Introduction to monitoring and diagnosis of electrical machines. 1 st lesson General Principles. Ewen Ritchie Krisztina Leban

Transcription

1 Introduction to monitoring and diagnosis of electrical machines and apparatus 1 st lesson General Principles Ewen Ritchie Krisztina Leban 1

2 1. Introduction to course 2. Introduction to diagnosis 3. Overview of fault incidence id 4. Detection Methods Lesson 1 5. Monitoring Methods 6. Sensors & Transducers 2

3 Introduction to this course Introductory diagnostics objectives, problems and some methods on line monitoring i and diagnosis i of the condition of electrical machinery and apparatus. 3

4 Requirements for this course Previous knowledge steady state modelling dynamic modelling measurements stationary androtating electric machinery and apparatus. 4

5 Purpose of this course Methods to diagnose incipient faults on electricalmachinesduring operation different diagnosticmethods Lb Laboratory models dl 5

6 Overview of this course The course is given in 5 parts: Lesson 1: Introduction & Methods of monitoring Lesson2 : Faults in electrical machines Lesson 3: Faults in power converters Workshop: Mini project Mathematical Approach Workshop Laboratory work 6

7 So What is diagnosis? 7

8 Objectives of Condition Monitoring and Diagnosis IncreasedSystem availability early detection of failures in the system/components Reduced ddirect maintenance and downtime costs Monitoring and reliable diagnosis of incipient faults on electrical apparatus and machines may. Correction of known faults as they become critical Planned downtime Loss of production reduced or avoided On line diagnosis may increase lifetime of the system by avoiding overloads and overheating 8

9 Objectives of Condition Monitoring and Diagnosis Generate new knowledge Enable more reliable automated methods for monitoring and diagnosis Enable staff without expert knowledge to make reliable judgements about machine condition Enable necessary downtime to be planned at an appropriate time 9

10 Methods for Condition Monitoring and Diagnosis Features for measurement could be RMS or instantaneous current, voltage or speed Vibration characteristics Temperature Measurement results may be analysed for example, using FFT, wavelets, dynamic and steady state models Analysis results may be assessed and diagnosed automatically using logic, fuzzy logic, neural networks, etc. 10

11 What are Incipient Faults? 11

12 Incipient Faults One that is at a very early stage For example A crack that has just begun Df Defective insulation allowing charge migration To detect an incipient fault will never be easy The signals will be small aberrations of normal functional signals Needs careful, precise measurements With a wide bandwidth To interpret the results make a diagnosis Advanced signal processing Separate required signals from background signals and noise Advanced numerical techniques to identify the exact cause of the signal 12

13 Introduction to Incipient Faults and Diagnosis Incipient failures may require: Urgent action E.g. disconnection of the faulty component Non urgent action Some part in the system needs repair or replacement in the near future 13

14 Overview of fault incidence Data is available only for large machines (P2>100 kw) [1 04, 1 05] Bearing Itis important to diagnose: Bearing faults Stator winding faults External device faults Other faults are of minor importance Much research has focussed on faults in rotor bars and end rings Stator Winding External device Rotor bars/rings Shaft/ coupling Other 14

15 Overview of fault incidence Two categories: Fault incidencefor rotating machines Electrical faults Stator and armature faults Rotor faults Mechanical faults Eccentricity faults Bearing and gearbox faults Fault incidence for transformers Winding faults Insulation faults Iron core faults 15

16 Fault incidence for rotating machines Stator and armature faults Reason for failure Insulation failure Phase to ground fault Phase to phase fault Turn to turn fault Symptoms Unbalanced air gap voltages and line currents Increased torque pulsation Decreased average torque Increased losses and reduction in efficiency Excessive heating Sudden failure due to fuses or relays tripping 16

17 Fault incidence for rotating machines Broken rotor bar and end ring faults Reason for failure Thermal stresses Magnetic stresses Residual stresses (manufacturing) Dynamic stresses (force related) Environmental stresses Mechanical stresses (fatigue etc.) Symptoms Unbalanced air gap voltages and line currents Increased torque pulsation Reducedaverage torque Increased losses and reduction in efficiency Excessive heating Excessive acoustic noise Excessive vibration 17

