JULY 2017

VOL. 75 • NO. 7
ASNT… CREATING A SAFER WORLD!™

TECHNICAL FOCUS

CHARACTERIZATION
Opportunities for NDE
NDE Capability
Characterizing Cracks
Characterization Using RUS
PCRT Inspections
Characterizing CMCs

THE AMERICAN SOCIETY FOR NONDESTRUCTIVE TESTING

JULY 2017
VOLUME 75 • NUMBER 7

TECH FOCUS ISSUE: CHARACTERIZATION

FOCUS LETTER

953 Focus Letter

“Materials characterization is the evaluation of the structure and
properties of a material system”
John Aldrin

FEATURES

862 Beyond NDT

Opportunities for Nondestructive Evaluation:
Quantitative Characterization
Eric A. Lindgren

870 CE Evaluation
NDE Characterization Capability Evaluation
John C. Aldrin and David S. Forsyth

TECHNICAL PAPERS
862
915 Characterizing Surface-breaking Cracks
through Eddy Current NDE and Model-based
Inversion
Erin K. Oneida, Eric B. Shell, John C. Aldrin, Harold A. Sabbagh, Elias H.
Sabbagh, and R. Kim Murphy

930 Toward Characterization of Single Crystal

Elastic Properties in Polycrystalline Materials
using Resonant Ultrasound Spectroscopy
T.J. Lesthaeghe, R.A. Adebisi, S. Sathish, M.R. Cherry, and P.A. Shade

941 Process Compensated Resonance Testing

Models for Quantification of Creep Damage
in Single Crystal Nickel-based Superalloys 953
Julieanne V. Heffernan, Leanne Jauriqui, Eric Biedermann, Alexander
Mayes, Richard Livings, Brent Goodlet, Siamack Mazdiyasni

953 Handheld Fourier Transform Infrared

Spectroscopic Characterization of Ceramic
Matrix Composites
Amanda K. Criner, Christine Knott, Megan Imel, Derek King, and Ming
Chen

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 835
 departments
PERSPECTIVE
ASNT SCOPE
841 President’s Letter
846 Section News “We look forward to positive
changes in the mutual recognition
847 Society Notes
programs later this year”
854 Awards and Honors
843 Director’s Letter
860 New ASNT Certificate “When ASNT undertakes travel, it
Holders is with the goal of cultivating a
meaningful relationship and 882
mutual respect for the host
country organizations”
RESOURCES
WHAT’S NEW
861 ASNT Exams
882 Product Gallery 884 Contact ASNT
883 Spotlight: Radiographic
Testing 896 Corporate Partners
885 Web Sightings 901 Calendar
887 Industry News 966 Employment Service
Carestream Non-Destructive 967 Service Directory
Testing Announces
Appointments, Element Appoints 976 Coming Attractions
Experienced Alcoa Executive to
846 Board, NDT Products of Canada 976 Ad Index
Acquires Air Dynamics
889 New Media
Heat Exchange Institute Updates
Standards for Air Cooled
Condensers with 2nd Edition
890 Standards Update
894 New Patents

Digital Materials Evaluation content is also available online at

www.asnt.org/materialsevaluation/access.
Subscription Questions?
IN THIS ISSUE
ASNT membership includes a one-year
Characterization subscription to Materials Evaluation.
Cover Photo Credit: Institutions, or others who wish to have
a subscription without becoming an
Eric B. Shell, KBRwyle ASNT member, may simply subscribe to
the journal through ASNT. To become a
member or subscribe to the journal,
contact ASNT at (800) 222-2768 or
see www.asnt.org/membershipoptions.
Back Issues & Article Copies
Back issues of Materials Evaluation
are available for purchase. See
www.asnt.org/ME for details, or
call (800) 222-2768. Copies of indi-
vidual articles may also be obtained
through ASNT: contact the librarian
at (800) 222-2768 for more
information.
Comments & Suggestions
Letters to the editor are welcome at any
time. Letters that are timely and signifi-
cant may be published in an issue of
Materials Evaluation. Not all letters are
suitable for publication, and ASNT
makes no claim regarding publication
of a given letter. Letters should be
sent to Editor Karen Balkin at
kbalkin@asnt.org.

