Condition monitoring and fault diagnosis of induction motors

Home | Articles | Forum | Glossary | Books

1.1 Introduction

Fig. 1 Induction motor crosscut along its length -- Stator winding Stator core Rotor end ring Bearing Shaft Rotor core Rotor bar

Electrical machines are one of the most growingly manufactured devices which are widely used not only in industries but also in domestic applications. Due to their higher efficiency level, they are common sources of electromechanical energy conversion. Various applications including control, automotive and power generation applications are served by means of these devices, and no one can deny their ability in being precisely controlled in terms of position, speed and torque which are some of the essential quantities in electromechanical devices.

Among all kinds of electrical machines, induction motor has been more widely used since the beginning stage of industrial development. However, permanent magnet-assisted motors are probably going to be a good substitute for induction motors due to their higher power densities and efficiencies. Nevertheless, induction motors are still one of the most reliable machines used in both line-start and inverter fed applications. A very good example of a latest application of this machine is doubly fed induction generator utilized in wind turbines in conjunction with inverters. High-speed induction motor for low- and medium-voltage power-trains is another example. Due to the rapidly increasing demand of energy consumption in the world, induction motors are required to operate in different environments which might apply electrical, magnetic, thermal and mechanical transients and stresses to the motor. Therefore, the motor likely encounters undesirable operating conditions which in turn makes maintenance procedure an inseparable part of any industrial development. Equally important, maintenance could not be precisely done unless a proper monitoring and diagnosis process is devised. Induction motors are of great robustness against the mentioned stressful factors caused by any unpredictable reason. However, situation changes and sever internal or external tensions can probably affect not only the motor structure, material and smooth healthy operation. Stress-causing factors interrupt the smooth and reliable operation of motors, and induction motors, although very reliable, are not an exception. In this situation, the motor becomes out of order if unhealthy conditions are not diagnosed and treated well. This undesirable condition is usually referred to as ''fault,'' and the procedure of diagnosis is called ''fault diagnosis procedure'' which is the outcome of a condition monitoring system. Fault-producing factors are divided into two general categories of external or internal, depending on the motor component they impact on. A general overview of different motor components is shown in Fig 1 in which the main components are highlighted as follows:

--stator core,

--stator winding,

--rotor core,

--rotor bars,

--rotor end ring,

--bearings and


First, let us generally describe different components and take a glance at corresponding possible stressful conditions. Fig 1 shows a squirrel cage induction motor in which rotor consists of the bars which are connected to each other through end rings at two ends. The end rings have the same material as the bars which are usually made of aluminum. The stator windings are made of copper and carry the stator current which flows into the motor and the network connected to the motor terminals. Depending on the number of motor phases, there should be the same number of network phases to supply the motor. On the basis of the governing rules of induction motors, the stator current produces electromotive force across the rotor bars; hence, the rotor bar currents are produced as the result of this phenomenon which is called ''induction.'' This is the reason why these types of machines are called induction motors. The interaction between the stator and rotor magnetic fields produces the electromagnetic torque which is the main reason of the mechanical rotation of the rotor. The stator winding is located inside the stator slots which are surrounded by the stator core made of silicon-steel material. Normally, the rotor and stator cores are of the same material unless a specific application is targeted. The motor shaft is concentric with the rotor core and fixed at the two ends of the motor housing by means of the bearings. There is another type of induction motor in which the rotor consists of windings instead of bars, and this topology is called ''wound-rotor'' induction motor. Any winding inside the motor consists of insulation layers which separate the conductors electrically to prevent short-circuit faults.

Any magnetic, thermal, electrical and mechanical asymmetry in the stator, the rotor, the shaft or the bearing disrupts the smooth operation of the motor. The asymmetry can be inherent or fault dependent. In the former case, which is normally of a mechanical nature, the motor structure is made inherently asymmetrical during manufacturing process. This is somewhat inevitable as the cause is usually unknown or irreducible. Sometimes, the asymmetry in motor quantities is caused by the impure sinusoidal nature of induction motors. For example, it is ideally preferred to have a totally sinusoidal winding or bars distribution; however, it is impossible in practice. Therefore, some sort of undesirable harmonic components is produced. On the other hand, these harmonic components are some of the best tools for monitoring the motor behavior during faulty states; hence, they are not thoroughly undesirable or useless. Another type of inherent asymmetry of induction motors is the presence of eccentric rotor or shaft which is not concentric with the stator center. This is caused by improper manufacturing process due to the fact that assembling apparatus is not ideally accurate.

