Ken Lee: 2013 Sandia National Laboratoies Wind Plant Reliability Workshop
- 1. Underestimated Low-Cycle Fatigue as a
Contributor to Premature Failures
Ken T. Lee, MSAE, Blade Design Technical Lead
Amir Bachelani, Blade Structural Engineer
Cody Moore, Loads & Dynamics Engineer
Kyle K. Wetzel, Ph.d., CEO/CTO
Sandia Blade Reliability Workshop 2013
Albuquerque, New Mexico
- 2. Outline
Wind Turbine Component Damage/Failure
Low-Cycle Fatigue – Premature Failures
Design Loads for Certification
Fatigue Analysis for Certification
Recommendations for Future Research
Copyright © 2013 Wetzel Engineering, Inc. All Rights Reserved.
- 3. WT Component Damage/Failure
Share of main components of
total number of failures
Damage/failure occurrences
attributed to:
electrical systems
rotor blades
gearbox
Structural in nature sustained
early in the life of these
components (sometimes
within 3 or 6 months to a
year) continuous and
costly maintenance and
repair
- 4. WT Component Damage/Failure
Wind Turbine
subassembly reliability:
WT reliability & downtime
varies across subcomponents
of WT
Failure frequencies of blades
and gearboxes lower
compared to other
subcomponents
Downtime per failure for rotor
blades and gearbox higher
compared to other
subcomponents
Pitch mechanism & electrical
system shown to be significant
sources.
Failure or fault-induced
shutdowns
- 5. Common WT Component Damage/Failure
Common Types of Rotor
Blade Damage/Failure:
Crack propagation due to inherent
design or manufacturing defects
Blade damage due to extreme
load buckling or blade-tower
strike
Failure of adhesive bonds
leading edge/trailing edge
splitting
Blade failure at the blade root
connection blade throw
- 6. Causes of WT Component Damage/Failure
Operating loads exceed design loads:
Underestimated Design Loads (Ultimate, Low-cycle fatigue, or High-cycle fatigue)
Mistakes in applying standard methods
Shortcomings in standard methods
Malfunctioning control safety systems
Improper site assessment
Insufficient assessment of structural integrity:
Mistakes in applying standard methods
Shortcomings in standard methods
- 7. Low-Cycle Fatigue (LCF)
Possible Contributor to Premature Wind Turbine
Component Failures
Focus: LCF – Low-cycle fatigue characterized as:
• High amplitude alternating stress
• Low cycle counts (approx. less than 104)
• Fatigue failure: propagation-dominated
- 8. LCF - Overview
Components designed to 20 years of life (according to GL, IEC, DNV)
Premature failure of WT subcomponents:
Rotor Blades
Gearbox
Structural design & analysis of WT Components, Current Provision:
Fatigue design focused on high-cycle fatigue (HCF) (i.e. S-N curve methods).
High amplitude loading events that occur at low cycle counts induce crack
growth rates that far exceed that predicted by S-N curve analyses.
- 9. LCF - Overview
Design Loads Simulations, Current Framework:
Does not capture high-amplitude transient events (less than 104 – 504 cycles)
Insufficient resolution of the “tails” of the loads spectra “Normal-operation” during
turbulent conditions
Under-reporting of loads induced by transient events, coherent inflow conditions, fault-
induced shutdowns.
Blade Load Spectra as per Current
Framework for Certification Design Loads:
Average Normalized Cumulative Blade Load
Spectra with/without LCF:
- 10. Design Loads for Certification
Case Study on the Impact of Emergency Stops on
Blade Fatigue Life
- 11. GL -Design Load Cases for Fatigue Analysis
•Power Production Load Case:
DLC 1.1; NTM; Vin < V < Vout
•Idling or Parked Load Case:
DLC 6.4; NTM; V<Vin and Vout < V < Vref
- 12. GL -Design Load Cases for Fatigue Analysis
•Transient Load Cases, Faults during Operation:
DLC 1.4; NWP; Grid-loss/E-stop – 20/year
DLC 1.8; NWP; Iced Conditions – 24 hours/year
DLC 2.1; NWP; Fault Occurrence – 24 hours/year
DLC 3.1; NWP; Start Up - 1000/year @ Vin, 50 @ Vr & Vout
DLC 4.1; Normal Shutdown - 1000/year @ Vin, 50 @ Vr & Vout
- 13. Possible Shortcomings in Current Guidelines
• WT components could possibly be failing prematurely from fatigue.
