Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Downloaded from https://iranpaper.

ir
https://www.tarjomano.com https://www.tarjomano.com

Static Eccentricity Fault Detection in Salient and Non-Salient Synchronous


Generators Using Harmonic Components

Seyed Mohamadsaeed Toufighian Jawad Faiz Amirhossein Erfani-Nik


2021 12th Power Electronics, Drive Systems, and Technologies Conference (PEDSTC) | 978-1-6654-0366-5/20/$31.00 ©2021 IEEE | DOI: 10.1109/PEDSTC52094.2021.9405971

toufighian@ut.ac.ir jfaiz@ut.ac.ir amirerfani@ut.ac.ir

Center of Excellence on Applied


Electromagnetic Systems
School of Electrical and Computer
Engineering,
College of Engineering,
University of Tehran
Tehran, Iran

Abstract— In this paper, a non-salient-pole and salient-pole Simultaneous occurrence of both types of fault leads to the
synchronous generator with static eccentricity fault are mixed eccentricity fault [2].
simulated using finite element analysis. Different severities of In synchronous generator analysis, there are three axes
the static eccentricity fault are considered and its effect on flux including stator symmetry axis, rotor symmetry axis and rotor
density, voltage and current is shown and compared with the rotational axis. In a healthy machine, these three axes
healthy generator. It was observed that the air gap magnetic flux
coincide. The minimum length of the air gap in the
pattern in the faulty generator is asymmetrical leading to
harmonics in the flux and voltage signals. The indexes for fault synchronous generator is the prominent pole relative to the
detection are introduced based on the no-load voltage and full stator [2].
load current signals. These indexes are sensitive to the The eccentricity fault causes symmetry between these
occurrence of the fault and can be used to detect the static three axes to be disappeared. Fig. 1 shows the healthy, SE,
eccentricity fault. and DE electric machine.

Keywords—static eccentricity, salient-pole synchronous


generator, non-salient-pole synchronous generator

I. INTRODUCTION
Synchronous generators are mainly used in power plants
to generate electrical power for domestic customers, and
industrial centers. This is the reason that it is one of the most
valuable and important equipment in power systems [1]. Fig. 1. (a) healthy state (b) static eccentricity (c) dynamic
Therefore, the correct operation of this generator in the power eccentricity
systems has a significant impact on the safe operation of
power system, and in return, any damage or interruption in In the SE fault, the rotor symmetry axis coincides with the
the operation has severe consequences. These are economic rotation axis and the stator symmetry axis is shifted relative
and safety consequences. Therefore, maintenance of to these two axes [9][10]. When this happens, the air gap
synchronous generators is essential to prevent interruption of length distribution around the rotor changes linearly with
power generation. Fault in generator reduces its efficiency time.
and leads to its aging [2]. Besides, stopping operation and This paper diagnoses the SE fault in salient-pole and non-
leaving the generator out of generation cause generator salient pole synchronous generators. The changes in the
prevents its progress, increases the life of the equipment and harmonics of voltage and current signals are caused by
loss of electrical power generation [1][3][4][5]. Eccentricity eccentricity fault. In Section II, the harmonic index relations
fault causes asymmetry in the air gap length. If this fault is are extracted. In Section III, general information on the
detected on time, the rotor and stator will not collide and the simulation of non-salient and salient pole synchronous
maintenance cost reduces. These changes in the air gap length generators is provided. Section IV identifies the fault in the
cause asymmetry in the air gap flux, distortion in voltage and synchronous generator with salient pole by means of the
current, vibration and changes in torque [6]. current and voltage, and Section V identifies this fault in the
The reasons for this fault include changes in the generator non-salient pole generator. Finally, Section VI concludes the
shaft position, inaccurate position of the rotor relative to the paper.
stator in the initial construction stages, stator core
displacement, bearings displacement and wearing. II. CALCULATION OF FAULT DIAGNOSIS INDEX
Eccentricity is one of the most common mechanical faults in Eccentricity fault disturbs the flux density symmetry
rotating electric machines that occurs in the rotor [7] leading to the distortion and harmonic changes in voltage and
[8][2][9][10] which is divided into two general static current. The reason is relationships between the magnetic
eccentricity (SE) and dynamic eccentricity (DE) faults.

