DOT/FAA/TC-15/63 A Quantitative Assessment of

Federal Aviation Administration

William J. Hughes Technical Center
Conventional and Advanced
Aviation Research Division
Atlantic City International Airport
New Jersey 08405
Nondestructive Inspection
Techniques for Detecting Flaws
in Composite Honeycomb
Aircraft Structures

December 2016

Final Report

This document is available to the U.S. public

through the National Technical Information
Services (NTIS), Springfield, Virginia 22161.

This document is also available from the

Federal Aviation Administration William J. Hughes
Technical Center at actlibrary.tc.faa.gov.

U.S. Department of Transportation

Federal Aviation Administration
 NOTICE

This document is disseminated under the sponsorship of the U.S.

Department of Transportation in the interest of information exchange. The
U.S. Government assumes no liability for the contents or use thereof. The
U.S. Government does not endorse products or manufacturers. Trade or
manufacturers’ names appear herein solely because they are considered
essential to the objective of this report. The findings and conclusions in
this report are those of the author(s) and do not necessarily represent the
views of the funding agency. This document does not constitute FAA
policy. Consult the FAA sponsoring organization listed on the Technical
Documentation page as to its use.

This report is available at the Federal Aviation Administration William J.

Hughes Technical Center’s Full-Text Technical Reports page:
actlibrary.tc.faa.gov in Adobe Acrobat portable document format (PDF).
 Technical Report Documentation Page
1. Report No. 2. Government Accession No. 3. Recipient's Catalog No.

DOT/FAA/TC-15/63
4. Title and Subtitle 5. Report Date

A QUANTITATIVE ASSESSMENT OF CONVENTIONAL AND ADVANCED December 2016

NONDESTRUCTIVE INSPECTION TECHNIQUES FOR DETECTING FLAWS
IN COMPOSITE HONEYCOMB AIRCRAFT STRUCTURES 6. Performing Organization Code

7. Author(s) 8. Performing Organization Report No.

Dennis P. Roach
9. Performing Organization Name and Address 10. Work Unit No. (TRAIS)

Sandia National Laboratories

FAA Airworthiness Assurance Center
Box 5800 MS-0615
Albuquerque, NM 87185

12. Sponsoring Agency Name and Address 13. Type of Report and Period Covered

FAA National Headquarters Final Report

950 L'Enfant Plaza N SW
950 L'Enfant Plaza
14. Sponsoring Agency Code
Washington, DC 20024
AIR-100
15. Supplementary Notes

The Federal Aviation Administration William J. Hughes Technical Center Aviation Research Division Technical Monitor was
David Westlund.
16. Abstract
The aircraft industry continues to increase its use of composite materials. This is most noteworthy in the area of principle
structural elements. The extreme damage tolerance and high strength-to-weight ratio of composites have motivated designers to
expand the role of fiberglass and carbon graphite in aircraft structures. This has placed greater emphasis on the development of
improved nondestructive inspection (NDI) methods that are more reliable and sensitive than conventional NDI. The FAA
Airworthiness Assurance NDI Validation Center (AANC) at Sandia Laboratories has been pursuing this goal via a host of studies
on inspection of composite structures. Through the FAA-AANC’s participation in the Commercial Aircraft Composite Repair
Committee Inspection Task Group, this team has been investigating the need for improved inspections of composite structures.
The majority of composite honeycomb structure inspections are performed visually and supplemented by tap test methods. Tap
testing, which uses a human-detected change in acoustic response to locate flaws, and more sophisticated NDI methods such as
ultrasonics or thermography have been applied to an increasing number of applications to detect voids, disbonds, and
delaminations in adhesively bonded composite aircraft parts. Low frequency bond testing and mechanical impedance analysis
tests are often used to inspect thicker laminates. A probability of detection experiment was completed to assess the performance
of both conventional and advanced NDI techniques. A series of composite honeycomb specimens with statistically relevant flaw
profiles was inspected using both human tap test equipment and new inspection techniques, which have recently been introduced
to automate and improve composite NDI. Industry-wide performance curves have been produced to establish: 1) how well current
inspection techniques are able to reliably find flaws in composite honeycomb structure, and 2) the degree of improvements
possible through the integration of more advanced NDI techniques and procedures. This study compared the results from a wide
array of NDI methods, and identified limitations and optimum applications for specific inspection methods while producing key
recommendations for improving the performance of NDI.
17. Key Words 18. Distribution Statement
Nondestructive inspection, honeycomb composites, solid This document is available to the U.S. public through the
laminate composites, probability of detection, disbonds, National Technical Information Service (NTIS), Springfield,
delaminations, impact, aircraft VA 22161. This document is also available from the Federal
Aviation Administration William J. Hughes Technical Center at
actlibrary.tc.faa.gov.
19. Security Classif. (of this report) 20. Security Classif. (of this page) 21. No. of Pages 22. Price
Unclassified Unclassified 350
Form DOT F 1700.7 (8-72) Reproduction of completed page authorized
 ACKNOWLEDGEMENTS

This program is sponsored by the Federal Aviation Administration (FAA) William J. Hughes Technical
Center under the direction of the technical monitor David Westlund. The approach used in this effort was
formulated in concert with the Commercial Aircraft Composite Repair Committee (CACRC) Inspection Task
Group, which consisted of the team members listed below, whose contributions are gratefully acknowledged.

Gerry Doetkott, Northwest Airlines

Alex Melton, Delta Air Lines
Richard Watkins, Delta Air Lines
Tom Dreher, Rolls Royce
Robert Stevens, United Airlines
Bruce Garbett, Airbus Industries
John Hewitt, Airbus Industries
Jim Hofer, Boeing
Jeff Kollgaard, Boeing
Glae McDonald, US Airways
Eric Bartoletti, American Airlines
Kirk Rackow, Sandia Labs AANC
Dennis Roach, Sandia Labs AANC

The author would like to recognize the data acquisition support provided by others at Sandia National
Laboratories AANC, including Phil Walkington, Tony Delong, Joe Dimambro, and Ciji Nelson. Special
thanks go to Kirk Rackow (retired Sandia National Laboratories) who played a critical role in the
development and implementation of this experiment. FAA oversight and extensive guidance for this effort
was provided by David Westlund and Dave Galella, FAA Project Managers, and Rusty Jones, FAA Senior
Technical Specialist in Nondestructive Inspection (NDI) and Composites. David Hsu and Dan Barnard of
Iowa State University provided many useful ideas on tap testing and appropriate performance testing. Thanks
also to David Caulkins and his colleagues in the Sandia National Laboratories composite shop for their
assistance in test specimen fabrication.

Advanced NDI tests were supported by Laser Technology Inc. (shearography), Honeywell (laser velocimetry
Structural Anomaly Mapping), Imperium Inc. (ultrasonic imaging Acoustocam), Iowa State University
(Computer-aided tap tester and air-coupled ultrasonics), Boeing (Mobile Automated Scanner in MIA and
Resonance mode), Thermal Wave Imaging (thermography), Sonatest (Rapidscan linear array ultrasonics),
Southern Research Institute (ultrasonic spectroscopy), Lockheed Martin (laser ultrasonics), Evisive
(microwave), GMA Industries (terahertz ultrasonic imaging), NDT Solutions Inc. (FlawInspecta phased array
ultrasonics), QUEST Integrated (induction thermography), and Digiray (digital x-ray laminography).

Finally, the author would like to thank the following airlines and maintenance facilities for participating in
this experiment: Delta Air Lines (four facilities), United Airlines (three facilities), American Airlines (two
facilities), Aloha Airlines, Alaska Airlines, US Airways, British Airways, KLM Airlines, Air France, FedEx
(two facilities), SR Technics, GKN Westland Aerospace, and FLS Aerospace Manchester.

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.

iii
 TABLE OF CONTENTS

Page

EXECUTIVE SUMMARY xxiv

1. INTRODUCTION AND BACKGROUND 1

1.1 Overview of Composite Honeycomb Flaw Detection Experiment 1

1.1.1 Description of CHE: 2

1.1.2 Results Obtained From CHE 3

1.2 Increasing Use of Composites in Aircraft Structures 4

1.3 Background on In-Service Inspection Needs for Composite Structures 9
1.4 Damage Tolerance Approach to Establish Inspection Intervals 9

1.4.1 Inspection intervals: 12

2. PURPOSE OF COMPOSITE HONEYCOMB FLAW DETECTION

EXPERIMENT 13

2.1 OEM Guidelines for Inspecting Composite Honeycomb Structures 19

3. CONVENTIONAL INSPECTION METHODS FOR COMPOSITE

HONEYCOMB STRUCTURE 22

3.1 Bond Testing 23

3.2 Manual Mechanical Tap Testing 25
3.3 Automated/Instrumented Mechanical Tap Testing 29
3.4 Low Frequency Bond Testing 32
3.5 High Frequency Bond Testing – Resonance Mode 33
3.6 Mechanical Impedance Analysis 34
3.7 Ultrasonic Inspection Methods 36

3.7.1 A-scan Mode: 36

3.7.2 B-scan Mode 38
3.7.3 C-scan Mode—Use of UT Scanning Technology 39

4. ADVANCED COMPOSITE HONEYCOMB INSPECTION TECHNIQUES 42

4.1 Instrumented and Automated Tap Test Devices 43

4.1.1 Woodpecker 46
4.1.2 Rapid Damage Detection Device 47
4.1.3 Computer-Aided Tap Test Device 48

iv
 4.2 Pulsed Thermography 53

4.2.1 Thermal Wave Imaging EchoTherm Thermography Inspection System 55

4.3 Line Scanning Thermography 61

4.3.1 MISTRAS Line Scan Thermography Inspection System 63

4.4 Lock-In Thermography 64

4.4.1 MoviTHERM Lock-In Thermography System 66

4.5 Shearography 67

4.5.1 Laser Technology Inc. LTI-5200 System 69

4.5.2 Dantec Dynamics Q-810 Laser Shearography System 72

4.6 Mobile Automated Scanner in Resonance and Mia Modes 74

4.7 Microwave 76
4.8 Structural Anomaly Mapping System 80
4.9 Phased Array and Linear Array Ultrasonics 84

4.9.1 Boeing MAUS FlawInspecta Linear Array UT System and the

Diagnostic Sonar FlawInspecta PA-UT System 87

4.10 Air-Coupled Ultrasonics 90

4.11 Laser Ultrasonics 96

4.11.1 LUS deployment 96

4.11.2 iPhoton LUS (iPLUS) 99

4.12 Laminography 101

5. COMPOSITE HONEYCOMB FLAW DETECTION EXPERIMENT DESIGN 106

5.1 Experiment Design Criteria 109

5.2 Experiment Test Specimen Design 110

6. COMPOSITE HONEYCOMB FLAW DETECTION EXPERIMENT

IMPLEMENTATION 123

7. RESULTS FROM COMPOSITE HONEYCOMB FLAW DETECTION

EXPERIMENT 134

8. CONCLUSIONS AND RECOMMENDATIONS 278

9. REFERENCES 284

v
APPENDIX A— EXPERIMENT BRIEFING AND INFORMATION PACKET

APPENDIX B—EXPERIMENT OBSERVATIONS

APPENDIX C—DISTRIBUTION LIST FOR DOT DOCUMENT

vi
 LIST OF FIGURES

Figure Page
1 Typical construction of composite sandwich structures 4
2 Use of composite structures on Airbus 320 series aircraft 5
3 Major composite structures on A380 aircraft 6
4 Summary of composite structures on Boeing 787 aircraft 6
5 Summary of composite structures on Cessna Citation III aircraft and conventional NDI
methods used to inspect them 7
6 Production of an all-composite fuselage section 7
7 Summary of advanced composite applications on A380 primary structures 8
8 Residual strength curve 11
9 Crack growth curve showing time available for fracture control 11
10 Probability of flaw detection versus flaw size 12
11 Effect of circumstances on probability of detection 13
12 Expansion in use of composite materials in aircraft construction 14
13 Sample sources of damage to composite structures 16
14 Sample damage from ground service vehicle impact 16
15 Sample damage from ground operations 17
16 Sample damage from impacts during flight 17
17 Sample damage from lightning strike 18
18 Sources of in-service damage to composite structures 18
19 Probability of impact energy as a function of take-off speed (based on runway debris
collected from four U.K. military air bases) 19
20 Effects of impact on composite structures 20
21 Example of external impact creating minor surface demarcation but significant
internal damage 20
22 Comparison between visible and backside damage (crushed core and backside fiber
fracture) in honeycomb structures 21
23 Tap test inspections of composite aircraft structure 22
24 Evolution of the tap test method in nondestructive testing 25
25 Mechanical tap hammer specified by Airbus 26
26 Mechanical tap hammer specified by Boeing 27

vii
27 Samples of homemade tappers used in the field including modification to Woodpecker
device (addition of wood piece) to ensure that automated tap is perpendicular to the
surface being inspected 27
28 General test pattern used for mechanical impact inspections 28
29 Detailed view of test pattern used for mechanical impact inspections 28
30 Sample composite honeycomb damage to be detected with mechanical tap test 29
31 Comparison of the surface stiffness deduced from the tap test with the stiffness
measured in a static load test 30
32 Force versus time history for various tapping forces in a good region and a damaged
region on a composite honeycomb panel 31
33 Instrumented tap test devices: DTH and Woodpecker 32
34 S-9 Sondicator low frequency bond test device 33
35 Olympus BondMaster device operating in HFBT/resonance test mode 34
36 Mechanical impedance analysis inspection with V-95 device 35
37 Schematic of pulse-echo ultrasonic inspection and A-scan signal showing reflection
of UT waves at assorted interfaces 38
38 A-scan waveform from bonded and disbonded portions of a composite structure 39
39 Schematic of C-scan setup for pulse-echo ultrasonic inspection 40
40 MAUS automated ultrasonic scanning system 41
41 Sample ultrasonic signals generated from: structure without damage and structure
with damage 41
42 Sample C-scan produced by an automated ultrasonic scanning device 42
43 Tapping head 44
44 Determination of the impact duration, τ, from an accelerometer output and a
photograph of an oscilloscope trace with a τ = 968 µs 44
45 Sample force pulses 45
46 Fourier transform of the impulse 45
3
47 Three examples of instrumented tap test devices: Woodpecker, RD , and CATT 46
48 Woodpecker automated tap test device 47
3
49 RD automated tap test device 48
50 Manual deployment of CATT device 48
51 Manual tap probe and hand-deployed cart for semi-automated scanning using CATT
system 49
52 CATT system being deployed by Iowa State University developers on honeycomb
flaw detection experiment test panels 49

viii
53 Close-up of CATT tapper and transfer of data from computer to grading sheets 49
54 CATT C-scan tap test image of surface contact stiffness of a composite honeycomb
sandwich panel containing engineered defects 51
55 Sample CATT C-scan of a 6-ply fiberglass test panel with flaw template overlay
showing hits and misses 51
56 Two-dimensional and three-dimensional C-scan images produced from a CATT
inspection of a B767 elevator 52
57 Assembled CATT tap test image of the inner and outer surfaces of a composite
rudder recovered from an aircraft accident 52
58 Principle of active pulsed thermography 54
59 Laboratory thermal wave imaging system inspecting composite flaw detection panels
and a portable field system inspecting an aircraft fuselage 54
60 Comparison of IR cameras for thermography inspection 55
61 TWI system equipment and inspection of aircraft 56
62 Sample thermography image showing a disbond in an aluminum fuselage tear strap
structure 57
63 FLIR A40 uncooled camera inspecting the honeycomb test panels and a sample IR
image from a fiberglass panel 57
64 Thermography image produced from inspection of composite laminate panel with
flaw profile as shown in drawing on the right 58
65 Sample thermography images showing damage in composite structures with
comparisons from three different IR cameras on 3-ply carbon honeycomb panel (all
flaws identified by all cameras) 59
66 Sequence of thermal wave images from DC-9 composite doubler inspection 60
67 Examples of thermal images generated after scanning a composite structure using the
LST technique 61
68 Setup of LST where IR camera and heat source move in tandem through the surface
to be inspected 62
69 Panel showing the observation gate selection with respect to the heat deposition
location 62
70 MISTRAS LST system—crawler used on composites 63
71 MISTRAS LST system—small area scanner 64
72 Results produced by MISTRAS LST system on a 32-ply panel with substructure
elements 64
73 Equipment setup used for typical lock-in thermography inspection 66
74 MoviTHERM lock-in thermography with halogen heat lamp being used as the
excitation source 67

ix
75 Results produced by lock-in thermography on a 32-ply panel with substructure
elements 67
76 LTI-5200 portable shearography system with camera on test specimen 68
77 Basic principles of shearography 69
78 Composite rudder inspection using LTI-5200 portable vacuum shearography system 70
79 Schematic of shearography inspection for near-side and far-side disbond detection 70
80 Near-side and far-side disbonds detected by LTI-5200 shearography system in A310
composite rudder 71
81 Close-up view of LTI-5200 shearography image showing flaws in a composite
honeycomb structure and a sample shearography result for 6-ply fiberglass panel
showing near-side and far-side flaw imaging 71
82 LTI-5200 shearographic inspection image of a scarfed repair to a honeycomb structure
with anomaly indications in the repair plies 72
83 Shearography image produced from inspection of composite laminate panel (0.11″
thick skin) with flaw profile as shown in drawing 72
84 Q-810 laser shearography system 73
85 Test specimen and Q-810 shearography image of wrinkles in a composite
laminate 73
86 MAUS inspections on aircraft fuselage section 74
87 MAUS system operating on an aircraft fuselage with sample C-scan image produced
from resonance inspection on a 6-ply carbon skin honeycomb panel 75
88 MAUS MIA inspections on honeycomb panel with sample C-scan image produced 76
89 X and Y scans of a composite honeycomb panel produced by the MAUS inspection
system deployed in resonance mode 76
90 Configuration of microwave inspection system on a laboratory scan table 77
91 Basic equipment setup for microwave inspection 78
92 Microwave C-scan of a fiberglass boom structure showing the presence of interply
delaminations 79
93 Microwave inspection results for fiberglass and carbon skin honeycomb panels with
engineered flaws (skins bonded to Nomex honeycomb) 79
94 Sample microwave inspection results for 3-ply and 12-ply fiberglass panels with
delamination, disbonds, potted core, and core splice 80
95 Schematic showing the principle of the SAM inspection method 81
96 SAM laser Doppler velocimetry camera and acoustic excitation device; composite test
specimen in the background 82
97 SAM system inspecting composite honeycomb panels and transfer of scan data onto
experiment grading sheets 82

x
98 Frequency response plots produced during the application of the SAM inspection
system on an engine cowling 83
99 SAM system inspecting an engine cowling and sample image showing flaws 83
100 Sample image produced by SAM system showing flaws in a fiberglass composite
honeycomb structure 84
101 PA-UT deployed in rolling wheel mechanism and contained in a single probe housing 85
102 Schematic showing the operation of an ultrasonic array that allows for the generation
and acquisition of multiple UT signals 85
103 Olympus OmniScan device with a 16:128 phased array transducer 86
104 Amplitude and time of flight data produced by OmniScan inspection of composite
laminate aircraft panel with flaw profile as shown 86
105 C-Scan images produced by OmniScan PA-UT inspection of 20-ply composite
laminate feedback panel with the flaw profile as shown 87
106 Diagnostic sonar FlawInspecta PA-UT inspection system 88
107 FlawInspecta linear array UT system deployed on MAUS V scanner platform—linear
array UT probe includes a delay line shoe 88
108 Composite honeycomb reference standard and sample FlawInspecta results—3-ply
carbon skin with 1″ thick core 89
109 Sample result from FlawInspecta PA-UT system on 6-ply carbon specimen 89
110 C-scan images produced by FlawInspecta MAUS V linear array UT system on a
32-ply composite laminate feedback panel with the flaw profile as shown 90
111 Schematic of air-coupled ultrasonic inspection of panel in through-transmission mode 92
112 Equipment setup for air-coupled ultrasonic inspections of composite honeycomb
panels shown in background 92
113 AC-UT applied to a composite panel test specimen in both through-transmission mode
and one-sided pitch/catch mode 93
114 AC-UT TTU test setup on fiberglass skin and foam core test specimen 94
115 AC-UT TTU C-scan image produced by 120 kHz inspection of composite foam core
test specimen 94
116 AC-UT TTU C-scan showing disbonds in skin-to-core bondline of a composite
honeycomb panel 95
117 Sample C-scan data produced by AC-UT method on composite honeycomb test
specimen 95
118 Schematic of laser UT system operation 96
119 Comparison of conventional and laser UT interrogation of components 98
120 Schematic of laser UT method and deployment in gantry system and rail system 98

xi
121 Inspection of a part using the iPLUS scan head and articulating robot 99
122 iPLUS laser-ultrasonic scan of a 16-ply composite laminate with impact damage 100
123 iPLUS laser-ultrasonic inspection of 0.111″ thick composite laminate test specimen;
photo on right shows the layout of the engineered flaws 100
124 iPLUS laser-ultrasonic image of a composite part containing an inclusion as
highlighted 101
125 Sample results from laser UT inspection of 3-ply fiberglass panel 101
126 Schematic of the reverse geometry laminography concept 102
127 Use of array of detectors and scanning beam to produce images of layers within the
part being inspected 103
128 Images highlighting the sensitivity of laminography and ability to produce slices for
improved detail recognition 103
129 Comparison of normal X-ray with laminography images showing ability of
laminography to detect damage in a 4″ thick Lucite part 104
130 Laminography slices of machined honeycomb core that simulate a disbond in the
fiberglass skin-to-core bond line 105
131 Laminography slices of interply delaminations in a 3-ply fiberglass honeycomb
panel 105
132 Laminography test setup and sample image showing all flaws detected in 3-ply
fiberglass panel 106
133 Test specimens used in composite honeycomb flaw detection experiment 107
134 Schematic of composite honeycomb NDI reference standards showing construction
and flaw profiles 108
135 Cross-section of honeycomb reference standard design 108
136 Impactors and impactor drop-table apparatus used to produce damaged area on test
panels 111
137 Sample results from calibration shots used to determine energy levels needed to
produce damage in composite honeycomb panels without producing surface
demarcations 112
138 Pyramid pattern matrix crack produced by impact of composite skin 112
139 Cross-section of sample impact damage in composite honeycomb experiment panels
showing crushed core but no skin deformation 112
140 Engineering drawing to evaluate honeycomb reference standard design and
fabrication 116
141 Construction of pillow insert for delamination flaws 117
142 Process for creating potted honeycomb core areas 118

xii
143 TTU inspection of honeycomb panel—assessment of methods to engineer repeatable
flaws 119
144 TTU C-scan showing signal attenuation produced by inserts; tests on 6-ply fiberglass
laminate 119
145 TTU C-scan showing signal attenuation produced by inserts; tests on 6-ply fiberglass
laminate 120
146 Specimen characterization—TTU C-scan of honeycomb test panel used to verify
fabrication and location of engineered flaws 121
147 Experiment instructions being provided to supplement the written experimenter
briefing and information packet 124
148 During inspections, each panel was supported around its perimeter by a foam frame
to provide uniform boundary conditions 126
149 Samples of homemade tappers used in the field and modification to Woodpecker
device to ensure that automated tap is perpendicular to the surface 126
150 Sample tap testing at maintenance depots and alternative tappers identified in study 127
151 Airline and aircraft maintenance depot participants in composite honeycomb flaw
detection experiment 128
152 Advanced NDI companies that participated in composite honeycomb flaw detection
experiment 128
153 Sample implementation of experiment in Alaska Airlines’ hangar environment 129
154 CHE being implemented at American Airlines’ maintenance facility 129
155 Typical experiment setup with separate inspector— Airbus facility hosting multiple
airlines 130
156 Inspectors completing inspections and marking flaw detections on the test
specimens—clockwise from upper right: Airbus tap hammer, Boeing tap hammer,
DTH, LFBT, Woodpecker, and MIA 131
157 Honeycomb panel with sample grid markings used by inspector during inspection of
panel along with flaw markings within the grid 132
158 Schematic showing the grading categories comparing experimenter flaw calls with
actual flaw information 133
159 Probability of detection curves showing the performance of multiple NDI devices
for a single type of test specimen: 6-ply carbon 137
160 Probability of detection curves for flaw detection performance of a single device
(Woodpecker) over the range of test specimen types 137
161 Probability of detection curves quantifying inspection improvements possible
through the application of advanced NDI techniques 140
162 Sample effect of inspection time on flaw detection performance (example for
Boeing tap hammer on 3-ply fiberglass) 141

xiii
163 Individual and cumulative PoD curve comparisons along with tabulated values for
the Airbus tap hammer deployed on 3-ply carbon test specimen set 145
164 Individual and cumulative PoD curve comparisons along with tabulated values for
the Airbus tap hammer deployed on 3-ply fiberglass test specimen set 146
165 Individual and cumulative PoD curve comparisons along with tabulated values for
the Airbus tap hammer deployed on 6-ply carbon test specimen set 147
166 Individual and cumulative PoD curve comparisons along with tabulated values for
the Airbus tap hammer deployed on 6-ply fiberglass test specimen set 148
167 Individual and cumulative PoD curve comparisons along with tabulated values for
the Airbus tap hammer deployed on 9-ply carbon test specimen set 149
168 Individual and cumulative PoD curve comparisons along with tabulated values for
the Airbus tap hammer deployed on 9-ply fiberglass test specimen set 150
169 Individual and cumulative PoD curve comparisons along with tabulated values for
the Boeing tap hammer deployed on 3-ply carbon test specimen set 151
170 Individual and cumulative PoD curve comparisons along with tabulated values for
the Boeing tap hammer deployed on 3-ply fiberglass test specimen set 152
171 Individual and cumulative PoD curve comparisons along with tabulated values for
the Boeing tap hammer deployed on 6-ply carbon test specimen set 153
172 Individual and cumulative PoD curve comparisons along with tabulated values for
the Boeing tap hammer deployed on 6-ply fiberglass test specimen set 154
173 Individual and cumulative PoD curve comparisons along with tabulated values for
the Boeing tap hammer deployed on 9-ply carbon test specimen set 155
174 Individual and cumulative PoD curve comparisons along with tabulated values for
the Boeing tap hammer deployed on 9-ply fiberglass test specimen set 156
175 Individual and cumulative PoD curve comparisons along with tabulated values for
LFBT deployed on 3-ply carbon test specimen set 157
176 Individual and cumulative PoD curve comparisons along with tabulated values for
LFBT deployed on 3-ply fiberglass test specimen set 158
177 Individual and cumulative PoD curve comparisons along with tabulated values for
LFBT deployed on 6-ply carbon test specimen set 159
178 Individual and cumulative PoD curve comparisons along with tabulated values for
LFBT deployed on 6-ply fiberglass test specimen set 160
179 Individual and cumulative PoD curve comparisons along with tabulated values for
LFBT deployed on 9-ply carbon test specimen set 161
180 Individual and cumulative PoD curve comparisons along with tabulated values for
LFBT deployed on 9-ply fiberglass test specimen set 162
181 Individual and cumulative PoD curve comparisons along with tabulated values for
MIA deployed on 3-ply carbon test specimen set 163

xiv
182 Individual and cumulative PoD curve comparisons along with tabulated values for
MIA deployed on 3-ply fiberglass test specimen set 164
183 Individual and cumulative PoD curve comparisons along with tabulated values for
MIA deployed on 6-ply carbon test specimen set 165
184 Individual and cumulative PoD curve comparisons along with tabulated values for
MIA deployed on 6-ply fiberglass test specimen set 166
185 Individual and cumulative PoD curve comparisons along with tabulated values for
MIA deployed on 9-ply carbon test pecimen set 167
186 Individual and cumulative PoD curve comparisons along with tabulated values for
MIA deployed on 9-ply fiberglass test specimen set 168
187 Individual and cumulative PoD curve comparisons along with tabulated values for
WichiTech DTH deployed on 3-ply carbon test specimen set 169
188 Individual and cumulative PoD curve comparisons along with tabulated values for
WichiTech DTH deployed on 3-ply fiberglass test specimen set 170
189 Individual and cumulative PoD curve comparisons along with tabulated values for
WichiTech DTH deployed on 6-ply carbon test specimen set 171
190 Individual and cumulative PoD curve comparisons along with tabulated values for
WichiTech DTH deployed on 6-ply fiberglass test specimen set 172
191 Individual and cumulative PoD curve comparisons along with tabulated values for
WichiTech DTH deployed on 9-ply carbon test specimen set 173
192 Individual and cumulative PoD curve comparisons along with tabulated values for
WichiTech DTH deployed on 9-ply fiberglass test specimen set 174
193 Individual and cumulative PoD curve comparisons along with tabulated values for
Woodpecker deployed on 3-ply carbon test specimen set 175
194 Individual and cumulative PoD curve comparisons along with tabulated values for
Woodpecker deployed on 3-ply fiberglass test specimen set 176
195 Individual and cumulative PoD curve comparisons along with tabulated values for
Woodpecker deployed on 6-ply carbon test specimen set 177
196 Individual and cumulative PoD curve comparisons along with tabulated values for
Woodpecker deployed on 6-ply fiberglass test specimen set 178
197 Individual and cumulative PoD curve comparisons along with tabulated values for
Woodpecker deployed on 9-ply carbon test specimen set 179
198 Individual and cumulative PoD curve comparisons along with tabulated values for
Woodpecker deployed on 9-ply fiberglass test specimen set 180
199 Comparison showing level of PoD improvement when single performance outlier is
removed from the calculation—Airbus tap hammer, 6-ply fiberglass 181
200 Comparison showing level of PoD improvement when performance outliers are
removed from the calculation—Airbus tap hammer, 9-ply fiberglass 181

xv
201 Comparison showing level of PoD improvement when single performance outlier is
removed from the calculation—LFBT, 6-ply fiberglass 182
202 Comparison showing level of PoD improvement when single performance outlier is
removed from the calculation—LFBT, 9-ply fiberglass 182
203 Cumulative PoD curve with 95% confidence bound— Airbus tap hammer, 3-ply
carbon 183
204 Cumulative PoD curve with 95% confidence bound— Airbus tap hammer, 3-ply
fiberglass 183
205 Cumulative PoD curve with 95% confidence bound— Airbus tap hammer, 6-ply
carbon 184
206 Cumulative PoD curve with 95% confidence bound— Airbus tap hammer, 6-ply
fiberglass 184
207 Cumulative PoD curve with 95% confidence bound— Airbus tap hammer, 9-ply
carbon 185
208 Cumulative PoD curve with 95% confidence bound— Airbus tap hammer, 9-ply
fiberglass 185
209 Cumulative PoD curve with 95% confidence bound— Boeing tap hammer, 3-ply
carbon 186
210 Cumulative PoD curve with 95% confidence bound— Boeing tap hammer, 3-ply
Fiberglass 186
211 Cumulative PoD curve with 95% confidence bound— Boeing tap hammer,6-ply
carbon 187
212 Cumulative PoD curve with 95% confidence bound— Boeing tap hammer, 6-ply
fiberglass 187
213 Cumulative PoD curve with 95% confidence bound— Boeing tap hammer,9-ply
carbon 188
214 Cumulative PoD curve with 95% confidence bound— Boeing tap hammer, 9-ply
fiberglass 188
215 Cumulative PoD curve with 95% confidence bound—LFBT, 3-ply carbon 189
216 Cumulative PoD curve with 95% confidence bound—LFBT, 3-ply fiberglass 189
217 Cumulative PoD curve with 95% confidence bound—LFBT, 6-ply carbon 190
218 Cumulative PoD curve with 95% confidence bound—LFBT, 6-ply fiberglass 190
219 Cumulative PoD curve with 95% confidence bound—LFBT, 9-ply carbon 191
220 Cumulative PoD curve with 95% confidence bound—LFBT, 9-ply fiberglass 191
221 Cumulative PoD curve with 95% confidence bound—MIA, 3-ply carbon 192
222 Cumulative PoD curve with 95% confidence bound—MIA, 3-ply fiberglass 192
223 Cumulative PoD curve with 95% confidence bound—MIA, 6-ply carbon 193

xvi
224 Cumulative PoD curve with 95% confidence bound—MIA, 6-ply fiberglass 193
225 Cumulative PoD curve with 95% confidence bound—MIA, 9-ply carbon 194
226 Cumulative PoD curve with 95% confidence bound—MIA, 9-ply fiberglass 194
227 Cumulative PoD curve with 95% confidence bound— WichiTech DTH, 3-ply
carbon 195
228 Cumulative PoD curve with 95% confidence bound— WichiTech DTH, 3-ply
fiberglass 195
229 Cumulative PoD curve with 95% confidence bound— WichiTech DTH, 6-ply
carbon 196
230 Cumulative PoD curve with 95% confidence bound— WichiTech DTH, 6-ply
fiberglass 196
231 Cumulative PoD curve with 95% confidence bound— WichiTech DTH, 9-ply
carbon 197
232 Cumulative PoD curve with 95% confidence bound— WichiTech DTH, 9-ply
fiberglass 197
233 Cumulative PoD curve with 95% confidence bound— Woodpecker, 3-ply carbon 198
234 Cumulative PoD curve with 95% confidence bound— Woodpecker, 3-ply fiberglass 198
235 Cumulative PoD curve with 95% confidence bound— Woodpecker, 6-ply carbon 199
236 Cumulative PoD curve with 95% confidence bound— Woodpecker, 6-ply fiberglass 199
237 Cumulative PoD curve with 95% confidence bound— Woodpecker, 9-ply carbon 200
238 Cumulative PoD curve with 95% confidence bound— Woodpecker, 9-ply fiberglass 200
239 Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 3-ply carbon (Tier 1 results) 201
240 Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 3-ply carbon (Tier 2 results) 202
241 Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 3-ply fiberglass (Tier 1 results) 203
242 Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 3-ply fiberglass (Tier 2 results) 204
243 Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 6-ply carbon (Tier 1 results) 205
244 Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 6-ply carbon (Tier 2 results) 206
245 Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 6-ply fiberglass (Tier 1 results) 207
246 Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 6-ply fiberglass (Tier 2 results) 208

xvii
247 Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 9-ply carbon (Tier 1 results) 209
248 Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 9-ply carbon (Tier 2 results) 210
249 Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 9-ply fiberglass (Tier 1 results) 211
250 Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 9-ply fiberglass (Tier 2 results) 212
251 Cumulative PoD curve comparisons along with tabulated values of all conventional
NDI devices on 3-ply carbon (average results for all inspectors) 213
252 Cumulative PoD curve comparisons along with tabulated values of all conventional
NDI devices on 3-ply fiberglass (average results for all inspectors) 214
253 Cumulative PoD curve comparisons along with tabulated values of all conventional
NDI devices on 6-ply carbon (average results for all inspectors) 215
254 Cumulative PoD curve comparisons along with tabulated values of all conventional
NDI devices on 6-ply fiberglass (average results for all inspectors) 216
255 Cumulative PoD curve comparisons along with tabulated values of all conventional
NDI devices on 9-ply carbon (average results for all inspectors) 217
256 Cumulative PoD curve comparisons along with tabulated values of all conventional
NDI devices on 9-ply fiberglass (average results for all inspectors) 218
257 Comparison of all mechanical tap test devices on 3-ply carbon 219
258 Comparison of all mechanical tap test devices on 3-ply fiberglass 219
259 Comparison of all mechanical tap test devices on 6-ply carbon 220
260 Comparison of all mechanical tap test devices on 6-ply fiberglass 220
261 Comparison of all mechanical tap test devices on 9-ply carbon 221
262 Comparison of all mechanical tap test devices on 9-ply fiberglass 221
263 Cumulative PoD curves for Airbus tap hammer along with tabulated values on all
panel types (average results for all inspectors) 222
264 Cumulative PoD curves for Boeing tap hammer along with tabulated values on all
panel types (average results for all inspectors) 223
265 Cumulative PoD curves for LFBT along with tabulated values on all panel types
(average results for all inspectors) 224
266 Cumulative PoD curves for MIA along with tabulated values on all panel types
(average results for all inspectors) 225
267 Cumulative PoD curves for WichiTech DTH along with tabulated values on all
panel types (average results for all inspectors) 226

xviii
268 Cumulative PoD curves for Woodpecker along with tabulated values on all panel
types (average results for all inspectors) 227
269 PoD curve comparisons for CATT along with tabulated values on all panel types 228
270 PoD curve comparisons for MAUS Resonance 1 inspections along with tabulated
values on all panel types (results shown for resonance experimenter #1 only) 229
271 PoD curve comparisons for SAM inspections along with tabulated values on all
panel types 230
272 PoD curve comparisons for shearography inspections along with tabulated values on
all panel types 231
273 PoD curve comparisons for thermography inspections along with tabulated values on
all panel types 232
274 PoD curve comparisons for MAUS IV MIA inspections along with tabulated values
on all panel types 233
275 PoD curve comparisons for AC-UT inspections (TTU mode) along with tabulated
values on all panel types 234
276 PoD curve comparisons for laminography inspections along with tabulated values on
all panel types 235
277 PoD curve comparisons for microwave inspections along with tabulated values on
all panel types 236
278 PoD curve comparisons for FlawInspecta UT array along with tabulated values on
all panel types 237
279 PoD curve comparisons for laminography showing the improvement when impact
and edge (inspection image overlap) flaws are removed from PoD calculation on
6-ply carbon 238
280 PoD curve comparisons for laminography showing the improvement when impact
and edge (inspection image overlap) flaws are removed from PoD calculation on
6-ply fiberglass 238
281 PoD curve comparisons for laminography showing the improvement when impact and
edge (inspection image overlap) flaws are removed from PoD calculation on 9-ply
carbon 239
282 PoD curve comparisons for laminography showing the improvement when impact
and edge (inspection image overlap) flaws are removed from POD calculation on
9-ply fiberglass 239
283 PoD curve comparisons along with tabulated values for laser-ultrasonic inspections
with exception that includes removing flaws located in center area of specimen from
PoD calculations on 3-ply fiberglass 240
284 PoD curve comparisons along with tabulated values for advanced inspection methods
and best performing conventional method on 3-ply carbon 241

xix
285 PoD curve comparisons along with tabulated values for advanced inspection methods
and best performing conventional method on 3-ply fiberglass 242
286 PoD curve comparisons along with tabulated values for advanced inspection methods
and best performing conventional method on 6-ply carbon 243
287 PoD curve comparisons along with tabulated values for advanced inspection methods
and best performing conventional method on 6-ply fiberglass 244
288 PoD curve comparisons along with tabulated values for advanced inspection methods
and best performing conventional method on 9-ply carbon 245
289 PoD curve comparisons along with tabulated values for advanced inspection methods
and best performing conventional method on 9-ply fiberglass 246
290 Scatter diagram showing effect of inspection time on PoD values for 3-ply panels 257
291 Scatter diagram showing effect of inspection time on PoD values for 6-ply panels 258
292 Scatter diagram showing effect of inspection time on PoD values for 9-ply panels 259
293 Scatter diagram showing effect of inspection time on false calls for 3-ply panels 260
294 Scatter diagram showing effect of inspection time on false calls for 6-ply panels 261
295 Scatter diagram showing effect of inspection time on false calls for 9-ply panels 262
296 Scatter diagram showing the effect of the number of false calls on PoD values for
3-ply panels 263
297 Scatter diagram showing the effect of the number of false calls on PoD values for
6-ply panels 264
298 Scatter diagram showing the effect of the number of false calls on PoD values for
9-ply panels 265

xx
 LIST OF TABLES

Table Page
1. Inspection methods applied to the composite honeycomb flaw-detection experiment 24
2. Digital tap test results from Mitsui Woodpecker comparing flaw detection in Nomex
and fiberglass cores 113
3. Digital tap test results from Mitsui Woodpecker showing relative stiffness values for
different cores 114
4. Test specimen matrix for composite honeycomb flaw-detection study 115
5. Sample S/N data from various flaws demonstrating the viability of each flaw for use
in the composite honeycomb flaw detection experiment 122
6. Inspectors’ NDI experience level for conventional methods 135
7. Advanced NDI methods deployed on composite honeycomb flaw-detection
experiment 135
8. Summary of overall flaw detection for 6-ply fiberglass construction — PoD levels
along with false calls and ability to correctly size flaws 139
9. Inspection summary—description of all figures and tables showing the variables
captured in the results 143
10. Average inspection times per 18″ x 18″ panel and max/min times for 3-ply panels
—conventional NDI methods 247
11. Average inspection times per 18″ x 18″ panel and max/min times for 6-ply panels
—conventional NDI methods 248
12. Average inspection times per 18″ x 18″ panel and max/min times for 9-ply panels
—conventional NDI methods 248
13. Average inspection times per 18″ x 18″ panel and max/min times for all panels—
advanced NDI methods (times shown in minutes) 249
14. Summary of individual and average false calls for all inspectors and false call
averages per inspection method and panel type—conventional NDI 250
15. Average number of false calls produced by conventional NDI devices on carbon and
fiberglass specimens 252
16. False calls generated by advanced inspection techniques (arranged by panel type) 252
17. Backside flaw detection for all inspectors and average backside flaws detected per
inspection method and panel type—conventional NDI 253
18. Summary of average backside flaw detection for conventional inspection techniques
listed by panel type 255
19. Backside flaw detection for advanced inspection techniques by panel type (note:
AC-UT was executed in through-transmission mode) 256

xxi
20. Summary of all NDI performance for 3-ply carbon and fiberglass test specimens
—flaw sizing, overall 90% PoD levels, and false calls 266
21. Summary of all NDI performance for 6-ply carbon and fiberglass test specimens
—flaw sizing, overall 90% PoD levels, and false calls 268
22. Summary of all NDI performance for 9-ply carbon and fiberglass test specimens
—flaw sizing, overall 90% PoD levels, and false calls 270
23. Summary of all NDI flaw-detection capability for 3-ply carbon and fiberglass test
specimens—overall flaw detection, highest detection (by individual inspector),
lowest detection (by individual inspector), overall 90% PoD levels, and false calls 272
24. Summary of all NDI flaw-detection capability for 6-ply carbon and fiberglass test
specimens—overall flaw detection, highest detection (by individual inspector),
lowest detection (by individual inspector), overall 90% PoD levels, and false calls 274
25. Summary of all NDI flaw-detection capability for 9-ply carbon and fiberglass test
specimens—overall flaw detection, highest detection (by individual inspector),
lowest detection (by individual inspector), overall 90%PoD levels, and false calls 276

xxii
 LIST OF ACRONYMS

AANC Airworthiness Assurance NDI Validation Center

AC Advisory circular
ARP Aerospace recommended practice
AC-UT Air-coupled ultrasonic testing
CACRC-ITG Commercial Aircraft Composite Repair Committee Inspection Task Group
CATT Computer-aided tap tester
CCD Charge-coupled device
CHE Composite Honeycomb Flaw Detection Experiment
DTA Damage tolerance analysis
DTH Digital tap hammer
FWC Full waveform capture
HFBT High frequency bond testing
iPLUS iPhoton laser ultrasonic system™
IR Infrared
LFBT Low frequency bond testing
LST Line Scan Thermography™
LUS Laser ultrasonics
MAUS Mobile Automated Scanner
MIA Mechanical impedance analysis
MRO Maintenance and repair organizations
NDI Nondestructive inspection
NDT Nondestructive test
NDTM Nondestructive testing manuals
OEM Original equipment manufacturer
PA-UT Phased array ultrasonic
PE-UT Pulse-echo ultrasonic
PoD Probability of detection
QMI Quality Material Inspection
RD3 Rapid Damage Detection Device
RF Radio frequency
RGX Reverse Geometry X-ray®
S/N Signal-to-noise
SAM Structural Anomaly Mapping
TTU Through-transmission ultrasonic
TWI Thermal Wave Imaging
UT Ultrasonic testing
WFD Widespread fatigue damage

xxiii
 EXECUTIVE SUMMARY

The aircraft industry continues to increase its use of composite materials. This is most
noteworthy in the area of principle structural elements. The extreme damage tolerance and high
strength-to-weight ratio of composites have motivated designers to expand the role of fiberglass
and carbon graphite in aircraft structures. This has placed greater emphasis on the development
of improved nondestructive inspection (NDI) methods that are more reliable and sensitive than
conventional NDI. The Federal Aviation Administration’s (FAA) Airworthiness Assurance NDI
Validation Center (AANC) at Sandia Laboratories has been pursuing this goal via a host of
studies on inspection of composite structures. Through the FAA-AANC’s participation in the
Commercial Aircraft Composite Repair Committee Inspection Task Group, this team has been
investigating the need for improved inspections of composite structures.

The majority of composite honeycomb structure inspections are performed visually and
supplemented by tap test methods. Tap testing, which uses a human-detected change in acoustic
response to locate flaws, and more sophisticated NDI methods such as ultrasonics or
thermography have been applied to an increasing number of applications to detect voids,
disbonds, and delaminations in adhesively bonded composite aircraft parts. Low frequency bond
testing and mechanical impedance analysis tests are often used to inspect thicker laminates. A
probability of detection experiment was completed to assess the performance of both
conventional and advanced NDI techniques.

A series of composite honeycomb specimens with statistically relevant flaw profiles was
inspected using both human tap test equipment and new inspection techniques, which have
recently been introduced to automate and improve composite NDI. Industry-wide performance
curves have been produced to establish: 1) how well current inspection techniques are able to
reliably find flaws in composite honeycomb structure, and 2) the degree of improvements
possible through the integration of more advanced NDI techniques and procedures. This study
compared the results from a wide array of NDI methods, and identified limitations and optimum
applications for specific inspection methods while producing key recommendations for
improving the performance of NDI.

xxiv
1. INTRODUCTION AND BACKGROUND

1.1 OVERVIEW OF COMPOSITE HONEYCOMB FLAW DETECTION EXPERIMENT

In 1991, the FAA established an Airworthiness Assurance NDI Validation Center (AANC) at
Sandia National Laboratories. Its primary mission is to support technology development,
validation, and transfer to industry to enhance airworthiness and improve the aircraft
maintenance practices of the commercial aviation industry. The rapidly increasing use of
composites on commercial airplanes coupled with the potential for economic savings associated
with their use in aircraft structures means that the demand for composite materials technology
will continue to increase. Inspecting these composite structures is a critical element in assuring
their continued airworthiness. The FAA-AANC at Sandia National Laboratories, in conjunction
with the Commercial Aircraft Composite Repair Committee Inspection Task Group (CACRC-
ITG), completed a study to assess conventional and advanced inspection methods as applied to
flaw detection in honeycomb composite structures.

Aircraft reliability is a critical concern that is receiving increased attention in today’s aviation
industry. More sophisticated methods and enhanced training of maintenance personnel are being
deployed in an effort to increase aircraft safety as well as aircraft availability. To safely extend
the operational life of their aircraft, operators may need to perform additional inspections.
Nondestructive inspection (NDI) becomes critical when addressing widespread fatigue damage
(WFD) for which requirements may call for the detection of exceedingly small damage. As
composite materials are increasingly used in principal structural elements, the application of NDI
methods suited for composite and bonded structures becomes more important. The evolution of
advanced NDI methods produces a number of inspection options that can enhance sensitivity,
improve reliability, minimize human factors concerns, and even reduce operating costs.

The aircraft industry continues to increase its use of composite materials. This is most
noteworthy in the area of principle structural elements. The extreme damage tolerance and high
strength-to-weight ratio of composites have motivated designers to expand the role of fiberglass
and carbon graphite in aircraft structures. This has placed greater emphasis on the development
of improved NDI methods that are more reliable and sensitive than conventional NDI and the
optimization of current inspection practices. The FAA-AANC has been pursuing this goal via a
host of studies addressing the inspection of composite structures. The FAA-AANC, in
collaboration with the CACRC-ITG, have been investigating the performance of conventional
inspection methods and determining the need for improved inspections of composite structures.

The majority of composite honeycomb structure inspections are performed visually and
supplemented by tap test methods. Tap testing, which uses a human-detected change in acoustic
response to locate flaws, and more sophisticated NDI methods (e.g., ultrasonics or
thermography), has been applied to an increasing number of applications to detect voids,
disbonds, and delaminations in adhesively bonded composite aircraft parts. Low-frequency bond
testing and mechanical impedance analysis tests are often used to inspect thicker laminates.

The FAA-ANNC conducted the Composite Honeycomb Flaw Detection Experiment (CHE) to
assess flaw detection in composite honeycomb aircraft structures using both conventional and
advanced NDI techniques. The CHE involves the use of a set of composite honeycomb test

1
specimens containing engineered flaws that were shipped to airlines and third-party maintenance
depots for the purpose of acquiring flaw detection data from aviation industry inspectors. A
series of composite honeycomb specimens with statistically relevant flaw profiles was inspected
using both human tap test equipment and new inspection techniques that have recently been
introduced to automate and improve composite NDI. The majority of the testing was in the form
of blind probability of detection (PoD) studies, whereas other portions of the testing determined
signal-to-noise ratios from which flaw detection could be inferred. The primary factors affecting
inspections were incorporated into this study, including composite materials and construction
type, flaw profiles (e.g., voids, heat damage, disbonds, delaminations, and impact damage),
mechanical interactions (e.g., impact and audible response), subsurface damage, and
environmental conditions. The critical phase of this effort used airline personnel to study PoD
performance in the field and to formulate improvements to conventional inspection practices.
Industry-wide performance curves have been produced to establish: 1) how well current
inspection techniques are able to reliably find flaws in composite honeycomb structure, and 2)
the degree of improvements possible through the integration of more advanced NDI techniques
and procedures.

The conventional composite inspection techniques studied were: Boeing and Airbus manual tap
hammers, low frequency bond testing (LFBT), high frequency bond testing (resonance), through-
transmission ultrasonics (TTU), and mechanical impedance analysis (MIA). In addition, 18
different advanced NDI methods were evaluated in these experiments, including: Mitsui
Woodpecker automated tap hammer, WichiTech Digital Tap Hammer (DTH), computer-aided
tap tester (CATT), thermography (pulsed and induction), pulse-echo ultrasonic testing (UT), air-
coupled UT, MAUS C-scanning in MIA and resonance mode, laser UT, shearography,
microwave, Structural Anomaly Mapping (SAM), UT linear array, digital acoustic video,
laminography, terahertz imaging, and BaNDIcoot resonance scanning.

The CACRC-ITG completed an effort to develop solid laminate and honeycomb NDI reference
standards [1] to aid the uniform and optimum application of aircraft NDI techniques. As a
follow-on activity, the CACRC-ITG completed a multiyear study to assess flaw detection
capabilities in composite honeycomb structure. This document summarizes the experiment
purpose, the test variables included in the CHE study, the experiment planning issues, the set of
test specimen designs, and a comprehensive set of results from the conventional and advanced
NDI methods evaluated in this experiment.

1.1.1 Description of CHE:

The test articles for this experiment were modeled after the general range of construction
scenarios found on commercial aircraft. A set of 44 composite honeycomb test specimens
containing engineered and natural flaws were manufactured. Flaws of various size, shape, and
type were randomly located across the specimen footprints. Details of the specimen construction
are as follows:

• Skin type: carbon graphite and fiberglass

• Skin thickness: panels have three-, six-, and nine-ply skins
• Core: 1" thick Nomex® core

2
• Paint: all panels are painted as per current aircraft specifications

The flaw types included in the test specimens are:

• Interply delaminations (pillow inserts).

• Machined core disbonds simulating the presence of an air gap.
• Pillow insert disbonds simulating tight contact but no adhesive strength (kissing disbond).
• Naturally formed impact damage (e.g., crushed core, disbonds, delaminations, and broken
fibers).

One critical technical challenge required a parallel research effort to determine how to produce
representative flaws in composite honeycomb structure. During the experiment, each blind
inspection was preceded by inspections on appropriate reference standards supplied by the
experiment monitors. The inspectors were given information on the manufactured flaws present
in the reference standards and allowed to use these specimens to ensure the proper operation of
their equipment.

Over the course of several years, the CHE traveled to airlines, third-party maintenance depots,
and aircraft manufacturers to acquire flaw detection data. The experiment was deployed in a
hangar to provide a representative inspection environment, including impediments such as poor
lighting and noise distractions. The experiment required approximately two to three days of each
inspector’s time. In general, inspectors were asked to locate and size hidden flaws in the test
specimens. After a sufficient number of inspectors completed the experiment, industry-wide
performance curves were established to determine how well current inspection techniques are
able to reliably find flaws in composite laminate structure. A total of 75 inspectors from 22
airlines as well as maintenance and repair organizations located around the world participated in
the experiment. Inspector feedback on performance provided excellent training for the
experimenter, whereas their results produced a valuable baseline of how well the industry is able
to inspect composite structures (e.g., flaw hits/misses, false calls, flaw sizing, effects of
construction scenarios, and effects of environment). The test program was intended to evaluate
the technical capability of the inspection procedures and equipment (i.e., NDI method). Key
insights regarding measures to improve inspection performance were obtained. The results are
published in this report as industry-wide performance measures, and all links to specific aircraft
maintenance depots have been permanently removed. These blind tests produced statistically
valid PoD curves that are representative of the industry as a whole. Furthermore, results from the
18 advanced NDI methods allowed the team to quantify the degree of inspection improvements
possible via the application of more sophisticated inspection methods and procedures. The
experiment results also allowed the team to determine which NDI methods possess unique
capabilities to satisfy specific inspection requirements.

1.1.2 Results Obtained From CHE

Each inspection technique was evaluated using the following performance attributes: 1) accuracy
and sensitivity; 2) data analysis capabilities; 3) versatility; 4) portability; 5) complexity; 6)
human factors; and 7) inspection time. The most important of these parameters was the
quantitative metrics, as they are objective standards that can be numerically counted. Accuracy is

3
the ability to detect flaws reliably and correctly in composite structures and repairs without
overcalling (false calls). Sensitivity is the extent to which the inspection system responds to
flaws as a function of size, type, and location in the structure (e.g., proximity to repair edges, and
underlying or adjacent structural elements). Each inspector’s flaw calls were used to identify hits
(calls with any amount of overlap between the call and the solution), misses (no call for an area
of a known flaw), false calls (call with no overlap of a flaw), and the degree of overlap between
experimenter calls and actual flaw areas. Typical results include PoD curves, tabulated flaw
detection and false call data, and scatter charts relating accuracy in delineating flaw size and
shape. All of the data were sorted by the attributes of the structure being inspected. For example,
some PoD curves compare conventional inspection methods for a particular construction type of
honeycomb panel. Another set of PoD curves may compare the results from several advanced
NDI methods with those obtained from one of the better-performing conventional methods.

1.2 INCREASING USE OF COMPOSITES IN AIRCRAFT STRUCTURES

Composite materials are increasingly becoming the material of choice for aircraft designers
because of their global benefits. Engineers estimate that building comparable fuselages with
aluminum would take thousands of components and fasteners and require extensive tooling and
dozens of technicians. An aircraft would weigh approximately 20% more and consume more
fuel. Through the use of composite technology construction, engineers can cut the number of
parts in an assembly in half. This results in significant cost savings. Other benefits of composite
technology include lower acquisition costs; lower operating costs; and improved maintainability,
reliability, and durability. Figure 1 shows a typical composite honeycomb construction with
laminate skins, produced from fiberglass, carbon, or other epoxy-impregnated materials, bonded
to a honeycomb core using an adhesive film. The laminate skins may be pre-cured prior to the
secondary honeycomb bonding process. The skins may also be co-cured where the various plies
in the skins are cured simultaneously with the adhesive layer between the skins and the core.

Figure 1. Typical construction of composite sandwich structures

New transport and commuter category aircraft, such as the Boeing 787 and the Airbus A350, are
being produced with a majority of their structure composed of composite materials. Typical
damage encountered in composite structures includes: 1) disbonds and delaminations stemming
from normal flight loads; 2) fluid ingress; 3) impact damage; 4) lightning strikes; 5) deterioration
from contact with fluids, such as paint strippers or hydraulic fluids; and 6) extreme heat and

4
ultraviolet exposure. Each of these elements can produce hidden damage that may be difficult to
visually detect yet are significantly detrimental to the strength of the structure.

