Reliability & Availability f Wi d T biof Wind Turbine

Electrical & Electronic Componentsp

Peter Tavner, Professor of New & Renewable EnergyEnergy GroupHead of School of Engineering, Durham University

“One has to consider causes rather than symptoms of undesirable events and avoid uncritical attitudes.” Prof Dr Alessandro BiroliniProf Dr Alessandro Birolini

1 of 25

Summary of Reliawindy•EU funded FP7 R&D project•€7.7M total funding

•10 organisations involved•3 years

•€5.5M EU funding

Aims:•Improve general understanding of wind turbine and farm reliability•Develop reliability models specific to wind turbines

•Increase MTBF •Increase Availability•Decrease MTTR •Decrease Cost of Energy

Important Onshore Critical Offshore

2 of 25

Summary of ReliawindyPartners:Partners:1.1. GamesaGamesa Wind Turbine1.1. GamesaGamesa2.2. EcotecniaEcotecnia3.3. LM LM GlasfiberGlasfiber

Wind TurbineManufacturers

4.4. Hansen TransmissionsHansen Transmissions5.5. ABB Machines & DrivesABB Machines & Drives

ComponentManufacturers

6.6. SKF SKF UKUK7.7. GarradGarrad HassanHassan88 RELEXRELEX8.8. RELEXRELEX9.9. Durham UniversityDurham University1010 SZTAKISZTAKI

ExpertKnowledge

10.10. SZTAKI SZTAKI

3 of 25

Availability, Operatory, p

Operability100%

MTBFLogistic Delay

i

ScheduledRepair

MTTF

MTTR

Time Time

Time0% MTTR

4 of 25

Availability, Manufacturery,

Operability100% MTBF

MTTF

MTTRTime

0% MTTR

5 of 25

Availability, Manufacturer?y,

Operability100%

MTBF

Op keyturned off

MTTF

MTTR

turned off

Time0% MTTR

6 of 25

Availability & ReliabilityM Ti T F ilM Ti T F il MTTFMTTF•• Mean Time To Failure, Mean Time To Failure, MTTFMTTF

•• Mean Time to Repair, or downtime Mean Time to Repair, or downtime MTTRMTTR•• Logistic Delay TimeLogistic Delay Time LDTLDT•• Mean Time Between Failures, Mean Time Between Failures,

MTTF ≈ MTBFMTTF ≈ MTBFMTBF=MTTF+MTTR+LDTMTBF=MTTF+MTTR+LDT

•• Failure rate, Failure rate, λ λ λλ = 1/MTBF= 1/MTBF•• Repair rate, Repair rate, μμ μμ=1/MTTR=1/MTTR

MTBF≈MTTF+MTTR=1/MTBF≈MTTF+MTTR=1/λ +1/ μλ +1/ μ•• Manufacturer’s or Inherent Availability, Manufacturer’s or Inherent Availability,

AAii=(MTBF=(MTBF--MTTR)/MTBF=1MTTR)/MTBF=1−−(λ/μ)(λ/μ)•• Operational or Technical Availability, Operational or Technical Availability,

AA =MTTF/MTBF < 1=MTTF/MTBF < 1−−(λ/μ)(λ/μ)AAoo=MTTF/MTBF < 1=MTTF/MTBF < 1−−(λ/μ)(λ/μ)•• Typical UK valuesTypical UK values

–– Operational or Technical Availability Operational or Technical Availability 97%, 97%, –– Manufacturer’s or Inherent Availability Manufacturer’s or Inherent Availability 98%98%

7 of 25

a u actu e s o e e t va ab tya u actu e s o e e t va ab ty 98%98%

Cost of Energy, COE•• COE, £/kWh= COE, £/kWh=

O&M+ {[(ICC*FCR) + LRC]/O&M+ {[(ICC*FCR) + LRC]/AEPnetAEPnet} } O&M=O&M=Cost of Operations & Maintenance £Cost of Operations & Maintenance £–– O&M=O&M=Cost of Operations & Maintenance, £Cost of Operations & Maintenance, £