978-1-6654-0366-5/21/$31.00 ©2021 IEEE

Authorized licensed use limited to: The University of Toronto. Downloaded on February 18,2023 at 06:06:36 UTC from IEEE Xplore. Restrictions apply.
Downloaded from https://iranpaper.ir
https://www.tarjomano.com https://www.tarjomano.com

flux, current and voltage. Interaction between the stator and B r (ϕ , t ) = (α1 + α 2cos 2 p (ϕ − ωr t ) ×
rotor fields leads to the following air gap magnetic flux
density: (1 + δ me cos (ωr t − ϕ ) ) .  µo J mr sin (ωs t − P ϕ ) .d ϕ =
(11)
B ( ϕ , t ) = B s ( ϕ , t ) + B r (ϕ , t ) (1) (α1 + α 2cos 2 p (ϕ − ωr t ) ×
where Br (φ, t) is the rotor magnetic field density and Bs (φ, t) µo J mr
is the stator magnetic field density. There is the following (1 + δ me cos (ωr t − ϕ ) ) (
P
cos (ωs t − P ϕ ))
relationship between the magnetic field, the transmittance Λ Considering ωs=Pωr and some mathematical
and current density J [9]: manipulation, the air gap magnetic flux density is the
B (ϕ,t ) = Λ(ϕ,θr ).  J (ϕ,t ) .d ϕ (2) interaction between the stator and rotor magnetic flux density
where the current density J is as follows: is given below:
µ (J +J )
J (ϕ ,t ) = µ J sin (ωt −Pϕ ) (3) B (ϕ,t ) = Bs (ϕ,t ) + Br (ϕ,t ) = o mr ms (α1 cos (ωst − Pϕ )
o m
P
where the Jm is the maximum current density. As a result,
  1  
the rotor magnetic density is as follows:  cos  ( −1)ωs t − (1− P ) ϕ  + 
α1 δmt   P  
B r (ϕ,t ) = Λ(ϕ,θr ).  µo J mr sin (ωr t − Pϕ ) .d ϕ (4) + +
2   1  
Also, the stator magnetic flux density is as follows: cos  ( +1)ωs t − (1+ P ) ϕ  
  P  
Bs (ϕ,t ) = Λ(ϕ,θr ).  µo J ms sin (ωs t − Pϕ ) .d ϕ (5)
α2 δmt
To model the symmetrical air gap distribution in the non- 2
( cos (ωst −( P ) ϕ) +cos (( 3P )ϕ + 3ωst ))
salient pole synchronous machine, the following
  1   
approximated function can be used [9]:  cos   −1ωs t + ( P −1) ϕ  + 
αδ  P   
g o (ϕ , θ r ) =
1
(6) + 2 mt 
4   
α1 + α 2cos 2 p (ϕ − θ r ) cos  ( 3P −1) ϕ +  − 3ωst  
1
  P   
 