836 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7

VOLUME 75 • NUMBER 7
J OURNAL STAF F
PUBLISHER: Dr. Arnold Bereson
PUBLICATIONS MANAGER: Timothy E. Jones
EDITOR: Karen Balkin
ASSOCIATE EDITOR: Haley Cowans
PRODUCTION/LAYOUT: Joy Grimm
RE V IE W B OARD
TECHNICAL EDITOR
Richard H. Bossi, The Boeing Company (retired)
TECHNICAL EDITOR (emeritus)
Emmanuel P. Papadakis, Quality Systems
Concepts
ASSOCIATE TECHNICAL EDITORS
John C. Aldrin, Computational Tools
Ali Abdul-Aziz, Kent State University
Narendra K. Batra, Naval Research
Laboratory (retired)
William C. Chedister, Chedister Associates
Yiming Deng, Michigan State University
John Z. Chen, Schlumberger
John C. Duke, Jr., Virginia Polytechnic
Trey Gordon, The Boeing Company
Dietmar Henning, Level III Service, LLC
Cara A.C. Leckey, NASA Langley Research
Center
Mani Mina, Iowa State University
William E. Mooz, Met-L-Chek Company
Yi-Cheng (Peter) Pan, Therm-O-Disc/EmersonInc.
Anish Poudel, Transportation Technology
Center, Inc.
S.I. Rokhlin, The Ohio State University
Donald J. Roth, GE Aviation
Ram P. Samy, NDE Information Consultants
Robert E. Shannon, Siemens Energy, Inc.
(retired)
Steven M. Shepard, Thermal Wave Imaging
Roderic K. Stanley, NDE Information
Consultants
Mike C. Tsao, University of Connecticut –
Avery Point
Glenn A. Washer, University of Missouri
Lianxiang Yang, Oakland University
Reza Zoughi, Missouri University of Science
and Technology
CONTRIBUTING EDITORS
Bruce G. Crouse, Inspection Services
Frank A. Iddings, Louisiana State University
(emeritus)
Robert E. Shannon, Siemens Energy, Inc.
Ripudaman Singh, Pratt & Whitney
Materials Evaluation (ISSN 0025-5327) is
published monthly by the American Society for
Nondestructive Testing, Inc. Periodical postage
paid at Columbus, Ohio, and additional mailing
offices. Posted under Canadian IPM #0312819.
POSTMASTER: Send address changes to
Materials Evaluation, 1711 Arlingate Lane,
PO Box 28518, Columbus, OH 43228-0518.
838 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
The American Society for Nondestructive Testing
www.asnt.org
ASN T MI SS I O N STAT E M E N T
ASNT exists to create a safer world by advancing scientific, engineering, and
technical knowledge in the field of nondestructive testing.
SO C IETY O F F I CE R S
CHAIR: Kevin D. Smith, Pratt & Whitney, 2017
PRESIDENT: David R. Bajula, Acuren Group, Inc., 2017
VICE PRESIDENT: David A. Mandina, Mandina’s Inspection Services, Inc., 2017
SECRETARY/TREASURER: Scott P. Cargill, Mistras Group, Inc., 2017
D IREC TO R S
Mohammed A. Abufour, Saudi Aramco, 2018
Martin T. Anderson, Global Diving & Salvage, Inc., 2019
B. Boro Djordjevic, Materials and Sensors Technologies, Inc., 2018
Gregory A. Garcia, Evraz North America, 2019
Mohsen Hassanein, Quality Control Co., 2019
Danny L. Keck, KCS Enterprises, 2019
Michael V. McGloin, NDT Enterprises, 2018
Marybeth Miceli, Metal Fatigue Solutions, 2019
Yi-Cheng (Peter) Pan, Therm-O-Disc/Emerson, Inc., 2019
Robert L. Saunders, Ellwood City Forge Co., 2017
Flynn Spears, Laser Technology, 2017
John Turner, FlawTech, Inc., 2017
William F. Via, Jr., Virginia Department of Transportation, 2019
Materials Evaluation is an archival journal in nondestructive testing/evaluation/inspection. The journal’s
technical articles are refereed by experts in their fields and the papers are abstracted by major technical
abstracting services, including: Acoustic Abstracts; Alloys Index; Aluminum Industry Abstracts; Applied Mechanics
Review; Applied Science and Technology Index; Cadscan; Corrosion Abstracts; Current Contents; Energy Science &
Technology; Engineered Materials Abstracts; Engineering Index; Exploration and Production Health, Safety and
Environment; Gas Processing and Pipelining; Highway Research Info Service; INIS Atomindex; INSPEC, Institution of
Electrical Engineers; ISMEC, Mechanical Engineering Abstracts; Index to Scientific Reviews; International Aerospace
Abstracts; Leadscan; Metals Abstracts; Metals Information; Nondestructive Testing Information Analysis Center;
Nonferrous Metals Alert; Offshore Technology; PASCAL; PIRA; Petroleum Abstracts; Polymers, Ceramics, Composites
Alert; Science Abstracts (Physics Abstracts, Electrical and Electronics Abstracts and Computer and Control
Abstracts); Science Citation Index; Solid State and Superconductivity Abstracts; Steels Alert; and Zincscan.
Subscriptions to Materials Evaluation (noncommissionable) to other than members of the Society: $135 per
year domestic; $245 (prepaid) per year international, which includes special handling outside the USA. Single
copy price: $9 for members ($12 for nonmembers), except for Buyers Guide issue in June ($21.25 for members,
$26.50 for nonmembers). Claims for replacement of lost or damaged copies must be made in writing, received
within 60 days of the date of publication. No more than two claims for replacement copies will be honored in a
single year. Printed in the United States of America. Copyright © 2017 by the American Society for
Nondestructive Testing, Inc.
The American Society for Nondestructive Testing, Inc. (ASNT) is not responsible for the authenticity or accuracy of
the information herein. Published opinions and statements do not necessarily reflect the opinions of ASNT.
Products or services that are advertised or mentioned do not carry the endorsement or recommendation of ASNT.
IRRSP, NDT Handbook, The NDT Technician and www.asnt.org are trademarks of the American Society for
Nondestructive Testing, Inc. ACCP, ASNT, Level III Study Guide, Materials Evaluation, Nondestructive Testing
Handbook, Research in Nondestructive Evaluation and RNDE are registered trademarks of the American Society
for Nondestructive Testing, Inc.
Authorization to photocopy fee-coded items for internal or personal use or the internal or personal use of
specific clients, is granted by the American Society for Nondestructive Testing, Inc., for libraries and other users
registered with the Copyright Clearance Center (CCC) Transactional Reporting Service, provided that the base fee
of $2.50 per copy plus $1.00 per page is paid directly to CCC, 27 Congress St., Salem, MA 01970; website
www.copyright.com. A fee code (0025-5327/97/$2.50/0) should be used in transactions with CCC and covers
all material to be photocopied beyond that photocopying permitted by Section 107 or 108 of the US Copyright
Law. This consent does not extend to other kinds of copying, such as copying for general distribution, for adver-
tising or promotional purposes, for creating new collective works or for resale.
 PRESIDENT’S
Positive Moves
IT’S BEEN ANOTHER CRAZY MONTH with lots of
positive moves in the right direction for certification
and credentialing.
In May, I attended the Conference on Equipment
Technology (COTEQ) in Rio de Janeiro, Brazil, and
represented ASNT as the USA Pan-American delegate.
COTEQ is a major international industry event for
nondestructive testing (NDT) technology and was
hosted by our partner, the Brazilian Association of
Non-Destructive Testing and Inspection (Abendi).
Representatives from Argentina, Colombia, Canada,
and ICNDT Chairman Sajeesh Babu all contributed.
We had a great meeting on alignment of the Americas
for technician certification and credentialing, and we
look forward to positive changes in the mutual recogni-
tion programs later this year; we hope for full
alignment with Canada by the end of the year.
In May, I also attended the International Chemical
& Petroleum Industry Inspection Topical (ICPIIT) confer-
ence held in Texas. This year, the conference was in
Galveston and included a tour of the Ocean Star Off-
Shore Rig, a working rig with loads of significant
history in offshore exploration and drilling. ICPIIT was a
great event, and although it involved me taking a red-
eye flight to get there, I wouldn’t have missed it for the
world. The Greater Houston Section of ASNT really puts
on a good event.
In June, I attended the International Conference on
Certification and Standardization on NDT in Vienna,
Austria, in addition to International Organization for
Standardization (ISO) meetings, further supporting our
mission for harmonization of NDT certification and
credentialing. As you may or may not know, ASNT has
taken on responsibility for the US position for the
TC-135 Technical Advisory Group (ISO TAGS).
Specifically, TC-135 addresses nondestructive testing,
and our involvement is paramount in establishing the
US position and influence on the ISO document
content concerning nondestructive testing method-
ology, technology, techniques, and personnel certifica-
tion requirements (training, experience, and testing).
Also in June, I attended the First World Congress
on Condition Monitoring held in London, England.
Numerous national societies were in attendance for
the first world event focused on condition monitoring.
Condition monitoring (CM) and structural health
letter
We look forward to positive
changes in the mutual recognition
programs later this year
monitoring (SHM) are closely related and many times
NDT plays a vital role in those assessments. Often
automated monitoring and signal processing are
involved, but even methods/techniques such as
acoustic emission, ultrasonic sensors, vibration
analysis sensors, and other techniques are used to
predict and monitor conditions prior to component or
system failure. I personally have focused much of my
career in “on-stream inspection,” which in many ways
is a CM approach. In the petrochemical and refining
industries, being able to inspect reactors and other
critical components at temperatures up to 750 °F is
not only paramount to the continued safe operation of
the processing unit or plant, but also provides millions
of dollars in savings as compared to shutting down for
inspections. I believe that CM and on-stream inspec-
tions will continue to expand as a viable means to
assure safety and economic efficiencies. Look for ASNT
to get more actively involved in CM and SHM.
Next month, I plan on completing the series of
“HEART,” and will talk about the positive direct and
indirect effects of T – Teamwork. Additionally, as the
series will then be complete, I will be reaching out to
companies across North America to request an oppor-
tunity to present to all levels of NDT technicians the
importance of “HEART” with a motivational presenta-
tion I am calling “The Ultimate Safety Moment.” If you
or your company would like me to speak to a group of
your technicians, please don’t hesitate to reach out to
me at president@asnt.org or dbajula@acuren.com; I
welcome the opportunity. I can also deliver this
message to local and international Sections as a draw
for Level II technicians to attend the meetings.
D AV I D R . B A J U L A
2016–2017 ASNT President
president@asnt.org
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 841
 DIRECTOR’S letter
Meaningful Travel
HERE IN THE UNITED STATES, this is the season when many people take time for personal and family
holidays. It’s often a time when great rates on travel packages can be found. The days are longer,
families look to spend important time together, and thoughts of sandy beaches or far-off getaways fill
our minds and spirits. There is an overwhelming need to get out, downshift, and breathe deeply in a
city that is not your own. So, as many members and those involved in nondestructive testing (NDT)
are beginning personal and family holidays and travel, this is also a great time to once again share a
bit about ASNT’s recent travels.
Over the past twelve months, officers, committed member volunteers, and senior staff have
traveled extensively for the benefit of the Society. The countries to which we traveled included Brazil, When ASNT
France, Australia, Canada, Japan, Korea, United Kingdom, and Austria.
When ASNT undertakes travel, it is with the goal of cultivating a meaningful relationship and undertakes travel,
mutual respect for the host country organizations, and then focusing that relationship into a partnership
of importance. Having personal meetings was a critical factor in ASNT’s successful implementation of
it is with the goal
computer-based testing (CBT). This initiative was executed with thoughtful consideration of impacted of cultivating a
international agencies and organizations. Being culturally competent and responsible helped craft the
pathway to provide unprecedented access to ASNT certification examinations around the world. meaningful
The NDT community now can schedule online examinations in 175 countries. The Society is
proud to bring this convenience to this industry of dedicated professionals that value ASNT certifica- relationship and
tion as integral in their career growth and excellence.
As ASNT President David R. Bajula mentioned in his June President’s Letter, his recent travel to
mutual respect for
Canada was fruitful in signing a memorandum of understanding (MOU) with the Canadian nonde- the host country
structive testing (NDT) certifying organization. MOUs are an example of a takeaway that provides
expectations of reciprocity for each signing organization. From members being able to attend events organizations
for the same registration fee as a member of the host society to each having exhibit space at the
host society’s events to promote its respective activities, MOUs are a great way to shorten the
distance between our sister societies and remain in effective dialogue and partnership.
Mutual recognition agreements (MRAs) are another important vehicle ASNT is using to deepen
connections with our global partners. As you read this column, there are meetings taking place with organ-
izations to discuss more tangible value-added benefits and sharing of resources and knowledge backed
by an agreed upon business plan that will make your ASNT membership and certifications even more
valuable. Look for an upcoming announcement of ASNT’s first MRA signing and what it means for you.
ASNT presently has 23 MOUs with NDT sister societies and peer societies in very populous
nations such as Japan, United Kingdom, Spain, Korea, India, Australia, and Brazil, as well as
moderate size organizations in Azerbaijan, Israel, Kazakhstan, and Ukraine, among others. ASNT
views each relationship as equally valuable, regardless of the size.
If you happen to fly to your holiday destination, drive, or take a train (and, no doubt, appreciate
the NDT quality checks that your vehicle has undergone to ensure your safety), you may pass an
ASNT volunteer or senior staffer en route to shaking hands with an international Society partner to
make your membership experience more robust and impactful.
We remain dedicated to making your membership more meaningful and fruitful, and as always
we welcome your comments and suggestions. Have a terrific summer!

