1. Results from WINDIE Experiment to Characterize
Inspection Methods for Wind Blades and Use of Probability
of Detection Studies to Quantify NDI Performance
Dennis Roach, Tom Rice, Stephen Neidigk,
Randy Duvall, Josh Paquette
Sandia National Labs
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.
2. Blade Reliability Collaborative - Objective
Create the ability for manufacturers to determine the
quality of their product before it leaves the factory & to
enhance the in-service inspection of wind blades
Required Relationship Between
Structural Integrity and
Inspection Sensitivity
Detectable Flaw Size
1 2 3 4 5 6 7 8 9 10
Nondestructive Inspection
Need this
overlap
1 2 3 4 5 6 7 8 9 10
Damage Tolerance
Allowable Flaw Size
3. Sandia Labs Wind Turbine Blade
Test Specimen Library
Engineered Test Specimens Wind Blade Specimens
4. Different Flaw Types Engineered into
NDI Feedback Specimens
Glass Beads Grease Mold Release Pillow Insert
Voids in
bond joint
Materials inserted into multiple layers
Glass beads
In bond joint
Waviness produced Dry fabric areas
by pre-cured
resin rods
Pull tabs in
bond joint
5. Spar Cap and Shear Web
NDI Feedback Specimen No. 2
EXAMPLES OF VARIOUS FLAW
DEPTHS IN SPAR CAP SECTION
INSPECTION SURFACE
1.01"
0.34"
(.25" MR)
1.35"
0.68"
0.67"
1.35"
0.34"
1.01"
1.35"
25% (OF FULL THICKNESS)
50% (OF FULL THICKNESS)
FLAT BOTTOM HOLE (FBH)
75% (OF FULL THICKNESS)
50%
USED VECTORPLY ELT 5500
24 PLIES OF MATERIAL (UNIAXIAL FIBER)
1.00" DIA
.50" DIA
1.00" DIA
1.50" DIA
1.000" 2.000"
2.000"
.40" (10mm) BONDLINE
2.00" DIA
1.50" DIA
1.00" DIA
INSPECTION SIDE
PERCENTAGE OF FULL
THICKNESS AT BONDLINE
(.100" SKIN AND .400" BOND
THICKNESS)
PILLOW INSERT
NDI REFERENCE STANDARD 2 FABRICATION DRAWING
SPAR CAP AND SHEAR WEB BLADE SCHEMATIC
PULL TABS
(DISBONDS IN ADHESIVE)
(DELAMS) (DELAMS) (BASED ON 24 PLIES OF UNIAXIAL MAT'L)
(DISBONDS IN ADHESIVE)
25%
(.125" MR)
1.00" DIA
2.00" DIA
ADHESIVE
SHEAR WEB
FLAT BOTTOM HOLES
.50" DIA
2 PLIES OF DOUBLE BIAS (DB)
1.00" (25mm) FOAM CORE
INTERFACE 1
INTERFACE 1
INTERFACE 2
1.00" DIA
25%
(B/W PLIES 18 & 19)
75%
(B/W PLIES 6 & 7)
75%
(.375" MR)
FLAT BOTTOM HOLES 4 PLY PILLOW INSERTS
25% (B/W PLIES
18 & 19)
50% (B/W PLIES
12 & 13)
75% (B/W PLIES
6 & 7)
75% (1.01" MR) 50% (.68" MR) 25% (.34" MR)
2.00" DIA
.50" DIA
1.50" DIA
1.50" DIA
1.00" DIA
2.00" DIA
.50" DIA
2.00" DIA
.50" DIA
1.50" DIA
1.50" DIA
1.00" DIA
2.00" DIA
2.00" DIA
.50" DIA
18.00"
~1.35" (34mm) UNIAXIAL (SPANWISE)
30.00"
(+45, +45)
2 PLIES OF DOUBLE BIAS (DB)
11-30-10
MR = MATERIAL REMAINING
PLY NO. 1 OF SPAR CAP
2 PLIES OF
DOUBLE BIAS (DB)
(+_45, _+45)
(+45, +45)
(NOTE: IF USING TEFLON BASED RELEASE FABRIC
WHEN CURING MAIN SPAR, BE SURE TO LIGHTLY
SAND SURFACE AREA WHERE SHEAR WEB BOND
WILL TAKE PLACE) 0.60"-1.00"
2.500"
(BASED ON 24 PLIES OF UNIAXIAL MAT'L)
2.750"
NOTE: PULL TABS (.007" THK) WILL EXTEND OUT FROM SPECIMEN
EDGE DURING CURE PROCESS, BE SURE TO USE SPECIAL
CARE NOT TO PUNCTURE VACUUM BAG (COVER SHARP
EDGES WITH BREATHER FABRIC) . PULL TABS REMOVED
AFTER CURE PROCESS.