where α1 and α2 are the coefficients depending on the
geometry of the machine, angle φ is a random position on the   1   
 cos   3 + ωs t + ( P +1) ϕ  + 
stator surface, θr is the angular position of the rotor which is αδ  P   
+ 2 mt 
equal to , p is the number of pole pairs and g0 is the 4    
cos  ( 3P +1) ϕ + 1− ωs t  
1
minimum air gap length. For eccentricity fault the minimum   P   
(12)
air distance is as follows [9]:  
g (ϕ ,θ r ) = g (ϕ , ωr t ) = where the harmonics 1 ± , 2 −1 and 3 ±
(1 − δ cos (ωr t − ϕ ) ) (7) are known as the fault detection indexes. If quadratic
g o × (1 − δ me cos (ωr t − ϕ ) ) =
me
expressions are not omitted in (9) and all terms are
α1 + α 2cos 2 p (ϕ − ωr t )
considered, harmonics in the form of 3 ± , 1±
Λ (φ, θr) is the inverse of the air gap and for eccentricity fault
(7) can be rewritten as follows: , 3± and 2 − 1 appear in the air gap magnetic
1 α + α 2cos 2 p (ϕ − ωr t ) flux density when the SE fault occurs. They can be used as
Λ(ϕ , θ r ) = = 1 (8)
g (ϕ , θ r ) (1 − δ me cos (ω r t − ϕ ) ) fault diagnosis indexes.
Therefore, due to aggressive and non-aggressive signals,
By applying the Taylor expansion, (8) can be simplified as using expensive sampling equipment for extracting the flux
follows: is necessary; however, the voltage is readily available. So, the
(ϕ ,θ r ) = (α1 + α 2cos 2 p (ϕ − ωr t ))* × SE fault can be detected using the voltage index. The
((1 + δ me cos (ωr t − ϕ ) ) + δ me cos (ωr t − ϕ ) +
2
(9) relationship between the magnetic flux density and induced
voltage at stator terminals is as follows:
3 4
δ me cos (ωr t − ϕ ) + δ me cos (ωr t − ϕ ) + …
d
V (ϕ ,t ) = (
B (ϕ ,t )×A
dt s
) (13)
Eqn. (9) is a 2nd-order equation which can be finally rewritten
as follows: where A is the effective surface through which the magnetic
flux passes. Considering (12), the induced voltage is
Λ(ϕ ,θ r ) = (α1 + α 2cos 2 p (ϕ − ωr t ) calculated as follows:
(10)
× (1 + δ me cos (ωr t − ϕ ) ) d
V (ϕ , t ) = ( B s (ϕ , t ) × A ) =
dt
Now, the rotor magnetic flux density can be shown as µo A ( J mr + J ms )
follows: ( −α1ωs sin (ωs t − P ϕ )
P
 1  1  
 − ( −1) sin  ( −1)ωs t − (1− P )ϕ  − 
α δ ω  P  
+ 1 mt s 
P
2  1  1  +
 ( +1) sin  ( +1)ωs t −(1+ P )ϕ  
 P  P  

Authorized licensed use limited to: The University of Toronto. Downloaded on February 18,2023 at 06:06:36 UTC from IEEE Xplore. Restrictions apply.
Downloaded from https://iranpaper.ir
https://www.tarjomano.com https://www.tarjomano.com

α 2 δ mt ωs more applicable than the current index for the SE fault


2
( − sin(ωs t −( P )ϕ )−3sin(( 3P )ϕ +3ωs t )) diagnosis.
 1   1    Fig. 5 shows the full-load current in the healthy (green)
  −1  sin   −1 ωs t + ( P −1)ϕ  − 
α 2 δ mt ωs   P   P    and 60% severity SE fault. By examining the current signal,
+  
4
  1 −3  sin  ( 3P −1)ϕ +  1 −3 ω t  
  P   
P 
 s 
  

  1   1   
 −  3+  sin   3+ ωs t + ( P +1)ϕ  − 
α 2 δ mt ωs   P   P   
+   (14)
4
 1− 1  sin  ( 3P +1)ϕ +  1− 1 ω t  
  P   
 P
 s  
   

As seen in (14), the frequency pattern is quite similar to the


magnetic flux density pattern and is repeated in the voltage.
It can be used as a fault detection index.
.
III. STATIC ECCENTRICITY FAULT DIAGNOSIS Fig. 3. Comparison of frequency spectrum of no-load voltage of
salient pole synchronous generator in healthy case (green) and 40%
severity SE fault (purple).
To investigate the SE fault, two non-salient poles
(125 MVA, 3000 rpm, 50 Hz, 2-pole) and salient pole (3MVA,
750 rpm, 50 Hz, 8-pole) synchronous generators are
simulated using Maxwell software. These two healthy
generators have been shown in Fig. 2.
Now, the salient-pole and the non-salient pole
synchronous generators in the no-load and full load modes
are examined and analyzed. The generator with the SE
severities of 10%, 20%, 40%, 30%, 50% and 60% are
considered and compared with the healthy generator. This
analysis is based on the frequency domain and it uses no-load
inducted current and voltage signals at the stator terminals.