DR. ARNOLD “ARNY” BERESON

ASNT Executive Director
abereson@asnt.org

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 843
 FOCUS letter
Characterization in NDT: An Introduction
to the Technical Focus Issue
WELCOME TO THE TECHNICAL FOCUS ISSUE on Characterization in Nondestructive Testing (NDT). The
primary goal of this issue is to highlight current work to develop and transition innovative techniques
to not simply detect critical material conditions, but to more completely characterize the part state.
Characterization in NDT can be divided into the subtopics of damage and materials characterization.
Damage characterization strives to evaluate the dimensions and location of critical discontinuities in
materials, like cracks, delaminations, and corrosion. Materials characterization is the evaluation of the
structure and properties of a material system. The vision of NDT providing characterization capability,
for any material system, in all three dimensions, at any scale, is essentially the “moon shot” challenge
for our community. Materials
Characterization in NDT is a very broad topic, with a long history. While characterization in NDT
has been investigated for many decades, successful transition of characterization capability has been characterization
difficult for some inspection problems. Advances in NDT sensor technology, scanning hardware, imaging
capabilities, and data analysis algorithms have provided improvements in resolution and coverage.
is the evaluation
Improvements have also been made through a better understanding of the interaction of nondestructive of the structure
evaluation (NDE) modalities with material systems through first-principle models. However, NDT simula-
tions have been slow to address the full complexity of typical inspections, and challenges remain to accu- and properties of
rately evaluate the key characterization properties with respect to other varying conditions under test.
Another goal of this issue is to revisit and build on two important technical focus issues in a material
Materials Evaluation dating back to 2008. The first issue that year, organized by Ali Abdul-Aziz,
addressed modeling of NDT using numerical methods such as the finite element method. The second
system.
issue, organized by Jeremy Knopp, was on Inverse Problems in NDT. The problem of material
parameter estimation can be considered an inverse problem, where a model is defined and fit to the
measurement data. This prior technical focus issue highlighted some of the early leaders in the NDT
community using inverse methods for characterization.
The first feature article in this issue, by Eric Lindgren, addresses the challenges and opportunities
for NDT characterization. Representative examples are presented that illustrate the need and value for
nondestructive methods to characterize the material, not simply detect discontinuities. Another key
aspect of delivering NDT for characterization is understanding the reliability. The second feature article
in this issue, by John Aldrin and David Forsyth, is on how to evaluate the capability of NDE techniques
for characterization.
Four technical articles are presented, highlighting the benefit of using models for NDT characterization.
Erin Oneida et al. present an approach with results using eddy current NDT with model-based inversion for
estimation of flaw dimensions and orientation. Two papers investigate the use of inverse methods with
resonance ultrasound spectroscopy measurements for materials characterization. Tyler Lesthaeghe et al.
present a framework and results for estimating elastic properties on samples with multiple grains, arbitrary
sample shape, and without restrictions to crystallographic orientation. Julieanne Heffernan et al. present
work on the quantification of creep damage in test parts using process compensated resonant testing with
inversion methods and vibrational pattern recognition analysis, distinguished creep from other variations
in the part populations. For evaluating the environmental degradation of ceramic matrix composite
materials, Amanda Criner et al. present a model-based approach using fourier transform infrared spec-
troscopy for evaluation of surface chemical changes in materials exposed to high temperatures. Following
the 2008 technical focus issues, it is hoped that these articles further demonstrate the value of modeling
and inversion for improved characterization in NDT.

