Upload
others
View
13
Download
0
Embed Size (px)
Citation preview
1
Nondestructive Evaluation:
USAF Perspectives on POD, Imaging,
and Characterization
E r i c L i n d g r e n
M a t e r i a l s a n d M a n u f a c t u r i n g D i r e c t o r a t e , A i r F o r c e R e s e a r c h L a b o r a t o r y
J u l y 1 3 , 2 0 1 9
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
2
Outline
• USAF Relevance
• POD Refresher• MIL HDBK 1823A
• Assumptions
• Method (Math)
• USAF Standardization• Why / How
• Image based NDI• Implementation
• POD Validation
• Characterization• Background
• Approach / Status
• MAPOD
• Summary
Photograph by author
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
3
US Air Force Relevance
• Nondestructive Inspection / Evaluation (NDI / NDE)
key input to risk management in USAF
• Aircraft Structural Integrity Program (ASIP)
• Propulsion Systems Integrity Program (PSIP)
• NDI / NDE used for new materials qualification
• Polymer matrix composites: 100% inspection
• Additive Manufacturing for critical parts
• Probability of Detection (POD): capability validation
• Facilitate with model-assisted POD (MAPOD)
• Image-based NDE / NDI methods in use
• Desired state: flaw / material characterization
• Augments value of data from NDE
• Accelerates decisions based on NDE
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
4
Example: USAF Approach for ASIP (MIL STD 1530Dc1)
• First flight to 1950s: design to static strength• Wright brothers
• Worked with finite use
• 1950s: fatigue in metals leads to safe life• DeHavilland Comet
• B-47 Stratofortress
• 1970s: flaws at manufacturing leads to durability and damage tolerance
• F-111
• F-5
• Commercial aviation uses variant of DADT
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
5
Durability and Damage Tolerance (DADT)
• Tolerate defects for some inspection-free period of service usage:
slow damage growth is USAF ASIP preferred approach
• NDE at critical locations based on DADT analysis to protect safety
How do we assess this value?
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
6
POD: A Key Factor in Risk Calculation
*www.afgrow.net
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
7
POD Refresher
Photograph by AuthorDistribution A, Unlimited Release. Case Number 88ABW-2019-3348
8
POD Introduction
Probability of Detection:
• Testing/evaluation procedures for assessing NDI capability
• Validation of NDI procedure when performed as intended
• Human factors affecting procedure
• Not human factors affecting inspector
• Objective: determine largest flaw missed during inspection
• Capability study, not a sensitivity study
• Improved sensitivity does not necessarily improve POD
• Statistics to determine value
• Must meet assumptions of math to be valid
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
9
MIL HDBK 1823A
• Current title “NDE System
Reliability Assessment”
• Personal perspective: “system
reliability” should be replaced by
“procedure validation”
• Provide guidance:
• How to set up a POD Study
• How to analyze the data
• Assumptions that MUST be met
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
10
Design of the experiment
• Includes all factors affecting detection
• Measurement system
• Probes, electrical noise, calibration
procedure, and many more
• Structure being assessed
• Variance: manufacturing, maintenance,
repair, modification, and use
• Defect being detected
• Geometry, morphology, and many more
Attributes of the data
• Independent samples
• Response increases with flaw size
• Linear models apply
• Noise is normal
• Constant variance
If data is binary, use hit/miss analysis
• Outcome: Probability that flaw of size
“x” will be called by inspector
Factors and Assumptions for Valid POD
≠ Find damage here
Sensors
Notch Plate
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
11
The Nature of Data
𝜀 is a
NORMAL
random
variable
(noise)
Parameters are
ESTIMATED,
not calculated
Ideal Data
Real POD
Curve
Ideal POD
Curve
Real Data
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
12
Confidence Bounds (Complex Math for Engineers)
Requires:
• Large data set
• Symmetric
distribution
Determines:
• Variance
• Standard Normal
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
13
POD: Cautions
Remember assumptions:
• Independent samples
• Response increases with flaw size
• Linear models apply
• Noise is normal
• Constant variance
Must have suitable data set to meet assumptions
Size
Response
POD
POD values can be calculated even if assumptions are
NOT met, but these values are wrongDistribution A, Unlimited Release. Case Number 88ABW-2019-3348
14
Additional Challenges for SHM POD
USAF definition: SHM is “a nondestructive inspection process or
technique that uses in-situ sensing devices to detect damage”
• Environment
• Temperature, loads, etc.
