Eccentricities F aults M agnetic S ignature o f a n I nduction Machine Determined with Flux
Farid Zidat - CEDRAT.
I
n the litterature we can find two approaches to make diagnosis: - Model approach: a specific method for automation engineers. Depending on the mechanism adopted, we can distinguish three branches in this method: monitoring by observers, by analytical redundancy and by parametric estimation. - Signal approach: this approach is based on measurable signals data, such as current, torque, stray flux, noise and vibration, temperature. T he principle of this method is to look for frequencies unique to the healthy or fault operation. Faults in electrical rotating machines can induce other phenomena such as noise and vibrations and possibly other faults like friction between the stator and the rotor or accelerated wear of insulations.
We shall focus on the signal approach. The idea is to extract the magnetic signature of the machine in its healthy and degraded mode, to spot specific frequencies of typical defaults. The contribution of Flux in this approach is two-fold – on the one hand, we can combine several faults and identify their magnetic signatures, and on the other, we can identify the performance of the machine in degraded mode.
In this article we are interested in the static and dynamic eccentricity fault in the induction motor. The quantities we will consider are: torque, leakage flux and magnetic forces measured outside the machine (such as accelerometer sensor). For each quantity, we will compare its signature in degraded mode with the healthy mode. We will devote a short paragraph of this article to demonstrating the new feature of Flux concerning double FFT; this feature is available in the dedicated vibro-acoustic coupling environment.
Figure 1: Mesh information.
In Flux, electromagnetic torque is calculated on the mechanical sets, and there is no need to create sensors for this quantity. Experimentally, to measure magnetic flux, a coil with a large number of turns is used. The measurement in this case is voltage, which is the image of magnetic flux.
In our projects, to simplify the models, we chose to create a sensor which makes the integral on the domain depth of the normal component of the stray flux density on a spatial path. To calculate the voltage from this sensor, we multiply the results by 2000 (turn number) and derivate it later.
Sensors definition» Electromagnetic torque
» Stray magnetic flux
The induction machine studied
The machine studied is a 2-pole induction motor, 3-phase star connected (*), characterized by:- Rated-load power, Pn = 7.5kW.
- Rated source voltage, Unf = 380V (phase to null value) - Rated source frequency, f1n = 50Hz.- Stator armature has 24 slots,- Rotor armature has 20 slots.
- Outer diameter of the stator magnetic core is 212mm.- Inner diameter of the stator is 120mm.
- Outer diameter of the rotor is 119mm; air-gap thickness is 0.5mm.- Inner diameter of the rotor magnetic core is 40mm.- Length of the stator and rotor magnetic core is 125mm.
Figure 2: Stray magnetic
flux sensor.
Flux simulations
The simulations are performed on three cases: the healthy case, static and the dynamic eccentricity case. This machine is simulated in Flux 2D in transient application in rated load. The geometry is done using Flux overlays; the mesh is then automatically defined. Mesh information is provided in Figure 1.
This motor is supplied with nominal current sources (In = 8.48 A rms) in order to speed up the calculations. Nominal rotor velocity is 2904 rpm. The calculations are performed during six electrical periods.
The static and dynamic eccentricity value is 10% (the air gap is 0.5 mm), these two faults are defined in two different Flux projects like indicated in CEDRAT News N°66 June2014
.
* Induction machine Flux tutorial (C:\Cedrat\FluxDocExamples_12.0\Examples2D\Tutorial_Technical\InductionMotor_1)
Experimentally, to measure vibrations we use an accelerometer; in Flux we do not have a dedicated sensor for measuring acceleration. Acceleration can be measured only on the mechanical sets. For our application, we chose to make a magnetic forces sensor. The magnetic pressure is integrated on the domain depth of a spatial path located on the external stator.
» External magnetic forces
Results
Space and temporal components of the radial magnetic flux density on respectively spatial path and point located in the middle of the air gap are given for the three cases: healthy, static and dynamic eccentricity cases.
Figure 4 illustrates the comparison of this quantity in the three cases. The comparison of their frequency spectrum is also given.Space harmonics corresponding to stator and rotor teeth are respectively (H23, H25) and (H19, H21). From Figure 5 we can conclude that eccentricities do not have a big influence on these harmonics. However, we do see substantial variation in magnitude of space harmonics H3 and H9.
(see continued on page 7)
» Air gap magnetic flux density
- 6 -
Figure 5-a: 3D curves of stator teeth magnetic forces.
Figure 4: Airgap flux density.
From the dedicated vibro-acoustic environment, we can make a contour on the stator teeth and calculate magnetic forces acting on them. This feature was already available in Flux 11.2, and an example appears in CEDRAT News - N°66 - June 2014. In the recent Flux 12 version, we can make a 3D curve of these forces versus time and angular position. It is also possible to make double FFT (time & space) of this 3D curve, so we can have a space and temporal harmonics of the stator teeth’s magnetic forces, for example.Figures (5-a) and (5-b) illustrate the 3D curve of stator teeth magnetic forces and their double FFT for the three cases. We can clearly see the influence of the degraded mode on stator teeth magnetic forces.