NOTES: 1 of 2
1. SPECIMEN CURED USING 14 IN. HG. VACUUM PRESSURE
AND VACUUM LEFT ON OVER NIGHT.
2. POST CURE SPECIMEN AT 70 C FOR 10 HOURS.
3. FINAL FLAT BOTTOM HOLE DEPTH MAY CHANGE DEPENDING
ON FINAL PART THICKNESS.
1.875"
2.750"
2.750"
2.750"
2.750"
(41)
(42)
(43)
(44)
(45)
(46)
(52)
(51)
(50)
(49)
(48)
(47)
(53)
(54)
(55)
(56)
(57)
(58)
(64)
(63)
(62)
(61)
(60)
(59)
(65) (66) (67) (68)
(69)
(70)
(71)
(72)
(73)
(74)
(75)
(76)
(77)
(78)
(79)
(80)
Flaw types to
include:
snowflaking,
porosity, resin-starved
regions,
voids, interply
delaminations,
spar and shear
web disbonds, ply
waviness
75% MR
Depth into adhesive
& shear web skin
50% MR
25% MR
6. NDI Feedback Specimens 1, 2 & 4 –
Shear Web & Foam Core Specimens
Shear Web/Spar with Disbonds and Delaminations
Laminate with Waviness
and Dry Regions
Foam Core with Disbonds
and Delaminations
7. Ultrasonic Deployment Progression
Single Element Transducer varying Diameter
• 500 KHz, 1 MHz, 1.5 MHz
Linear Encoded Phased Array
• 500 KHz,1 MHz, 1.5 MHz
• Multiple linear encoders
• 16, 32, 42 and 64 Elements
• 5 to 10 Water Box options
Automated and Semi-Automated X-Y Scanning
• MAUS V Automated Scanner
• OmniScan X-Y Glider
• Marrietta Automated Scanner
8. On-Blade Phased Array UT Inspections
16 Meter Station on
Fiberglass Spar Cap Blade
Adhesive Void
Between Spar
Cap and
Shear Web
Spar Cap Cross Section Schematic
Showing the Spar Cap, Adhesive
Bond Line and Shear Webs
Scanning Direction
Vertical Strip C-Scan Image
Showing Adhesive Void in
Upper Bond Line
Sealed water box and 1.5L16 Phased Array probe was used to
detect missing adhesive in bond lines
9. Wind Inspection NDI Experiment (WINDIE) -
Advanced NDI Vendor Participation
Advanced NDI Methods
for Wind Turbine Blades
Over 30 agencies invited
22 accepted invitation
Report completed detailing
advanced NDI screening
Inspection Method Company
P Linear Array UT 3D Matrix Eye Toshiba
P Phased Array UT Olympus NDT
Acoustic Emission iHMSi
P Air Coupled UT ISU
ANDSCAN-Robot Genesis Systems
Bandicoot CSIRO
Custom Systems Exova
P Digital Acoustic Video (Acoustacam) Imperium
Digital Image Correlation (DIC) Dantec Dynamics
Flaw Inspecta UT Array NDT Solutions Inc
P Focused Probe Immersion UT GE Inspection Technologies
Guided Ultrasonics Guided Ultrasonics
P Induction Thermography System & Air Coupled UT Boeing
Induction Thermography System (ITS) Quest Integrated Inc
P IR Inspection Ssytem (IRIS) Vista Engineering Technologies
Laminography Digiray
P Laser UT iPhoton
Line Thermography Mistras Group
Linear Array UT USUT Labs/Veracity
P Lock-In Thermography moviMED/MoviTherm
MAUS MIA Mode AANC
MAUS Resonance Mode AANC
P Microwave GE Global Research
P Microwave NDE Evisive
P Millimeter Wave Inspection Tool Physical Optics Corp. (POC)
P Phased Array UT AANC
Pulse Echo UT QinetiQ
P RapidScan2 (Phased Array Wheel Probe) Sonatest/R-CON NDT
Rotor Blade CT System iHMSi
P Shearography Dantec Dynamics
Shearography Laser Technology Inc
Sonic IR WSU
Terahertz Teraview
P Terahertz Radiation (T-Ray) Iowa State University
Terrahertz Imaging GMA Industries
Thermography AANC
P Thermography Thermal Wave Imaging