• Time variance in performance
• Includes durability
• Validation of Capability*
• Required for ASIP driven applications
• Qualification
• Initial documents available**
*Lindgren, et.al., “Demonstrating Capability Validation Protocol for in-situ Damage Detection,” presented at the 2011 ASIP Conference, San Antonio, TX
**Brausch and Steffes, “Demonstration, Qualification, and Airworthiness Certification of Structural Damage Sensing (SDS) Systems for
Air Force Applications, ” AFRL-RX-WP-TM-2013-0062
Installation:
• Qualification / certification
• Training
• Human factors
Sustainment:
• Training
• Maintenance / Calibration
• Durability
• Repair / modification
• Technical orders
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
15
Reminder: How USAF Uses POD for ASIP
When to Inspect
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
Calculation of Risk
16
USAF Standardization
Photograph by AuthorDistribution A, Unlimited Release. Case Number 88ABW-2019-3348
17
Need for Standardized Capability
When critical to ensuring safety (i.e. risk)
USAF structures challenge:
• Large number of aircraft
• High mobility of inspectors
• Increased inspection requirements with aging fleet
• Possible variance in capability as a function of weapon system
USAF Solution:
• Structures Bulletin EN-SB-08-012• Baseline common procedures/equipment
• Assumed capability for general classes of structural geometric features
Inventory: 1017*
*http://www.af.mil/AboutUs/FactSheets.aspx
Inventory: 428*
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
18
Foundation of EN-SB-08-012
Best Practices common inspection methods using standardized equipment
• See "Recommended Processes and Best Practices for NDI of Safety-of-
Flight Structures", AFRL-RX-WP-TR-2008-4373
• Includes inspection implementation and capability estimation
• Detection capability quantified for most standard practices
• Part specific procedures reference appropriate standard practice
Applicability
• Detection capability assumptions based on standard equipment, training,
and procedures across USAF programs
• Not applicable for specialized equipment or procedures: they should have separate
validation
• Supersedes Table XXXII of JSSG 2006: Joint Services Specification Guide –
Aircraft Structures
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
19
Content of EN-SB-08-012
• Introduction
• Members of USAF Capability Task Group
• Description of assumptions in training, procedures, and
assessments made in determining the capability values
• Additional steps to be followed if alternative values of capability
for specific inspections
• Description of applicability
• Detection of fatigue cracks in metallic structures to support damage
tolerance or durability analysis
• Field and Depot, performed by USAF personnel
• Not for contractor personnel or commercial procedures
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
20
Content of EN-SB-08-012 (cont)
• NDI capability values, a90/95 for calculating inspection intervals by
inspection method, including
• Method specific requirements, e.g. surface preparation
• Material type, e.g. aluminum, titanium, or steel
• Representative geometry, e.g. flat surface, radii, edges, bolt hole
• Eddy current probes, e.g. pencil probes, conformal probes
• Fluorescent Penetrant Inspection
• Whenever possible, the a90/95 values and POD curves determined via
guidance of MIL HDBK 1823A
• Alternative published values can be used
• Values can be adjusted when consensus that published values are not
representative
• Includes guidance for the use of:
• Ultrasound, Magneto-optical Imaging, Visual, Computed RadiographyDistribution A, Unlimited Release. Case Number 88ABW-2019-3348
21
Representative Case Study: Raised Head Fasteners*
• Inspection around raised head fasteners
• New probes conform to geometry of
fasteners
• Engineering / demonstration
• Validation via POD IAW MIL HDBK 1823A
• 51 inspectors, 3 USAF depots, 2200 fasteners,
44 cracks
• Impact -- detectable crack size decreased:
0.200” to 0.100” exposed crack length
*Forsyth et.al., available at: www.meetingdata.utcdayton.com/agenda/asip/2010/proceedings/presentations/P4213.pdf
Representative structure
Raised head fastener probes and kit
POD: pencil/raised head fastener probes
POD results:
Raised head
fastener
probes
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
22
Benefit to USAF*
Standardized:
• Assessment of inspection capability
• Assessment of methods to improve
capability
• Consensus for all USAF weapon
systems
• Focused efforts to improve capability
• Additional examples in Dr. Jones’
2015 ICAF paper
*K. Jones, J.C. Brausch, W.A. Fong, B.L. Harris, “Probing the Future,: Better F-16 Inspections using Conformal Eddy Current Inspection Tools,” Proceedings of the 28th International Conference of Aerospace Fatigue Symposium Helsinki, Finland, June 2015.