Unlike the inherent asymmetries or faulty conditions, there are some cases where an additional factor, besides the improper manufacturing process, causes some components to fail to operate correctly. One of the very well-known situations is the insulation failure of the stator or rotor winding. As a result, two or more turns of the stator or rotor windings are short circuited. This is usually caused by aging or wearing of insulation due to thermal or electrical stresses applied to the insulation. The presence of the short-circuit fault is a reason for a highly unbalanced current distribution in the stator and/or the rotor, causing an asymmetrical operation of motor quantities including torque, speed, magnetic flux, current and voltage if an inverter-fed application is studied. This kind of failure necessitates a regular condition monitoring and quality investigation of motor components. The short-circuit fault is a very good example of electrical faults. As amplitude of motor current might dramatically increase depending on the location of the fault, over saturated core and consequent hot spots are generated in motor core material as well. The harmonic distortion is another obvious result of this type of fault. As an initial tool used for the diagnostic purposes, zero sequence current might be used to detect and determine the short-circuit fault and its level. However, the short-circuit fault is still one of the hot topics in the field of condition monitoring of electrical machines. In the wound rotor motors, the rotor can also be subjected to the short circuit fault as there are turns and coils separated from each other by means of insulations. Therefore, both the rotor and the stator should be monitored for addressing possible short-circuit faults in wound-rotor motors.

The other type of unhealthy conditions caused by mechanical stresses is the deficiency of the rotor bars, end rings and bearings. Mechanical stresses are mostly observed in the form of disconnected joints, unaligned centers and corrosion.

The broken bars, the eccentric rotor and the bearing wear are the best examples of the mentioned mechanical defects. As the broken bar fault directly affects the rotor bars current, it is sometimes considered as an electrical type of fault. However, the eccentricity fault is always of a mechanical nature. Any undesirable friction force or aging is the major reason of improper operation of bearings leading to an eccentric rotor if suitable maintenance is not performed. As a result of bearing deficiency, the eccentricity fault might occur. In the case of eccentricity fault, the air gap length becomes asymmetrical around the rotor circumference. Hence, the rotor is subjected to an unbalanced magnetic pull which in turn affects the motor current, torque, speed or any other quantity. If the air gap length varies in time at different positions around the rotor circumference, a dynamic type of the eccentricity fault exists. In fact, the rotor center rotates about the stator center. On the other hand, if the rotor center is fixed at a point close to the stator center and does not change position, it is called the static eccentricity. Both the static and dynamic types produce time- or frequency-domain variations in mechanical, electric and magnetic variables of the motor. One of the very significant variations is that of the motor torque and speed which lead to a highly unsmooth operation. This is not a desirable situation in industrial environment. Besides, it causes the other motor components such as bearings or housing to have defects if higher fault levels exist or lower levels of faults are not detected and treated properly. Therefore, many efforts are made to address the issues associated with monitoring and diagnosis of induction motors and their fault levels and types. Under some major fault levels, typically a static eccentricity of 50% or above, the rotor might rub the stator in high-speed levels due to centrifugal force, and the whole motor operation fails. This can be a catastrophe for industries. Thus, diagnosing faults in their initial levels should be an inseparable part of the diagnosis procedure.

Sometimes, due to inappropriate manufacturing and casting process, joints between bars and end rings are not tight enough to stay connected. In other words, the joint between rotor bars and end rings is broken and no current passes through broken joints. This phenomenon is called broken bars fault and is generally considered as an electrical fault while it can also be assumed a mechanical fault as well as it is initially caused by a mechanical deficiency. Although the broken bar fault is considered as the less probable fault compared to the eccentricity and short-circuit faults, it is of a great interest not only in industries but also in academia. Therefore, many researches have devoted themselves to addressing the pros and cons of this type of fault along with those related to other types of faults. The reason for holding a condition monitoring process is actually the safety and maintenance-related costs which might be unbearable if a fault takes place and is not diagnosed afterward. The motor quantities are considerably affected by the broken bars fault in both the time and frequency domains. What happens is the production or magnification of undesirable backward magnetic fields caused by asymmetrical current distribution in the rotor. It means that additional torque-producing harmonic components are generated.

Many efforts have been carried out since 30 or 40 years ago, to carefully find the main reasons and also practically possible diagnostic procedures to detect the fault. However, none of them could come up with a comprehensive approach which can be referred to in terms of diagnosing the fault in every possible situation. The reason is that the number of influential parameters affecting the motor behavior under fault conditions is very large, and almost none of the researchers could have gathered and organized existing information in a significant way by means of which one can select a proper procedure to diagnose the fault. The following are some of the main influential factors in broken bars fault diagnosis procedure:

--fault level,

--fault location ,

--motor load level,

--motor speed which can be a function of motor load or be separately changed by means of drives and

--supply mode including

* line-start mode

* open-loop speed control mode

* closed-loop speed control mode.

In addition to the mentioned factors, internal adjustments of any control-loop such as the cutoff frequency of regulators, along with the bandwidth of controller of any motor-drive system, are the other significant reasons for writing the present guide in which all the mentioned parameters will be carefully discussed. Moreover, the drives capability in separately controlling the motor speed and torque levels will be investigated. The increasing number of influential factors necessitates the presence of a generalized approach which enables the user to deal with different types of faults. The idea is to discriminate the concepts of detecting, determining, locating and diagnosing through analyzing various supply modes and operating conditions.

The ''detection'' is to find an incipient fault during the motor operation. The ''determination'' is to find the fault severity, and ''fault-location'' deals with finding the fault location. The combination of detection, determination and fault-location detection processes is called ''diagnosis procedure'' which is a very important concept and will be defined in the next sections.