P Through Transmission AANC
Ties to QinetiQ Triton Systems
P TSCOUT (Thick Section Comp. UT) & PAC UT Mistras Group
UT and IR Systems TecScan
UT Spectroscopy QinetiQ
Various NDI IHI Southwest Technologies
P Vibro Thermography Resodyne
P MAUS Phased Array UT AANC
P RotoArray - Phased Array UT GE Inspection Technologies
10. WINDIE – Advanced NDI Screening Activity
Information gathered during round-robin inspections:
• Flaw detection peformance (type, sensitivity)
• Duration of inspection
• Fieldability (contact/noncontact)
• Deployment issues
• Inspection difficulties
• Cost of new system
• Accessories needed to make fieldable
• Ease of data interpretation
11. WINDIE – Specimens Used for NDI Comparison
REF-STD-4-135-SNL-1
(wrinkles & dry areas)
REF-STD-2-127-173-SNL-1
12. Phased/Linear Array Ultrasonics
Olympus OmniScan Toshiba MatrixEye
• Ultrasonic probes consists of 16 to 256 individual elements
• Can produce A, B, and C-scans
• Low frequency (0.5 to 1.5 MHz) for deep penetration
• Multiple deployment options
13. Sandia Labs
Method: MAUS V PE Focus Probe
with Water Column
Sandia Labs
Method: Phased Array UT
25mm Water Box
14. Phased Array UT – Display and Deployment
Olympus 1.5Mhz,
42 element probe
GE Phased Array UT RotoArray
Sonatest RapidScan 2
15. Shearography
Sandwich core specimen
Thick laminate with bond lines
• Uses vacuum, heat, vibration to monitor the surface of a
structure for changes in the surface strain field/displacement
• Wide area interferometric imaging technique that is capable of
detecting micron-sized displacements
16. Thermography
• Thermography relies on the heat absorption characteristics of the
structure and changing IR images/heat transfer curves to indicate the
presence of defects
17. Terahertz Radiation
C-scans gated around the
returning time of flaw sets
DIA.
• 50 GHz – 4 THz frequency range with air-coupled, high penetration
• Flaws detected through frequency attenuation, phase shift and time of
flight
• Changes in THz signatures can indicate degradation of material as well
• Pitch-catch mode allows for single-sided inspections
Time gate range
?
740 X 400 @ 1.5mm
18. WINDIE Technology Assessment
Improved flaw detection:
Advanced NDI
Hybrid inspection approach - stack multiple
methods which address array of flaw types
(data fusion)
19. Wind Turbine Blade Flaw Detection Experiment
Wind Energy
Blade Reliability Collaborative (BRC)
Detection of Hidden
Flaws in Composite Wind Turbine
Blade Structure
Tom Rice, Dennis Roach, Stephen Neidigk,
Randy Duvall and Josh Paquette
Sandia National Labs
Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly
owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy's National Nuclear Security
Administration under contract DE-AC04-94AL85000
20. An Experiment to Assess Flaw Detection
Performance in Wind Turbine Blades (POD)
Purpose
• Generate industry-wide performance curves to quantify:
how well current inspection techniques are able to reliably find
flaws in wind turbine blades (industry baseline)
the degree of improvements possible through integrating more
advanced NDI techniques and procedures.