JOHN C. ALDRIN
Technical Focus Issue Editor
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 845
 ASNT Scope
provides readers with
updates on ASNT
members, sections, and
section news
activities. We depend on
member contributions
for this section. Send Charlotte
updates, announcements, CHARLOTTE, NORTH CAROLINA
and photos regarding On 20 March, the Charlotte Section had its
your Section, people, monthly meeting at Sticky Fingers Ribhouse.
awardees, obituaries, The guest speaker was Jeremy Renshaw
and other updates to from Electric Power Research Institute
presaward@asnt.org. Charlotte, who presented on “Dry Fuel
Please include Scope Storage Canister Inspection.” Renshaw
News in the subject line,
shared information about the challenges
and your name and
for examining dry fuel storage canisters in
contact information.
nuclear power plants, present state, field
trials, qualitative and quantitative overview
of NDE technologies in the field, and new Jeremy Renshaw from Electric Power
technologies under development. The Research Institute Charlotte (left) with John F.
meeting was sponsored by Olympus and Burke (right). Renshaw presented on dry fuel
was attended by 13 participants. storage canister inspection to the Charlotte
Section during its 20 March meeting.
SECTION HIGHLIGHT

The Colorado Section held

its March meeting at
Transportation Technology
Center, Inc. (TTCI), a
secure 52-square mile
facility that has 48 miles
of test track and also
houses state-of-the-art
laboratory and training
facilities. TTCI is a wholly
owned subsidiary of the
Association of American
Railroads and is consid-
ered one of the world’s
foremost rail-related
research, testing, and The Colorado Section enjoyed a demonstration of the cracked wheel detector research at
training organizations. Transportation Technology Center, Inc., as part of their 9 March meeting.