References 1–3 describe a successful effort to develop an industry-wide set of composite

reference standards. The standards are being used in NDI equipment calibration for damage
assessment and post-repair inspection of commercial aircraft composites. Final review of these
honeycomb and solid laminate standards was completed, and several aircraft manufacturers have
already adopted these standards into their maintenance manuals. The activity described herein
complements the composite reference standard development effort. The purpose of this
experiment was to assess the ability of conventional and emerging NDI techniques to inspect for
flaws in representative composite structures. The experiment established the sensitivities and
limitations of applicable NDI methods. Other observations accumulated during the test program
will allow for inspection improvements via optimized procedures and practices.

Figures 2–5 depict the increasing use of composite materials in aircraft manufacture and
highlight the wide range of composite structures on commercial aircraft. The photos in figures 6
and 7 show several finished composite aircraft components. They underscore the degree of
complexity associated with these structures and the size of components that are being fabricated
from composites.

Figure 2. Use of composite structures on Airbus 320 series aircraft

5
 Figure 3. Major composite structures on A380 aircraft

Figure 4. Summary of composite structures on Boeing 787 aircraft

6
Figure 5. Summary of composite structures on Cessna Citation III aircraft and
conventional NDI methods used to inspect them

Figure 6. Production of an all-composite fuselage section

7
Figure 7. Summary of advanced composite applications on A380 primary structures

8
1.3 BACKGROUND ON IN-SERVICE INSPECTION NEEDS FOR COMPOSITE
STRUCTURES

Composites have many advantages pertaining to their use as aircraft structural materials. These
advantages include their high specific strength and stiffness, resistance to damage by fatigue
loading, and resistance to corrosion. In addition, new analyses, operational experience, and
aircraft safe-life extension programs may produce additional NDI requirements. The expanded
use of composite structures, coupled with difficulties associated with damage tolerance analysis
(DTA) of composites, create a greater need for NDI methods that can effectively identify
degradation and damage in composite structures. This must be balanced with the need for simple,
low-cost NDI methods for detecting damage in composite structures and repair configurations.
Recent developments in advanced NDI techniques have produced a number of new inspection
options. Many of these methods can be categorized as wide-area techniques that produce two-
dimensional flaw maps of the structure. New inspection techniques available today or in the
immediate future hold promise for reducing the direct maintenance costs while improving the
capacity for detecting damage. Improved NDI techniques could also help detect damage in its
early stages, thus, improving safety and reducing the costs associated with the restoration of a
larger affected area.

The reliability, safety, and availability of aircraft can be improved, if deemed necessary, through
the application of more sophisticated NDI methods and with enhanced procedures and improved
training of maintenance personnel. This study compared the results from a wide array of NDI
methods, and identified limitations and optimum applications for specific inspection methods.
Reference 4 previously addressed the application of conventional pulse-echo UT NDI methods to
establish an aviation industry performance baseline for flaw detection capability.

1.4 DAMAGE TOLERANCE APPROACH TO ESTABLISH INSPECTION INTERVALS

Today’s transport category aircraft were designed using the damage tolerance approach, such that
they can meet continuing structural airworthiness requirements for an indefinite period. This
approach is predicated on the use of an effective inspection and corrective maintenance program
that effectively ensures structural integrity over the life of the aircraft. Damage tolerance is the
attribute of the structure that permits it to retain its required residual strength without detrimental
structural deformation for a period of use after the structure has sustained a given level of
fatigue, corrosion, and accidental or discrete source damage. The maintenance program may be
adjusted to reflect real-time operational experience and analytical findings through the use of
modern analysis tools, testing, and trends assessment of historical operation. Effective
maintenance programs can ensure that airplane structures continue to meet the required ultimate
strength, fatigue, and damage tolerance requirements.

Inspection requirements (sensitivity and inspection intervals) are driven by DTA. However, the
multiple plies of composite material, composite lamina (anisotropic) response characteristics,
and adhesive layers make the analysis quite complex and hinder the calculation of an exact DTA.
It is difficult to determine the effects of flaw size and the point at which a flaw size/location
becomes critical. This is especially true of disbond, delamination, and porosity flaws. Thus, an
increased emphasis is placed on quantifying the probability that a flaw of a particular size and

9
location will be detected by a piece of NDI equipment. In any surveillance of aircraft structure,
there are three main aspects to the inspection requirements:

1. The DTA that determines the flaw onset and growth data (especially critical flaw size
information)
2. The sensitivity, accuracy, and repeatability of NDI techniques, which, in concert with the
DTA, establishes the minimum inspection intervals
3. The impediments with which the NDI techniques must contend while achieving the
required level of sensitivity.

In addition to this report, detailed discussions on damage tolerance assessments for composite
materials are presented in references 5–9.

Damage tolerance is the ability of an aircraft structure to sustain damage, without catastrophic
failure, until such time that the component can be repaired or replaced. The U.S. Federal
Aviation Requirements (i.e., Title 14 Code of Federal Regulations Part 25) specify that the
residual strength shall not fall below limit load, PL, which is the load anticipated to occur once in
the life of an aircraft. This establishes the minimum permissible residual strength, σP = σL. To
varying degrees, the strength of composite doubler repairs is affected by crack, disbond, and
delamination flaws. The residual strength as a function of flaw size can be calculated using
fracture mechanics concepts. Figure 8 shows a sample residual strength diagram. The residual
strength curve is used to relate this minimum permissible residual strength, σP, to a maximum
permissible flaw size, aP.

A damage control plan is needed to safely address any possible flaws that may develop in a
structure. NDI is the tool used to implement the damage control plan. Once the maximum
permissible flaw size is determined, the additional information needed to properly apply NDI is
the flaw growth versus time or number of cycles. Figure 9 contains a flaw growth curve. The
first item of note is the total time, or cycles, required to reach aP. A second parameter of note is
ad, which is the minimum detectable flaw size. A flaw smaller than ad would likely be undetected
and, thus, inspections performed in the time frame prior to nd would be of little value. The time,
or number of cycles, associated with the bounding parameters ad and aP is set forth by the flaw
growth curve and establishes H(inspection). Safety is maintained by providing at least two
inspections during H(inspection) to ensure flaw detection between ad and aP.

10
 Figure 8. Residual strength curve

Figure 9. Crack growth curve showing time available for fracture control

11
1.4.1 Inspection intervals:

An important NDI feature highlighted by figure 9 is the large effect that NDI sensitivity has on
the required inspection interval. Two sample flaw detection levels, ad (1) and ad (2), are shown
along with their corresponding intervals, n1and nd2. Because of the gradual slope of the flaw
growth curve in this region, it can be seen that the inspection interval H1(inspection) can be
much larger than H2(inspection) if NDI can produce just a slightly better flaw detection
capability. As the detectable flaw size provides the basis for the inspection interval, it is essential
that quantitative measures of flaw detection are performed for each NDI technique applied to the
structure of interest. This quantitative measure is represented by a PoD curve such as the one
shown in figure 10. Regardless of the flaw size, the PoD never quite reaches 1 (100% PoD).
Inspection sensitivity requirements normally ask for a 90–95% POD at aP. For any given
inspection task, the PoD is affected by many factors, such as the skill and experience of the
inspector, accessibility to the structure, exposure of the inspection surface, and confounding
attributes (e.g., underlying structure or the presence of fasteners). Thus, the effects of
circumstances on PoD must be accounted for in any NDI application and associated damage
control plan. Figure 11 shows how increasingly difficult circumstances can degrade the PoD of
an NDI technique.

Figure 10. Probability of flaw detection versus flaw size

12
 1

Probability of
Detection increasingly
(PoD) difficult
circumstances

0.01
Flaw Size
Figure 11. Effect of circumstances on probability of detection

2. PURPOSE OF COMPOSITE HONEYCOMB FLAW DETECTION EXPERIMENT

Composites have many advantages for use as aircraft structural materials, including their high
specific strength and stiffness, resistance to damage by fatigue loading, light weight, and
resistance to corrosion. The percentage of composite materials used on new aircraft continues to
rise, most noteworthy in the area of principle structural elements. This expanded use, coupled
with difficulties associated with DTA of composites, create a greater need for NDI methods that
can effectively identify degradation and damage in composite structures. This must be balanced
with the need in both the commercial and military aircraft industries for simple, low-cost
nondestructive test (NDT) methods for detecting damage in composite structures and repair
configurations.

The rapidly increasing use of composites on commercial airplanes coupled with the potential for
economic savings associated with their use means that the demand for composite materials
technology will continue to increase. Inspecting these composite structures is a critical element
in assuring their continued airworthiness. Many composite honeycomb structure inspections are
performed by tap test methods that use a human-detected change in acoustic response to locate
flaws. More sophisticated NDI methods could be applied to improve the damage detection in
adhesively bonded composite aircraft parts. Towards that end, the CACRC-ITG, chaired by the
FAA-AANC at Sandia Laboratories, completed a study to assess current inspection capabilities
and the need for improved inspections of composite structures. The use of more advanced NDI to
supplement or substitute for existing inspections can be introduced through new original
equipment manufacturer (OEM) requirements.

13
The primary motivation for this program is to address the extensive and increasing use of
composites on commercial aircraft and the associated increase in the array of NDI methods used
to inspect them. Figure 12 shows how the use of composite materials has risen dramatically over
the last decade. The end result of this experiment is an assessment of the NDI flaw detection
capability in composite honeycomb structures, along with insights that can be used to improve
the performance of composite inspection methods.

Goal—use airline inspectors to establish industry-wide NDI performance curves that quantify:

• How well current inspection techniques are able to reliably find flaws in composite
honeycomb structure.
• The degree of improvements possible through the integration of more advanced NDI
techniques and procedures.

The related goals include: improve composite laminate inspection procedures and performance,
and develop structured comparisons between results from handheld inspection equipment and
automated scanning systems. Overall, the results from this study will provide input and
recommendations to the FAA regarding guidance (e.g., Advisory Circulars) that can enhance the
composite inspection process. Thus, this study is driven by a desire to improve aircraft safety.
Airlines and OEMs can use these results to guide NDI deployment and training, define what
flaws/damage can be reliably found by inspectors, and reduce the human factors issues to
produce improved NDI performance in the field.

Figure 12. Expansion in use of composite materials in aircraft construction

14
The primary sources of damage to composite structures are:

• Normal and abnormal flight loads.

• Fluid contamination and ingress.
• Surface coating removal/erosion.
• Impact (in-flight and on the ground).
hail, birds, tools, runway debris, tire separation, and ground handling equipment
• Lightning strikes.
• Heat and ultraviolet light exposure.
• Corrosion effects from adjacent metals in conductive joints (carbon materials)
• Maintenance errors.

Sample damage found in composite structures is shown in figures 13–18. Information from one
airline report indicates an average of eight composite damage events per aircraft, with 87% of
those stemming from impact. Figure 19 shows data relating the probability of an aircraft being
impacted by runway debris alone. The data indicate probability of impact that reaches the 25–
30% range. The costs associated with the repair of such impact damage averages $200,000 per
aircraft. Another report indicates that fuselage damage is incurred every 1,000 flights in wide
body aircraft and every 4,600 flights in narrow body aircraft.

The inspection challenges associated with the composite damage described above include:

• Subsurface delaminations and disbands.

• Hidden, subsurface damage.
• Small amounts of moisture.
• Cluster of damage where each individual damage point is quite small.
• Heat damage that affects resin matrix.
• Weak bonds (manufacturing or environmentally induced).

Impact damage can be especially hard to detect because this damage mode often produces
subsurface damage while leaving no external surface demarcations or visual clues. Figures 20‒
22 describe the physics behind this impact damage scenario and include photos of this type of
“blind” damage in both solid laminate and honeycomb structure. For example, hailstorm damage
can produce subsurface interply delaminations, whereas low-velocity, high-mass impacts (e.g.,
ground handling equipment) can produce substructure damage (e.g., stringer-to-skin disbonds,
frame fracture), both of which can be challenging to detect.

15
Figure 13. Sample sources of damage to composite structures

Figure 14. Sample damage from ground service vehicle impact

16
 Figure 15. Sample damage from ground operations

Figure 16. Sample damage from impacts during flight

17
 Figure 17. Sample damage from lightning strike

Figure 18. Sources of in-service damage to composite structures

18
 Figure 19. Probability of impact energy as a function of take-off speed (based on runway
debris collected from four U.K. military air bases)

2.1 OEM GUIDELINES FOR INSPECTING COMPOSITE HONEYCOMB STRUCTURES

The nondestructive testing manuals (NDTM) produced by the OEMs (e.g., Boeing, Airbus,
Embraer, and Bombardier) describe the inspection techniques developed for the detection of
composite damage before they reach critical levels [10, 11]. The procedures described in the
NDTM provide step-by-step instructions on how to apply each inspection method. These
procedures are developed through maintenance engineering experience, comparisons with
similar installations on service equipment, and engineering evaluation. They are refined and
changed as required during the service life of the equipment by a continuous evaluation of the
performance of the equipment, the results of scheduled inspections, and through study of failure
data.

The inspection requirements are stated in such a manner as to establish what is to be inspected,
what inspection conditions are needed, what NDI method(s) are to be used, and how the methods
are to be deployed. However, it must be emphasized that the reliability of the inspection results
depends on the proper application of the NDI method by trained, experienced, and proficient
inspectors. For example, interpretations of radiographs and the readout of other nondestructive
testing equipment, such as ultrasonic and eddy current, require much skill and must be performed
only by trained and experienced personnel. The operator must have training in both the basic
theory and the practical application of the NDT methods he is using and a good working
knowledge of aircraft structure. The inspectors must understand the failure characteristics of the
composite structures and be aware of limitations and capabilities of the nondestructive testing
methods they are using [10–12].

19
 Figure 20. Effects of impact on composite structures

Figure 21. Example of external impact creating minor surface demarcation but significant
internal damage
20
 Figure 22. Comparison between visible and backside damage (crushed core and backside
fiber fracture) in honeycomb structures

Figure 23 shows a tap test inspection of a composite engine shroud. Tap testing has been used for
many years and is still one of the primary methods used to assess the health of composite
honeycomb structure. During tap test inspections, low-energy impacts are applied to the surface
of the honeycomb sandwich structure. Subtle variations in the audible response from the
structure are detected by the inspector and used to infer the presence of flaws. The subjective
nature of this approach, coupled with the high probability of human factors issues adversely
affecting its performance, were the main reasons that prompted this study. The desire to validate
advanced NDI techniques for more demanding composite inspections was another major reason
for undertaking this research effort.

This study used a series of composite honeycomb specimens with statistically relevant flaw
profiles to evaluate flaw detection via widely used tap test methods. Tap tests were conducted
using both human impact techniques and NDI equipment that has recently been introduced to
automate and improve acoustic tap testing. The effort focused on understanding the factors
influencing composite honeycomb inspections and used NDI experimentation to assess flaw
detection performance. Some portions of the testing took the form of blind PoD studies whereas
other portions determined signal-to-noise ratios from which flaw detection could be inferred. The
primary factors affecting NDI that were included in this study were: composite materials,
laminate type and thickness, flaw types and profiles, mechanical interactions (impact and audible
response), and environmental conditions. One phase of this effort used airline personnel to study
tap test PoD in the field and to formulate improvements in this critical inspection procedure. The
tap test results were compared with quantitative data from other NDI devices. Whereas tap
testing was the initial focus of this effort, other composite inspection techniques, such as LFBT,
resonance, through-transmission and pulse-echo UT, and MIA were applied to complete a
comprehensive assessment of flaw detection in composite honeycomb structures using
conventional NDI. In addition, improvements obtained through the use of advanced NDI
methods were quantified through the application of instrumented tap testing, pulsed
thermography, induction thermography, vibro thermography, resonance scanning, MIA
scanning, pulse-echo UT imaging, through-transmission UT, UT spectroscopy, phased array UT,
linear array UT, terahertz UT, laser UT, laser velocimetry, shearography, digital X-ray, and
microwave. Sections 3 and 4 describe each of the techniques listed in more detail.

21
 Figure 23. Tap test inspections of composite aircraft structure

3. CONVENTIONAL INSPECTION METHODS FOR COMPOSITE HONEYCOMB

STRUCTURE

NDI tests enable direct and individual inspections of parts and assemblies in as-manufactured
and in-service condition. They identify any possible defects without affecting the integrity of the
structure. This section presents the array of conventional NDI techniques that are applicable to
flaw detection in composite honeycomb structures. The method of inspection selected for a
particular area depends on several factors, including accessibility, type of defect, material,
geometry, structural configuration, area to be inspected, NDI device deployment, and degree of
sensitivity required. Optimum inspection methods provide maximum detection sensitivity while
requiring a minimum of airplane disassembly or component removal.

Recent developments in more advanced NDI techniques have produced a number of new
inspection options. Many of these methods can be categorized as wide-area techniques that
produce two-dimensional flaw maps of the structure. New inspection techniques available today
or in the immediate future hold promise for reducing the direct maintenance costs while
improving the capacity of detecting damage. Improved NDI techniques could also help detect
damage in its early stages, thus reducing the costs associated with the restoration of a larger
affected area. The set of advanced NDI methods that are applicable to composite honeycomb
inspections are presented in section 4.

The conventional and advanced NDI methods that were applied to the CHE are summarized in
table 1. This table contains a list of applicable in-service NDI methods for composite honeycomb

22
structures and a summary concerning their application for: 1) investigating the quality and
integrity of these structures, and 2) determining their viability for continued service.

3.1 BOND TESTING

The inspection, maintenance, and repair of bonded structures are critical tasks for the aviation
industry [13]. Bonded structures can take a variety of forms—from adhesively bonded metallic
or composite plates to sandwich structures with various face sheets and core material.
Honeycomb sandwich structures are widely used on aerospace structures, especially as light
control surfaces such as ailerons, spoilers, rudders, and trailing edge flaps. The main objective
for the NDI of bonded structures is to detect the presence of delaminations in composite
laminates and disbonds between adhesively bonded layers and between the face sheet and
sandwich core. Delaminations and disbonds in adhesively bonded structures almost invariably
lead to a reduction in the stiffness of the structure, especially the contact stiffness on the surface.
This change in stiffness is exploited by a number of NDI techniques. Two main causes for the
delamination and disbond damage are low-velocity impacts occurring during maintenance, such
as that due to dropped tools, and impact damage caused by hail, runway debris, and bird strikes.
Depending on the face sheet thickness and stiffness, impact damage of honeycomb sandwich
structures may not leave clear visual indications on the surface (referred to as barely visual
impact damage or BVID). One of the NDI goals is, therefore, to characterize the severity of the
defects or damage in terms of their size, shape, location, and severity. In applying the NDI
methods, accessibility of the inspection area plays an important role. Certain techniques, such as
through-transmission air-coupled UT testing, can only be applied when two-sided access is
available. For thick honeycomb sandwiches with only one-sided access, the detection of far-side
disbonds becomes more challenging.

23
 Table 1. Inspection methods applied to the composite honeycomb flaw-detection
experiment

Item No. Conventional NDI Techniques Company/Devices Used

1 Manual tapping Airbus and Boeing tap hammers
2 Low frequency bond test (LFBT) S-9R, S-5 Sondicator, BondMaster
3 Mechanical impedance analysis (MIA) V-95, BondMaster
4 Instrumented tap hammer WichiTech Digital Tap Hammer (DTH)
5 Instrumented and automated tap hammer Mitsui Woodpecker

Item No. Advanced NDI Techniques Company/Devices Used

*Acoustography -
1 Imperium AcoustoCam™
digital acoustic video (DAV)
QMI SONDA 0070CX AIRSCAN Pulser and
2 Air-coupled UT
Receiver System
3 Instrumented and automated tap hammer Computer-Aided Tap Tester (CATT)
Digiray® Motionless Laminography X-ray
4 Advanced scanning X-ray system
(MLX)™
5 Laser UT Lockheed Martin Laser Ultrasonic Technology
6 Mechanical impedance analysis (MIA) Boeing Mobile Automated Scanner (MAUS)
7 Microwave Evisive Scan Microwave
8 Phased array UT NDT Solutions FlawInspecta® UT Array
9 Resonance Boeing Mobile Automated Scanner (MAUS)
Laser Technology Inc. Vacuum Stress
10 Shearography
Shearography
Structural Anomaly Mapping (SAM) Honeywell Structural Anomaly Mapping
11
System (SAM) System
12 *Terahertz Imaging GMA Industries Benchtop System
Thermal Wave Imaging Inc. Flash
13 Thermography
Thermography System
*Did not participate in the blind Composite Honeycomb Detection Experiment

The NDI techniques for testing bonded structures may be divided into three categories: 1)
mechanical methods of testing; 2) low-frequency methods of testing; and 3) emerging NDI
methods with bond testing capability [14, 15]. The mechanical methods include the manual
(qualitative) tap test, instrumented (quantitative) tap test, and portable load-displacement devices
such as the elasticity laminate checker. The low-frequency methods refer to several established
test methods that do not require a coupling fluid or gel between the transducer and the part
surface; these include the MIA, membrane resonance method, and pitch-catch method. In
addition to the low-frequency “sonic” methods, UT testing using the low end of the frequency
spectrum (typically less than 0.5 MHz) is sometimes used on bonded structures as well. In the
past 10–20 years, many newer NDI methods have emerged that have proved to be effective in

24
the inspection of bonded structures; these methods include shearography [16], thermography and
thermal wave imaging [17], air-coupled ultrasound, and laser peening tests [18].

3.2 MANUAL MECHANICAL TAP TESTING

In NDI of adhesively bonded structures, such as honeycomb panels, the oldest and still widely
used method is the simple tap test. The tap test is a local method in which the surface contact
stiffness is determined by tapping. Tapping a location without damage will produce a crisp and
solid sound whereas a damaged region will produce a dull sound. These subtle variations in
audible response from the structure are detected by the inspector and used to infer the presence
or absence of flaws. The hearing-based and manually operated tap test, although inexpensive and
quick to perform, is hampered by operator-dependent subjectivity and background noise within
the work environment. Research over the years on the mechanical response of a tap test has led
to the evolution of instrumented methods, with imaging capability, that are less dependent on an
operator’s subjective interpretations. Figure 24 describes this transition to more sophisticated and
sensitive tap test devices. The physical response to a tap involves the force data and sound data.
Although the acoustic spectrum has been used in some instrumented tap test systems, most of the
instrumentation of the tap test focuses on the force response [19–24].

Hearing-based “coin tap”

Sensory judgment

Instrumented tap test

Electronic signal, contact time

Mechanized tapping unit

Solenoid or magnet driven

Tap test with imaging

Merging contact time and position

Figure 24. Evolution of the tap test method in nondestructive testing

The mechanical tap test is a manual method wherein a small diameter rod or hammer with a
spherical tip is used to tap the surface of a structure while the human ear is used to monitor the
audible results (see figure 23). Tap testing has been used for many years and is still one of the
primary methods used to assess the health of composite structures. During tap test inspections,
low-energy impacts are applied to the surface of the structure being inspected. Subtle variations
in the audible response from the structure are detected by the inspector and used to infer the
presence of flaws. The audible sound resonating from the part will be characteristic of the mass,
cohesive stiffness, and the cross-sectional thickness of the part or assembly being inspected. The

25
characteristics of the impact are dependent on the local impedance of the structure and on the
mass of the tap test device used. When a defective area is tapped, the higher structural vibration
modes are not excited as strongly as when a structurally sound area is tapped. The sound
produced from a defective area has less high-frequency content, and the structure sounds duller.

The tap test inspection is a simple and fast inspection method that is based on the difference in
sound produced by an impacted clean structure and the same structure when it contains damage.
In using the tap test tool—that is, tapping the entire surface of the inspection area—it is possible
to have the following situations:

• Distinct clear sound indicates that the area tested is sound.

• Dull, damped sound indicates the presence of a flaw.

Manual tap testing using a coin or other metal pieces is the most widely practiced inspection for
bonded structures and composites. The equipment usually used in the tap test inspection is a
special hammer. For hearing-based manual tap test of bonded structures on aircraft,
manufacturers recommend various convenient tapper size and mass in their service manuals.
Figures 25 and 26 show the official Airbus and Boeing tap test devices that are deployed
manually. Inspectors sometimes use devices that deviate from the tap hammers specified in the
OEM NDT manuals. Figure 27 shows some of the alternate tappers that were discovered during
the course of this investigation. These alternative tap devices were not assessed in this study.
Figures 28–30 show the deployment of a tap test inspection along with the typical types of
damage detected with this method [25]. Tap testing has been used for many years and is still one
of the primary methods used to assess the health of composite structures. The tap test inspection
is useful to detect subsurface disbonding and delamination in honeycomb sandwich construction.
Despite its qualitative nature, it remains the most convenient and cost-effective method for
inspecting bonds. Tap testing is most effective on honeycomb structure with thin face sheets. The
subjective nature of this approach and the possibility of human factors issues adversely affecting
its performance were among the main factors prompting this study.

Figure 25. Mechanical tap hammer specified by Airbus

26
 Figure 26. Mechanical tap hammer specified by Boeing

Figure 27. Samples of homemade tappers used in the field, including modification to
Woodpecker device (addition of wood piece) to ensure that automated tap is perpendicular
to the surface being inspected

27
 Figure 28. General test pattern used for mechanical impact inspections

Figure 29. Detailed view of test pattern used for mechanical impact inspections

28
 Figure 30. Sample composite honeycomb damage to be detected with mechanical tap test

3.3 AUTOMATED/INSTRUMENTED MECHANICAL TAP TESTING

A commonly cited model for the tap test is the simple spring model, for which a tapper mass, (m)
is bounced off the surface represented by a spring constant (k). In this model, the force-time
curve is taken to be one half cycle of the mass-spring oscillation. The contact time is therefore:

τ = π (m/k)1/2 (1)

Thus, the physical basis of the tap test is that the time of contact (τ) between the tapper and
surface (i.e., the width of the force-versus-time curve) is a function of the stiffness constant (k) of
the surface and the mass (m) of the tapper [24]. For a given m, a good structure with a higher
stiffness (k) will produce a shorter contact time (τ) and a higher-pitched “crispy” sound.
Conversely, a damaged region with a lower stiffness will lead to a longer contact time and,
therefore, a lower-pitched dull sound. As a result, the local contact stiffness (k) can be deduced
from the contact time (τ) measured by an instrumented tap test device. The significance of this
relationship is that a meaningful engineering quantity of the structure, the local stiffness, can be
obtained from a tap test using any impactor of a known mass without the need for reference
standards.

The stiffness deduced from the tap test has been compared with the stiffness at the same location
measured in static load tests [26]. The measurements were made of a number of honeycomb
sandwich panels with carbon and graphite face sheets of various thicknesses. The comparison of
the stiffness deduced from the tap test and measured by static loading is shown in figure 31. It
was demonstrated using aircraft composite parts that the stiffness (k) deduced from the tap test

29
was in good agreement with the contact stiffness determined in mechanical load-displacement
tests. Thus, an instrumented tap test can produce a quantity that is indicative of the integrity, or
damage condition, of the structure.

The physical quantity most central to the tap test of structures is the contact time or impact
duration between the tapping mass and the part surface. When a surface is tapped with a different
force, the amplitude of the vibration changes, but the time of contact (τ), which is related to the
period of frequency of the vibration, remains approximately the same. The contact time at a
location with internal damage is often much longer than that at an undamaged region. The
contact time can be measured by incorporating an accelerometer or force sensor into the tapping
mass. The contact time of a tap depends on the local contact stiffness of the surface and the mass
of the tapper, but it is relatively insensitive to the velocity or the force of the tap.

Figure 32(a) shows the time of contact response on an undamaged part of a composite
honeycomb panel tapped by an accelerometer. The larger amplitudes correspond to taps of
greater force. The Y-axis shows the force or acceleration while the width of the curve on the X-
axis represents the contact time for the impactor. Figure 32(b) shows the time of contact response
of an impact-damaged portion of the panel, as tapped by the same accelerometer. The damage in
the honeycomb core caused a reduction of the contact stiffness (k) and, therefore, led to a longer
time of contact (τ).
6
Stiffness k from static load test (MN/m)

k static= 1.03 k tap - 0.13

5 2
R = 0.92

0 1 2 3 4 5 6

Stiffness k from tap test (MN/m)

Figure 31. Comparison of the surface stiffness deduced from the tap test with the stiffness
measured in a static load test

30
The range of contact time obtained with a small mass, of the order of 15 grams, on composite
honeycomb sandwiches with glass or carbon face sheets up to 7 or 8 plies is several hundred
microseconds. The sensitivity of the tap test on structures with thicker face sheets (greater than
~9–12 plies) is too low to be useful even with instrumented units. As the thickness of the skin
laminates increases, it becomes more difficult to hear subtle differences between pristine and
damaged structure. Thus, the performance of the tap test method diminishes as the skin becomes
thicker. This experiment includes a variety of honeycomb skin thicknesses to quantify the change
in performance corresponding to three-, six-, and nine-ply carbon and fiberglass skins. The
results in figure 32 show that the contact time—the parameter used to detect hidden damage—is
approximately constant for different tapping forces in either a good region or in a damaged
region. The contact time of a handheld tapping mass has proven to be surprisingly consistent
among different operators despite the differences in tapping force, velocity, strike angle, and
grasping force [24, 26].

Electronic tap test instruments, such as those shown in figure 33, have been developed to
automate the inspection process. These devices use the physics discussed above and a
force/acceleration transducer to quantify the stiffness of the local impact region. Changes in
stiffness can be observed by the user to infer the presence of damage. Some of these instruments
measure the duration of the impact while others measure the frequency content of the tap signal.
These will be discussed further in section 4.

(a) Undamaged region (b) Damaged region

Figure 32. Force vs. time history for various tapping forces in a good region and a damaged
region on a composite honeycomb panel

31
 Figure 33. Instrumented tap test devices: DTH (left) and Woodpecker (right)

3.4 LOW FREQUENCY BOND TESTING

Low Frequency Bond Testing (LFBT) refers to ultrasonic bond testers that operate below 100
kHz and are generally called sonic bond testers. Figure 34 shows an inspection of a composite
panel using an S-9 LFBT device. Sonic bond testers generally do not require the use of liquid
couplant (dry-coupled), and operate in the audio or near-audio frequency range. Different
techniques for transmitting and receiving energy have been developed for LFBT applications.
Each technique introduces a pressure wave into the specimen and then detects the transmitted or
reflected wave. The pitch-catch impulse test method uses a dual-element, point contact, non-
couplant, low-frequency sonic probe. One element transmits acoustic waves into the test part,
and a separate element receives the sound. The sound propagates in a complex wave mode across
the test piece between the probe tips. The return signals are processed, and the difference
between the effects of good and bad areas of the part along the sound path are analyzed and
compared. A complex wave front is generated internally in the material as a result of velocity
characteristics, acoustical impedance, and thickness. The time and amount of received energy is
affected by the changes in material properties, such as thickness, disbonds, and discontinuities.
The instrument processes the received impulse and displays the received information on a phase
and amplitude meter. Figure 34 shows the LFBT transducer deployed on the MAUS V scanning
device. LFBT signals, which contain both amplitude and phase information from each location
on the test article, can also be translated into a two-dimensional, color-coded C-scan image.

32
 Figure 34. S-9 Sondicator low-frequency bond test device

3.5 HIGH FREQUENCY BOND TESTING – RESONANCE MODE

HFBT is often referred to as resonance testing and uses interrogating frequencies of 25–500 kHz.
It is similar in application to contact ultrasonics in that a transducer with a hardware face is
acoustically coupled to the item under inspection using a liquid couplant. HFBT uses special
narrow bandwidth transducers, which, when coupled to the item under test, produce a continuous
sound field in the material. The test material, in turn, provides a mass loading on the transducer,
increasing the transducer bandwidth as well as changing the transducer’s resonant frequency.
Anomalies (such as disbonds) or changes in material thickness result in changes to the transducer
loading that cause changes in transducer resonance. These changes are subsequently detected as
differences in phase and amplitude of the electronic detection circuits. Acoustic impedance
changes can be thought of as variation in the ability to transmit sound between the probe and the
material under test. Changes in the materials acoustic impedance cause a corresponding change
in the electrical impedance of the transducer; it is these electrical impedance changes that are
monitored by the instrument. HFBT has been proven to be effective for inspection of multilayer
metal and non-metal laminates for the detection of disbonds as well as multi-ply, non-metallic

33
composite structure for the detection of inter-ply delaminations. Figure 35 shows the use of a
BondMaster device in resonance test mode to inspect a composite panel. The data on the right
show the typical display, known as the “flying dot” response, used to determine the presence of
damage. The constantly changing path of the flying dot can make the signal interpretation
difficult. This is especially true in the areas where signal changes corresponding to changes in
part geometry can be misinterpreted. The challenge associated with flaw detection from this type
of signal is, in some part, the cause for a reduction in the performance of this NDI method for
handheld resonance testing. When these data are transformed into a resonance C-scan image, the
test results, as shown in this report, are quite good.

Figure 35. Olympus BondMaster device operating in HFBT/resonance test mode

3.6 MECHANICAL IMPEDANCE ANALYSIS

MIA is the method of bond testing that compares the stiffness of a structure in contact with the
probe tip. The probe, in direct contact with the test surface, sends sonic vibrations into the
structure and then measures how these vibrations resonate within the structure. A flaw in the
structure causes a reduction in its mechanical impedance (stiffness), which changes the displayed
signal. The stiffness of the bonded structure is a function of thickness, geometry, elastic
variables, and densities of the bonded components. The bonded structure under test is vibrated.
Disbonds or other anomalies normally cause a reduction in mechanical impedance (stiffness) and
can result in a phase or amplitude change to the displayed signal depending on the frequency of
the probe.

The mechanical impedance of a structure, defined as the ratio of the applied force (F) and the
resultant velocity (v) is a measure of the resistance to motion of the part. Like the tap test, the
MIA method also exploits the reduced stiffness, and, hence, a reduction in the resistance to
motion, of a structure containing damage. For example, when the same force is applied to the

34
face sheet of a composite honeycomb panel over a region of fractured core, the resultant velocity
will be higher, and the mechanical impedance will be lower than a region of no damage. Like
electrical impedance, the mechanical impedance is also frequency dependent. In commercial
MIA instruments, the probe typically contains a driving piezoelectric element and a receiving
piezoelectric element. An oscillatory voltage of a certain frequency is applied to the driving
element; the receiving element—sandwiched between the driving element and the structural
surface with a spring loaded force—then produces a response voltage with an amplitude and
phase dependent on the mechanical impedance of the structure. In practice, the spring-loaded
probe is moved slowly over the part surface to detect disbonds or delaminations. The V-95 MIA
device, which uses a simple meter deflection and alarm system based on probe calibration of a
known, undamaged area, is shown in figure 36.

Figure 36. Mechanical impedance analysis inspection with V-95 device

In commercial MIA instruments, the probe typically contains a driving piezoelectric element and
a receiving piezoelectric element. The probe consists of two piezoelectric crystals with a driver
positioned behind the receiver within the same holder. An oscillatory voltage of a certain
frequency is applied to the driving element; the receiving element, sandwiched between the
driving element and the structural surface with a spring-loaded force, produces a response
voltage with an amplitude and phase dependent on the mechanical impedance of the structure.
The driver converts electrical energy into sonic vibrations, and the receiver, in direct contact with
the test surface, converts the modified vibrations into electrical signals for processing by the
instrument. If the probe is placed on an infinitely stiff structure, and the driver crystal is set to
vibrate at a given frequency, the receiver crystal will compress and expand in opposition to the
driver crystal (180° phase shift) at maximum signal amplitude. If the probe is now placed on an
infinitely flexible structure (free air), and the driver set to vibrate at a given frequency, the
receiver crystal will simply move back and forth in space but will not be compressed or
expanded and, thus, produce no output. Somewhere between these two extremes lies reality—
and, in general, a defect will produce a signal containing amplitude proportional to its stiffness
with a possible phase change. The displayed information can be impedance plane (flying dot),
meter deflection, or horizontal bar graph. Alarm thresholds can be used to provide audible or

35
visual warnings. With the aid of some simple position tracking devices and software, certain
inspection devices are able to generate two-dimensional images of the inspected area [27].

3.7 ULTRASONIC INSPECTION METHODS

UT inspections can be used for bond evaluation. With respect to composite honeycomb
inspections, only certain forms of UT are normally applied. TTU and PE-UT inspection
techniques are suitable to detect disbonds in many cases, with the latter’s utilization on
honeycomb structure being limited to thick laminates. TTU is commonly used for sandwich
structures when access to both sides can be accommodated. Nonlinear ultrasound and
spectroscopy methods have been investigated for bond testing, with the specific intention of
extracting greater information about the bond condition. Parameters such as modulus, density,
thickness, and resonant frequency can be useful in the characterization of the consistency of the
bonding process. At this time, these methods are not routinely used for honeycomb inspections.
Air-coupled inspection is an accepted method for in-service inspection of control surfaces for
skin-to-core disbonds. The low frequency (50–400 kHz) is effective despite the significant
reflection loss at an air-solid interface due to the acoustic impedance difference between air and
material. There are two types of air-coupled transducers—piezoceramic-based (disk or
composite) and capacitive type—while most of the field applications use the piezoceramic type.
The air-coupled ultrasonic NDI was implemented in the through-transmission mode with the
transducers mounted on a yoke for aircraft components to afford two-sided access. A general
discussion on UT inspection follows.

In general, UT inspection uses high-frequency sound waves as a means of detecting anomalies in

parts. UT test equipment usually operates in the range of 200 kHz–25 MHz. The speed with
which the sound waves travel through a material is dependent on the composition and density of
the material. The speed of sound in carbon graphite composite material is approximately 0.117
in/µs. Thus, the time it takes for an UT pulse to travel from the front surface to the back surface
and back to the front surface of a 0.1″ thick composite laminate (0.2″ total travel) is
approximately 1.7 µs. In PE UT inspections, short bursts of high-frequency sound waves are
introduced into materials for the detection of surface and subsurface flaws in the material. The
sound waves travel through the material with some attendant loss of energy (attenuation) and are
reflected at interfaces. The reflected beam is displayed and then analyzed to define the presence
and location of flaws. Sound is transmitted into the test item by means of a transducer. The
reflected waves are then received by a transducer, often the same transducer for PE-UT, and
converted back into electrical signals for display.

3.7.1 A-scan Mode:

UT testing involves one or more of the following measurements—time of wave transit (or
delay), path length, frequency, phase angle, amplitude, impedance, and angle of wave deflection
(reflection and refraction). In conventional PE UT, pulses of high-frequency sound waves are
introduced into a structure being inspected. A-scan signals represent the response of the stress
waves, in amplitude and time, as they travel through the material. As the waves interact with
defects within the solid, portions of the pulse’s energy are reflected back to the transducer and
the flaws are detected, amplified, and displayed on a computer screen. The interaction of the UT
waves with defects and the resulting time versus amplitude signal produced on the computer

36
screen depends on the wave mode, its frequency, and the material properties of the structure.
Flaw size can be estimated by comparing the amplitude of a discontinuity signal with that of a
signal from a discontinuity of known size and shape. Flaw location (depth) is determined from
the position of the flaw echo along a calibrated time base. In the pitch-catch UT method, one
transducer introduces a pressure wave into the specimen, and a second transducer detects the
transmitted wave. A complex wave front is generated internally in the material as a result of
velocity characteristics, acoustical impedance, and thickness. The time and amount of energy is
affected by the changes in material properties, such as thickness, disbonds, and discontinuities.
The mechanical vibration (ultrasound) is introduced into the specimen through a couplant and
travels by wave motion through the specimen at the velocity of sound. If the pulses encounter a
reflecting surface, some or all of the energy is reflected and monitored by the transducer. The
reflected beam, or echo, can be created by any normal or abnormal (flaw) interface. Complete
reflection, partial reflection, scattering, or other detectable effects on the UT waves can be used
as the basis of flaw detection. Most instruments with an A-scan display allow the signal to be
displayed in its natural radio frequency (RF) form, as a fully rectified RF signal, or as either the
positive or negative half of the RF signal.

In most pulse-echo systems, a single transducer acts alternately as the sending and receiving
transducer. If the pulses encounter a reflecting surface, some or all of the energy is reflected and
monitored by the transducer. The reflected beam, or echo, can be created by any normal (e.g.,
layer in a multilayered structure) or abnormal (flaw) interface. Figure 37 is a schematic of the
pulse-echo technique. It shows the interaction of UT waves with various interfaces within a
structure and the corresponding A-scan waveforms that are displayed on a UT inspection
instrument. Complete reflection, partial reflection, scattering, or other detectable effect on the
UT waves can be used as the basis of flaw detection. In addition to wave reflection, other
variations in the wave that can be monitored include time of transit through the test piece,
attenuation, and features of the spectral response [28, 29]. Sometimes it is advantageous to use
separate sending and receiving transducers for pulse-echo inspection. The term pitch-catch is
often used in connection with separate sending and receiving transducers. The degree of
reflection depends largely on the physical state of the materials forming the interface. Cracks,
delaminations, shrinkage cavities, pores, disbonds, and other discontinuities that produce
reflective interfaces can be detected.

37
 1 2
3 4

Transducer A UT Response Signal

1
5

Transducer B UT Response Signal

UT Pulse
Generator

Ultrasonic
Transducer

UT Gel
A B
Couplant
Flaw
Pool 1

Aircraft Skin 2 5
Adhesive Layer 3

Tear Strap 4

Figure 37. Schematic of pulse-echo ultrasonic inspection and A-scan signal showing
reflection of UT waves at assorted interfaces

Figure 38 contains two A-scan signals produced by the handheld UT inspection of a composite
specimen that contained intentional, engineered flaws at discrete locations. Changes in the A-
Scan signal (i.e., lack of reflected signal from the back wall), caused by the presence of the
disbond, are clearly visible. Key portions of the signal in figure 38 are identified to highlight how
the A-Scan can be used to detect disbonds and delaminations. The primary items of note are: 1)
the unique signature of the amplitude vs. time waveform which allows the user to ascertain the
transmission of the UT pulse through various layers of the test article and which indicates a good
bond and 2) the absence of signature waveforms indicating a disbond.

3.7.2 B-scan Mode

The B-scan presentations provide a profile (cross-sectional) view of the test specimen. In the B-
scan, the time-of-flight (travel time) of the sound energy is displayed along the vertical axis, and
the linear position of the transducer is displayed along the horizontal axis. From the B-scan, the
depth of the reflector and its approximate linear dimensions in the scan direction can be
determined. Thus, it can be considered a cross-sectional view of the part showing the depth and
in-plan dimension of the various interfaces within the part—both natural and those produced by
internal damage. The B-scan is typically produced by establishing a trigger gate on the A-scan.

38
 Backwall echo
Main Pulse from composite Ring down from
aluminum

No Disbond Present
Backwall echo No Ring down
Main Pulse from composite from aluminum

Disbond Present

Figure 38. A-scan waveform from bonded and disbonded portions of a composite structure

3.7.3 C-scan Mode—Use of UT Scanning Technology

It is sometimes difficult to clearly identify flaws using UT A-scan signals alone. Small porosity
pockets commonly found in composites, coupled with signal fluctuations caused by material
nonuniformities, can create signal interpretation difficulties. Significant improvements in
disbond and delamination detection can be achieved by taking the A-scan signals and
transforming them into a single C-scan image of the part being inspected. C-scans are two-
dimensional images (area maps) produced by digitizing the point-by-point signal variations of an
interrogating sensor while it is scanned over a surface. A computer converts the point-by-point
data into a color representation and displays it at the appropriate point in an image. Specific
“gates” can be set within the data acquisition software to focus on response signals from
particular regions within the structure. C-scan area views provide the inspector with easier-to-use
and more reliable data with which to recognize flaw patterns. This format provides a quantitative
display of signal amplitudes or time-of-flight data obtained over an area. The X-Y position of
flaws can be mapped and time-of-flight data can be converted and displayed by image processing
equipment to provide an indication of flaw depth. A variety of PC-based manual and automated
scanning devices can provide position information with digitized UT signals [30].

The basic C-scan system is shown schematically in figure 39. The scanning unit containing the
transducer is moved over the surface of the test piece using a search pattern of closely spaced
parallel lines. A mechanical linkage connects the scanning unit to X-axis and Y-axis position
indicators, which feed position data to the computer. The echo signal is recorded, versus its X-Y

39
position on the test piece, and a color-coded image is produced from the relative characteristics
of the sum total of signals received. Typically, a data collection gate is established on the A-scan
signal, and the amplitude, or time-of-flight of the signal, is recorded at regular intervals as the
transducer is scanned over the test piece. The relative signal amplitude, or the time-of-flight, is
displayed as a shade of gray or a color for each of the positions where data were recorded. The
C-scan presentation provides an image that indicates the reflection and scatter of the sound
within and on the surfaces of the test piece and their features relative to the gate settings.

A photograph of an automated (motorized) scanner, the Boeing MAUS system, inspecting an

aircraft fuselage section is shown in figure 40. The entire UT C-scan device is attached to the
structure using suction cups connected to a vacuum pump. The unit is tethered to a remotely
located computer for control and data acquisition. Figure 41 shows a comparison of A-scan
signals from damaged and undamaged portions of a composite structure that were produced by
the PE-UT method. Note the clear reflection peak produced by uninterrupted signal travel to the
back wall in the “undamaged” A-scan signal. Compare this to the A-scan signal from the
“damaged” region where the amplitude of the back wall signal is decreased, and a new
intermediate peak (reflection) is observed. Both of these A-scan changes indicate the presence of
damage or other anomaly. Figure 42 shows a sample C-scan image (based on amplitude) from a
PE-UT inspection of a composite fuselage structure containing stringers and frame shear ties
(see figure 40). Dark spots and irregularly shaped regions of non-uniform color indicate the
presence of impact damage in this panel. The value of using two-dimensional color coding,
stemming from the sum total of the A-scan signals, to identify and size composite flaws is
evident in this C-scan image.

Pulser Circuit Clock Depth Gate

Scanner Control &

Transducer Motion
Echo Intensity
X-Y Mechanical Receiver-Amplifier
Linkage Circuit

X-Axis andY-Axis
Position Indicator

Composite Doubler UT Transducer

UT Coupling to
Structure

Aluminum Substrate
C-Scan Display Compiled
Flaw Anomoly from Pulse-Echo Responses
at Discrete X-Y Locations
on Test Specimen

Figure 39. Schematic of C-scan setup for pulse-echo ultrasonic inspection

40
 Figure 40. MAUS automated ultrasonic scanning system

Figure 41. Sample ultrasonic signals generated from: a) structure without damage and b)
structure with damage

41
 Figure 42. Sample C-scan produced by an automated ultrasonic scanning device

4. ADVANCED COMPOSITE HONEYCOMB INSPECTION TECHNIQUES

This section presents the array of advanced, state-of-the-art inspection methods that are
applicable to flaw detection in composite honeycomb structures. Advanced NDI methods that
have shown great promise for inspecting composite honeycomb structures—and in some
instances gained acceptance for specific applications—include thermography, shearography,
scanning UT, acoustic laser Doppler velocimetry (SAM system), phased array UT, laser UT, and
acoustography. The advanced NDI techniques that were evaluated ranged from an automated,
sensor-based form of tap testing (CATT) to C-scan technology for improved flaw identification
(MAUS).

Advanced NDI methods that were evaluated were: Laser Technology Inc. (shearography),
Honeywell (laser velocimetry – SAM), Imperium, Inc. (UT imaging AcoustoCam), Iowa State
University (CATT and air-coupled UT), Boeing (MAUS in MIA and Resonance mode), Thermal
Wave Imaging (thermography), Sonatest (linear array UT), Southern Research Institute (UT
spectroscopy), Lockheed Martin (laser UT), Sonatest (Rapidscan linear array UT), Evisive
(microwave), GMA Industries, Inc. (terahertz UT imaging), NDT Solutions Inc. (FlawInspecta
phased array UT), QUEST Integrated (induction thermography), and Digiray (digital x-ray
laminography).

42
4.1 INSTRUMENTED AND AUTOMATED TAP TEST DEVICES

The manual, mechanical tap test method described in section 3 is an NDI practice that is very
simple and inexpensive to apply for inspecting composites. However, it has several drawbacks in
that it is dependent on the inspector’s proper deployment, their subjective hearing of the audible
response, and their interpretation of that audible tone. This approach is unable to provide
quantitative data. The results are also subject to interference from workplace noise. Another
difficulty with the qualitative, manual tap test is that hidden substructures such as ply overlap,
core splice, ribs, and spars of a part can often lead to confusing acoustic responses that can
hamper actual flaw detection. To address these concerns and enhance the performance of tap
testing technology, a number of new devices have been developed with varying degrees of tap
testing automation. These devices use digital readouts, solenoid-driven impacts, and computer
data analysis to aid the tap test process. These devices are presented in the following section.

As the tap test is a local inspection technique, full coverage of the inspection area requires
tapping in a grid pattern with a “pixel” size appropriate for the flaw size of interest. As figures 28
and 29 indicate, aircraft maintenance manuals often call for this mode of inspection coverage.
However, a raster scan using a handheld impactor can be tedious and impractical for larger areas.
A mechanized tapper is therefore desirable. Over the years, a number of instrumented tap test
devices have been developed to acquire quantitative tap response data and to use that data to
automate the interpretation process.

These mechanized tappers have substantially alleviated the operator fatigue problem often
associated with the tap test. A mechanized device not only ensures uniformity of the tapping
force and angle of strike but is indispensable in a system that generates tap test images. The basic
premise behind the instrumented tap test devices is the same. The force data and sound data of
the tap test are acquired with electronic circuits so as to eliminate the dependence on the
operator’s sensory response. Certain features of the digital data are extracted and used in the
detection of flaws, damage, and property degradation. Further processing of the data, using
Fourier spectral analysis, for example, has also been used. However, the most beneficial
instrumentation of the tap test is still the acquisition of the contact time electronically. By
incorporating an accelerometer into the tapping device, the contact time may be measured with a
simple circuit without the need for acquiring a digitized force-time curve. For example, a counter
may be started when the output voltage exceeds a certain threshold and stopped when the voltage
falls below the threshold. The value of the contact time can be used directly in the differentiation
of damaged and undamaged regions based on the established contact time of a good region for
the tapper mass used. Alternatively, the stiffness deduced from the contact time, a tapper mass-
independent quantity, may be used as the differentiating parameter.

Figure 43 shows a sample cross-section of a solenoid-driven tapping device and the inclusion of
a force transducer or accelerometer on the tip to record the impact response of the structure being
inspected. Figure 44 shows a voltage waveform of an impact that allows the impact duration—
related to the stiffness of the structure—to be measured. Changes in the stiffness of an area,
which produce changes in the duration of impact, can be observed in the plots contained in figure
45. As the structure becomes less stiff, the duration of impact gets longer, and this can be related
to the presence of damage. Some devices have used changes in the frequency response spectrum
to infer the presence of damage. In this case, the Fourier transform of the acceleration/force
43
versus time plot is used to produce the plots shown in figure 46. In both cases, the differences
between the “good” and “bad” areas can be automatically detected in the instrumented tap test
device and various feedbacks can be used (e.g., alarms, lights) to signal the presence of damage
or, at least, the need for additional scrutiny of the area.

Electromagnetic
Relay

Piezoelectric
Force Gauge
Profiled Head

Structure Being
Inspected
Figure 43. Tapping head

Figure 44. Determination of the impact duration, τ, from an accelerometer output and a
photograph of an oscilloscope trace with a τ = 968 µs

44
 COMPRESSION

Good Area
Defective
FORCE Area
P1
P2

TENSION T1 T2 TIME
Figure 45. Sample force pulses

Over the years, a number of instrumented tap test devices have been developed. Instrumented tap
test devices, like the Woodpecker, DTH, and CATT have been developed to improve the
performance of the tap test method. These devices use embedded sensors in the hammer heads to
quantify the stiffness of the contact area and display this stiffness as a relative numeric value.
Structural anomalies appear as distinct variations in the stiffness measured by the sensor as
determined by the impact force-time characteristics. This approach greatly reduces the human
factors concerns and results in more repetitive inspections.

FORCE
(log scale)
Good Area

Defective
Area

FREQUENCY
Figure 46. Fourier transform of the impulse

A mechanized, and at one time commercialized, tap test device, called the Tapometer, uses a
force transducer in a solenoid-driven impactor [31]. The system makes use of the Fourier
transform of the force-time history for differentiating defective and good regions on a structure.
When combined with a motorized scanner, the device was shown to produce good tap test
images of honeycomb sandwich structures. A handheld, battery-powered tap tester, called the
Mitsui Woodpecker [32], employs a solenoid-driven hammer that contains an accelerometer for
measuring the contact time during which the hammer is in contact with the surface. A hand-

45
operated digital hammer, called the WichiTech DTH or Rapid Damage Detection Device (RD3),
is also commercially available [33, 34]. The DTH uses a digital LCD readout to display the value
of the contact time. For this device, the impactor is a hand-deployed plastic hammer that contains
an acceleration sensor. Finally, the CATT, developed at Iowa State University, uses a
piezoelectric accelerometer fitted with a hemispherical tip as a handheld impactor and also in a
semi-automated, cart-deployed magnetic tapper. The CATT uses the contact time data and
converts it into the local stiffness based on the mass of the impactor [35]. Figure 47 shows three
of the four systems described above.

(a) (b) (c)

Figure 47. Three examples of instrumented tap test devices: a) Woodpecker, b) RD3,
and c) CATT

4.1.1 Woodpecker

The Woodpecker device, developed by Mitsui, is shown in figure 48. It uses automatic tapping,
produced by an electric solenoid, to drive the hammer tip. The device uses a calibration feature
to store a response corresponding to an undamaged region, which it then compares with the
response from subsequent inspection regions. The device then displays green (good), yellow
(possible damage), and red (probable damage) lights, along with audible alarms, to relay the
results from this relative comparison in real time. Thus, it does not depend on the audible
responses, which may be difficult to distinguish from each other, especially in noisy
environments. The quantitative values associated with the impacts are transmitted to the
Woodpecker’s monitoring unit. A user-selected number of taps can be averaged into a single
impact duration number to better normalize the results and minimize the detrimental effects of
noise sources (e.g., a non-perpendicular deployment angle). An attachment to the Woodpecker
displays the numerical value of the contact time in microseconds. This separate monitoring unit
can be used for displaying and storing measured values. This unit can also be connected to a
computer so that a grid of measurements can be logged corresponding to various locations on a
structure. In this manner, the duration of impact values can be color-coded according to specific
bins, and two-dimensional images (C-scans) can be produced of the structure’s surface. The
Woodpecker is small and lightweight to permit operation with one hand.

46
4.1.2 Rapid Damage Detection Device

The RD3 developed by Boeing, and distributed by WichiTech, consists of a lightweight hammer
containing an accelerometer connected by cable through the handle to a handheld module that
contains digital logic components and a liquid crystal digital display (see figure 49). The
accelerometer in the head of the hammer translates the force time pulse at the hammer head due
to each tap into a voltage pulse. A programmable array logic integrated circuit receives the signal
and measures the pulse amplitude. The display resets and shows a new value after each hammer
tap. If the minimum set point is not reached (i.e., the tap is too light), a zero (0) is displayed.
With this technique, the hand-held hammer is retained as the basic means of obtaining tap data.
By instrumenting a traditional tap hammer with a force transducer and associated electronics,
quantitative, objective data can be obtained simply and inexpensively. RD3 is a low-cost
instrumented tap hammer that provides a quantitative measure of the hammer/composite impulse
time that can be correlated to delaminations or other stiffness-reducing damage in the structure.
The instrumented tap hammer supplements the tonal discrimination of the operator with a
numeric readout that can readily be related to local part quality. The user is able to compare the
contact time at a region of unknown condition with that of a known good region. The effect of
background noise and operator differences on the inspection results can be minimized.