18 Fault incidences for rotating machines Eccentricity faults Reason for failure Symptoms Unequal air gap between Unbalanced air gap voltages stator and rotor and line currents Static (fixed min. air gap) Increased torque pulsation Dnamic(rotating Dynamic min. air Decreased average torque gap) Increased losses and Manufacturing fault reduction in efficiency Post manufacturing fault Excessive heating 18

19 Fault incidence for rotating machines Bearing and gearbox faults Reasons for failure Rotor asymmetry faults=eccentricity faults (see previous slide) Outer bearing race defect Rolling element defect Inner bearing race defect Gear train defect Teeth defect Oil contamination Symptoms Rotor asymmetry faults Eccentricity related faults Acoustic noise Excessive vibration 19

20 Bearing gfaults Example of a bearing 20

21 Contact fatigue Occurs on the rolling elements Bearing gfaults Leads to pitting At the top the picture shows a micro crater cross section At the bottom, the picture shows micro craters on the surface 21

22 Bearing gfaults Electric pitting i due to Induced current flowing through the bearings due to a zero sequence component. 22

23 Wear Bearing faults Due to sliding bt between rolling elements and bearing raceway and bearing cage Scuffing High temperature due to friction leads to loss of lubrication, exchange of material and metal adhesion Plastic indentation (brinelling) Excessive loads or impacts on bearing, leads to vibration and noise Corrosion Is a chemical action on different parts Fracture Due to overload, improper fitting practice, bending fatigue or component defect 23

24 Gear faults Contact fatigue Dominant failure mode Polluted or insufficient lubrication Plastic deformation Excessive load acts on the gear Bending fatigue Excessive stress beyond the yield point of the material Results in broken teeth 24

25 Wear Seizing Scuffing Gear faults High temperature due to friction between sliding teeth, leads to plastic flow of material Pitting Contact fatigue stress resulting from repeated application of force. Pitting, a)incipient, b) advanced, c) severe - peeling 25

26 Wear Abrasion Gear faults Scoring of the teeth by foreign matter Scoring Due to sharp edges on the teeth Pitting, a)incipient, b) advanced, c) severe - peeling 26

27 Fault incidence for transformers Winding faults Phase to Phase fault Phase to ground fault Winding to winding fault Turn to turn fault Insulation faults Ageing of oil Ageing of Nomex/Mylar/paper Iron core faults 27

28 Fault incidence for transformers Winding faults Symptoms Reason for failure Winding insulation failure Leads to: Phase to Phase fault Phase to ground fault Winding to winding fault Turn to turn fault Unbalanced voltages and line currents Change in open circuit impedance in the low frequency area weakened magnetic field in the core induced eddy current running in the conductive loop short circuit between windings or winding turns. Sudden disconnection due to fuse or relay 28

29 Fault incidence for transformers Insulation faults Reason for failure Ageing of oil Ageing of Nomex/Mylar/paper Symptoms Change in open circuit impedance in the low and frequency area, due to changein the capacitances in the transformers due to change of the dielectric constant in the system. Partial discharges 29

30 Fault incidence for transformers Causes of iron core faults Thermal stresses Magnetic stresses Dynamic stresses Environmental stresses Mechanical stresses Symptoms Change of total reluctance of the transformer Increased iron and copper losses Overheating of insulation 30

31 Detection methods Many different detection methods can be applied, some of them are: Classical FFT analysis Current signature analysis Current vector analysis Instantaneous power Impedance transfer function Partial ldischarge signals Vibration spectra signals Shaft Condition Monitoring Monitoring negative sequence currents in three phase systems Temperature measurements Infrared recognition Radio frequency emission monitoring Chemical analysis of fluids Acoustic noise measurements Detection using search coils 31

32 Detection methods example Problems, failures and possible on line monitoring techniques for IM drive Ref: El Hachemi Benbouzid, M., A review of induction motors signature analysis as a medium for faults detection, Industrial Electronics, IEEE Transactions on, Volume: 47 Issue: 5, Oct. 2000, 32 Page(s):

