prognostic control to enhance offshore wind turbine operations and maintenance strategies

Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company,for the United States Department of Energy’s National Nuclear Security Administration

under contract DE-AC04-94AL85000.

D. Todd Griffith1, Nate Yoder2, Brian Resor1, Jon White1, Josh Paquette1,

Alistair Ogilvie1, Valerie Peters1

1Sandia National Laboratories2ATA Engineering

European Wind Energy Association (EWEA) 2012 Annual Event

April 16-19,2012; Copenhagen, Denmark

Prognostic Control to Enhance Offshore Wind Turbine

Operations and Maintenance Strategies

Background• Operations and maintenance (O&M) costs for offshore wind plants are estimated to be

2-5x those for onshore plants and represent 20-30% of the total levelized cost of energy [1]

• Increased loading and environmental harshness• Difficulty of access

• Structural health monitoring of offshore turbines as part of a condition based maintenance paradigm could provide significant cost reductions

• Reduce unscheduled maintenance • Improve supply chain management• Smart turbine load management

• Operate and maintain turbines to maximize overall profit

• [Sensing, CBM/SHM, Solid Mechanics, Controls]

[3]

[2]

Simulation Campaign: Broad View

The most relevant damage and fault conditions for simulation and further study are being identified using other

Sandia National Laboratories projects.

Simulation Campaign

Sandia/DOE Blade Reliability

Collaborative (BRC)

Sandia/DOE Continuous Reliability

Enhancement for Wind (CREW)

[Sensing, CBM/SHM, Solid Mechanics, Controls]

Multi-scale Simulation of Damage

Equivalent Blade Model

(SNL BPE)


(SNL BPE)

0 0.2 0.4 0.6 0.8 110

4

106

108

1010

1012

BlFract

Fla

pw

ise

Stif

fne

ss, F

lpS

tff (

N*m

2 )

NREL 5MW Baseline PropertiesSNL 5MW All Glass FE Model PropertiesSNL 5MW w/Carbon FE Model Properties

Healthy / Damaged

Blade Model

(ANSYS)

Healthy / Damaged

Blade Model

(ANSYS)

Full Turbine Simulations

(FAST or ADAMS)


(FAST or ADAMS)

Offshore Turbine Model

(NREL 5MW)


(NREL 5MW)

Ground

Water

High Fidelity

Simulations

(ANSYS)

High Fidelity

Simulations

(ANSYS)

Local Sensitivity

(ANSYS)

Local Sensitivity

(ANSYS)

Global Operating Sensitivity

(FAST or ADAMS)


(FAST or ADAMS)

12

34

56

2 3 4 5 6 7 8 9

0

1

2

3

4

5

Measurement IDDamage State

Ske

wne

ss D

iffer

ence

[%

]

Damage Mitigating

Control

(FAST)

Damage Mitigating

Control

(FAST)

75808590951000

0.2

0.4

0.6

0.8

1

Bla

de

Ro

ot

Ou

t Of P

lan

e B

en

din

g M

om

en

tN

orm

aliz

ed

Fa

tigu

e D

am

ag

e

Percent of Turbine Rating

Healthy/Damage Blade Model• Case Study #1: Blade models

with variable length trailing edge disbonds created

[4]

Sandia 5-MW NuMAD Blade Model• A NuMAD blade model of a 61.5 m blade was created using existing blade geometry

from the DOWEC[5,6] study for the preliminary model development• First model was heavier than NREL 5-MW reference wind turbine[7] so UD carbon spar

caps were added

0

20

40

60

80

100

120

0 10 20 30 40 50 60 70

Num

ber o

f lay

ers (

-)

Blade Span (m)

GELCOAT

TRIAX-SKINS

TRIAX-ROOT

UDCarbon

UD_GLASS_TE

FOAM

Model Version Total Mass

NREL 5-MW Blade 17,740 kg

Sandia Fiberglass 5-MW Blade 25,630 kg

Sandia 5-MW With Carbon Spar Caps 16,381 kg


(FAST or ADAMS)


