. Industry Applications, vol. 31, no.2, pp. 311-318, March/April 1995. [8] H.A. Toliyat and T.A. Lipo, Transient analysis of cage induction machines under stator, rotor bar and end ring faults, IEEE Trans. On Energy Conversion, vol. 10, no.2, pp. 241-247, June 1995. [9] H.A. Toliyat, M.S. Arefeen and A.G. Parlos, A method for dynamic simulation of air-gap eccentricity in induction machines, IEEE Trans. Industry Applications, vol. 32, no.4, pp. 910-918, July/Aug. 1996. [10] N.A. Al-Nuaim and H.A. Toliyat, A novel method for modeling dynamic air-gap eccentricity in synchronous machines based on modified winding function theory, IEEE Trans. On Energy Conversion, vol. 13, no.2, pp. 156-162, June 1998. [1]

120Hz -40 30Hz 60Hz -60 90Hz 150Hz -80

240Hz 300Hz

480Hz

-100 0 -20 PSD (dB) 100 200 240Hz 300 400 360Hz 480Hz 500

120Hz -40 30Hz 60Hz -60 90Hz 150Hz -80

-100 0 -20 -40 PSD (dB) -60 -80 120Hz 30Hz 60Hz 90Hz 150Hz 240Hz 210Hz 300Hz 360Hz 480Hz 100 200 300 400 500

-100 0 100 200 300 Frequency (Hz) 400 500

Figure 15. Experimental PSD of field current under full-load, 0.8 lagging pf condition with balanced supply (top), with unbalanced supply (middle) and 4turn fault (bottom) for a standard field winding.

2367