To prepare the readers' minds and make them interested and aware of various aspects of the diagnosis procedure, let us discuss in more detail. Induction motors are electromechanical devices which generally consist of the following inputs and outputs:

--input electric variables (voltage and current) and

--output mechanical variables (torque and speed).

The aforementioned variables are measurable outside the motor by means of electrical and mechanical sensors. Depending on the sensor type and how it is installed to measure motor performance, two concepts of invasive and noninvasive sampling techniques are introduced. Generally, mechanical signals measurement requires invasive types of sensors while electrical signals measurement is usually handled in a noninvasive way. Sometimes, the motor torque is measured by means of dynamometers whose inertia is relatively larger than that of the motor shaft. This is an undesirable situation in which some of potential fault-related harmonic components might be filtered out. Motor torque might not be a good medium for the diagnosis purposes at some points. Thus, proper signal selection depending on the type of available sensors and even the supply mode should be considered. This is another major focus of this guide. It is also possible to measure magnetic quantities such as the motor flux variations. The flux, if measured properly, would be the most promising signal to analyze any kind of fault as both the stator and the rotor deficiencies are reflected directly by this quantity. The flux is measured in both invasive and noninvasive ways by means of flux estimation techniques or search coils. Furthermore, measurements, either offline or online, depend on the sensor type, measured signal and signal processing technique. Of course, online approaches are mostly preferred while there are several offline methods applicable to inverter-fed motors.

As another significant factor, the motor supply mode might widely change the motor behavior, depending on the operating condition. Unlike the line-start operation in which the motor speed is a function of the motor torque, the torque and the speed are controlled independently in inverter-fed motors. On the other hand, the voltage and the current are not smooth sinusoidal signals anymore. Instead, a PWM-type voltage, along with a ripple-included current, is applied to motors in inverter-fed applications. This means that any diagnosis procedure which is based on motor torque, speed, voltage and current is indeed affected by inverters and also control strategies. Sometimes, proper ideas might be taken from conventional diagnosis approaches applied to the line-start mode, but they are not generally applicable to any kind of inverter-fed system.

On the other hand, induction motors generally operate in two modes including the transient and steady state. Analyzing different faults in these two modes of operation requires a very deep knowledge of various processing tools including time, frequency and time-frequency domains. Besides, extracting a proper fault indicator is achievable only if an inclusive behavioral study is conducted. The sections of the guide are organized in a way to provide such a knowledge based on which the final goal which is helping readers to understand and get familiar with fault diagnosis tools is achieved. It is worth noting that no one can claim a single diagnosis approach is applicable to all fault types and conditions; and the main goal of this guide is to address practical diagnosis techniques, conditions and procedures. To achieve this goal, mathematical, simulation-based and experimental analysis of faulty induction motors in different operating conditions is provided by means of the following tools:

--analytical formulation of healthy and faulty induction motors,

--winding function-based modeling and analysis of healthy and faulty motors,

--finite element-based modeling and analysis of healthy and faulty motors and

--experimental implementation and analysis of healthy and faulty motors.

As this guide is organized to discuss fault-related materials, it is expected that potential readers would have the primary knowledge of line-start and inverter-fed induction motors, although the guide includes a useful section addressing the line start and inverter-fed applications and their effects on induction motors behavior.

To deal with this, a general routine is proposed in order to gather required information and knowledge. As the starting point, a proper analysis tool should be provided. This is handled by mathematical, experimental and simulation-based materials included in the Sections 2, 3 and 5. In practice, it is hard or sometimes impossible to measure a signal inside the motor. For example, the air gap flux density requires a highly invasive technique to be measured. In this case, the quantities will be obtained by means of the finite element or winding function package. Electrical, magnetic and mechanical quantities are investigated in different operating conditions including various faults, load and speed levels. During the experimental implementation of faults, conventional and new sensor types are discussed, and the way they are used during the process is explained. In the case of simulation processes, interesting and practical implementing techniques and details are addressed and provided for readers.

Having measured or sampled signals, signal processing techniques including time, frequency and time-frequency domains are explained in Section 7. The fast Fourier, wavelet and the Hilbert-Huang transforms are discussed and the corresponding codes are provided.

A comprehensive database of induction motors operating in the line-start or inverter-fed modes are gathered in the Sections 8-10. The sections include different aspects of a diagnosis procedure as follows:

--online or offline methods,

--line-start motors,

--inverter-fed motors including the open-loop and closed-loop schemes,

--invasive or noninvasive methods,

--different fault types,

--experimental implementation of different faults,

--simulation process of faulty motors,

--time, frequency and time-frequency analysis of faulty motors,

--different fault features,

--different fault severities,

--the impact of different load levels on fault features,

--the impact of speed variation on fault features,

--the impact of drive parameters including proportional and integral gains as well as bandwidth of controller on fault features,

--fault diagnosis in transient and steady-state modes and

--loss characterization of faulty motors.

Top of Page

    NEXT   Guide Index HOME