846 MATERIALS EVALUATION • JUNE 2017


Chicago Section Chairman Charles P.
Johnson of Karden Sales (left) expresses
appreciation to Larry Strouse of the Federal
Railroad Administration.
Chicago
CHICAGO, ILLINOIS
The Chicago Section of ASNT met on
13 March with 18 guests and members in
attendance at the Crazy Pour in Villa Park,
Illinois. The guest speaker was Larry H.
Strouse from the Federal Railroad
Administration, part of the Department
of Transportation.
Strouse is an ASNT NDT Level III in
six methods and an American Welding
Society Certified Welding Inspector. He is
also certified by the American Society for
Quality in five disciplines and is an ASNT
Fellow. His talk for the evening was
entitled “Nondestructive Testing in the
Railroad Industry – A Federal Perspective.”
The discussion detailed the various activi-
ties of the Federal Railroad Administration,
especially with regards to tank cars and
the hazmat regulations governing the
inspection of them. Specific inspection
intervals are set out and a number of NDT
methods are recognized, such as liquid
penetrant testing, magnetic particle
testing, ultrasonic testing, radiographic
testing, and visual testing. The American
Association of Railroads is responsible for
having qualified repair shops to carry out
the inspections under Appendix T of their
regulations.
Strouse indicated that there are about
350 000 to 400 000 tank cars in service
and about 350 repair stations with the
United States, Canada, and Mexico. A
Society
Notes
single accident can be catastrophic, so
probability of detection studies are being
conducted to try to set a 90% confidence
level that cracks will be found. Efforts are
underway to try to reduce the critical flaw
size through the use of nondestructive
testing. The personnel qualifications are
called out as SNT-TC-1A; however, the
CP-189 requirements are applied so that
the hours and training are mandatory
rather than a recommended practice.
Cleveland
CLEVELAND, OHIO
The Cleveland Section held its 20 March
meeting at Bucci’s J-Bella in Strongsville,
Ohio. Nineteen members and guests
attended this meeting. They had the
pleasure of having Dave Marcyjanik give
a very informative and entertaining presen-
tation regarding the Pittsburgh Air National
Guard and their operations, mainly
refueling of many types of aircraft from
F 18 to C S’s. Marcyjanik also gave a very
insightful look into Nadcap’s stance on
open book tests in accordance with
NAS 410.
Colorado
DENVER, COLORADO
The Colorado Section held its March
meeting at Transportation Technology
Center, Inc. (TTCI) on 9 March. Safety is
TTCI’s highest priority. At the beginning
of this meeting, a briefing was conducted
by TTCI safety manager, Srba Lalic, to
acquaint all meeting attendees and to
ensure everyone knew how to respond in
case of an emergency.
Semih Kalay, Senior Vice-President
of TTCI, welcomed ASNT members and
provided a brief overview of TTCI. TTCI is a
wholly owned subsidiary of the Association
of American Railroads, and is considered
one of the world’s foremost rail-related
research, testing, and training organizations.
TTC is a secure 52-square mile facility that
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N
ASNT Elections
ASNT voting members can vote elec-
tronically through a secure website
in the upcoming ASNT elections. All
eligible members will be able to vote
at their convenience online anytime
and from anywhere during the
seven-week balloting period. By
1 July 2017, all members for whom
ASNT has an e-mail address will
receive a balloting notification e-mail
that will include voting instructions,
a voter login identification, and a
link to the election web page. Paper
ballots will be mailed to all voting
members for whom ASNT has no
e-mail address, as well as by
request. The completed ballots
must be received on or before Friday,
18 August. Candidate information
appears in the June issue of
Materials Evaluation on pp. 652–
655. See the ASNT website at
www.asnt.org/election for more
information.
Update Your Address
The best way to change or update
your address for ASNT communica-
tions is to login to the MyASNT sec-
tion of the ASNT website. Visit
www.asnt.org/myasnt, create an
online account if you do not already
have one, and then click on the
“Contact Information” option in the
menu on the left side of the screen.
Address edit and delete options can
be found there.
Submit NDT Pics
Submissions are now being
accepted for NDT Pics, a depart-
ment in Materials Evaluation
that allows members of the NDT
industry to share their work and
experiences visually. Large, high-
quality photos or images (page
sized, 300 dpi) or questions about
the process can be submitted to
kbalkin@asnt.org.
847
section news
has 48 miles of test track and also houses
state-of-the-art laboratory and training
facilities. TTCI provides its customers—
which include freight and passenger
railroads, freight and passenger rail
suppliers, and government agencies from
throughout North America and the world—
with railway research, consulting, testing,
system engineering, inspections, training,
and technical support. Gregory A. Garcia,
ASNT Board of Directors member, recog-
nized TTCI for organizing this section
meeting and tour by presenting a Colorado
Section appreciation plaque to Kalay.
Anish Poudel, longtime ASNT member,
then provided a brief overview of the
cracked wheel detector research at TTCI.
The presentation started by providing a
brief overview of the wheel failures and the
need for reliable, efficient, and real-time
848 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
nondestructive evaluation systems for
timely detection, characterization, and
reporting of critical flaws in wheels to
assure safety and integrity of the railroads.
Several technologies that were explored
and studied by TTCI for potential deploy-
ment in North American railroad environ-
ment were presented. Poudel, with the
help of ASNT member Brian Lindeman,
also provided live demonstration of the
state-of-the-art Tycho automated cracked
wheel detector system at TTC.
Columbia River
RICHLAND, WASHINGTON
The Columbia River Section meeting was
held on 9 March at Crazy Moose Casino.
The meeting was dedicated as a section
business meeting. Topics discussed
included ideas for reaching the Section’s
growth goal by reaching out to similar
organizations and schools, as well as
publicizing Section events.
Greater Houston
HOUSTON, TEXAS
The Greater Houston Section hosted
50 members and guests at their monthly
meeting and International Chemical and
Petroleum Industry Inspection Technology
(ICPITT) committee meeting on 8 March at
the Hotel Icon in Houston. Conference
Chair C. Deal Moore presided over the
ICPIIT meeting, Board Chair Jeffrey A.
Wagner ran the Business Meetings, and
Vice Chair Rick D. Arnett ran the Executive
Committee.
As this month was the Section’s
Awards Night, the Greater Houston ASNT
Technicians of the Year and Scholarship
Students from Ocean Corporation were awarded scholarships by the Greater Houston Section.
From left: Michael Tran, Brandon Hernandez, ASNT member Jerry Fulin, Brooke Ann Beesley,
and Kristofer Christoferson.
Award winners were honored. The Greater
Houston Section encourages the study of
nondestructive testing (NDT) by providing
financial assistance to students pursuing
an NDT program of study which leads to a
degree or certificate in NDT at a junior
college or NDT training facility within the
Houston area. The Sherlock Scholarship,
awarded to students enrolled in an NDT-
related course at a university or college,
memorializes the late Charles N. Sherlock,
who was a colleague and friend to many
members of the Greater Houston Section.
The Smith Scholarship, awarded to
students enrolled in an NDT-related course
at a university or college, memorializes
the late Gerald E. Smith, who was also a
colleague and friend to many members
of the Greater Houston Section.
Section Directors Jerry Fulin presented
the Scholarship winners and Samuel R.
(Roger) Jordan presented the Technicians
of the Year. Greater Houston recognizes
Level I, II, and III technicians separately.
Winners included Jason B. Burleson
Twelve members and guests attended the Hawaii Section’s 2 March meeting.
(Level I), Tyler Smith (Level II), and John Z.
Chen (Level III). Scholarship winners
included Brandon Hernandez, George
Rodriguez, Michael Tran, Kristofer
Christoferson, Delsi Arana, Joe Garza,
Matthew Ortiz, Jacinto Torres, Amulfo
Mendoza, Tim Riemer, Manuel Gutierrez,
Xavier Benitez, Maximilano Ortiz, Abel
Solis, Dale Stein III, Brook Ann Beesley,
and Amador Ramirez.
At the end of the awards, Jerry Fulin
presented a short program on “The UT
tribulation of very large valve castings.”
Hawaii
