Chapter 4

Inter-turn Fault Detection of Three Phase Induction Motor

4.1 Introduction

The internal faults of Induction Motor (IM) accounts for more than 70% in proportion

of total IM failures. Internal faults include stator as well as rotor faults in an IM. Rotor

faults are related to broken rotor bar/end-ring. They are caused by a combination of

various stresses that act on the rotor and these stresses can be identified as

environmental, electromagnetic, thermal, dynamic, environmental and mechanical.

Therefore these leads to low-frequency torque harmonics, which increases noise and

vibration [186, 187]. Occurrence of inter turn fault and failure of the cooling system

arises due to increase in stator temperature. Two thermal profile indicators with thermal

analysis of the infrared thermography (IRT) images is accomplished in this work [188].

Inter turn fault cause a large circulating fault current in the shorted turns, leading to

localized thermal overloading. This one can cause open-circuit failures (melting of

conductors), and electrical fire. Voltage unbalance produces negative sequence current,

which decreases the motor efficiency and accelerates motor degradation due to

increased thermal/mechanical stresses [189, 190]. In fact, a more precise model of the

machine is necessary for an accurate analysis of the machine behavior in both healthy

and faulty cases [191]. A detailed analysis of short circuit faults requires a precise

model. The models of the induction machine, such as the multi winding model, multi-

turns and the model of Park [192] are not practical to make changes in the electrical

stator and rotor. They represent the electrical behavior of the equivalent induction
Chapter 4: Inter – turn fault detection of three phase induction motor

machine. They do not take into account the electric or magnetic phenomena such as

induced currents, magnetic saturation and the effect of complex geometry.

Inter turn fault cause a large circulating fault current in the shorted turn, leading to

localized thermal overloading. This one can cause open-circuit failures (melting of

conductors), short circuit faults (insulation damage) in the electrical fire [193]. In this

chapter the three phase induction motor analysis is done by Ansys-Maxwell software in

healthy as well as in faulty condition. An experimental set up has been made for the

analysis of current signature of the induction motor in 16 channel CRO, current

signature analysis has been done by the FFT and THD evaluation is also accomplished.

Finally the fault classification task has been performed by ANN and SVM.

4.2. Design of Three Phase Induction Motor Using Ansys Software

Finite Element Analysis (FEM) is a computer based numerical technique for calculating

the parameters of electromagnetic devices. It can be used to calculate the flux density,

flux linkages, inductance, torque; induced emf etc., in the FEM, the large

electromagnetic device is broken down into many small elements. The behavior of an

individual element can be described with a relatively simple set of equations [194]. The

computer can solve this large set of simultaneous equations. From the solution, the

computer extracts the behavior of the individual elements.

The FEM provides detailed information about the machine nonlinear effects (based on

its geometry and material properties). This modeling approach is capable of obtaining

an accurate and complete description of an electrical machine [195, 196]. The magnetic

circuit is modeled by a mesh of small elements. The field values are then assumed to be

a simple function of position within these elements, enabling interpolation of results.

The time required to calculate the field distribution may be very long, depending on the

Electrical Engineering, IIT(BHU) Varanasi Page | 74

Chapter 4: Inter – turn fault detection of three phase induction motor

number of elements considered [197]. A compromise must be reached between using

finer meshes to achieve higher accuracy and the processing resources needed to achieve

reasonable simulation times. The FEM is very flexible, especially for new designs

incorporating new shapes. However long time simulation requirements reduce its

attractiveness for a case when a control algorithm needs to be incorporated [198].

A model of the considered IM (1 hp,3-phase, 440 volt,1400 rpm,36 stator slots,28 rotor

slots) has been constructed using ANSYS software in 3D as shown in Figure 4.1.

Figure 4.1: ANSYS 3-D model of the considered induction motor.

4.2.1 Performance Analysis of Three Phase Induction Motor in Healthy Condition

The motor performace are carried out by simulation results by Ansys software in

healthy condtions. The various performance of induction motor are depicted in

following figures.

Electrical Engineering, IIT(BHU) Varanasi Page | 75

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.2: Stator winding currents in induction motor.

Figure 4.3: Flux linkages in induction motor.

Electrical Engineering, IIT(BHU) Varanasi Page | 76

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.4: Profile of induced voltages in various phases.

Figure 4.5: Speed profile of induction motor.

Electrical Engineering, IIT(BHU) Varanasi Page | 77

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.6: Torque profile of induction motor.

Figure 4.7: Electrical and Mechanical power developed in induction motor.

