Investigation of Broken Rotor Bar Faults in Three-Phase Squirrel-Cage Induction Motors

It is well understood that squirrel-cage induction motors are rugged, reliable, cheap, and thus widely used in industrial and manufacturing processes. However, electrical and mechanical faults pose a particular challenge to the industry and end users which often interrupts the productivity and requires maintenance. In literature, rotor faults have been shown to account for a large portion of induction motor failures, sometimes they are the single biggest cause of failure in the field. Rotor bar faults generally arise from repeated operating stresses which can be electrical, mechanical, thermal or environmental by nature. The causes of rotor bar and end-ring breakage include: (a) magnetic stresses caused by electromagnetic forces, (b) thermal stresses due to abnormal operating duty, including overload and unbalance, (c) inadequate casting, fabrication procedures or overloading, (d) contamination and abrasion of rotor because of poor operating conditions, (e) lack of maintenance. Most failures will increase the current and stress in the adjacent bars, progressively deteriorating the rotor part and degrading the motor’s overall performance. Without doubt, it is of great importance to appreciate the mechanisms and characteristic changes of broken bar faults. Furthermore, an online fault diagnostic system is highly desired to meet the reliability requirements, as well as reducing the costs of maintenance and increasing field efficiency.


Introduction
It is well understood that squirrel-cage induction motors are rugged, reliable, cheap, and thus widely used in industrial and manufacturing processes.However, electrical and mechanical faults pose a particular challenge to the industry and end users which often interrupts the productivity and requires maintenance.In literature, rotor faults have been shown to account for a large portion of induction motor failures, sometimes they are the single biggest cause of failure in the field.Rotor bar faults generally arise from repeated operating stresses which can be electrical, mechanical, thermal or environmental by nature.The causes of rotor bar and end-ring breakage include: (a) magnetic stresses caused by electromagnetic forces, (b) thermal stresses due to abnormal operating duty, including overload and unbalance, (c) inadequate casting, fabrication procedures or overloading, (d) contamination and abrasion of rotor because of poor operating conditions, (e) lack of maintenance.Most failures will increase the current and stress in the adjacent bars, progressively deteriorating the rotor part and degrading the motor's overall performance.Without doubt, it is of great importance to appreciate the mechanisms and characteristic changes of broken bar faults.Furthermore, an online fault diagnostic system is highly desired to meet the reliability requirements, as well as reducing the costs of maintenance and increasing field efficiency.

Broken bar faults in squirrel-cage induction motors
During the past twenty years, there have been continuing efforts at studying and diagnosing faults in induction motors and, in particular, substantial research work is devoted to induction motor bar breakages and the development of non-intrusive diagnostic techniques (Elkasabgy et al., 1992;Bellini et al., 2001;Said et al., 2000).Some research work was based on the finite element (FE) techniques (Mohammed et al., 2006;Mirafzal & Demerdash, 2004;ying xie, 2009;Sprooten & Maun, 2009;Bentounsi & Nicolas, 1988;ying xie, 2010), and more information may be retrieved for diagnostic purpose.It is well established to use line currents as an indicative parameter (Kliman et al., 1988) which can provide insight into the basis of a non-invasive condition monitoring system for the early detection.Other research effort has been focused on the motor current signature analysis (Costa, et al., 2004;Walliser & Landy, 1994;Bacha et al., 2004;Thomson & Fenger, 2001) in order to detect electrical and mechanical faults in induction motors.Another issue reported in literature is the temperature-rise-related failure which has also received much research attention.For example, some research has been conducted on totally enclosed fan-cooled (TEFC) induction motors by thermal sensitivity analysis (Mueller et al., 1995;Boglietti et al., 2005;Staton et al., 2005;).In this reference, the thermal design issues were reviewed and optimisation design algorithms were also developed.References (Alberti & Bianchi, 2008) propose a coupled thermal-magnetic analysis of an induction motor with the primary goal of achieving a rapid and accurate prediction of the IM performance.The heating problem of a motor when one of rotor bars is totally broken was simulated in some papers (Casimir et al., 2004); and some papers investigated the heating characteristics and heat distribution of the motor with healthy and broken rotors (Cho et al., 1992;Lopez-Fdez et al., 1999;Antal & Zawilak, 2005).Indeed, there are some technical challenges when analysing electrical motor thermal fields under broken-bar fault conditions although many papers have covered this area of research.In particular, there is little work on the influence of breaking bars on the temperature-rise of electrical motors using quantitative methods.This chapter will bridge the gap.This chapter is also set out to discuss early diagnostic techniques to identify the faults.