» Magnetic forces on stator teeth
Figure 5-b: Double FFT of stator teeth magnetic forces.
In Flux post processing, we can directly plot electromagnetic torque on the mechanical sets. In Figure 6, we show the torque on the last period and its time spectrum for the three cases: healthy, static and dynamic eccentricities.
(see continued on page 8)
» Electromagnetic torque
- 7 -
The magnetic forces measured by the sensor for the three cases: Healthy, static and dynamic eccentricity, are shown in figure 8. Temporal FFT is also shown in the same figure for the three cases.The static and dynamic eccentricities have the same influence on the magnetic forces measured. There are three new frequencies (H17 and H19 and H21) that appear with the presence of the
eccentricities. Note that harmonic H23 is greatly modified in the
presence of the eccentricity default.
» Magnetic forces sensor
Figure 6: Electromagnetic torque.
Static eccentricity has a visible influence on electromagnetic torque; we can see the increase in the amplitude of many harmonics, in particular harmonic H19 and H21. The dynamic eccentricity default particularly affects harmonics: H9, H10, H11, H22 and H23.
Stray magnetic flux measured by the sensor for the three cases: healthy, static and dynamic eccentricity, is shown in figure 7. Temporal FFT is also shown in the same figure for the three cases.
» Stray magnetic flux sensor
Figure 8: Magnetic forces sensor.
Conclusion
Thanks to the possibilities offered by mechanical sets in Flux, we can model eccentricities and study their magnetic signature, and of course evaluate machine performances in the presence of the defaults. In this article, the emphasis is on the magnetic signature of the eccentricity defaults on torque, the stray magnetic flux sensor and on the magnetic forces sensor. From this brief study, we note harmonics due to eccentricity defaults that we find in all the measurements; these harmonics are a temporal harmonics H19, H21 and H23. After constructing this induction machine, it is necessary to instrument it with torque or force or leakage flux sensor and monitor these frequencies. If there is a significant variation of these frequencies; then this machine is probably degraded.
We can also conclude that it is better to monitor the eccentricity default by measuring magnetic forces and/or electromagnetic torque, because the influence of this degraded mode is more visible on these quantities.
A new feature has been highlighted in this article, namely 3D curves of magnetic forces versus space & time and its double FFT. This feature may be useful for people who are interested in performing vibro-acoustic studies on their rotating machines.
Figure 7: Electromagnetic torque.
In this case, dynamic eccentricity has a visible influence on the stray magnetic flux measured; we can see the increase in the amplitude of many harmonics, in particular even ones.
- 8 -
Eccentricities F aults M agnetic S ignature o f a n I nduction Machine Determined with Flux
Farid Zidat - CEDRAT.
I
n the litterature we can find two approaches to make diagnosis: - Model approach: a specific method for automation engineers. Depending on the mechanism adopted, we can distinguish three branches in this method: monitoring by observers, by analytical redundancy and by parametric estimation. - Signal approach: this approach is based on measurable signals data, such as current, torque, stray flux, noise and vibration, temperature. T he principle of this method is to look for frequencies unique to the healthy or fault operation. Faults in electrical rotating machines can induce other phenomena such as noise and vibrations and possibly other faults like friction between the stator and the rotor or accelerated wear of insulations.
We shall focus on the signal approach. The idea is to extract the magnetic signature of the machine in its healthy and degraded mode, to spot specific frequencies of typical defaults. The contribution of Flux in this approach is two-fold – on the one hand, we can combine several faults and identify their magnetic signatures, and on the other, we can identify the performance of the machine in degraded mode.
In this article we are interested in the static and dynamic eccentricity fault in the induction motor. The quantities we will consider are: torque, leakage flux and magnetic forces measured outside the machine (such as accelerometer sensor). For each quantity, we will compare its signature in degraded mode with the healthy mode. We will devote a short paragraph of this article to demonstrating the new feature of Flux concerning double FFT; this feature is available in the dedicated vibro-acoustic coupling environment.
Figure 1: Mesh information.
In Flux, electromagnetic torque is calculated on the mechanical sets, and there is no need to create sensors for this quantity. Experimentally, to measure magnetic flux, a coil with a large number of turns is used. The measurement in this case is voltage, which is the image of magnetic flux.
In our projects, to simplify the models, we chose to create a sensor which makes the integral on the domain depth of the normal component of the stray flux density on a spatial path. To calculate the voltage from this sensor, we multiply the results by 2000 (turn number) and derivate it later.
Sensors definition» Electromagnetic torque
» Stray magnetic flux
The induction machine studied
The machine studied is a 2-pole induction motor, 3-phase star connected (*), characterized by:- Rated-load power, Pn = 7.5kW.
- Rated source voltage, Unf = 380V (phase to null value) - Rated source frequency, f1n = 50Hz.- Stator armature has 24 slots,- Rotor armature has 20 slots.
- Outer diameter of the stator magnetic core is 212mm.- Inner diameter of the stator is 120mm.
- Outer diameter of the rotor is 119mm; air-gap thickness is 0.5mm.- Inner diameter of the rotor magnetic core is 40mm.- Length of the stator and rotor magnetic core is 125mm.
