Wind turbine induction generator bearing fault detection using stator current analysis

Wind turbine induction generator bearing fault detection using

stator current analysis

By

School of Electrical and Electronic Engineering The University of Manchester

D.S. Vilchis-Rodriguez, S. Djurovic, A.C. Smith

Content

1. Wind generator failure figures

2. Ball bearing frequencies

3. Mathematical model

4. Simulation results

5. Experimental results

6. Fault detection improvement

7. Conclusions

Wind turbine reliability

Feng Y. and Tavner P., “Introduction to Wind Turbines and their Reliability & Availability”, Warsaw, EWEC 2010, 2010.

Wind generator failure occurrence

1-2 MW >2 MW

Alewine K. and Chen W., “Wind Turbine Generator Failure Modes Analysis and Occurrence”, Windpower 2010, Dallas, Texas, May 24-26, 2010.

Rolling bearing race frequencies

1 cos2b b

o rc

N Df f

D

Outer race Inner race

1 cos2b b

i rc

N Df f

D

Bearing fault mechanical effects

Shaft displacement Rolling element drop

Air-gap modulation

Air-gap variations Periodic eccentricity

IG modelling for condition monitoring purposes

• Based on coupled-circuit approach

• Localized bearing faults are modelled as temporary eccentricity variations

• Axial asymmetry is taken into account in the model by averaging both machine ends eccentricity

• This approach makes it possible to analyze with detail incipient bearing faults

Bearing fault simulation results

0 50 100 150 200 250 300 350 400 450 500

10-6

10-4

10-2

100

stator current frequency spectrum

I s [n

orm

alize

d]

Frequency [Hz]

healthy1 mm2 mm3 mm4 mm5 mm6 mm7 mm

fo- f

s

fundamental

fs+ f

o

2fo- f

s 3fo - f

s

fs + 2fo

slot harmonicslot harmonic

fs + 3f

o

4fo - f

s5f

o - f

s

fs + 4f

o

6fo - f

s

fs + 5f

o

127 128 129 130 131 132 133 134 135 136 137

10-5

stator current frequency spectrum

I s [no

rm

alized

]

Frequency [Hz]

healthy1 mm2 mm3 mm4 mm5 mm6 mm7 mm

fs+f

o

Stator current frequency spectrum

Principal bearing fault frequency

detail

f s of f kf

1,2,3...k

Test rig layout

Laboratory test bed(viewed from above)

Load side bearing

Test rig description

Artificial bearing fault Test rig bearing data

Drive-end Non-drive-end

SKF 6313 SKF 6214

Nb = 8 Nb = 10

fo=3.07fr fo=4.11fr

fi=4.93fr fi=5.89fr

Bearing faultMeasured Frequency spectrum

Vibration spectrumStator line current spectrum

110 115 120 125 130 1350

0.5

1

1.5

2

2.5x 10-3 Stator current frequency spectrum, 1600 rpm

Frequency [Hz]

I s [n

orm

aliz

ed

]

healthyfaulthy

2fo-f

sfo+f

s

80 90 100 110 120 130 140 150 160 170

10-4

10-3

10-2

10-1

100

Frequency [Hz]

Acc

ele

ratio

n [

m/s

2]

Vibration frequency spectrum 1600 rpm

healthyfaulthy

2fo

fo

Instantaneous complex current signal

0 0.02 0.04 0.06 0.08 0.1-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

Time [s]

Ma

gn

itud

e [

no

rma

lize

d]

Intantaneous complex signal

is(t)

is(t-t)

magnitude (envelope)

Stator current and current envelope frequency spectrums

127 128 129 130 131 132 133 134 135 1360

0.5

1

1.5

2x 10

-3 Is normalized frequency spectrum

Frequency [Hz]

I s [nor

mal

ized

]

77 78 79 80 81 82 83 84 85 860

0.005

0.01

0.015

0.02

0.025Complex envelope spectrum

Frequency [Hz]

mag

nitu

de [n

orm

aliz

ed]

faulthyhealthy

faulthyhealthy

fo

fs+f

oStator current

spectrum

Complex signalmagnitudespectrum

Complex signal magnitude frequency spectrumper phase

Stator currentsComplex signal magnitude spectrum

81 82 83 84 85 860

2

4

6

8

10

12

14

16

x 10-3 Complex envelope frequency spectrum, 1630 rpm

Frequency [Hz]

ma

gn

itud

e [

no

rma

lize

d]

Ias

Ibs

Ics

fo

Instantaneous negative sequence magnitude

Instantaneous symmetrical components

02

2

1 1 111

31

a

b

c

I t i t

I t a a i t

I t a a i t

02

2

1 1 111

31

a a

b b

c c

I t i t ji t t

I t a a i t ji t t

I t a a i t ji t t

Real valued instantaneous symmetrical components

Complex valued instantaneous symmetrical components

2

3j

a e

Complex signals frequency spectrum

0

0.02

0.04

0.06

mag

nitu

de

Frequency spectrum, 1630 rpm

0

1

2

3

4x 10

-3

mag

nitu

de

79 80 81 82 83 84 85 86 87 880

0.005

0.010

0

0.005

Frequency [Hz]

mag

nitu

de

fo

fo

fo

a) Current envelope spectrum average

b) Complex valued Instantaneous

negative sequence spectrum

c) Real valued Instantaneous

negative sequence spectrum

Fault severity analysis

Artificial bearing fault Fault frequency amplitude variation

0 1 2 3 4 5 6 71

1.5

2

2.5

3

3.5

4

4.5

5x 10-3

Defect width [mm]

f o a

mp

litu

de

[n

orm

aliz

ed

]

Fault frequency amplitude versus fault severity

Conclusions

• An IG analytical model was developed and a commercial machine test rig was used to verify the findings

• Research shows that there are frequency components in IG steady state stator current that are directly related to existence of bearing fault.

• Simulation and experimental data indicate that conventional CSA is not well suited for bearing fault detection.

• The use of complex signals is shown to considerably improve the fault detection using stator current analysis.

Thank You