Fig. 4. Comparison of frequency spectrum of no-load voltage of


salient-pole synchronous generator in healthy case (green) and SE
fault of 60% (purple).

it is observed that the changes in (2k‒1) fs frequency


pattern is large and the harmonics 3, 9 and 17 have the higher
changes and can be introduced as index for the SE fault
diagnosis. Table 1 shows the changes of the main harmonics
for different severities SE fault.

(a) (b)
Fig. 2. Magnetic flux distribution in healthy (a) salient-pole
generator, (b) non-salient pole generator.

IV. FAULT DIAGNOSIS IN SALIENT POLE SYNCHRONOUS


GENERATOR
Fig.3 and Fig.4 show the salient pole synchronous
generator no-load voltage with the SE severities of 40%, 60%
which have been compared with the healthy generator. A
change in the harmonic components pattern can be observed
in the figure.
It is clear that by increasing the SE fault severity, the
harmonic pattern increases more which indicate the
sensitivity of the introduced index to the SE fault severity
change. Comparison of the no-load and on-load conditions
shows that the no-load voltage is more sensitive than on-load
case. This is the reason that only no-load voltage is used here. Fig. 5. Comparison of frequency spectrum of full-load current of
Synchronous generator current is also used to detect the SE salient pole synchronous generator in healthy mode (green) and
fault only in the on-load case. Therefore, the voltage index is 60% severity SE fault (purple).

Authorized licensed use limited to: The University of Toronto. Downloaded on February 18,2023 at 06:06:36 UTC from IEEE Xplore. Restrictions apply.
Downloaded from https://iranpaper.ir
https://www.tarjomano.com https://www.tarjomano.com

TABLE I
Comparison of current harmonics of salient-pole
synchronous generator at different severities of SE fault

Frequency
Healthy 10% 40% 60%
(Hz)
150 0.11 0.37 1.60 3.60
250 3.03 3.38 3.39 3.62
350 22.01 22.4922.3922.67
450 0.33 0.09 0.19 0.34
550 1.04 0.95 0.88 1.06
650 0.12 0.12 0.11 0.17
750 2.47 2.47 2.51 2.94
850 0.11 0.12 0.34 0.61 Fig. 7. Comparison of frequency spectrum of no-load voltage of
non-salient pole synchronous generator in healthy mode (green)
and 50% severity SE fault (purple)
V. FAULT DIAGNOSIS OF NON-SALIENT POLE SYNCHRONOUS
GENERATOR
The non-salient synchronous generator with the SE fault
and healthy modes are investigated in this section. Similar to
the fault detection in the salient pole generator, here the no-
load voltage and full load current signals are used for 30%
and 50% severities SE fault.
Fig. 6 and Fig. 7 show the no-load voltage of the non-
salient pole synchronous generator in the healthy case and
30% and 50% severities SE fault. Changes are observed in
the fundamental and fractional components of these signals.
Fig. 8 also compares the non-salient synchronous
generator current in the healthy mode (green) and 30%
severity SE (purple). It is observed that in the faulty mode,
the changes at higher frequencies are larger.
Here again, because the changes if the index (2k‒1) fs, the
harmonics of the currents are greater, these harmonics are Fig. 8. Comparison of spectrum of frequency in full-load current of
compared to the healthy case in Table 2 at different fault non-salient pole synchronous generator in healthy mode (green)
severity. and 30% severity SE fault (purple).
It is observed that by increasing the fault severity, the
harmonics are also increased; there are more changes in the TABLE II
3, 9 and 17 harmonics Comparison of current harmonics of non-salient pole synchronous
generator at different severity SE fault

Frequency Healthy 10% 30% 50%


(Hz)
150 0.14 0.33 2.24 3.34
250 22.41 22.42 21.99 22.02
350 27.87 27.86 28.00 29.19
450 0.02 0.04 0.05 0.30
550 4.30 4.32 4.38 4.38
650 0.25 0.26 0.36 0.34
750 0.04 0.04 0.16 0.23
850 0.81 0.84 0.87 0.91

VI. CONCLUSION
Fig. 6. Comparison of frequency spectrum of no-load voltage of This paper introduced some indexes to identify the static
non-salient pole synchronous generator in healthy mode (green) eccentricity faults in non-salient and salient pole synchronous
and 30% severity SE fault (purple). generators. Flux equations were used to extract these indexes.