HONOLULU, HAWAII
The Hawaii Section was called to order by
Chair Derek Barbadillo on 2 March at Buzz’s
Original Steakhouse in Pearl City, Hawaii.
Twelve members and guests attended. The
Hawaii Section newsletter drafted by Jayme
J. Martin was the main topic for the evening,
along with finalizing the Section Bylaws,
suggestions for speakers, and the “Who’s
Next” Campaign.
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 849
Participate
Participate in American
National Standards
Development
ASNT’s Standards Development
Committee (SDC) develops ASNT’s
standards, including ANSI/ASNT CP-
106: Nondestructive Testing —
Qualification and Certification of
Personnel, ANSI/ASNT CP-105; ASNT
Standard Training Outlines of
Nondestructive Testing Personnel.
ANSI/ASNT CP-189: ASNT Standard
for Qualification and Certification of
Nondestructive Testing Personnel and
ANSI/ASNT ILI-PQ: In-line Inspection
Personnel Qualification and
Certification.
If you wish to join the SDC and
participate in the development of
American National Standards,
contact SDC Secretary Charles
Longo at clongo@asnt.org. More
information is available at
www.asnt.org/publications/
standards/standards.htm.
Patents
Have you been awarded a
patent?
If you have recently been granted a
new patent by a government patent
office, we invite you to let us know
about it. We are looking for patents
that describe innovations in the
science and practice of nondestruc-
tive testing. You can send a few
paragraphs describing the invention
and its range of applications, and a
copy of the patent document (or if it
was issued by the United States
Patent and Trademark Office, you
can just give us the patent number).
E-mail to ndeguy@yahoo.com with
“ASNT M.E. New Patents” in the
subject line.
For more information on the
patents, go to the U.S. Patent and
Trademark Office website at
www.uspto.gov.
850 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
section news
Attendees of the Kuwait Section’s technical presentation on 10 April in Fahaheel, Kuwait.
Kuwait Following gourmet pizza sponsored
K U W A I T C I T Y, K U W A I T
by Test Equipment Distributors, David N.
The Kuwait Section organized its third Hamilton gave an informative talk on the
technical presentation in the current various testing that his employer, Northwest
program year (July 2016 to June 2017) on Testing, provides to the public through the
10 April at the Kohinoor Restaurant in building industry. He provided examples
Fahaheel, Kuwait, on “EMAT Guided Wave of some concrete and rebar testing and
Application For Incoloy, Alloy Furnace Tube explained the new use of carbon fiber rein-
and Pipe Support Inspection.” The presen-
tation was delivered by S. Gunasekaran, a
forced polymers and the tests these new
materials undergo. Hamilton explained the
renowned and valued NDT professional. use of radiation to test the density of soil as
Shanmugom S. Murugan introduced the well as the use of ultrasonics to test the
speaker to the audience. integrity and characteristics of concrete to
The program was sponsored by ensure the materials meet the various ASTM
the Section and about 17 professionals standards imposed for the safety of the
attended the meeting. Section Chair public. Hamilton’s presentation was very
Al-Enezi Nayef Ma’ashi appreciated the well received by the group.
speaker’s effort to deliver the presentation
in a comprehensive and interactive manner.
Lewis & Clark
PORTLAND, OREGON
On 10 April, eight members of the Lewis &
Clark Section met at Flying Pie Pizzeria in
Portland, Oregon for the monthly section
meeting. During the meeting, Emery E.
Roberts, Region 16 Regional Director, gave a
brief talk regarding upcoming opportunities
provided by ASNT for Section leadership and
the benefits of the various Sections sending
representatives to the upcoming Regional
Planning Meeting. Roberts also highlighted
David N. Hamilton (left) received a small
the improvements ASNT has made in the
token of appreciation for his presentation to
way of electronic certification exams and the Lewis & Clark Section from Michael
instantaneous results of those exams. Dunbar, Section Treasurer.
ASNT President David R. Bajula addressed the Minnesota Section at its 23 March dinner meeting.
Minnesota
MINNEAPOLIS, MINNESOTA
Seventeen members, students, and guests
attended the Minnesota Section dinner
meeting on 23 March at the American
Legion in Chanhassen, Minnesota. The
Section was honored to host David R.
Bajula, 2016–2017 ASNT President, as the
featured speaker. Bajula presented an
interesting and informative presentation
titled “The Ultimate Safety Moment,”
including an overview and discussion on
honesty, excellence, accuracy, respect, and
teamwork (H.E.A.R.T.); ASNT’s strategic
plans and goals; ACCP certification
program; and ASNT’s Center for Excellence.
The Section Board also met to elect a new
slate of Officers/Directors and to identify
meeting and tour dates for the upcoming
year.
Mohawk Hudson
S C H E N E C T A D Y, N E W Y O R K
The April meeting of the Mohawk Hudson
Section included a visit to Quad/Graphics,
a printing company in Saratoga Springs,
New York, followed by dinner at El
Mexicano Restaurant, also in Saratoga
Springs. Quad/Graphics is an American
printing company based in Sussex,
Wisconsin. It was founded by Harry and
Attendees of the North Atlantic Section’s meeting, held at the Silverwing (part of the Eddyfi Group) facilities in Swansea, Wales.
Members and guests of the April meeting of the Mohawk Hudson Section.
Elizabeth Quadracci. The company has 39
printing facilities in the United States, as
well as facilities throughout Europe, Canada,
and Latin America. Quad/Graphics prints
numerous magazines, including
BusinessWeek, Milwaukee Magazine (a title
the company owns and self-publishes),
Time, Sports Illustrated, People, and Maria
Trinidad. One of their locations outside
Wisconsin is in Saratoga Springs, where
many of the nation’s top catalogers and
publishers are concentrated.
Narragansett Bay
PROVIDENCE, RHODE ISLAND
The Narragansett Bay Section held a
meeting on 28 February at the Atwood Grill
in Johnston, Rhode Island. There were 25
members in attendance. Past President
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N
and Chair of the Board of Directors Kevin D.
Smith gave a talk on Alloy Analysis with XRF
and LIBZ technology. Smith brought along
equipment to demonstrate. An evening of
learning and fun was enjoyed by all.
At the 28 March meeting, Stephen Pflanz,
the regional sales manager for Carestream
NDT, gave a talk on changing from film radi-
ography to computed radiography. The
meeting was attended by 28 members.
North Atlantic
SHEFFIELD, UNITED KINGDOM
Members of the North Atlantic Section
travelled to Silverwing (part of the Eddyfi
Group) for a section meeting. Based in the
Wales’ second city, Swansea, Silverwing
has a global presence in the tank inspec-
tion market.
851
 section news
Jim Costain, Silvering’s VP Sales,
together with Jonathan Jones, Training
Manager, showed the Section around their
impressive facilities. In addition to
building and shipping equipment from
their Swansea base, Silverwing, in partner-
ship with the South West School of NDT,
offer accredited training and examinations
in magnetic flux leakage.
Following the tour, Neil Pearson took
the Section through the tank mapping
software, which processes data from the
tank floor inspection systems. The data
handling capabilities triggered a lot of
discussion and questions. The Section
visit coincided with the release of the
Scorpion II, a magnetic UT crawler for tank
walls utilizing a dry coupled probe. In a
demonstration, Pearson removed the
batteries so that the engineers present
were able to inspect the inside of the
equipment. Attendees were also able to
explore the Phased Eddy Current Array
system – LYFT and the FloorMap.
The section would like to thank Jim
Costain, Jonathan Jones, Stefan
Parmentier, and Neil Pearson for hosting
the visit. (Officially, it was highly educa-
tional–but really, it was like leaving
children alone in a toy shop!)
Old Dominion Section Chair Doug A. McAvoy, Jr. (left) presents a traditional Jefferson Cup to
Alan Caulder, who presented at the Section’s 27 March meeting.
852 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Old Dominion
RICHMOND, VIRGINIA
On 27 March, 12 members and guests
of the Old Dominion Section gathered at
Calabash Seafood in Mechanicsville,
Virginia, for a technical presentation on
full matrix capture and the total focusing
method for ultrasonic testing from Alan
Caulder of Advanced OEM Solutions.
Region 3 and 4 Regional Director William F.
Via, Jr. was also in attendance, and
brought the Section up to speed on the
actions of ASNT. At the end of the meeting,