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
23
Image-based NDI
Photograph by AuthorDistribution A, Unlimited Release. Case Number 88ABW-2019-3348
24
Image-based NDI in USAF
Two general classes of image-based NDI
• Image direct from data capture system• Fluorescent Penetrant
• Radiography
• Magnetic Particle
• Thermography/ Shearography
• Visual
• Image from transposing other data• Ultrasonic B and C-scans
• Eddy current C-scans
• Computed tomography
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
25
Image Case Study*
Note: Requirement removed with
center wing replacement program
in the mid 2000’s
• Large area requirement
• ~8,800 fasteners in upper wing
• ~10,000 fasteners in lower wing
• Fasteners both raised and flush
• Fatigue cracks in both layers
• Skins and stringer flanges
• 0.070” corner cracks*E. A. Lindgren, et.al., “Validation and Deployment of Automated Ultrasonic Inspection of the C-130 Center Wing,” Proceedings of the Aircraft Structural Integrity Program Conference, Savannah, GA December 2003.
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
26
Approach: Portable Ultrasonic C-scan*
*E. A. Lindgren, et.al., “Validation and Deployment of Automated Ultrasonic Inspection of the C-130 Center Wing,” Proceedings of the Aircraft Structural Integrity Program Conference, Savannah, GA December 2003.
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
27
Criteria for Defect Detection*
• Elongated gate for first and
second layer indications
• Side lobes in amplitude C-scan
• Must have correct location
relative to fastener
• Time-of-flight gradient
• Shows reflections from crack as a
function of transducer position
• Initial analysis automated to
assist inspectors with large
data sets
Amplitude C-scan
Time-of-Flight C-scan
*E. A. Lindgren, et.al., “Validation and Deployment of Automated Ultrasonic Inspection of the C-130 Center Wing,” Proceedings of the Aircraft Structural Integrity Program Conference, Savannah, GA December 2003.
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
28
• 216 holes, 6 samples
• 36 per sample
• 432 inspection sites
• 3 different first layer thickness
• 0.15”, 0.25”, 0.35”
• Second layer constant
• Fasteners:
• Three samples: raised head
• Three samples: flush head
• 144 flaws: 24 over 36 holes
• 18 no, 12 one, and 6 two flaws
• 36 flaws per layer: near and far
on first layer, plus near and far
on second layer
• Flaw size distribution: uniform
log scale
• Four each of following (inches):
0.01, 0.013, 0.017, 0.023, 0.030,
0.039, 0.052, 0.068, and 0.090
Design of Experiment*
*E. A. Lindgren, et.al., “Validation and Deployment of Automated Ultrasonic Inspection of the C-130 Center Wing,” Proceedings of the Aircraft Structural Integrity Program Conference, Savannah, GA December 2003.
Set criteria and MIL HDBK 1823A methods apply
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
29
POD Execution and Results*
Execution
• 5 inspectors
• In Depot environment
• Upper / lower wing configuration
• Production equipment
Results
• 90/95 was 0.067” or smaller
• Exact number depended on location and fastener type
*E. A. Lindgren, et.al., “Validation and Deployment of Automated Ultrasonic Inspection of the C-130 Center Wing,” Proceedings of the Aircraft Structural Integrity Program Conference, Savannah, GA December 2003.