(FAST or ADAMS)

12

34

56

2 3 4 5 6 7 8 9

0

1

2

3

4

5


Ske

wne

ss D

iffer

ence

[%

]



(SNL BPE)


(SNL BPE)

0 0.2 0.4 0.6 0.8 110

4

106

108

1010

1012

BlFract

Fla

pw

ise

Stif

fne

ss, F

lpS

tff (

N*m

2 )


Healthy / Damaged

Blade Model

(ANSYS)

Healthy / Damaged

Blade Model

(ANSYS)


(FAST or ADAMS)


(FAST or ADAMS)


(NREL 5MW)


(NREL 5MW)

Ground

Water

High Fidelity

Simulations

(ANSYS)

High Fidelity

Simulations

(ANSYS)

Local Sensitivity

(ANSYS)

Local Sensitivity

(ANSYS)

Damage Mitigating

Control

(FAST)

Damage Mitigating

Control

(FAST)

75808590951000

0.2

0.4

0.6

0.8

1

Bla

de

Ro

ot

Ou

t Of P

lan

e B

en

din

g M

om

en

tN

orm

aliz

ed

Fa

tigu

e D

am

ag

e


Reduced Order Blade Models• High fidelity blade models reduced to

equivalent beam elements for full turbine simulations

BPE

100s of DOF10s to 100s of thousands of DOF

BPE Stiffness Changes due to Trailing Edge Disbonds

• A decrease of over 1% seen in a blade section for the 0.75 m disbond• For the 6 m disbond two sections have decreases larger than 7.5%, max=13.3%

020

4060

2

4

6

0

5

10

Elem. Loc. [m]

Percent Decrease in Torsional Stiffness

Disbond Length [m]

Per

cen

t D

ecre

ase

[%]

020

4060

2

4

60

5

10

Disbond Length [m]

Percent Decrease in Torsional Stiffness

Elem. Loc. [m]

Per

cen

t D

ecre

ase

[%]

Stiffness Changes due to Trailing Edge Disbonds

• Even large TE disbonds caused very small decreases of less than 0.5% in both flap- and edge-wise bending stiffnesses

020

4060

2

4

6

0

5

10

Elem. Loc. [m]

Percent Decrease in Flap-wise Stiffness

Disbond Length [m]

Per

cen

t D

ecre

ase

[%]

020

4060

2

4

6

0

5

10

Elem. Loc. [m]

Percent Decrease in Edge-wise Stiffness

Disbond Length [m]

Per

cen

t D

ecre

ase

[%]


(FAST or ADAMS)


(FAST or ADAMS)

12

34

56

2 3 4 5 6 7 8 9

0

1

2

3

4

5


Ske

wne

ss D

iffer

ence

[%

]



(SNL BPE)


(SNL BPE)

0 0.2 0.4 0.6 0.8 110

4

106

108

1010

1012

BlFract

Fla

pw

ise

Stif

fne

ss, F

lpS

tff (

N*m

2 )


Healthy / Damaged

Blade Model

(ANSYS)

Healthy / Damaged

Blade Model

(ANSYS)


(FAST or ADAMS)


(FAST or ADAMS)


(NREL 5MW)


(NREL 5MW)

Ground

Water

High Fidelity

Simulations

(ANSYS)

High Fidelity

Simulations

(ANSYS)

Local Sensitivity

(ANSYS)

Local Sensitivity

(ANSYS)

Damage Mitigating

Control

(FAST)

Damage Mitigating

Control

(FAST)

75808590951000

0.2

0.4

0.6

0.8

1

Bla

de

Ro

ot

Ou

t Of P

lan

e B

en

din

g M

om

en

tN

orm

aliz

ed

Fa

tigu

e D

am

ag

e


Full Turbine Simulations• Reduced order blade models integrated into 5-MW

offshore turbine models in FAST and ADAMS

Ground

Water

[1]