33 Current Signature Analysis A non invasive, online or off line monitoring technique Diagnosis of problems in e.g. induction machines turn to turn faults, broken rotor bars, and static and/or dynamic eccentricity. From measured instantaneous values, current spectra are derived by use of FFT analysis Analysing the spectra different faults can be found, different kinds of faults affect different frequencies in the current spectrum Reference : Lucian Mihet Popa, Birgitte Bak Jensen, Ewen Ritchie and Ion Boldea, Current Signature Analysis to Diagnose Incipient Faults in Wind Generator Systems 33

34 Current signature analysis 34

35 Current Vector Analysis Monitoring i the stator tt Vlt Voltage and Current Park s Vector is claimed in the literature to detect eccentricity faults in induction motor, Reference: A.J.M Cardoso, E.S. Saraiva, Computer aided detection of air gap eccentricity in operating three phase induction motors by Park s Vector Approach, IEEE Trans. Ind. Appls., pp , vol. 29, no. 5, Sept./Oct

36 Instantaneous power Reliable interpretation of the current spectra is difficult Distortion of the currentwaveform causedby the abnormalitiesinthe in the drivesystem is usually minute. An alternative is the instantaneous power Theory, simulations, and laboratory experiments, have shown that the instantaneous power carries more information than the current. Instantaneous power should improve the reliability of diagnosis. Almost all the fault harmonics are contained in the frequency band Hz. A great advantage as the fault harmonics domain is well bounded. The power spectra are noisy, so an instantaneous power FFT, at this stage, does not bring important improvement. More oework is required ed to improve poethis method. References: S.F. Legowski, A.H.M. Sadrul Ula and A.M. Trzynadlowski, Instantaneous power as a medium for the signature analysis of induction motors, IEEE Transactions on Industry Applications, Vol. 32, no. 4, July/August 1996, pp Andrzej jm. Trzynadlowski and Ewen Ritchie, Comparative investigation of diagnostic media for induction motors: a case of rotor cage faults, IEEE Transactions on industrial electronics, vol. 47, no. 5, October 2000, pp

37 Impedance transfer function Diagnosis i may be by studying di changes in the impedance transfer function of an electrical machine. The impedance may be given by a parametric transfer function: As a block diagram description either in the Laplace plane orasa discrete transfer function inthe z domain or as a discrete transfer function in the z domain Orit may be given non parametrically: As for instance a frequency spectrum (Bode plot) obtained by FFT analysis or frequency sweep analysis 37

38 Partial Discharge Signals Partial discharge tests can determine if motor and generator stator windings have insulation problems. A deteriorated winding has a PD activity, which can be 30 times higher This great difference enables staff to identify machines needing further maintenance. Advantages of on line partial discharge tests. Machines in good condition get no attention. The effect is lower maintenance costs. After implementing on line partial discharge tests the time between major machine inspections may be increased. Regular monitoring i provides early detection of problems Enables remedial action to prolong winding life Scheduling of rewind operations during routine shutdown periods. Reference: M. Fenger, G.C. Stone and B.A. Lloyd, Experience with continuous partial discharge monitoring of stator windings, Electrical Insulation ConferenceandElectrical Manufacturing&Coil Winding Conference, 2001, pp D.G. Edwards, On line diagnosis of defects in the stator winding insulation structures of high voltage rotating machines, ICEM 92, Manchester, UK (15 17 September, Vol. 3, pp

39 Vibration Spectra Signals Vibration signals may be monitored to detect eccentricity related faults. High frequency vibration components may indicate static or dynamic eccentricity Reference: J.R. Cameron, W.T. Thomson, A.B. Dow, Vibration and current monitoring for detecting air gap eccentricity in large induction motors, IEE Proceedings, pp , vol. 133, pt. B, no. 3, May,