Figure 48. Woodpecker automated tap test device

47
 Figure 49. RD3 automated tap test device

4.1.3 Computer-Aided Tap Test Device

The CATT collects quantitative data relating to the surface stiffness of the component and
completely eliminates the reliance on the audio signal [35, 36]. It measures the duration of a tap
impact and relates this quantity to an effective spring stiffness of the structure. It can be deployed
manually over a prescribed grid or in a semi-automated fashion using a rolling wheel that
controls both the magnitude and spacing of the impacts. Figures 50–53 show both modes of
deployment for the CATT system. The instrumented tapper consists of an accelerometer
connected to a battery-operated signal conditioning circuit. The voltage waveform of an impact
is digitized, and the impact duration is measured, as shown in figure 44. The output voltage
waveform of an impact (i.e., the force-time history of a tap) resembles one half cycle of a sine
wave. The impact duration, τ, is measured near the zero baseline using a high gain and low
threshold.

Pre-amp

A/D Board
and Display
Software

COMPUTER

Accelerometer

Brass Tip

Sample

Figure 50. Manual deployment of CATT device

48
 SA100 Inspection Cart

Computer Interface Box

Computer
Manual Tap Probe

Figure 51. Manual tap probe and hand-deployed cart for semi-automated scanning using
CATT system

Figure 52. CATT system being deployed by Iowa State University developers on
honeycomb flaw detection experiment test panels

Figure 53. Close-up of CATT tapper and transfer of data from computer to grading sheets

49
To take advantage of the visual interpretive power of the inspector, the impact duration of an
instrumented tapper is measured and fed into a Microsoft Excel spreadsheet to produce C-scan
images. A graphical user interface is used to produce a C-scan image by displaying the relative
stiffness values acquired across the scan as color-coded values of impact durations. The system
significantly improves the quality of the inspection by producing a visual image of the condition
of the structure. To initiate a scan, a transparent plastic sheet with a printed grid is taped over the
region to be tapped. The number of taps in the length and width directions is entered into the
computer. The operator then taps each square of the grid with the handheld accelerometer tapper
in a systematic manner (for example, from left to right and top to bottom) until all the squares are
tapped. The manual tap probe is used to inspect small or confined areas and surfaces with a
significant curvature. The semi-automated version of the CATT system uses a magnet-operated
cart that impacts the surface with the accelerometer at a pre-set spacing interval as the cart is
pushed over the surface (see figures 51 and 52). Using the cart, a large area can be inspected very
rapidly. Each tap triggers the A/D board and prompts the program to record one impact duration
value for the corresponding location. A C-scan type image is generated as the tap scan
progresses. The measured values of the impact duration are displayed in 16 different colors. The
resulting image shows a damaged area as a region of higher impact duration values. Such an
image also reveals internal substructures of the composite part, including core splices, ply drop-
off or build-up, reinforcing doublers, and septum changes. Based on a mechanical model of the
impact response, the impact duration image can be converted into an image of an effective spring
constant that indicates the local stiffness of the structure. Tap scan images of different parts of a
component or of different components may be stored in the computer for future reference and
assessment of damage progression.

Images can also be produced to display the local stiffness (an effective spring constant) of the
structure and the percentage loss of stiffness caused by damage. Figure 54 shows a tap test image
of a carbon-fiber-reinforced polymer honeycomb sandwich panel containing six engineered
flaws that simulate disbonds and delaminations. The figure is an image of the local stiffness (k)
converted from the measured contact time using the impactor mass. Figure 54 shows that the
background stiffness of the panel is approximately 1.4 MN/m while at the two worst flaws, the
stiffness has decreased to approximately 0.5 MN/m. The main advantage of an image display of
the tap test results is that the visual and analytical power of the operator is immediately put to
use. The size, shape, and severity of the defects can be easily assessed. In addition, the
distinction between defects and normal substructures is much easier to ascertain using an image
that displays both the defects and the normal substructures such as ply overlaps, core splices, and
repair patches. Figure 55 shows a similar C-scan based on the color-coded results from the
duration of impact parameters. The C-scan is plotted with the X and Y locations of the part
displayed to allow for flaw location and sizing. For the purposes of this CHE, all CATT results
were transferred to grading sheets using the data transfer process shown in figure 53. Test
specimen flaw layout drawings were also overlaid on the CATT C-scans, as shown in figure 55,
to allow for accurate grading of the CATT performance.

The C-scan images can be displayed as two-dimensional or three-dimensional surface plots. For
example, figure 56 reveals structural features (stiffeners), damage, and flaws within a
component. An example of an intensive application of tap test imaging in the field is shown in
figure 57. The image shows the inner and outer surfaces of a composite rudder that was

50
recovered from an aircraft accident that involved both the vertical stabilizer and rudder
components. Parts of the rudder were broken apart so the tap test scans, made with the
semiautomatic magnetic cart of the CATT system over intact regions, were assembled afterward.
The assembled image in figure 4-15 represents more than 60 scans of 2 x 3 ft (60 x 90 cm) each
and covered a total scanned area of approximately 400 ft2 (37 m2). The overall image revealed
considerable information about the damage distribution and fracture orientation on the rudder.

0.5

1.5
Stiffness
2.5

3.5
(MN/m)
4.5
0.00-0.18
5.5
0.18-0.36
6.5
0.36-0.53
7.5
0.53-0.71
8.5 in.
0.71-0.89
9.5

10.5
0.89-1.07

11.5 1.07-1.24
12.5 1.24-1.42
13.5 1.42-1.60
14.5 1.60-1.78
15.5

16.5
0.25

1.25
2.25

3.25
4.25

5.25

6.25

7.25

8.25

9.25
10.25

11.25
12.25

13.25

14.25
15.25

16.25

in.

Figure 54. CATT C-scan tap test image of surface contact stiffness of a composite
honeycomb sandwich panel containing engineered defects
Local Stiffness, K, MN/m

0.75
1.5

Missed 2.25
3
Missed 3.75
4.5
5.25
6
6.75
7.5
8.25 in.
9
9.75
10.5
11.25
12
12.75
13.5
14.25
15
15.75
16.5
2.5
0.25
1
1.75

3.25
4
4.75
5.5
6.25
7
7.75
8.5
9.25

10.75
11.5
12.25
13
13.75
14.5
15.25
16
10

PANEL 6F-1

in.

Figure 55. Sample CATT C-scan of a 6-ply fiberglass test panel with flaw template overlay
showing hits and misses
51
Figure 56. Two-dimensional and three-dimensional C-scan images produced from a CATT
inspection of a B767 elevator

Figure 57. Assembled CATT tap test image of the inner and outer surfaces of a composite
rudder recovered from an aircraft accident

52
4.2 PULSED THERMOGRAPHY

Thermography is an NDI method that uses thermal gradients to analyze the physical
characteristics of a structure such as internal defects. This is done by converting a thermal
gradient into a visible image by using a thermally sensitive detector such as an infrared (IR)
camera [37, 38]. Flash thermography relies on the heat absorption characteristics of the structure
to indicate the presence of defects. In thermographic NDI, part of the IR band of the
electromagnetic spectrum is used to map the surface temperature of an inspected item. The
temperature distribution on a structure can be measured optically by the radiation that it produces
at IR wavelengths. Many defects affect the thermal properties of materials. Examples are
corrosion, disbonds, cracks, impact damage, panel thinning, and fluid ingress into composite or
honeycomb materials. In general, a source of energy is used to create a temperature difference
between the specimen and the surrounding environment. Variations in the structure or material
properties result in variations in heat flow and surface temperature, which are recorded by the IR
camera. Figure 58 shows a schematic of a thermographic inspection system and highlights the
physics of flaw detection.

Thermographic inspection is accomplished using high-power flash lamps or other heat source, an
IR video camera, and image processing hardware and software—all of which are controlled by a
personal computer. By the judicious application of external heat sources, common aircraft
defects can be detected by an appropriate IR survey. The heat source, such as flash lamps, is used
to raise the surface temperature of the structure. The subsequent heat transfer into the material is
affected by any defects that may be present. The resulting temperature distribution is then
recorded by the IR camera and displayed on the computer monitor. As the heat diffuses through
the structure, the surface temperature is monitored for a period of time by an IR camera. In
practice, the computer actually obtains several images at progressively later times after each
flash. Areas that appear hotter than normal may indicate the presence of a delamination or
disbond beneath the surface that is preventing heat diffusion into deeper layers. By using a
computer to analyze and manipulate the IR data captured over time, subtle variations can be
enhanced in the image. Typical computer enhancements include analysis of the first and second
derivatives of the heat-versus-time signatures at each point in the time sequence to produce
images showing rates of change. Through the use of temperature-versus-time images produced
by the thermography system, it is possible to determine the depths of disbonds, delaminations,
and other flaws in a structure. Typical gantry-based and handheld thermographic inspection
systems are shown in figure 59.

Thermographic inspection procedures on aircraft parts can be used to detect certain local changes
in materials that occur in homogenous parts. These may typically be considered (but not
exclusively) as voids, inclusions, disbonds, fluid ingress or contamination, foreign objects, and
damaged or broken structural assemblies. Thermographic inspection can be carried out on almost
every type of material used in the construction of aircraft. The means of excitation, the detection
method, and the inspection parameters can be varied depending on the material to be inspected
and the flaws to be detected.

53
 Figure 58. Principle of active pulsed thermography

Figure 59. Laboratory thermal wave imaging system inspecting composite flaw detection
panels and a portable field system inspecting an aircraft fuselage

Some of the advantages of the thermography inspection method are enumerated below:

1. Thermography can be performed without physical contact with the surface.

2. Single images can include relatively large areas (1–2 ft2), allowing for rapid inspections
of large surface areas.
3. Two-dimensional images of the inspected surface help the operator visualize the location
and extent of any defect.

The primary disadvantages of thermography are:

54
1. It is often necessary to apply a high-emissivity coating during inspections to obtain an
acceptable image (steps have been taken to minimize the labor time associated with this
task).
2. Damage to layers deep within a structure is more difficult to detect than damage in
surface layers because the larger mass of material tends to dissipate the applied heat
energy.

After presenting the thermography principles and equipment, it is worthwhile to discuss some
specifics on the critical component: the IR camera. An IR camera is a non-contact device that
detects IR energy (heat) and converts it into an electronic signal, which is then processed to
produce a thermal image on a video monitor and perform temperature calculations. Heat sensed
by an IR camera can be very precisely quantified, or measured, to monitor thermal performance
as well as identify and evaluate the relative severity of heat-related problems. Recent innovations
(in particular, detector technology) pertaining to the incorporation of built-in visual imaging,
automatic functionality, and IR software development deliver more cost-effective thermal
analysis solutions. A brief comparison of some IR cameras used for thermographic inspection
systems is provided in figure 60.

A40 Merlin Mid Phoenix

Detector Material: Vanadium Oxide (VOx) Indium Antimonide (InSb) Indium Antimonide (InSb)
Detector Cooling: Uncooled Microbolometer Integral Stirling or LN2 Integral Stirling or LN2
Spectral Range: 7.5-13 micron 3-5 micron 3-5 micron
Thermal Sensitivity: 0.08° C 0.025 °C 0.025 °C
Focal Plane Array: 320 x 240 320 x 256 640 x 512
Frame Rate: 60 Hz 60 Hz 30 Hz
Weight: 3.1 lbs 9 lbs Camera: 7 lbs & RTIE: 6 lbs
Size: 8.2” x 4.3” x 3.6” 9.8” x 5.5” x 5.0” Camera: 7.5” x 4.4” x 5.2”

Figure 60. Comparison of IR cameras for thermography inspection

4.2.1 Thermal Wave Imaging EchoTherm Thermography Inspection System

In the solid laminate flaw detection experiment, a turnkey thermography inspection system—the
Thermal Wave Imaging (TWI)—was used to assess the merits of thermography to detect flaws in
composite honeycomb construction. Figure 61 shows a photo of this inspection device and
example applications on aircraft. The TWI ThermoScope® and EchoTherm® NDI systems are
designed for in-service applications and are integrated hardware and software systems for
analyzing and measuring physical properties of materials using pulsed thermography. The
system includes TWI’s Thermographic Signal Reconstruction® (TSR) processing technique,
which increases spatial and temporal resolution of a thermogram sequence.

55
Figures 62 and 63 show sample results from thermographic inspections on bonded tear straps and
composite honeycomb structure, respectively. Figure 62 shows how a disbond between an
aircraft skin and the substructure tear strap affects the thermographic image by changing the heat
transfer in that local region. Similarly, the IR image in figure 63 indicates the various flaws that
were engineered into the honeycomb panel. Figures 64 and 65 contain additional IR images of
various flaws in composite honeycomb and composite laminate structures. One of the limitations
of thermography is the depth of penetration of the inspection. For composite laminates, the
inspection depth limit is approximately 0.2″. Only flaws that manifest themselves as variations in
the surface temperature of the structure can be readily imaged by the IR camera. Novel heating
methods are currently being used to infuse higher levels of heat energy into the structure and
improve the detection of deeper flaws.

Figure 61. TWI system equipment and inspection of aircraft

56
Figure 62. Sample thermography image showing a disbond in an aluminum fuselage tear
strap structure

Figure 63. FLIR A40 uncooled camera inspecting the honeycomb test panels
and a sample IR image from a fiberglass panel

57
The TWI IR system was applied to a bonded, composite doubler repair that was installed on a
DC-9 fuselage section in the FAA-AANC hangar. Figure 66 shows a schematic of the 10-ply
doubler highlighting the size, shape, and location of the embedded flaws. The resultant sequence
of images produced by a TWI inspection is also contained in figure 66. The features seen in the
images associated with the earlier times are defects closest to the outside surface of the patch
(note appearance of flaws #1 and #2 in the first few frames). The disbonds, located at the base of
the doubler, and deeper delaminations appear in the later frames, corresponding to their delayed
effect on the thermal field. All six embedded flaws were identified in the TWI images, and flaws
smaller than 0.5” in diameter could be detected.

Pulsed Thermography

Figure 64. Thermography image produced from inspection of composite laminate panel
with flaw profile as shown in drawing on the right

58
 Water Ingress in a Composite Impact Damage in a Solid
Honeycomb Structure Laminate Composite Structure

FLIR A40 Uncooled IR Camera FLIR Phoenix IR Camera Radiance HS IR Camera

6
7

8
5
4

3
2

Figure 65. Sample thermography images showing damage in composite structures with
comparisons from three different IR cameras on 3-ply carbon honeycomb panel
(all flaws identified by all cameras)

59
 gate 1: 5 f gate 2: 20 f gate 3: 43 f gate 5: 115 f

gate 7: 205 f gate 9: 305 f gate 11: 413 f gate 14: 660 f
frame time (f) = 1/60 of a second

Figure 66. Sequence of thermal wave images from DC-9 composite doubler inspection

60
4.3 LINE SCANNING THERMOGRAPHY

LST is a non-contact inspection method based on dynamic thermography. The LST technique
provides a quick and efficient methodology to scan wide areas rapidly. The technique has been
used for the inspection of composite propellers, sandwich panels, motor case tubes, and wind
turbine blades, among other applications. Figure 67 shows some examples of composite
structures scanned using the LST technique.

Figure 67. Examples of thermal images generated after scanning a composite structure
using the LST technique

After heat deposition in a dynamic thermography technique, internal flaws in the material show
up by variations in both the surface temperature distribution and transient surface temperature
decay rate. LST is a dynamic thermography technique patented by NASA [39, 40]. This
technique deposits heat along a thin line that is swept from edge to edge of the surface under
inspection. An IR camera moves in tandem with the heat source at a set speed, and it is able to
capture the thermal profile of the sample after the heat deposition takes place. A diagram of the
basic setup is shown in figure 68, where one can observe that the field of view of the camera is
restricted to an area of the sample surrounding the heat application region. The image on the left
shows a side view of the heat source, IR camera, and surface being studied. The image on the
right shows the LST thermal image generated by stacking a selected pixel line captured in every
frame. During the scan, the temperature of the region swept by the heat source increases,
whereas the surface temperature of the region in front of the heat application remains constant. In
LST, the scanning speed and heat intensity should be optimized to match the heat diffusion in the
inspected material. A thin material with good thermal conductivity will require fast scanning
speed and significant heat deposition. Conversely, a thick material or a material with lower
thermal conductivity will require a slower scan with a reduced heat deposition intensity.

The LST technique produces a series of images of the whole area scanned. Each image in the
series shows the surface temperature distribution at a given time after heat deposition. The
images are generated by defining an observation window or a given pixel line from all frames

61
acquired from the camera during the scan. The final image or image of the whole area scanned is
formed by stacking the selected pixel line from all the frames captured during the scan. When
using images with a sensor resolution of 240 x 320 pixels, a maximum of 240 images of the
whole area can be constructed. The time elapsed between consecutive pixel lines depends on the
scanning speed and the camera frame rate. Figure 69 shows an example of the images that can be
generated using the LST technique following heat deposition. The images show the same scale
and were generated using different observation windows. Each image in the series shows the
surface temperature distribution of the whole area scanned at a given time after heat deposition;
the time is defined by the distance between the heat application, observation gate, and speed at
which the scan is set. The LST thermal image is generated by stacking the selected observation
line from all frames recorded during the scan. The panel on the right in figure 69 shows a
collection of LST thermal images generated from different observation gates. The images show
the same scale and represent how the surface temperature drops after heat deposition.

Figure 68. Setup of LST where IR camera and heat source move in tandem through the
surface to be inspected

Figure 69. Panel showing the observation gate selection with respect to the heat deposition
location

62
Observation of a defect using LST requires proper optimization of the scanning parameters (i.e.,
scan velocity, and heat deposition intensity), as these determine the section of the cooling curve
that will be observed. The amount of heat deposited over the surface should be sufficient to
produce a thermal gradient between the defect and the sound area. In particular, when scanning
thin materials displaying good thermal conductivities, the scanning speed should be set higher
than the speed used on materials that have lower thermal diffusivities. Scanning at high speeds
provides observation of earlier times after heat deposition; scanning at lower speeds provides
images corresponding to latter observation times.

4.3.1 MISTRAS Line Scan Thermography Inspection System

Figures 70 and 71 show the MISTRAS Line Scan Thermography™ (LST) scanner used for
scanning composites. The LST technique requires that the camera moves in sync with the lamp
used for depositing the heat over the surface of interest. The movement is controlled using a
motor. These components have been organized in different ways depending on the structure to be
scanned and the size of the area of interest. MISTRAS has fabricated mainly three different
systems for inspecting different structures. The first one is a gantry-type system capable of
performing vertical scans of up to 1.5 m long and 40.64 cm in width. The system employs a
cooled IR camera working in the mid-wave IR range (3–5 micrometers). The lamp employed in
the system corresponds with a quartz lamp 40.6 cm long. The second LST system is a small area
scanner that offers a 30.5 cm x 81.2 cm area scan and uses suction cups for attaching to the
surface of interest. This scanner uses a microbolometer working in the 8–12 micrometers range.
Finally, the third MISTRAS LST system is a motorized crawler designed to scan flat areas. This
scanner offers a 4–6 cm wide field of view and can cover scan lengths of up to 12 m in a single
scan. This scanner can hold a cooled camera or a microbolometer. This crawler can be easily
modified to scan composite structures of different thicknesses, like fiber glass wind turbine
blades, for which it is necessary to wait significant time for observation after the heat deposition.
This is achieved by adding a train that increases the separation between the observation area and
the heat deposition location.

Figure 70. MISTRAS LST system—crawler used on composites

63
 Figure 71. MISTRAS LST system—small area scanner

Figure 72 shows some sample inspection results from the MISTRAS LST system applied to a
32-ply (0.23″ thick skin) solid laminate composite panel with a 58-ply (0.192″ thick) upper
stringer and 50-ply (0.125″ thick) lower stringer. Most of the flaws are detected in the LST
image, with the deeper flaws providing the biggest challenge to LST detection.

Figure 72. Results produced by MISTRAS LST system on a 32-ply panel with substructure
elements

4.4 LOCK-IN THERMOGRAPHY

The principle of lock-in thermography is based on the application of a periodic input energy
wave (i.e., thermal emitter, ultrasound, microwave, eddy current, flash lamp) to the surface of the
object being inspected and an examination of the resulting local temperatures on the surface of
the object [41]. When the wave generated by the input energy penetrates the object’s surface, it is
absorbed, and the phase angle of the wave is shifted. When the input wave reaches internal areas

64
of the object where a delamination or inclusion is present, the thermophysical properties are not
homogeneous in relation to the surrounding material—hence, the input wave is partially
reflected. The reflected portion of the wave interferes with the incoming input wave at the
surface of the object and causes an interference pattern in the local surface temperature. In turn,
this oscillates at the same frequency as the thermal wave. The internal structure of the object
being examined can then be derived by evaluating the phase shift of the local surface
temperatures in relation to the input energy wave. However, the ability to derive internal
thermophysical inconsistencies within the object requires the input energy source be used at an
optimal frequency. This depends on both the thermophysical characteristics of the object, as well
as its thickness. A schematic of a typical equipment setup for lock-in thermography is shown in
figure 73. The dynamic stimulus can be applied from a wide variety of sources when using lock-
in thermography. For composite inspection, this includes halogen lamps, ultrasound, and
mechanical stimulation.

For lock-in thermography, the recorded temperature information gathered by the IR camera is
transformed into the frequency domain. The measured temporal evolution in each pixel of the
temperature is Fourier-transformed for all images of the recorded sequence. Phase and amplitude
information are derived and presented as an image [41].

Below are some of the advantages of the lock-in thermography method:

• Summation results in noise-filtering, which enhances the contrast in inspection results.

• Depth range for phase information is twice that of pulse thermography mode.
• Lock-in allows detection of thermal waves with a sensitivity that is 100–1,000 times
greater than the best thermal camera—down to μ-Kelvin range.
• The phase image is insensitive to external effects, such as sunlight, reflections depending
on surface finish, dirt, and emissivity differences—problems common to conventional
thermography.
• The phase information is insensitive to uneven distribution of the applied heat.
• Large areas can be examined within a few minutes from a distance via non-contact
measurement.
• A less-costly, uncooled IR camera is normally sufficient.
• Affordable heat sources are widely available (e.g., halogen lamps).
• Visualization of deep defects is possible.

65
 )

Figure 73. Equipment setup used for typical lock-in thermography inspection

4.4.1 MoviTHERM Lock-In Thermography System

The most common excitation source for lock-in thermography is the halogen lamp. Figure 74
shows the MoviTHERM lock-in thermography equipment featuring a FLIR camera SR2 SC7650
with a Hedler 2500 watt lamp. Lock-in thermography can be used to detect damages such as
delamination, inclusions, and impact in composite structures. Figure 75 shows inspection results
from a 32-ply (0.23″ thick skin) solid laminate composite panel with a 58-ply (0.192″ thick)
upper stringer and 50 ply (0.125″ thick) lower stringer. The engineered flaw profile is also shown
on the left for comparison purposes.

An important excitation source used in lock-in thermography is ultrasound. Typical settings for
this method are 100 watts at 20 kHz with a 200 ms burst frequency for synchronization. A
disadvantage of the UT technique is that it can be destructive, and care is required during
excitation of the part. A powerful tool for laboratory and factory measurements is the use of
mechanical excitation for heat generation through the thermoelastic effect. Applications of this
technique include measurement of fatigue limits, imaging of stress patterns, crack propagation
studies, and imaging of vibration patterns.

66
 Figure 74. MoviTHERM lock-in thermography with halogen heat lamp being used as the
excitation source

Figure 75. Results produced by lock-in thermography on a 32-ply panel with substructure
elements

4.5 SHEAROGRAPHY

Shearography is a wide-area interferometric imaging technique that is capable of detecting

micron-sized displacements in the surface of a structure. Shearography equipment, shown in
figure 76, monitors the surface of a structure for any changes in the surface strain field. Stressing
the material in the appropriate way ensures that the subsurface anomalies are manifested on the

67
surface of the structure. Shearography is implemented by comparing two interference patterns on
a detector plane, typically “before” and “after” an object motion. If the motion, and subsequent
out-of-plane deformations, cause changes in the optical path, then the speckle patterns differ.
These images can be compared by subtraction or other algorithms to obtain an image of the
object with fringe patterns superimposed. These fringe patterns can then be used to identify the
presence, size, and depth of flaws in a structure.

A typical shearography system uses a charge-coupled device (CCD) camera with a shearing lens,
which is completely integrated into a compact measurement head, to view laser light reflected
from the surface under inspection. The object under test is illuminated with laser light, and
images from the object at different states of loading are taken. The loading of the surface is
created by different excitation methods such as vacuum, thermal, vibration, or mechanical load,
which induce some deformation of the outer surface. Such deformations are locally altered by
the presence of subsurface defects (e.g., disbonding or delaminations in composites). A
comparison of the different images captured before and after loading allows a deformation
gradient to be calculated. This deformation gradient can be a sensitive measure for identifying
local defects. Overlapping sheared images are produced in the interferometric process. Two
overlapped portions of the sheared images combine and interfere to produce a speckle pattern.
When an applied stress deforms the specimen, the speckle pattern is slightly modified. A
comparison of the two speckle patterns (stressed and unstressed) produces a fringe pattern that
depicts the relative displacement of the area being inspected. Figure 77 shows the basic
principles of shearography.

Figure 76. LTI-5200 portable shearography system with camera on test specimen

68
 Figure 77. Basic principles of shearography

Shearography inspections can detect anomalies like disbonds, delaminations, voids, separation of
structural components, wrinkles, kissing disbonds, impact damage, internal corrosion, crushed
core, changes in sections, and core splices [42]. With the use of the CCD-camera technique, no
photo laboratory is required. This makes it possible to use shearography for real-time, NDI of
structures. Laser shearography views only the surface and does not penetrate into the material.
As a result, subsurface defects must affect the surface strain field to be detectable.

4.5.1 Laser Technology Inc. LTI-5200 System

The LTI-5200 is designed for large-area inspections of bonded metallic or composite structures.
Inspection rates of 14 m² per hour and the capability to inspect face sheet, core bond lines (near
and far side), core splice joints, and bonded repairs make this system well-suited for composite
applications. The LTI-5200 is a compact, portable vacuum shearography system designed for
large-area inspections of aerospace, marine, and rail composite and cored sandwich structures
and components. The LTI-5200 vacuum attaches in any orientation. Increasing vacuum level
allows imaging and measurement of subsurface defects. Figure 78 shows the LTI-5200
inspecting a composite honeycomb aircraft rudder assembly. Figure 79 shows a schematic of this
setup where detection of both near-side and far-side honeycomb disbonds are possible. Figures
80–82 show samples of shearography images that identify flaws in composite honeycomb
panels, while figure 83 shows shearography images of a damaged solid laminate composite
structure.

69
Figure 78. Composite rudder inspection using LTI-5200 portable vacuum shearography
system

Figure 79. Schematic of shearography inspection for near-side and far-side disbond
detection

70
Figure 80. Near-side and far-side disbonds detected by LTI-5200 shearography system in
A310 composite rudder

Figure 81. Close-up view of LTI-5200 shearography image showing flaws in a composite
honeycomb structure and a sample shearography result for 6-ply fiberglass panel showing
near-side and far-side flaw imaging

71
 Figure 82. LTI-5200 shearographic inspection image of a scarfed repair to a honeycomb
structure with anomaly indications in the repair plies

Shearography

Figure 83. Shearography image produced from inspection of composite laminate panel
(0.11″ thick skin) with flaw profile as shown in drawing

4.5.2 Dantec Dynamics Q-810 Laser Shearography System

The Q-810 laser shearography system, shown in figure 84, is oriented toward use on composite
materials over large surface areas. It can detect defects such as delaminations, disbonds, kissing
bonds, wrinkling, impact damage, and crushed core with no surface preparation. The turnkey
optical systems are non-contact and full-field, and will work on such materials as carbon-fiber,
glass-fiber, laminates, honeycomb, foam, metal, and GLARE (glass laminate aluminum
reinforced epoxy). The integrated systems are optimized for large surface area inspections—for
example on aircraft fuselages, wings, control surfaces, ship hulls, wind turbine blades, and rocket
components.

72
 Figure 84. Q-810 laser shearography system

The full-field inspection rate of the Q-810 laser shearography system is approximately 300 mm x
200 mm every 10 seconds. With adaptive seals, the Q-810 can be used on flat as well as highly
curved surfaces. The system operates independently of the local environmental conditions and
can be used for production or in-field inspections. The interferometric technique measures
microscopic surface deformations caused by internal flaws when a small loading is applied to the
object. This can be done using thermal, pressure, vibration, or mechanical excitation. The results
are displayed live as the material responds to the excitation. Further image processing is also
available for export and reporting. Figure 85 shows a sample shearography image produced by
the Q-810 system while inspecting a composite laminate that contains wrinkles.

Figure 85. Test specimen (left) and Q-810 shearography image of wrinkles in a composite
laminate (right)

73
4.6 MOBILE AUTOMATED SCANNER IN RESONANCE AND MIA MODES

Section 3.7 discusses the value of using two-dimensional NDI images to more clearly and
reliably identify and characterize damage in structures. It is sometimes difficult to clearly
identify flaws using screen signal displays alone. These include A-scan signals, meter
deflections, flying dot displays, and impedance plan plots. Small porosity pockets commonly
found in composites, coupled with signal fluctuations caused by material non-uniformities, can
create signal interpretation difficulties. Significant improvements in damage detection can be
achieved by taking the analog display signals and transforming them into a single C-scan image
of the part being inspected. C-scans are two-dimensional images (area maps) produced by
digitizing the point-by-point signal variations of an interrogating sensor while it is scanned over
a surface. A computer converts the point-by-point data into a color representation and displays it
at the appropriate point in an image. A mechanical linkage connects the scanning unit to X-axis
and Y-axis position indicators, which feed position data to the computer. The transducer signal is
recorded, versus its X-Y position on the test piece, and a color-coded image is produced from the
relative characteristics of the sum total of signals received.

This C-scan capability was included in the MAUS scanner to produce two-dimensional color-
coded images of both MIA and resonance inspection modes (see sections 3.5 and 3.6). A
photograph of an automated (motorized) scanner, the Boeing MAUS system, inspecting an
aircraft fuselage is shown in figure 86. The entire UT C-scan device is attached to the structure
using suctions cups connected to a vacuum pump. The unit is tethered to a remotely located
computer for control and data acquisition. The scanning unit containing the transducer is moved
over the surface of the test piece using a search pattern of closely spaced parallel lines.

Figure 86. MAUS inspections on aircraft fuselage section

74
The MAUS, shown in figures 87 and 88, is a portable, multimodal, large-area scanning system
that integrates several inspection techniques into a single package. This system can be used in a
variety of production manufacturing and aircraft maintenance environments for process quality
inspections, damage assessment, aging structure evaluation, and repair validation programs.
Unique features of the MAUS V system include equipment portability, ease of setup, inspection
versatility, and rapid inspection rates. It incorporates an X-Y scanner to match the transducer
position with its corresponding signal so that C-scans can be constructed for the surface being
inspected. The probe is held in place using a gimble arrangement on a spring-loaded mount that
allows the probe to accurately follow, and maintain proper orientation, over curved surfaces. The
scanner motor moves the sensors across an inspection surface. As the scanner is moved forward
or backward, a strip of data is collected by each sensor. The total width of the data strip is
defined by the scanner stroke and the number of sensors attached to the scanner. The MAUS
flexible track provides fully automated hands-free scanning capability. It attaches to the part
surface using vacuum pressure created from a shop air source. The system incorporates PE-UT,
eddy current, resonance, LFBT, and MIA inspection modalities. The MAUS was applied to the
CHE using both MIA and resonance modalities. Sample C-scans produced by the MAUS,
deployed in both resonance and MIA inspection modes, are shown in figures 87–89.

Figure 87. MAUS system operating on an aircraft fuselage with sample C-scan image
produced from resonance inspection on a 6-ply carbon skin honeycomb panel

75
 Missed

Figure 88. MAUS MIA inspections on honeycomb panel with sample C-scan image
produced

Figure 89. X and Y scans of a composite honeycomb panel produced by the MAUS
inspection system deployed in resonance mode

4.7 MICROWAVE

Microwave inspection works by using a specialized transducer to bathe the material of interest in
microwave energy of an essentially constant frequency [43]. Several different system setups for
microwave inspection are shown in figures 90 and 91. The energy is reflected from each
interface between materials possessing differing dielectric constants within the specimen. The
reflected energy is superimposed, creating a signal that is acquired as an analog voltage that is

76
then digitized. This signal is sampled at numerous discrete locations across the sample to create a
two-dimensional image of the surface that highlights damage or other anomalies.

The ability of microwaves to penetrate inside dielectric materials makes microwave inspections
an NDT technique very suitable for interrogating structures made of composites. Additionally,
the sensitivity of microwaves to the presence of dissimilar layers in such materials allows for
accurate thickness measurement and variation detection. The quality of the experimental images
captured with these systems has demonstrated the potential of the technique for material NDT
purposes. Basically, these systems use an antenna (a horn antenna used in the first experiments
or open-ended rectangular waveguide used in recent years) to illuminate the composite with
electromagnetic waves—for this particular application, the electromagnetic wavelength goes
from 1–100 mm—and monitor the reflected waves. The electromagnetic waves penetrate deep
into the dielectric material where they interact with its interior and reflect back to the antenna.
The properties of the reflected wave will convey the needed information about the composite at
hand. The imaging mechanism is based on the idea that microwaves are very sensitive to
discontinuities in the material space and the presence of water (the water reflects specularly with
the wavelength of microwaves).

Microwave NDT techniques may be conducted on a contact or non-contact basis. In addition,

these techniques are conducted from only one side of the sample (reflection techniques).
Furthermore, when compared with UT techniques, microwave NDT approaches require no
coupling material (glass or water) and do not suffer from signal attenuation. Indeed, microwaves
have good propagation in the air.

Figure 90. Configuration of microwave inspection system on a laboratory scan table

77
 Figure 91. Basic equipment setup for microwave inspection

Microwave techniques are able to detect voids, delaminations, porosity variation in a variety of
materials, impact damage, and water infiltration—all problems that affect composite materials. It
can also provide the possibility of process control during the manufacturing of composites so that
the final product may not need any scrutiny and may only require occasional testing once under
some loading. Finally, microwave NDT techniques do not require a high level of expertise from
an operator and can be conducted in real time with simple, portable hardware. The main
limitation of the microwave method is that it is limited to non-conductive materials. Thus, it has
been successfully applied to fiberglass composite structures but cannot be used to inspect
carbon-graphite composites. Figure 92 shows microwave results highlighting delamination in a
fiberglass laminate boom structure. Figure 93 and 94 show results from the microwave tests on
Sandia National Laboratories’ composite honeycomb panels ranging in thickness from 3–12
plies. Note that the microwave method performs better on fiberglass skin panels than on the
carbon skin panels; this is because the carbon material does not possess dielectric properties
appropriate for the propagation of the microwave scan waves.

78
 Figure 92. Microwave C-scan of a fiberglass boom structure showing the presence of
interply delaminations

Figure 93. Microwave inspection results for fiberglass and carbon skin honeycomb panels
with engineered flaws (skins bonded to Nomex honeycomb)

79
Figure 94. Sample microwave inspection results for 3-ply and 12-ply fiberglass panels with
delamination, disbonds, potted core, and core splice

4.8 STRUCTURAL ANOMALY MAPPING SYSTEM

The SAM system uses laser velocimetry to measure out-of-plane motion in a dynamically loaded
structure [44–45]. By studying localized or unusual velocities/displacements, it is possible to
image and identify anomalies or flaws in a structure. The excitation (loading) is applied in a non-
contact, acoustic manner to provide uniform, wide-area loading and inspections. The SAM
system was developed to provide an automated, wide-area inspection method that does not
require contact with the airframe structure. The system consists of a database coordination
system, a laser-guided robotically controlled sensor delivery system, an acoustic source for
structural excitation, and a scanning laser vibrometer sensor suite. The database coordination
system includes software for command/control, signal processing, and data base management.
The results, consisting of anomaly size, classification, and location, are visualized on a three-
dimensional airframe model or in a two-dimensional C-scan image that represents the surface
area covered in the inspection scan. The database also holds information pertaining to previous
similar inspections of the airframe and any other similar airframes in the same fleet. This allows
for quick comparisons to previous measurements so that change detection can be used to monitor
changes in the airframe that may occur from operation.

Figure 95 depicts the principle behind the SAM device. This system uses acoustic excitation,
along with laser Doppler velocimetry instrumentation, to selectively excite and identify flaws
that manifest themselves in unique, out-of-plane deformations in the structure. This non-contact
technique relies on the vibration response of a structure to a remotely administered acoustic
shock. A device called a spark-gap generator delivers a rapid series of acoustic pulses that are
focused in a cone toward the part several feet away. A scanning laser vibrometer monitors the
part’s surface in response to these acoustic shocks, mapping the vibration data as different color
shades. Areas of reduced stiffness produce changed patterns, indicating the presence of a defect

80
or other anomaly. Figures 96 and 97 show the SAM device inspecting one of the composite
honeycomb panels for the flaw detection experiment, including how the results scans from the
inspections were traced onto the data logging sheets

Figure 95. Schematic showing the principle of the SAM inspection method

When using the scanning laser vibrometer, the forced frequency response data at two distinct
excitation amplitudes are processed to identify areas of the panel that exhibit significant
nonlinear response characteristics. These local nonlinearities are shown to coincide with
subsurface damage in the panel. Because the nonlinearities distort the frequency response, it is
possible to locate the damage without comparing the response to an undamaged, or baseline,
measurement. In addition, rudimentary modeling has shown results confirming this reference-
free method of detecting damage by direct comparison of frequency response functions excited
by multiple amplitude signals, particularly at higher modal frequencies [44]. Figure 98 shows
raw frequency response data comparing undamaged and damaged portions of a honeycomb
engine cowling. These data are used to produce color-coded response maps from all points on the
structure. Figure 99 shows the C-scans produced from the frequency domain signatures. The
C-scan is able to display the areas with engineered defects, as indicated on the photo of the
engine cowling. A sample SAM scan from the CHE is shown in figure 100, where the subsurface
disbond and delamination damage is shown as hot spots in the color palette.

81
 Figure 96. SAM laser Doppler velocimetry camera and acoustic excitation device;
composite test specimen in the background

Figure 97. SAM system inspecting composite honeycomb panels and transfer of scan data
onto experiment grading sheets

82
 Figure 98. Frequency response plots produced during the application of the SAM
inspection system on an engine cowling

Figure 99. SAM system inspecting an engine cowling and sample image showing flaws

83
 Figure 100. Sample image produced by SAM system showing flaws in a fiberglass
composite honeycomb structure

4.9 PHASED ARRAY AND LINEAR ARRAY ULTRASONICS

Conventional UT transducers for NDI commonly consists of either a single active element that
both generates and receives high-frequency sound waves or two paired elements: one for
transmitting and one for receiving. Phased array probes, on the other hand, typically consist of a
transducer assembly with 16 to as many as 256 small individual elements that can each be pulsed
separately. A phased array system will also include a sophisticated computer-based instrument
that is capable of driving the multi-element probe, receiving and digitizing the returning echoes,
and plotting that echo information in various standard formats. Unlike conventional flaw
detectors, phased array systems can sweep a sound beam through a range of refracted angles or
along a linear path, or dynamically focus at a number of different depths, thus increasing both
flexibility and capability in inspection setups. The main difference between a phased array and a
linear array is that linear arrays aren’t capable of steering the sound beam at different angles or
focusing the beam. Thus, the sound waves stay parallel to each other regardless of the depth.

Phased array ultrasonics (PA-UT) involves the use of multiple signals from a contained series of
transducers (phased arrays) to produce diagnostic images in the form of UT C-scans. The
operation is similar to handheld UT; however, the simultaneous use of multiple sensors allows
for rapid coverage and two-dimensional images from which to assess structural integrity. A
linear array of UT sensors is placed within a single, scanning probe. The width of the linear
probe array determines the swath of the inspection “scan” as the probe is moved along the
surface. A compression wave beam is electronically scanned along the array at pulse repetition
frequencies in excess of 10 kHz. The response of each individual sensor is monitored and
assessed using the UT wave analysis approaches described above. High-speed pulsing combined
with rapid data capture permits the linear array to be quickly moved over the structure. The
individual responses from each UT sensor are integrated to produce a real-time C-scan image of
the covered area. An example of a PA-UT inspection device deployed by Sonoscan in a rolling
wheel arrangement is shown in figure 101. The physics of how the UT array works is depicted in
figure 102. By carefully controlling the generation of UT signals and data acquisition from select
elements in the array, it is possible to produce customized focusing of the array to improve the

84
sensitivity of the inspection. Electronic focusing permits optimizing the beam shape and size at
the expected defect location, thus further optimizing probability of flaw detection. The ability to
focus at multiple depths also improves flaw sizing of critical defects in volumetric inspections.
Focusing can significantly improve the signal-to-noise ratio in challenging applications and
electronic scanning across many groups of elements allows for C-scan images to be produced
very rapidly.

Figure 101. PA-UT deployed in rolling wheel mechanism (left) and contained in a single-
probe housing (right)

transducer

Figure 102. Schematic showing the operation of an ultrasonic array that allows for the
generation and acquisition of multiple UT signals

85
The Olympus OmniScan PA-UT system, shown in figure 103, is one example of a PA-UT
device. Figures 104 and 105 show sample results produced by the OmniScan PA-UT system
from the inspection of carbon laminate test specimens that contain engineered flaws. Damage in
the parts is shown in the photos and schematics, while the accompanying C-scan images show
the ability of the inspection method and equipment to detect the flaws.

Figure 103. Olympus OmniScan device with a 16:128 phased array transducer

Figure 104. Amplitude (right) and time of flight (left) data produced by OmniScan
inspection of composite laminate aircraft panel with flaw profile as shown

86
 Time of Flight

Amplitude Amplitude with Gate

in Substructure
(deeper flaw detection)

Figure 105. C-Scan images produced by OmniScan PA-UT inspection of 20-ply composite
laminate feedback panel with the flaw profile as shown

4.9.1 Boeing MAUS FlawInspecta Linear Array UT System and the Diagnostic Sonar
FlawInspecta PA-UT System

The FlawInspecta system, shown in figures 106 and 107, was designed to address the
requirement for a rapid, low-cost UT phased/linear array inspection system. The imager is a
laptop-based device that allows for easy transfer of images to other applications or via the
Internet to remote locations. It works with a wide range of integrated arrays and is suited for
applications ranging from rapid large-area flaw detection to high-speed, low-cost corrosion
mapping with 100% coverage. The data acquisition is fast enough to allow for interactive B-scan
imaging or rapid C-scanning—typically 40,000 mm2/s (64 in.2/s) for 1 mm pixels—for manual
coverage of large areas. The system is also able to perform full waveform capture (FWC), in
which the full A-scan (RF or rectified) is acquired and stored for every point on the inspection
surface. This volumetric representation offers the ultimate in data acquisition for archiving and
offline review yet is achieved at similar data rates. The FlawInspecta system is capable of a pulse
rate of 30 kHz, corresponding to a scan rate of 10 in./second or 19.3 ft2/minute, with a 128-
element array. Smart arrays can be used with a wide range of conventional equipment such as

87
flaw detectors, but their full capability is not realized unless used with a real-time imaging
system such as the FlawInspecta. At the heart of the system is Diagnostic Sonar’s FIRE-
technology (Flaw Imaging and Reconstruction Engine) for real-time full-waveform acquisition
and B-scan imaging. A position sensor attachment to the array extends this capability to C-scans
for mapping of inspection areas. The latest FlawInspecta uses proprietary FIRE-technology to
provide fast, manual imaging and mapping with FWC—typically more than 1 m2/minute for 1
mm pixels. This FIRE-technology has now been integrated into other proprietary ultrasound
mapping systems and provides an easy upgrade path to high-performance acquisition for users of
these systems.

Figure 106. Diagnostic sonar FlawInspecta PA-UT inspection system

Figure 107. FlawInspecta linear array UT system deployed on MAUS V scanner

platform—linear array UT probe includes a delay line shoe

88
In trial tests on composite honeycomb test specimens, such as the one shown in figures 108 and
109, the FlawInspecta system was able to detect nearly all of the engineered defects. Some of the
indications are best detected while observing the B-scan display. The majority of the defects in
the composite solid laminate test specimen can be seen on the C-scan images shown in figure
110. The FlawInspecta is a high-speed UT array system with dynamic real-time B-scan
capability as well as FWC and C-scan capability.

The MAUS V C-scan system uses an OEM version of the FlawInspecta UT array system that is
controlled by the MAUS software via a dynamic link library. The combination of the MAUS V
system with the FlawInspecta system allows for the rapid inspection of large areas of composite
structures in a seamless easy-to-use package. One of the differences between the Diagnostic
Sonar FlawInspecta and the add-on FlawInspecta to the MAUS V is the phased array capability.
The MAUS V FlawInspecta does not have phased array capability and is deployed in a linear
array. The addition of the NDT Solutions-designed vacuum-assisted couplant delivery and
recovery system provides for excellent coupling of large arrays to the parts with the ability to
recover and recycle the couplant to nearly eliminate the watery mess of flowing copious amounts
of water required to couple large transducers to the inspection area.

Figure 108. Composite honeycomb reference standard and sample FlawInspecta results—
3-ply carbon skin with 1″ thick core

Figure 109. Sample result from FlawInspecta PA-UT system on 6-ply carbon specimen
(dashed lines represent missed flaws)
89
Figure 110. C-scan images produced by FlawInspecta MAUS V linear array UT system on
a 32-ply composite laminate feedback panel with the flaw profile as shown

4.10 AIR-COUPLED ULTRASONICS

UT inspection is a nondestructive method in which beams of high-frequency sound waves are

introduced into materials for the detection of surface and subsurface flaws in the material. The
sound waves, normally at frequencies between 0.1 and 25 MHz, travel through the material with
some attendant loss of energy (attenuation) and are reflected at interfaces. The reflected or
transmitted beam is displayed and then analyzed to define the presence and location of flaws. UT
testing involves one or more of the following measurements: time of wave transit (or delay), path
length, frequency, phase angle, amplitude, impedance, and angle of wave deflection (reflection
and refraction).

When sound passes across an interface between two materials, only a portion of the sound is
transmitted while the rest of the sound is reflected. The proportion of the sound that is
transmitted depends on how close the acoustic impedance of the two materials matches. UT
inspections traditionally involve the use of a couplant material often deployed in a continuous

90
manner using a water pump, immersion tank, or a water squirter system to properly transmit the
UT wave from the transducer into the part being inspected. Water is a fairly good match for most
commonly used materials—for example, typically around half the sound energy is transmitted at
the interface between water and a carbon laminate. After four solid-liquid interfaces (from the
probe, to the couplant, to the test piece, and then back again), there is still a small, sufficient
amount of the original energy left to allow for accurate measurements. UT inspection by
immersion or squirter systems cannot always be conveniently applied in the field. For these
practical and operational reasons, non-contact, air-coupled ultrasonic testing (AC-UT) has the
distinct advantage of being a couplant-free UT inspection technique [46–49]. For this reason, it is
an attractive alternative for certain applications, even though AC-UT also suffers from several
significant disadvantages, the most significant of which is the attenuation and loss of signal that
accompanies air transmission of the UT signals. However, if the sound has to move between the
test piece and air (which has very low acoustic impedance), only a small percentage of the sound
energy is transmitted. Typically, the overall path loss may be 100 dB higher using air as a
couplant than when water is used.

Thus, the main limitations of AC-UT are the large reflection loss at the air-solid interface and the
large attenuation of high-frequency ultrasound in air. The latter has limited the application of
AC-UT to frequencies mainly below 1 MHz or so. It is therefore necessary to minimize losses at
every stage to achieve acceptable signal-to-noise levels for AC-UT for the inspection. Despite
the enormous reflection loss at an air-solid interface due to their acoustic impedance difference,
the advances in transducer technology and electronics are gradually making AC-UT a viable
NDI technique for bonded structures in the field. There are two types of air-coupled transducers:
piezoceramic-based (disk or composite) and capacitive transducers. Most of the field
applications use the piezoceramic type.

AC-UT has the potential for being developed into a practical NDI tool for aircraft inspection for
which portable equipment is needed to inspect water-incompatible materials [48]. However,
because of the tremendous difference in transmitted and received signal amplitudes, and the
inherent difficulties in achieving adequate transducer/amplifier isolation and recovery, no current
air-coupled NDT systems work in single-probe mode; separate transmit and receive transducers
are always used. In the NDE of aerospace structures, where internal flaws and defects are to be
detected and imaged, the current state of AC-UT technology uses mainly transmission mode that
requires two-sided access, as shown in figure 111.

AC-UT, which transmits the interrogating UT wave into the test article without the use of a
liquid couplant, was applied to the composite honeycomb specimens in this study. A
commercially available UT system from Quality Material Inspection, Inc. (QMI) was used in
conjunction with a scanning frame and software produced by Sonix, Inc. The inspections were
made in through-transmission mode, as depicted in figure 111, using the hardware setup shown
in figure 112. In this case, there are separate sending and receiving transducers. As a result, it
should be noted that the AC-UT results presented here correspond to the capabilities of this
inspection method when: a) both sides of the structure are accessible, and b) it is possible to
fixture the probes such that they are in proper alignment. Cracks, delaminations, shrinkage
cavities, pores, disbonds, and other discontinuities that produce reflective interfaces can be
detected.

91
 R

QMI
Sonda-007CX Airscan

Figure 111. Schematic of air-coupled ultrasonic inspection of panel in through-

transmission mode

Figure 112. Equipment setup for air-coupled ultrasonic inspections of composite

honeycomb panels shown in background

In the through-transmission mode of deployment for AC-UT, the transducers are mounted on a
yoke for aircraft components that allow for two-sided access. With the aid of portable position
encoding devices and software in a laptop computer, C-scan images of AC-UT inspections can
be obtained in the field [35]. A summary of the AC-UT hardware and setup used in these
inspections is as follows:

92
• Scanner with UTEX Winspect software
• QMI SONDA 007CX pulser/receiver
• QMI focused UT probes
• Both 120 kHz and 225 kHz frequencies

The photos in figure 113 show the AC-UT test setup for both TTU and one-sided pitch/catch
mode. The AC-UT TTU method was applied to a fiberglass skin foam core specimen with the
flaw profile shown in figure 114. This specimen contains a wide array of flaw types found in
composite sandwich construction. The resulting C-scan image from the AC-UT TTU is shown in
figure 115. This image shows that the AC-UT TTU inspection revealed only the pull tab
(skin-to-core disbonds) and pillow insert (interplay delamination) flaws, along with the through
holes in the core (red dots), which acted as resonance cavities to produce high amplitudes in the
TTU signals. The grease contamination and micro-balloon (porosity) flaws were not detected.

Figure 113. AC-UT applied to a composite panel test specimen in both through-
transmission mode (left photo) and one-sided pitch/catch mode (right photo)

AC-UT scanning has been used in nondestructive imaging of composite honeycomb sandwiches
and repairs on such structures. A considerable fraction of composite components and control
surfaces on aircraft are honeycomb sandwiches. AC-UT can readily transmit through such
honeycomb structures. Very little sound energy can be transmitted through two face sheets
separated by air. Practically all the energy transmitted through a honeycomb sandwich has
propagated along the cell walls of the honeycomb core. This does not seem to hamper the ability
of AC-UT to image disbonds between the face sheet and the core or defects within the face
sheets. Figure 116 shows six disbond defects in an aluminum honeycomb sandwich with 0.032″
face sheets and 3/8″ cells. The image is obtained with focused 120 kHz air-coupled, 3/4″
diameter transducers.

93
 Figure 114. AC-UT TTU test setup on fiberglass skin and foam core test specimen

REF-STD-1-050-
TPI-1
Figure 115. AC-UT TTU C-scan image produced by 120 kHz inspection of composite foam
core test specimen

The set of composite panels from the CHE was examined using AC-UT. A commercially
available UT system from QMI was used in conjunction with a scanning frame and software
produced by Sonix, Inc. The inspections were all made in through-transmission mode using the
equipment setup shown in figures 111–113. This means that both face sheets of the honeycomb
sandwich are inspected simultaneously. Thus, flaws from both sides of the panel are
superimposed in the resultant AC-UT image, and there is no way to determine which face sheet
contains a particular indication. The C-scan images have a pixel size of 0.1″ and took
approximately 18 minutes to complete the 2.25 ft2 area. Each panel was scanned using a pair of
120 kHz planar probes. Some panels were also scanned using 120 kHz focused probes. The
transducers were placed at a distance of 1″ from the panels. Figure 117 shows a sample C-scan
produced by AC-UT TTU on a panel from the CHE. All flaws in the panel were detected. Note
that this includes the flaws located on the backside of the panel. This is due to the deployment of
the AC-UT method in TTU mode such that both sides of the panel were inspected
simultaneously.

94
Figure 116. AC-UT TTU C-scan showing disbonds in skin-to-core bondline of a composite
honeycomb panel

Figure 117. Sample C-scan data produced by AC-UT method on composite honeycomb test
specimen

95
4.11 LASER ULTRASONICS

LUS is generally defined as a technology in which one laser generates UT waves and another
laser coupled to a detection system detects the associated UT displacements [50–52]. There are
four main issues that have limited the adoption of LUS for the inspection of composites:

1. The lack of reliability of various prototypes used to validate the technology for
production
2. The acquisition cost of the LUS equipment
3. The small but significant differences between conventional and LUS signals
4. The unavailability of fieldable (portable) system for use in hangar environments, even as
gantry-based LUS systems for production environments have demonstrated excellent
results

4.11.1 LUS deployment

LUS is a non-contact technique that uses a scanning laser beam to quickly move across the part
in a uniform coverage pattern (see figure 118). Ultrasound is generated by pulsing the laser
beam, causing the surface layer to rapidly expand and contract through thermal expansion. The
absorbed laser energy is converted into heat in the top 10–100 µm of the surface. The resultant
temperature rise creates a local expansion of the material in the frequency of ultrasound
(1–10 MHz). Thus, a longitudinal UT wave is introduced into the part. Echoes from this wave,
when they again reach the surface, are sensed by a coaxial detection laser and converted to
images proportional to the echo strength. Laser light scattered off the surface is analyzed by an
interferometer to extract the UT signals that are “imprinted” on the laser as phase and frequency
modulations caused by the moving surface. The UT signals that are extracted are basically the
same as those obtained with conventional UT systems. The two laser beams can be indexed over
the material with a scanner to produce standard C-scan images [51].

Generation Laser
Composite
Composite
Detection Laser
LaserUT Signal

1.0
Amplitude [normalized]

0.5

Interferometer 0.0

-0.5

2D
2DScanner
Scanner Optics
Optics -1.0
-2.0 0.0 2.0 4.0
Time [us]
6.0 8.0 10.0

Figure 118. Schematic of laser UT system operation

This data transmission and acquisition do not require the laser beam to be deployed
perpendicular to the structure as in other UT methods. Thus, it is possible to scan complex parts
96
without detailed contour following. Ultrasound propagates perpendicular to the surface
regardless of the laser incident angle (up to + 45o). Currently, the laser UT systems are deployed
using a gantry system that provides a high-speed two-dimensional optical scanner to index the
beams over the part. This allows for rapid inspections and generation of the C-scan images.
Hand-scanning using UT can be slow and tedious, leading to human factors concerns with
respect to coverage and human vigilance. In addition, water-coupled UT can be difficult and
time-consuming to implement on complex-shaped parts. Figure 119 compares traditional UT
inspections to a laser UT interrogation. The first LUS systems mounted on robots [15] used
gantry-type robots. Optical alignment of the CO2 laser beam in the optical scanner must be
precisely maintained to obtain valid UT results. The CO2 laser cannot be efficiently transmitted
by optical fibers. Therefore, the most obvious solution is to move the CO2 laser along with the
optical scanner. This approach requires gantry robots because only this type of robot can move
equipment as large and heavy as an industrial CO2 laser. Gantry robots present several
disadvantages, the most important being the cost. The gantry robot is typically the single most
expensive element of an LUS system that includes such a robot. Several different deployments of
the laser UT inspection technique and a schematic showing the ability to inspect parts without
maintaining a perpendicular inspection orientation are shown in figure 120.