Experimental setup
A dedicated experimental test bench has been designed for testing squirrel-cage induction motor faults.One stator and three originally identical rotors have been employed to study the behaviours of the induction motor with or without broken bars.Prior to the start of the testing process, two out of the three rotors are damaged deliberately by drilling holes in the bars on all their depth and used with the same stator to ensure the testing accuracy.The healthy rotor is considered here as a reference.The laboratory test setup and broken bar rotors are shown in Fig. 1.

Search coil and voltage detection techniques
The measured value of air-gap flux density can be obtained by the search coil technology, and it can be applied to motors operating under different loads.The search coil is inserted around the stator tooth tip, as is shown in Fig. 2. The value of air-gap flux density is obtained by analyzing the induced voltage waveform in the coil, which could help to detect the presence of broken bars.
The technique uses a search coil mounted on the internal stator tooth tip and the analysis of the induced voltage waveform to detect the presence of broken bars.The induced voltage in search coil is given by where im B is the amplitude of the fundamental component and the i th harmonic component, i E is the effective value of voltage induced by the fundamental air-gap flux and the i th harmonic air-gap flux, D is the inner diameter of the stator core, s n is the synchronous speed, L is the effective length of the search coil.

Influence of broken bar faults on the magnetic field distribution
The following assumptions have been made in the solution procedure.


Displacement current is neglected because the frequency of the source is very low.


The rotor bars are insulated from the rotor core, and there is no direct electrical contact between the rotor bars and the rotor core.


The leakage on the outer surface of the stator and the inner surface of the rotor is neglected.


The 2-D domain is considered, and the magnetic vector potential and the current density have only the axial z component.
The 2-D model of the motor is employed, and the fundamental equation describing the space and time variations of the vector potential has the following form.
where D is the region of analysis, 1  is the outside circumferential of the stator and inside circumferential of the rotor (the Dirichlet boundary conditions), z J is current density, σ is the conductivity of the conductors, z A is magnetic vector potential, μ is the permeability of the material (Ning Yuquan,2002;Tang Yunqiu,1998;Yan Dengjun et al., 2003;Bangura & Demerdash 1999;Gao Jingde et al., 1993;ying xie, 2009).
The distribution of magnetic field of the motor for the case of no broken bars is symmetrical under the rated load conditions and the locked rotor conditions, while the symmetry of magnetic field distribution is distorted in the case of broken bars.Samples of flux density distributions across the cross-section of the motor are shown in Fig. 3-4.From them, we know that the magnetic saturation around the broken bars is more severe than that of the rated load.In addition to changes in the broken bar regions, the field distributions at other positions in the stator and rotor core are also distorted and increased to some extent, while these are less significant in the case of the rated load.Fig. 8 is the elliptical flux density at positions A, B and C, from it, we can see that the trace of elliptical flux density in case of healthy motor is nearly the same; however it is disorderly and unsystematic when broken bar fault happened.All radii of elliptical flux density vector for broken bars are greater than those for the healthy case, which is due to the local heavy magnetic saturation appearing in the vicinity of the bar breakages.

Operation characteristics of induction motors with broken bar fault
The effect of the broken bar in three-phase cage-rotor induction motors on the motor's operating performances is investigated under both the rated load conditions and the locked rotor conditions.A 2-D Time-Stepping Coupled Finite Element Method (TSCFEM) is employed for predictive characterization of rotor broken bars in induction motors.Simulation results based on detailed theoretical analysis are confirmed by the experimental results.

Stator currents
In the generalized rotating field theory, a backward-rotating field can be produced by the broken rotor bar faults and then lower sideband components in the stator current spectrum at double slip-frequency is introduced.Figs.9-10 show experimental and simulated transient phase currents at rated load.One can notice that the amplitude of stator current fluctuations with time compared to that in the healthy cage.However, while the tests are performed at standstill, the fault-specific sideband components of stator currents do not appear near the fundamental component.Therefore, the stator current for healthy rotor at standstill is very similar to that for faulty rotors.