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
30
Image-based NDI: Lessons Learned
• Current methods for evaluating capability apply
• Select detection criteria
• Determine type of response: a vs. a-hat or hit/miss
• Leverage all available data
• e.g. use A-and/or B-scans for UT, impedance plane for ET
• Images can help detection processes
• Human eye very sensitive to change
• Caution when trying to automate image analysis
• Especially true when data direct from imaging system: relying on pixel
values to determine if defect is present
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
31
Characterization (MSA)
Photograph by AuthorDistribution A, Unlimited Release. Case Number 88ABW-2019-3348
32
Characterization: MSA
Materials State Awareness: Digitally-enabled Reliable Nondestructive Quantitative Materials / Damage Characterization Regardless of Scale
• National Academies Workshop: 2007 (Wood Hole, MA)
• Defense Materials, Manufacturing, and Infrastructure Workshop: 2014 (Washington, DC)
• Hosted by the National Academies
Motivation:
• Condition-based Maintenance
• Enhanced risk management
• New materials qualification (e.g. additive manufacturing)
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
33
Approaches for Characterization
Multiple thoughts on technical method:
• Image-based pixel counts
• Provides an initial approximation
• Advanced signal processing
• Used in some applications in power generation
• Inversion of NDE signal
• Typically ill-posed: factors affect NDE signal
not related to defect
• USAF focused on Model-based inversion
• Use data to models to assist in addressing
confounding factors
Representative fatigue crack
Representative corrosion (intergranular)
Representative composite impact damage
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
34
USAF Technical Approach to MSA
Augmented Materials Design, Processing, and Performance
Efficient and Effective ASIP/PSIP/MX Actions
Model-driven Quantitative Representation of Material/Damage State with Statistical Metrics
3D Representation and Validation
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.20
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
crack length (in)
PO
D
MAPOD
exp.
Signal Analysis and Uncertainty Quantification
NDE Damage / Materials
Characterization
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
35
Considerations for Characterization
• One-dimensional experiments
• Limits change to one factor
• Excellent to determine sensitivity to that one factor
• Historical focus of lots of talks at QNDE
• Example: residual stress measurements with eddy current
• Correlation between conductivity and residual stress, but
• Also correlation between conductivity and microstructure, loads,
temperature, etc.
• Characterization problems are typically multi-dimensional
• NDE modalities are sensitive to many factors
• Need to address confounding factors
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
36
AFRL Characterization Projects
Past
• Microstructure from surface measurements (eddy current and Rayleigh waves)
• Consider capability of SRAS from Univ. Nottingham
Present
• Eddy current for crack sizing: turbine engine components and evolving to bolt-hole
eddy current
• Bulk wave ultrasound for impact damage in polymer matrix composites
Future
• Volumetric microstructure
• Localized atomic changes in ceramic-based materials
Multiple talks at QNDE 2019, plus in previous years in the proceedings
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
37
Validation of Characterization
• Consider metrics of accuracy in length, depth, and width,
plus volumetric location in a part for risk management
• Analogy to a 3D POD
• Focus on only important parameters
• For fatigue cracks: length and depth
• For location, depends on geometry
• Active efforts within AFRL to address the process
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
38
Model-assisted POD (MAPOD)
Photograph by AuthorDistribution A, Unlimited Release. Case Number 88ABW-2019-3348
39
MAPOD
• Motivation: reduce time and cost of POD studies
• Approach:
• Transfer functions
• Full model assist
• Working group formed in 2004
• AFRL, NASA, and FAA
• Over 100 contributions
• Active for ~10 years
• Renewed interest to address POD of SHM
MAPOD
Logo was
not created
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
40
Unified Approach (Framework) for MAPOD
• Combines considerations
of transfer functions and
full-model assist
• Discussed in MIL HDBK
1823A
• Sample preparation and
flaw independence driving
renewed interest for SHM
• Presentation on Tuesday
afternoon with examples
Distribution A, Unlimited Release. Case Number 88ABW-2019-3348
41
Summary
POD, Imaging, and Characterization
• USAF Relevance
• POD Refresher
• USAF Capability Standardization
• Image based NDI
• Implementation
• POD Validation
• Characterization
• Background
• Approach / Status
• MAPOD
USAF Pay-off
• Safety of structures
• Integrated to realize ASIP/PSIP risk
(safety) metrics
• Availability of aircraft
• Extended inspection intervals
• Augmented management capability
• Reduced cost / time of inspection
• With NO compromise to safety
• New materials qualificationDistribution A, Unlimited Release. Case Number 88ABW-2019-3348
42
Discussion
Photograph by AuthorDistribution A, Unlimited Release. Case Number 88ABW-2019-3348