The LUS technology offers flexibility and faster inspection cycle time. However, even though
LUS signals are UT signals, there are some differences with UT signals obtained with
conventional UT transducers. Those differences must be understood when trying to adapt
processing and analysis techniques coming from decades of experience using piezoelectric
transducers. Key advantages of laser UT are:

1. The ability to scan quickly in a non-contact mode, all the way to the edge of a part
2. The ability to launch a through-thickness longitudinal wave even when the laser beam
impinges on the surface at an angle.

This means that the laser beam can be directed at the surface of complex shapes and scan them
efficiently without the need for contour following, complicated scan shoes, or angled water jet
arrangements. A laser beam can also be directed through apertures to scan the interior of a
structure. Disadvantages of laser UT include:

1. Sensitivity to surface coatings (variations in coatings can affect the strength of the UT
signal)
2. Maximum sensitivity, requiring tuning for each structure type
3. Need for system expertise/training to ensure alignment to produce uniform signal
4. Safety concerns necessitating personnel exclusion zones
5. Potential damage to the part surface from laser if not used with caution

97
Figure 119. Comparison of conventional and laser UT interrogation of components

Ultrasound propagates
perpendicular to surface
regardless of the laser
incident angle (up to + 45o)

Articulated robot system on a linear rail

Freq. Range = 0.5 MHz – 20 MHz

Figure 120. Schematic of laser UT method and deployment in gantry system and rail
system

98
4.11.2 iPhoton LUS (iPLUS)

The iPhoton LUS concept, called iPLUS™, was used to conduct the LUS inspections listed in
this report. An iPLUS III system is shown in figures 120 and 121. It uses a beam delivery system
mounted on an articulated robot. To increase the working envelope, the robot, beam delivery
system, and CO2 laser are mounted on a linear rail. The linear rail provides an almost unlimited
working envelope to the iPLUS system in one direction. Articulated robots provide flexibility not
possible with gantry-based approaches. Some applications require the inspection of composite
substructures inside larger structures, such as stringers inside a fuselage. Therefore, the iPLUS
configuration was developed as a response to these applications. In the iPLUS III systems, the
beam delivery system is composed of two standard beam delivery systems joined together on
axis 3 of the robot. This approach, combined with a cantilevered linear rail, provides more than
6 m of penetration inside a structure (e.g., a fuselage). For the inspection of parts, the iPLUS
scan head is positioned using the articulating robot. Once the scan head is in position, the
scanning is carried out by moving the laser beams along the surface of the sample using only the
two mirrors of the scanner. The scan area is defined by the angular movement, θx and θy, of the
scanner’s two mirrors. This process is illustrated in figures 121. When the scanning of one area is
completed, the robot moves the scan head to the next pose to scan the next area.

Figure 121. Inspection of a part using the iPLUS scan head and articulating robot

Figures 122–125 provide sample images produced by laser UT inspections on various composite
parts, some of which contain substructure elements. Note that surface and subsurface structural
details are imaged in the scan. The clarity of the flaws and sensitivity for flaw detection down to
0.25″ diameter are depicted in the C-scan images. Figure 122 shows the iPLUS LUS results from
a 16-ply, 12″ x 12″ composite laminate panel that was damaged by simulated hail impact. The
top left and right are the amplitude and time-of-flight C-scan images, respectively. The bottom
graphics show an A-scan and B-scan corresponding to the line in the top C-scans. Figure 125 is
just a flaw map produced by the LUS inspection as applied to a honeycomb panel in the CHE. It
can be seen that all but one of the smallest 0.25″ diameter flaws was detected.

99
Figure 122. iPLUS laser-ultrasonic scan of a 16-ply composite laminate with impact
damage

Figure 123. iPLUS laser-ultrasonic inspection of 0.111″ thick composite laminate test
specimen; photo on right shows the layout of the engineered flaws

100
 Figure 124. iPLUS laser-ultrasonic image of a composite part containing an inclusion as
highlighted

Figure 125. Sample results from laser UT inspection of 3-ply fiberglass panel

4.12 LAMINOGRAPHY

A motionless laminography system, developed by Digiray®, uses a reverse geometry scanning

X-ray source and an array of 64 small detectors to generate a sequence of laminographic slices
that may be reconstructed to produce three-dimensional X-ray images. The technology facilitates
an inspection through an object, layer-by-layer, to detect and pinpoint the location of a feature or
defect without moving either the object or the X-ray source. Laminography has been shown to
image defects in composites, porosity voids in adhesive bonds, and cracks and voids in metals.
NASA Ames has used laminography to inspect the thermal protection tiles for the X-37

101
development project. In trial tests with Boeing, the equipment was used to detect cracks caused
by foam impact on a reinforced carbon panel representative of the leading-edge heat shield on
the Space Shuttle wings.

Figure 126 illustrates the Digiray Reverse Geometry X-ray® (RGX) concept. The reverse
geometry shows the object close to the source instead of close to the detector as in conventional
systems. The result is that most of the scattered X-rays are not detected by the system. The single
detector may be replaced by an array of detectors that are simultaneously exposed to X-rays of
the object shown in figure 127. The diagram in figure 127 shows the X-ray source on the left. A
scanning beam generated in a raster pattern using a magnetically deflected electron beam inside
the X-ray tube produces X-rays. The fan beam is then transmitted through a thin anode window
and the adjacent object. When these X-rays impinge on the detectors, they produce an output
signal that enters an amplifier. An analog-to-digital converter digitizes the amplifier’s output.
The resulting digitized signals are then synchronized with the raster pattern generated in the
X-ray tube and stored in the computer’s memory. The result is a two-dimensional pattern of
picture elements associated with each detector in the array. The patterns are shifted and added
together to create a layer-by-layer view of the object. The photos in figure 128 show the
difference between a transmission X-ray on the left and laminography on the right. Using
laminography, the two sides of a U.S. quarter are shown with the image of the head separated
from the image of the tail.

Figure 126. Schematic of the reverse geometry laminography concept

102
Figure 127. Use of array of detectors and scanning beam to produce images of layers within
the part being inspected

Figure 128. Images highlighting the sensitivity of laminography and ability to produce
slices for improved detail recognition

The basic features of laminography are that it:

• Looks through an object layer-by-layer during inspection.

• Detects the location of defects rapidly.
• Eliminates most X-ray scattering (see figure 126).
• Produces high-resolution digital images.
• Captures 1,000 X-ray slices, each 1 mil thick, with a single exposure.
• Enables quick setup—alignment of tube and detectors is fixed.
• Requires structure to be removed from aircraft.
• Has contrast sensitivity < 0.3%—contrast ratio is 1500:1.

103
Digiray developed this equipment by combining:

1. An electronically scanned beam X-ray source placed close to the object and 64 small
detectors placed at a distance; this geometry is the reverse of conventional X-ray and
greatly reduces X-ray scattering effects, thus attaining sharper X-ray images
2. Multiple detectors (an array of 64 photomultipliers), each “viewing” the object from a
different angle
3. Special software that generates laminography without the need to move the object or the
detectors

The result of these features is a specialized X-ray inspection method that provides two-
dimensional spatial resolution of 12 microns; resolution in the third dimension of 8 mils; and
contrast sensitivity of 0.3%. In operation, the Digiray Motionless Laminography X-ray (MLX)
system displays one image after another of multiple layers of the object. In a single short X-ray
exposure, it captures 1,000 X-ray slices that are 8 mils thick. These slices may be displayed
individually or in a consecutive layer-by-layer motion picture. It also provides zoom
magnification capability without loss of resolution by electronically compressing the raster into a
smaller area. The reduced-size raster can then be moved about the face of the screen so that it
can pan within the field of view without using any mechanical means. The operator sets
instantaneous electronic controls to zoom and pan.

Figure 129 shows laminography slices of a Lucite wedge made up of seven steps ranging 1-4″. A
four-hole penetrameter is mounted on each step, with holes ranging 0.5–2T or 1–4% change in
density compared with the total density of each step. Each hole is visible on all seven steps.

Figure 129. Comparison of normal X-ray with laminography images showing ability of
laminography to detect damage in a 4″ thick Lucite part

104
Damage detection in composite honeycomb parts is shown in figures 130 and 131, which
highlight two of the flaw types found in the CHE. Machined core regions were used as one
method to produce gross disbond flaws between the skin and the core structure, while pillow
inserts were used to create interply delaminations in the skins. The test setup for the experiment
panels is shown in the photo contained in figure 132. Nine shots were used to construct a
complete image of the 2′ x 2′ panel where various flaws within the sample panel can be seen in
the laminography image.

Figure 130. Laminography slices of machined honeycomb core that simulate a disbond in
the fiberglass skin-to-core bond line

Figure 131. Laminography slices of interply delaminations in a 3-ply fiberglass honeycomb

panel

105
 MLX Detectors

Test Specimen

X-Ray
Tube

Figure 132. Laminography test setup and sample image showing all flaws detected in 3-ply
fiberglass panel

5. COMPOSITE HONEYCOMB FLAW DETECTION EXPERIMENT DESIGN

The CHE was developed to assess the performance of both conventional and advanced NDI
techniques in detecting voids, disbonds, delaminations, and impact damage in adhesively bonded
composite aircraft parts. A series of composite honeycomb specimens with statistically relevant
flaw profiles were inspected using both human tap test techniques and NDI equipment
introduced to automate and improve acoustic tap testing. The majority of the testing was in the
form of blind PoD studies while other portions of the testing determined signal-to-noise (S/N)
ratios from which flaw detection could be inferred. The primary factors affecting inspections
were incorporated into this study, including composite materials and construction type, flaw
profiles, mechanical interactions (impact and audible response), and subsurface damage and
environmental conditions. The critical phase of this effort used airline personnel to study PoD
performance in the field and formulate improvements to conventional inspection practices.

The test articles for this experiment were modeled after the general range of construction
scenarios found on commercial aircraft. A set of 44 composite honeycomb test specimens
containing engineered and natural flaws were manufactured. Flaws of various size, shape, and
type were randomly located across the specimen footprints. Details of the specimen construction
are as follows:

• Skin type: carbon graphite and fiberglass

• Skin thickness: panels have three-, six-, and nine-ply skins
• Core: 1″ thick Nomex core
• Paint: all panels are painted as per current aircraft specifications

106
The flaw types included in the test specimens are:

• Interply delaminations (pillow inserts).

• Machined core disbonds simulating an air gap skin-to-core disband.
• Pillow insert disbonds simulating tight contact but no adhesive strength (kissing disbond).
• Naturally formed impact damage (e.g., crushed core, disbonds, delaminations, broken
fibers).

One critical technical challenge required a parallel research effort to determine how to produce
representative flaws in composite honeycomb structure. For example, an extensive test series
was completed to produce impact damage that was subsurface only (core crush and/or backside
fiber fracture) and not accompanied by any surface demarcations.

The CHE traveled to airlines and third-party maintenance depots to acquire flaw detection data
as provided by qualified aviation inspectors. The experiment required approximately 2–3 days of
each inspector’s time. In general, inspectors were asked to locate and size hidden flaws in the test
specimens. The test program was intended to evaluate the technical capability of the inspector,
inspection procedures, and equipment (i.e., NDI method). The inspections emphasized flaw
detection methods applicable to composite honeycomb structures. During the experiment, each
blind inspection was preceded by inspections on appropriate reference standards supplied by the
experiment monitors. These feedback specimens contain all of the same construction and flaw
types as those found in the blind PoD tests specimens. The inspectors were given information on
the manufactured flaws present in the reference standards and were allowed to use these
specimens to ensure the proper operation of their equipment and allow them to become
comfortable with the inspection demands before moving on to the blind PoD specimens.
Figure 133 shows some of the composite honeycomb test specimens while figures 134 and 135
show the honeycomb reference standards. The same reference standard design was used for both
carbon and fiberglass skins, and repeated for three, six, nine, and 12 plies [53–55].

Side View Showing Nomex

Honeycomb Core Between
Two Laminate Skins

Figure 133. Test specimens used in composite honeycomb flaw detection experiment

107
 1.00" DIA BREAK CORNERS 1.000"
6X 4X

FIBERGLASS CORE
FIBERGLASS 3/16" CELL
3
SIZE
LAMINATE 4LB/FT CELL DENSITY
(FLAW SIDE)

FOAM TAPE
FIBERGLASS LAMINATE
FOR QA (UNFLAWED SIDE)
INSPECTION

11.000"
NOMEX CORE
3/16" CELL
3
SIZE
3LB/FT CELL DENSITY

5.500"
(ALL-AROUND)
.250"

EDGE SEAL
12.000"

MACHINE CORE (DEPTH OF 0.250" IN HONEYCOMB)

PILLOW INSERTS (LOCATED TWO PLIES UP FROM CORE MATERIAL)

POTTED CORE

CORE SPLICE

Figure 134. Schematic of composite honeycomb NDI reference standards showing

construction and flaw profiles

MACHINED CORE

HONEYCOMB
PILLOW INSERT
LAMINATE POTTED CORE

SEALANT
CORE SPLICE

FOAM TAPE
LAMINATE (18 dB LARGER THAN
12dB LARGER THAN UNFLAWED SITE)
UNFLAWED SITE 6dB LESS THAN
UNFLAWED SITE SHOULD UNFLAWED SITE
BE 18dB LESS THAN TAPE 12 dB LARGER THAN
UNFLAWED SITE
SEALANT

Figure 135. Cross-section of honeycomb reference standard design

108
These tests were conducted using NDI equipment that the inspectors are experienced in using for
this type of inspection. The effort focuses on understanding the factors influencing the
performance of NDI methods (device and inspector) when applied to the inspection of composite
honeycomb structures. The experiment results evaluated inspection performance attributes,
including accuracy and sensitivity (e.g., flaw hits, misses, false calls, and flaw sizing) and
usability features such as versatility, portability, complexity, and inspection time (human
factors). This report includes the results from the application of conventional honeycomb
inspection methods as well as the results from the application of advanced NDI methods.

5.1 EXPERIMENT DESIGN CRITERIA

The list of criteria that were used to design this experiment are as follows:

• Conventional and advanced inspection techniques are being assessed.

• Use information from completed composite reference standard study [53–56] to produce
the critical set of composite honeycomb panels (i.e., use only design variables that affect
NDI).
• Use knowledge of general composite inspections and tap tester responses to eliminate or
minimize additional variables to arrive at a workable number of test specimens.
• Sufficient number, type, and sizes of flaws are included to produce a statistically relevant
PoD analysis. Specimens contain sufficient unflawed regions to allow for assessment of
the probability of flaw detection as well as the likelihood of false calls.
• Flaw types to be studied: 1) interply delaminations, 2) gross skin-to-core disbonds, 3)
tight/subtle skin-to-core disbonds simulating tight contact but no adhesive strength
(kissing disbond), and 4) impact damage (crushed core, disbonds, delaminations, broken
fibers).
• Engineered flaws were placed on both sides of the honeycomb panels. This reduced
specimen costs (reduced material and fabrication time) and minimized the number of
panels to transport.
• Application of NDI:

‒ Blind application of techniques to study hits, misses, false calls, and flaw sizing;
blind tests will employ some procedural guidance to ensure uniformity of results.
The procedures will also study:

o Use of no grids on the specimens (inspecting “freehand”).

o Use of 1″ grid spacing on the specimens to control NDT coverage.
o Use of a 0.5″ grid spacing on the specimens to assess improvements
stemming from decreasing tap test spacing.

‒ Apply NDI devices with knowledge of flaw locations to determine quantitative

S/N ratios; flaw detection can be inferred by studying S/N levels.

• Conventional composite inspection techniques studied were: Boeing and Airbus manual
tap hammers, LFBT, HFBT (resonance), and MIA. In addition, 15 different advanced
NDI methods were evaluated in these experiments, including: Mitsui Woodpecker

109
 automated tap hammer, WichiTech DTH, CATT, thermography, shearography, PE-UT,
AC-UT, MAUS C-scanning in MIA and resonance mode, laser UT, shearography,
microwave, laser Doppler velocimetry, UT linear array, and laminography.

5.2 EXPERIMENT TEST SPECIMEN DESIGN

To implement a realistic experiment, it was necessary to design representative specimens that

include a full spectrum of variables found on composite aircraft structures. Another important
factor in the specimen design was to determine the most prevalent flaw types found on this type
of structure and develop ways to engineer representative flaws. This included determining the
various flaw sizes required for the statistical analysis.

While the size of flaw, or damage, that must be detected is affected by many parameters, the
general goal for composite inspections is to detect flaws that are 1″ diameter or larger. Many of
the NDI reference standards in OEM NDT manuals use 1″ diameter flaws to guide equipment
setup. In addition, the CACRC ITG members generally conclude that 1″ flaw detection provides
a good center point for this CHE SLE. Thus, the flaw sizes in the experiment design were
established with 1″ diameter at the center. Larger and smaller flaws were included such that PoD
values smaller than 1” (as small as 0.25″) and PoD values larger than 1″ (as large as 2″) could be
ascertained.

The list of criteria for the change of the test specimens is as follows:

1. Specimen types – modeled after range of construction found on aircraft; carbon graphite
and fiberglass skin over Nomex core.
2. Skin thickness – panels have three-, six-, and nine-ply skins.
3. Paint – all panels were painted as per current aircraft specifications.
4. Flaw types – statistically relevant flaw distribution, with sizes ranging from 0.2 in.2 to 3
in.2

a. Interply delaminations
b. Skin-to-core air gap disbonds (machined core disbonds simulating presence of air
gap)
c. Skin-to-core “kissing” disbonds (pillow insert disbonds simulating tight contact
but no adhesive strength)
d. Impact damage – crushed core, disbonds, delaminations, broken fibers

5. A minimum of 2″ separation between flaws is maintained to eliminate signal cross-talk;

include a few flaw pairs that are closely clustered to study ability to define boundaries of
flaws.
6. A minimum of 1.5″ distance from flaws to edge of panels is maintained; includes a few
instances of flaws close to edge to study flaw detection near a natural edge.
7. Desired flaw size to detect is 1″ diameter; minimum flaw size is half this value
(0.5″ dia.), maximum flaw size is twice this value (2″ dia.). The 0.5″–2.0″ dia. flaws were
divided into five discrete sizes—0.5″, 0.75″, 1.0″, 1.5″, and 2.0″—and distributed
throughout the specimen set. One large 3″ dia. flaw was added to provide a flaw that
should be found by all techniques.
110
8. Proper flaw spacing, including the need to produce sufficient unflawed regions to
determine false call rates, required 44 specimens that are 18″ W x 18″ H.
9. Flaw engineering – The methods used to engineer controlled flaws include “loose”
disbonds (machined core), “tight” or kissing disbonds (pillow inserts placed between
honeycomb core and adhesive), impact damage (fracture, delamination/disbond, crushed
core), and delaminations (pillow inserts placed between adjacent composite plies). In
various circumstances, the engineered flaws may be more or less difficult to detect than
actual aircraft flaws. A three-year study was completed to develop methods for producing
repeatable, realistic flaws.
10. To provide the proper challenge to the tap test and artificially showing NDI in a poor
light, it was decided to add impact flaws to the specimen matrix. The impact panels
provide “natural” disbonds and crushed core damage scenarios to the experiment. A
focused impact study on representative honeycomb sandwich structures was used to
guide the impact levels such that subsurface damage could be achieved without creating
surface demarcations. Post-impact TTU was used to characterize the flaw profiles in the
test specimens.
11. Impact specimens – Two impact panels were added for each of the six- and nine-ply
families. Nominal impactor sizes of 0.75″, 1″, 1.5”, and 2″ diameter were used to create
representative flaws of different sizes. As the impact panels represent normal
manufactured structure, the laminate and honeycomb core were co-cured in an autoclave
at elevated pressures (45 psi versus the 11 psi panels with engineered flaws that represent
worst-case repair cures). Figure 136 shows the different impactors along with the drop-
table apparatus that was used to produce repeatable, controlled impact energy for the
impact panels. Figure 137 shows one of the impact calibration tables that was produced
using trial impacts on duplicate composite honeycomb panels. Similar curves were made
for fiberglass and carbon skin panels of three-, six-, and nine-ply thickness. Then, these
predetermined energy levels were used to produce the actual impact damage on the
experiment honeycomb panels. Sample damage produced by the impacts are shown in
figures 138 (fiber and matric cracking) and 139 (crushed core and disbonds).

Figure 136. Impactors and impactor drop-table apparatus used to produce damaged area
on test panels

111
 Impact Damage: Severe
3” Diameter Steel Impactor
5 Energy Levels

70.6 in-lbs

Energy (in-lbs)

50.6 in-lbs

Barely Visible
32.5 in-lbs

21.6 in-lbs

10.8 in-lbs • Carbon skin

• 3/8” cell size
• 1” honeycomb core

Flaw Size Produced

Figure 137. Sample results from calibration shots used to determine energy levels needed to
produce damage in composite honeycomb panels without producing surface demarcations

Low Energy Impact

0
0
+ 45
-45
0
90
90
0
-45
+ 45
0
0
P id P tt M ti C kf i t
Figure 138. Pyramid pattern matrix crack produced by impact of composite skin

Figure 139. Cross-section of sample impact damage in composite honeycomb experiment

panels showing crushed core but no skin deformation

12. Specimen reduction – The specimen matrix was reduced by determining the core type
that provides the most difficult flaw detection for the tap test: Nomex. Nomex is softer
than fiberglass, thus its loss via a disbond will produce a relatively smaller change in

112
 stiffness. A series of tap tests were conducted using the Woodpecker device on an array
of honeycomb specimens. The results are shown in tables 2 and 3. With the exception of
a single anomaly, disbonds in the Nomex cores displayed a lower signal and were more
difficult to detect. Note that in the thicker plies, the loss of core has less effect on the
stiffness so the signal change is less. Thus, it was determined that the specimen set would
only include Nomex core. It is expected that the PoD results for fiberglass core will be
slightly better than that calculated for Nomex core.

Table 2. Digital tap test results from Mitsui Woodpecker comparing flaw detection in
Nomex and fiberglass cores

3-Ply Carbon Skin* 3-Ply Fiberglass Skin*

Pillow Insert Machined Pillow Insert Machined
Core Type Flaw Core Flaw Flaw Core Flaw
Nomex 41/56 (15) 41/73 (32) 41/63 (22) 41/81 (40)
Fiberglass 35/51 (16) 35/70 (35) 36/52 (16) 36/80 (44)

6-Ply Carbon Skin* 6-Ply Fiberglass Skin*

Pillow Insert Machined Pillow Insert Machined
Core Type Flaw Core Flaw Flaw Core Flaw
Nomex 35/38 (3) 35/42 (7) 34/41 (7) 34/45 (11)
Fiberglass 34/37 (3) 34/41 (7) 31/38 (7) 31/43 (12)
* Information presented as “Baseline Signal/Flaw Signal” with differential shown in parentheses

113
 Table 3. Digital tap test results from Mitsui Woodpecker showing relative stiffness values
for different cores

6-Ply Fiberglass Skin

Core Type Signal from Unflawed Region*
No Core 65
Nomex Core 35
Fiberglass Core 32.5
* Lower number (shorter pulse duration) = stiffer core
Softer core = harder to detect loss of core (harder to find disbonds)

13. Final specimen matrix – The final test specimen matrix is shown in table 4. The two
specimen types produce a total of 44 specimens. Table 4 shows the engineered and
“natural” impact specimen combinations for the following variables: skin type (carbon,
fiberglass), skin thickness (three, six, and nine plies), honeycomb type (Nomex), and
damage type (engineered disbonds and delaminations, and impact flaws). Notice that the
number of specimen configurations was reduced by applying several NDI assumptions
that are described in the following section. This allowed us to have a manageable number
of specimens in the study. A summary of the test specimen configurations follows:

a. Engineered honeycomb test specimens:

i. 6 specimens x 3 configurations (3, 6, and 9 plies) = 18 specimens for

carbon skin with Nomex core
ii. 6 specimens x 3 configuration (3, 6, and 9 plies) = 18 specimens for
fiberglass skin with Nomex core

b. Natural impact test specimens:

i. 2 specimens x 2 configurations (6 and 9 plies) = 4 specimens for carbon

skin with Nomex core
ii. 2 specimens x 2 configurations (6 and 9 plies) = 4 specimens for fiberglass
skin with Nomex core

114
 Table 4. Test specimen matrix for composite honeycomb flaw-detection study

Engineered Specimens
Resulting Number
Skin Number of Plies Honeycomb Core
of Configurations
Carbon 3, 6, 9 Nomex 3
Fiberglass 3, 6, 9 Nomex 3
6 configurations x 6 each =
36 specimens total
Flaws = MC disbonds, PI disbonds, PI delaminations.
Flaw Sizes = ½", 3/4", 1", 1.5", 2.0" diameter.
Flaw Numbers = 51 per configuration; spread over 6 specimens to allow for assessment of false calls.

“Natural” Impact Specimens

Resulting Number
Skin Number of Plies Honeycomb Core
of Configurations
Carbon 6, 9 Nomex 2
Fiberglass 6, 9 Nomex 2
4 configurations x 2 each =
8 specimens total
Flaws = ID impact disbonds, IC impact crushed core.
Flaw Sizes = as generated by impactors with 0.75", 1", 1.5", and 2.0" diameter.
Flaw Numbers = 8 per configuration, concentration on 1" diameter requirement.

14. Methods for engineering flaws – Inspection results from a custom, preliminary suite of
specimens were used to determine the best way of producing realistic and reliable
skin-to-core disbond and interply delamination flaws. A series of inspections were carried
out on trial standards that were manufactured with various candidate methods for
engineering delamination flaws. Figure 140 shows the engineering drawing for one of
these honeycomb specimens for evaluating flaw insertion methods. One carbon specimen
and one fiberglass skin specimen were produced with this flaw layout. The three methods
employed to engineer the delamination flaws were: 1) pillow insert consisting of Kapton
tape around four layers of tissue paper; 2) brass shims coated with a silicon mold release
to prevent bonding to the plies; and 3) Teflon disk inserts. Each flaw method was used to
generate three like delamination flaws to test for repeatability as well as to statistically
determine the amount of NDI signal disruption generated by the flaw method. Note also
that the trial specimen includes potted core and core splice areas. To expand the
utilization of these standards, potted core and core splice areas were included as a tool to
aid in the interpretation of NDI signals. This will help minimize false calls caused by the
presence of potted cores or core splices that can alter NDI equipment readings.

115
 SKIN TO CORE DISBONDS
DELAMINATION INSERTS (PULL TABS)
CORE SPLICE HONEYCOMB
POTTED CORE CORE
DISBONDS

10.000"
20.000"

COMPOSITE LAMINATES
PILLOW INSERTS TEFLON INSERTS BRASS SHIMS

POTTED CORE CORE SPLICE

MACHINED CORE DISBONDS PULL TAB

Figure 140. Engineering drawing to evaluate honeycomb reference standard design and
fabrication

a. Disbonds – Machining the honeycomb core (recessing) away from the laminate
provides the best way of producing reliable skin-to-core disbond flaws. This
method also produces flaw sites that can support tap testing. Pull tabs were also
included as an alternate disbond engineering method; however, the required
proximity of the pull tab flaws to the edge of the specimen produces boundary
condition effects in some NDI methods and limits the value of these types of
flaws.
b. Delaminations – The goal was to realistically and repeatably produce interply
delamination flaws. The methods were repeated multiple times to assess
consistency. The methods employed to engineer the delamination flaws were as
follows: 1) pillow insert; 2) brass shims coated with a silicon mold release agent;
and 3) Teflon inserts (individual inserts with 0.003″, 0.005″, and 0.008″
thicknesses, and two plies of stacked 0.003″ inserts). Figure 141 shows how the
pillow inserts were fabricated. Experiments determined that the four layers of
tissue paper are needed to produce a uniform and repeatable interruption (i.e.,
flaw indication) of an interrogating NDI signal.

116
 E
dia ffec
m tive
et
er
t

Tape 'adhesive' faces

to be in contact
t= .050" [1.3]

Four plies of
.002" [.05] thick
tissue paper

Two layers of
.002" [.05] thick
heat resistant tape
(e.g. Kapton tape)

Figure 141. Construction of pillow insert for delamination flaws

c. Potted core and core splice (used as reference not calibration) – The trial
specimen included potted core (BMS 5-28 Type 7 material) and core splice (AF-
3028 material) areas. To expand the utilization of these standards, potted core and
core splice areas were included as a tool to aid the interpretation of NDI signals.
A process was developed wherein the potting material was placed in a vacuum
chamber just prior to insertion in the core cells with a syringe. This produced a
uniform potting, and no porosity was visible in subsequent TTU inspections.
Figure 142 shows the process for creating a potted core area in the honeycomb. A
profile of a machined core region to create disbonds is also shown in this
schematic.

117
 1.00" [25.4] DIA.

FILL IN CELLS TO APPROXIMATE 1.00" [25.4]

DIAMETER (FILL CONSERVATIVELY TO ASSURE AT
LEAST 1.00" [25.4] POTTED AREA)

CELLS FILLED TO TOP WITH

CORE POTTING MATERIAL
TOP VIEW

SYRINGE TO INJECT MATERIAL

INTO CELLS

CELLS TO BE FILLED WITH

POTTING MATERIAL
MACHINED CORE

CELL TOP

HONEYCOMB
CELL BOTTOM

SIDE VIEW TAPE (OPPOSITE SIDE

FROM MACHINE CORE)

Figure 142. Process for creating potted honeycomb core areas

TTU was used to evaluate the flaw engineering methods as it has the best resolution and the
ability to provide quantitative signal attenuation values. TTU results for the trial honeycomb
panels (see figure 140 for design drawing) are shown in figure 143. All of the flaws can be
clearly seen; however, the Teflon inserts and brass shim inserts, second and fourth row down,
respectively, are smaller than their original 1” diameter area. This is probably due to ingress of
the adhesive around the perimeter of these inserts and an associated coupling of the UT signal.
As a result, attenuation levels in the Teflon and brass shim areas are less than those measured in
the pillow inserts regions. Furthermore the pillow insert regions were able to retain their
1″ diameter size. The machined core disbond areas were also very consistent with uniform signal
attenuation levels. Finally, the potted core regions shown in the top row provided enhanced
signal coupling and an associated decrease in UT attenuation through the panel. This is expected,
and the results show a very uniform and repeatable signal level to support NDT calibration.

Two other laminate skin specimens were produced to complete the evaluation of various
delamination inserts. TTU results from scanning these two skins are shown in figures 144 and
145. This exercise determined that the use of multiple layers (i.e., four plies of tissue in pillow
inserts and two plies of Teflon stacked together) produce better signal attenuation and more
repeatable flaws than single-ply inserts, even if the single plies are thicker. These results indicate
that the best method(s) to engineer delaminations are: 1) pillow inserts with four plies of tissue,
and 2) Teflon inserts with two to three plies of 0.003″ material. Measured signal attenuation
through these disbond flaws ranged from 26–46 dB. BondMaster and S-9 inspections also
support the use of these flaw insertion methods. A minimum of 18 dB attenuation is required at
the flaw sites.

118
  Pull Tabs

 Potted Core

 Teflon Inserts

 Pillow Inserts

 Brass Shims

 Machined Core

 Core Splice
Fiberglass Skin Carbon Skin
Figure 143. TTU inspection of honeycomb panel—assessment of methods to engineer
repeatable flaws

Figure 144. TTU C-scan showing signal attenuation produced by inserts; tests on 6-ply
fiberglass laminate

119
 48dB 50dB 54dB 54dB 54dB
Pillow Insert 0.013”
24 dB 24 dB 26 dB 32 dB 27 dB
Teflon (3mil; 2 plies) 0.006”
20 dB 27 dB 23 dB 21 dB 18 dB
Separator Ply (2 plies) 0.006”
54 dB 54 dB 54 dB 54 dB 54 dB
British Aero Pillows 0.029”
11dB 11dB 11dB 11dB 10 dB
Bubble Backing Ply (1 ply) 0.013”
4 dB 4 dB 3 dB 3 dB 3 dB
DASA Shrink Film (1 ply) 0.002”
9 dB 18 dB 9 dB 9 dB 9 dB
DASA Shrink Film (2 plies) 0.004”

Figure 145. TTU C-scan showing signal attenuation produced by inserts; tests on 6-ply
fiberglass laminate (numbers in right hand column represent thicknesses of inserts)

15. Test specimen characterization—to determine if the engineered flaws are adequate and
able to be detected, a comprehensive flaw characterization effort was conducted. In one
set of tests, TTU inspections were completed on a set of composite honeycomb test
specimens. Figure 146 shows one sample TTU image in which signal attenuation values
were ascertained to make sure that each flaw produced at least 12 dB of signal loss. A
second flaw characterization step involved the use of S/N ratios to infer whether or not a
particular flaw should and could be detected by normal NDI means.

To intercompare the results from different NDI methods that use different indicators to
infer the presence of defects, each inspection measured the S/N of each defect versus the
surrounding good structure. The noise level was determined by examining the output
variation corresponding to inspections along adjacent sections of good structure. This was
compared to the signal obtained during inspections of the flawed areas.

BS = base signal; peak signal at unflawed area

NS = noise signal; (max-min)/2 over range of
unflawed area in each quadrant
FS = flaw signal; peak signal at each flaw site
S/N = signal-to-noise ratio

FS − BS
S/N = (2)
NS

120
Data from equipment providing digital signal values (e.g., Woodpecker, S-9 Sondicator) were
logged directly into a data acquisition computer. Several different inspection methods were
applied to each flaw that was used in the CHE so that numerous S/N values for each flaw could
be calculated. Normally, an inspector will look for S/N values in the vicinity of 3 to make a flaw
call. Thus, the S/N assessment conducted here used an S/N = 3 as the threshold for determining
the suitability of including an engineered flaw in this study. All flaws were determined to
produce S/N values in excess of 3, so it was not necessary to exclude any flaws from the
experiment. Testing using this scheme did not require calibration on a “median” or “neutral”
reference standard. The key measurement for each case was the difference between unflawed
areas of the test panel and the defect area. Sample data from S/N testing from different flaw
types using the MIA mode of inspection are presented in table 5.

Figure 146. Specimen characterization—TTU C-scan of honeycomb test panel used to

verify fabrication and location of engineered flaws

121
 Table 5. Sample S/N data from various flaws demonstrating the viability of each flaw for
use in the composite honeycomb flaw detection experiment

Specimen Flaw Equipment Mean at Std. Dev. at Peak Value Signal-to-

Flaw Type
No. No. Setup Null Area Null Area at Defect Noise
6CI-1 307 1.046" IM Fine Gain = .96 80.00 5.00 20.00 12.00
6CI-1 308 1.118" IM Fine Gain = .96 80.00 5.00 16.00 12.80
6CI-1 309 0.658" IM Fine Gain = .96 80.00 5.00 55.00 5.00
6CI-2 313 1.490" IM Fine Gain = .96 80.00 4.00 9.00 17.75
6CI-2 314 1.078" IM Fine Gain = .96 80.00 4.00 14.00 16.50
6FI-1 315 0.802" IM Fine Gain = .90 80.00 4.00 40.00 10.00
6FI-1 316 1.810" IM Fine Gain = .90 80.00 4.00 11.00 17.25
6FI-1 317 1.002" IM Fine Gain = .90 80.00 4.00 46.00 8.50
6FI-1 318 1.730" IM Fine Gain = .90 80.00 4.00 20.00 15.00
6FI-1 319 0.814" IM Fine Gain = .90 80.00 4.00 39.00 10.25
6FI-2 322 1.060" IM Fine Gain = .92 80.00 5.00 45.00 7.00
9CI-1 323 1.040" IM Fine Gain = .58 80.00 6.00 39.00 6.83
9CI-1 324 1.980" IM Fine Gain = .58 80.00 6.00 10.00 11.67
9CI-1 325 0.790" IM Fine Gain = .58 80.00 6.00 37.00 7.17
9CI-1 326 1.042" IM Fine Gain = .58 80.00 6.00 26.00 9.00
9CI-2 330 1.038" IM Fine Gain = .58 80.00 5.00 20.00 12.00
9FI-1 331 1.610" IM Fine Gain = .54 80.00 4.00 50.00 7.50
9FI-1 332 0.536" IM Fine Gain = .54 80.00 4.00 63.00 4.25
9FI-1 333 1.760" IM Fine Gain = .54 80.00 4.00 49.00 7.75
9FI-2 336 0.786" IM Fine Gain = .54 80.00 5.00 55.00 5.00
9FI-2 337 1.020" IM Fine Gain = .54 80.00 5.00 58.00 4.40
9FI-2 338 1.710" IM Fine Gain = .54 80.00 5.00 37.00 8.60

Specimen Flaw Equipment Mean at Std. Dev. at Peak Value Signal-to-

Flaw Type
No. No. Setup Null Area Null Area at Defect Noise
6F-5 188 2.00" MC Fine Gain = 1.06 80.00 6.00 19.00 10.17
6F-5 189 1.40" PI-DE Fine Gain = 1.06 80.00 6.00 13.00 11.17
6F-5 190 1.00" PI-DE Fine Gain = 1.06 80.00 6.00 42.00 6.33
6F-5 191 0.50" MC Fine Gain = 1.06 80.00 6.00 55.00 4.17
6F-5 192 1.00" MC Fine Gain = 1.06 80.00 6.00 20.00 10.00
6F-5 193 1.40" MC Fine Gain = 1.06 80.00 6.00 19.00 10.17
6F-5 194 2.00" PI-DE Fine Gain = 1.06 80.00 6.00 7.00 12.17
6F-5 195 0.75" MC Fine Gain = 1.06 80.00 6.00 32.00 8.00
6F-6 196 0.75" MC Fine Gain = 1.02 80.00 5.00 36.00 8.80
6F-6 197 0.75" MC Fine Gain = 1.02 80.00 5.00 34.00 9.20
6F-6 198 1.00" PI-DE Fine Gain = 1.02 80.00 5.00 9.00 14.20
6F-6 199 0.50" PI-DE Fine Gain = 1.02 80.00 5.00 35.00 9.00
6F-6 200 1.40" PI-DI Fine Gain = 1.02 80.00 5.00 27.00 10.60
6F-6 201 3.00" MC Fine Gain = 1.02 80.00 5.00 18.00 12.40
6F-6 202 0.50" PI-DI Fine Gain = 1.02 80.00 5.00 55.00 5.00
6F-6 203 1.40" MC Fine Gain = 1.02 80.00 5.00 20.00 12.00
6F-6 204 0.75" PI-DI Fine Gain = 1.02 80.00 5.00 56.00 4.80

122
6. COMPOSITE HONEYCOMB FLAW DETECTION EXPERIMENT IMPLEMENTATION

The CHE traveled to airlines, third-party maintenance depots, and NDI developer labs to acquire
flaw detection data. A sufficient number of inspectors performed the blind tests to produce
statistically valid PoD curves representative of the industry as a whole. Testing on all
conventional and advanced NDI methods was completed, and the data were analyzed to produce
a final, comprehensive evaluation of composite honeycomb inspections. A total of eight to ten
inspectors deployed each conventional NDI technique to arrive at the final overall performance
assessments. The experiment was deployed in maintenance hangars to provide representative
inspection environments and inspection impediments such as poor lighting, noise, or distractions.
Although the experiment made no attempt to uncouple the effects of these variables, it did
include inspectors of different age and experience levels. The experience level of the
experimenters was varied as a result of the volume of participants, but this variable was not
controlled.

The test program was intended to evaluate the technical capability of the inspection procedures
and the equipment (i.e., NDI technique). Evaluation of inspector-specific or environment-
specific factors associated with performing this inspection was not the primary objective of this
experiment. However, notes were taken by the experiment monitor when such factors seemed to
influence results or when unplanned events occurred that could impact the results of the
inspection. Specific notice was taken if issues such as deployment or maneuverability adversely
affected the outcome of the inspection.

All inspectors were provided with the appropriate inspection procedures from the Boeing and
Airbus maintenance manuals. While these procedures contain specific deployment information,
such as the recommended tap spacing, the inspectors were allowed to interpret and apply the
procedures as they saw fit. Specific deployment of the inspection devices was not enforced by
the experiment monitors. Use of and adherence to OEM recommended procedures were noted by
the experiment monitors. Each blind inspection process was preceded by inspections on
appropriate reference standards supplied by the experiment monitors. The inspectors were given
information on the manufactured flaws present in the reference standards and were allowed to
use these specimens to ensure the proper operation of their inspection equipment. The reference
standards have the same construction as the blind test specimens and include similar flaws. Thus,
they also can be used to allow inspectors to become familiar with an inspection device and learn
about specific equipment’s response to various composite structures and flaws within those
structures.

A set of experiment protocols were written to guide the implementation of the CHE. The
experiment protocols ensured that the information provided to all experiment participants was
consistent and comprehensive, such that all participants received similar guidance and inspection
aids. The experiment protocols also provided step-by-step guidance to the experiment monitors
so that all data and observations associated with the tests were acquired in a consistent manner.
Appendix A contains the “Experiment Briefing and Information Packet” for the CHE. This
information was sent to each inspector in advance of the experiment and also reviewed with the
inspectors during airline briefing sessions, which were completed at each site prior to beginning
the NDI tests. The set of Honeycomb NDI Reference Standards, with flaw locations clearly

123
marked, were also sent out in advance so that experiment participants could conduct inspections
to familiarize themselves with the composite structure and flaw detection requirements.

Prior to conducting the tests, all experimenters were given a briefing on the CHE.

Figure 147 shows one of the briefings being provided to inspectors. This briefing explained the
purpose of the experiment and the process the inspectors would be using to indicate their flaw
findings. The briefing was used at each facility to ensure a consistent presentation on the
experiment goals and a thorough explanation of how the experiment would proceed. It also
allowed the inspectors to ask questions. At this time, the inspectors were introduced to the
inspection devices that they could optionally use. Inspectors could also decide to deploy their
own composite honeycomb inspection equipment and transducers.

Figure 147. Experiment instructions being provided to supplement the written

experimenter briefing and information packet

124
Guidance and instructions provided to inspectors were:

• Inspection coverage should be 100% of the panel with the exception of a small 1″ band
around the perimeter of the panels where edge effects may create problems.
• The Composite Honeycomb Reference Standards, or equivalent, should be used to set up
the equipment. Minor equipment adjustments stemming from in-situ calibration on the
parts being inspected are allowed.
• Inspectors should draw the entire size/shape of the flaw (i.e., delineate the edges).
• Reference standards should be used as an aid to determine where to make flaw call edges.
This is based on the diameter of the probe and how much of the probe needs to be over
the flaw to react/detect.
• Inspectors do not need to determine the type of flaw—just the location, size, and shape of
the suspected anomaly.
• Inspectors should ignore any visual clues (surface anomalies in the paint or small surface
marks) and avoid using these as flaw detection aids. It was indicated that such anomalies
may be intentionally planted to add complexity to the inspection. Guidance to the
inspectors instructed them to indicate only those flaws detected by their inspection
device.

The inspections were conducted on the series of 18″ x 18″ panels described in section 5. The
panels were placed on a foam pad, as shown in figure 148, to produce uniform boundary
conditions across all experimenters. Experimenters were asked to work at a pace that is
comfortable for them. Although experiment monitors did note start and stop times for each panel
inspection, time-to-inspect was a secondary variable of the experiment. Inspectors were
instructed to take whatever time was necessary to ensure that any and all flaws in the test
specimens were found. Experiment monitors logged observations using the data sheets shown in
appendix B. For example, monitors noted: 1) the tap spacing used by the inspector; 2) whether
visual clues were used by the experimenters to aid in flaw detection; and 3) whether use of and
adherence to OEM recommended procedures were used. An array of acceptable tap test devices
(see official Boeing and Airbus tap hammers in section 3) were provided, and the inspectors
chose one. This eliminated the use of flashlights, washers, and other devices that do not meet the
size and weight criteria (see figures 149 and 150).

125
 TEST SPECIMEN

FOAM FRAME

CENTER TEST SPECIMEN ON FOAM FRAME

Figure 148. During inspections, each panel was supported around its perimeter by a foam
frame to provide uniform boundary conditions

Addition of wood piece

Figure 149. Samples of homemade tappers used in the field and modification to
Woodpecker device to ensure that automated tap is perpendicular to the surface

126
Figure 150. Sample tap testing at maintenance depots and alternative tappers identified in
study

A total of 75 inspectors from 22 airlines and maintenance and repair organizations located
around the world participated in the experiment. The maintenance facilities included Delta Air
Lines (four facilities), United Airlines (three facilities), American Airlines (two facilities), Aloha
Airlines, Alaska Airlines, US Airways, British Airways, KLM Airlines, Air France, FedEx (two
facilities), SR Technics, GKN Westland Aerospace, and FLS Aerospace Manchester. The
participating companies, including those that deployed advanced NDI methods, are shown in
figures 151 and 152. The typical setups for the experiment deployment are shown in
figures 153–156, where each inspector has a workstation on which to set up their equipment and
test specimens. Inspectors were asked to inspect each test specimen and provide any information
they could about the presence of applicable flaws. If they determined that flaws were present,
they then provided size and shape information about each detected flaw. The results were
marked directly on the test specimen using markers provided by the experiment monitors. Figure
157 shows sample sets of flaw marks on one of the honeycomb test specimens along with grid
marks that some inspectors used to guide the surface coverage of their inspection transducers.

Inspector feedback on performance provided excellent training for the experimenter, while their
results produced a valuable baseline for how well the industry is able to inspect composite
honeycomb structures using conventional NDI methods (flaw hits/misses, false calls, flaw sizing,
effects of construction scenarios, and effects of environment). Furthermore, results from the 18
advanced NDI methods allowed the team to quantify the degree of inspection improvements
possible via the application of more sophisticated inspection methods and procedures. The
experiment results also allowed the experimenters to determine which NDI methods possessed
unique capabilities to address specific inspection requirements.

127
Figure 151. Airline and aircraft maintenance depot participants in composite honeycomb
flaw detection experiment

Figure 152. Advanced NDI companies that participated in composite honeycomb flaw
detection experiment

128
Figure 153. Sample implementation of experiment in Alaska Airlines’ hangar environment

Figure 154. CHE being implemented at American Airlines’ maintenance facility

129
Figure 155. Typical experiment setup with separate inspector—
Airbus facility hosting multiple airlines

130
 Figure 156. Inspectors completing inspections and marking flaw detections on the test
specimens—clockwise from upper right: Airbus tap hammer, Boeing tap hammer, DTH,
LFBT, Woodpecker, and MIA

131
 Figure 157. Honeycomb panel with sample grid markings used by inspector during
inspection of panel along with flaw markings within the grid

The CHE was deployed with conventional NDI methods for composite honeycomb structures
and also completed using a wide array of advanced NDI methods. The conventional and
advanced NDI methods evaluated with these tests included: conventional tap test (Boeing- and
Airbus-designated tap hammers), automated/instrumented tap test (Mitsui Woodpecker,
WichiTech DTH, and CATT), HFBT/resonance (S-5, S-9, Sondicator, BondMaster), LFBT
(BondMaster), MIA (MIA-3000, V-95 Bondcheck, BondMaster), scanning in resonance and
MIA mode (Boeing MAUS), Thermography (TWI EcoTherm), shearography (Laser Technology

132
Inc. LTI-5200), microwave (Evisive microwave scanner), laser Doppler velocimetry
(SAM system), linear array UT (Boeing FlawInspecta Linear Array), AC-UT (QMI SONDA
007CX), LUS (iPLUS), and laminography (Digiray RGX). This report describes the PoD results
from both the conventional and advanced NDI methods.

The overall performance attributes that were evaluated include: 1) accuracy and sensitivity (e.g.,
hits, misses, false calls, and sizing), 2) versatility and complexity (e.g., human factors), 3) data
analysis capabilities, 4) portability, and 5) inspection time. Test results were graded to evaluate
the accuracy of quantitative measurements and assess qualitative measurement parameters. Each
inspector’s flaw calls were used to identify hits (calls with any amount of overlap between the
call and the solution), misses (no call for an area of a known flaw), false calls (call with no
overlap of a flaw), and the degree of overlap between experimenter calls and actual flaw areas.
Typical results include PoD curves, tabulated flaw detection and false call data, and scatter charts
relating accuracy in delineating flaw size and shape. All of the data were sorted by the attributes
of the structure being inspected. Figure 158 is a grading parameter drawing that shows how the
hits/misses/false calls results were to be graded. The percentage of flaw covered was another
variable of primary interest. Errors in lateral extent of flaw and maximum linear extent of
overcall were variables of secondary concern and not considered part of the current grading plan.

NOTES:
1. THE CIRCLES REPRESENT THE FLAWED AREAS AND THE SHADED
AREAS REPRESENT THE AREAS CALLED AS FLAWS.

PERCENTAGE COVERED
< 25% 25% - 49% 50% - 74% 75% - 99% 100%
1 2 3 4 5

NOTE: ALL THESE ARE CONSIDERED HITS (YES-DETECTED)

FLAW SIZING DIAGRAM

Figure 158. Schematic showing the grading categories comparing experimenter flaw calls
with actual flaw information

133
7. RESULTS FROM COMPOSITE HONEYCOMB FLAW DETECTION EXPERIMENT

Each inspection technique that was applied in this blind flaw detection experiment was evaluated
using the following performance attributes: 1) accuracy and sensitivity, 2) data analysis
capabilities, 3) versatility, 4) portability, 5) complexity, 6) human factors, and 7) inspection time.
The most important of these parameters are the quantitative metrics, as they are objective
standards that can be numerically counted or quantified. Accuracy is the ability to detect flaws
reliably and correctly in composite structures and repairs without overcalling (false calls).
Sensitivity is the extent to which the inspection system responds to flaws as a function of size,
type, and location in the structure (e.g., proximity to repair edges, underlying or adjacent
structural elements). During the tests, the inspectors attempt to optimize their accuracy and
sensitivity without producing a significant number of false calls. More than 70 inspectors from
airline maintenance facilities and NDI developers around the world participated in the blind tests.
The results were used to establish an industry inspection baseline indicating the current
performance of aviation maintenance practices.

To acquire flaw detection data, the test panels were shipped to airlines, third-party maintenance
depots, and NDI developer labs around the world. The inspectors performed the blind tests to
produce statistically valid PoD curves that were representative of the industry as a whole. At
aircraft maintenance depots, inspectors of different ages and experience levels (see table 6)
participated in the experiment. In addition, 14 inspectors applied 11 different advanced NDI
methods (see table 7) to determine PoD performance. The NDI techniques that were evaluated
ranged from manually applied tap testing (e.g., Airbus tap hammer and Boeing tap hammer) to
an automated, sensor-based form of tap testing (e.g., Woodpecker, DTH, and CATT) and C-scan
technology for improved flaw identification (MAUS). The conventional devices that were used
in this experiment were randomly assigned to each inspector with the exception of LFBT
devices, for which previous experience and knowledge of the device used was required. It should
be noted that all results shown for thermography in this report are for pulsed thermography, and
there were three different cameras used. The three cameras were a FLIR A40 uncooled IR
(results shown as thermography uncooled), FLIR Indigo Phoenix large format array (results
shown as thermography FLIR), and a Radiance HS (results shown as thermography).

134
 Table 6. Inspectors’ NDI experience level for conventional methods

Inspector Experience Level

Years of NDI
No. of Inspectors
Experience
1‒4 15
5‒8 13
9‒12 11
13‒16 11
17‒20 2
21‒24 2
25 or greater 2

Table 7. Advanced NDI methods deployed on composite honeycomb flaw-detection

experiment

Advanced NDI Methods

Air-coupled UT
Computer-aided tap tester (CATT)
FlawInspecta UT array
Laminography
Laser UT
Microwave
Mobile Automated Scanner (MAUS)
Mechanical impedance analysis (MIA)
Shearography
Structural anomaly mapping (SAM) system
Thermography

Each inspector’s flaw calls were used to identify hits (calls with any amount of overlap between
the call and the solution), misses (no call for an area of a known flaw), false calls (call with no
overlap with a flaw), and the ability to properly size flaws (degree of overlap between
experimenter calls and actual flaw areas). Typical results include PoD curves, tabulated flaw
detection and false call data, and scatter charts relating the effects of inspection time on false
calls and PoD values along with the effect of false calls on PoD values. All of the data were
sorted by the attributes of the structure being inspected: 1) three-ply fiberglass skin; 2) three-ply
carbon skin; 3) six-ply fiberglass skin; 4) six-ply carbon skin; 5) nine-ply fiberglass skin; and 6)
nine-ply carbon skin. The set of graphs in this section presents all of the detailed results for all
aspects of the solid laminate flaw detection experiment. These include the PoD curves for each

135
inspector as well as the resulting cumulative PoD curve. Tabulated results are also provided to
summarize various aspects of the experiments. The tables present the percentage of flaws
detected for each flaw size in the different inspection and composite honeycomb construction
categories. The tables also show the inspector’s ability to properly size each flaw they detected.
For example, of all the flaws they found in the Constant Thickness category, 21% were correctly
sized (100% coverage). Additional tables show the False Calls for each inspector completing the
Thin and Thick Laminate experiments as well as an average false call rate broken down into the
different geometry categories and sizes.

Overall, the PoD results were consistent with only a few outliers. This is fairly common for
human performance assessment experiments. A large number of variables were studied and
isolated to determine their impact on PoD values. Figure 159 shows a sample PoD curve from
the completed experiments. The PoD curve relates the probability of finding a flaw to the size of
the flaw for a particular specimen construction (in this case, six-ply carbon skin). The curves
show that to achieve a 90% chance of finding a flaw, the flaw size must be between 1.7″ and 3.0″
in diameter depending on the inspection device used. The graph in figure 160 shows another way
of presenting the data. The series of PoD curves indicate the performance of a single inspection
device—the Woodpecker automated tap instrument—over the entire range of specimens. The
greatest sensitivity is found in the thinnest three-ply fiberglass skin panels while the worst flaw
detection occurred in the thickest nine-ply carbon skin panels. This phenomenon was typical for
most inspection methods deployed. This is probably because a defect creates the greatest loss of
stiffness in the weakest fiberglass structure, and the effect of similar-sized defects is lessened as
the skin thickness increases and the stiffer carbon material is used. The entire set of PoD curves,
such as the samples shown in figures 159 and 160, established a baseline of how well current
inspection methods are able to detect flaws in composite honeycomb structures. This is
essentially a measure of current flaw detection at aircraft maintenance depots with these devices.
Overall, for the conventional methods, the MIA mode of inspection produced the best results for
the range of specimen types tested, with the exception being on the nine-ply carbon where the
WichiTech DTH performed the best. Except for instances in which only large flaws need to be
detected, the limit of manual tap testing was established at around six plies (PoD values for six-
ply panels ranged from 2.3 to greater than 3). In addition to quantifying the capabilities of
conventional inspection methods, this study highlighted many potential procedural and
equipment deployment improvements.