Rotor-bar currents
When the rotor is rotating, each rotor bar passes every stator slot, so that each bar will be equally influenced by all the stator-driven flux waves, and all the currents of the rotor bars at rated load are sensibly uniform around the rotor periphery.At standstill, however, the current amplitude of the trouble-free rotor varies with position around the rotor periphery and is not equal, which is different from the rated load.The variation of the rotor current in fault at standstill in accordance with it at rated load and the current amplitude at standstill increases more serious, which can be seen from Figure 14.

Magnetic force on the rotor
There have been a variety of methods for calculating local magnetic forces, i.e. the methods based on the virtual work principle, on the Maxwell stresses or on the forces acting on equivalent sources (magnetizing current or magnetic charges).In this section, the method of virtual work is employed in the process of the magnetic force calculation.There is the magnetic force on the rotor bars because of the induced current in the bars.In the twodimensional magnetic field, the magnetic force can be expressed as follows.
where k is unit number, k f is the magnetic force of the unit k , k S is the area of the unit k , J is the induced current density on the rotor bars, l is the length of the bars, k B is the magnetic flux density of the unit k .The magnetic force corresponding to Eq.3 may be simplified, and it becomes

Torques
The torque variation at rated load is given in Fig. 18, and the torque is smooth at no fault, and torque ripple can be observed in faulty conditions.The torque tendency at rotor-locked conditions is different to that at rated load condition, see Fig. 19.The torque waveforms are almost identical.Through further observations, the average torque is reduced at lockedrotor conditions (from 12.28, 11.23 to 10.22 Nm, respectively).It becomes clear that the average torque continues to decrease, impacting on the loading capability of the motor.

Core loss of the motor
The variation of iron core loss at rated load with time is shown in Fig. 20.For motor with healthy bars, the core loss of stator is stable under steady state.When broken bars fault happened, the starting core loss of stator is significantly higher than normal motor, and the core loss is fluctuant with time rather than smooth under steady state.The amplification of this distortion is directly related to the number of broken bars, and this was mainly due to deformation of electromagnetic field deduced by broken bars fault, and the magnetic saturation and higher harmonic component around the broken bars.From it we can note the stator and rotor core losses are fluctuant with time whatever the motor is normal or not, which is different from the rated load conditions, for motor with healthy bars, the core loss of stator and rotor is stable under steady state when the motor is operating in the rated load.When broken bars fault happens, the core losses of stator and rotor are significantly higher than normal motor at standstill, and the fluctuation is more intense.In addition, the rotor core losses can not be ignored at standstill.

Influence of broken bar faults on the thermal field distribution
For TEFC (Totally Enclosed Fan-Cooled) induction motor, the 2-D thermal analysis is well accepted.Then the difficulty in calculating the thermal field is reduced to some extent and the simulation time is beneficially reduced.In terms of the calculation results of electromagnetic field and some empirical formulas, the heat losses can be obtained.The steady temperature distributions of the motor operating at the rated load are calculated shown as Fig. 23.It can be seen that the rotor temperature is highest, and the temperature distribution tendencies of the faulty conditions are similar to that of the motor with healthy rotor.Therefore the broken bar fault has an unobvious influence on the total temperature distribution tendency of the motor.The air-gap temperature distribution along radial is given in Fig. 25.From it the temperature gradient of the air-gap along radial is rather large.The temperature distribution throughout stator slot along radial of the motor cross section is given as Fig. 26.