Expected Results - evaluate performance attributes
1) accuracy & sensitivity (hits, misses, false calls, sizing)
2) versatility, portability, complexity, inspection time (human factors)
3) produce guideline documents to improve inspections
4) introduce advanced NDI where warranted
21. Wind Blade NDI Probability of Detection Experiment
- Blind experiment: type, location and size of flaws are not know by inspector
- Statistically relevant flaw distribution – Probability of Detection (POD)
- Used to analytically determine the performance of NDI techniques – hits,
misses, false-calls, flaw sizing, human factors, procedures
Experimental Design Parameters
• Representative design and manufacturing
• Various parts of blade such as spar cap,
bonded joints, leading and trailing edge
• Statistically valid POD (number, size of flaws
and inspection area)
• Random flaw location
• Two days to perform experiment
• Deployment
Fabrication Considerations
• Realistic, random flaw locations
• Portable sample set
• Range of thickness
• Material types (fiberglass, carbon and various
adhesives)
• Who will manufacture
Designed to be applicable to various blade construction
Spar Caps & Shear Web Box Spar & Shear Webs
22. An Experiment to Assess Flaw Detection
Performance in Aircraft Composite Structures
737 Composite Horiz. Stabilizer
A380 Fuselage Section 19
Thickness Range:
12 – 64 plies
Simple Tapers
Complex tapers
Substructure Flaws
Curved Surfaces
Array of flaw types
24. Solid Laminate Flaw Detection
Experiment Implementation
PODs calculated for overall laminate,
by thickness family, by substructure
effects, by complex geometry effects,
by flaw types, etc.
25. POD Curves for 20-32 Ply
Solid Laminate Family
False Calls: Constant thickness = 0.8/inspector
Complex Geometry = 0.3/inspector
12 ft.2 inspection area
Overall:
POD[90/95] = 0.82” dia.
Individual and Cumulative Comparisons
Flaw Size (Diameter in Inches)
Probability of Detection
What improvements
will advanced NDI
provide?
Thermography
26. Wind Blade Probability of Detection Experiment
First design iteration of
POD experiment 2012
Review Committee
NREL
UpWind
DOE
Clipper
LM Wind Power
Gamesa
Molded Fiberglass
SNL
TPI Composites
GE – Global Research
Vestas
Sandia
Second iteration
incorporating review
committee’s suggestions
Specimens fabricated,
characterized and ready
(11 specimens)
Ensure representative blade construction and materials
27. Wind Blade Probability of Detection Experiment
What We Need
• Qualified Inspectors
Wind blade manufacturing companies
Blade service companies
Wind farms
NDI equipment development labs
• 2-2½ days of your time
How Does This Benefit You?
• Training perspective, inspections on representative blade
structure
• Inspector will receive feedback on how they performed
• PoD Value, smallest flaw size detectable with 95%
confidence
• Number of flaws detected
• Number of flaws missed
• Number of false calls, if any
• Flaw sizing
• Location and type of flaws missed
28. Wind POD Experiment is UNDERWAY
Completed fabrication of 11 POD Specimens
• 11 POD specimens with spar cap and shear web geometry
• Thickness ranges from 8 Plies (0.45” thick laminate, 0.85” thick with
adhesive bond line) to 32 Plies (1.80” thick laminate, 2.20” thick with
adhesive bond line)
• All panels painted with wind turbine blade paint (match inspection surface)
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Example Wind POD Curve - All Flaws - All Construction Types
POD Maximum Likelihood Estimate
POD Uncertainty - 95% Confindence Bound
0 0.5 1 1.5 2 2.5 3
Probability of Detection
Flaw Size (Diameter in Inches)
29. Wind Blade Probability of Detection Experiment
If you are interested in participating in
this experiment or have other
questions, please contact me using the
following:
Tom Rice
Phone: (505) 844-7738
Email: tmrice@sandia.gov
Editor's Notes
Dennis – You could animate the carbon samples to pop in