Electrical Engineering, IIT(BHU) Varanasi Page | 78

Chapter 4: Inter – turn fault detection of three phase induction motor

4.2.2 Performance Analysis of Three Phase Induction Motor in Faulty Condition

The motor performace are carried out by simulation results by Ansys software in faulty

condtions( 5%, 10%, 15% and 20% of total turns of the phase A). The various

performance of induction motor under 5% of total turns shorted in phase A are depicted

in following figures.

Figure 4.8: Stator winding currents in faulty (5% turn short) condition.

Electrical Engineering, IIT(BHU) Varanasi Page | 79

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.9: Flux linkages in faulty condition (5% turns short).

Figure 4.10: Stator currents in faulty condition (5% turns short).

Electrical Engineering, IIT(BHU) Varanasi Page | 80

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.11: Speed profile of in faulty condition (5% turns short).

Figure 4.12: Torque profile of in faulty condition (5% turns short).

Electrical Engineering, IIT(BHU) Varanasi Page | 81

Chapter 4: Inter – turn fault detection of three phase induction motor

4.3 Experimental Set-up for Inter-turn Faults in Three Phase Induction Motor

In order to discriminate the healthy and faulty condition in real time experimental data are

obtained from the induction motor of the specifications as shown in Table 4.1 and

experimental test bed is shown in Figure 4.13. The inter-turn fault severity is considered as

5%, 10%, 15% and 20% of total turn short in phase A from the neutral point.Total number

of conductor in phase A winding is 420 (4 coils/phase). I have taken only one coil having

105 turn in phase A of the stator winding for taking out the tappings (21,42,63,84 turns)

have been short circuited for 5%, 10%, 15% and 20% short circuit condition.

Table 4.1: Parameters of IM for inter turn fault

Variable Value

Rated Power 1 HP

Rated Voltage 440 V

Frequency 50 Hz

Rated Speed 440 rpm

Number of Pole 4

No. of turns per phase 420

Types of winding Progressive

Connection type Star

No. of stator slots 36

No. of rotor slots 28

The signature of the stator current have been taken from the current probe to 16 channel

digital storage oscilloscope and analyzed by FFT. The Total Harmonic Distortion (THD)

have been calculated in normal as well as in faulty condition as shown in figures below.

Electrical Engineering, IIT(BHU) Varanasi Page | 82

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.13: Experimental test bed.

Figure 4.14: Stator currents in healthy condtion.

Electrical Engineering, IIT(BHU) Varanasi Page | 83

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.15: Current signal, FFT and THD of phase A under healthy condition

Figure 4.16: Current signal, FFT and THD of phase B under healthy condition

Electrical Engineering, IIT(BHU) Varanasi Page | 84

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.17: Current signal, FFT and THD of phase C under healthy condition

Figure 4.18: Stator currents under 5% short circuit condition.

Electrical Engineering, IIT(BHU) Varanasi Page | 85

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.19: Current signal, FFT and THD of phase A under 5% faulty condition

Figure 4.20: Current signal, FFT and THD of phase B under 5% faulty condition

Electrical Engineering, IIT(BHU) Varanasi Page | 86

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.21: Current signal, FFT and THD of phase C under 5% faulty condition

Figure 4.22: Stator currents under 10% short circuit condtion.

Electrical Engineering, IIT(BHU) Varanasi Page | 87

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.23: Current signal, FFT and THD of phase A under 10% faulty condition

Figure 4.24: Current signal, FFT and THD of phase B under 10% faulty condition

Electrical Engineering, IIT(BHU) Varanasi Page | 88

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.25: Current signal, FFT and THD of phase C under 10% faulty condition

Figure 4.26: Stator currents under 15% short circuit condition.

Electrical Engineering, IIT(BHU) Varanasi Page | 89

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.27: Current signal, FFT and THD of phase A under 15% faulty condition

Figure 4.28: Current signal, FFT and THD of phase B under 15% faulty condition

Electrical Engineering, IIT(BHU) Varanasi Page | 90

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.29: Current signal, FFT and THD of phase C under 15% faulty condition

Figure 4.30: Stator currents under 20% short circuit condition.