Conclusions
In this chapter, the application of a Time-Stepping Coupled Finite Element Method for predictive characterization of effects of rotor broken bars has been presented in a comprehensive fashion.The FE analysis has clearly showed that the effect of the broken-bar fault on motor electromagnetic, mechanical performance, and temperature field.Core losses and current profiles of the stator and rotor, the magnetic force and torque in the rotor bar are also affected by the presence of broken bar faults and the motor performance would deteriorate as the number of broken rotor bars increases.Simulation results based on detailed theoretical analysis are validated by the experimental results.
Experimental test and simulation results have illustrated the reason why the broken bar faults are severe and the likelihood of fault propagation to the adjacent bars.From the results in the work, one can appreciate that the broken bar position has a great impact on the motor's operation, especially on the stator current and starting torque.This further confirms the capability of the proposed numerical models which have accounted for the impact of harmonic components of air-gap flux density.Clearly, this research has also highlighted a necessity for advanced online diagnostic techniques to detect the broken bar faults since these are a common and severe type of mechanical faults to break down the induction motors in service.
However, it needs to point out that this chapter has taken use of a 2-D finite element method to analyze the induction motor's electro-magnetic, thermal, mechanical performance, which is proved to be suitable.If more complex problems are involved such as overhang region bar faults, a 3-D finite element method may be required.This is the further work of this research.

Fig. 1 .
Fig. 1.The experimental setup and the three rotors.

Fig. 2 .
Fig. 2. The search coil positions.(a) The sketch map of the search coil technology, (b) measuring position in the experiment Under rated load conditions, the air-gap field is dissymmetrical in the case of broken bars and the harmonic components of air-gap flux density vary significantly.The third harmonic components for the case of a faulty motor are pronouncedly higher and ripple more severely than that of the healthy condition.The appearance of a faulty third harmonic component is clearly an effective method of detecting broken bars, and the test results showed the faulty third harmonic component of air-gap flux density becomes more significant as the number of the broken bars is increased.The results of the first and third harmonic of the air-gap flux density are shown in TableIfor the sample simulation and experiment.(The first harmonics for the case of normal and abnormal conditions from experiment are also fluctuated with time, and the maximal values are listed in the Table1.)(WeiliLi et al., 2007 )

Fig. 6 .Fig. 8 .
Fig. 6.Typical time variation of flux density waveform of x and y component at rated load with and without broken bar fault.(a) positions A, (b) positions B , (c) positions C (ying xie,2009)

Fig. 13 .
Fig. 13.The simulated rotor bar current at rated load.

Fig. 14 .
Fig. 14.The simulated rotor current at standstill: (a) healthy cage, (b) one broken bar, (c) two broken bars tangential component and the normal component of magnetic force, t,k Ba n d n,k B are the tangential component and the normal component of magnetic flux density respectively.Therefore, the magnetic force of the unit k , the magnetic force distribution on the rotor bar at rated load and at standstill is computed, and the position of broken bars is shown in the Fig.15.The magnetic force distributions on the rotor bar at rated load and at standstill are computed by the FE method and the results are shown in the Fig.16-17 for comparison.It can be noticed that the bars with the highest magnetic force are those immediately adjacent to the broken bars, whether the motor operating under rated load conditions or standstill conditions.Consequently, such non-uniform distribution of the force inevitably leads to excessive mechanical stress in the bars, and the bars would become more susceptible to additional wearing and eventual breaking.

Fig. 15 .
Fig. 15.The serial number of the stator tooth and rotor tooth.

Fig. 21 .Fig. 22 .
Fig. 20.Variations of stator core losses versus time before and after broken bars at rated load.(a) healthy motor cage, (b) a one-broken-bar fault, (c) a continuous two-broken-bar (ying xie, 2009) On figure 21-22, we present the variation of stator and rotor iron core loss with time at locked-rotor for healthy and broken bars faulty condition respectively.

Fig. 24 Fig. 24 .
Fig.24are the steady rotor temperature distributions of the motor at the above three states.The rotor temperature distribution of the motor with a healthy rotor is not complete symmetry because of the quasi-stationary-state treatment of the air-gap and the incomplete symmetry of the motor house.But the whole rotor solving region is quite small which is due to the large thermal conductivities of the rotor core and rotor bar.It can be found that the lowest temperatures are in the positions of broken bars in the whole rotor solving region from Fig.24 (b) and (c).It indicates that with the increase of the broken bar number, the temperature-rise at the same position of the motor increases.It can be predicted that the temperature-rises of the stator windings and the rotor will increase dramatically in the case of the motor with serious adjacent broken bars fault.
Fig. 25.The air-gap temperature distribution along radial

Table 1 .
The fundamental and third harmonic component of air gap flux density comparison for simulation and experiment at rated load www.intechopen.comFiniteElement Analysis -From Biomedical Applications to Industrial Developments 480