• Current guidelines do not account for higher frequency rates of E-stops and/or
Fault-Induced shutdown procedures
• E-stops due to grid loss could have a higher impact on fatigue life
• E-stops of more than 20/year, averaging 1 E-stop/day (360/year)
Early life of WT operation, debugging of control and safety systems
Control system malfunction in response to abnormal inflow conditions
• Impact of an E-stop or a pitch control system fault during a gusting wind (with or
without turbulent inflow) not considered.
- 14. Case Study of Impact of Increased Frequency of
E-Stop Procedures
• Fatigue life of two turbines
analyzed the impact of stricter
provisions for Emergency
Shutdown (E-stops)
20 cases/year of DLC 1.1 NTM (GL-
2010), grid loss timed at highest
point of My
Increase cycle counts of DLC 1.4
(NWP) Estops from 20/year to
360/year (7200 in turbine lifetime)
1 case/year of DLC 1.5 EOG1 (GL-
2010), grid loss timed at highest
point of My
21 cases/year of grid loss during a
gusting wind were added to the
analysis
- 18. Current Fatigue Analyses Guidelines
Fiber-Reinforced Composite Laminates:
Characteristic S-N curve established for laminate. (GL 5.5.3.3.1)
Goodman diagram constructed using this curve. (GL 5.5.3.3.1)
Fatigue damage calculation using Miner’s rule (IEC 61400-1, 7.6.3.2)
95 % survival probability with a confidence level of 95 % used as basis for SN-
curve (IEC 61400-1, 7.6.3.2)
Adhesives:
3 samples (being representative for the jointed components in geometry
and material)
Minimum number of load cycles of N=106. (GL 5.5.6.10)
- 19. •HCF estimation is captured. LCF response of WT
subcomponents/blade composites should also be considered in
guidelines.
•Testing on pristine laminates. Manufacturing defects not tested.
•Structural damage/failures due to:
Manufacturing that is out of QC
Failure to design for manufacturing quality that can be realistically
controlled.
•Current Safety Factors (SF’s) result in excessively conservative
designs in areas not really the source of problems.
•Heavy reliance on SF’s still fails to address on-going problems
industry has with blade reliability.
Current Fatigue Analyses Guidelines
- 20. Types of Damage & Defects
• Delamination
•Voids & Cracks in Laminates/Adhesive Bonds
• Wavy fiber plies, bridging of fibers
• Porosity, discontinuities in laminates
- 21. Crack Growth in Fiber-Reinforced Plastics
•Region I – Matrix Cracking
•Region II – Matrix-Fiber Interface
Cracking
•Region III – Fiber Cracking
- 22. Crack Growth Modes in Adhesives
•Mode I – Opening Stresses (Peel Stress)
•Mode II – Shear Stresses
•Mixed Mode Loading of I and II
- 23. Low Cycle Fatigue (LCF) - Unidirectional
Composites
• Normal S-N curve does not properly capture LCF.
• A bi-linear S-N curve can be used as a basis to capture structural
response from LCF.
- 25. Crack Growth - Adhesives
Crack Length Range, mm Best Fit Equation, mm/cycle
0 – 20
0.99
.
20 – 40
2.77
.
40 – 80
6.70
.
- 26. Future Research Directions & Recommendations
Design Fatigue Loads Estimation:
Fatigue loads estimations methods should analyze:
Increased probability of high amplitude transient events resulting from:
Fault-induced shutdown procedures
Control system fault or Emergency shutdown during a gusting wind
and/or coherent inflow with wind directional changes
Rare occurrences of extreme oblique inflow
Increased cycle counts of high amplitude loading to capture extreme
statistics where peak load will occur during the early stages in the
operating life of WTG.
Increase resolution of the tails of fatigue loads spectra.
- 27. Future Research Directions & Recommendations
Structural Design for Fatigue:
Structural influences of manufacturing defects, response to
LCF
Crack growth modeling
Fracture mechanics
Improved S-N curve analyses:
Non-pristine laminates
Adhesive bonds
Sandwich cores
Anticipated manufacturing tolerances on structurally critical
members
Analyses should define tolerances expected of the
manufacturing QMS
Establish SF’s that rely less on testing laboratory coupons to
establish material strength and properties, more on testing of
components and subsystems that more closely reflect actual
QC to establish.