20 m



(SNL BPE)


(SNL BPE)

0 0.2 0.4 0.6 0.8 110

4

106

108

1010

1012

BlFract

Fla

pw

ise

Stif

fne

ss, F

lpS

tff (

N*m

2 )


Healthy / Damaged

Blade Model

(ANSYS)

Healthy / Damaged

Blade Model

(ANSYS)


(FAST or ADAMS)


(FAST or ADAMS)


(NREL 5MW)


(NREL 5MW)

Ground

Water

High Fidelity

Simulations

(ANSYS)

High Fidelity

Simulations

(ANSYS)

Local Sensitivity

(ANSYS)

Local Sensitivity

(ANSYS)

Damage Mitigating

Control

(FAST)

Damage Mitigating

Control

(FAST)

75808590951000

0.2

0.4

0.6

0.8

1

Bla

de

Ro

ot

Ou

t Of P

lan

e B

en

din

g M

om

en

tN

orm

aliz

ed

Fa

tigu

e D

am

ag

e



(FAST or ADAMS)


(FAST or ADAMS)

12

34

56

2 3 4 5 6 7 8 9

0

1

2

3

4

5


Ske

wne

ss D

iffer

ence

[%

]

0 10 20 30 40 50 60-5

0

5

1 2 3 4 5 6 7 8 9

Blade Span [m]

Bla

de C

hord

[m

]

FAST Sensitivity results

• Skewness of pitching moment on damaged blade demonstrated highest sensitivity to presence and size of disbond

12

34

56

2 3 4 5 6 7 8 9

0

1

2

3

4

5


Ske

wne

ss D

iffer

ence

[%

]

23

1

2

1

3

1

1

ˆ

n

ii

n

ii

xxn

xxn

Full Offshore Turbine Simulations using ADAMS

• 1,007 total response measurements both on- and off-rotor were investigated

• Pitching moments once again showed large changes in moments

1020

3040

5060

2

4

60

5

10

15

20

Disbond Length [m]

Absolute Percent Change in Standard Deviation Due to Disbond

Measurement Location [m]

Per

cent

Cha

nge

[%]

1020

3040

5060

2

4

60

100

200

300

Disbond Length [m]

Absolute Percent Change in Skewness Due to Disbond


Per

cent

Cha

nge

[%]

1020

3040

5060

2

4

60

1

2

3

4

Disbond Length [m]

Absolute Percent Change in Mean Due to Disbond


Per

cent

Cha

nge

[%]



(SNL BPE)


(SNL BPE)

0 0.2 0.4 0.6 0.8 110

4

106

108

1010

1012

BlFract

Fla

pw

ise

Stif

fne

ss, F

lpS

tff (

N*m

2 )


Healthy / Damaged

Blade Model

(ANSYS)

Healthy / Damaged

Blade Model

(ANSYS)


(FAST or ADAMS)


(FAST or ADAMS)


(NREL 5MW)


(NREL 5MW)

Ground

Water

High Fidelity

Simulations

(ANSYS)

High Fidelity

Simulations

(ANSYS)

Local Sensitivity

(ANSYS)

Local Sensitivity

(ANSYS)

Damage Mitigating

Control

(FAST)

Damage Mitigating

Control

(FAST)

75808590951000

0.2

0.4

0.6

0.8

1

Bla

de

Ro

ot

Ou

t Of P

lan

e B

en

din

g M

om

en

tN

orm

aliz

ed

Fa

tigu

e D

am

ag

e



(FAST or ADAMS)


(FAST or ADAMS)

12

34

56

2 3 4 5 6 7 8 9

0

1

2

3

4

5


Ske

wne

ss D

iffer

ence

[%

]

ANSYS Strain Field Investigation• Disbond results in

localized difference in strain localized around the edges of the disbond



(SNL BPE)