The results shown in this section are broken down into the following composite honeycomb
structure categories:

• 3C = three-ply carbon
• 3F = three-ply fiberglass
• 6C = six-ply carbon
• 6F = six-ply fiberglass
• 9C = nine-ply carbon
• 9F = nine-ply fiberglass

136
 Cumulative PoD of All Conventional NDI Devices for 6 Ply Carbon
Airbus Tap Hammer Boeing Tap Hammer LFBT MIA Wichitech DTH Woodpecker
1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Figure 159. Probability of detection curves showing the performance of multiple NDI
devices for a single type of test specimen: 6-ply carbon

Cumulative PoD - Woodpecker for All Panel Types

3 Ply Fiberglass 3 Ply Carbon 6 Ply Fiberglass 6 Ply Carbon 9 Ply Fiberglass 9 Ply Carbon

0.9

0.8
Probability of Detection

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Figure 160. Probability of detection curves for flaw detection performance of a single
device (Woodpecker) over the range of test specimen types

137
With this data set as a foundation, it was then possible to quantify the improvements via the
application of advanced NDI techniques. Figure 161 compares the performance of the best
conventional NDI technique, the DTH for the nine-ply carbon set shown, with those obtained
from the array of advanced NDI techniques evaluated. The 90% PoD level was achieved at a
2.2″ flaw diameter for the best, conventional NDI technique (WichiTech DTH). However, that
PoD number was reduced to less than 0.7″ diameter when the best performing inspection device,
in this case thermography, was deployed. This level of inspection improvement was very
consistent, and the POD numbers were reduced 66–72% in each of the composite construction
scenarios studied. Furthermore, the advanced NDI techniques were all wide-area inspection
techniques as opposed to the conventional methods that are more localized. This minimizes some
of the human factors issues by producing more uniform and rapid coverage of the entire surface
area.

Table 8 provides a summary of the overall accuracy of the evaluated NDI techniques for one of
the honeycomb construction scenarios: six-ply fiberglass. The PoD results are shown alongside
the false call data. Normally, there are adjustments that can be made in equipment’s signal
thresholds such that smaller flaws might be identified. However, this may be accompanied by a
corresponding increase in the number of false indications produced by the instrument. Thus, the
inspections become a delicate balance between optimizing flaw detection and maintaining the
number of false calls within acceptable levels.

Another indicator of equipment accuracy pertains to the ability of the NDI technique to correctly
size the flaws that it finds. Flaw sizing is important when considering damage tolerance and
critical flaw growth in the structure. Table 8 summarizes the flaw sizing capability of each
inspection device by placing each instrument’s flaw coverage into five different sizing bins. Each
bin describes the degree of overlap between an instrument’s predicted flaw size and the actual
engineered flaw in the specimen. The 100% column, for example, lists the percentage of called
flaws that were correctly sized by the inspection method while the 50–74% column lists the
percentage of called flaws that were only 50–74% of the actual flaw size. The table shows that
thermography, laminography, microwave, and the MAUS automated scanner (MIA mode)
produced the best flaw sizing results for the six-ply fiberglass set.

138
 Table 8. Summary of overall flaw detection for 6-ply fiberglass construction—
PoD levels along with false calls and ability to correctly size flaws

Flaw Sizing and False Call Summary Table for 6-Ply Fiberglass
Flaw Coverage
90%
99‒ 74‒ 49‒ False
Inspection Device 100% <25% PoD
75% 50% 25% Calls
Level
Airbus Tap Hammer 28% 30% 27% 13% 0% 2.44 2.9
Boeing Tap Hammer 21% 34% 25% 16% 4% 2.33 4.7
LFBT 28% 29% 20% 18% 5% 2.55 3.3
MIA 26% 26% 26% 18% 4% 1.49 1.9
WichiTech DTH 32% 39% 19% 8% 2% 1.71 1.6
Woodpecker 31% 28% 20% 14% 7% 2.05 0.1
CATT 28% 38% 19% 13% 2% 1.10 1.0
MAUS Resonance 1 47% 31% 4% 4% 14% 0.55 9.0
SAM 11% 40% 32% 9% 8% 0.84 8.0
Shearography 49% 27% 15% 9% 0% ≤0.50 0.0
Thermography 75% 15% 5% 5% 0% 0.70 3.0
MAUS Resonance 2 39% 33% 18% 9% 0% 2.07 2.0
MAUS MIA 84% 8% 8% 0% 0% 1.48 6.0
Laminography 86% 6% 0% 2% 6% 1.43 2.0
Microwave 75% 17% 2% 0% 6% 0.93 12.0
Thermography
55% 24% 2% 2% 17% 2.19 2.0
(uncooled)
Thermography (FLIR) N/A N/A N/A N/A N/A ≤0.50 4.0
FlawInspecta UT
31% 13% 25% 6% 25% ≥3.00 0.0
Array
Air-coupled UT 21% 43% 34% 2% 0% 0.90 0.0
Laser UT N/A N/A N/A N/A N/A N/A N/A

139
 Comparison of Advanced Inspection Techniques
Thermography (FLIR) 97% detection
with Best Conventional NDI Result on 9 Ply Carbon
Thermography Air Coupled UT MAUS Reson. (1)
Laminography Shearography Wichitech DTH
1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

9-Ply Carbon
90% POD
Device False Calls
Value
Thermography (FLIR) 0 ≤.50
Thermography 0 0.64
Air-coupled UT 2 1.26
MAUS Resonance 1 0 1.32
Laminography 0 1.39
Shearography 0 2.00
WichiTech DTH 4.4 2.15
(Highlighted = best conventional method for comparison purposes)

Figure 161. Probability of detection curves quantifying inspection improvements possible

through the application of advanced NDI techniques

140
During the testing, inspectors were asked to work at their normal pace so that no time restrictions
were placed on the experiment. However, time-of-inspection was recorded as a secondary
variable. Figure 162 shows one of the plots relating inspection time to the resulting 90% PoD
level. This graphic shows that flaw detection improved (i.e., PoD levels decreased) as the
inspector took more time to inspect each panel. Most of the improvements are in the shorter
inspection times in which increasing the inspection time from 5 to 15 minutes for each 2.25 ft.2
panel could reduce the inspection PoD by as much as 60%. The improvement levels off to the
point where increased inspection time does not yield better results, so optimum inspection rates
can be determined by this type of data.

3.00
90% PoD Value

2.50
2.00
1.50
1.00
0.50
0.00
0 5 10 15 20 25 30
Inspection Time

Figure 162. Sample effect of inspection time on flaw detection performance

(example for Boeing tap hammer on 3-ply fiberglass)

The comprehensive set of analyzed data produced during this study includes the following
performance data for the conventional and advanced inspection methods (see table 9). A
comparison of performance results for all conventional and advanced NDI methods will be
provided at the end of this section detailing the following:

• PoD curves comparing the performance of individual inspectors and a cumulative result
for a particular inspection method in each of the specific composite honeycomb structure
categories
• PoD curves comparing the overall cumulative performance of several conventional
inspection methods when performance outliers are removed from the calculation in a
couple of the specific composite honeycomb structure categories
• PoD curves comparing the overall cumulative performance of conventional inspection
methods with the 95% confidence bound
• PoD curves comparing the performance of advanced NDI techniques in each of the
specific composite honeycomb structure categories
• PoD curves comparing the overall cumulative performance of conventional inspection
techniques in each of the specific composite honeycomb structure categories
• PoD curves comparing the overall cumulative performance of all mechanical tap test
devices in each of the specific composite honeycomb structure categories

141
• PoD curves comparing the overall cumulative performance of a specific conventional
inspection technique over all of the composite honeycomb structure categories
• PoD curves comparing the overall performance of a specific advanced inspection
technique over all of the composite honeycomb structure categories
• PoD curves comparing the performance of laminography with all hit/miss data versus
hit/miss data with impact flaws removed from data set, and when edge flaw misses were
changed from misses to hits (most likely due to lack of inspection image overlap and not
inspection method sensitivity). Edge flaws for this purpose are defined as flaws that were
located at the edges of their system images (C-scans) where the resolution was less
than optimum
• PoD curves comparing the performance of laser UT on three-ply carbon and fiberglass
along with the exceptions on the three-ply fiberglass that included the removal of some
flaws from the data set in the center region of the panels, where their system could not
attenuate the laser light level sufficiently because of the specular nature of the glossy
paint
• PoD curves comparing the performance of the best conventional inspection method to the
best performing advanced NDI methods in each of the specific composite honeycomb
structure categories
• Tabulated values comparing inspection times for each of the conventional and advanced
NDI techniques in each of the specific composite honeycomb structure categories
• Tabulated values summarizing individual and average false calls for each composite
honeycomb structure category; false call averages arranged by inspection method and
panel type
• Tabulated values summarizing detection of far-side flaws for each composite honeycomb
structure category. Far-side flaws are flaws on the side of the panel opposite the
inspection surface. This is a secondary variable, and detection of these flaws was not
required in the experiment. However, positive indications in these regions were tabulated
• Scatter diagrams showing effect of inspection time on PoD and false call values for each
composite honeycomb structure category
• Scatter diagrams showing effect of the number of false calls on PoD values for each
composite honeycomb structure category
• Tabulated performance results summary—flaw sizing capability alongside PoD values
and false calls for each composite honeycomb structure category
• Tabulated performance results summary—flaw detection capability alongside PoD values
and false calls for each composite honeycomb structure category

142
 Table 9. Inspection summary—description of all figures and tables showing the variables
captured in the results

Inspection Summary – Figure/Table Performance Data

Composite Honeycomb
Figure/Table No. Inspection Method(s) Variation /Description
Structure Category
Individual and cumulative POD curve
Full suite (3-, 6-, and 9-ply
Figures 163‒168 Airbus tap hammer comparisons along with tabulated values of
carbon and fiberglass)
POD and false calls
Individual and cumulative POD curve
Full suite (3-, 6-, and 9-ply
Figures 169‒174 Boeing tap hammer comparisons along with tabulated values of
carbon and fiberglass)
POD and false calls
Individual and cumulative POD curve
Low frequency bond test Full suite (3-, 6-, and 9-ply
Figures 175‒180 comparisons along with tabulated values of
(LFBT) carbon and fiberglass)
POD and false calls
Individual and cumulative POD curve
Mechanical impedance Full suite (3-, 6-, and 9-ply
Figures 181‒186 comparisons along with tabulated values of
analysis (MIA) carbon and fiberglass)
POD and false calls
Individual and cumulative POD curve
WichiTech Digital Tap Full suite (3-, 6-, and 9-ply
Figures 187‒192 comparisons along with tabulated values of
Hammer (DTH) carbon and fiberglass)
POD and false calls
Individual and cumulative POD curve
Full suite (3-, 6-, and 9-ply
Figures 193‒198 Woodpecker comparisons along with tabulated values of
carbon and fiberglass)
POD and false calls
Airbus tap hammer and 6- and 9-ply carbon and Cumulative POD curves with and without
Figures 199‒202
LFBT fiberglass performance outliers
Full suite (3-, 6-, and 9-ply Cumulative POD curves with 95% confidence
Figures 203‒238 All conventional methods
carbon and fiberglass) bound

POD curve comparisons within a specific

All advanced NDI Full suite (3-, 6-, and 9-ply
Figures 239‒250 composite honeycomb structure along with
methods carbon and fiberglass)
tabulated values of POD and false calls

Cumulative POD curve comparisons within a

Full suite (3-, 6-, and 9-ply specific composite honeycomb structure
Figures 251‒256 All conventional methods
carbon and fiberglass) category along with tabulated values of POD
and average number of false calls

All mechanical impact tap Full suite (3-, 6-, and 9-ply Cumulative POD curve comparisons
Figures 257‒262
test methods carbon and fiberglass) (conventional) versus advanced (CATT)

Cumulative POD curve comparison of a single

Full suite (3-, 6-, and 9-ply device on all composite honeycomb structure
Figures 263‒268 All conventional methods
carbon and fiberglass) categories along with tabulated values of POD
and number of false calls
POD curve comparison of advanced NDI
method (individually) on all composite
All advanced NDI Full suite (3-, 6-, and 9-ply
Figures 269‒278 honeycomb structure categories along with
methods carbon and fiberglass)
tabulated values of POD and number of false
calls

143
 Table 9. Inspection summary—description of all figures and tables showing the variables
captured in the results (continued)

Inspection Summary – Figure/Table Performance Data

Composite Honeycomb
Figure/Table No. Inspection Method(s) Variation /Description
Structure Category
Overall POD curve comparison with certain
6- and 9-ply carbon and exceptions including impact flaw removal and
Figures 279‒282 Laminography
fiberglass flaws located in areas where there was lack of
image overlap
Overall POD curve comparison with
Figure 283 Laser UT 3-ply carbon and fiberglass exception for center area misses due to system
inability to attenuate the laser light sufficiently
POD curve comparisons along with tabulated
Best conventional method
Full suite (3-, 6-, and 9-ply values of advanced NDI and best conventional
Figures 284‒289 and advanced NDI
carbon and fiberglass) method in each composite honeycomb
methods
structure category
Tabulated values comparing inspection times
All convetional and Full suite (3-, 6-, and 9-ply
Table 10‒13 in each of the specific composite honeycomb
advanced NDI methods carbon and fiberglass)
structure categories

Tabulated values summarizing individual and

All convetional and Full suite (3-, 6-, and 9-ply
Table 14‒16 average false calls for each composite
advanced NDI methods carbon and fiberglass)
honeycomb structure category

Tabulated values summarizing detection of

All convetional and Full suite (3-, 6-, and 9-ply
Table 17‒19 farside (backside) flaws for each composite
advanced NDI methods carbon and fiberglass)
honeycomb structure category
Scatter diagrams showing the effect of
All conventional NDI Full suite (3-, 6-, and 9-ply inspection time on POD and false call values
Figures 290‒295
methods carbon and fiberglass) for each composite honeycomb structure
category
Scatter diagrams showing the effect of the
All conventional NDI Full suite (3-, 6-, and 9-ply
Figures 296‒298 number of false calls on POD values for each
methods carbon and fiberglass)
composite honeycomb structure category
Tabulated performance results summary with
All convetional and Full suite (3-, 6-, and 9-ply flaw sizing capability alongside POD values
Table 20‒22
advanced NDI methods carbon and fiberglass) and false calls for each composite honeycomb
structure category
Tabulated performance results summary with
All convetional and Full suite (3-, 6-, and 9-ply flaw detection capability alongside POD
Table 23‒25
advanced NDI methods carbon and fiberglass) values and false calls for each composite
honeycomb structure category

144
The PoD curves and summary tabulations in figures 163–198 compare the performance of
individual inspectors for a particular inspection device/method in each of the specific composite
honeycomb structure categories. In each figure, the cumulative PoD result is displayed as a bold
solid line. The individual inspector results for the desired 90% PoD level are listed above the
overall cumulative industry baseline.
Airbus Tap Hammer PoD Curve Comparisons 3C

0.9

0.8

0.7 AT-1
Probability of Detection

AT-2
0.6 AT-3
AT-4
0.5 AT-5
AT-6
0.4 AT-7
AT-8
0.3 Cumulative

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Airbus Tap Hammer 3-Ply Carbon

Inspector False Calls 90% POD Value
AT-1 0 2.93
AT-2 1 1.52
AT-3 1 2.20
AT-4 0 2.16
AT-5 0 2.33
AT-6 1 1.98
AT-7 3 2.81
AT-8 0 >3.00
Cumulative All Inspectors 2.34

Figure 163. Individual and cumulative PoD curve comparisons along with tabulated values
for the Airbus tap hammer deployed on 3-ply carbon test specimen set

145
 Airbus Tap Hammer PoD Curve Comparisons 3F

0.9

0.8

0.7 AT-1
Probability of Detection

AT-2
0.6 AT-3
AT-4
0.5 AT-5
AT-6
0.4 AT-7
AT-8
0.3 Cumulative

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Airbus Tap Hammer 3-Ply Fiberglass

Inspector False Calls 90% POD Value
AT-1 1 2.66
AT-2 0 1.42
AT-3 1 2.45
AT-4 0 2.44
AT-5 0 1.60
AT-6 0 1.27
AT-7 0 2.46
AT-8 0 1.75
Cumulative All Inspectors 2.03

Figure 164. Individual and cumulative PoD curve comparisons along with tabulated values
for the Airbus tap hammer deployed on 3-ply fiberglass test specimen set

146
 Airbus Tap Hammer PoD Curve Comparisons 6C

0.9

0.8

0.7 AT-1
Probability of Detection

AT-2
0.6 AT-3
AT-4
0.5 AT-5
AT-6
0.4 AT-7
AT-8
0.3 Cumulative

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Airbus Tap Hammer 6-Ply Carbon

Inspector False Calls 90% POD Value
AT-1 2 >3.00
AT-2 3 2.07
AT-3 1 >3.00
AT-4 4 2.46
AT-5 4 2.11
AT-6 9 >3.00
AT-7 2 2.74
AT-8 3 >3.00
Cumulative All Inspectors >3.00

Figure 165. Individual and cumulative PoD curve comparisons along with tabulated values
for the Airbus tap hammer deployed on 6-ply carbon test specimen set

147
 Airbus Tap Hammer PoD Curve Comparisons 6F

0.9

0.8

0.7 AT-1
Probability of Detection

AT-2
0.6 AT-3
AT-4
0.5 AT-5
AT-6
0.4 AT-7
AT-8
0.3 Cumulative

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Airbus Tap Hammer 6-Ply Fiberglass

Inspector False Calls 90% POD Value
AT-1 1 >3.00
AT-2 2 1.84
AT-3 13 >3.00
AT-4 0 2.50
AT-5 1 1.31
AT-6 0 1.90
AT-7 6 2.50
AT-8 0 1.90
Cumulative All Inspectors 2.44

Figure 166. Individual and cumulative PoD curve comparisons along with tabulated values
for the Airbus tap hammer deployed on 6-ply fiberglass test specimen set

148
 Airbus Tap Hammer PoD Curve Comparisons 9C

0.9

0.8

0.7 AT-1
Probability of Detection

AT-2
0.6 AT-3
AT-4
0.5 AT-5
AT-6
0.4 AT-7
AT-8
0.3 Cumulative

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Airbus Tap Hammer 9-Ply Carbon

Inspector False Calls 90% POD Value
AT-1 13 >3.00
AT-2 2 2.31
AT-3 26 2.42
AT-4 1 2.39
AT-5 1 >3.00
AT-6 3 2.74
AT-7 6 >3.00
AT-8 2 >3.00
Cumulative All Inspectors >3.00

Figure 167. Individual and cumulative PoD curve comparisons along with tabulated values
for the Airbus tap hammer deployed on 9-ply carbon test specimen set

149
 Airbus Tap Hammer PoD Curve Comparisons 9F

0.9

0.8

0.7 AT-1
Probability of Detection

AT-2
0.6 AT-3
AT-4
0.5 AT-5
AT-6
0.4 AT-7
AT-8
0.3 Cumulative

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Airbus Tap Hammer 9-Ply Fiberglass

Inspector False Calls 90% POD Value
AT-1 13 >3.00
AT-2 11 1.31
AT-3 23 >3.00
AT-4 2 2.04
AT-5 3 2.57
AT-6 2 2.81
AT-7 2 >3.00
AT-8 1 >3.00
Cumulative All Inspectors >3.00

Figure 168. Individual and cumulative PoD curve comparisons along with tabulated values
for the Airbus tap hammer deployed on 9-ply fiberglass test specimen set

150
 Boeing Tap Hammer PoD Curve Comparisons 3C

0.9

0.8
BT-1
0.7
Probability of Detection

BT-2
BT-3
0.6
BT-4
BT-5
0.5
BT-6
BT-7
0.4
BT-8
0.3 BT-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Boeing Tap Hammer 3-Ply Carbon

Inspector False Calls 90% POD Value
BT-1 0 2.03
BT-2 1 1.78
BT-3 3 2.09
BT-4 1 1.96
BT-5 1 1.44
BT-6 2 2.07
BT-7 0 2.17
BT-8 1 1.98
BT-9 0 2.17
Cumulative All Inspectors 2.03

Figure 169. Individual and cumulative PoD curve comparisons along with tabulated values
for the Boeing tap hammer deployed on 3-ply carbon test specimen set

151
 Boeing Tap Hammer PoD Curve Comparisons 3F

0.9

0.8
BT-1
0.7
Probability of Detection

BT-2
BT-3
0.6
BT-4
BT-5
0.5
BT-6
BT-7
0.4
BT-8
0.3 BT-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Boeing Tap Hammer 3-Ply Fiberglass

Inspector False Calls 90% POD Value
BT-1 4 1.05
BT-2 3 1.49
BT-3 3 1.88
BT-4 0 1.67
BT-5 0 0.94
BT-6 6 1.90
BT-7 0 1.19
BT-8 0 2.58
BT-9 0 1.27
Cumulative All Inspectors 1.57

Figure 170. Individual and cumulative PoD curve comparisons along with tabulated values
for the Boeing tap hammer deployed on 3-ply fiberglass test specimen set

152
 Boeing Tap Hammer PoD Curve Comparisons 6C

0.9

0.8
BT-1
0.7
Probability of Detection

BT-2
BT-3
0.6
BT-4
BT-5
0.5
BT-6
BT-7
0.4
BT-8
0.3 BT-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Boeing Tap Hammer 6-Ply Carbon

Inspector False Calls 90% POD Value
BT-1 0 2.74
BT-2 4 2.74
BT-3 6 2.39
BT-4 4 2.92
BT-5 8 1.62
BT-6 13 2.33
BT-7 2 2.51
BT-8 2 2.58
BT-9 3 2.45
Cumulative All Inspectors 2.59

Figure 171. Individual and cumulative PoD curve comparisons along with tabulated values
for the Boeing tap hammer deployed on 6-ply carbon test specimen set

153
 Boeing Tap Hammer PoD Curve Comparisons 6F

0.9

0.8
BT-1
0.7
Probability of Detection

BT-2
BT-3
0.6
BT-4
BT-5
0.5
BT-6
BT-7
0.4
BT-8
0.3 BT-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Boeing Tap Hammer 6-Ply Fiberglass

Inspector False Calls 90% POD Value
BT-1 0 2.81
BT-2 6 2.89
BT-3 6 >3.00
BT-4 1 1.60
BT-5 7 1.70
BT-6 17 2.12
BT-7 2 1.88
BT-8 2 2.50
BT-9 1 2.03
Cumulative All Inspectors 2.33

Figure 172. Individual and cumulative PoD curve comparisons along with tabulated values
for the Boeing tap hammer deployed on 6-ply fiberglass test specimen set

154
 Boeing Tap Hammer PoD Curve Comparisons 9C

0.9

0.8
BT-1
0.7
Probability of Detection

BT-2
BT-3
0.6
BT-4
BT-5
0.5
BT-6
BT-7
0.4
BT-8
0.3 BT-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Boeing Tap Hammer 9-Ply Carbon

Inspector False Calls 90% POD Value
BT-1 3 2.96
BT-2 2 2.36
BT-3 7 2.89
BT-4 2 2.70
BT-5 14 2.18
BT-6 14 2.72
BT-7 5 2.73
BT-8 1 >3.00
BT-9 1 >3.00
Cumulative All Inspectors 2.88

Figure 173. Individual and cumulative PoD curve comparisons along with tabulated values
for the Boeing tap hammer deployed on 9-ply carbon test specimen set

155
 Boeing Tap Hammer PoD Curve Comparisons 9F

0.9

0.8
BT-1
0.7
Probability of Detection

BT-2
BT-3
0.6
BT-4
BT-5
0.5
BT-6
BT-7
0.4
BT-8
0.3 BT-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Boeing Tap Hammer 9-Ply Fiberglass

Inspector False Calls 90% POD Value
BT-1 1 2.56
BT-2 7 2.75
BT-3 6 >3.00
BT-4 1 2.89
BT-5 25 1.73
BT-6 18 >3.00
BT-7 3 >3.00
BT-8 1 2.70
BT-9 0 2.35
Cumulative All Inspectors 2.90

Figure 174. Individual and cumulative PoD curve comparisons along with tabulated values
for the Boeing tap hammer deployed on 9-ply fiberglass test specimen set

156
 LFBT PoD Curve Comparisons 3C

0.9

0.8 LFBT-1
LFBT-2
0.7
Probability of Detection

LFBT-3
LFBT-4
0.6
LFBT-5
LFBT-6
0.5
LFBT-7
LFBT-8
0.4
LFBT-9
0.3 LFBT-10
LFBT-11
0.2 Cumulative

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

LFBT 3-Ply Carbon

Inspector False Calls 90% POD Value
LFBT-1 0 0.93
LFBT-2 0 1.44
LFBT-3 14 >3.00
LFBT-4 0 >3.00
LFBT-5 0 2.03
LFBT-6 0 1.01
LFBT-7 2 >3.00
LFBT-8 2 0.96
LFBT-9 0 1.25
LFBT-10 0 >3.00
LFBT-11 7 1.33
Cumulative All Inspectors 2.10

Figure 175. Individual and cumulative PoD curve comparisons along with tabulated values
for LFBT deployed on 3-ply carbon test specimen set

157
 LFBT PoD Curve Comparisons 3F

0.9

0.8 LFBT-1
LFBT-2
0.7
Probability of Detection

LFBT-3
LFBT-4
0.6
LFBT-5
LFBT-6
0.5
LFBT-7
LFBT-8
0.4
LFBT-9
0.3 LFBT-10
LFBT-11
0.2 Cumulative

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

LFBT 3-Ply Fiberglass

Inspector False Calls 90% POD Value
LFBT-1 0 0.90
LFBT-2 0 1.58
LFBT-3 8 >3.00
LFBT-4 0 2.20
LFBT-5 1 1.49
LFBT-6 0 1.20
LFBT-7 0 2.76
LFBT-8 1 1.59
LFBT-9 0 1.51
LFBT-10 0 0.69
LFBT-11 0 1.81
Cumulative All Inspectors 1.77

Figure 176. Individual and cumulative PoD curve comparisons along with tabulated values
for LFBT deployed on 3-ply fiberglass test specimen set

158
 LFBT PoD Curve Comparisons 6C

0.9

0.8 LFBT-1
LFBT-2
0.7
Probability of Detection

LFBT-3
LFBT-4
0.6
LFBT-5
LFBT-6
0.5
LFBT-7
LFBT-8
0.4
LFBT-9
0.3 LFBT-10
LFBT-11
0.2 Cumulative

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

LFBT 6-Ply Carbon

Inspector False Calls 90% POD Value
LFBT-1 0 2.77
LFBT-2 0 1.76
LFBT-3 33 2.03
LFBT-4 1 2.11
LFBT-5 1 1.97
LFBT-6 0 1.85
LFBT-7 1 >3.00
LFBT-8 0 2.28
LFBT-9 2 1.32
LFBT-10 0 2.84
LFBT-11 34 >3.00
Cumulative All Inspectors 2.24

Figure 177. Individual and cumulative PoD curve comparisons along with tabulated values
for LFBT deployed on 6-ply carbon test specimen set

159
 LFBT PoD Curve Comparisons 6F

0.9

0.8 LFBT-1
LFBT-2
0.7
Probability of Detection

LFBT-3
LFBT-4
0.6
LFBT-5
LFBT-6
0.5
LFBT-7
LFBT-8
0.4
LFBT-9
0.3 LFBT-10
LFBT-11
0.2 Cumulative

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

LFBT 6-Ply Fiberglass

Inspector False Calls 90% POD Value
LFBT-1 0 >3.00
LFBT-2 0 1.14
LFBT-3 24 >3.00
LFBT-4 0 1.87
LFBT-5 6 >3.00
LFBT-6 0 1.24
LFBT-7 1 2.08
LFBT-8 2 1.67
LFBT-9 1 1.05
LFBT-10 0 2.50
LFBT-11 2 2.92
Cumulative All Inspectors 2.55

Figure 178. Individual and cumulative PoD curve comparisons along with tabulated values
for LFBT deployed on 6-ply fiberglass test specimen set

160
 LFBT PoD Curve Comparisons 9C

0.9

0.8 LFBT-1
LFBT-2
0.7
Probability of Detection

LFBT-3
LFBT-4
0.6
LFBT-5
LFBT-6
0.5
LFBT-7
LFBT-8
0.4
LFBT-9
0.3 LFBT-10
LFBT-11
0.2 Cumulative

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

LFBT 9-Ply Carbon

Inspector False Calls 90% POD Value
LFBT-1 0 2.49
LFBT-2 0 2.25
LFBT-3 1 >3.00
LFBT-4 3 2.46
LFBT-5 1 1.86
LFBT-6 0 2.18
LFBT-7 4 2.92
LFBT-8 0 2.36
LFBT-9 0 1.45
LFBT-10 0 2.45
LFBT-11 33 >3.00
Cumulative All Inspectors 2.61

Figure 179. Individual and cumulative PoD curve comparisons along with tabulated values
for LFBT deployed on 9-ply carbon test specimen set

161
 LFBT PoD Curve Comparisons 9F

0.9

0.8 LFBT-1
LFBT-2
0.7
Probability of Detection

LFBT-3
LFBT-4
0.6
LFBT-5
LFBT-6
0.5
LFBT-7
LFBT-8
0.4
LFBT-9
0.3 LFBT-10
LFBT-11
0.2 Cumulative

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

LFBT 9-Ply Fiberglass

Inspector False Calls 90% POD Value
LFBT-1 0 >3.00
LFBT-2 0 1.94
LFBT-3 0 >3.00
LFBT-4 3 >3.00
LFBT-5 0 2.10
LFBT-6 0 2.44
LFBT-7 0 1.68
LFBT-8 0 2.11
LFBT-9 3 1.79
LFBT-10 0 >3.00
LFBT-11 25 >3.00
Cumulative All Inspectors 3.00

Figure 180. Individual and cumulative PoD curve comparisons along with tabulated values
for LFBT deployed on 9-ply fiberglass test specimen set

162
 MIA PoD Curve Comparisons 3C

0.9

0.8
MIA-1
0.7 MIA-2
Probability of Detection

MIA-3
0.6 MIA-4
MIA-5
0.5 MIA-6
MIA-7
0.4 MIA-8
MIA-9
0.3 MIA-10
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

MIA 3-Ply Carbon

Inspector False Calls 90% POD Value
MIA-1 1 1.23
MIA-2 1 1.79
MIA-3 0 2.11
MIA-4 4 1.23
MIA-5 0 1.44
MIA-6 0 1.23
MIA-7 0 1.40
MIA-8 0 1.35
MIA-9 0 2.40
MIA-10 0 1.26
Cumulative All Inspectors 1.46

Figure 181. Individual and cumulative PoD curve comparisons along with tabulated values
for MIA deployed on 3-ply carbon test specimen set

163
 MIA PoD Curve Comparisons 3F

0.9

0.8
MIA-1
0.7 MIA-2
Probability of Detection

MIA-3
0.6 MIA-4
MIA-5
0.5 MIA-6
MIA-7
0.4 MIA-8
MIA-9
0.3 MIA-10
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

MIA 3-Ply Fiberglass

Inspector False Calls 90% POD Value
MIA-1 2 0.95
MIA-2 1 1.15
MIA-3 0 0.96
MIA-4 1 0.97
MIA-5 0 1.05
MIA-6 0 1.05
MIA-7 1 0.91
MIA-8 0 1.48
MIA-9 0 0.87
MIA-10 0 0.97
Cumulative All Inspectors 1.04

Figure 182. Individual and cumulative PoD curve comparisons along with tabulated values
for MIA deployed on 3-ply fiberglass test specimen set

164
 MIA PoD Curve Comparisons 6C

0.9

0.8
MIA-1
0.7 MIA-2
Probability of Detection

MIA-3
0.6 MIA-4
MIA-5
0.5 MIA-6
MIA-7
0.4 MIA-8
MIA-9
0.3 MIA-10
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

MIA 6-Ply Carbon

Inspector False Calls 90% POD Value
MIA-1 0 1.93
MIA-2 0 1.76
MIA-3 0 1.88
MIA-4 4 1.75
MIA-5 0 1.72
MIA-6 0 1.72
MIA-7 1 1.71
MIA-8 1 2.03
MIA-9 6 1.55
MIA-10 1 1.62
Cumulative All Inspectors 1.76

Figure 183. Individual and cumulative PoD curve comparisons along with tabulated values
for MIA deployed on 6-ply carbon test specimen set

165
 MIA PoD Curve Comparisons 6F

0.9

0.8
MIA-1
0.7 MIA-2
Probability of Detection

MIA-3
0.6 MIA-4
MIA-5
0.5 MIA-6
MIA-7
0.4 MIA-8
MIA-9
0.3 MIA-10
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

MIA 6-Ply Fiberglass

Inspector False Calls 90% POD Value
MIA-1 1 1.65
MIA-2 0 1.55
MIA-3 0 1.39
MIA-4 0 1.33
MIA-5 0 2.00
MIA-6 0 1.51
MIA-7 1 1.18
MIA-8 0 1.66
MIA-9 15 1.47
MIA-10 2 1.16
Cumulative All Inspectors 1.49

Figure 184. Individual and cumulative PoD curve comparisons along with tabulated values
for MIA deployed on 6-ply fiberglass test specimen set

166
 MIA PoD Curve Comparisons 9C

0.9

0.8
MIA-1
0.7
Probability of Detection

MIA-2
MIA-3
0.6
MIA-4
MIA-5
0.5
MIA-6
MIA-7
0.4
MIA-8
0.3 MIA-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

MIA 9-Ply Carbon

Inspector False Calls 90% POD Value
MIA-1 0 >3.00
MIA-2 1 2.08
MIA-3 3 2.45
MIA-4 6 >3.00
MIA-5 1 2.32
MIA-6 4 3.00
MIA-7 4 >3.00
MIA-8 5 >3.00
MIA-9 N/A N/A
MIA-10 4 2.40
Cumulative All Inspectors 1.49

Figure 185. Individual and cumulative PoD curve comparisons along with tabulated values
for MIA deployed on 9-ply carbon test specimen set

167
 MIA PoD Curve Comparisons 9F

0.9

0.8
MIA-1
0.7 MIA-2
Probability of Detection

MIA-3
0.6 MIA-4
MIA-5
0.5 MIA-6
MIA-7
0.4 MIA-8
MIA-9
0.3 MIA-10
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

MIA 9-Ply Fiberglass

Inspector False Calls 90% POD Value
MIA-1 0 >3.00
MIA-2 2 2.43
MIA-3 2 1.86
MIA-4 5 2.22
MIA-5 0 1.97
MIA-6 2 >3.00
MIA-7 15 1.83
MIA-8 0 2.50
MIA-9 36 1.7
MIA-10 6 1.51
Cumulative All Inspectors 2.00

Figure 186. Individual and cumulative PoD curve comparisons along with tabulated values
for MIA deployed on 9-ply fiberglass test specimen set

168
 Wichitech DTH PoD Curve Comparisons 3C

0.9

0.8
DTH-1
0.7
Probability of Detection

DTH-2
DTH-3
0.6
DTH-4
DTH-5
0.5
DTH-6
DTH-7
0.4
DTH-8
0.3 DTH-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia in Inches)

DTH 3-Ply Carbon

Inspector False Calls 90% POD Value
DTH-1 0 2.92
DTH-2 5 >3.00
DTH-3 0 2.20
DTH-4 0 1.46
DTH-5 0 2.31
DTH-6 0 1.62
DTH-7 0 >3.00
DTH-8 0 1.94
DTH-9 0 1.26
Cumulative All Inspectors 2.28

Figure 187. Individual and cumulative PoD curve comparisons along with tabulated values
for WichiTech DTH deployed on 3-ply carbon test specimen set

169
 Wichitech DTH PoD Curve Comparisons 3F

0.9

0.8
DTH-1
0.7
Probability of Detection

DTH-2
DTH-3
0.6
DTH-4
DTH-5
0.5
DTH-6
DTH-7
0.4
DTH-8
0.3 DTH-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia in Inches)

DTH 3-Ply Fiberglass

Inspector False Calls 90% POD Value
DTH-1 0 1.91
DTH-2 8 2.36
DTH-3 0 1.44
DTH-4 0 1.22
DTH-5 0 2.29
DTH-6 0 1.12
DTH-7 0 2.39
DTH-8 0 2.92
DTH-9 0 1.00
Cumulative All Inspectors 1.81

Figure 188. Individual and cumulative PoD curve comparisons along with tabulated values
for WichiTech DTH deployed on 3-ply fiberglass test specimen set

170
 Wichitech DTH PoD Curve Comparisons 6C

0.9

0.8
DTH-1
0.7
Probability of Detection

DTH-2
DTH-3
0.6
DTH-4
DTH-5
0.5
DTH-6
DTH-7
0.4
DTH-8
0.3 DTH-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia in Inches)

DTH 6-Ply Carbon

Inspector False Calls 90% POD Value
DTH-1 0 2.36
DTH-2 6 2.37
DTH-3 0 1.79
DTH-4 1 2.17
DTH-5 0 2.66
DTH-6 0 1.99
DTH-7 0 2.05
DTH-8 1 1.83
DTH-9 0 2.10
Cumulative All Inspectors 2.18

Figure 189. Individual and cumulative PoD curve comparisons along with tabulated values
for WichiTech DTH deployed on 6-ply carbon test specimen set

171
 Wichitech DTH PoD Curve Comparisons 6F

0.9

0.8
DTH-1
0.7
Probability of Detection

DTH-2
DTH-3
0.6
DTH-4
DTH-5
0.5
DTH-6
DTH-7
0.4
DTH-8
0.3 DTH-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia in Inches)

DTH 6-Ply Fiberglass

Inspector False Calls 90% POD Value
DTH-1 0 1.84
DTH-2 8 1.41
DTH-3 0 1.54
DTH-4 0 1.50
DTH-5 0 2.40
DTH-6 6 1.01
DTH-7 0 2.11
DTH-8 0 1.55
DTH-9 0 1.48
Cumulative All Inspectors 1.71

Figure 190. Individual and cumulative PoD curve comparisons along with tabulated values
for WichiTech DTH deployed on 6-ply fiberglass test specimen set

172
 Wichitech DTH PoD Curve Comparisons 9C

0.9

0.8
DTH-1
0.7
Probability of Detection

DTH-2
DTH-3
0.6
DTH-4
DTH-5
0.5
DTH-6
DTH-7
0.4
DTH-8
0.3 DTH-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

DTH 9-Ply Carbon

Inspector False Calls 90% POD Value
DTH-1 0 2.59
DTH-2 5 2.04
DTH-3 0 2.20
DTH-4 0 1.86
DTH-5 0 2.24
DTH-6 31 1.72
DTH-7 0 2.25
DTH-8 0 2.43
DTH-9 4 1.79
Cumulative All Inspectors 2.15

Figure 191. Individual and cumulative PoD curve comparisons along with tabulated values
for WichiTech DTH deployed on 9-ply carbon test specimen set

173
 Wichitech DTH PoD Curve Comparisons 9F

0.9

0.8
DTH-1
0.7
Probability of Detection

DTH-2
DTH-3
0.6
DTH-4
DTH-5
0.5
DTH-6
DTH-7
0.4
DTH-8
0.3 DTH-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

DTH 9-Ply Fiberglass

Inspector False Calls 90% POD Value
DTH-1 0 2.51
DTH-2 12 2.21
DTH-3 0 1.82
DTH-4 0 1.76
DTH-5 0 2.01
DTH-6 32 1.63
DTH-7 0 2.38
DTH-8 0 2.10
DTH-9 0 1.77
Cumulative All Inspectors 2.07

Figure 192. Individual and cumulative PoD curve comparisons along with tabulated values
for WichiTech DTH deployed on 9-ply fiberglass test specimen set

174
 Woodpecker PoD Curve Comparisons 3C

0.9

0.8
WOOD-1
0.7
Probability of Detection

WOOD-2
WOOD-3
0.6
WOOD-4
WOOD-5
0.5
WOOD-6
WOOD-7
0.4
WOOD-8
0.3 WOOD-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Woodpecker 3-Ply Carbon

Inspector False Calls 90% POD Value
WOOD-1 0 1.70
WOOD-2 0 1.94
WOOD-3 0 1.56
WOOD-4 0 2.33
WOOD-5 0 2.04
WOOD-6 0 1.70
WOOD-7 0 1.42
WOOD-8 0 1.32
WOOD-9 0 1.92
Cumulative All Inspectors 1.78

Figure 193. Individual and cumulative PoD curve comparisons along with tabulated values
for Woodpecker deployed on 3-ply carbon test specimen set

175
 Woodpecker PoD Curve Comparisons 3F

0.9

0.8
WOOD-1
WOOD-2
Probability of Detection

0.7
WOOD-3
0.6 WOOD-4
WOOD-5
0.5
WOOD-6
0.4 WOOD-7
WOOD-8
0.3
WOOD-9
0.2 Cumulative

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Woodpecker 3-Ply Fiberglass

Inspector False Calls 90% POD Value
WOOD-1 0 1.06
WOOD-2 0 1.60
WOOD-3 0 1.30
WOOD-4 3 1.27
WOOD-5 0 0.97
WOOD-6 0 1.13
WOOD-7 0 2.68
WOOD-8 0 1.12
WOOD-9 0 1.93
Cumulative All Inspectors 1.45

Figure 194. Individual and cumulative PoD curve comparisons along with tabulated values
for Woodpecker deployed on 3-ply fiberglass test specimen set

176
 Woodpecker PoD Curve Comparisons 6C

0.9

0.8
WOOD-1
0.7
Probability of Detection

WOOD-2
WOOD-3
0.6
WOOD-4
WOOD-5
0.5
WOOD-6
WOOD-7
0.4
WOOD-8
0.3 WOOD-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Woodpecker 6-Ply Carbon

Inspector False Calls 90% POD Value
WOOD-1 0 2.30
WOOD-2 0 2.03
WOOD-3 0 2.44
WOOD-4 0 2.39
WOOD-5 1 1.98
WOOD-6 1 2.00
WOOD-7 4 3.00
WOOD-8 0 2.45
WOOD-9 0 2.26
Cumulative All Inspectors 2.32

Figure 195. Individual and cumulative PoD curve comparisons along with tabulated values
for Woodpecker deployed on 6-ply carbon test specimen set

177
 Woodpecker PoD Curve Comparisons 6F

0.9

0.8
WOOD-1
0.7
Probability of Detection

WOOD-2
WOOD-3
0.6
WOOD-4
WOOD-5
0.5
WOOD-6
WOOD-7
0.4
WOOD-8
0.3 WOOD-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Woodpecker 6-Ply Fiberglass

Inspector False Calls 90% POD Value
WOOD-1 0 1.71
WOOD-2 0 2.37
WOOD-3 0 1.96
WOOD-4 0 2.54
WOOD-5 0 1.77
WOOD-6 0 1.75
WOOD-7 0 1.87
WOOD-8 0 2.15
WOOD-9 1 2.14
Cumulative All Inspectors 2.05

Figure 196. Individual and cumulative PoD curve comparisons along with tabulated values
for Woodpecker deployed on 6-ply fiberglass test specimen set

178
 Woodpecker PoD Curve Comparisons 9C

0.9

0.8
WOOD-1
0.7
Probability of Detection

WOOD-2
WOOD-3
0.6
WOOD-4
WOOD-5
0.5
WOOD-6
WOOD-7
0.4
WOOD-8
0.3 WOOD-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Woodpecker 9-Ply Carbon

Inspector False Calls 90% POD Value
WOOD-1 0 2.61
WOOD-2 1 2.60
WOOD-3 2 2.59
WOOD-4 2 2.72
WOOD-5 1 2.36
WOOD-6 9 2.93
WOOD-7 0 >3.00
WOOD-8 8 2.44
WOOD-9 2 >3.00
Cumulative All Inspectors 2.76

Figure 197. Individual and cumulative PoD curve comparisons along with tabulated values
for Woodpecker deployed on 9-ply carbon test specimen set

179
 Woodpecker PoD Curve Comparisons 9F

0.9

0.8
WOOD-1
0.7
Probability of Detection

WOOD-2
WOOD-3
0.6
WOOD-4
WOOD-5
0.5
WOOD-6
WOOD-7
0.4
WOOD-8
0.3 WOOD-9
Cumulative
0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Woodpecker 9-Ply Fiberglass

Inspector False Calls 90% POD Value
WOOD-1 5 2.25
WOOD-2 0 2.75
WOOD-3 5 2.10
WOOD-4 16 2.88
WOOD-5 3 2.49
WOOD-6 13 2.82
WOOD-7 5 2.36
WOOD-8 6 2.08
WOOD-9 7 2.48
Cumulative All Inspectors 2.52

Figure 198. Individual and cumulative PoD curve comparisons along with tabulated values
for Woodpecker deployed on 9-ply fiberglass test specimen set

Several NDI devices, such as the Airbus tap hammer and the LFBT equipment, included some
outlier data where an inspector’s performance was quite higher than the other inspectors. The
PoD curves in figures 199–202 compare the cumulative PoD performance of all inspectors for a
particular inspection method with performance curves produced when performance outliers are
removed from the calculation. The results are broken down into the various composite

180
honeycomb structure categories. The improvements observed when these outliers are removed
indicate that additional training and experience with these devices could increase PoD levels.
Cumulative POD Comparison of All Inspectors With and Without
Performance Outlier Removed from Calculation - Airbus Tap Hammer (6F)

0.9

0.8

0.7
Probability of Detection

0.6

POD with outlier removed

0.5
Cumulative POD
0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Figure 199. Comparison showing level of PoD improvement when single performance
outlier is removed from the calculation—Airbus tap hammer, 6-ply fiberglass

Cumulative POD Comparison of All Inspectors With and Without

2 Performance Outliers Removed from Calculation - Airbus Tap Hammer (9F)

0.9

0.8

0.7
Probability of Detection

0.6

POD with 2 outliers removed

0.5
Cumulative POD
0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Figure 200. Comparison showing level of PoD improvement when performance outliers are
removed from the calculation—Airbus tap hammer, 9-ply fiberglass

181
 Cumulative POD Comparison of All Inspectors With and Without
Performance Outlier Removed from Calculation - LFBT (6F)

0.9

0.8

0.7
Probability of Detection

0.6
POD with outlier removed
0.5
Cumulative POD
0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Figure 201. Comparison showing level of PoD improvement when single performance
outlier is removed from the calculation—LFBT, 6-ply fiberglass

Cumulative POD Comparison of All Inspectors With and Without

Performance Outlier Removed from Calculation - LFBT (9F)

0.9

0.8

0.7
Probability of Detection

0.6

POD with outlier removed

0.5
Cumulative POD
0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Figure 202. Comparison showing level of PoD improvement when single performance
outlier is removed from the calculation—LFBT, 9-ply fiberglass

Normally, when using PoD curves to establish the performance of an NDI method, it is common
to use the 90% PoD level with a 95% confidence. The additional confidence bound consideration
182
shifts the PoD slightly to the left. The PoD curves in figures 203–238 compare the cumulative
PoD performance when the 95% Confidence Bound is imposed with the maximum likelihood
estimate PoD when there is no required confidence level. The results are broken down into the
various composite honeycomb structure categories.
Airbus Tap Hammer Cumulative PoD Curve with 95% Confidence Bound 3C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 203. Cumulative PoD curve with 95%

confidence bound—Airbus tap hammer, 3-ply carbon

Airbus Tap Hammer Cumulative PoD Curve with 95% Confidence Bound 3F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 204. Cumulative PoD curve with 95%

confidence bound—Airbus tap hammer, 3-ply fiberglass

183
 Airbus Tap Hammer Cumulative PoD Curve with 95% Confidence Bound 6C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 205. Cumulative PoD curve with 95%

confidence bound—Airbus tap hammer, 6-ply carbon

Airbus Tap Hammer Cumulative PoD Curve with 95% Confidence Bound 6F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 206. Cumulative PoD curve with 95%

confidence bound—Airbus tap hammer, 6-ply fiberglass

184
 Airbus Tap Hammer Cumulative PoD Curve with 95% Confidence Bound 9C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 207. Cumulative PoD curve with 95%

confidence bound—Airbus tap hammer, 9-ply carbon

Airbus Tap Hammer Cumulative PoD Curve with 95% Confidence Bound 9F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 208. Cumulative PoD curve with 95%

confidence bound—Airbus tap hammer, 9-ply fiberglass

185
 Boeing Tap Hammer Cumulative PoD Curve with 95% Confidence Bound 3C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 209. Cumulative PoD curve with 95%

|confidence bound—Boeing tap hammer, 3-ply carbon

Boeing Tap Hammer Cumulative PoD Curve with 95% Confidence Bound 3F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 210. Cumulative PoD curve with 95%

confidence bound—Boeing tap hammer, 3-ply Fiberglass

186
 Boeing Tap Hammer Cumulative PoD Curve and 95% Confidence Bound 6C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 211. Cumulative PoD curve with 95%

confidence bound—Boeing tap hammer,6-ply carbon

Boeing Tap Hammer Cumulative PoD Curve with 95% Confidence Bound 6F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 212. Cumulative PoD curve with 95%

confidence bound—Boeing tap hammer, 6-ply fiberglass

187
 Boeing Tap Hammer Cumulative PoD Curve with 95% Confidence Bound 9C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 213. Cumulative PoD curve with 95%

confidence bound—Boeing tap hammer,9-ply carbon

Boeing Tap Hammer Cumulative PoD Curve with 95% Confidence Bound 9F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 214. Cumulative PoD curve with 95%

confidence bound—Boeing tap hammer, 9-ply fiberglass

188
 LFBT Cumulative PoD Curve with 95% Confidence Bound 3C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 215. Cumulative PoD curve with 95% confidence bound—LFBT, 3-ply carbon

LFBT Cumulative PoD Curve with 95% Confidence Bound 3F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Figure 216. Cumulative PoD curve with 95% confidence bound—LFBT, 3-ply fiberglass

189
 LFBT Cumulative PoD Curve with 95% Confidence Bound 6C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Figure 217. Cumulative PoD curve with 95% confidence bound—LFBT, 6-ply carbon

LFBT Cumulative PoD Curve with 95% Confidence Bound 6F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 218. Cumulative PoD curve with 95% confidence bound—LFBT, 6-ply fiberglass

190
 LFBT Cumulative PoD Curve with 95% Confidence Bound 9C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 219. Cumulative PoD curve with 95% confidence bound—LFBT, 9-ply carbon

LFBT Cumulative PoD Curve with 95% Confidence Bound 9F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 220. Cumulative PoD curve with 95% confidence bound—LFBT, 9-ply fiberglass

191
 MIA Mode Cumulative PoD Curve with 95% Confidence 3C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 221. Cumulative PoD curve with 95% confidence bound—MIA, 3-ply carbon

MIA Mode Cumulative PoD Curve and 95% Confidence Bound 3F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 222. Cumulative PoD curve with 95% confidence bound—MIA, 3-ply fiberglass

192
 MIA Mode Cumulative PoD Curve with 95% Confidence Bound 6C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 223. Cumulative PoD curve with 95% confidence bound—MIA, 6-ply carbon

MIA Mode Cumulative PoD Curve with 95% Confidence Bound 6F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 224. Cumulative PoD curve with 95% confidence bound—MIA, 6-ply fiberglass

193
 MIA Mode Cumulative PoD Curve with 95% Confidence Bound 9C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 225. Cumulative PoD curve with 95% confidence bound—MIA, 9-ply carbon

MIA Mode Cumulative PoD Curve with 95% Confidence Bound 9F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 226. Cumulative PoD curve with 95% confidence bound—MIA, 9-ply fiberglass

194
 Wichitech DTH Cumulative PoD Curve with 95% Confidence Bound 3C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia in Inches)

Figure 227. Cumulative PoD curve with 95%

confidence bound—WichiTech DTH, 3-ply carbon

Wichitech DTH Cumulative PoD Curve with 95% Confidence Bound 3F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia in Inches)

Figure 228. Cumulative PoD curve with 95%

confidence bound—WichiTech DTH, 3-ply fiberglass

195
 Wichitech DTH Cumulative PoD Curve with 95% Confidence Bound 6C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 229. Cumulative PoD curve with 95%

confidence bound—WichiTech DTH, 6-ply carbon

Wichitech DTH Cumulative PoD Curve with 95% Confidence Bound 6F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia in Inches)

Figure 230. Cumulative PoD curve with 95%

confidence bound—WichiTech DTH, 6-ply fiberglass

196
 Wichitech DTH Cumulative PoD Curve with 95% Confidence Bound 9C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 231. Cumulative PoD curve with 95%

confidence bound—WichiTech DTH, 9-ply carbon

Wichitech DTH Cumulative PoD Curve with 95% Confidence Bound 9F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 232. Cumulative PoD curve with 95%

confidence bound—WichiTech DTH, 9-ply fiberglass

197
 Woodpecker Cumulative PoD Curve with 95% Confidence Bound 3C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 233. Cumulative PoD curve with 95%

confidence bound—Woodpecker, 3-ply carbon

Woodpecker Cumulative PoD Curve with 95% Confidence Bound 3F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 234. Cumulative PoD curve with 95%

confidence bound—Woodpecker, 3-ply fiberglass

198
 Woodpecker Cumulative PoD Curve with 95% Confidence Bound 6C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 235. Cumulative PoD curve with 95%

confidence bound—Woodpecker, 6-ply carbon

Woodpecker Cumulative PoD Curve with 95% Confidence Bound 6F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 236. Cumulative PoD curve with 95%

confidence bound—Woodpecker, 6-ply fiberglass

199
 Woodpecker Cumulative PoD Curve with 95% Confidence Bound 9C

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 237. Cumulative PoD curve with 95%

confidence bound—Woodpecker, 9-ply carbon

Woodpecker Cumulative PoD Curve with 95% Confidence Bound 9F

0.9

0.8

0.7
Probability of Detection

0.6

POD(a)
0.5
95% Conf.