Electrical Engineering, IIT(BHU) Varanasi Page | 91

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.31: Current signal, FFT and THD of phase A under 20% faulty condition

Figure 4.32: Current signal, FFT and THD of phase B under 20% faulty condition

Electrical Engineering, IIT(BHU) Varanasi Page | 92

Chapter 4: Inter – turn fault detection of three phase induction motor

Figure 4.33: Current signal, FFT and THD of phase C under 20% faulty condition

4.4 Brief Explanation of Different Pattern Classification Schemes

4.4.1 ANN as a Pattern Classification Scheme

Artificial neural networks model is inspired from biological learning process of the

human brain and have been developed in form of parallel distributed network. During

the training process of the neural networks are fitted to the data by learning algorithms.

In this work, supervised learning has used 70 % of training data and 30 % data used for

testing. In this work, the traditional back propagation algorithm has been used which is

regarded by the usage of a given out-put that compared to the predicted output and by

adjusted of all parameters according to comparison. The parameters of ANNs such as

weights are usually initialized with random values drawn from a standard normal

distribution. The first step of the ANN training process is that to compute an output for

Electrical Engineering, IIT(BHU) Varanasi Page | 93

Chapter 4: Inter – turn fault detection of three phase induction motor

given inputs and its current weights and output is compared with predicted output. The

second step is to calculate the error and as per algorithm weights are adjusted and

checked error below the threshold value. However, the training time is relatively long

and it is also susceptible to local minimum traps [199, 200].

4.4.2 SVM as a Pattern Classification Scheme

The Support Vector Machine (SVM) in is a new kind of classifier that is motivated by

two concepts. First, transforming data into a high-dimensional space can transform

complex problems (with complex decision surfaces) into simpler problems that can use

linear discriminant functions. Second, SVMs are motivated by the concept of training

and using only those inputs that are near the decision surface since they provide the

most information about the classification. It is a kind of learning machine based on

statistical learning theory. The basic idea of applying SVM to pattern classification can

be stated as follows: first map the input vectors into one features space, possible in

higher space, either linearly or nonlinearly, which is relevant with the kernel function.

Then, within the feature space from the first step, seek an optimized linear division, that

is, construct a hyperplane which separates two classes. It can be extended to multi-class.

SVMs training always seek a global optimized solution and avoid over fitting, so it has

ability to deal with a large number of feature [201-203]. In this work, supervised

learning has used 70 % of training data and 30% data used for testing by using RBF

kernel.

Electrical Engineering, IIT(BHU) Varanasi Page | 94

Chapter 4: Inter – turn fault detection of three phase induction motor

4.4.3 Classification of signals by ANN and SVM

For fault classification the samples of stator currents at different faults as well as

loading conditions has been taken. Total number of features for training was 80 and 20

for testing, 10 samples of the signal after being normalized was taken as input vectors

for proper feature selection.

The feed forward artificial neural network (ANN) with back propagation algorithm has

been adopted for classification of faults. The acceleration factor, learning rate and

momentum is choosen 0.2, 0.1 & 0.8 respectively. After 65 iteraition error curve

converges which indicates proper freezing of the weights. Once the neural network is

trained out of the 80 datasets with 100 datasets available, the efficacy of the algorithm is

tested for the remaining 20 datasets and the classification accuracy is obtained as 87.26

%. Similarly the support vector machine (SVM) is implemented according to Kuhan-

Tucker condition. Here the regulation parameters and the slack variable is choosen as

0.2 and 0.8 respectively. In the higher dimensional feature space, the fault and no fault

situation is clearly discriminated with minimum overlapping class. The classification

accuracy is obtained is 93.21% which is superior than that of ANN.

4.5 Result & Discussion

In the steady state the current signal from the hardware setup is shown in a 16 channel

CRO which indicates the clear phenomenon of steady state scenario further the tapping

at 5%, 10%, 15% and 20% are created and corresponding signal is retrieved subjected

to short at a particular phase thus, the signal retrieved after the short circuit indicates the

fault scenario and the corresponding FFT analysis is done to evaluate the harmonic

content of the signal by means THD. It is important to discriminate the fault with that of

Electrical Engineering, IIT(BHU) Varanasi Page | 95

Chapter 4: Inter – turn fault detection of three phase induction motor

the steady state scenario. Thus the intelligent pattern classification scheme such as

ANN, SVM was considered for subsequent classification of the fault.

4.6 Conclusion

This chapter discusses the classification scheme by means of ANN and SVM to

discriminate fault condition with that of no fault condition. The experimental prototype

has developed in electrical machine lab and subsequent signal was retrieved for

classification objective. The accuracy of the obtained results clearly indicates the

discrimination between the fault and no-fault condition. Further the reliability is an

important aspect for the condition monitoring objective is discussed in subsequent

chapter.

Electrical Engineering, IIT(BHU) Varanasi Page | 96