40 Shaft Condition Monitoring A continuous monitoring that uses shaft current and voltage profiles in operating units to determine if there are incipient problems in electrical rotating machinery Shaft Condition Monitoring offers a means to prevent costly parts replacement from shaft current damage, frosting, pitting and spark tracking, and other failures. Shaft Condition Monitoring provides Early Warning which allows maintenance personnel to catch ththe onset of problems and prevent failures, before their effect is known by temperature and vibration sensors. Reference: Marian NEGREA, Fault Diagnostics of Electrical AC Machines, Literature Survey, Helsinki University of Technology,Laboratory of Electromechanics,P.O Box 3000, FIN HUT, Finland 40

41 Monitoring negative sequence currents in three phase systems The theory of symmetrical components provides powerful analysis techniques for simplifying calculation on unbalanced or faulted power systems. Tolyiat and Lipo claim by modeling and experiment that rotor and stator faults of electrical machines result in asymmetryofthemachine the machine impedancescausingcausing the machine to draw unbalanced phase currents. This is the result of negative sequence currents flowing in the line as was also reported by S.Williamson and P.Mirzoian. However, negative sequence current may be caused by voltage unbalance, machine saturation etc. 41

42 Monitoring negative sequence currents in three phase systems Symmetricalcomponentquantities component ofthree phase systems include the positive, negative and zero sequence components. The phase value is the vector sum of its zero, positive and negative sequence components as expressed in (1) for voltage VA. The relationship expressed in (1) is also true for current if I is substituted for V A common example for turn fault detection is the decomposition, and subsequent analysis, of the three, phase currents and voltages into the positive, negativeandzero sequencevaluesand values. V A V V V A0 A1 A2 42

43 Monitoring negative sequence currents in three phase systems Equation (1) may be expanded by substituting the unit vector by a =e +j120 as expressed in (2) in matrix notation. V V A V A 1 2 A1 1 a a VB 3 2 A2 1 a a V C V 2 1 Ref:S.Williamson and P.Mirzoian, Analysis of cage induction motor with stator winding faults, IEEE PES PES, Summer Meeting, July H.A.Toliyat, T.A.Lipo, Transient Analysis of Cage Induction Machines under Stator, Rotor Bar and End Ring Faults, IEEE Trans. On Energy Conv., Vol. 9, No. 4, June (2) 43

44 Temperature measurements Reliable detection and monitoring of hotspots allows action to be taken in the event of overheating. Some electricity distributors require direct measurement of hotspot temperaturesin new designs of transformer. Conventional electrical temperature sensors may be placed at the top and bottom of a transformer but cannot detect hotspots directly. Only fibre optic sensors, being all dielectric dielectric, are permitted. Distributed sensors employing Brillouin scattering, Bragg gratings or Raman effects are proposed for multiple installations but are very expensive. A single optical fibre with an arrayof sensors may be fittedintointo the windings duringmanufactureor refurbishment, allowing quasi distributed temperature measurements to be made accurately. Reference:

45 Infrared recognition Infrared cameras are widely used for commercial and industrial applications such as mechanical and electrical equipment monitoring and predictive maintenance. Infrared thermometers measure the surface temperature of an object from a distance, making them useful in any electrical maintenance operation. May prevent equipment failures andunscheduleddown down time in the following applications. Electric Motors To detect overheated motors, inspect supply connections and circuit breakers (or fuses) for normal temperatures. Motor Bearings Detect hot spots, enabling repairs or replacements to be scheduled before problems lead to equipment failure. Motor Winding Insulation Prolong the life of winding insulation by measuring its temperature. Phase to Phase Measurement Check cables and connectors for equal phase tophase temperatures for induction motors, large computers, and other equipment. Transformers Check thewindingsof air cooled unitsfor hotspots thatindicate winding flaws. Reference: 38&cat_id=4.4 4 Date: 31/

46 Radio frequency emission monitoring Reference: ures/radio_system_90402.pdf Date

47 Chemical analysis of fluids When oil is used for insulation in for instance power transformers an analyze of dissolved gasses is an efficient diagnostic method. Different methods exist some of them are using gas chromatography, others are measuring the thermal conductivity of some of the dissolvedgasses or uses a flame ionisation detector. When other types of insulation material based on cellulosic materials are used, then a furfuraldehyde analysis can be used, since the paper material will release furanoid compounds during ageing. The analysis is done using a high performance liquid chromatography. 47