0.4

0.3

0.2

0.1

0
0 1 2 3 4
Flaw Size (Dia. in Inches)

Figure 238. Cumulative PoD curve with 95%

confidence bound—Woodpecker, 9-ply fiberglass

200
The PoD curves, along with tabulated values in figures 239‒250, display and compare the
performance of the advanced NDI techniques in each of the specific composite honeycomb
structure categories. Note that PoD curves are not plotted for specific methods when the
detection percentage is exceptionally high. In these cases, it is not possible to generate a curve fit
for this data because of the extremely high flaw detection rate. For these methods, the flaw
detection rate is listed at the top of the PoD plots. However, the 90% PoD values for all methods
are listed in each figure. For each category of 3F, 3C, 6F, 6C, 9F, and 9C, the results are divided
into Tier 1 (top set of performers) and Tier 2 (second set of performers) groups. This allows for
greater clarity in the PoD plots. It can be seen that the best performing advanced NDI methods
produce 90% PoD levels at flaws that are less than 0.5″ diameter.
MAUS Reson. (1) 100% detection
Advanced Inspection Technique Comparisons 3C Shearography 98% detection
Thermography (FLIR) 98% detection
Thermography
Thermography Uncooled Air Coupled UT and MAUS Resonance (2)

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

3-Ply Carbon
Device False Calls 90% POD Value
MAUS Resonance 1 3 ≤.50
Shearography 0 ≤.50
Thermography (FLIR) 0 ≤.50
Thermography 2 ≤.50
Thermography (uncooled) 0 0.52
Air-coupled UT 0 0.52
MAUS Resonance 2 2 0.53

Figure 239. Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 3-ply carbon (Tier 1 results)

201
 Advanced Inspection Technique Comparisons 3C
Laminography MAUS MIA Laser UT
CATT Flaw Inspecta UT Array S.A.M.
Microwave
1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

3-Ply Carbon
Device False Calls 90% POD Value
Laminography 0 0.66
MAUS MIA 0 0.77
Laser UT 4 0.84
CATT 0 1.19
FlawInspecta UT Array 3 1.24
SAM 2 1.26
Microwave 9 2.13

Figure 240. Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 3-ply carbon (Tier 2 results)

202
 Thermography 100% detection
Advanced Inspection Technique Comparisons 3F
Thermography (FLIR) 98% detection
Microwave 98% detection
Thermography Uncooled Shearography S.A.M. Air Coupled UT
1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

3-Ply Fiberglass
Device False Calls 90% POD Value
Thermography 0 ≤.50
Thermography (FLIR) 0 ≤.50
Microwave 2 ≤.50
Thermography (uncooled) 0 0.52
Shearography 1 0.56
SAM 0 0.65
Air-coupled UT 0 0.69

Figure 241. Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 3-ply fiberglass (Tier 1 results)

203
 Advanced Inspection Technique Comparisons 3F

MAUS Reson. (1) MAUS MIA Laminography CATT

MAUS Reson. (2) Laser UT Flaw Inspecta UT Array

0.9

0.8
Probability of Detection

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

3-Ply Fiberglass
Device False Calls 90% POD Value
MAUS Resonance 1 5 0.69
MAUS MIA 0 0.69
Laminography 0 0.97
CATT 4 1.04
MAUS Resonance 2 6 1.18
Laser UT 0 1.25
FlawInspecta UT Array 0 1.78

Figure 242. Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 3-ply fiberglass (Tier 2 results)

204
 Advanced Inspection Technique Comparisons 6C
Thermography (FLIR) 100%
Air Coupled UT Shearography detection
Thermography 97% detection
Laminography MAUS MIA MAUS Reson. (1) 98% detection

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

6-Ply Carbon
Device False Calls 90% POD Value
Thermography (FLIR) 1 ≤.50
Thermography 2 ≤.50
MAUS Resonance 1 12 ≤.50
Air-coupled UT 0 0.87
Shearography 0 1.00
Laminography 0 1.10
MAUS MIA 6 1.32

Figure 243. Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 6-ply carbon (Tier 1 results)

205
 Advanced Inspection Technique Comparisons 6C

S.A.M. Thermography Uncooled CATT

Flaw Inspecta UT Array MAUS Reson. (2) Microwave
1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

6-Ply Carbon
Device False Calls 90% POD Value
SAM 37 1.63
Thermography (uncooled) 2 1.70
CATT 1 1.85
FlawInspecta UT Array 0 1.92
MAUS Resonance 2 4 2.27
Microwave 10 >3.00
Laser UT N/A N/A

Figure 244. Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 6-ply carbon (Tier 2 results)

206
 Advanced Inspection Technique Comparisons 6F Shearography 100% detection
Thermography (FLIR) 98% detection

MAUS Reson. (1) Thermography S.A.M. Air Coupled UT Microwave

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

6-Ply Fiberglass
Device False Calls 90% POD Value
Shearography 0 ≤.50
Thermography (FLIR) 4 ≤.50
MAUS Resonance 1 9 0.55
Thermography 3 0.70
SAM 8 0.84
Air-coupled UT 0 0.90
Microwave 12 0.93

Figure 245. Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 6-ply fiberglass (Tier 1 results)

207
 Advanced Inspection Technique Comparisons 6F

CATT Laminography MAUS MIA

MAUS Reson. (2) Thermography Uncooled Flaw Inspecta UT Array

0.9

0.8
Probability of Detection

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

6-Ply Fiberglass
Device False Calls 90% POD Value
CATT 1 1.10
Laminography 2 1.43
MAUS MIA 6 1.48
MAUS Resonance 2 2 2.07
Thermography (uncooled) 2 2.19
FlawInspecta UT Array 0 >3.00
Laser UT N/A N/A

Figure 246. Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 6-ply fiberglass (Tier 2 results)

208
 Advanced Inspection Technique Comparisons 9C Thermography (FLIR) 97% detection

Thermography Air Coupled UT MAUS Reson. (1)

Laminography Shearography CATT
1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

9-Ply Carbon
Device False Calls 90% POD Value
Thermography (FLIR) 0 ≤.50
Thermography 0 0.64
Air-coupled UT 2 1.26
MAUS Resonance 1 0 1.32
Laminography 0 1.39
Shearography 0 2.00
CATT 4 2.36

Figure 247. Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 9-ply carbon (Tier 1 results)

209
 Advanced Inspection Technique Comparisons 9C
Flaw Inspecta UT Array Thermography Uncooled MAUS MIA
MAUS Reson. (2) Microwave S.A.M.
1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

9-Ply Carbon
Device False Calls 90% POD Value
FlawInspecta UT Array 1 2.85
Thermography (uncooled) 0 >3.00
MAUS MIA 8 >3.00
MAUS Resonance 2 10 >3.00
Microwave 23 >3.00
SAM 74 >3.00
Laser UT N/A N/A

Figure 248. Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 9-ply carbon (Tier 2 results)

210
 Thermography (FLIR) 97% detection
Advanced Inspection Technique Comparisons 9F

MAUS Reson. (1) Thermography Shearography Air Coupled UT S.A.M. Laminography

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

9-Ply Fiberglass
Device False Calls 90% POD Value
Thermography (FLIR) 1 ≤.50
MAUS Resonance 1 2 0.60
Thermography 2 67.00
Shearography 0 0.77
Air-coupled UT 1 1.01
SAM 17 1.04
Laminography 1 1.69

Figure 249. Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 9-ply fiberglass (Tier 1 results)

211
 Advanced Inspection Technique Comparisons 9F

Microwave CATT Flaw Inspecta UT Array MAUS MIA Thermography Uncooled MAUS Reson. (2)

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

9-Ply Fiberglass
Device False Calls 90% POD Value
Microwave 8 2.00
CATT 0 2.27
FlawInspecta UT Array 1 >3.00
MAUS MIA 8 >3.00
Thermography (uncooled) 4 >3.00
MAUS Resonance 2 11 >3.00
Laser UT N/A N/A

Figure 250. Individual PoD curve comparisons along with tabulated values for advanced
inspection methods on 9-ply fiberglass (Tier 2 results)

Figures 163–198 show the individual inspector results and the resulting cumulative 90% PoD
values for each of the conventional inspection methods studied in the CHE. Figures 251–256
provide comparison PoD performance plots for the five conventional NDI methods: Boeing tap
hammer, Airbus tap hammer, LFBT, MIA, WichiTech DTH, and Woodpecker. The data are
divided into the specific composite honeycomb structure categories. It can be seen that the false
calls are quite low, and the best 90% POD levels range from 1–2″ diameter, depending on the
construction scenario.

212
 Cumulative PoD's of All Conventional Devices 3C

Airbus Tap Hammer Boeing Tap Hammer LFBT MIA Wichitech DTH Woodpecker

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

3-Ply Carbon
Device False Calls 90% POD Value
Airbus tap hammer 0.8 2.34
Boeing tap hammer 1.0 2.03
LFBT 2.3 2.10
MIA 0.6 1.46
WichiTech DTH 0.6 2.28
Woodpecker 0.0 1.78

Figure 251. Cumulative PoD curve comparisons along with tabulated values of all
conventional NDI devices on 3-ply carbon (average results for all inspectors)

213
 Cumulative PoD's of All Conventional Devices 3F

Airbus Tap Hammer Boeing Tap Hammer LFBT MIA Wichitech DTH Woodpecker
1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

3-Ply Fiberglass
Device False Calls 90% POD Value
Airbus tap hammer 0.3 2.03
Boeing tap hammer 1.8 1.57
LFBT 0.9 1.77
MIA 0.5 1.04
WichiTech DTH 0.9 1.81
Woodpecker 0.3 1.45

Figure 252. Cumulative PoD curve comparisons along with tabulated values of all
conventional NDI devices on 3-ply fiberglass (average results for all inspectors)

214
 Cumulative PoD's of All Conventional Devices 6C
Airbus Tap Hammer Boeing Tap Hammer LFBT MIA Wichitech DTH Woodpecker
1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

6-Ply Carbon
Device False Calls 90% POD Value
Airbus tap hammer 3.5 >3.00
Boeing tap hammer 4.7 2.59
LFBT 6.5 2.42
MIA 1.3 1.76
WichiTech DTH 0.9 2.18
Woodpecker 0.7 2.23

Figure 253. Cumulative PoD curve comparisons along with tabulated values of all
conventional NDI devices on 6-ply carbon (average results for all inspectors)

215
 Cumulative PoD's of All Conventional Devices 6F

Airbus Tap Hammer Boeing Tap Hammer LFBT MIA Wichitech DTH Woodpecker

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

6-Ply Fiberglass
Device False Calls 90% POD Value
Airbus tap hammer 2.9 2.44
Boeing tap hammer 4.7 2.33
LFBT 3.3 2.55
MIA 1.9 1.49
WichiTech DTH 1.6 1.71
Woodpecker 0.1 2.05

Figure 254. Cumulative PoD curve comparisons along with tabulated values of all
conventional NDI devices on 6-ply fiberglass (average results for all inspectors)

216
 Cumulative PoD's of All Conventional Devices 9C

Airbus Tap Hammer Boeing Tap Hammer LFBT MIA Wichitech DTH Woodpecker

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

9-Ply Carbon
Device False Calls 90% POD Value
Airbus tap hammer 6.8 >3.00
Boeing tap hammer 5.4 2.88
LFBT 3.8 2.61
MIA 3.1 2.97
WichiTech DTH 4.4 2.15
Woodpecker 2.8 2.76

Figure 255. Cumulative PoD curve comparisons along with tabulated values of all
conventional NDI devices on 9-ply carbon (average results for all inspectors)

217
 Cumulative PoD's of All Conventional Devices 9F

Airbus Tap Hammer Boeing Tap Hammer LFBT MIA Wichitech DTH Woodpecker

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

9-Ply Fiberglass
Device False Calls 90% POD Value
Airbus tap hammer 7.1 >3.00
Boeing tap hammer 6.9 2.90
LFBT 2.8 3.00
MIA 6.8 2.00
WichiTech DTH 4.9 2.07
Woodpecker 6.7 2.52

Figure 256. Cumulative PoD curve comparisons along with tabulated values of all
conventional NDI devices on 9-ply fiberglass (average results for all inspectors)

Figures 257–262 provide comparison PoD performance plots for the mechanical tap test
techniques only: Boeing tap hammer, Airbus tap hammer, WichiTech DTH, Woodpecker, and
the CATT system. The data are divided into the specific composite honeycomb structure
categories. It can be seen that the automated CATT system, with its C-scan display of data,
produced the best results and that the best 90% POD values range from 1.0–2.3″ diameter,
depending on the construction scenario.

218
 Cumulative POD's of All Mechanical Impact TapTest Devices - 3 Ply Carbon

Airbus Tap Hammer Boeing Tap Hammer Wichitech DTH Woodpecker CATT

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Figure 257. Comparison of all mechanical tap test devices on 3-ply carbon

Cumulative POD's of All Mechanical Impact TapTest Devices - 3 Ply Fiberglass

Airbus Tap Hammer Boeing Tap Hammer Wichitech DTH Woodpecker CATT

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Figure 258. Comparison of all mechanical tap test devices on 3-ply fiberglass

219
 Cumulative POD's of All Mechanical Impact Tap Test Devices - 6 Ply Carbon

Airbus Tap Hammer Boeing Tap Hammer Wichitech DTH Woodpecker CATT

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Figure 259. Comparison of all mechanical tap test devices on 6-ply carbon

Cumulative POD's of All Mechanical Impact Tap Test Devices - 6 Ply Fiberglass

Airbus Tap Hammer Boeing Tap Hammer Wichitech DTH Woodpecker CATT

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Figure 260. Comparison of all mechanical tap test devices on 6-ply fiberglass

220
 Cumulative POD's of All Mechanical Impact Tap Test Devices - 9 Ply Carbon

Airbus Tap Hammer Boeing Tap Hammer Wichitech DTH Woodpecker CATT

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Figure 261. Comparison of all mechanical tap test devices on 9-ply carbon

Cumulative POD's of All Mechanical Impact Tap Test Devices - 9 Ply Fiberglass

Airbus Tap Hammer Boeing Tap Hammer Wichitech DTH Woodpecker CATT

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Figure 262. Comparison of all mechanical tap test devices on 9-ply fiberglass

221
The results from this experiment can also be divided to show the performance of each individual
conventional inspection method for each of the composite honeycomb structure categories.
Figures 263–268 compare the overall cumulative performance of a specific conventional
inspection technique and show the results from all six construction types (i.e., 3F, 3C, 6F, 6C,
9F, and 9C) on a single plot. Tabulated 90% PoD levels are also compared along with the false
calls associated with composite honeycomb type. These data show the rise in 90% PoD levels as
the inspection becomes more challenging (i.e., skin gets thicker).

Cumulative PoD's for Airbus Tap Hammer on All Panel Types

Cumulative PODs for Airbus Tap Hammer on All Panel Types
1

0.9

0.8

0.7
Probability of Detection

3 Ply Carbon
0.6
3 Ply Fiberglass
6 Ply Carbon
0.5
6 Ply Fiberglass
9 Ply Carbon
0.4
9 Ply Fiberglass
0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Airbus Tap Hammer

Panel Type False Calls 90% POD Value
3-ply carbon 0.8 2.34
3-ply fiberglass 0.3 2.03
6-ply carbon 3.5 >3.00
6-ply fiberglass 2.9 2.44
9-ply carbon 6.8 >3.00
9-ply fiberglass 7.1 >3.00

Figure 263. Cumulative PoD curves for Airbus tap hammer along with tabulated values on
all panel types (average results for all inspectors)

222
 Cumulative PODs for
Cumulative Boeing
PoD's Tap Hammer
for Boeing on Allon
Tap Hammer Panel Types
Panel Types

0.9

0.8

0.7
Probability of Detection

3 Ply Carbon
0.6
3 Ply Fiberglass
6 Ply Carbon
0.5
6 Ply Fiberglass
9 Ply Carbon
0.4
9 Ply Fiberglass
0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Boeing Tap Hammer

Panel Type False Calls 90% POD Value
3-ply carbon 1.0 2.03
3-ply fiberglass 1.8 1.57
6-ply carbon 4.7 2.59
6-ply fiberglass 4.7 2.33
9-ply carbon 5.4 2.88
9-ply fiberglass 6.9 2.90

Figure 264. Cumulative PoD curves for Boeing tap hammer along with tabulated values on
all panel types (average results for all inspectors)

223
 Cumulative
CumulativePODs
PoD'sfor
forLFBT
LFBTon
onAll
AllPanel
PanelTypes
Types

0.9

0.8

0.7
Probability of Detection

3 Ply Carbon
0.6
3 Ply Fiberglass
6 Ply Carbon
0.5
6 Ply Fiberglass
9 Ply Carbon
0.4
9 Ply Fiberglass
0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

LFBT
Panel Type False Calls 90% POD Value
3-ply carbon 2.3 2.10
3-ply fiberglass 0.9 1.77
6-ply carbon 6.5 2.42
6-ply fiberglass 3.3 2.55
9-ply carbon 3.8 2.61
9-ply fiberglass 2.8 3.00

Figure 265. Cumulative PoD curves for LFBT along with tabulated values on all panel
types (average results for all inspectors)

224
 Cumulative PODs
Cumulative for MIA
PoD's on on
for MIA AllAll
Panel Types
Panel Types

0.9

0.8

0.7
Probability of Detection

3 Ply Carbon
0.6
3 Ply Fiberglass
6 Ply Carbon
0.5
6 Ply Fiberglass
9 Ply Carbon
0.4
9 Ply Fiberglass
0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

MIA
Panel Type False Calls 90% POD Value
3-ply carbon 0.6 1.46
3-ply fiberglass 0.5 1.04
6-ply carbon 1.3 1.76
6-ply fiberglass 1.9 1.49
9-ply carbon 3.1 2.97
9-ply fiberglass 6.8 2.00

Figure 266. Cumulative PoD curves for MIA along with tabulated values on all panel types
(average results for all inspectors)

225
 Cumulative
CumulativePODs
PoD'sfor
forWichiTech
WichitechDTH
DTHon
onAll
AllPanel
PanelTypes
Types

0.9

0.8

0.7
Probability of Detection

3 Ply Carbon
0.6
3 Ply Fiberglass
6 Ply Carbon
0.5
6 Ply Fiberglass
9 Ply Carbon
0.4
9 Ply Fiberglass
0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia in Inches)

WichiTech DTH
Panel Type False Calls 90% POD Value
3-ply carbon 0.6 2.28
3-ply fiberglass 0.9 1.81
6-ply carbon 0.9 2.18
6-ply fiberglass 1.6 1.71
9-ply carbon 4.4 2.15
9-ply fiberglass 4.9 2.07

Figure 267. Cumulative PoD curves for WichiTech DTH along with tabulated values on all
panel types (average results for all inspectors)

226
 Cumulative PODs
Cumulative forforWoodpecker
PoD's Woodpecker on
on All
All Panel Types
Panel Types

0.9

0.8

0.7
Probability of Detection

3 Ply Carbon
0.6
3 Ply Fiberglass
6 Ply Carbon
0.5
6 Ply Fiberglass
9 Ply Carbon
0.4
9 Ply Fiberglass
0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Woodpecker
Panel Type False Calls 90% POD Value
3-ply carbon 0.0 1.78
3-ply fiberglass 0.3 1.45
6-ply carbon 0.7 2.32
6-ply fiberglass 0.1 2.05
9-ply carbon 2.8 2.76
9-ply fiberglass 6.7 2.52

Figure 268. Cumulative PoD curves for Woodpecker along with tabulated values on all
panel types (average results for all inspectors)

227
The results from this experiment can also be divided to show the performance of each individual
advanced inspection method for each of the composite honeycomb structure categories. Figures
269–278 compare the overall cumulative performance of a specific advanced inspection
technique and show the results from all six construction types (i.e., 3F, 3C, 6F, 6C, 9F, and 9C)
on a single plot. Tabulated 90% PoD levels are also compared along with the false calls
associated with composite honeycomb type. These data show the rise in 90% PoD levels as the
inspection becomes more challenging (i.e., skin gets thicker). Overall, when both 90% PoD
levels and false calls are considered, thermography provided the best overall performance. Note
that PoD curves are not plotted for specific methods when the detection percentage is
exceptionally high. In these cases, it is not possible to generate a curve fit for these data because
of the extremely high flaw detection rate. For these methods, the flaw detection rate is listed at
the top of the PoD plots.
PoD's for CATT on All Panel Types
PODs for CATT on All Panel Types
1

0.9

0.8

0.7
Probability of Detection

3 Ply Carbon
0.6 3 Ply Fiberglass
6 Ply Carbon
0.5
6 Ply Fiberglass
0.4 9 Ply Carbon
9 Ply Fiberglass
0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

CATT
Panel Type False Calls 90% POD Value
3-ply carbon 0.0 1.19
3-ply fiberglass 4.0 1.04
6-ply carbon 1.0 1.85
6-ply fiberglass 1.0 1.10
9-ply carbon 4.0 2.36
9-ply fiberglass 0.0 2.27

Figure 269. PoD curve comparisons for CATT along with tabulated values on all panel
types

228
 3 Ply Carbon 100% Detection
PoD's for MAUS Resonance 1 on Panel Types
PODs for MAUS Resonance 1 on All Panel Types 6 Ply Carbon 98% Detection

0.9

0.8

0.7
Probability of Detection

0.6
3 Ply Fiberglass
6 Ply Fiberglass
0.5
9 Ply Carbon
9 Ply Fiberglass
0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

MAUS Resonance 1
Panel Type False Calls 90% POD Value
3-ply carbon 3.0 ≤.50
3-ply fiberglass 5.0 0.69
6-ply carbon 12.0 ≤.50
6-ply fiberglass 9.0 0.55
9-ply carbon 0.0 1.32
9-ply fiberglass 2.0 0.60

Figure 270. PoD curve comparisons for MAUS Resonance 1 inspections along with
tabulated values on all panel types (results shown for resonance experimenter #1 only)

229
 PODs for SAMfor
PoD's onS.A.M.
All Panel Types
on All Panel Types

0.9

0.8

0.7
Probability of Detection

3 Ply Carbon
0.6
3 Ply Fiberglass
6 Ply Carbon
0.5
6 Ply Fiberglass
9 Ply Carbon
0.4
9 Ply Fiberglass
0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

SAM
Panel Type False Calls 90% POD Value
3-ply carbon 2.0 1.26
3-ply fiberglass 0.0 0.65
6-ply carbon 37.0 1.63
6-ply fiberglass 8.0 0.84
9-ply carbon 74.0 >3.00
9-ply fiberglass 17.0 1.04

Figure 271. PoD curve comparisons for SAM inspections along with tabulated
values on all panel types

230
 PODs forfor
PoD's Shearography ononAllAllPanel
Shearography PanelTypes
Types 3 Ply Carbon 98% Detection
6 Ply Fiberglass 100% Detection

0.9

0.8

0.7
Probability of Detection

0.6
3 Ply Fiberglass
6 Ply Carbon
0.5
9 Ply Carbon
9 Ply Fiberglass
0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Shearography
Panel Type False Calls 90% POD Value
3-ply carbon 0.0 ≤.50
3-ply fiberglass 1.0 0.56
6-ply carbon 0.0 1.00
6-ply fiberglass 0.0 ≤.50
9-ply carbon 0.0 2.00
9-ply fiberglass 0.0 0.77

Figure 272. PoD curve comparisons for shearography inspections along with tabulated
values on all panel types

231
 3 Ply Fiberglass 100% Detection
PODsPoD's for Thermography
for Thermography on All
on All Panel
Panel Types
Types 6 Ply Carbon 97% Detection

0.9

0.8

0.7
Probability of Detection

0.6
3 Ply Carbon
6 Ply Fiberglass
0.5
9 Ply Carbon
9 Ply Fiberglass
0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Thermography
Panel Type False Calls 90% POD Value
3-ply carbon 2.0 ≤.50
3-ply fiberglass 0.0 ≤.50
6-ply carbon 2.0 ≤.50
6-ply fiberglass 3.0 0.70
9-ply carbon 0.0 0.64
9-ply fiberglass 2.0 0.67

Figure 273. PoD curve comparisons for thermography inspections along with tabulated
values on all panel types

232
 PoD's forMAUS
PODs for MAUS MIA
MIA on on All Panel
All Panel TypesTypes

0.9

0.8
Probability of Detection

0.7
3 Ply Carbon
0.6 3 Ply Fiberglass
6 Ply Carbon
0.5
6 Ply Fiberglass
0.4 9 Ply Carbon
9 Ply Fiberglass
0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

MAUS MIA
Panel Type False Calls 90% POD Value
3-ply carbon 0.0 0.77
3-ply fiberglass 0.0 0.69
6-ply carbon 6.0 1.32
6-ply fiberglass 6.0 1.48
9-ply carbon 8.0 >3.00
9-ply fiberglass 8.0 >3.00

Figure 274. PoD curve comparisons for MAUS IV MIA inspections along with tabulated
values on all panel types

233
 PODs
PoD's for Air for AC-UT
Coupled (TTU Mode) (TTU
Ultrasonics on All Mode)
Panel Types
on All Panel Types

0.9

0.8
Probability of Detection

0.7
3 Ply Carbon
0.6 3 Ply Fiberglass
6 Ply Carbon
0.5
6 Ply Fiberglass
9 Ply Carbon
0.4
9 Ply Fiberglass
0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

AC-UT
Panel Type False Calls 90% POD Value
3-ply carbon 0.0 0.52
3-ply fiberglass 0.0 0.69
6-ply carbon 0.0 0.87
6-ply fiberglass 0.0 0.90
9-ply carbon 2.0 1.26
9-ply fiberglass 1.0 1.01

Figure 275. PoD curve comparisons for AC-UT inspections (TTU mode) along with
tabulated values on all panel types

234
 PODs forfor
PoD's Laminography on All Panel
Laminography on AllTypes
Panel Types

0.9

0.8

0.7
Probability of Detection

3 Ply Carbon
0.6
3 Ply Fiberglass
6 Ply Carbon
0.5
6 Ply Fiberglass
9 Ply Carbon
0.4
9 Ply Fiberglass

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Laminography
Panel Type False Calls 90% POD Value
3-ply carbon 0.0 0.66
3-ply fiberglass 0.0 0.97
6-ply carbon 0.0 1.10
6-ply fiberglass 2.0 1.43
9-ply carbon 0.0 1.39
9-ply fiberglass 1.0 1.69

Figure 276. PoD curve comparisons for laminography inspections along with tabulated
values on all panel types

235
 PoD's forMicrowave
PODs for Microwaveon on All Panel
All Panel TypesTypes 3 Ply Fiberglass 98% Detection

3 Ply Carbon 6 Ply Carbon 6 Ply Fiberglass 9 Ply Carbon 9 Ply Fiberglass

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3 3.5
Flaw Size (Dia. in Inches)

Microwave
Panel Type False Calls 90% POD Value
3-ply carbon 9.0 2.13
3-ply fiberglass 2.0 0.50
6-ply carbon 10.0 >3.00
6-ply fiberglass 12.0 0.93
9-ply carbon 23.0 >3.00
9-ply fiberglass 8.0 2.00

Figure 277. PoD curve comparisons for microwave inspections along with tabulated values
on all panel types

236
 PoD's forfor
PODs FlawInspecta UTArray
FlawInspecta UT Arrayonon
All All Panel
Panel TypesTypes

3 Ply Carbon 3 Ply Fiberglass 6 Ply Carbon 6 Ply Fiberglass 9 Ply Carbon 9 Ply Fiberglass

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3 3.5
Flaw Size (Dia. in Inches)

FlawInspecta UT Array
Panel Type False Calls 90% POD Value
3-ply carbon 3.0 1.24
3-ply fiberglass 0.0 1.78
6-ply carbon 0.0 1.92
6-ply fiberglass 0.0 >3.00
9-ply carbon 1.0 2.85
9-ply fiberglass 1.0 >3.00

Figure 278. PoD curve comparisons for FlawInspecta UT array along with tabulated values
on all panel types

The laminography inspections were completed edge to edge with no overlap between adjacent
shots. This made detection of flaws at these edge junctions difficult and was probably the cause
for the flaw misses in these areas. In addition, impact flaws were very difficult for laminography
to detect, more so than any other flaw type. It is believed that if more time were spent looking at
various time slices, there possibly could have been a higher detection rate. The PoD curves in
figures 279–282 compare the overall cumulative performance of laminography while considering
the different inspection challenges noted above. The plots show the results produced when all
flaws are considered and compares them with the curves produced when certain exceptions are
removed from the flaw data. These include removing impact flaw data (difficult to detect) and

237
flaws located in areas where there was a lack of image overlap that resulted in misses. It can be
seen that these adjustments result in a significant improvement in the 90% PoD levels. This
indicates that laminography improvements in impact detection and changes in inspection
deployment procedures could make laminography one of the top performing inspection options.
Laminography 6 Ply Carbon Inspection Results

All Flaws No Impact Flaws No Edge Flaw Misses

1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3 3.5
Flaw Size (Dia. in Inches)

Figure 279. PoD curve comparisons for laminography showing the improvement when
impact and edge (inspection image overlap) flaws are removed from PoD calculation on
6-ply carbon

Laminography 6 Ply Fiberglass Inspection Results

All Flaws No Impact Flaws No Edge Flaw Misses

1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3 3.5
Flaw Size (Dia. in Inches)

Figure 280. PoD curve comparisons for laminography showing the improvement when
impact and edge (inspection image overlap) flaws are removed from PoD calculation on
6-ply fiberglass

238
 Laminography 9 Ply Carbon Inspection Results

All Flaws No Impact Flaws No Edge Flaw Misses

1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3 3.5
Flaw Size (Dia. in Inches)

Figure 281. PoD curve comparisons for laminography showing the improvement when
impact and edge (inspection image overlap) flaws are removed from PoD calculation on
9-ply carbon

Laminography 9 Ply Fiberglass Inspection Results

All Flaws No Impact Flaws No Edge Flaw Misses

1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3 3.5
Flaw Size (Dia. in Inches)

Figure 282. PoD curve comparisons for laminography showing the improvement when
impact and edge (inspection image overlap) flaws are removed from POD calculation on
9-ply fiberglass

239
Similarly, when the laser UT method was applied to the three-ply fiberglass test specimens,
multiple flaws in the center region of specimens were missed. This is probably because of the
difficulties faced by experimenters engaged in data acquisition in this region. The PoD curve and
tabulated values shown in figure 283 compare the overall cumulative performance of laser UT.
The plots show the results produced when all flaws are considered and compares them with the
curves produced when certain exceptions are removed from the flaw data. These include
removing flaws located near the center area of the test specimens (difficulty in detecting due to
system’s inability to attenuate the laser light from the shiny painted surface). If these flaws are
omitted from the PoD calculations, the detection percentage improves, and the 90% PoD level
reduces to 0.77″ diameter. Once again, this example shows that the laser UT performance could
be improved with some changes in the system tuning and inspection procedures.
PODs
PoD'sfor
forLaser UTUT
Laser on on
All Panel Types
All Panel Types

0.9

0.8

0.7
Probability of Detection

0.6
3 Ply Carbon
0.5 3 Ply Fiberglass - Original
3 Ply Fiberglass - w/Exceptions
0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

Laser UT
Panel Type False Calls 90% POD Value
3-ply carbon 4.0 0.84
3-ply fiberglass 0.0 1.25
6-ply carbon N/A N/A
6-ply fiberglass N/A N/A
9-ply carbon N/A N/A
9-ply fiberglass N/A N/A

Figure 283. PoD curve comparisons along with tabulated values for laser-ultrasonic
inspections with exception that includes removing flaws located in center area of specimen
from PoD calculations on 3-ply fiberglass

240
The final comparison matches the top performing advanced NDI methods in each honeycomb
construction category with the best performing conventional NDI method. The PoD curves and
tabulated values in figures 284–289 show the improvements that can be achieved through the
application of advanced NDI methods to inspection composite honeycomb structures. In general,
the level of improvement becomes higher as the inspection challenge increases
(i.e., skin becomes thicker and moves from fiberglass to carbon). The 90% POD improvements
range from 50–75%.
Comparison of Advanced Inspection Techniques MAUS Reson. (1) 100% detection
Shearography 98% detection
with Best Conventional NDI Result on 3 Ply Carbon Thermography (FLIR) 98% detection
Thermography
Thermography Uncooled, Air Coupled UT, and MAUS Resonance (2)
MIA
1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

3-Ply Carbon
Device False Calls 90% POD Value
MAUS Resonance 1 3.0 ≤.50
Shearography 0.0 ≤.50
Thermography (FLIR) 0.0 ≤.50
Thermography 2.0 ≤.50
Thermography (uncooled) 0.0 0.52
Air-coupled UT 0.0 0.52
MAUS Resonance 2 2.0 0.53
MIA 0.6 1.46

(Highlighted = best conventional method for comparison purposes)

Figure 284. PoD curve comparisons along with tabulated values for advanced inspection
methods and best performing conventional method on 3-ply carbon

241
 Comparison of Advanced Inspection Techniques Thermography 98-100% detection
with Best Conventional NDI Result on 3 Ply Fiberglass Microwave 98% detection

Thermography Uncooled Shearography S.A.M. Air Coupled UT MIA

1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

3-Ply Fiberglass
Device False Calls 90% POD Value
Thermography 0.0 ≤.50
Thermography (FLIR) 0.0 ≤.50
Microwave 2.0 ≤.50
Thermography (uncooled) 0.0 0.52
Shearography 1.0 0.56
SAM 0.0 0.65
Air-coupled UT 0.0 0.69
MIA 0.5 1.04

(Highlighted = best conventional method for comparison purposes)

Figure 285. PoD curve comparisons along with tabulated values for advanced inspection
methods and best performing conventional method on 3-ply fiberglass

242
 Comparison of Advanced Inspection Techniques
with Best Conventional NDI Result on 6 Ply Carbon Thermography 97-100% detection
Air Coupled UT Shearography MAUS Reson. (1) 98% detection
Laminography MAUS MIA
MIA
1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

6-Ply Carbon
Device False Calls 90% POD Value
Thermography (FLIR) 1.0 ≤.50
Thermography 2.0 ≤.50
MAUS Resonance 1 12.0 ≤.50
Air-coupled UT 0.0 0.87
Shearography 0.0 1.00
Laminography 0.0 1.10
MAUS MIA 6.0 1.32
MIA 1.3 1.76

(Highlighted = best conventional method for comparison purposes)

Figure 286. PoD curve comparisons along with tabulated values for advanced inspection
methods and best performing conventional method on 6-ply carbon

243
 Comparison of Advanced Inspection Techniques Shearography 100% detection
Thermography (FLIR) 98% detection
with Best Conventional NDI Result on 6 Ply Fiberglass
MAUS Reson. (1) Thermography S.A.M. Air Coupled UT Microwave MIA
1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

6-Ply Fiberglass
Device False Calls 90% POD Value
Shearography 0.0 ≤.50
Thermography (FLIR) 4.0 ≤.50
MAUS Resonance 1 9.0 0.55
Thermography 3.0 0.70
SAM 8.0 0.84
Air-coupled UT 0.0 0.90
Microwave 12.0 0.93
MIA 1.9 1.49

(Highlighted = best conventional method for comparison purposes)

Figure 287. PoD curve comparisons along with tabulated values for advanced inspection
methods and best performing conventional method on 6-ply fiberglass

244
 Comparison of Advanced Inspection Techniques
Thermography (FLIR) 97% detection
with Best Conventional NDI Result on 9 Ply Carbon
Thermography Air Coupled UT MAUS Reson. (1)
Laminography Shearography Wichitech DTH
1

0.9

0.8

0.7
Probability of Detection

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

9-Ply Carbon
Device False Calls 90% POD Value
Thermography (FLIR) 0.0 ≤.50
Thermography 0.0 0.64
Air-coupled UT 2.0 1.26
MAUS Resonance 1 0.0 1.32
Laminography 0.0 1.39
Shearography 0.0 2.00
CATT 4.0 2.36
WitchiTech DTH 4.4 2.15

(Highlighted = best conventional method for comparison purposes)

Figure 288. PoD curve comparisons along with tabulated values for advanced inspection
methods and best performing conventional method on 9-ply carbon

245
 Comparison of Advanced Inspection Techniques Thermography (FLIR) 97% detection
with Best Conventional NDI Result on 9 Ply Fiberglass

MAUS Reson. (1) Thermography Shearography Air Coupled UT S.A.M. Laminography MIA

0.9

0.8
Probability of Detection

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5 3
Flaw Size (Dia. in Inches)

9-Ply Fiberglass
Device False Calls 90% POD Value
Thermography (FLIR) 1.0 ≤.50
MAUS Resonance 1 2.0 0.60
Thermography 2.0 0.67
Shearography 0.0 0.77
Air-coupled UT 1.0 1.01
SAM 17.0 1.04
Laminography 1.0 1.69
MIA 6.8 2.00

(Highlighted = best conventional method for comparison purposes)

Figure 289. PoD curve comparisons along with tabulated values for advanced inspection
methods and best performing conventional method on 9-ply fiberglass

246
While the time to complete the inspections was considered a secondary variable, timing data
were recorded so that some trends analysis could be completed. Tables 10–13 compare
inspection times for each of the conventional and advanced NDI techniques in each of the
specific composite honeycomb structure categories. Results for the conventional NDI methods
indicate fairly consistent results for the average time to inspect each 18″ x 18″ panel, as well as
the extreme values of minimum and maximum inspection times. Some of the advanced NDI
methods that used automated scanning deployment completed the inspections in less time;
however, when data analysis time was included in the timing calculations, many advanced NDI
methods took slightly longer than the conventional NDI methods.

Table 10. Average inspection times per 18″ x 18″ panel and max/min times for 3-ply panels
—conventional NDI methods

3-Ply Carbon Inspection Time Comparisons

Inspection Device Low Insp. Time High Insp. Time Ave. Insp. Time
Airbus tap hammer 5.2 16.8 11.7
Boeing tap hammer 5.2 25.2 12.8
LFBT 6.8 23.5 13.8
MIA 7.8 20.3 13.7
WichiTech DTH 9.7 23.3 14.7
Woodpecker 6.8 22.7 15.7

3-Ply Fiberglass Inspection Time Comparisons

Inspection Device Low Insp. Time High Insp. Time Ave. Insp. Time
Airbus tap hammer 6.0 16.7 10.9
Boeing tap hammer 5.0 25.8 13.6
LFBT 6.7 28.7 16.6
MIA 9.7 32.5 18.0
WichiTech DTH 8.7 30.5 16.7
Woodpecker 7.8 21.3 15.7
(Times shown in minutes)

247
Table 11. Average inspection times per 18″ x 18″ panel and max/min times for 6-ply panels
—conventional NDI methods

6-Ply Carbon Inspection Time Comparisons

Inspection Device Low Insp. Time High Insp. Time Ave. Insp. Time
Airbus tap hammer 6.5 21.6 12.3
Boeing tap hammer 5.1 26.6 11.8
LFBT 6.8 30.1 14.2
MIA 9.4 24.8 13.9
WichiTech DTH 5.8 20.8 13.1
Woodpecker 5.8 22.0 13.0

6-Ply Fiberglass Inspection Time Comparisons

Inspection Device Low Insp. Time High Insp. Time Ave. Insp. Time
Airbus tap hammer 6.4 23.9 12.8
Boeing tap hammer 6.4 28.3 12.9
LFBT 6.5 20.9 12.2
MIA 9.0 34.3 16.6
WichiTech DTH 8.0 21.4 14.5
Woodpecker 4.4 23.6 15.1
(Times shown in minutes)

Table 12. Average inspection times per 18″ x 18″ panel and max/min times for 9-ply panels
—conventional NDI methods

9-Ply Carbon Inspection Time Comparisons

Inspection Device Low Insp. Time High Insp. Time Ave. Insp. Time
Airbus tap hammer 4.3 16.3 9.6
Boeing tap hammer 6.8 28.4 12.0
LFBT 9.5 19.3 12.4
MIA 8.4 16.8 12.2
WichiTech DTH 6.8 27.4 13.6
Woodpecker 5.0 22.5 12.5
(Times shown in minutes)

248
Table 12. Average inspection times per 18″ x 18″ panel and max/min times for 9-ply panels
—conventional NDI methods (continued)

9-Ply Fiberglass Inspection Time Comparisons

Inspection Device Low Insp. Time High Insp. Time Ave. Insp. Time
Airbus tap hammer 4.4 16.1 10.1
Boeing tap hammer 5.6 28.4 11.6
LFBT 8.0 20.4 12.2
MIA 8.9 41.5 16.5
WichiTech DTH 7.1 23.5 15.0
Woodpecker 5.1 21.0 13.3
(Times shown in minutes)

Table 13. Average inspection times per 18″ x 18″ panel and max/min times for all panels—
advanced NDI methods (times shown in minutes)

Advanced Inspection Methods’ Time Comparisons (all panel types)

3-Ply 6-Ply 6-Ply 9-Ply
3-Ply Carbon 9-Ply Carbon
Fiberglass Carbon Fiberglass Fiberglass
Ave. Insp. Ave. Insp.
Inspection Device Ave. Insp. Ave. Insp. Ave. Insp. Ave. Insp.
Time Per Time Per
Time Per Time Per Time Per Time Per
Panel Panel
Panel Panel Panel Panel
CATT 21.0 23.8 23.8 24.6 25.4 25.9
MAUS Resonance 1 N/A N/A N/A N/A N/A N/A
SAM 34.3 29.0 27.5 30.3 31.8 26.5
Shearography 15.5 19.8 14.8 14.1 10.8 10.5
Thermography 12.0 12.0 13.0 12.0 13.8 13.9
MAUS Resonance 2 15.0 15.0 15.0 15.0 15.0 15.0
MAUS MIA 40.3 41.0 40.8 42.9 49.1 49.9
Laminography 36.0 36.0 36.0 36.0 36.0 36.0
Microwave 44.0 44.0 44.0 44.0 44.0 43.0
Thermography
N/A N/A N/A N/A N/A N/A
(uncooled)
Thermography
N/A N/A N/A N/A N/A N/A
(FLIR)
FlawInspecta UT
8.0 9.0 8.0 8.0 9.0 6.0
Array
Air-coupled UT 18.0 18.0 18.0 18.0 18.0 18.0
Laser UT N/A N/A N/A N/A N/A N/A

249
 Table 14 summarizes the false calls generated by the individual inspector for a particular
inspection device/method in each of the specific composite honeycomb structure categories. It
also tabulates the average number of false calls for each inspection method. Table 15 compares
the average false call rates for each of the inspection methods and highlights the lowest (yellow
highlight) and highest (red highlights) false call rates in each of the honeycomb construction
categories. The Woodpecker device produced the lowest overall false call rate while the Boeing
tap hammer produced the highest overall false call rate. Table 16 compares the false calls
produced by each of the advanced NDI methods. The lowest number of false calls occurred with
the laminography and CATT system, while, in general, the number of false calls did not seem to
increase as the inspections became more challenging (i.e., thicker skin laminates).
Table 14. Summary of individual and average false calls for all inspectors and
false call averages per inspection method and panel type—conventional NDI

Airbus Tap Hammer False Calls

Panel Type
AT-1 AT-2 AT-3 AT-4 AT-5 AT-6 AT-7 AT-8 Ave.
/Insp. ID
3-ply carbon 0 1 1 0 0 1 3 0 0.8
3-ply fiberglass 1 0 1 0 0 0 0 0 0.3
6-ply carbon 2 3 1 4 4 9 2 3 3.5
6-ply fiberglass 1 2 13 0 1 0 6 0 2.9
9-ply carbon 13 2 26 1 1 3 6 2 6.8
9-ply fiberglass 13 11 23 2 3 2 2 1 7.1
Total 30 19 65 7 9 15 19 6 21.3

Boeing Tap Hammer False Calls

Panel Type
BT-1 BT-2 BT-3 BT-4 BT-5 BT-6 BT-7 BT-8 BT-9 Ave.
/Insp. ID
3-ply carbon 0 1 3 1 1 2 0 1 0 1.0
3-ply fiberglass 4 3 3 0 0 6 0 0 0 1.8
6-ply carbon 0 4 6 4 8 13 2 2 3 4.7
6-ply fiberglass 0 6 6 1 7 17 2 2 1 4.7
9-ply carbon 3 2 7 2 14 14 5 1 1 5.4
9-ply fiberglass 1 7 6 1 25 18 3 1 0 6.9
Total 8 23 31 9 55 70 12 7 5 24.4

LFBT False Calls

Panel Type LFBT- LFBT- LFBT- LFBT- LFBT- LFBT- LFBT- LFBT- LFBT- LFBT- LFBT-
Ave.
/Insp. ID 1 2 3 4 5 6 7 8 9 10 11
3-ply carbon 0 0 14 0 0 0 2 2 0 0 7 2.3
3-ply fiberglass 0 0 8 0 1 0 0 1 0 0 0 0.9
6-ply carbon 0 0 33 1 0 0 1 0 2 0 34 6.5
6-ply fiberglass 0 0 24 0 6 0 1 2 1 0 2 3.3
9-ply carbon 0 0 1 3 1 0 4 0 0 0 33 3.8
9-ply fiberglass 0 0 0 3 0 0 0 0 3 0 25 2.8
Total 0 0 80 7 8 0 8 5 6 0 101 19.5

250
 Table 14. Summary of individual and average false calls for all inspectors and
false call averages per inspection method and panel type—conventional NDI (continued)

MIA False Calls

Panel Type
MIA-1 MIA-2 MIA-3 MIA-4 MIA-5 MIA-6 MIA-7 MIA-8 MIA-9 MIA-10 Ave.
/Insp. ID
3-ply carbon 1 1 0 4 0 0 0 0 0 0 0.6
3-ply fiberglass 2 1 0 1 0 0 1 0 0 0 0.5
6-ply carbon 0 0 0 4 0 0 1 1 6 1 1.3
6-ply fiberglass 1 0 0 0 0 0 1 0 15 2 1.9
9-ply carbon 0 1 3 6 1 4 4 5 N/A 4 3.1
9-ply fiberglass 0 2 2 5 0 2 15 0 36 6 6.8
Total 4 5 5 20 1 6 22 6 57 13 13.9

WitchiTech DTH False Calls

Panel Type
DTH-1 DTH-2 DTH-3 DTH-4 DTH-5 DTH-6 DTH-7 DTH-8 DTH-9 Ave.
/Insp. ID
3-ply carbon 0 5 0 0 0 0 0 0 0 0.6
3-ply fiberglass 0 8 0 0 0 0 0 0 0 0.9
6-ply carbon 0 6 0 1 0 0 0 1 0 0.9
6-ply fiberglass 0 8 0 0 0 6 0 0 0 1.6
9-ply carbon 0 5 0 0 0 31 0 0 4 4.4
9-ply fiberglass 0 12 0 0 0 32 0 0 0 4.9
Total 0 44 0 1 0 69 0 1 4 13.2

Woodpecker False Calls

Panel Type WOOD- WOOD- WOOD- WOOD- WOOD- WOOD- WOOD- WOOD- WOOD-
Ave.
/Insp. ID 1 2 3 4 5 6 7 8 9
3-ply carbon 0 0 0 0 0 0 0 0 0 0.0
3-ply fiberglass 0 0 0 3 0 0 0 0 0 0.3
6-ply carbon 0 0 0 0 1 1 4 0 0 0.7
6-ply fiberglass 0 0 0 0 0 0 0 0 1 0.1
9-ply carbon 0 1 2 2 1 9 0 8 2 2.8
9-ply fiberglass 5 0 5 16 3 13 5 6 7 6.7
Total 5 1 7 21 5 23 9 14 10 10.6

251
 Table 15. Average number of false calls produced by conventional NDI devices on carbon
and fiberglass specimens

Conventional Inspection Technique False Call Averages

Airbus Boeing
Panel Type / WichiTech
Tap Tap LFBT MIA Woodpecker
Insp. Method DTH
Hammer Hammer
3-ply carbon 0.8 1.0 2.3 0.6 0.6 0.0
3-ply fiberglass 0.3 1.8 0.9 0.5 0.9 0.3
6-ply carbon 3.5 4.7 6.5 1.3 0.9 0.7
6-ply fiberglass 2.9 4.7 3.3 1.9 1.6 0.1
9-ply carbon 6.8 5.4 3.8 3.1 4.4 2.8
9-ply fiberglass 7.1 6.9 2.8 6.8 4.9 6.7
Total 21.3 24.4 19.5 13.9 13.2 10.6
Yellow highlights = lowest average number of false calls
Red highlights = highest average number of false calls

Table 16. False calls generated by advanced inspection techniques

(arranged by panel type)

Advanced Inspection Technique False Calls

Air- Thermo
Panel Type / Shearo Lamino Laser FlawInspecta Thermography
coupled graphy
Insp. Method graphy graphy UT UT Array (uncooled)
UT (FLIR)
3-ply carbon 0 0 0 4 3 0 0
3-ply fiberglass 1 0 0 0 0 0 0
6-ply carbon 0 0 0 N/A 0 1 2
6-ply fiberglass 0 2 0 N/A 0 4 2
9-ply carbon 0 0 2 N/A 1 0 0
9-ply fiberglass 0 1 1 N/A 1 1 4
Total 0 3 3 4 5 6 8

252
 Table 16. False calls generated by advanced inspection techniques
(arranged by panel type) (continued)

Advanced Inspection Technique False Calls

Panel Type / Thermo MAUS MAUS MAUS
CATT Microwave SAM
Insp. Method graphy MIA Resonance 1 Resonance 2
3-ply carbon 2 0 0 3 2 9 2
3-ply fiberglass 0 4 0 5 6 2 0
6-ply carbon 2 1 6 12 4 10 37
6-ply fiberglass 3 1 6 9 2 12 8
9-ply carbon 0 4 8 0 10 23 74
9-ply fiberglass 2 0 8 2 11 8 17
Total 9 10 28 31 35 64 138

The test panels used in this experiment contained flaws on both sides of the honeycomb core.
Each side was treated as a separate specimen number, as front-side flaw detection was the
primary assessment being conducted. However, some NDI methods were able to detect flaws on
the backside of the specimens (i.e., signal penetration provided some resolution of damage on the
far side of the test specimen). Such detections of far-side flaws—flaws on the side of the panel
opposite the inspection surface—were recorded. Tables 17–19 summarize detection of far side
flaws for each composite honeycomb structure category. This is a secondary variable, and
detection of these flaws was not required in the experiment.

Table 17. Backside flaw detection for all inspectors and average backside flaws detected
per inspection method and panel type—conventional NDI

Airbus Tap Hammer Backside Calls

Panel Type AT-1 AT-2 AT-3 AT-4 AT-5 AT-6 AT-7 AT-8 Ave.
3-ply carbon 0 7 0 4 3 0 1 0 1.9
3-ply fiberglass 0 0 0 3 0 0 3 0 0.8
6-ply carbon 0 13 1 9 5 4 10 0 5.3
6-ply fiberglass 1 11 2 4 9 3 11 0 5.1
9-ply carbon 1 12 12 13 15 9 10 0 9.0
9-ply fiberglass 3 15 11 10 21 9 15 1 10.6
Total 5 58 26 43 53 25 50 1 32.6

253
 Table 17. Backside flaw detection for all inspectors and average backside flaws detected
per inspection method and panel type—conventional NDI (continued)

Boeing Tap Hammer Backside Calls

Panel Type BT-1 BT-2 BT-3 BT-4 BT-5 BT-6 BT-7 BT-8 BT-9 Ave.
3-ply carbon 1 1 1 2 9 1 2 3 2 2.4
3-ply fiberglass 3 1 1 0 4 0 0 2 1 1.3
6-ply carbon 0 1 3 1 16 2 2 6 3 3.8
6-ply fiberglass 2 3 2 1 8 1 7 2 9 3.9
9-ply carbon 1 0 3 1 22 3 10 2 2 4.9
9-ply fiberglass 2 2 2 2 22 6 16 8 6 7.3
Total 9 8 12 7 81 13 37 23 23 23.7

LFBT Backside Calls

LFBT- LFBT- LFBT- LFBT- LFBT- LFBT- LFBT- LFBT- LFBT- LFBT- LFBT-
Panel Type Ave.
1 2 3 4 5 6 7 8 9 10 11
3-ply carbon 6 0 0 0 0 0 0 5 10 15 1 3.4
3-ply fiberglass 0 0 1 0 0 0 0 8 3 21 1 3.1
6-ply carbon 5 0 8 3 3 1 14 18 23 18 9 9.3
6-ply fiberglass 0 0 1 0 0 8 3 16 8 13 1 4.5
9-ply carbon 14 0 19 1 4 10 5 19 11 22 1 10.1
9-ply fiberglass 14 0 21 0 0 0 0 17 2 19 6 6.7
Total 39 0 50 4 7 19 22 83 57 108 19 37.1

MIA Backside Calls

MIA- MIA- MIA- MIA- MIA- MIA- MIA- MIA- MIA- MIA-
Panel Type Ave.
1 2 3 4 5 6 7 8 9 10
3-ply carbon 11 3 1 7 3 1 4 4 3 7 4.4
3-ply fiberglass 1 2 2 3 0 0 2 1 0 5 1.6
6-ply carbon 18 2 2 3 3 0 5 3 7 4 4.7
6-ply fiberglass 12 3 4 2 0 0 5 2 7 4 3.9
9-ply carbon 7 4 2 1 2 1 2 1 N/A 7 3.0
9-ply fiberglass 14 4 6 2 6 1 6 3 10 8 6.0
Total 63 18 17 18 14 3 24 14 27 35 23.3

WichiTech DTH Backside Calls

Panel Type DTH-1 DTH-2 DTH-3 DTH-4 DTH-5 DTH-6 DTH-7 DTH-8 DTH-9 Ave.
3-ply carbon 3 0 1 4 0 6 1 7 4 2.9
3-ply fiberglass 3 2 1 0 1 0 1 1 2 1.2
6-ply carbon 3 1 7 3 1 11 3 8 4 4.6
6-ply fiberglass 3 3 5 5 1 10 2 4 6 4.3
9-ply carbon 6 7 7 8 4 11 6 13 10 8.0
9-ply fiberglass 5 6 7 7 6 11 6 13 8 7.7
Total 23 19 28 27 13 49 19 46 34 28.7

254
 Table 17. Backside flaw detection for all inspectors and average backside flaws detected
per inspection method and panel type—conventional NDI (continued)

Woodpecker Backside Calls

WOOD- WOOD- WOOD- WOOD- WOOD- WOOD- WOOD- WOOD- WOOD-
Panel Type Ave.
1 2 3 4 5 6 7 8 9
3-ply carbon 0 0 1 0 3 1 1 1 1 0.9
3-ply fiberglass 0 0 0 1 0 0 1 1 0 0.3
6-ply carbon 1 0 1 4 2 4 1 2 2 1.9
6-ply fiberglass 3 0 2 2 3 6 1 1 0 2.0
9-ply carbon 1 1 2 1 5 1 1 3 2 1.9
9-ply fiberglass 3 1 4 7 9 2 4 9 6 5.0
Total 8 2 10 15 22 14 9 17 11 12.0

Positive indications (i.e., backside flaws correctly detected) in these regions reflect an NDI
penetration through the 1″ honeycomb core that can be useful in regions where far-side access is
not possible. The Airbus tap hammer, with its lower frequency excitation, and LFBT method
produced the highest number of backside flaw detections for the conventional NDI methods.
AC-UT, as it is a through-transmission method, and shearography produced the best backside
flaw detection among the advanced NDI methods.

Table 18. Summary of average backside flaw detection for conventional inspection
techniques listed by panel type

Average Number of Backside Calls Per Device

Airbus Boeing
WichiTech
Device/Panel Type Tap Tap LFBT MIA Woodpecker
DTH
Hammer Hammer
3-ply carbon 1.9 2.4 3.4 4.4 2.9 0.9
3-ply fiberglass 0.8 1.3 3.1 1.6 1.2 0.3
6-ply carbon 5.3 3.8 9.3 4.7 4.6 1.9
6-ply fiberglass 5.1 3.9 4.5 3.9 4.3 2.0
9-ply carbon 9.0 4.9 10.1 3.0 8.0 1.9
9-ply fiberglass 10.6 7.3 6.7 6.0 7.7 5.0
Total 32.6 23.7 37.1 23.3 28.7 12.0
Yellow highlights = lowest number of backside calls
Red highlights = highest number of backside calls

255
 Table 19. Backside flaw detection for advanced inspection techniques by panel type
(note: AC-UT was executed in through-transmission mode)

Advanced Inspection Techniques Backside Calls

Air- MAUS
Panel Type / MAUS
Coupled Shearography SAM Microwave CATT Resonance
Insp. Method MIA
UT 1
3-ply carbon 49 26 20 5 12 9 0
3-ply fiberglass 47 27 21 27 10 4 1
6-ply carbon 46 26 23 0 11 7 3
6-ply fiberglass 45 26 22 25 10 7 8
9-ply carbon 41 24 18 2 7 4 0
9-ply fiberglass 44 26 34 25 3 3 3
Total 272 155 138 84 53 34 15

Advanced Inspection Techniques Backside Calls

MAUS
Panel Type / FlawInspecta Thermography Thermography Lamino Thermo Laser
Resonance
Insp. Method UT Array (uncooled) (FLIR) graphy graphy UT
2
3-ply carbon 0 0 0 0 0 0 0
3-ply fiberglass 0 0 0 0 0 0 0
6-ply carbon 1 0 0 0 0 0 N/A
6-ply fiberglass 3 0 0 0 0 0 N/A
9-ply carbon 2 2 1 0 0 0 N/A
9-ply fiberglass 0 0 0 0 0 0 N/A
Total 6 2 1 0 0 0 0

Figures 290–295 are scatter diagrams showing the effect of inspection time on PoD and false call
performance for each composite honeycomb structure category. These are plots of each
inspector’s average time of inspection versus their resulting performance. For the PoD category,
the trends analyses showed that there was some improvement in PoD levels as the inspection
time per panel increased from 5 minutes to 15 minutes. However, increased inspection times
beyond 15 minutes did not provide much improvement in the PoD. Thus, the optimum inspection
rate for a 2.25 ft.2 area is approximately 15 minutes. For the false call category, the trends
analyses did not show any reduction in false call rates as the inspection times were increased.
This indicates that the false call rate was much more dependent on individual interpretations than
on the rate of making the interpretations.

