anomalert motor monitoring

27
Bently Nevada AnomAlert TM Motor Anomaly Detector

Upload: chrisengdahl

Post on 20-Jan-2015

2.011 views

Category:

Documents


7 download

DESCRIPTION

Model-based motor monitoring from Bently Nevada

TRANSCRIPT

Page 1: AnomAlert Motor Monitoring

Bently Nevada AnomAlertTM

Motor Anomaly Detector

Page 2: AnomAlert Motor Monitoring

2

Motor Failure modes

Typical distribution of motor failure modes.

Motors up to 4Kv

RollerBearingFailures

51%

StatorFailures 25%

Rotor Failures 6%

Others18%

MECHANICAL FAULTS

•Bearings– Contamination– Stress, Load, Fatigue– Vibration– Misalignment– Heat– Lubrication– Electrical discharge

• Rotor– Mass unbalance– Rotor bow– Uneven cooling

• External Misalignment– Foundation crack– Grouting degradation– Wrong thermal offset

ELECTRICAL FAULTS

•Electrical Unbalance– Voltage unbalance

– Rotor bar failure

• Stator Problems:– Loose Iron

– Stator Eccentricity

– Shorted Turns

• Windings– Heat– Inverters– Supply Voltage

problems– Load– Contamination

• Rotor Problems:– Broken/Cracked

Rotor Bars

– Loose Rotor Bars

– Eccentric Rotor

Page 3: AnomAlert Motor Monitoring

3

Condition Monitoring Methodologies•Vibration•Temperature•Motor circuit analysis> current> voltage

•Thermography•Ultrasound•Partial Discharge•Lubrication analysis•Insulation Resistance Testing

Page 4: AnomAlert Motor Monitoring

4

Condition Monitoring Methodologies•Vibration•Temperature•Motor circuit analysis> current> voltage

•Thermography•Ultrasound•Partial Discharge•Lubrication analysis•Insulation Resistance Testing

AnomAlert provides one

technology that detects anomalies

and the cause.

Page 5: AnomAlert Motor Monitoring

5

Technology – Model Based Fault Detection

Measured Current

Voltage ΣΣ

Predicted Current

Diff

+

-

MOTOR

AnomAlert

Compares ACTUAL motor behavior with PREDICTED behavior to detect Anomalies and diagnose type of fault.

Page 6: AnomAlert Motor Monitoring

6

MODEL

Frequency (Hz)

INPUT

OUTPUT

Measured Current

Voltage ΣΣ

Predicted Current

Diff

+

-

MOTOR

AnomAlert

The inputs and outputs of the system are treated as complex dynamic signals

Technology – Model Based Fault Detection

Page 7: AnomAlert Motor Monitoring

7

Frequency (Hz)

Fault Identification

MODEL

Frequency (Hz)

INPUT

OUTPUT

Measured Current

Voltage ΣΣ

Predicted Current

Diff

+

-

MOTOR

AnomAlert

Technology – Model Based Fault Detection

Page 8: AnomAlert Motor Monitoring

8

Three assessments are made:• Inputs (Line voltage analysis)• Outputs (Motor current, Power factor)• Power Spectral Density difference.

Frequency (Hz)

Fault Identification

MODEL

Frequency (Hz)

INPUT

OUTPUT

Measured Current

Voltage ΣΣ

Predicted Current

Diff

+

-

MOTOR

AnomAlert

Technology – Model Based Fault Detection

Page 9: AnomAlert Motor Monitoring

9

Frequency (Hz)

Fault Identification

MODEL

Frequency (Hz)

INPUT

OUTPUT

Measured Current

Voltage ΣΣ

Predicted Current

Diff

+

-

MOTOR

AnomAlert Fault type is identified from frequency content

Technology – Model Based Fault Detection

Page 10: AnomAlert Motor Monitoring

10

Frequency (Hz)

Fault Identification

MODEL

Frequency (Hz)

INPUT

OUTPUT

Measured Current

Voltage ΣΣ

Predicted Current

Diff

+

-

MOTOR

AnomAlert

Extensive motor database is used to set threshold envelope for Current PSD at 8 standard deviations.