Caulder was presented with a traditional
Jefferson Cup from the Section.
Pacific Northwest
SE ATT L E , WA S H I N GTO N
Twenty-five members of the Pacific
Northwest Section gathered at the Old
Spaghetti Factory in Tukwila, Washington,
to listen to the presentation entitled “Full
Matrix Capture and the Total Focusing
Method” by Alan Caulder of Advanced
OEM Solutions. Full Matrix Capture (FMC)
is hardware data acquisition of phased
array ultrasonic testing data and Total
Focusing Method (TFM) is a software post-
processing of the data captured by FMC.
The data may be post processed years
after capture, if necessary with new
Vice Chair Kathy Ferguson (left) presents an ASNT pen and pencil set and section coin to the
evening’s speaker Alan Caulder.
From left: Pittsburgh Section Chair John A. Markanich; James V. Pellegrino, RJ Lee Group, Inc.;
and Donald C. Stoll, AWS Chairman.
generation software. TFM is an improve-
ment of the synthetic aperture focusing
technique. Indications, which may be inter-
preted as a planar discontinuity when
processed with the TFM algorithms, may
turn out to be two or more rounded indica-
tions. The trained operator can thus make
a correct interpretation of data. Among the
members in attendance were six students
and two instructors from the nondestruc-
tive testing club at Clover Park Technical
College in Lakewood, Washington,
Puyallup campus. After Caulder’s presenta-
tion, he answered questions from the
audience.
Pittsburgh
P I T T S B U R G H , P E N N S Y LV A N I A
The Pittsburgh Section held a meeting on
14 March. It was a joint meeting with
American Welding Society at the RJ Lee
Group, Inc. in Monroeville, Pennsylvania.
The Section would like to thank the host,
James V. Pellegrino, and many others at RJ
Lee Group, Inc., for the tour of their facili-
ties and the excellent presentations at the
various stations. RJ Lee Group, Inc., is an
industrial forensics analytical laboratory
and scientific consulting firm. Dinner was
held at Palmieri’s Restaurant. w x
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 853
awards & honors
Larry G. Olsen Ralph M. Booth II
Charles N. Sherlock
Meritorious Service
Recognition
2017 winner Larry G. Olsen is the quality
engineer supplier at Northrop Grumman
Corporation. He holds ASNT NDT Level III
certification in PT, RT, and UT. Olsen is a
dedicated member of ASNT, serving in
myriad roles at both the national level and
within the San Diego Section. He currently
serves as Board Member and Director at
Large of the San Diego Section, where he
is an RT instructor and is involved in the
website, publications (including the
yearbook and newsletter), and student
outreach.
Olsen has served in leadership roles in
the San Diego Section for over two
decades, beginning with his position as
Board Member and Director at Large from
1995–1996. Since then, Olsen has served
as Secretary, Treasurer, Chair, and Chair of
the Regional Planning Committee for the
San Diego Section, as well as serving as
Board Member and Director at Large from
2000–2006 and from 2011 to the present.
In that time, he has given numerous
presentations, taken on leadership roles at
Regional Planning Meetings, and been
854 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Jennifer Anaya
actively involved in student outreach.
Olsen has also served as Chair of the
Awards and Nominating Committees
within the San Diego Section. In 2001,
Olsen received the San Diego Section
Chairman’s Meritorious Service Award.
On the national level, Olsen has been
a leader in the Section Operations Council,
beginning as the Chair of the Section
Management Division from 1998–2002.
He has also been a frequent speaker at
the Regional Planning Conference, and
most recently served on the Section
Operations Council and Awards
Committees. Olsen is also an ASNT Fellow
(2005) and an ASNT Mentor (2007).
Olsen received a certificate in
Radiologic Technology from Creighton
Memorial Hospital in 1972. In addition to
his longtime work with ASNT, Olsen is a
member of ASM International and ASTM
International, and has been a member of
the American Welding Society.
About the Charles N. Sherlock
Meritorious Service Recognition
The Charles N. Sherlock Meritorious
Service Recognition provides recognition
for an individual’s outstanding voluntary
service to the Society, through single or
aggregate activities, though not neces-
sarily in any single year. Activities
performed as a function of an elected
office or paid position shall not be consid-
ered. In 2004, the recognition was
renamed in honor of its first recipient,
Charles N. Sherlock. Except in extreme
extenuating circumstances, no more than
one recognition shall be made in any
calendar year. Nominations for this recog-
nition may be made by members of the
headquarters staff, Board of Directors,
councils, the Awards Committee, local
sections, or by any ASNT member.
Recognition for the
Advancement of Active
Military and Veterans in NDT
2017 winner Ralph M. Booth II is an
NDT Engineer/Responsible Level 3 at
Senior Aerospace Ketema. He holds
ASNT NDT Level III certification in PT and
RT, and holds NAS 410 Level 3 certifica-
tion in the same methods. He also
holds Level 3 VT certification from his
employer, as well as Visual Weld
Inspector Specialist/Trainer certification.
He is a member of ASNT’s Veterans
Ad-Hoc Committee.
 Booth began his nondestructive
testing career when he completed the
Nondestructive Inspectors School in the
US Marine Corps in September 2002.
Shortly after, he was deployed to Iraq as a
nondestructive inspection (NDI) technician.
After completing a tour of instructor duty at
NDI School in Pensacola, Florida, Booth
entered the nondestructive testing (NDT)
field with a temporary part time job as a
Level II PT inspector for Chromalloy in San
Diego, California. Booth went on to work in
the X-ray industry at Senior Aerospace
Ketema in El Cajon, California. While he
was working these full-time jobs, Booth
obtained training and certification until he
reached Level III.
Booth is dedicated to helping veterans
transition into the workforce, and helps
those aspiring to move into the NDT field
with understanding what qualifications
they need, where to get training, and
where to find a job. His ability to reach
large numbers of military NDI technicians
has made him a widely known resource for
helping military veterans find employment
in NDT. Booth joined Jocelyn Langlois’
Veterans Committee at the 2016 Annual
Conference in Long Beach, California. He is
also an active member of the San Diego
Section.
About the Recognition for the
Advancement of Active Military
and Veterans in NDT
The Award for the Advancement of Active
Military and Veterans in NDT recognizes
individuals or organizations who
encourage active military and veterans to
enter and thrive in the field of NDT, either
through creation and/or implementation
of programs that provide educational or
career advancement opportunities in NDT
that may not otherwise be available to
active military and veterans or through
leading by example of how active military
and veterans can not only have a career in
NDT but also contribute greatly to NDT,
ASNT, and creating a safer world.
Recognition for the
Advancement of Women in
NDT
2017 winner Jennifer Anaya is the vice
president and co-owner of Alloyweld
Inspection Company Inc., located in
Bensenville, Illinois. Anaya grew up around
nondestructive testing since her father
started a nondestructive testing laboratory
and welding shop in 1968. Her first task in
nondestructive testing was to process film
by hand after school and on weekends.
After high school graduation, her father
asked her to help out temporarily in the
X-ray department. That temporary service
has now lasted 33 years. For 15 years,
Anaya performed film X-ray inspection. In
1998, she obtained one of the first digital
computed radiographic systems. She has
made it her mission to hire as many
women as possible for positions at the
company. Over the years, she has trained
many women in X-ray and key positions in
Alloyweld. Many of those women were
trained to perform both film and digital
X-ray inspection, including her daughter
Amanda, who has recently begun a career
in radiography.
Along with raising four daughters
and serving as a local school board
president, Anaya also pioneered the way
in computed tomography in 2013 after
obtaining a new system. She is an
NAS-410, Level 3 in film, computed radiog-
raphy, and digital DDA radiography, and
has been qualified as an AWS Certified
Radiographic Interpreter since 2009.
Anaya also serves on ASTM E07.01 and as
secretary for the ASTM E07.02 committee
for Reference Radiographs. Her local
involvement with ASNT includes serving as