–– ICC=ICC=Initial Capital Cost, £Initial Capital Cost, £–– FCRFCR=Fixed Charge Rate, interest, %=Fixed Charge Rate, interest, %–– LRCLRC==LevelisedLevelised Cost of Replacement, replacing unavailable Cost of Replacement, replacing unavailable

generation, £generation, £–– AEPAEP=Annualised Energy Production kWh=Annualised Energy Production kWh–– AEPAEP=Annualised Energy Production, kWh=Annualised Energy Production, kWh

•• COE , £/kWh = COE , £/kWh = O&M(O&M(λλ, , 1/1/μ) μ) + {[(ICC*FCR) + LRC(+ {[(ICC*FCR) + LRC(λλ,, 1/1/μ)μ)]/]/AEPnetAEPnet(A((A(μμ, , 1/1/λλ)} )}

8 of 25

Cost of Energy, COE•• Reduce failure rate, Reduce failure rate, λ,λ,

Reliability MTBFReliability MTBF 1/1/λλ increases andincreases andReliability MTBF, Reliability MTBF, 1/1/λ,λ, increases andincreases andAvailability, Availability, AA, improves, improves

•• Increase repair rate, Increase repair rate, μ,μ,Downtime MTTFDowntime MTTF 1/1/μ μ reducesreduces andandDowntime MTTF, Downtime MTTF, 1/1/μ, μ, reducesreduces,, andandAvailability, Availability, AA, improves, improvesTh f dTh f d COECOE•• Therefore reduces Therefore reduces COECOE

9 of 25

The Bathtub CurveFailure

Intensity, λ

PopulationTurbinefailures ofnumber Total

⎟⎟⎠

⎞⎜⎜⎝

⎛

Most turbines

(years) Period OperatingPopulationTurbine ⎠⎝=aλ

lie here

te)t( βρβλ −=

Early Life Useful Life Wear-out Period

Time, t

10 of 25

(β < 1) (β = 1) (β > 1)

Trend in Turbine Failure Rates with timewith time

11 of 25

Reliability & Size, EU

12 of 2612 of 25

Wind TurbineWind Turbine Configurations

13 of 25

Reliability & Subassemblies , EUy

IndustrialIndustrialReliability figures

14 of 26

Reliability & Downtime & Subassemblies, EU,

Electrical System LWK Failure Rate, approx 6000 Turbine Years

ISET Pivot Diagram Failure Rate and Downtime from 2 Large Surveys of European Wind Turbines

Hydraulic System

Other

Electrical Control WMEP Failure Rate, approx 7800 Turbine YearsLWK Downtime, approx 6000 Turbine Years

WMEP Downtime, approx 7800 Turbine Years

M h i l B k

Rotor Hub

Yaw System

Gearbox

Rotor Blades

Mechanical Brake

Drive Train

Generator

1 0.75 0.5 0.25 0 2 4 6 8 10 12 14 Annual failure frequency Do ntime per fail re (da s)

15 of 25

Annual failure frequency Downtime per failure (days)

RELIAWIND Training Course, 1st June 2009, Helsinki

Reliability & Time, LWK

0

LWK, E66, converter

0.8

1.0

/ yea

r]

ime:

9 y

ears

0.6

ty [

failu

res

actu

al e

laps

ed t

demonstrated reliabilitydemonstrated reliabilitydemonstrated reliability

0.2

0.4

lure

inte

nsit a

0 20 40 60 80 100

0.0

fai industrial range

16 of 25total test time [turbines * year]

Reliability & Time, LWK GeneratorsGenerators

0.8

1.0

LWK, E40, generator

res

/ yea

r]

ed ti

me:

11

yea

rs

0.8

1.0

LWK, E66, generator

res

/ yea

r]

sed

time:

7 y

ears

.00.

20.

40.

6

failu

re in

tens

ity [

failu

r

actu

al e

laps

e

industrial range

0.0

0.2

0.4

0.6

failu

re in

tens

ity [

failu

actu

al e

laps

industrial range

0 100 200 300 400 500

0

total test time [turbines * year]

0 20 40 60 80

0

total test time [turbines * year]

1.0

LWK, V27/225, generator

ear]

12 y

ears

81.