48 Acoustic Emission Measurements Acoustic emission (AE) is a wellknown technique that is used to detect and locate acoustic sources in for instance power transformers. Advantages of this technique include: Applied on line Non invasive More sensitive than electric methods for on site tests Locate the source(s) in a three dimensional plot Can be used on manufacturer facilities or repair/refurbishment shops to locate a defect when detected by electric methods The performance of this technique is enhanced when used in conjunction with Dissolved Gas Analysis (DGA). AE detects: Partial discharge Arcing Hot spots Loose connections Static electrification in GSU transformers Core clamping problems Reference: TRANSFORMER_BROCHURE.PDF Date:

49 Detection using search coils A method of detecting shorted turns consists of a search coil placed in the airgap A recording of the voltage signal was made when the rotor field winding was excited dfor open circuited i and short circuited i stator winding. Shorted turns may be detected with the generator on load. The method depends on elimination of the distortion factor due to the air gap flux density waveform. Also broken rotor bar faults can be detected by time and frequency domain analysis of induced voltage in search coils placed internally around stator tooth top and yoke and externally on motor frame. References: J.Penman, H.G.Sedding, B.A.Lloyd, W.T.Fink, Detection and location of interturn short circuits in the stator windings of operating motors, IEEE Trans. Energy Conv., Vol. 9, no. 4, Dec. 1994,pp

50 Model based diagnostic methods In model based diagnosis a model of the component/machine to be diagnosed must be set up and analysed. A parametric mathematicalmodel model, an equivalent circuit model, or an artificial intelligence model. Parameters which cannot be measured directly may be calculated and used for fault prediction. In condition monitoring, parameter values may be tracked, and the remaining lifetime of the machine predicted. Model based methods can be useful both for steady state conditions and dynamic operation. 50

51 Methods of monitoring 51

52 A monitoring system; some considerations: Which properties are to be monitored? Structural effects and responses Component damage or ageing What to measure to monitor these properties? Direct measurements of these parameters Indirect measurements of these parameters Type ofsensors Which sensitivity/resolution is necessary 52

53 Measurement System The user of a measurement system should ldbe able to do the following: Define a test procedure Select relevant instruments Be able to complete the test Initialize the different parameters Analyse the test results 53

54 Test Procedure Define a test procedure What is the PURPOSE of the test? Which parameters must be measured? In which order are they to be measured? Are they dependent on each other? Are there several measurements that use the same instruments on the same test? (for instance if parameter variations are to be studied) Is subsequent signal conditioning necessary? Draw the test arrangement and describe the test procedure 54

55 Instruments Choose relevant tinstruments t Which parameters are to be measured? Which accuracy and resolution must the measurements have? How fast should the instruments be? Where should the instruments be placed? Is automatic data collection required? 55

56 Method To be able to complete the test It is important that the test is reproducible, therefore it is important tthat t the previously mentioned information is complete and recorded. (remember to write down the instrument identification numbers) Remember to record any changes to the test procedure or the test arrangement during the test Save all measured values; you could come in doubt later 56

57 Method Initialize different parameters Initialise instruments Reset counters and timers Initialise different parameters for automated measurements Set instruments (measuring ranges, offset, etc.), whether hth they are controlled manually or automatically 57

58 Analyse the test tresults Method Calculatethe the parameters required from the values measured Carry out accuracy calculations Document the test results Perform a critical evaluation of the results Draw a conclusion based on the results 58

59 Test Report A formal test report shouldinclude Title, Author and Date Abstract Contents Introduction with background, limitations i i and objective Instruments and test procedures used Circuit diagrams and algorithms Results Discussion of the results Conclusion References Appendices 59