Figure 2: Stray magnetic
flux sensor.
Flux simulations
The simulations are performed on three cases: the healthy case, static and the dynamic eccentricity case. This machine is simulated in Flux 2D in transient application in rated load. The geometry is done using Flux overlays; the mesh is then automatically defined. Mesh information is provided in Figure 1.
This motor is supplied with nominal current sources (In = 8.48 A rms) in order to speed up the calculations. Nominal rotor velocity is 2904 rpm. The calculations are performed during six electrical periods.
The static and dynamic eccentricity value is 10% (the air gap is 0.5 mm), these two faults are defined in two different Flux projects like indicated in CEDRAT News N°66 June2014
.
* Induction machine Flux tutorial (C:\Cedrat\FluxDocExamples_12.0\Examples2D\Tutorial_Technical\InductionMotor_1)
Experimentally, to measure vibrations we use an accelerometer; in Flux we do not have a dedicated sensor for measuring acceleration. Acceleration can be measured only on the mechanical sets. For our application, we chose to make a magnetic forces sensor. The magnetic pressure is integrated on the domain depth of a spatial path located on the external stator.
» External magnetic forces
Results
Space and temporal components of the radial magnetic flux density on respectively spatial path and point located in the middle of the air gap are given for the three cases: healthy, static and dynamic eccentricity cases.
Figure 4 illustrates the comparison of this quantity in the three cases. The comparison of their frequency spectrum is also given.Space harmonics corresponding to stator and rotor teeth are respectively (H23, H25) and (H19, H21). From Figure 5 we can conclude that eccentricities do not have a big influence on these harmonics. However, we do see substantial variation in magnitude of space harmonics H3 and H9.
(see continued on page 7)
» Air gap magnetic flux density
- 6 -
Figure 5-a: 3D curves of stator teeth magnetic forces.
Figure 4: Airgap flux density.
From the dedicated vibro-acoustic environment, we can make a contour on the stator teeth and calculate magnetic forces acting on them. This feature was already available in Flux 11.2, and an example appears in CEDRAT News - N°66 - June 2014. In the recent Flux 12 version, we can make a 3D curve of these forces versus time and angular position. It is also possible to make double FFT (time & space) of this 3D curve, so we can have a space and temporal harmonics of the stator teeth’s magnetic forces, for example.Figures (5-a) and (5-b) illustrate the 3D curve of stator teeth magnetic forces and their double FFT for the three cases. We can clearly see the influence of the degraded mode on stator teeth magnetic forces.
» Magnetic forces on stator teeth
Figure 5-b: Double FFT of stator teeth magnetic forces.
In Flux post processing, we can directly plot electromagnetic torque on the mechanical sets. In Figure 6, we show the torque on the last period and its time spectrum for the three cases: healthy, static and dynamic eccentricities.
(see continued on page 8)
» Electromagnetic torque
- 7 -
The magnetic forces measured by the sensor for the three cases: Healthy, static and dynamic eccentricity, are shown in figure 8. Temporal FFT is also shown in the same figure for the three cases.The static and dynamic eccentricities have the same influence on the magnetic forces measured. There are three new frequencies (H17 and H19 and H21) that appear with the presence of the
eccentricities. Note that harmonic H23 is greatly modified in the
presence of the eccentricity default.
» Magnetic forces sensor
Figure 6: Electromagnetic torque.
Static eccentricity has a visible influence on electromagnetic torque; we can see the increase in the amplitude of many harmonics, in particular harmonic H19 and H21. The dynamic eccentricity default particularly affects harmonics: H9, H10, H11, H22 and H23.
Stray magnetic flux measured by the sensor for the three cases: healthy, static and dynamic eccentricity, is shown in figure 7. Temporal FFT is also shown in the same figure for the three cases.
» Stray magnetic flux sensor
Figure 8: Magnetic forces sensor.
Conclusion
Thanks to the possibilities offered by mechanical sets in Flux, we can model eccentricities and study their magnetic signature, and of course evaluate machine performances in the presence of the defaults. In this article, the emphasis is on the magnetic signature of the eccentricity defaults on torque, the stray magnetic flux sensor and on the magnetic forces sensor. From this brief study, we note harmonics due to eccentricity defaults that we find in all the measurements; these harmonics are a temporal harmonics H19, H21 and H23. After constructing this induction machine, it is necessary to instrument it with torque or force or leakage flux sensor and monitor these frequencies. If there is a significant variation of these frequencies; then this machine is probably degraded.
We can also conclude that it is better to monitor the eccentricity default by measuring magnetic forces and/or electromagnetic torque, because the influence of this degraded mode is more visible on these quantities.
A new feature has been highlighted in this article, namely 3D curves of magnetic forces versus space & time and its double FFT. This feature may be useful for people who are interested in performing vibro-acoustic studies on their rotating machines.
Figure 7: Electromagnetic torque.
In this case, dynamic eccentricity has a visible influence on the stray magnetic flux measured; we can see the increase in the amplitude of many harmonics, in particular even ones.
- 8 -