0

LWK, V39/500, generator

ear]

11 y

ears

0.2

0.4

0.6

0.8

ilure

inte

nsity

[fa

ilure

s / y

e

actu

al e

laps

ed ti

me:

1

0.2

0.4

0.6

0.8

ailu

re in

tens

ity [

failu

res

/ ye

actu

al e

laps

ed ti

me:

0 100 200 300 400

0.0

total test time [turbines * year]

fai

industrial range

0 100 200 300 400 500

0.0

total test time [turbines * year]

fa

industrial range

Figure 4.4: Variation between the failure rates of generator subassemblies, in the LWK

population of German WTs, using the PLP model. The upper two are low speed direct drive generators while the lower two are high speed

17 of 25

The upper two are low speed direct drive generators while the lower two are high speed indirect drive generators.

Reliability & Time, LWKGearboxesGearboxes

0.8

1.0

LWK, TW600, gearbox

/ ye

ar]

time:

12

yea

rs

0.8

1.0

LWK, V39/500, gearbox

/ yea

r]

ime:

11

yea

rs

0.2

0.4

0.6

failu

re in

tens

ity [

failu

res

actu

al e

laps

ed t

industrial value

0.2

0.4

0.6

failu

re in

tens

ity [

failu

res

actu

al e

laps

ed ti

industrial value

0 100 200 300

0.0

total test time [turbines * year] 0 100 200 300 400 500

0.0

total test time [turbines * year]

1.0

LWK, N52/N54, gearbox

ars 1.0

LWK, Micon M530, gearbox

ars

0.4

0.6

0.8

nten

sity

[fa

ilure

s / y

ear]

actu

al e

laps

ed ti

me:

7 y

ea

0.4

0.6

0.8

nten

sity

[fa

ilure

s / y

ear]

actu

al e

laps

ed ti

me:

12

yea

0 20 40 60 80 100

0.0

0.2

total test time [turbines * year]

failu

re in

industrial value

0 50 100 150 200 250

0.0

0.2

total test time [turbines * year]

failu

re in

industrial value

Figure 4 5: Variation between the failure rates of gearbox subassemblies using the

18 of 25

Figure 4.5: Variation between the failure rates of gearbox subassemblies, using the PLP model, in the LWK population of German WTs.

Reliability & Time, LWK ElectronicsElectronics

60.

81.

0

LWK, E40, electronics

ures

/ ye

ar]

psed

tim

e: 1

1 y

ears

0.8

1.0

LWK, E66, electronics

res

/ yea

r]

sed

time:

7 y

ears

0 100 200 300 400 500

0.0

0.2

0.4

0.6

failu

re in

tens

ity [

failu

actu

al e

lap

industrial range

0.0

0.2

0.4

0.6

failu

re in

tens

ity [

failu

actu

al e

laps

industrial range

0 100 200 300 400 500

total test time [turbines * year] 0 20 40 60 80

total test time [turbines * year]

0.8

1.0

LWK, TW 1.5s, electronics

/ yea

r]

me:

5 y

ears

00.

20.

40.

6

failu

re in

tens

ity [

failu

res

/

actu

al e

laps

ed ti

m

industrial range

0 10 20 30 40 50

0.0

total test time [turbines * year] Figure 4.6: Variation between the failure rates of electronics subassemblies, or converter,

using the PLP model, in the LWK population of German WTs. The upper two are low speed direct drive generators with fully rated converters while the lower

hi h d i di d i i h i ll d

19 of 25

two are high speed indirect drive generators with partially rated converters.