60 Opportunity Which properties may be Indicators of the condition monitored for electrical machines? Power P [W] Reactive Power Q [VAR] Current I [A] Stator, rotor and/or magnetization flux linkages ψs ψr ψm [Wb] Electromagnetic Torque Te [Nm] Impedance Z [Ω) Voltage U [V] Frequency f [Hz] 60

61 Opportunity What Parameters could be measured? Stator and rotor currents Stator voltages Shaft speed Mechanical torque Stator and rotor temperature Vibration It must be decided what to measure, and how. Measured results must be scaled, may be converted A/D and stored. The data will first be processed to synthesise suitable indicators such as a power, reactive power, flux, impedance etc. 61

62 Sensors/transducers Definition of a transducer 1. A transducer is a device which transforms nonelectrical energy into electrical energy or vice versa Chemical Magnetic Optical Electrical Thermal Mechanical 62

63 Sensors/transducers Definition of a transducer 2. A transducer is a device which transforms energy from one domain into another. Typical energy domains are mechanical, electrical, chemical, fluid, and thermal Chemical Magnetic Optical Thermal Mechanical Electrical 63

64 Sensors/transducers Definition of a transducer 3. A transducer is a device which transforms energy from one type to another, even if both energy types are in the same domain. Chemical Magnetic Optical Thermal Mechanical Electrical 64

65 Use of transducer Sensors/transducers Measurand Sensor/ Transducer Signal Conditioning i Recorder/ Display/ Processor Example of transducer Transducer Sensing Element Amplifier Amplifier Noise Filters + Transducer Output Power Supply 65

66 Sensors/transducers Characteristics ti of sensors Measurement range Minimum and maximum values for a sensor s input and output give the measuring range Span Are the variations for the input and the output Input span= I max I min Output span = O max O min Linearityit A sensor is linear if values for input and output are placed on a straight line 66

67 Characteristics of sensors Linearity continued Sensors/transducers The ideal straight line connects the minimum point (I min,o min ) and the maximum point (I max,o max ) and satisfies the equation: O O min O O max min I max I min The ideal output will then be: O ideal K I a ( I I ) min O max With: K a O O max O min I I max K min min I min O min a K I min I max 67

68 Characteristics of sensors Hysteresis Defined as the difference in the output between measured values for increasing and decreasing values Temperature response There are two possibilities for errors One which offsets the zero value One which changes the span or measuring range For small differences both can be regarded as linearity errors Repeatability and reproducibility The possibility to be able to do the same measurement without differences arising ii from the first measurement to the next, under the same conditions. Normally given as a fraction of the full scale range 68

69 Signal conditioning A variety of signal processing strategies are applicable to time varying data. Some simple ones are : Signal amplification Attenuation Isolation Filtering Multiplexing Linearization Sensor conditioning Offsetting Noise reduction Some more sophisticated ones are: Spectrum analysis The Wigner Ville Distribution Wavelet Decomposition 69

70 Data collection When collecting data the following must be taken into account: The bandwidth of the system must be sufficient to ensure that the dynamicsofthesystem system can befollowed The resolution must be adequate to achieve the required accuracy There should be enough channels to measure all the required signals Simultaneous sampling is preferred, especially if the measured values will be used to calculate other parameters, for instance when an impedance is to be calculated from a measured current and voltage signal 70

71 Data storage For datastorage the following should be considered: Must all the measured values be stored, or only average or limit values Shouldit be possibleto time track the data afterwards, off line for instance for comparison with other measurements made simultaneously at a different site How large a memory is needed? Is it allowable to overwrite data in the memory? 71

72 Example of monitoring system A basic stator current monitoring system configuration. Reference: El Hachemi Benbouzid, M., A review of induction motors signature analysis as a medium for faults detection, Industrial Electronics, IEEE Transactions on, Volume: 47 Issue: 5, Oct. 2000, Page(s):

73 Summary In this lesson 1 of the course the following were discussed: Why and how a diagnostic system can be set up, including a discussion of how to measure and what should be taken into account when setting up a measuring system Different faults in electrical machines, how they arise andhow theycan bedetected Different fault detection methods were introduced 73