Authorized licensed use limited to: The University of Toronto. Downloaded on February 18,2023 at 06:06:36 UTC from IEEE Xplore. Restrictions apply.
Downloaded from https://iranpaper.ir
https://www.tarjomano.com https://www.tarjomano.com

Although the flux signals can be used to detect the fault,


capturing these signals is costly and therefore, voltage and
current signals were considered. It was observed that at the no-
load voltage and full-load current, the fundamental and
fractional harmonics are variable compared to the fault
occurred. Also, the 3, 9 and 17 currents harmonics are very
changeable with respect to the fault, and these main harmonics
can be used as the fault diagnosis indexes.
REFERENCES
[1] I. Tabatabaei, J. Faiz, H. Lesani, and M. T. Nabavi-Razavi,
“Modeling and simulation of a salient-pole synchronous
generator with dynamic eccentricity using modified
winding function theory,” IEEE Trans. Magn., vol. 40, no.
3, pp. 1550–1555, 2004.
[2] J. Faiz, B. M. Ebrahimi, M. Valavi, and H. A. Toliyat,
“Mixed eccentricity fault diagnosis in salient-pole
synchronous generator using modified winding function
method,” Prog. Electromagn. Res., vol. 11, pp. 155–172,
2009.
[3] B. M. Ebrahimi, M. Etemadrezaei, and J. Faiz, “Dynamic
eccentricity fault diagnosis in round rotor synchronous
motors,” Energy Convers. Manag., vol. 52, no. 5, pp.
2092–2097, 2011.
[4] M. Babaei, J. Faiz, B. M. Ebrahimi, S. Amini, and J.
Nazarzadeh, “A detailed analytical model of a salient-pole
synchronous generator under dynamic eccentricity fault,”
IEEE Trans. Magn., vol. 47, no. 4, pp. 764–771, 2011.
[5] I. Sadeghi, H. Ehya, and J. Faiz, “Analytic method for
eccentricity fault diagnosis in salient-pole synchronous
generators,” in 2017 International Conference on
Optimization of Electrical and Electronic Equipment
(OPTIM) & 2017 Intl Aegean Conference on Electrical
Machines and Power Electronics (ACEMP), 2017, pp.
261–267.
[6] H. A. Toliyat and N. A. Al-Nuaim, “Simulation and
detection of dynamic air-gap eccentricity in salient-pole
synchronous machines,” IEEE Trans. Ind. Appl., vol. 35,
no. 1, pp. 86–93, 1999.
[7] A. Polat, Y. D. Ertugrul, and L. T. Ergene, “Analysis of
static eccentricity in squirrel cage induction motors,” in 4th
International Conference on Power Engineering, Energy
and Electrical Drives, 2013, pp. 1504–1508.
[8] J. Faiz, M. Babaei, J. Nazarzadeh, B. M. Ebrahimi, and S.
Amini, “Time-stepping finite-element analysis of dynamic
eccentricity fault in a three-phase salient pole synchronous
generator,” Prog. Electromagn. Res., vol. 20, pp. 263–284,
2010.
[9] J. Faiz, M. Babaei, J. Nazarzadeh, B. M. Ebrahimi, and S.
Amini, “Diagnosis and Magnetic Field Analysis of Small
Power Salient-Pole Synchronous Generator with Static
Eccentricity Using Time-Stepping Finite-Element
Method,” Electromagnetics, vol. 31, no. 3, pp. 173–191,
2011.
[10] H. Ehya, I. Sadeghi, and J. Faiz, “Online condition
monitoring of large synchronous generator under
eccentricity fault,” in 2017 12th IEEE Conference on
Industrial Electronics and Applications (ICIEA), 2017, pp.
19–24.

Authorized licensed use limited to: The University of Toronto. Downloaded on February 18,2023 at 06:06:36 UTC from IEEE Xplore. Restrictions apply.

You might also like