Reliability of Electronics,Important Root CausesImportant Root Causes

1.Components 2. Environmental conditions•Stochastic variation of the wind•Diurnal variation of the weather•Geographical locationGeographical location

3. Control•Excessive I/O from converters•Uncoordinated Alarms from excessive I/O

Failure root cause distribution for power electronics

from E Wolfgang, 2007

excessive I/O • False alarms causing unnecessary trips

20 of 25

Downtime of Wind TurbinesG 1994 2004Germany 1994-2004

35

40

25

30

bine

15

20

st p

er T

ur

5

10

Hou

rs L

os

total

subassembly failures0Sep-94 Oct-95 Dec-96 Jan-98 Feb-99 Mar-00 Apr-01 May-02 Jun-03 Aug-04

H subassembly failures

21 of 25

Variable Load of Wind PowerLine Side InverterLine Side Inverter

22 of 25

Variable Load of Wind PowerGenerator Side InverterGenerator Side Inverter

23 of 25

Conclusions•• Definitions of Availability are open to interpretationDefinitions of Availability are open to interpretation•• Definitions of Availability are open to interpretationDefinitions of Availability are open to interpretation•• Unreliability > 1 failure/turbine/year is commonUnreliability > 1 failure/turbine/year is common•• Unreliability increases with turbine sizeUnreliability increases with turbine size•• Such unreliability will be unacceptable offshoreSuch unreliability will be unacceptable offshore•• Offshore we need unreliability < 0.5 failure/turbine/yearOffshore we need unreliability < 0.5 failure/turbine/year•• Unreliability concentrated mainly in the Drive Train including electricsUnreliability concentrated mainly in the Drive Train including electrics•• Some unreliable subassemblies are surprising:Some unreliable subassemblies are surprising:

For example gearboxes are not unreliableFor example gearboxes are not unreliable–– For example gearboxes are not unreliableFor example gearboxes are not unreliable–– But gearbox failures cause large downtime and costsBut gearbox failures cause large downtime and costs–– But electrical parts are unreliableBut electrical parts are unreliable–– Cause less downtime but significant costs, their downtime will increase Cause less downtime but significant costs, their downtime will increase

ff hff hoffshoreoffshore•• For electrical parts the root causes from these surveys are not clear:For electrical parts the root causes from these surveys are not clear:

–– ComponentsComponents–– Environmental conditionsEnvironmental conditions–– ControlsControls

•• But the highly variable loading on turbines is clearly a factorBut the highly variable loading on turbines is clearly a factor•• And false alarms are almost certainly a factorAnd false alarms are almost certainly a factor•• PrePre testing is essential to eliminate early life failurestesting is essential to eliminate early life failures

24 of 25

•• PrePre--testing is essential to eliminate early life failurestesting is essential to eliminate early life failures

Thank you• www.reliawind.eu

i d k/• www.supergen-wind.org.uk/• P. J. Tavner, C. Edwards, A. Brinkman, and F. Spinato. Inuence of wind

speed on wind turbine reliability. Wind Engineering, 30(1), 2006.• P. J. Tavner, J. P.Xiang, and F. Spinato. Reliability analysis for wind turbines.

Wind Energy, 10(1), 2007.• F. Spinato, P. J. Tavner, and G.J.W van Bussel. Reliability-growth analysis of

wind turbines from field data. Proceedings of AR2TS conference, Loughborough, 2007.

• P J Tavner, G J W van Bussel, F Spinato, Machine and converter reliabilities in WTs. Proceedings of IEE PEMD Conference, Dublin, April 2006.g f f , , p

• B Hahn, M Durstewitz, K Rohrig , Reliability of wind turbines, experiences of 15 years with 1,500 WTs, in ‘Wind Energy’ (Springer, Berlin, 2007), ISBN 10 3-540-33865-9.

• Hansen, A D., Hansen, L H. ,Wind turbine concept market penetration over 10 years (1995–2004), Wind Energy, 2007; 10:81–9710 years (1995 2004), Wind Energy, 2007; 10:81 97

• Ribrant J., Bertling L.M.: Survey of failures in wind power systems with focus on Swedish wind power plants during 1997–2005, IEEE Trans. Energy Conversion, 2007, EC22 (1), pp. 167–173

• Wolfgang, E. Examples for failures in power electronics systems, in EPE Tutorial ‘Reliability of Power Electronic Systems’ April 2007Tutorial Reliability of Power Electronic Systems , April 2007.

• Beckendahl, P, SKIIP, an Intelligent Power Module for wind turbine inverters, EPE Wind Chapter Mtg, Stockholm, May 2009

25 of 25