256
 3.50 3.50

3.00
Airbus Tap Hamme r 3.00
Airbus Tap Hamme r

90% PoD Value

2.50 2.50

90% PoD Value

2.00 2.00

1.50 1.50

1.00 1.00

0.50 0.50

0.00 0.00
3.50 3.50

3.00
Boeing Tap Hamme r 3.00 Boeing Tap Hamme r
90% PoD Value

90% Po D Value
2.50
2.50
2.00
2.00
1.50
1.50
1.00
1.00
0.50
0.50
0.00
0.00
3.50 3.50 0 5 10 15 20 25 30 35

3.00
LFBT 3.00 LFBT
90% PoD Value

2.50

90% PoD Value

2.50

2.00 2.00

1.50 1.50

1.00 1.00

0.50
0.50
0.00
0.00
3.50 3.50

MIA 3.00
MIA
3.00
2.50
90% P o D V alu

90% PoD Value

2.50

2.00 2.00

1.50 1.50

1.00 1.00

0.50 0.50

0.00 0.00
3.50
3.50
Wichitech DTH 3.00
Wichitech DTH
3.00
90% P o D V alu

2.50
90% PoD Value

2.50

2.00 2.00

1.50 1.50

1.00 1.00

0.50 0.50
0.00 0.00
3 .5 0 3.50

3 .0 0 Woodpecker 3.00 Woodpecker

90% PoD Value
9 0 % P oD V a l

2 .5 0 2.50

2 .0 0 2.00

1 .5 0 1.50

1 .0 0 1.00

0 .5 0 0.50

0 .0 0 0.00
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35

Insp. Time (min.) Insp. Time (min.)

3 Carbon
3-Ply Carbon 33-Ply
Ply Fiberglass
Fiberglass
Figure 290. Scatter diagram showing effect of inspection time
on PoD values for 3-ply panels

257
 3.50 3.50

3.00
Airbus Tap Hamme r 3.00
Airbus Tap Hamme r

90% PoD Value

90% PoD Value

2.50 2.50

2.00 2.00

1.50 1.50

1.00 1.00

0.50 0.50
0.00 0.00
3.50 3.50 0 5

3.00
Boeing Tap Hamme r 3.00
Boeing Tap Hamme r
90% PoD Value

90% PoD Value

2.50 2.50

2.00 2.00

1.50 1.50

1.00 1.00

0.50 0.50

0.00 0.00
3.50 0 5 10 15 20 25 30 35 3.50 0 5 10 15 20 25
3.00 LFBT 3.00 LFBT
90% PoD Value

90% PoD Value

2.50 2.50

2.00 2.00

1.50 1.50

1.00 1.00

0.50 0.50

0.00 0.00
3.50 0 5 10 15 20 25 3.50
3.00
MIA 3.00 MIA
90% PoD Value

2.50 90% PoD Value 2.50

2.00 2.00
1.50 1.50
1.00 1.00
0.50 0.50
0.00 0.00
3.50 3.50

3.00 Wichitech DTH 3.00 Wichitech DTH

90% PoD Value
90% PoD Value

2.50 2.50

2.00 2.00

1.50 1.50

1.00 1.00

0.50 0.50

0.00 0.00
3.50 3.50
3.00
Woodpecker Woodpecker
3.00
90% PoD Value

2.50
90% PoD Value

2.50
2.00 2.00
1.50 1.50
1.00 1.00
0.50 0.50
0.00 0.00
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
Insp. Time (min.) Insp. Time (min.)
66-Ply
Ply Carbon
Carbon 66-Ply
Ply Fiberglass
Fiberglass

Figure 291. Scatter diagram showing effect of inspection time

on PoD values for 6-ply panels

258
 3.50 3.50

3.00
Airbus Tap Hamme r 3.00
Airbus Tap Hamme r
90% PoD Value

90% PoD Value

2.50 2.50

2.00 2.00

1.50 1.50

1.00 1.00

0.50 0.50

0.00 0.00
3.50 3.50 0 5 10 15 20 25 30 35

3.00
Boeing Tap Hamme r 3.00
Boeing Tap Hamme r
90% PoD Value

2.50 2.50

90% PoD Value

2.00 2.00
1.50 1.50
1.00 1.00
0.50 0.50
0.00 0.00
3.50 0 5 10 15 20 25 30 35 3.50
3.00
LFBT 3.00
LFBT
90% PoD Value

2.50 2.50

90% PoD Value

2.00 2.00
1.50 1.50
1.00 1.00
0.50 0.50
0.00 0.00
3.50 0 5 10 15 20 25 30 35 3.50
3.00 MIA 3.00
MIA
90% PoD Value

2.50
90% PoD Value

2.50
2.00 2.00 42 min.
1.50 1.50
1.00 1.00
0.50 0.50
0.00 0.00
3.50
3.50
3.00 Wichitech DTH Wichitech DTH
3.00
90% PoD Value

2.50
90% PoD Value

2.50
2.00
2.00
1.50
1.50
1.00
1.00
0.50
0.50
0.00
0.00
3.50
Woodpecker 3.50
3.00
3.00
Woodpecker
90% PoD Value

2.50
90% PoD Value

2.50
2.00
2.00
1.50
1.50
1.00
1.00
0.50
0.50
0.00
0.00
0 5 10 15 20 25 30 35
0 5 10 15 20 25 30 35
Insp. Time (min.) Insp. Time (min.)
99-Ply
Ply Carbon
Carbon 9 9-Ply
Ply Fiberglass
Fiberglass

Figure 292. Scatter diagram showing effect of inspection time

on PoD values for 9-ply panels

259
 30 30

25
Airbus Tap Hamme r 25
Airbus Tap Hamme r

Number of False Calls

Number of False Calls 20 20

15 15

10 10

5 5

0 0
30 30 25 30 35

25
Boeing Tap Hamme r 25
Boeing Tap Hamme r

Number of False Calls

Number of False Calls

20 20

15 15

10 10

5 5

0 0
30 30 0 5 10 15 20 25 30 35

25
LFBT 25
LFBT

Number of False Calls

Number of False Calls

20 20

15 15

10 10

5 5

0 0
30 30 0 5 10 15 20 25 30 35

25
MIA 25
MIA
Number of False Calls

Number of False Calls

20 20

15 15

10 10

5 5

0 0
30 30 0 5 10 15 20 25 30 35

25
Wichitech DTH 25
Wichitech DTH
Number of False Calls

Number of False Calls

20 20

15 15

10 10

5 5

0 0
30 30 0 5 10 15 20 25 30 35

25
Woodpecker Woodpecker
Number of False Calls

25
Number of False Calls

20 20

15 15
10 10

5 5

0 0
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
Insp. Time (min.) Insp. Time (min.)
33-Ply
Ply Carbon
Carbon 3 PlyFiberglass
3-Ply Fiberglass

Figure 293. Scatter diagram showing effect of inspection time

on false calls for 3-ply panels

260
 30 30
Airbus Tap Hamme r 25
Airbus Tap Hamme r
25

Number of False Calls

Number of False Calls 20 20

15 15

10 10

5 5
(2)
0 0
30 30 0 10 1 20 2 30 3

25
Boeing Tap Hamme r 25
Boeing Tap Hamme r
Number of False Calls

Number of False Calls

20 20

15 15

10 10

5 5

0 0
30 0 10 1 20 2 30 3 30 0 10 1 20 2 30 3
34 33
25
LFBT 25
LFBT
Number of False Calls

Number of False Calls

20 20

15 15

10 10

5 5

0 0
30 0 5 10 15 20 25 30 35 30 0 5 10 15 20 25 30 35

25
MIA 25
MIA
Number of False Calls
Number of False Calls

20 20

15 15

10 10

5 5

0 0
30 30 0 10 1 20 2 30 3

25
Wichitech DTH 25
Wichitech DTH
Number of False Calls

Number of False Calls

20 20

15 15

10 10

5 5

0 0
30 30

25
Woodpecker 25
Woodpecker
Number of False Calls

Number of False Calls

20 20

15 15

10 10

5 5

0 0
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35

Insp. Time (min.) Insp. Time (min.)

66-Ply
Ply Carbon
Carbon 66-Ply
Ply Fiberglass
Fiberglass

Figure 294. Scatter diagram showing effect of inspection time

on false calls for 6-ply panels

261
 30 30
Airbus Tap Hamme r Airbus Tap Hamme r
25 25

Number of False Calls

Number of False Calls 20 20

15 15

10 10

5 5

0 0
30 30

25
Boeing Tap Hamme r 25
Boeing Tap Hamme r

Number of False Calls

Number of False Calls

20 20

15 15

10 10

5 5

0 0
30 30
33
25
LFBT 25
LFBT

Number of False Calls

Number of False Calls

20 20

15 15

10 10

5 5

0 0
30 30

25
MIA 25
MIA 36 42
Number of False Calls

Number of False Calls

20 20

15 15

10 10

5 5

0 0
30 30
31 32
25
Number of False Calls

25
Number of False Calls

20
Wichitech DTH Wichitech DTH
20

15 15

10 10

5 5

0 0
30
30
25
Woodpecker Woodpecker
Number of False Calls

25
Number of False Calls

20
20
15
15
10
10
5
5
0
0
0 5 10 15 20 25 30 35
0 5 10 15 20 25 30 35
Insp. Time (min.) Insp. Time (min.)
99-Ply
Ply Carbon
Carbon 99-Ply
Ply Fiberglass
Fiberglass

Figure 295. Scatter diagram showing effect of inspection time

on false calls for 9-ply panels

262
The scatter diagrams in figures 296–298 show that there is no relation between the number of
false calls and the resulting PoD values for each composite honeycomb structure category. This
is primarily because the inspectors were not trying to improve PoD performance by marking a
large number of areas; rather, they were trying to earnestly discern flaw regions from pristine
regions.

3-Ply Carbon 3-Ply Fiberglass

Figure 296. Scatter diagram showing the effect of the number of false calls
on PoD values for 3-ply panels

263
 6-Ply Carbon 6-Ply Fiberglass

Figure 297. Scatter diagram showing the effect of the number of false calls
on PoD values for 6-ply panels

264
 9-Ply Carbon 9-Ply Fiberglass

Figure 298. Scatter diagram showing the effect of the number of false calls
on PoD values for 9-ply panels

265
Tables 20–25 summarize the overall performance results including flaw sizing capability, 90%
PoD values, and false call rates for each NDI method in this study and broken down by the
various composite honeycomb structure categories. These tables provide the best single view of
the capability of conventional and advanced NDI methods to detect flaws/damage in composite
honeycomb structures.

Table 20. Summary of all NDI performance for 3-ply carbon and fiberglass test
specimens—flaw sizing, overall 90% PoD levels, and false calls

Flaw Sizing and False Call Summary Table for 3-Ply Carbon
Flaw Coverage
99%‒ 74%‒ 49%‒ 90% POD False
Inspection Device 100% <25%
75% 50% 25% Level Calls
Airbus tap hammer 42% 40% 15% 3% 1% 2.34 0.8
Boeing tap hammer 38% 42% 15% 5% 1% 2.03 1.0
LFBT 34% 26% 21% 13% 6% 2.10 2.3
MIA 34% 37% 15% 11% 3% 1.46 0.6
WichiTech DTH 46% 38% 12% 4% 1% 2.28 0.6
Woodpecker 38% 41% 12% 7% 1% 1.78 0.0
CATT 12% 53% 14% 16% 5% 1.19 0.0
MAUS Resonance 1 68% 19% 13% 0% 0% ≤.50 3.0
SAM 26% 38% 21% 10% 5% 1.26 2.0
Shearography 74% 18% 8% 0% 0% ≤.50 0.0
Thermography 84% 12% 4% 0% 0% ≤.50 2.0
MAUS Resonance 2 56% 38% 6% 0% 0% 0.53 2.0
MAUS MIA 91% 4% 4% 0% 0% 0.77 0.0
Laminography 92% 8% 0% 0% 0% 0.66 0.0
Microwave 90% 0% 0% 3% 7% 2.13 9.0
Thermography (uncooled) 92% 6% 0% 2% 0% 0.52 0.0
Thermography (FLIR) N/A N/A N/A N/A N/A ≤.50 0.0
FlawInspecta UT Array 79% 16% 5% 0% 0% 1.24 3.0
Air-coupled UT 4% 59% 20% 16% 0% 0.52 0.0
Laser UT 93% 7% 0% 0% 0% 0.84 4.0
(Highlighted = best performers)

266
 Table 20. Summary of all NDI performance for 3-ply fiberglass test specimens—flaw
sizing, overall 90% PoD levels, and false calls (continued)

Flaw Sizing and False Call Summary Table for 3-Ply Fiberglass
Flaw Coverage
99%‒ 74%‒ 49%‒ 90% POD False
Inspection Device 100% <25%
75% 50% 25% Level Calls
Airbus tap hammer 32% 40% 21% 5% 2% 2.03 0.3
Boeing tap hammer 40% 35% 17% 7% 2% 1.57 1.8
LFBT 32% 29% 21% 15% 3% 1.77 0.9
MIA 34% 37% 19% 8% 1% 1.04 0.5
WichiTech DTH 35% 44% 16% 4% 1% 1.81 0.9
Woodpecker 42% 39% 12% 5% 2% 1.45 0.3
CATT 28% 58% 12% 2% 0% 1.04 4.0
MAUS Resonance 1 61% 22% 17% 0% 0% 0.69 5.0
SAM 21% 38% 25% 15% 2% 0.65 0.0
Shearography 59% 31% 8% 2% 0% 0.56 1.0
Thermography 75% 20% 6% 0% 0% ≤.50 0.0
MAUS Resonance 2 58% 37% 5% 0% 0% 1.18 6.0
MAUS MIA 89% 6% 4% 0% 0% 0.69 0.0
Laminography 93% 5% 2% 0% 0% 0.97 0.0
Microwave 92% 6% 0% 2% 0% ≤.50 2.0
Thermography (uncooled) 86% 10% 2% 2% 0% 0.52 0.0
Thermography (FLIR) N/A N/A N/A N/A N/A ≤.50 0.0
FlawInspecta UT Array 67% 25% 6% 3% 0% 1.78 0.0
Air-coupled UT 4% 36% 45% 15% 0% 0.69 0.0
Laser UT 82% 18% 0% 0% 0% 1.25 0.0
(Highlighted = best performers)

267
 Table 21. Summary of all NDI performance for 6-ply carbon and fiberglass test
specimens—flaw sizing, overall 90% PoD levels, and false calls

Flaw Sizing and False Call Summary Table for 6-Ply Carbon
Flaw Coverage
99%‒ 74%‒ 49%‒ 90% POD False
Inspection Device 100% <25%
75% 50% 25% Level Calls
Airbus tap hammer 18% 26% 23% 24% 8% >3.00 3.5
Boeing tap hammer 20% 33% 27% 16% 4% 2.59 4.7
LFBT 28% 24% 23% 19% 6% 2.42 6.5
MIA 18% 33% 26% 21% 2% 1.76 1.3
WichiTech DTH 34% 36% 19% 9% 2% 2.18 0.9
Woodpecker 31% 31% 20% 15% 4% 2.32 0.7
CATT 29% 47% 18% 5% 0% 1.85 1.0
MAUS Resonance 1 48% 34% 10% 8% 0% ≤.50 12.0
SAM 3% 49% 31% 10% 8% 1.63 37.0
Shearography 62% 18% 8% 10% 2% 1.00 0.0
Thermography 82% 7% 4% 7% 0% ≤.50 2.0
MAUS Resonance 2 49% 27% 22% 3% 0% 2.27 4.0
MAUS MIA 90% 6% 4% 0% 0% 1.32 6.0
Laminography 90% 2% 0% 0% 8% 1.10 0.0
Microwave 43% 19% 0% 24% 14% >3.00 10.0
Thermography (uncooled) 74% 11% 2% 7% 7% 1.70 2.0
Thermography (FLIR) N/A N/A N/A N/A N/A ≤.50 1.0
FlawInspecta UT Array 64% 12% 2% 19% 19% 1.92 0.0
Air-coupled UT 23% 49% 17% 11% 11% 0.87 0.0
Laser UT N/A N/A N/A N/A N/A N/A N/A
(Highlighted = best performers)

268
 Table 21. Summary of all NDI performance for 6-ply carbon and fiberglass test
specimens—flaw sizing, overall 90% PoD levels, and false calls (continued)

Flaw Sizing and False Call Summary Table for 6-Ply Fiberglass
Flaw Coverage
99%‒ 74%‒ 49%‒ 90% POD False
Inspection Device 100% <25%
75% 50% 25% Level Calls
Airbus tap hammer 28% 30% 27% 13% 2% 2.44 2.9
Boeing tap hammer 21% 34% 25% 16% 4% 2.33 4.7
LFBT 28% 29% 20% 18% 5% 2.55 3.3
MIA 26% 26% 26% 18% 4% 1.49 1.9
WichiTech DTH 32% 39% 19% 8% 2% 1.71 1.6
Woodpecker 31% 28% 20% 14% 7% 2.05 0.1
CATT 28% 38% 19% 13% 24% 1.10 1.0
MAUS Resonance 1 47% 31% 4% 4% 14% 0.55 9.0
SAM 11% 40% 32% 9% 8% 0.84 8.0
Shearography 49% 27% 15% 9% 0% ≤.50 0.0
Thermography 75% 15% 5% 5% 0% 0.70 3.0
MAUS Resonance 2 39% 33% 18% 9% 0% 2.07 2.0
MAUS MIA 84% 8% 8% 0% 0% 1.48 6.0
Laminography 86% 6% 0% 2% 6% 1.43 2.0
Microwave 75% 17% 2% 0% 6% 0.93 12.0
Thermography (uncooled) 55% 24% 2% 2% 17% 2.19 2.0
Thermography (FLIR) N/A N/A N/A N/A N/A ≤.50 4.0
FlawInspecta UT Array 31% 13% 25% 6% 25% >3.00 0.0
Air-coupled UT 21% 43% 34% 2% 0% 0.90 0.0
Laser UT N/A N/A N/A N/A N/A N/A N/A
(Highlighted = best performers)

269
 Table 22. Summary of all NDI performance for 9-ply carbon and fiberglass test
specimens—flaw sizing, overall 90% PoD levels, and false calls

Flaw Sizing and False Call Summary Table for 9-Ply Carbon
Flaw Coverage
99%‒ 74%‒ 49%‒ 90% POD False
Inspection Device 100% <25%
75% 50% 25% Level Calls
Airbus tap hammer 35% 25% 17% 18% 5% >3.00 6.8
Boeing tap hammer 19% 20% 27% 28% 6% 2.88 5.4
LFBT 32% 23% 17% 20% 8% 2.61 3.8
MIA 14% 22% 25% 30% 9% 2.97 3.1
WichiTech DTH 41% 30% 19% 6% 3% 2.15 4.4
Woodpecker 22% 27% 26% 19% 6% 2.76 2.8
CATT 12% 31% 38% 19% 0% 2.36 4.0
MAUS Resonance 1 44% 28% 5% 19% 5% 1.32 0.0
SAM 0% 41% 22% 11% 26% >3.00 74.0
Shearography 53% 21% 16% 7% 2% 2.00 0.0
Thermography 86% 9% 0% 5% 0% 0.64 0.0
MAUS Resonance 2 50% 31% 15% 4% 0% >3.00 10.0
MAUS MIA 74% 19% 6% 0% 0% >3.00 8.0
Laminography 90% 2% 2% 0% 6% 1.39 0.0
Microwave 24% 20% 8% 16% 32% >3.00 23.0
Thermography (uncooled) 71% 0% 6% 6% 16% >3.00 0.0
Thermography (FLIR) N/A N/A N/A N/A N/A ≤.50 0.0
FlawInspecta UT Array 71% 9% 6% 9% 6% 2.85 1.0
Air-coupled UT 17% 55% 19% 4% 4% 1.26 2.0
Laser UT N/A N/A N/A N/A N/A N/A N/A
(Highlighted = best performer)

270
 Table 22. Summary of all NDI performance for 9-ply carbon and fiberglass test
specimens—flaw sizing, overall 90% PoD levels, and false calls (continued)

Flaw Sizing and False Call Summary Table for 9-Ply Fiberglass
Flaw Coverage
99%‒ 74%‒ 49%‒ 90% POD False
Inspection Device 100% <25%
75% 50% 25% Level Calls
Airbus tap hammer 33% 27% 23% 13% 5% >3.00 7.1
Boeing tap hammer 18% 30% 24% 21% 6% 2.90 6.9
LFBT 25% 26% 24% 20% 5% 3.00 2.8
MIA 14% 22% 30% 23% 10% 2.00 6.8
WichiTech DTH 43% 32% 18% 6% 1% 2.07 4.9
Woodpecker 35% 28% 16% 11% 11% 2.52 6.7
CATT 13% 23% 30% 33% 0% 2.27 0.0
MAUS Resonance 1 45% 33% 10% 2% 10% 0.60 2.0
SAM 0% 75% 23% 2% 0% 1.04 17.0
Shearography 65% 21% 12% 2% 0% 0.77 0.0
Thermography 80% 5% 7% 7% 0% 0.67 2.0
MAUS Resonance 2 39% 29% 11% 21% 0% >3.00 11.0
MAUS MIA 72% 8% 12% 8% 0% >3.00 8.0
Laminography 92% 6% 0% 0% 2% 1.69 1.0
Microwave 87% 13% 0% 0% 0% 2.00 8.0
Thermography (uncooled) 60% 4% 4% 4% 28% >3.00 4.0
Thermography (FLIR) N/A N/A N/A N/A N/A ≤.50 1.0
FlawInspecta UT Array 58% 0% 3% 16% 26% >3.00 1.0
Air-coupled UT 12% 47% 31% 10% 0% 1.01 1.0
Laser UT N/A N/A N/A N/A N/A N/A N/A
(Highlighted = best performer)

271
Table 23. Summary of all NDI flaw-detection capability for 3-ply carbon and fiberglass test
specimens—overall flaw detection, highest detection (by individual inspector), lowest
detection (by individual inspector), overall 90% PoD levels, and false calls

Flaw Detection Percentage Summary Table for 3-Ply Carbon

Overall Highest Lowest
90% POD False
Inspection Device Detection Detection Detection
Level Calls
Percentage Percentage Percentage
Airbus tap hammer 52% 76% 25% 2.34 0.8
Boeing tap hammer 64% 78% 47% 2.03 1.0
LFBT 63% 82% 35% 2.10 2.3
MIA 79% 86% 73% 1.46 0.6
WichiTech DTH 61% 78% 47% 2.28 0.6
Woodpecker 65% 75% 51% 1.78 0.0
CATT 84% N/A N/A 1.19 0.0
MAUS Resonance 1 100% N/A N/A ≤.50 3.0
SAM 82% N/A N/A 1.26 2.0
Shearography 98% N/A N/A ≤.50 0.0
Thermography 96% N/A N/A ≤.50 2.0
MAUS Resonance 2 94% N/A N/A 0.53 2.0
MAUS MIA 90% N/A N/A 0.77 0.0
Laminography 94% N/A N/A 0.66 0.0
Microwave 59% N/A N/A 2.13 9.0
Thermography (uncooled) 96% N/A N/A 0.52 0.0
Thermography (FLIR) 98% N/A N/A ≤.50 0.0
FlawInspecta UT Array 75% N/A N/A 1.24 3.0
Air-coupled UT 96% N/A N/A 0.52 0.0
Laser UT 90% N/A N/A 0.84 0.0

272
Table 23. Summary of all NDI flaw-detection capability for 3-ply carbon and fiberglass test
specimens—overall flaw detection, highest detection (by individual inspector), lowest
detection (by individual inspector), overall 90% PoD levels, and false calls (continued)

Flaw Detection Percentage Summary Table for 3-Ply Fiberglass

Overall Highest Lowest
90% POD False
Inspection Device Detection Detection Detection
Level Calls
Percentage Percentage Percentage
Airbus tap hammer 60% 75% 41% 2.03 0.3
Boeing tap hammer 69% 86% 45% 1.57 1.8
LFBT 71% 92% 51% 1.77 0.9
MIA 84% 90% 79% 1.04 0.5
WichiTech DTH 66% 82% 47% 1.81 0.9
Woodpecker 70% 86% 55% 1.45 0.3
CATT 84% N/A N/A 1.04 4.0
MAUS Resonance 1 92% N/A N/A 0.69 5.0
SAM 94% N/A N/A 0.65 0.0
Shearography 96% N/A N/A 0.56 1.0
Thermography 100% N/A N/A ≤.50 0.0
MAUS Resonance 2 84% N/A N/A 1.18 6.0
MAUS MIA 92% N/A N/A 0.69 0.0
Laminography 88% N/A N/A 0.97 0.0
Microwave 98% N/A N/A ≤.50 2.0
Thermography (uncooled) 96% N/A N/A 0.52 0.0
Thermography (FLIR) 98% N/A N/A ≤.50 0.0
FlawInspecta UT Array 71% N/A N/A 1.78 0.0
Air-coupled UT 92% N/A N/A 0.69 0.0
Laser UT 86% N/A N/A 1.25 0.0

273
Table 24. Summary of all NDI flaw-detection capability for 6-ply carbon and fiberglass test
specimens—overall flaw detection, highest detection (by individual inspector), lowest
detection (by individual inspector), overall 90% PoD levels, and false calls

Flaw Detection Percentage Summary Table for 6-Ply Carbon

Overall Highest Lowest
90% POD False
Inspection Device Detection Detection Detection
Level Calls
Percentage Percentage Percentage
Airbus tap hammer 41% 58% 24% >3.00 3.5
Boeing tap hammer 43% 69% 31% 2.59 4.7
LFBT 57% 80% 46% 2.42 6.5
MIA 63% 71% 59% 1.76 1.3
WichiTech DTH 56% 73% 44% 2.18 0.9
Woodpecker 44% 53% 39% 2.32 0.7
CATT 64% N/A N/A 1.85 1.0
MAUS Resonance 1 98% N/A N/A ≤.50 12.0
SAM 66% N/A N/A 1.63 37.0
Shearography 85% N/A N/A 1.00 0.0
Thermography 97% N/A N/A ≤.50 2.0
MAUS Resonance 2 63% N/A N/A 2.27 4.0
MAUS MIA 81% N/A N/A 1.32 6.0
Laminography 88% N/A N/A 1.10 0.0
Microwave 36% N/A N/A >3.00 10.0
Thermography (uncooled) 78% N/A N/A 1.70 2.0
Thermography (FLIR) 100% N/A N/A ≤.50 1.0
FlawInspecta UT Array 71% N/A N/A 1.92 0.0
Air-coupled UT 90% N/A N/A 0.87 0.0
Laser UT N/A N/A N/A N/A N/A

274
Table 24. Summary of all NDI flaw-detection capability for 6-ply carbon and fiberglass test
specimens—overall flaw detection, highest detection (by individual inspector), lowest
detection (by individual inspector), overall 90% PoD levels, and false calls (continued)

Flaw Detection Percentage Summary Table for 6-Ply Fiberglass

Overall Highest Lowest
90% POD False
Inspection Device Detection Detection Detection
Level Calls
Percentage Percentage Percentage
Airbus tap hammer 47% 66% 20% 2.44 2.9
Boeing tap hammer 50% 69% 34% 2.33 4.7
LFBT 59% 85% 32% 2.55 3.3
MIA 72% 81% 56% 1.49 1.9
WichiTech DTH 61% 81% 42% 1.71 1.6
Woodpecker 50% 64% 34% 2.05 0.1
CATT 81% N/A N/A 1.10 1.0
MAUS Resonance 1 97% N/A N/A 0.55 9.0
SAM 90% N/A N/A 0.84 8.0
Shearography 100% N/A N/A ≤.50 0.0
Thermography 93% N/A N/A 0.70 3.0
MAUS Resonance 2 56% N/A N/A 2.07 2.0
MAUS MIA 83% N/A N/A 1.48 6.0
Laminography 80% N/A N/A 1.43 2.0
Microwave 88% N/A N/A 0.93 12.0
Thermography (uncooled) 71% N/A N/A 2.19 2.0
Thermography (FLIR) 98% N/A N/A ≤.50 4.0
FlawInspecta UT Array 27% N/A N/A >3.00 0.0
Air-coupled UT 90% N/A N/A 0.90 0.0
Laser UT N/A N/A N/A N/A N/A

275
Table 25. Summary of all NDI flaw-detection capability for 9-ply carbon and fiberglass test
specimens—overall flaw detection, highest detection (by individual inspector), lowest
detection (by individual inspector), overall 90%PoD levels, and false calls

Flaw Detection Percentage Summary Table for 9-Ply Carbon

Overall Highest Lowest
90% POD False
Inspection Device Detection Detection Detection
Level Calls
Percentage Percentage Percentage
Airbus tap hammer 40% 53% 17% >3.00 6.8
Boeing tap hammer 38% 59% 24% 2.88 5.4
LFBT 51% 66% 34% 2.61 3.8
MIA 39% 54% 27% 2.97 3.1
WichiTech DTH 50% 64% 41% 2.15 4.4
Woodpecker 35% 44% 29% 2.76 2.8
CATT 44% N/A N/A 2.36 4.0
MAUS Resonance 1 80% N/A N/A 1.32 0.0
SAM 46% N/A N/A >3.00 74.0
Shearography 73% N/A N/A 2.00 0.0
Thermography 95% N/A N/A 0.64 0.0
MAUS Resonance 2 44% N/A N/A >3.00 10.0
MAUS MIA 53% N/A N/A >3.00 8.0
Laminography 83% N/A N/A 1.39 0.0
Microwave 42% N/A N/A >3.00 23.0
Thermography (uncooled) 53% N/A N/A >3.00 0.0
Thermography (FLIR) 97% N/A N/A ≤.50 0.0
FlawInspecta UT Array 59% N/A N/A 2.85 1.0
Air-coupled UT 80% N/A N/A 1.26 2.0
Laser UT N/A N/A N/A N/A N/A

276
Table 25. Summary of all NDI flaw-detection capability for 9-ply carbon and fiberglass test
specimens—overall flaw detection, highest detection (by individual inspector), lowest
detection (by individual inspector), overall 90%PoD levels, and false calls (continued)

Flaw Detection Percentage Summary Table for 9-Ply Fiberglass

Overall Highest Lowest
90% POD False
Inspection Device Detection Detection Detection
Level Calls
Percentage Percentage Percentage
Airbus tap hammer 44% 73% 7% >3.00 7.1
Boeing tap hammer 41% 69% 25% 2.90 6.9
LFBT 49% 71% 24% 3.00 2.8
MIA 52% 64% 36% 2.00 6.8
WichiTech DTH 55% 68% 42% 2.07 4.9
Woodpecker 41% 54% 27% 2.52 6.7
CATT 51% N/A N/A 2.27 0.0
MAUS Resonance 1 86% N/A N/A 0.60 2.0
SAM 81% N/A N/A 1.04 17.0
Shearography 88% N/A N/A 0.77 0.0
Thermography 93% N/A N/A 0.67 2.0
MAUS Resonance 2 47% N/A N/A >3.00 11.0
MAUS MIA 42% N/A N/A >3.00 8.0
Laminography 83% N/A N/A 1.69 1.0
Microwave 80% N/A N/A 2.00 8.0
Thermography (uncooled) 42% N/A N/A >3.00 4.0
Thermography (FLIR) 97% N/A N/A ≤.50 1.0
FlawInspecta UT Array 32% N/A N/A >3.00 1.0
Air-coupled UT 86% N/A N/A 1.01 1.0
Laser UT N/A N/A N/A N/A N/A

277
8. CONCLUSIONS AND RECOMMENDATIONS

The rapidly increasing use of composites on commercial airplanes, coupled with the potential for
economic savings associated with their use, means that the demand for composite materials
technology will continue to increase. Inspecting these composite structures is a critical element
in assuring their continued airworthiness. Extensive damage tolerance analysis associated with
the use of composites in primary structures will likely identify areas requiring more sensitive
health monitoring. Many composite honeycomb structure inspections are performed by tap test
methods that use a human-detected change in acoustic response to locate flaws. More
sophisticated nondestructive inspection (NDI) methods could be applied to improve the damage
detection in adhesively bonded composite aircraft parts. The Composite Honeycomb Flaw
Detection Experiment (CHE) was developed to assess the performance of both conventional and
advanced NDI techniques in detecting voids, heat damage, disbonds, delaminations, and impact
damage in adhesively bonded composite aircraft parts.

The conventional composite inspection techniques studied were: Boeing and Airbus manual tap
hammers, low frequency bond testing, high frequency bond testing (resonance), Mitsui
Woodpecker automated tap hammer, WichiTech digital tap hammer (DTH), and mechanical
impedance analysis (MIA). In addition, 14 different advanced NDI methods were evaluated in
these experiments including: computer assisted tap hammer (CATT), thermography (pulsed and
induction), pulse-echo ultrasonics (PE-UT), air-coupled UT (AC-UT), Mobile Automated
Scanner (MAUS) C-scanning in MIA and resonance mode, laser UT, shearography, microwave,
structural anomaly mapping (SAM), UT linear array, digital acoustic video, laminography, and
terahertz imaging.

The CHE traveled to airlines, third-party maintenance depots, and aircraft manufacturers to
acquire flaw detection data. The experiment was deployed in a hangar to provide a representative
inspection environment including impediments such as poor lighting and noise distractions. A
total of 75 inspectors from 22 airlines and maintenance and repair organizations (MRO) located
around the world participated in the experiment. Inspector feedback on performance provided
excellent training for the experimenter while their results produced a valuable baseline of how
well the industry is able to inspect composite structures (e.g., flaw hits/misses, false calls, flaw
sizing, effects of construction scenarios, and effects of environment). These blind tests produced
statistically valid probability of detection (PoD) curves representative of the industry as a whole.
Furthermore, results from the 18 advanced NDI methods allowed the team to quantify the degree
of inspection improvements possible via the application of more sophisticated inspection
methods and procedures. The experiment results also allowed the team to determine which NDI
methods possess unique capabilities to address specific inspection requirements.

Inspectors conducted tests in accordance with Boeing and/or Airbus maintenance guidelines but
were otherwise free to apply nondestructive test (NDT) techniques and interpret the data
according to their own standards. Using 90% PoD as a benchmark, the study generally found that
the more sophisticated the technique, the smaller the flaw that could be detected. Yet all the
techniques achieved 90% PoD for flaws 3″ in size. This level of reliability may be sufficient for
many secondary structures. With automated tap testers, the 90% PoD flaw size decreased to
approximately 1.5″. The largest factors affecting PoD levels were determined to be the thickness
of the skin and the material type (e.g., fiberglass or carbon skins).
278
A summary of the overall experiment results, observations, and recommendations associated
with these results follows.

• This program assessed current industry capabilities by quantifying flaw detection

performance in composite laminate structures. Overall, the results from the CHE
produced a capability baseline for current NDI techniques and quantified improvements
stemming from advanced NDI.
• This experiment provides overall PoD values for inspecting composite honeycomb
structures so that the aviation industry can: 1) better understand what type of damage
detection is possible for specific inspection scenarios; 2) adjust inspection procedures to
optimize performance; and 3) smartly enhance inspector preparation and training to
generate the performance improvements possible via optimized NDI deployment,
sufficient knowledge of the inspection idiosyncrasies, and increased exposure to realistic,
composite inspection demands.
• The viability of certain NDI methods, selected to meet specific application demands, and
the quantification of performance, must be continually pursued. Toward that end, this
CHE is available for continued testing. All future testing will have the results from this
assessment to serve as the basis of comparison and help quantify NDI improvements.

‒ While the size of flaw, or damage, that must be detected is affected by many
parameters (e.g., structure type, location on aircraft, and stress and fatigue levels),
the general goal for composite honeycomb inspections is to detect flaws that are
1″ diameter or larger. Many of the NDI reference standards in original equipment
manufacturer (OEM) NDT manuals use 1″ diameter flaws to guide equipment
setup. In addition, the Commercial Aircraft Composite Repair Committee
Inspection Task Group members generally concur that a 1″ flaw detection
provides a good center point for the CHE. Thus, the flaw sizes in the design were
established with a 1″ diameter at the center. Larger and smaller flaws were
included such that PoD values smaller than 1″ (as small as 0.25″) and PoD values
larger than 1″ (as large as 2″) could be ascertained.

• Of the conventional NDI methods for composite honeycomb, the top performers and 90%
PoD levels for each category are listed below:

‒ 3-ply fiberglass – MIA (PoD90 = 1.0″ dia.)

‒ 3-ply carbon – MIA (PoD90 = 1.5″ dia.)
‒ 6-ply fiberglass – MIA (PoD90 = 1.5″ dia.)
‒ 6-ply carbon – MIA (PoD90 = 1.8″ dia.)
‒ 9-ply fiberglass – MIA (PoD90 = 2.0″ dia.)
‒ 9-ply carbon – WichiTech DTH (PoD90 = 2.2″ dia.)

• For the manually deployed tap hammers only, the top performance level that can be
expected for each category are listed below. It should be noted that the transition from
manually deployed tapping to automated, with C-scan imaging via the CATT system,
produced improvements by 20–50% over the values listed below:

279
 ‒ 3-ply fiberglass – Boeing tap hammer (PoD90 = 1.6″ dia.)
‒ 3-ply carbon – Boeing tap hammer (PoD90 = 2.0″ dia.)
‒ 6-ply fiberglass – Boeing tap hammer (PoD90 = 2.3″ dia.)
‒ 6-ply carbon – Boeing tap hammer (PoD90 = 2.6″ dia.)
‒ 9-ply fiberglass – Boeing tap hammer (PoD90 = 2.9″ dia.)
‒ 9-ply carbon – Boeing tap hammer (PoD90 < 2.9″ dia.)

• Of the advanced NDI methods, the top performers and 90% PoD levels for each category
are listed below. In general, the level of improvement over conventional NDI methods
becomes higher as the inspection challenge increases (i.e., skin becomes thicker and
moves from fiberglass to carbon). The 90% PoD improvements range from 50–75% over
conventional NDI methods.

‒ 3-ply fiberglass – Thermography and microwave (PoD90 < 0.5″ dia.)

‒ 3-ply carbon – Thermography, MAUS Resonance, Shearography, and AC-UT
(PoD90 < 0.5″ dia.)
‒ 6-ply fiberglass – Thermography, MAUS Resonance, and Shearography UT
(PoD90 < 0.5″ dia.)
‒ 6-ply carbon – Thermography and MAUS Resonance (PoD90 < 0.5″ dia.)
‒ 9-ply fiberglass – Thermography (PoD90 < 0.5″ dia.)
‒ 9-ply carbon – Thermography (PoD90 < 0.5″ dia.)

• It should be noted that the advanced NDI techniques that were evaluated are in different
states of maturity. Overall, the improvement in flaw detection ranged from 66–72% over
the conventional methods. Automated deployment and data presentation/analysis reduces
many human factors concerns such as ensuring surface area coverage and interpretation
of inspection results (i.e., 100% coverage and flaw recognition on C-scan images).
• Advanced NDI generally produced slower inspections for wide areas. Inspection times
typically exceeded the times achieved by the conventional methods with the exception
being thermography and shearography. This is mostly attributed to the increase in initial
setup time plus the data analysis associated with the more sophisticated inspection
systems. Some methods required multiple shots/images to get full coverage of an
inspection area. Special care is needed to ensure that there is sufficient overlap between
adjacent images. Laminography’s performance would probably improve if there were a
larger overlap of inspection images to ensure complete coverage of the panel.
• Most stiffness-based inspection methods like tap tests, resonance, and MIA performed
better on fiberglass than carbon skin. This is because carbon is stiffer than fiberglass so
the effect of damage in fiberglass is more pronounced than in carbon-skinned
honeycomb. By the same reasoning, the inspection performance declined as the skins of
the honeycomb structure became thicker. A defect creates the greatest loss of stiffness in
the weakest fiberglass structure, and the effect of similar-sized defects is lessened as the
skin thickness increases and the stiffer carbon material is used.
• Some of the inspection methods included inspector performance levels that were much
higher than all of the other inspectors. When the performance outliers were removed from
the PoD calculations, the 90% PoD levels could be improved by as much as 30%.

280
 Airlines will need to assess their own inspection personnel to determine where their
performance lies amid the scatter of PoD curves shown in this report.
• Effect of inspection rate on PoD – To some degree, PoD can improve as inspection time
per area of honeycomb structure increases. Results showed that flaw detection improved
(i.e., PoD levels decreased) as the inspector took more time to inspect each panel. Most
of the improvements were in the shorter inspection times—increasing the inspection time
from
5 to 15 minutes for each 2.25 ft.2 panel could reduce the inspection PoD by as much as
60%. The improvement levels off to the point where increased inspection time does not
yield better results, so optimum inspection rates determined by this type of data is a
coverage of approximately 7 minutes per square foot of inspection region.
• False calls – Overall, the false call rates for most of the NDI methods were deemed to be
quite low. For the conventional methods, the Woodpecker device produced the lowest
overall false call rate while the Boeing tap hammer produced the highest overall false call
rate. For the advanced NDI methods, the lowest number of false calls occurred with the
laminography and CATT system while, in general, the number of false calls did not seem
to increase as the inspections became more challenging (i.e., thicker skin laminates). The
trend analyses did not show any reduction in false call rates as the inspection times were
increased. This indicates that the false call rate was much more dependent on individual
interpretations than on the rate of making the interpretations. There is no relation between
the number of false calls and the resulting PoD values for each composite honeycomb
structure category so the false call rate had little effect on PoD values. Note that in certain
cases, false call averages were dominated by one or two inspectors with extremely high
numbers of false calls. Airlines will need to assess their own inspection personnel to
determine where their performance lies amid the variation in false call levels listed in this
report.
• Inspection surface coverage – Some inspectors marked grids on their panels to aid in
coverage of inspection area. Other inspectors used indexing methods with straight edge
and tic marks around the panel perimeter. While most inspectors had good coverage of
the inspection surface, it was noted that some followed a random pattern—resulting in the
detection of small flaws while some large flaws were missed.
• Setup procedures – Boeing and Airbus inspection procedures were provided for the
inspectors to use; however, very few of the inspectors referenced these procedures during
their testing. One result is that some of the spacing for the manual tap testing was not as
small as the 0.5″ recommended in the OEM procedures. It was noted that most of the
inspectors used reference standards to set up their equipment and obtain a reference for
the expected flaw response.
• Tap testing of composite honeycomb structures – The tap test inspection technique has
evolved from a hearing-based manual test into instrumented systems in which the
electronic circuits have removed the dependence on the human ear. The mechanized
impactors have ensured uniformity of the tapping and increased the inspection throughput
while improving flaw detection performance. The tap test will eventually evolve into
automated systems capable of imaging and quantitative analysis. However, the enhanced
capabilities should not be achieved at the expense of the original advantages of simplicity
and ease of field use.

281
• Amount of overall time spent inspecting composites – The duration of this experiment
was longer, and thus more tedious, than what would normally be expected of an
inspector. The inspections lasted 2–3 days and involved 99 ft.2 of inspection region.
When subjected to exceptionally long inspections, it is not unusual for the inspector’s
attention to wane at times, which increases the possibility of missing a flaw. The
recommendation is that wide-area inspections associated with large composite structures
be divided into a series of smaller inspection regions to allow for the necessary inspection
focus. In addition, some of the more demanding inspections that involve larger regions or
complex structure should be inspected using a two-man team. Discussions on signal
quality and interpretation between the two inspectors should improve the overall flaw
detection performance.
• Lack of exposure to composite honeycomb inspections – While all of the inspectors that
participated in this experiment were trained and qualified to inspect composite
honeycomb structures, some did not have extensive exposure to such inspections. This is
because the commercial fleet does not include a lot of composite honeycomb inspections.
Thus, the experiment monitors noted some variation in the inspectors’ comfort level in
conducting these inspections. Use of the NDI reference standards or NDI feedback
specimens, which was provided for unlimited inspector use, helped alleviate this issue.
However, it does indicate that additional training and exposure to composite honeycomb
inspections—and the unique challenges associated with inspecting composite
structures—could help improve these PoD results even further

‒ There is a general concern that the lack of routine exposure to composite

inspections makes it difficult for the inspectors to maintain the necessary level of
expertise. Furthermore, exposure to available flaw specimens is viewed as a
method to keep the inspectors ready for when an aircraft needs inspection because
of damage. So, in addition to formal composite NDI training classes, aircraft
inspectors should conduct routine practice inspections on representative
composite structures that contain realistic damage. Such test specimens should be
more complex and varied than the existing NDI reference standards and contain
known, but non-uniformly spaced, flaw profiles. Industry teams that allow for
participation by OEMs, airlines, and MROs should carry out an initiative to
develop such test specimens along with specifications for specimen acquisition
and use. Added exposure to available flaw specimens is viewed as a way to keep
the inspectors ready, well-trained, and current on composite inspections

• Use of aids to ensure proper coverage – The inspection procedures discuss proper
coverage of the inspection area and even suggest the use of grids or other methods to
ensure that the inspection transducer is moved over the entire surface area. In addition,
conformable straight edges and rulers were provided to the inspectors for their use. Some
inspectors completed their work using simple freehand (unguided) motion over the entire
surface area of each specimen. Some inspectors divided the test specimens into quadrants
while still moving the transducer in a freehand motion so that they could better monitor
their coverage and transducer movement. Some inspectors used straight edges to guide
their transducer movement, while some inspectors also added tick marks to ensure that
they moved their straight edge in 0.5″ increments along the test specimens. Finally, some

282
 inspectors used straight edges in some regions and freehand in other regions (the
percentage of each was not logged, but this combined practice was noted). The inspection
results showed a significant improvement in PoD for inspectors that used straight edges.
It was similarly observed that the inspection performance decreases when the inspectors
attempt to accurately cover the entire inspection area using a freehand method. As the
inspection regions become smaller, this effect will start to decrease; however, this does
not diminish the value of the finding described here. When inspectors are inspecting large
composite honeycomb areas, they should use some form of guides or grids to ensure
proper coverage of the inspection area.
• Keys to improving inspection performance – Successful efforts to transition inspectors
from “average” to “good” or “outstanding” performance levels will have a significant
effect on PoD[90/95] levels. Overall, the identified, potential measures to improve
inspectors’ performance on composite inspections include: increased training,
apprenticeships, exposure to representative inspections, enhanced procedures, and
inspector teaming and awareness training on inspection obstacles.
• Training – The issues described above can also be addressed via additional personnel
training. Some of the training can be in the form of composite awareness training to
instruct inspectors on composite materials, composite structure fabrication, and typical
aircraft composite construction designs. Other forms of training can stress procedural
aspects of the inspections such as the use of NDI deployment aids and the proper use of
drawings to assist in signal interpretation.

‒ Based on previous Federal Aviation Administration (FAA) Airworthiness

Assurance NDI Validation Center (AANC) studies and discussions with airlines,
it was determined that a majority of the industry does not have additional, special
inspector qualification/certification to qualify personnel for conducting composite
inspections. Most companies use the normal qualification program for general
NDI inspection as qualification for composite inspection. Specialized certification
for aircraft NDI professionals who inspect composite structures should be
considered.
‒ Airlines and third-party MROs requested additional guidance related to composite
NDI training from the OEMs, the FAA, and industry groups in the areas of
specific instrument training, specific methods training, repair inspections,
composite construction training, and reference standard fabrication and use.
Programs supporting the evolution of such training should be initiated and
pursued in an industry-wide approach.
‒ Some of the specific composite NDI training needs can be addressed by more on-
the- job training and apprentice programs. Most airlines and MROs do not place
inspectors in their composite shops. An apprentice program could rotate
inspectors into composite shops so that they can learn about composite
construction while exploring the effects of different construction scenarios on
NDI.

• Based on input from the aviation industry, the FAA, working with OEMs and industry
groups, should consider publishing an advisory circular (AC) or produce a new aerospace
recommended practice (ARP) providing enhanced training guidelines specific to the

283
 inspection of composite structures. The majority of airlines think that additional training
should take place for composite inspection, so an AC or ARP outlining enhanced training
guidelines could be very useful to the industry. It will be necessary to determine an
appropriate way for this to be referenced by existing training standards such as ATA-105,
NAS-410, SNT-TC-1A, and EN-4179.

In conclusion, through the cooperative efforts of the FAA-AANC and the aircraft manufacturers,
airlines, and repair stations and research organizations, the program described herein represents a
successful, harmonized approach by the aviation industry worldwide.

One of the primary technical challenges was to produce the outstanding industry support that
allowed this team to successfully conduct the traveling experiment. The end result is a
comprehensive assessment of the ability of conventional and emerging NDI techniques to inspect
for flaws in composite honeycomb structures. The lessons learned can be used to develop
inspection improvements via optimized procedures and practices. These improvements can
produce both engineering and economic benefits to aircraft maintenance processes.

Industry-wide performance curves have now been established that determine: 1) how well
current inspection techniques are able to reliably find flaws in composite honeycomb structure,
and 2) the degree of improvements possible through the integration of more advanced NDI
techniques and procedures. This study established the current baseline for the aviation industry.
It quantified the performance of conventional NDI techniques that are currently being applied to
composites at aircraft maintenance depots and paved the way for improved industry inspections
via optimized procedures and practices. In addition, the superior capabilities of a host of
advanced NDI techniques were determined. If greater flaw detection sensitivity is needed,
advanced NDI methods are available now to address those needs. This investigation quantified
NDI performance to show that these NDI methods are ready to be applied to meet the
surveillance needs of today’s complex composite structures.

9. REFERENCES

1. Roach, D.P., Dorrell, L.R., Kollgaard, J., and Dreher, T., “Improving Aircraft Composite
Inspections Using Optimized Reference Standards,” SAE Airframe Maintenance and
Repair Conference, November 1998, SAE Technical Paper 98AEMR-34.

2. Roach, D. and Rackow, K., “Composite Honeycomb NDI Reference Standards,” SAE
Aerospace Recommended Practice ARP5606, in conjunction with CACRC Inspection
Task Group, March 2001.

3. Roach, D. and Rackow, K., “Solid Composite Laminate NDI Reference Standards,” SAE
Aerospace Recommended Practice ARP5606, in conjunction with CACRC Inspection
Task Group, March 2001.

4. Roach, D., Rice, T., and Rackow, K., “A Quantitative Assessment of Conventional NDI
Techniques for Detecting Flaws in Composite Laminate Aircraft Structures,” FAA report
DOT/FAA/AR-xx/xx, pending FAA publication.

284
5. Baker, A.A., “Fatigue Studies Related to Certification of Composite Crack Patching for
Primary Metallic Aircraft Structure,” FAA-NASA Symposium on Continued
Airworthiness of Aircraft Structures, FAA report DOT/FAA/AR-97-2, I, July 1997.

6. Fredell, R.S., “Damage Tolerant Repair Techniques for Pressurized Aircraft Fuselages,”
PhD Dissertation, Delft University of Technology, 1994.

7. Rice, R., Francini, R., Rahman, S., et al., “Effects of Repair on Structural Integrity,” FAA
report DOT/FAA/CT-93/79, December 1993.

8. Jones, R., Chiu, C., and Paul, J., “Designing for Damage Tolerant Bonded Joints,”
Composite Structures, Vol. 25, 1993.

9. Chiu, W.K., Rees, D., Chalkley, P., and Jones, R., “Designing for Damage Tolerant
Repairs,” ARL Aircraft Structures Report 450, August 1992.

10. Boeing Non-Destructive Testing Manual, NDTM ATR 72 Revision 33, Boeing
Commercial Aircraft Company.

11. Airbus Non-Destructive Testing Manual, A319/A320/A321, Airbus.

12. Migliore, R., “Investigation and Evaluation of the NDT State of the Art to Improve ATR
Aircraft Maintenance,” MS Thesis, Universita Degli Studi Di Napoli, May 2007.

13. Armstrong, K.B, Graham, L.B., and Cole, II, W.F., Care and Repair of Advanced
Composites, 2nd edition, Society of Automotive Engineers, Inc., Warrendale, PA, 2005.

14. Hsu, D., Barnard, D., and Roach, D., ASNT Industry Handbook, Aerospace NDT, “Bond
Testing,” American Society of Nondestructive Testing, Columbus, OH, 2014.

15. Cawley, P., “Low Frequency NDT Techniques for the Detection of Disbonds and
Delaminations,” British Journal of NDT, 1990, pp. 454-461.

16. Newman, J.W., “Holographic and Shearographic Applications in Aerospace

Manufacturing,” Materials Evaluation, 2005, Vol. 63 (7).

17. Thomas, R.L., Favro, L.D., Han, X., and Zhong, O., “Comprehensive Composite
Materials,” Pergamon/Elsevier Science, Oxford, U.K., 2000, Vol. 5.

18. Bossi, R., Housen, K., and Shepherd, W., “Using Shock Loads to Measure Bond Joint
Strength,” Materials Evaluation, 2002, pp. 1333–1338.

19. Adams, R.D., Allen, A.M., and Cawley, P., “Nondestructive Inspection of Composite
Structures by Low Velocity Impact,” Rev. Prog. Quantitative Evaluation, Thompson,
D.O. and Chimenti, D.E., eds., NY, 1986, pp. 1253–1258.

285
20. Cawley, P. and Adams, R.D., “Testing of Structures by Impact,” U.S. Patent No.
4,542,639, September 24, 1985.

21. Georgeson, G.E., Lea, S., and Hansen, J., “Electronic Tap Hammer for Composite
Damage Assessment,” Rempt, R.D. and Broz, A.L., eds., Proceedings of the SPIE
Conference, Vol. 2945, 1996.

22. Barnard, D., Roach, D., and Hsu, D., “Impact Testing: Evolution of a Long-Standing
Technique,” Journal of Materials Evaluation, July 2009.

23. Georgeson, G.E., Hansen, J.M., Kollgaard, J.R., Lea, S.W., and Bopp, J.R., “Damage
Detection Device and Method,” U.S. Patent No. 6,748,791, June 15, 2004.

24. Hsu, D.K., Barnard, D.J., Peters, J.J., and Dayal, V., “Physical Basis of Tap Test as a
Quantitative Imaging Tool for Composite Structures on Aircraft,” Rev. Progress in
Quantitative NDE, Vol. 5, American Institute of Physics, Melville, NY, 2000.

25. Boeing Non Destructive Testing Manual, NDTM ATR 72, Revision 33.

26. Peters, J.J., Nielsen, Z.A., and Hsu, D.K., “Comparison of Local Stiffness of Composite
Honeycomb Sandwich structures Measured by Tap Test and Mechanical Test,” Rev.
Progress in Quantitative NDE, Vol. 120, American Institute of Physics, Melville, NY,
2001.

27. Barnard, D.J., Peters, J.J., and Hsu, D.K., “Towards a Generic Manual Scanner for
Nondestructive Inspection,” Rev. Progress in Quantitative NDE, Vol. 24B, American
Institute of Physics, Melville, NY, 2005.

28. Roach, D.P., Moore, D., and Walkington, P., “Nondestructive Inspection of Bonded
Composite Doublers for Aircraft,” Proceedings of SPIE Conference on Nondestructive
Evaluation of Aging Aircraft, December 1996.

29. Roach, D., Beattie, A., and Dahlke, L., et al., “Emerging Nondestructive Inspection
Methods for Aging Aircraft,” Department of Energy SAND Report 92-2732, March
1994; FAA report DOT/FAA/CT-94/11, October 1994.

30. Palmer, D.D. and Wood, N.O., “Development of MAUS Enhancements for Large Area
Wing Inspections,” Air Force Structural Integrity Conference, December 1999.

31. Cawley, P. and Adams, R.D., “Testing of Structures by Impact,” U.S. Patent No.
4,542,639, September 24, 1985.

32. Mitsuhashi, K., Jyomuta, C., Oka, F., and Nishikawa, H., “Method and Apparatus for
Impact-Type Inspection of Structures,” U.S. Patent No. 5,048,320, September 17, 1991.

286
33. Georgeson, G.E., Lea, S., and Hansen, J., “Electronic Tap Hammer for Composite
Damage Assessment,” Proceedings of SPIE Conference, Vol. 2945, 1996.