Technology – Model Based Fault Detection

Page 11: AnomAlert Motor Monitoring

11

Frequency (Hz)

Fault Identification

MODEL

Frequency (Hz)

INPUT

OUTPUT

Measured Current

Voltage ΣΣ

Predicted Current

Diff

+

-

MOTOR

AnomAlert

Threshold Overlay on PSD Plot

Technology – Model Based Fault Detection

Page 12: AnomAlert Motor Monitoring

12

PSD Analysis

• In motor current spectral analysis, faults which cause dynamic change in air-gap create a frequency modulation to the line frequency. Other faults generate unique frequency symptoms

Page 13: AnomAlert Motor Monitoring

13

PSD Analysis

• AnomAlert uses the residual PSD spectrum for high resolution detection of potential problems.

Line Frequency

Page 14: AnomAlert Motor Monitoring

14

PSD Analysis

•Frequency bands are automatically generated to match known common fault characteristics, with threshold values in each band set by historical (empirical) database.

M1 : Looseness

M2 : Unbalance, Misalignment, Transmission Elements

M3 : Rotor Fault

M5 : Stator Fault

M4 : Unbalance, Misalignment, Transmission Elements

M6 : Bearing Fault

M8 : Bearing Fault

M10 : Bearing Fault

M9 : Other Fault

M7 : Other Fault

M11 : Other Fault

M12 : Other Fault

Line Frequency

Page 15: AnomAlert Motor Monitoring

15

MONITOROK

WATCH LINE ( Supply voltage problem )Temporary changes in supply voltage cause this alarm. If alarm is persistent check ; harmonic levels – capacitor - isolation of cables- motor connector or terminal slackness -contacts of the contactor

WATCH LOAD ( Changes in process is observed )If process is not altered deliberately, check; leakages – valve & vane misadjustments - Pressure gauge fault – Manometer – filters getting dirty (fans, compressors)

Examine 1 (First level alarm)Maintenance should be scheduled. Check imbalance – misalignment – bearing/ bearing housing – motor shaft - broken rotor bar - isolation of stator windings- over lubrication and lubrication leakages through oil belt Driven equipment mechanical problems (gear box, compressor, fan blades, pump seals, conveyor chain tension problem -.....etc

Examine 2 (Second level alarm) After this alarm, maintenance action is required.

MONITOR

Motor status display

Page 16: AnomAlert Motor Monitoring

16

1. Install & Commission

2. Train – 10 days 3. Run

Motor acts as a sensor

AnomAlert automatically build a mathematical models of the motor, which describe the electromechanical behavior of the motor-driven system.

Electrical and Mechanical anomalies automatically detected

How AnomAlert Works

Page 17: AnomAlert Motor Monitoring

17

AnomAlert Clustering Algorithm

Power Factor

Frequency

C1

Motor Operating Curve

Gain (A/V)

•During the learning period AnomAlert treats each operating point of the motor as a cluster in the three dimensional space (powerfactor, gain, supply frequency).

Page 18: AnomAlert Motor Monitoring

18

AnomAlert Clustering Algorithm

Power Factor

Frequency

C1C2

C3

Motor Operating Curve

Gain (A/V)

•During the learning period AnomAlert treats each operating point of the motor as a cluster in the three dimensional space (powerfactor, gain, supply frequency).

Page 19: AnomAlert Motor Monitoring

19

AnomAlert Clustering Algorithm

Power Factor

Frequency

C1C2

C3

C4

Motor Operating Curve

Gain (A/V)

•During the learning period AnomAlert treats each operating point of the motor as a cluster in the three dimensional space (powerfactor, gain, supply frequency).