(SNL BPE)

0 0.2 0.4 0.6 0.8 110

4

106

108

1010

1012

BlFract

Fla

pw

ise

Stif

fne

ss, F

lpS

tff (

N*m

2 )


Healthy / Damaged

Blade Model

(ANSYS)

Healthy / Damaged

Blade Model

(ANSYS)


(FAST or ADAMS)


(FAST or ADAMS)


(NREL 5MW)


(NREL 5MW)

Ground

Water

High Fidelity

Simulations

(ANSYS)

High Fidelity

Simulations

(ANSYS)

Local Sensitivity

(ANSYS)

Local Sensitivity

(ANSYS)


(FAST or ADAMS)


(FAST or ADAMS)

12

34

56

2 3 4 5 6 7 8 9

0

1

2

3

4

5


Ske

wne

ss D

iffer

ence

[%

]

Damage Mitigating

Control

(FAST)

Damage Mitigating

Control

(FAST)

75808590951000

0.2

0.4

0.6

0.8

1

Bla

de

Ro

ot

Ou

t Of P

lan

e B

en

din

g M

om

en

tN

orm

aliz

ed

Fa

tigu

e D

am

ag

e


Smart Turbine Load Management (1)• Turbine load management

• Fatigue damage implications are considered first

• Others: residual strength, deflection, etc.

75808590951000.75

0.8

0.85

0.9

0.95

1

Bla

de

Ro

ot

Ou

t Of P

lan

e B

en

din

g M

om

en

tN

orm

aliz

ed

Fa

tigu

e D

am

ag

e E

qu

iva

len

t Lo

ad


75808590951000

0.2

0.4

0.6

0.8

1

Bla

de

Ro

ot

Ou

t Of P

lan

e B

en

din

g M

om

en

tN

orm

aliz

ed

Fa

tigu

e D

am

ag

e


Repa

ir Co

st

Defect Size

Blade Removal

Blade Replacement

Up-Tower Repair

Ground Repair

No Repair(not to scale)

Damage Equivalent

Load

Fatigue Damage

Smart Turbine Load Management (2)• Prognostic Control: damage mitigating controls strategy to

• Increase life• Increase energy capture• Improved maintenance planning

0

1000

2000

3000

4000

5000

6000

0 5 10 15 20 25

Gene

rato

r Pow

er (k

W)

Wind Speed (m/s)

Normal Operation

Mode 1

Mode 2

Mode 3

Summary• A multi-scale methodology was developed to investigate the sensitivity of

operational response measurements; Case Study #1: Trailing Edge Disbond• Addresses sensor selection and damage characterization through sensitivity

analysis at both fine and coarse scales of the model• Prognostic control strategies were considered to manage turbine loads in the

presence of damage for improved offshore turbine O&M strategies• The simulation campaign is currently being exercised for additional forms of

damage in blade, tower and foundation as well as operating fault conditions (e.g. rotor imbalance)

• Cost benefit analysis and alternate prognostic controls are being explored

19

1. W. Musial, R. Thresher, B. Ram. Large-Scale Offshore Wind Energy for the United States: Assessment of Opportunities and Barriers. CO, Golden: National Renewable Energy Laboratory, 2010.

2. http://www.eurocopter.co.uk3. http://www.oceanpowermagazine.net4. http://www.netcomposites.com5. Kooijman, H.J.T., Lindenburg, C., Winkelaar, D., and van der Hooft, E.L., “DOWEC 6 MW Pre-Design: Aero-elastic modeling of the

DOWEC 6 MW pre-design in PHATAS,” ECN-CX--01-135, DOWEC 10046_009, Petten, the Netherlands: Energy Research Center of the Netherlands, September 2003.

6. Lindenburg, C., “Aeroelastic Modeling of the LMH64-5 Blade,” DOWEC-02-KL-083/0, DOWEC 10083_001, Petten, the Netherlands: Energy Research Center of the Netherlands, December 2002.