34. Georgeson, G.E., Hansen, J.M., Kollgaard, J.R., Lea, S.W., and Bopp, J.R., “Damage
Detection Device and Method,” U.S. Patent No. 6,748,791, June 15, 2004.

35. Hsu, D.K., Barnard, D.J., Peters, J.J., and Hudelson, N.A., “Nondestructive Inspection
and the Display of Inspection Results,” U.S. Patent No. 6,327,921, December 11, 2001.

36. Peters, J.J., Nielsen, Z.A., and Hsu, D.K., “Comparison of Local Stiffness of Composite
Honeycomb Sandwich structures Measured by Tap Test and Mechanical Test,” Rev.
Progress in Quantitative NDE, Vol. 120, American Institute of Physics, Melville, NY,
2001.

37. Thomas, R., Favro, L., Han, X. and Zhong, O. “Thermal Methods Used in Composite
Inspection,” Comprehensive Composite Materials, 2000.

38. Favro, L., Han, X., and Thomas, R., “Quantitative Thermal-Wave Measurement of
Defects in Composite Aircraft Structures,” 44th International SAMPE Symposium and
Exhibition, 1999.

39. Cramer, K.E., Winfree, W.P., Reid, D., and Johnson, J., “Thermographic Detection and
Quantitative Characterization of Corrosion by Application of Thermal Line Source,”
SPIE Conference on Nondestructive Evaluation of Utilities and Pipelines III, 277–768,
1999.

40. Cramer, K.E. and Winfree, W.P., “Method and Apparatus for the Portable Identification
of Material Thickness and Defects using Spatially Controlled Heat Application,” U.S.
Patent No. 6000844, 1999.

41. Tarin, M. and Rotolante, R., “NDT in Composite Materials with Flash, Transient, and
Lock-in Thermography,” FLIR Technical Series, Application Note for Research &
Science, 2011.

42. Mayer, T., Scherling, D., and Sun, J., “Shearography Testing on Aerospace CFRP
Components,” Journal of Inspection Physics, June 2002.

43. Stakenborghs, R., “Microwave Inspection Method and Its Application to FRP,”
Proceedings of MTI AmeriTAC, 2013.

44. Underwood, S., Adams, D., Koester, D., Plumlee, M. and Zwink, B., “Structural Damage
Detection in a Sandwich Honeycomb Composite Rotor Blade Material Using Three-
Dimensional Laser Velocity Measurements,” Proceedings of the American Helicopter
Society 65th Annual Forum, 2009.

45. “CT Scans for Airplanes,” Popular Mechanics, September 2003.

287
46. Peters, J., Barnard, D., and Hsu, D. “Development of a Fieldable Air-Coupled Ultrasonic
Inspection System,” Rev. Progress in Quantitative NDE, Vol. 23, American Institute of
Physics, Melville, NY, 2004.

47. Roach, D., and Walkington, P., “Hardware and Software Innovations to Improve
Ultrasonic Penetration in Highly Attenuative Structures,” Journal of Nondestructive
Testing, November 2006, Presented at European Conference on Nondestructive Testing,
September 2006.

48. Hsu, D.K., Kommareddy, V., Barnard, D.J., Peters, J.J., and Dayal, V., “Aerospace NDT
Using Piezoceramic Air-Coupled Transducers,” Proceedings of the 16th World
Conference on NDT, 2004.

49. Kommareddy, V., Peters, J., and Hsu, D., “Air-Coupled Ultrasonic Measurements in
Composites,” Microtechnologies for the New Millennium, International Society for
Optics and Photonics, April 2005.

50. Calder, C.A. and Wilcox, W.W., “Noncontact Material Testing Using Laser Energy
Deposition and Interferometry,” Materials Evaluation, 1980.

51. Fiedler, C.J., Ducharme, T., Kwan, J., “The Laser-Ultrasonic Inspection System (LUIS)
at the Sacramento Air Logistics Center,” Rev. Progress in Quantitative NDE, Vol. 16,
American Institute of Physics, Melville, NY, 1997.

52. Dubois, M., Drake, T., and Osterkamp, M., “Low-Cost Ultrasonic Inspection of
Composites for Aerospace Applications with LaserUT™ Technologies,” Journal of
JSNDI, Volume 57, Number 1, 2008.

53. Roach, D. and Rackow, K., “Development and Utilization of Composite Honeycomb and
Solid Reference Standards for Aircraft Applications,” Department of Energy SAND
Report SAND2003-2112, June 2004.

54. SAE International, “Composite Honeycomb NDI Reference Standards,” Aerospace

Recommended Practice 5606, September 2001.

55. SAE International, “Solid Composite Laminate NDI Reference Standards,” Aerospace
Recommended Practice 5605, September 2001.

56. Roach, D.P., Dorrell, L.R., Kollgaard, J., and Dreher, T., “Improving Aircraft Composite
Inspections Using Optimized Reference Standards,” SAE Technical Paper 98AEMR-34,
SAE Airframe Maintenance and Repair Conference, November 1998.

288
APPENDIX A—EXPERIMENT BRIEFING AND INFORMATION PACKET

CACRC Inspection
Task Group

Detection of Hidden
Flaws in Aircraft
Composite Honeycomb Structure
EXPERIMENTER BRIEFING AND
INFORMATION PACKET

Experiment Coordinators: D. Roach (505)844-6078

K. Rackow

FAA Airworthiness Assurance Center

Sandia National Labs

A-1
III. Experimenter Briefing and Information
The inspection category for evaluation in this experiment is the inspection for representative
disbonds, interply delaminations, and impact flaws in composite honeycomb structures. The test
articles are modeled after the general range of construction scenarios found on commercial
aircraft. The test program is intended to evaluate the technical capability of the inspection
procedures and the equipment (i.e. NDI technique). Evaluation of inspector specific or
environment specific factors associated with performing this inspection are not the primary
objective of this experiment. However, notice will be taken by the experiment monitor if such
factors seem to influence results or if unplanned events occur which could impact the results of
the inspection. Specific notice will be taken if issues such as deployment or maneuverability
adversely affect the outcome of the inspection

For this experiment a set of test specimens containing engineered flaws has been manufactured.
The inspections will be conducted on a series of 18" X 18" panels that will be placed on a foam
pad to produce uniform boundary conditions across all experimenters. You will be asked to
inspect each test specimen and provide any information you can about the presence of applicable
flaws. If you determine that flaws are present, you should then provide size and shape
information about each detected flaw. The results should be marked directly on the test specimen
using markers provided by the experiment monitors. Inspectors should use any positive
indications to find flaws as small as 1/2" in diameter. Experimenters should work at a pace that
is comfortable for them. Although monitors will note start and stop times for your inspection,
time to inspect is a secondary variable of the experiment. Inspectors should take whatever time is
necessary to assure that any and all flaws in the test specimens are found.

1. Test Specimens and the Flaw Detection Experiment

Engineered Specimens - Engineered specimens have been manufactured that mimic the
inspection applications of interest and include realistic flaws found in those structures. Specific
information on the construction of the test panels follows. Experimenters will be told the
configuration of each panel they inspect.
• Skin Type - carbon graphite and fiberglass
• Skin Thickness - Panels have 3, 6, and 9 ply skins.
• Core - 1" thick Nomex core.
• Paint - All panels are painted as per current aircraft specifications.
• Specimen Deployment - During testing, panels will be placed on a foam pad to support the
entire footprint. (see Figure 5)
• Flaw Detection - Inspectors should use any positive indications to find flaws as small as 1/2"
in diameter.
• Tapping Device - We will provide an array of acceptable tappers (meet Boeing/Airbus
specs) and the inspectors will get to choose their tap device. This will eliminate the use of
flashlights, washers and other devices that do not meet the size and weight criteria. Some
testing with non-standard devices may also be conducted in order to form a basis of
comparison with results obtained using recommended tapper devices.

A-2
Equipment Calibration and Familiarization - Each blind inspection process will be preceded by
inspections on appropriate reference standards supplied by the experiment monitors. The
inspector will be given information on the manufactured flaws present in the reference standards
and will be allowed to use this specimen for check-out of their inspection equipment. The
reference standards have the same construction as the blind test specimens and include similar
flaws. Thus, they also can be used to allow inspectors to become familiar with an inspection
device and learn about a specific equipment's response for various composite structures and
flaws within those structures. Figures 2 and 3 show the typical flaw profile found in all of the
reference standards. The design is the same for fiberglass and carbon skin and there is a separate
specimen for 3 ply, 6 ply, and 9 ply laminates. All of the blind test specimens for this experiment
have Nomex honeycomb core so only the bottom, Nomex core portion of each reference
standard is needed for equipment set-up. Figure 3 is a cross section of the reference standard
showing how the "pillow inserts" are used to simulate interply delaminations and "machined
core" regions are used to simulate skin-to-core disbonds. The reference standards will be used as
a training tool prior to starting the experiment and will also be used by inspectors during the
course of the experiment to set-up their equipment.

1.00" DIA BREAK CORNERS 1.000"

6X 4X

FIBERGLASS CORE
FIBERGLASS 3/16" CELL SIZE
3
LAMINATE 4LB/FT CELL DENSITY
(FLAW SIDE)

FOAM TAPE
FIBERGLASS LAMINATE
FOR QA (UNFLAWED SIDE)
INSPECTION

11.000"
NOMEX CORE
3/16" CELL SIZE
3LB/FT 3 CELL DENSITY

5.500"
(ALL-AROUND)
.250"

EDGE SEAL
12.000"

MACHINE CORE (DEPTH OF 0.250" IN HONEYCOMB)

PILLOW INSERTS (LOCATED TWO PLIES UP FROM CORE MATERIAL)

POTTED CORE

CORE SPLICE

Figure 2: Final Design of Honeycomb Reference Standards;

(Same Designs for Carbon & Fiberglass Skin and Repeated for 3, 6, 9, and 12 Plies)

A-3
 MACHINED CORE

HONEYCOMB
PILLOW INSERT
LAMINATE POTTED CORE

SEALANT
CORE SPLICE

FOAM TAPE
LAMINATE (18 dB LARGER THAN
12dB LARGER THAN UNFLAWED SITE)
UNFLAWED SITE 6dB LESS THAN
UNFLAWED SITE SHOULD UNFLAWED SITE
BE 18dB LESS THAN TAPE 12 dB LARGER THAN
UNFLAWED SITE
SEALANT

Figure 3: Cross Section of Honeycomb Reference Standard Design

2. Performance Metrics

Multiple performance attributes will be discussed in the final report for this experiment. These
are given in the table below and are briefly discussed following the table. Quantitative metrics
(standards applied to events that can be numerically counted or quantified) will be applied when
appropriate but many of the performance attributes will be discussed using qualitative metrics
(standards that rely on human judgments of performance). Where practical, qualitative
assessments will be based on predetermined criteria to ensure grading consistency. The intent is
to provide useful summaries of the major factors that would influence the user communities’
perception of the viability of the technique or specific equipment. Because different users may
have different priorities, we will not rank or prioritize the various measures.

Quantitative Metrics - objective standards applied to events that can be numerically counted or
quantified.

Qualitative Metrics - subjective standards that rely on human judgments of performance; where
practical, qualitative assessments will be based on predetermined criteria to ensure grading
consistency.

A-4
 STRUCTURED
EXPERIMENT EVALUATION
CRITERIA
1. Accuracy and Sensitivity
2. Data Analysis Capabilities
3. Versatility
4. Portability
5. Complexity
6. Human Factors
7. Inspection Time

1. Accuracy and Sensitivity

Accuracy is the ability to detect flaws reliably and correctly in composite structures and
repairs without overcalling (false calls). Sensitivity is the extent to which the inspection
system responds to flaws as a function of size, type, and location (e.g., proximity to repair
edges, underlying or adjacent structural elements) in the structure.

Test results will be graded to evaluate the accuracy of quantitative measurements and to
assess qualitative measurement parameters. The test results will identify hits (calls with any
amount of overlap between the call and the solution), misses (no call for an area of a known
flaw), false calls (call with no overlap of a flaw), degree of overlap between experimenter
calls and actual flaw areas, and accuracy of quantitative call.

2 Data Analysis Capabilities

Data analysis capabilities define how well the inspection system and process can correctly
characterize flaws. Analysis capabilities include, but are not limited to, the ability to identify
the flaw size (e.g., lateral extent), flaw location, and flaw type (i.e., distinguish between
disbonds and delaminations; backside flaw detection). Quantitative aspects of the data
analysis capabilities are provided by evaluating the accuracy and sensitivity as discussed
above. Also, the repeatability, reliability, degree of automation, data storage and retrieval
capabilities and constraints, and subjective interpretation requirements are considered when
assessing the data analysis capabilities.

3. Versatility
Versatility is the capability of the inspection system to be easily adapted for application to
varying inspection tasks and conditions (e.g., varying surface conditions, specimen
orientations and accessibility). Versatility is primarily assessed using qualitative metrics,
such as calibration and equipment reconfiguration requirements to address differing
inspection applications (e.g. inspection of fiberglass skin versus carbon skin). Furthermore,
variations in system performance due to changes in the surface condition (e.g., paint
variations, front and/or back surface contaminants, surface scratches or dents), and specimen
configuration (e.g., accessibility and orientation).

A-5
4. Portability
Portability is the capability of the inspection system to be easily moved and used in standard
aircraft inspection applications. Portability is assessed using qualitative metrics such as the
inspection system’s size, weight, apparent ease of use in each evaluated inspection
application, and inspection restrictions (i.e., limitations created by power requirements,
tethering or remote control issues, safety, or other factors that may restrict equipment usage).
Equipment storage and shipment requirements will also be considered when evaluating the
system portability.

5. Complexity
Complexity is the intricacy of the tasks required to perform the inspections and data analysis.
The inspection system should be suitable for use by qualified airline NDI personnel. Also,
the inspection process should be efficient, repeatable, and reliable. Complexity is assessed
using qualitative metrics, such as: the number of people required to perform the inspection;
the number and difficulty of the range of tasks required for the inspection (including setup,
calibration, system reconfiguration for changing inspection requirements, data acquisition,
and data analysis); the number of simultaneous tasks required; tasks requiring unusual
manipulative skills (as compared to traditional inspection needs) or which place the inspector
in awkward positions that may be uncomfortable; and tasks that require advanced
interpretative skills (including calibration, data acquisition, and data analysis - both
qualitative and quantitative).

6. Human Factors
For purposes of this evaluation, human factors include procedures or equipment (hardware or
software) related inspection elements that may act as a source of human error. Environmental
factors such as temperature, noise, and lighting level will not be considered. The Human
Factors criterion is assessed subjectively considering: man-machine interface issues (e.g.,
data presentation clarity and ease of interpretation, presentation speed, layout and usability of
knobs and dials, opportunities for operational or interpretative errors, glare effects, safety to
the inspector and others in the surrounding area, etc.); written procedure usability (e.g.,
clarity, correctness, correlation to tasks actually performed); inspector education, training
(initial and recurring) and experience requirements; objective versus subjective calibration,
inspection, and analysis processes.

7. Inspection Time
Inspection time is assessed quantitatively. Set up, clean up, inspection, and analysis time will
be measured. This includes re-calibration and equipment reconfiguration time to move to
differing inspection applications.

3. Experimenter Flaw Calls and Data Logging

The purpose of this experiment is to determine the capability of various inspection methods to
detect and measure flaws in composite honeycomb aircraft structure. The Composite Flaw
Detection Experiment will travel to airlines, third party maintenance depots, aircraft
manufacturers, and NDI developer labs to acquire flaw detection data.

A-6
For this experiment a set of test specimens containing engineered flaws has been manufactured.
The inspections will be conducted on a series of 18" X 18" panels that will be placed on a foam
pad to produce uniform boundary conditions across all experimenters. You will be asked to
inspect each test specimen and provide any information you can about the presence of applicable
flaws. If you determine that flaws are present, you should then provide size and shape
information about each detected flaw. The results should be marked directly on the test specimen
using markers provided by the experiment monitors. If instructed by the experiment monitors,
inspection results can also be marked on a full-scale sheet of tracing paper. Registration
points/lines should be used on the tracing paper to assure location accuracy of the flaws. Also,
test specimen numbers should be logged onto each log sheet. Figure 4 shows a sample set of
flaw marks on one of the honeycomb test specimens. This study would like to assess
performance for flaws as small as 1/2" in diameter. Inspectors should use any positive
indications to find flaws as small as 1/2" in diameter. It is not necessary to track small
anomalies, such as porosity, that are less than 1/2" in length.

Figure 4: Sample Set of Inspector's Flaw Marks on a Honeycomb Panel

Specimen Deployment -

During the inspections, all panels will be placed on a foam pad to produce uniform boundary
conditions across all experimenters. The pad, supplied in the specimen storage cases, should be
assembled as per Figure 5 to support the entire perimeter of the test specimen.

The order of inspections will be set forth by the experiment monitors. The inspection order may
be varied as far as the 3, 6, or 9 ply specimens, however, once started an inspector will complete

A-7
all specimens of a common skin type/thickness continuously before moving on to the next set of
specimens. Since the test specimens are painted and the reference standards are unpainted, there
may be some very slight differences in equipment response. To accommodate this, each type of
test specimen (i.e. 3 ply carbon, 3 ply fiberglass, 6 ply carbon, 6 ply fiberglass, 9 ply carbon, 9
ply fiberglass) will have an unflawed area marked by a square outline. This unflawed region of
the specimen can be used by inspectors to confidently null their equipment. Change in
equipment responses, corresponding to flaws, can then be observed as per the response
variations found on the reference standards.

TEST SPECIMEN

FOAM FRAME

CENTER TEST SPECIMEN ON FOAM FRAME

Figure 5: During Inspections, Place Each Panel Such That It Is Supported Around Its
Perimeter By a Foam Frame. This Will Provide Uniform Boundary Conditions.

Additional guidance for inspectors performing this experiment are as follows:

• Experimenters should work at a pace that is comfortable for them. Although monitors will
note start and stop times for your inspection, time to inspect is a secondary variable of the
experiment.
• Applicable procedures from OEM manuals will be provided as a reference tool. Inspectors
should use their own judgment as to how to perform the inspection (i.e. a strict procedure
will not be enforced).
• Inspection coverage should be 100% of the panel with the exception of a small 1" band
around the perimeter of the panels where edge effects may create problems.
• The CACRC Composite Honeycomb Reference Standards, or equivalent, should be used to
set-up the equipment. Minor equipment adjustments stemming from in-situ calibration on the
parts being inspected are allowed.
• Inspectors should draw the entire size/shape of the flaw (i.e. delineate the edges).

A-8
• Reference standards should be used as an aid to determine where to make flaw call edges.
This is based on the diameter of the probe and how much of the probe needs to be over the
flaw in order to react/detect.
• Inspectors do not need to determine the type of flaw just the location, size, and shape of the
suspected anomaly.
• Inspectors should ignore any visual clues (surface anomalies in the paint or small surface
marks) and to avoid using these as flaw detection aids. Such anomalies may be intentionally
planted to add complexity to the inspection. Inspectors should only make a call on those
flaws that are highlighted by their inspection device.

Test results will be graded to evaluate the accuracy of quantitative measurements and to assess
qualitative measurement parameters. The test results will identify hits (calls with any amount of
overlap between the call and the solution), misses (no call for an area of a known flaw), false
calls (call with no overlap of a flaw), and the degree of overlap between experimenter calls and
actual flaw areas. Figure 6 is a grading parameter drawing that shows how the hits-misses-false
calls results will be graded. Percentage of flaw covered will be another variable of primary
interest. Error in lateral extent of flaw and maximum linear extent of overcall are variables of
secondary concern and are not currently being considered as part of the grading plan.

NOTES: DISTANCE FALSE CALL

1. THE SHAPES REPRESENT THE FLAWED AREAS AND THE SHADED IS FROM CLOSEST FLAWED AREA
AREAS REPRESENT THE AREAS CALLED AS FLAWS. MAX. LINEAR EXTENT
OF OVERCALL
FLAW NUMBER
FALSE CALL .811"
AREA
ACTUAL FLAW
AREA

1.242"

(25) FLAWLINKED TO OTHER FLAWS BY

ERROR IN LATERAL THE SAME CONTINOUS CALL
EXTENT OF FLAW

.705"
1.493"

MAX. CLOSEST SEPARATION TO ANY

ADJACENT FLAW

PERCENTAGE COVERED
< 25% 25% - 50% 51% - 75% 76% - 100% 100%
1 2 3 4 5

NOTE: ALL THESE ARE CONSIDERED (YES-DETECTED)

Y - COORD
TAP TEST INSPECTION
DATACLARIFICATION
FLAWTYPES DWG. NO. XXX
MC = MACHINE CORE DISBOND K. RACKOW, 845-9204
X - COORD PI-DE = PILLOW INSERT/DELAMINATION D. ROACH, 844-6078
PI-DI = PILLOW INSERT/DISBOND
DATE: 10-20-00 REV. #1 SCALE: 1/2
IM = ENGINEERED IMPACT
AANC
SANDIA NATIONAL LABORATORIES

Figure 6: Schematic Showing the Grading Categories Comparing Experimenter

Flaw Calls with Actual Flaw Information

A-9
 4. Sample NDI Procedures for Tap Testing and Ultrasonic Inspection of
Composite Honeycomb Structures

Attached are a series of Boeing and Airbus inspection procedures. The procedures are for
general deployment of NDT equipment that is relevant to this flaw detection experiment. The
NDI procedures are included here as general information to aid inspectors in preparing for the
flaw detection experiment. It is not expected that these procedures are sufficient to train an
inexperienced inspector. Rather, they provide additional background and guidance to inspectors
who are already familiar with the equipment and have experience in performing these type of
composite honeycomb inspections. The NDI Reference Standards provided with this experiment
can be used in lieu of the standards described in the attached procedures.

A-10
 APPENDIX B—EXPERIMENT OBSERVATIONS

Ref. Std
Exp. used
NDI With for Set-
Exp. Device up Innovative Procedure or Practice Accessory Information

Formal linescan pattern of inspection at 1/2" Often spoke with others (while not looking down) &
10 None Yes increments-rotate panel 90 deg. then moved woodpecker along the panel. Poss. Cause
repeat.Local sizing done on flaws. of some missed flaws.

Uses hand grid marks to guide coverage.

Used larger dia. Side of tool. Initial random
When tapping panel edges, panel is moved off of
scan proc. Over panel by selected areas,
6.5 ? Yes later divided panel into 4" squares & tapped
foam frame. 9 Ply panels used greater tapping
force to detect defects (taps very hard).
zigzag across each sq. in 1/4"-1/2"
increment.

Both ends of tool used to tap, mostly small

11 ? Yes end used. Formal scan zigzag scan pattern
of approx. 1-1/2"-2" band.

Initial equipment calibration 30 minutes and

6 ? Yes after that 5 minutes when changing material
or no. of plies.

8 ? Yes

Used a straight edge for a guide on 9 ply

Cautioned danger of using excessive gain on 3-ply
1 ? Yes panels. Initial calibration and 9 plys 30 min.
panel #8 (3F-2). Ref.:B767 NDT manual 51-00-05
and 5 min. or less for others.

Lack of good response from woodpecker on

Possibly moving device faster than it can acquire
12 None Yes 9 ply panels:used yellow indicator for flaw
averages.
detection here.

Use of earplugs occasionally to filter out adjacent

1 ? Yes Very methodical. noise-feels it helps focus on sound & sound
anomalies in his panel.

Moved probe very fast in sliding motion-possibly to

Difficult to cal & get consistent signal during
the point of producing false alarms since V-95 will
9 ply insps. Used area just short of alarm
22 Yes Yes (below 60 but not alarming) to make flaw
alarm over small poros. Pockets/anomalies if it is
moving but meter will slide back to 60-80 if probe is
calls.
allowed to dwell over the area of interest.

Fast rapid testing in no set order-coverage could

Unique use of tissue earplugs to filter out
be improved with slower methodical tapping,
4 Yes Yes extraneous noise & adjacent tapping
however this more rapid coverage appears to be
response.
typical of tap testing.

Deaf in one ear. Difficult @ 9 plies especially with

Grid marked on panel to assist inspection-
adjacent tapping noise. More exception than the
16 ? Yes 10mm increments. Inspection carried out
rule-very good coverage assured through use rule
using steel rule.
& grid & careful attention to coverage.

B-1
 Exp. Ref. Std
NDI
With used for Innovative Procedure or Practice Accessory Information
Exp.
Device Set-up

Scans made in random lines across panel &

Tendency to keep working while talking to others.
sized as soon as found, using both ends of
Inconsistent method of using tool, sometimes just
4 ? Yes tool. Tap spacing not as small as
used tool weight to tap & other times used the
recommended 1/2"; dropping tapper rather
hand force to ??? The tool.
than using it as an impactor.

Regular scan line pattern of 1/2" separation.

35 ? Yes Defects sized when found.

Typical scan speed approx. 12ft/min. on all panels.

Good coverage but possibly moving probe
As a result the alarm would not sound if system
quicker than it can respond-results will tell.
20 Yes Yes Scan made in regular line scan pattern of
was normally set-up on the ref. stds. Dave adjusted
his device so that alarm would sound on ref. std.
1/2" index defects sized as located.
flaws-then did test panels.

Panel scanned in lines with approx. 1"

between each tap then local sizing of located
15 ? Yes defects-also random scanning anywhere on
panel. Scan mode with large dia. Defect
sizing done with large and small diameters.

Careful coverage; on 9 ply, set-up equipment using

60% @ flaw rather 80%@ unflawed to increase
Marked quadrants on panel to use for
5 None Yes inspection coverage aid. Good Coverage
sensitivity; noted false alarms caused by probe
movement & dwelled at each spot to avoid making
false calls.

Marks area 1" inside of edge & mid-points on

2nd day using device not comfortable with it's
panels before starting insp. Moves device
11 None Yes with fast pace. Scans to fast for woodpecker
capability (just not happy w/it). Getting yellow
indication on good areas of ref. Std. Near flaws.
to respond properly.

He used a 2 x 4 the 1st day only as a guide

He looked at visual indications on impact panels.
for tapping along the edge. He paid attention
5 None Yes to the digital readout & not the change in
Covers area regions instead of lines of tapping
(last day).
pitch of each tap.

Switched to Boeing "phase only" calibration @

2 probe types tried. Wasn't sure which probe
beginning of 3 ply graphite panels. Orig. cal was
was best for this application. Didn't know
per Zetec op manual. Boeing technique is
12 ? Yes how attenuator for DTE probe was used.
contained in Boeing 747 NDT manual, Part 4, 57-
DTE probe didn't seem to work properly
22-01. Noise prob. on panel 26 new instrument
w/attenuator.
produced same results.
Used milled core to set threshold. He had
Noticed erratic signals part way thru panel 10
difficulties calibrating on the 6 ply fiberglass
(tenth panel) and identified problem with cable.
7 Yes Yes std. Obtained indications between potted
Cable was replaced. Lost signal from probe
core and pillow inserts. Signal to noise was
switched to S-PC-P11 #963927
marginal.
Broke panels into 5 or 6 segments and used
raster scan within seg. markings were only at Expressed concern about being able to find 1/2"
top of panel. After tapping panel insp. would flaws if they were similar to insert flaws on stds.
16 ? Yes slide hammer across surface to listen for Indicated that he would not normally look for these
changes in vibration pitch. Did this on some conditions during in-service inspections. Became
panels - indicated he found some defects he very bored during second day of experiment.
missed with tapping.
Expressed frustration in using digital hammer.
Spent approx. 1 hour familiarizing himself Difficult to watch display while tapping in consistent
w/tap hammer & stds. Indicated difficulty in pattern. Ultimately used tap hammer in convent.
7 ? Yes establishing reject threshold. 10% variation manner w/digital data as back-up to evaluate
was often within noise. indications. (10% variation from baseline as
threshold)

B-2
 Exp. Ref. Std
NDI
With used for Innovative Procedure or Practice Accessory Information
Exp.
Device Set-up
Used raster type scan pattern on entire
panel. Repeat. inspection twice-once with
large end and then small end. Changed scan Spent 1hr 10min. Familiarizing himself w/Boeing &
5.5 ? Yes pattern approx. midway thru -divided panel in Airbus procedures and evaluated tap hammers.
left & right halves and did rast. scan on each
half.

3-ply marked some visual indications before

actual woodpecker insp. He looked over Missed second day of testing due to illness and
11 None Yes manual before starting insps. Marked points caught up in final 2 days.
for sizing of flaws.

Looked for visual clues before scanning

Didn't read over manual on device just verbal
panel. Marked cl's on probe to delineate
instructions from SNL. 9 ply panels he dropped
edge of flaw. Somewhat random coverage
gain so flaw would alarm on device. (Mach. Core
25 None Yes pattern. 2nd day marked quads. on panel for
would alarm w/80% setting but not PI) Probe was a
cov. & seems to be marking with less
lot more sensitive during inspection with this set-
concern for sizing (No dots for sizing just
up.
quick circle on flawed area.)
Did not use audible alarm, instead used
Comments: Seems real sensitive to deployment
alarm light and variance in meter indication.
Yes, angle of device & more delay in signal response on
24 Yes He used straight line visual coverage to
thicker plies. Somewhat disappointed in
10 yrs. inspect panels. 6 ply stds were difficult to get
equipment. Cal. Date Feb. 7, 99
flaw response from.

Used a piece of paper to guide and

determine coverage of panel (straight edge). Gets real consistent nos. on readout with each tap.
3 None Yes Wrote tap values on paper to keep straight Gave us a very serious effort during inspection.
what should be called a flaw.

0 None Yes Self-guided testing

Has good coverage of panel and serious Noted on ref. Std. signal affected by probe
effort on panels. Originally tried several orientation & pressure. Scan speed also an issue
10 Yes Yes machines to find the best one. Not happy and more so on 6 ply than 3 ply. Ran on his own
with set-up but best available. after start up (second shift)

Asked about using grid overlay for coverage

(uses one on aircraft) but decided not to use
He did not feel equipment was reliable enough or
6 None Yes on panels. Moves probe a little fast on 3-ply,
capable of inspecting 9-ply laminates.
coverage good. Says not calling small flaws
on 6-ply, paint bubbles, etc.

He was distracted some by tapping of Wichitech at

Yes, Taps panel in halves to insure coverage.
same time. Tried and preferred using his own
Seemed to find pitch change & then hit
17 tapping Yes general area & then mark flaw. (Not real
tapper (easier to detect flaws and hear w/his) Only
in gen. used Ref. Std. For initial set-up of each panel type
concerned on exact flaw sizing.)
(never repeated)

All plies, listens to several taps, hears

different pitch & verifies by looking at digital
13 None Yes readout. Has good coverage of panel. Used Did a good job of defining edges of flaws.
digital readout more frequently on thicker
laminates.

0 None Yes Self-guided testing

B-3
 Exp. Ref. Std
NDI
With used for Innovative Procedure or Practice Accessory Information
Exp.
Device Set-up

Had good coverage of panel, tapped panels

Uses same design hammer in current work
in direction (horiz.) then when completed
5 Yes Yes would rotate panel 90 deg. And inspect
assignment only it was made of brass (heavier).
Good conscientious worker.
again.

Good coverage of panel area. Slight angle of

No previous exp. w/WP. Read user's manual. Did
deployment with device but later noted
see visual clue but didn't get indication from WP so
1.5 None Yes critical to keep device straight. Set-up on std.
he did not mark panel. Stated gray panels more
and sometimes did re-cal on specimens
difficult to inspect.
especially if square marked on it.
Good coverage of panel area. Read users
manual over. Initial set-up (1st few panels)
Reviewed user's manual more than once. For 9 ply
was with gain b/w 80-100 and then changed
7.5 None Yes to 80 after that. Scans panel in one direction
panels he had to turn gain down to 74-76 in order
to get alarm to respond to flaw.
& then rotates panel 90 deg. and scans
again.

Good coverage of panel, re-verifies all

Yes, Took a fair amount to set-up inspection device for
1.5 Yes indications and takes time inspecting panels
panel configuration.
18 mo. (note longer insp. times).

Took a lot longer to set-up to inspect 6-ply carbon

Good coverage of panel, not as concerned
panels versus the 3-ply & says delam and potting
Yes, with sizing of flaw as finding flaw. Sometimes
4 Yes when scanning would lift front point of probe
much harder to detect in 6-ply, but can easily find
7yrs. disbond on std. Noted gray paint on panels less
(3-point probe) off of panel surface.
noise than white paint.

Good coverage of panel. After grading

Could not get equipment to alarm on delam of Ref.
results it should be noted that he had a high
Yes, Std during set-up (3C). I think typical for all
13 Yes detection of backside flaws (even called a
construction scenarios. Has used this equip. only a
6yrs. few PI-D and PI-DEL 1.00"- 2.00" backside
few times, only when S-9R is not available.
on 9 plies fiber.?).

Good coverage of panel, didn't use audible Definite problem with probe at end of 6 ply panels
alarm and initially moved probe at a pretty and constant flaw signal when moving probe over
5 None Yes fast pace. Looked for visual clues at the start panel. Battery light was coming on a lot even after
of several panels. Had problem with device replacing with a new battery. Noise at probe seems
at end of 6 ply panels. louder.

Good coverage of panel, takes time. Used

Had to switch out initial probe after determination
Yes, United's S-9R set-up procedure. Looked in
12 Yes our provided book w/procedures, but was not
that it was bad. Then changed out probe pitch
6yrs. catch points.
enough information.

Good coverage of panel and did catch

problem with 6 ply panel having different Used the plus or minus 10% for making calls.
13 None Yes response values from std, and it turned out Frustrated on 9 ply panels, due to lack of
he had wrong panel type for std. (was paying instrument response repeatability.
attention to detail)

? ? Yes Self-Monitored

Backside flaw detection for this advanced

? ? Yes Self -Monitored
technique includes the .75 size.

Exp. Ref. Std

NDI
With used for Innovative Procedure or Practice Accessory Information
Exp.
Device Set-up

B-4
 Started w/panel #26 on day 2. Because Changed from 1/2" to a 1" probe on the second
panel has a lot of porosity it was difficult to day. Took relatively long to calibrate but then went
3 3 yrs Yes set device with standard and inspect that through panels quickly. 3 yrs. experience using S-
particular panel. 9R.

Used 1/2" probe for inspections. 3-4 yrs.

12 3.5 yrs Yes Used audio alarm. experience on S-9R. Inspector uses mostly S-9R &
TTU devices in his NDT work.

No previous exp. w/V-95. Read through

Used mylar strip to guide probe on initial
manual/worked with device prior to experiment.
pass-then without to draw flaw. Drew lines
Noticed improved response after 9v battery was
initially to eliminate edge effect-later panels
2 None Yes drew quadrants for coverage - 2nd day (6
replaced. Stated that manual instructions denoted
flashing light meant low battery-battery light on
ply) drew 8 segments. Very good panel
device did not flash, it stayed lit. Noted
coverage.
imperfections in paint surface affected inspection.
Inspection setting Average 1/4. Good
Noted that leaning on device gave false results.
coverage of panels. No previous experience
Considerable improvement in response after
6.5 None Yes with this device. Tap Hammer and Heatcon
changing batteries-battery light was 'on' - wasn-t
coin used mostly by this inspector in his daily
sure it meant low batteries.
work.
1st morning hangar background noise <40dB.
Looks over panel for visual indications. Uses
Moved inspection station away from ventilation
round side of hammer for most tapping with
system after midday of first day. Says they don't
pointed side used for smaller flaw sizes.
11 20 Yes Good coverage of panel (at least early on).
typically tap anything over 3 plies (Boeing does not
allow it). He noted difficulty in detecting pitch
On 6 ply panels started to use 3"-4" grid
change of taps late in day. Insp. also A & P split
pattern on panel like insp. 51.
duties.
Lead Man. Read over Woodpecker manual before
Used good scan speed (not too fast) and
starting inspections. Conscientious about work and
good coverage of panels and most of the
results. On 9 ply panels to make a call he would
time scanned 2nd time in 90 deg. direction
insure it was red signal or if yellow would scan in
14 None Yes from first scan. Used two different aids to
90 degree direction to verify consistent repeatable
scan with so woodpecker would stay
signal. Several times each day people would come
perpendicular to panel. Re-cal more frequent
talk to him during inspections. Insp. also A & P split
on 9 plies.
duties.

B-5
 APPENDIX C—DISTRIBUTION LIST FOR DOT DOCUMENT

Frances Abrams Masahiro Asano

US Air Force Japan Airlines
WPAFB, OH Tokyo, Japan

Paul Acres Cindy Ashforth

Lockheed-Martin FAA
Ft Worth, TX Seattle, WA

Douglas Adams Hesham Azzam

Vanderbilt University HAHN Spring Ltd
Nashville, TN Southampton, United Kingdom

Tasdiq Ahmed John Bakuckas

Thermal Wave Imaging FAA WJ Hughes Technical Center
Ferndale, MI Atlantic City Int’l Airport, NJ

Aydin Akdeniz Rocky Ballew

Boeing United Airlines
Seattle, WA San Francisco, CA

Dick Alberts Tom Barber

Digiray Corporation Delta Airlines
Danville, CA Minneapolis, MN

Nick Amabile Dan Barnard

US Navy Iowa State Univ – CNDE
Lakehurst, NJ Ames, IA

Jay Amos David Barrett

Cessna Aircraft Co. US Navy
Wichita, KS Patuxent River MD

Shreyas Ananthan Bob Barry

US Department of Energy Bell Helicopter
Washington, DC Ft. Worth, TX

Paulo Anchieta da Silva Eric Bartoletti

Embraer Southwest Airlines
São José dos Campos, Brazil Grapevine, TX

Jim Arnold Zachary Bender

United Airlines Delta Air Lines
Houston, TX Atlanta, GA

Yoshiaki Asako Rob Bergman

Mitsubishi GE Energy
Addison, TX Schenectady, NY
C-1
Phil Berkley John Bohler
GKN Westland Aerospace Delta Air Lines
United Kingdom Atlanta, GA

Malcolm Berner Christian Boller

Delta Air Lines Fraunhofer Institute
Atlanta, GA Saarbrücken, Germany

Blake Bertrand Mike Borgman

Boeing Spirit Aviation
Seattle, WA Wichita, KS

Subra Bettadapur Richard Bossi

US Navy Boeing
Patuxent River, MD Seattle, WA

Anne Birt Francis Boudreault-Leclerc

QinetiQ Olympus NDT
Farnborough, United Kingdom Québec, Canada

Werner Bischoff John Brausch

Lufthansa Technik AG US Air Force
Hamburg, Germany WPFB, OH

Wolfgang Bisle Nick Brinkhoff

Airbus Cessna Aircraft Co.
Breman, Germany Wichita, KS

James Bitner Alistair Burns

Olympus NDT Air New Zealand
Kennewick, WA Auckland, New Zealand

Sara Black Rex Carlton

High Performance Composites Delta Airlines
Denver, CO Minneapolis, MN

Kay Blohowiak Charles Buynak

Boeing US Air Force
Seattle, WA WPAFB, OH

Clemens Bockenheimer Chris Carella

Airbus UTC Aerospace Systems
Toulouse, France Vergennes, VT

Bryce Boe Sander Carneiro

Raytheon Aircraft Co. Agencia Nacional de Aviação Civil
Wichita, KS São José dos Campos, Brazil

C-2
Christopher Chandler Ed Cosgro
Delta Airlines Petroleum Helicopters, Inc.
Atlanta, GA Lafayette, LA

Che-Yin Chang Richard Costantino

China Airlines UTC Aerospace Systems
Taoyuan, Taiwan Chula Vista, CA

Fu-Kuo Chang Joe Cotter

Stanford University UPS
Stanford, CA Louisville, KY

Randy Chappelear Danny Crab

Delta Air Lines Cargolux Airlines
Atlanta, GA Luxembourg

Carlos Chaves Elliott Cramer

Embraer NASA - LaRC
São José dos Campos, Brazil Hampton, VA

BoChye Cher John Cramer

Singapore Air Kalitta Air
Singapore Detroit, MI

Eric Chesmar Eric Cregger

United Airlines Boeing
San Francisco, CA Seattle, WA

George Clamser Matt Crompton

Delta Airlines Dantec Dynamics
Atlanta, GA Holtsville, NY

Heath Coker Curt Davies

Delta Airlines FAA WJ Hughes Technical Center
Atlanta, GA Atlantic City Int’l Airport, NJ

Ron Cook Mark Davis

American Airlines Sikorsky Aircraft
Tulsa, OK Stratford, CT

Jeff Cornell Paul Davis

Aviation Technical Services Delta Airlines
Seattle, WA Atlanta, GA

Vicente Cortes Russell Day

Airbus Kalitta Air
Madrid, Spain Detroit, MI

C-3
Mark Derriso Tom Eischeid
US Air Force General Electric
Wright-Patterson AFB, OH Lewistown, PA

Mike Derby Rebeca Elford

US Dept of Energy United States Air Force
Washington, DC Kirtland AFB, NM

Matt Dill John Ellington

Nordam FedEx
Tulsa, OK Indianapolis, IN

Leo Dominguez Robert Fabyan

American Airlines/TAESL Kalitta Air
Fort Worth, Texas Detroit, MI

Fernando Dotta Tim Fallon

Embraer US Navy
São José dos Campos, Brazil Patuxent River, MD

Steve Douglas Bennett Feferman

FAA Laser Technology Inc.
Washington, DC Norristown, PA

Christopher Dragan Luis Fernandes

Air Force Institute of Technology TAP Portugal
Warsaw, Poland Portugal

Tommy Drake Joy Finnegan

iPhoton Aviation Maintenance
Fort Worth, TX Rockville, MD

Tom Dreher Carl Fisher

Rolls Royce Engine FedEx
Indianapolis, IN Memphis, TN

Marc Dubois Tom Flournoy

iPhoton FAA WJ Hughes Technical Center
Fort Worth, TX Atlantic City Int'l Airport, NJ

Don Duncan Brian Flinn

US Airways University of Washington
Charlotte, NC Seattle, WA

Paul Ebert Rafael Fávaro Foltran

ST Aerospace Agencia Nacional de Aviação Civil
San Antonio, TX São José dos Campos, Brazil

C-4
Peter Foote Keith Gilmore
Cranfield University United Airlines
Cranfield, United Kingdom San Francisco, CA

Andrew Freese Juan Gomez

Air New Zealand United Airlines
Aukland, New Zealand Orlando, FL

Mark Freisthler Thomas Gonzales

FAA FedEx
Renton,WA Los Angeles, CA

Scott Fung Steve Goncz

FAA Sky West
Renton, WA Salt Lake City, UT

Steve Galea Grant Gordon

Defence Science and Technology Org Honeywell
Melbourne, Australia Phoenix, AZ

Yolanda de Frutos Galindo Nathalie Gouret

Airbus Airbus
Madrid, Spain Blagnac Cedex, France

Dave Galella John Graff

FAA WJ Hughes Technical Center Delta Airlines
Atlantic City Int'l Airport, NJ Minneapolis, MN

Rachel Gayle Dennis Granger

United States Air Force US Army
Kirtland AFB, NM Redstone Arsenal, AL

Marc Genest Robert Grant

National Research Council Canada FAA
Ottawa, Canada Ft. Worth, TX

Gary Georgeson Philip Griggs

Boeing GE Aviation
Seattle, WA Cincinnati, OH

Roger Gibreal Courtney Guasti

Aviation Technical Services US Army
Seattle, WA Redstone Arsenal, AL

Brad Gilliland Mike Gutierrez

General Electric Federal Express
Pleasant Hill, MO Los Angeles, CA

C-5
Fred Guzman Scott Herbert
Delta Airlines AAR Corp.
Minneapolis, MN Indianapolis, IN

Jason Habermehl Steve Hicks

Olympus NDT Timco
Quebec, Canada Greensboro, NC

Bob Hager Derek Highet

Delta Air Lines Cathay Pacific Airlines
Minneapolis, MN Hong Kong, China

Colin Hanna Keiji Hirabayashi

Bombardier All Nippon Airways
Belfast, United Kingdom Tokyo, Japan

Tim Harris Jim Hofer

Boeing Boeing
Ft. Lauderdale, FL Huntington Beach, CA

Eric Haugse Wolfgang Hoffman

Boeing European Aviation Safety Agency
Seattle, WA Cologne, Germany

Dale Hawkins Ed Hohman

FAA Bell Helicopter
Washington, DC Fort Worth, TX

Pekka Hayrinen Mike Hoke

Finnair ABARIS Training
Helsinki, Finland Reno

Rudolf Henrich Quincy Howard

Airbus Boeing
Bremen, Germany Seattle, WA

Nick Heminger Scott Huddleston

Aviation Technical Services US Army
Seattle, WA Redstone Arsenal, AL

Daniel Hebért Jeong-Beom Ihn

Transport Canada Boeing
Ottawa, Ontario, Canada Seattle, WA

Dirk Heider Takahiro Ikeda

University of Delaware Toshiba
Newark, DE Yokohama, Japan

C-6
Larry Ilcewicz Seth Kessler
FAA Metis Design Corporation
Renton, WA Cambridge, MA

Yutaka Iwahori Hyonny Kim

Japan Aerospace Exploration Agency UC San Diego
Tokyo, Japan La Jolla, CA

Dan Jacobson Tim Kinsella

San Diego Composites Falcon Jet
San Diego, CA Little Rock, AR

Bill Jappe James Kissel

Boeing Delta Airlines
Huntington Beach, CA Minneapolis, MN

Patrick Johnston Rene Klieber

NASA Langley Research Center SR Technics
Hampton, VA Zurich, Switzerland

Kevin Jones Kenneth Knopp

Gulfstream Aerospace Corp. FAA WJ Hughes Technical Center
Savannah, GA Atlantic City Int'l Airport, NJ

Rusty Jones Hiroshi Kobayashi

FAA All Nippon Airways
Washington, DC Tokyo, Japan

Ray Kaiser Kendall Koerner

Delta Air Lines Spirit Aerosystems
Minneapolis, MN Wichita, KS

Frank Kane Alan Koh

United States Air Force Singapore Air
Kirtland AFB, NM Singapore

Hirokuza Karasawa Jeff Kollgaard

Toshiba Boeing
Yokohama, Japan Seattle, WA

Kazunori Kato Jerzy Komorowski

Japan Airlines National Research Council Canada |
Tokyo, Japan Ottawa, ON, Canada

Russell Keller Ajay Koshti

Boeing NASA-Johnson Space Center
Seattle, WA Houston, TX

C-7
Mike Krehbiel John Linn
American Airlines Boeing
Tulsa, OK Seattle, WA

Paul Kulowitch Jack Little

US Navy Evisive Inc.
Patuxent River, MD Baton Rouge, LA

André Lamarre John Lundeen

Olympus NDT US Navy
Québec, Canada Patuxent River, MD

Bob Lasser Robert Luiten

Imperium KLM Airlines
Beltsville, MD Amsterdam, The Netherlands

Francois Landry John Lundeen

Bell Helicopter US Navy
Montreal, Canada Patuxent River, MD

Dy Le Doug Lutz
US Army General Electric
Aberdeen Proving Ground, MD Lewistown, PA

Ray Leseck Renato Maia

US Airways Embraer
Neville Island, PA São José dos Campos, Brazil

Arne Lewis Ben Manning

Boeing Express Jet
Seattle, WA Houston, TX

Obdulia Ley Carol Martineau

Mistras FAA
Princeton, NJ Washington, DC

Marco Liberatoscioli Marcias Martinez

Alitalia Technical University of Delft
Rome, Italy Delft, The Netherlands

Glenn Light Ryan Mather

Southwest Research Institute Timco
San Antonio, TX Macon, GA

Eric Lindgren Junya Matsuda

US Air Force Japan Airlines
WPAFB, OH Tokyo, Japan

C-8
Shin Matsumoto Scott Miller
Toshiba Alaska Airlines
Tokyo, Japan Seattle, WA

Jim Mazza Eric Mitchell

US Air Force American Airlines
WPAFB, OH Tulsa, OK

Sergio Mayer Yoichi Mizuma

Embraer Japan Airlines
San Jose dos Campos, Brazil Tokyo, Japan

Glae McDonald Elyse Moody

US Airways Aviation Week Overhaul & Maintenance
Charlotte, NC New York, NY

Jim Mcfeat Calvin Moore

BAe Systems US Air Force
Bristol, United Kingdom Tinker Air Force Base, OK

Robert Mcquire Tom Moran

FAA US Air Force
Atlantic City, NJ WPAFB, OH

Jason Meade Matt Moye

United Airlines US Air Force
Houston, TX Tinker Air Force Base, OK

Alexander Melton Tommy Mullis

Delta Air Lines US Air Force
Atlanta, GA Warner Robins, GA

Thomas Mensah Francois Museux

Georgia Aerospace Airbus
Atlanta, GA Blagnac, France

Steve Micich Yosuke Nagao

AAR Corp. Japan Aerospace Exploration Agency
Indianapolis, IN Tokyo, Japan

Clark Miller Tamotsu Nagasaka

Southwest Airlines All Nippon Airways
Highland Village, TX Tokyo, Japan

Ronald Miller John Newman

Delta Airlines Laser Technology Inc.
Minneapolis, MN Norristown, PA

C-9
Bill Nicol Dorsey Perkins
MoviMED Southwest Airlines
Irvine, CA Grapevine, TX

Steve Nolet Will Perry

TPI Composites General Electric
Warren, RI Lewistown, PA

Ronan O'Higgins, Hartmut Peters

University of Limerick, Lufthansa Technik AG
Limerick, Republic of Ireland Hamburg, Germany

Toshimichi Ogisu Keith Phillips

Fuji Heavy Industries Airbus
Tochigi, Japan Bristol, United Kingdom

Paul Oulton Steve Phillips

United Airlines Kalitta Air
San Francisco, CA Detroit, MI

Christophe Paget Jérôme Pinsonnault

Airbus Bombardier
Bristol, United Kingdom Montreal, Canada

Georgios Papageorgiou David Piotrowski

Olympic Airways Delta Air Lines
Athens, Greece Atlanta, GA

Rob Pappas Jan Popp

FAA Lufthansa
Washington, DC Hamburg, Germany

Mohd. Alamin Pardi Bill Prosser

Malaysia Airlines NASA Langley Research Center
Selangor, Malaysia Hampton, VA

Mick Patino Bernd Rackers

American Airlines Airbus
Tulsa, OK Bremen, Germany

Kieran Patton Tom Reep

Shannon Aerospace Zetec
County Clare, Ireland Issaquah, WA

Luiz Perin Kevin Rees

Embraer U.S. Army
São José dos Campos, Brazil Corpus Christi, TX

C-10
Jeff Register Patrick Safarian
General Electric FAA
Minneapolis, MN Renton, WA

Joerg Reinersmann Lamia Salah

General Electric Wichita State Univ. - NIAR
Huerth, Germany Wichita, KS

Paul Risso Liming Salvino

United Airlines US Navy
San Francisco, CA Singapore

Kurt Robinson Fernando Santos

Delta Air Lines NDT Expert
Atlanta, GA Toulouse, France

Ana Rodriguez Luis Santos

Airbus Embraer
Madrid, Spain São José dos Campos, Brazil

Raul Rojas Jeffery Schaff

Delta Airlines Sikorsky Aircraft
Atlanta, GA Stratford, CT

Craig Rolfson Carlyn Schlottman

Delta Airlines Boeing
Minneapolis, MN Seal Beach, CA

Ralph Rotolante George Schneider

MoviTherm Sikorsky Aircraft
Boxborough, MA Stratford, CT

Jean Rouchon Bob Scoble

European Aviation Safety Agency United Airlines
Toulouse Cedex, France San Francisco, CA

Ricardo Rulli Dachar Sertpunnuak

Embraer Thai Airways International
São José dos Campos, Brazil Bangkok, Thailand

Rick Russell Steve Shepard

NASA Thermal Wave Imaging
Kennedy Space Center, FL Ferndale, MI

Bob Saathoff Jimmy Shiver

Cessna Aircraft Company UTC Aerospace Systems
Wichita, KS Birmingham, AL

C-11
Vilmar da Silva do Valle Bob Stakenborghs
Embraer Evisive, Inc.
San Jose dos Campos, Brazil Baton Rouge, LA

Walt Sippel Raymond Stolarz

FAA JetBlue Airways
Renton, WA Jamaica, NY

Eskil Skoglund Larry Sullivan

DolphiTech UTC Aerospace Systems
Raufoss, Norway Chula Vista, CA

Jesse Skramstad Paul Swindell

NDT Solutions, Inc. FAA WJ Hughes Technical Center
New Richmond, WI Atlantic City Int'l Airport, NJ

Duane Slabaugh Ralph Sykes

Delta Airlines Lockheed Martin
Atlanta, GA Marietta, GA

Art Smith Nobuo Takeda

AAR Corp. University of Tokyo
Indianapolis, IN Tokyo, Japan

Scott Smotherman Chinnaphan Thattiyaphong

ST Aerospace Mobile Thai Airways International
Mobile, AL Bangkok, Thailand

Hideki Soejima Robert Thomason

Fuji Heavy Industries ST Aerospace
Tochigi, Japan San Antonio, TX

David Sokol Jeffery Thompson

LSP Technologies Boeing
Dublin, OH Seattle, WA

Holger Speckmann Darrell Thornton

Airbus UPS
Bremen, Germany Louisville, KY

Giancarlo Spera Zuhair Tibi

Alitalia Jet Blue
Rome, Italy New York, NY

Reinhardt Spiegel Sam Tucker

Airbus United Airlines
Stade, Germany San Francisco, CA

C-12
Andrew Vechart Lorenz Wenk
Honeywell Aerospace AT Airbus
Golden Valley, MN Hamburg, Germany

Victor Vilents Clemens Westerkamp

PK Design University of Applied Sciences
Moscow, Russia Osnabrueck, Germany

John Vogt Dave Westlund

Nordam FAA WJ Hughes Technical Center
Tulsa, OK Atlantic City Int'l Airport, NJ

Dennis von Seelen Kyle Wetzel

Lufthansa Airlines Wetzel Engineering
Hamburg, Germany Lawrence, KS

Chinh Vuong Al Williams

FAA ST Aerospace Mobile
Ft. Worth, Tx Mobile, AL

Simon Waite Scott Williams

European Aviation Safety Agency Southwest Airlines
Köln, Germany Grapevine, TX

Rusty Waldrop William Winfree

US Coast Guard NASA Langley Research Center
Elizabeth City, NC Hampton, VA

Patrick Walter Buzz Wincheski

Texas Christian University NASA Langley Research Center
Ft. Worth, TX Hampton, VA

Rick Wampler Ian Won

Delta Air Lines FAA
Atlanta, GA Renton, WA

Thomas Walz Roy Wong

Dantec Dynamics Bombardier
Holtsville, NY Montreal, Quebec Canada

Ben Wang Nancy Wood

Florida State Univ Boeing
Tallahassee, FL St. Louis, MO

Ed Weinstein John Vogt

FAA WJ Hughes Technical Center Nordham
Atlantic City Int’l Airport, NJ Tulsa, OK

C-13
Jun Yamanaka Rick Young
Japan Airlines NASA Langley Research Center
Tokyo, Japan Hampton, VA

Lei Yue
Taikoo Xiamen Aircraft Engineering
Xiamen Fujian, China

Sandia National Labs:

1522 Ciji Nelson

1522 David Moore
1522 Kevin Rolfe
1527 Colin McConnell
1833 David Calkins
1833 Michael Kelly
1833 William Miller
6121 David Minster
6121 Joshua Paquette
6122 Daniel Laird
6000 Jim Chavez
6600 Billy Marshall
6610 Jeff Danneels
6620 Roberto Mata
6620 Willy Morse
6620 Dennis Roach
6621 Mark Retter
6622 Robert Baca
6623 Mark Soo Hoo
6625 Barry Boughton
6626 Steve Heffelfinger
6626 Andrea Dorado
6626 Randy Duvall
6626 Stephen Neidigk
6626 Tom Rice
6630 Brad Parks
9532 Recorded Information Management

C-14