Page 20: AnomAlert Motor Monitoring

20

AnomAlert Clustering Algorithm

Power Factor

Frequency

C1C2

C3

C4

Motor Operating Curve

Gain (A/V)

• AnomAlert continuously compare real data with the clusters already defined during learning, any value out of the cluster will drive an error event.

Page 21: AnomAlert Motor Monitoring

21

The P-F Interval – Motor Mechanical Failures

Audible noise 1-4 weeks

Heat by touch 1-5 days

P1 P2

P5

P6

F

Lube Analysis 1-6 months

P3

P = Potential FailureFirst indication that a functional failure is occurring, or is about to.

F = Functional FailureThe point at which the asset fails to deliver to it’s intended purpose

Motor portable CM technology 4-8 weeks

P4

P7Con

d itio

n

Time (not linear scale)

P

ProtectionRelays

Vibration 1-9 months

Electrical / Mechanical

Anomaly Modeling. 2 – 3 months

AnomAlert – Motor Anomaly Detection

IR Thermography 6-8 weeks

Page 22: AnomAlert Motor Monitoring

22

Maintenance Planning

Typical 75kW motor uses over US$50,000 in electricity annually, of which up to 5% may be saved by correcting motor defects, unbalance, misalignment, etc..AnomAlert identifies motor and mechanical load anomalies, supporting an efficiency entitlement analysis where power efficiency improvements can be tracked.

Value of AnomAlert ?

Efficiency Optimization

AnomAlert can replace some PM inspection tasks and minimize the need for intrusive inspections, increasing availability.Unplanned downtime is reduced with accurate detection and monitoring of motor anomalies not well addressed with conventional PdM techniques.

Page 23: AnomAlert Motor Monitoring

23

Inaccessible Machines

While large motors are typically already well instrumented, medium and smaller motors are very well matched to the classes of anomalies detected by AnomAlert.

Good-fit Applications for AnomAlert

Motors below 4kV

AnomAlert uses the motor as a “transducer”, responding to anything that causes dynamic changes in the air-gap, including both motor problems and problems with the driven machine.For Submerged pumps and Cryogenic pump applications, which are inaccessible and hostile to instrumentation, AnomAlert is an ideal monitoring solution.

The failure modes typically seen on belt-driven, step-down gearbox or directly coupled medium and smaller motor driven machines are well matched to the detection techniques used by AnomAlert.

Belt-driven machines, and step-down gearboxes

Page 24: AnomAlert Motor Monitoring

24

Medium Voltage – above 500V

Low Voltage – up to 500V Inverter – Low Voltage

AnomAlert Model Types

Inverter – Medium Voltage

Measurement CTs required, but voltage can be a direct connection to the monitor

Measurement CTs and Voltage PT are usually already installed. Connect to the extra secondary winding.

Hall Effect Current sensors need to be fitted. Voltage can be directly connected to the monitor.

Hall Effect Current sensors and Voltage PT need to be fitted.

3 X3 X

Power supplyCT Hall Effect Sensor

3 XCT PT

3 X

PTHall Effect

Sensor

Page 25: AnomAlert Motor Monitoring

25

AnomAlert Connection Diagram

Page 26: AnomAlert Motor Monitoring

26

AnomAlert Architecture

RS485

RS485

Ethernet

Media Converter

Typical arrangement is RS485 multidrop with media converter connection to monitoring software.

Page 27: AnomAlert Motor Monitoring

Conclusion• AnomAlert enables maintenance planning to manage motor faults as well as the driven machine.

• Energy Efficiency entitlement, and change in driven load can be identified and tracked as a key deliverable of this solution

• It is a complement to other CM technologies to monitor the status of the motor using it as a sensor, with no added instrumentation. It can be installed in the MCC or near to it.

•Traffic light display for alert in the field and different levels of notification in the System 1 to provide a quick overview of the motor status.

• All the information can be use in ruledesk to have automatic diagnostic capability as any other system integrated to System1