7. Jonkman, J.; Butterfield, S.; Musial, W.; and Scott, G., "Definition of a 5-MW Reference Wind Turbine for Offshore System Development," NREL/TP-500-38060, Golden, CO: National Renewable Energy Laboratory, February 2009.

References

Extra Slides



(SNL BPE)


(SNL BPE)

0 0.2 0.4 0.6 0.8 110

4

106

108

1010

1012

BlFract

Fla

pw

ise

Stif

fne

ss, F

lpS

tff (

N*m

2 )


Healthy / Damaged

Blade Model

(ANSYS)

Healthy / Damaged

Blade Model

(ANSYS)


(FAST or ADAMS)


(FAST or ADAMS)


(NREL 5MW)


(NREL 5MW)

Ground

Water

High Fidelity

Simulations

(ANSYS)

High Fidelity

Simulations

(ANSYS)

Local Sensitivity

(ANSYS)

Local Sensitivity

(ANSYS)

Damage Mitigating

Control

(FAST)

Damage Mitigating

Control

(FAST)

75808590951000

0.2

0.4

0.6

0.8

1

Bla

de

Ro

ot

Ou

t Of P

lan

e B

en

din

g M

om

en

tN

orm

aliz

ed

Fa

tigu

e D

am

ag

e



(FAST or ADAMS)


(FAST or ADAMS)

12

34

56

2 3 4 5 6 7 8 9

0

1

2

3

4

5


Ske

wne

ss D

iffer

ence

[%

]

Equivalent Beam Properties with BPE

6

• A Matlab based interface between NuMAD and BPE was created so that equivalent beam properties could be extracted from NuMAD models

• The developed model has effective blade structural properties that closely approximate those used in the NREL 5-MW reference wind turbine blades[7]

0 0.2 0.4 0.6 0.8 110

1

102

103

104

BlFract

Ma

ss D

en

sity

, BM

ass

De

n (

kg/m

)


0 0.2 0.4 0.6 0.8 110

6

107

108

109

1010

1011

BlFract

Ed

ge

wis

e S

tiffn

ess

, Ed

gS

tff (

N*m

2 )


0 0.2 0.4 0.6 0.8 110

4

106

108

1010

1012

BlFract

Fla

pw

ise

Stif

fne

ss, F

lpS

tff (

N*m

2 )


0 0.2 0.4 0.6 0.8 110

5

106

107

108

109

1010

BlFractT

ors

ion

al S

tiffn

ess

, GJS

tff (

N*m

2 )


Damaged Beam Properties with BPE

7

• BPE was modified to allow for the automated insertion of variable length disbonds at a user specified location

• Using BPE variety one healthy and a series of damaged blade models were created• 36 damaged blades with trailing edge disbonds extending between 0.125 and 6 m

outboard from max chord were created • BPE modified slightly to improve performance over localized stiffness

discontinuities

Full Offshore Turbine Simulations using ADAMS

• Average responses also show significant changes

• Changes localized to damaged region

• Other responses and sensitivity measures currently being investigated

13

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-200

-180

-160

-140

-120

-100

-80

Rotations

Net

Pitc

hing

Mom

ent

[kN m

]

Blade 1 Mean Local Pitching Moment at 15.8 m

1020

3040

5060

2

4

60

0.2

0.4

0.6

0.8

1

Disbond Length [m]

Max Standardized Difference in Blade 1 Pitching Moments


Sta

ndar

dize

d D

iffer

ence

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-170

-160

-150

-140

-130

-120

-110

-100

-90

-80

Rotations

Net

Pitc

hing

Mom

ent

[kN

m]

Blade 1 Mean Local Pitching Moment at 20 m

Healthy

0.625 m1.25 m

1.875 m

2.5 m

4 m6 m

+/- Std Dev

prognostic control to enhance offshore wind turbine operations and maintenance strategies

Documents