chair of the Chicago Section from 1999–
2000 and as treasurer ever since then.
She has worked many outreach programs,
including the IIT Engineering Week
program each year, which introduces grade
school and high school students, girls in
particular, to nondestructive testing as a
career, as well as at the local high school
through ASNT. On a national level, she has
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N
attended many digital imaging and ASNT
National Conferences and recently made a
presentation on computed tomography at
the 75th Anniversary of ASNT in Long
Beach, California.
About the Recognition for the
Advancement of Women in NDT
The Award for the Advancement of Women
in NDT recognizes individuals or organiza-
tions who encourage women to enter and
thrive in the field of NDT, either through
creation and/or implementation of
programs that provide educational or
career advancement opportunities in NDT
that may not otherwise be available to
women or through leading by example of
how women can not only have a career in
NDT but also contribute greatly to NDT,
ASNT, and creating a safer world. w x
Write Us
Back to Basics Articles
Needed
Materials Evaluation is soliciting
submissions for its “Back to Basics”
department. “Back to Basics” are
tutorial articles written to introduce
the reader to the fundamentals of
an NDT method, application, or
technology, or to act as a refresher
for those already experienced in the
subject. Articles or ideas may be
sent to: Materials Evaluation, ASNT,
1711 Arlingate Lane, P.O. Box
28518, Columbus, OH 43228-0518;
(800) 222-2768 X207; fax (614)
274-6899; e-mail kbalkin@asnt
.org.
855
awards & honors
ASNT Award Winners
Complete information on all
awards is on pp. 250–265 of the
February 2017 issue of Materials
Evaluation. Other 2017 ASNT
award winners may be
announced at a later date in the
Awards & Honors pages.
ASNT Announces Award Winners
The American Society for
Nondestructive Testing is pleased to
announce the following award winners.
These award winners will be recognized
at the ASNT Annual Conference 2017 in
Nashville, Tennessee.
50-year Member Recognition
The 50-year Member Recognition
acknowledges individuals who have
dedicated their careers to ASNT and its
mission of making the world a safer
place. This recognition was established
to honor loyal members with 50 or
more years of membership who have
contributed to the nondestructive
testing (NDT) industry and the Society
through their knowledge and dedica-
tion decade upon decade. Recognitions
are presented at the awards banquet at
the ASNT Annual Conference.
The recipients of the 2017 50-year
Member Recognition are Thomas J.
Davis, Thomas J. Flaherty, Jerry L.
Nelson, Roy Odell, and Noel A. Tracy.
ASNT Fellow
A Fellow of ASNT is an individual
member of the Society who has demon-
strated outstanding professional
distinction and who has made
continued significant contributions to
the advancement of NDT in areas such
as management, engineering, science,
education, administration, or planning.
856 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
The 2017 class of ASNT Fellows is
John C. Aldrin, Martin T. Anderson,
David A. Fry, Edward R. Generazio,
Thomas J. Larkin, Xavier Maldague,
Lawrence E. Mullins Sr., Mark A.
Ormrod, and Mark R. Pompe.
Recognition for the Advancement
of Active Military and Veterans in
NDT
The Recognition for the Advancement of
Active Military and Veterans in NDT
recognizes individuals or organizations
who encourage active military and
veterans to enter and thrive in the field
of NDT, either through creation and/or
implementation of programs that
provide educational or career advance-
ment opportunities in NDT that may not
otherwise be available to active military
and veterans or through leading by
example of how active military and
veterans can not only have a career in
NDT but also contribute greatly to NDT,
ASNT, and creating a safer world.
The winner of the 2017 Recognition
for the Advancement of Active Military
and Veterans in NDT is Ralph M.
Booth II.
Recognition for the Advancement
of Women in NDT
The Recognition for the Advancement of
Women in NDT recognizes individuals or
organizations who encourage women to
enter and thrive in the field of NDT,
either through creation and/or imple-
mentation of programs that provide
educational or career advancement
opportunities in NDT that may not
otherwise be available to women or
through leading by example of how
women can not only have a career in
NDT but also contribute greatly to NDT,
ASNT, and creating a safer world.
The winner of the 2017 Recognition
for the Advancement of Women in NDT
is Jennifer Anaya.
Charles N. Sherlock Meritorious
Service Recognition
The Charles N. Sherlock Meritorious
Service Recognition provides recogni-
tion for an individual’s outstanding
voluntary service to the Society,
through single or aggregate activities,
though not necessarily in any single
year. In 2004, the recognition was
renamed in honor of its first recipient,
Charles N. Sherlock.
The winner of the 2017 Charles N.
Sherlock Meritorious Service
Recognition is Larry G. Olsen.
Engineering Undergraduate
Award
The Engineering Undergraduate Award
was created to provide an incentive to
engineering undergraduate students
enrolled in colleges and universities in
the United States with recognized engi-
neering programs..
The winners of the 2017
Engineering Undergraduate Award are
Christopher Ingemi, Elliott Jost, and
Krysztof Mazur.
Faculty Grant
The Faculty Grant is awarded to provide
incentive to faculty members of engi-
neering or science programs at US
universities and colleges to revise their
existing NDT courses or to develop new
courses to teach NDT and NDE.
The winners of the 2017 ASNT
Faculty Grant are Yiming Deng and
Lalita Upda, of Michigan State
University, based on their proposal for
“Development of Electromagnetic NDE
Sensor Technologies Course at
Michigan State University”; and Tsuchin
Philip Chu, of Southern Illinois
University Carbondale, based on his
proposal for “Development of a New
Course for Engineering Students, ME/CE
486: Introduction to Nondestructive
Testing.”
Fellowship Award
The ASNT Fellowship is a cash award
granted to an educational institution to
fund specific research in NDT at the post-
graduate level (M.S. or Ph.D.). Academic
institutions with graduate educational
research programs are invited to submit
proposals each year.
The winners of the 2017 ASNT
Fellowship Award are Brigham Young
University (Brian Mazzeo, advisor; Jeffrey
Barton, student); Texas A&M University
(Dallas N. Little and Stephen Sebesta,
advisors; Bryan Wilson, student); Virginia
Tech (Pablo A. Tarazaga, advisor; Rodrigo
Sarlo, student); Michigan State University
(Lalita Upda, advisor; Saptarshi
Mukherjee, student); and Missouri
University of Science & Technology (Reza
Zoughi, advisor; John R. Gallion, student).
George C. Wheeler Excellence in
Personnel Certification Recognition
The George C. Wheeler Excellence in
Personnel Certification Recognition recog-
nizes individuals who have made signifi-
cant contributions to the development
and advancement of professionalism of
ASNT’s programs for the qualification and
certification of personnel for NDT. This
Recognition honors George C. Wheeler,
who authored “A Guide to Personnel
Qualification and Certification.”
The recipient of the 2017 George C.
Wheeler Excellence in Personnel
Certification Recognition is Darrell W.
Harris.
Mehl Honor Lecture
ASNT established the Mehl Honor Lecture
in honor of the outstanding research in
the field of gamma radiography and
countless other contributions by Robert F.
Mehl. This honor is presented every odd-
numbered year by the Board of Directors
to outstanding persons in the field of NDT.
The lecture is traditionally presented
during the ASNT Annual Conference.
The recipient of the 2017 Mehl Honor
Lecture is Richard H. Bossi.
Lou DiValerio Technician of the Year
Recognition
The Lou DiValerio Technician of the Year
Recognition gives recognition to ASNT
members who are deserving technicians,
encouraging their continued participation
in the Society. This recognition was
renamed in 1995 to honor Louis R.
DiValerio, who remarkably impacted the
Society in his 23 years of membership
with his dedication and willingness to
serve. Nominees for the recognition must
hold a position of employment that corre-
sponds to general guidelines similar to
those described by Recommended Practice
No. SNT-TC-1A for defining the primary
duties and responsibilities of Level I,
Level II, and Level III technicians. The
recognition recipient is an individual
who has distinguished himself or herself
by showing exceptional merit either as an
NDT technician or through service to the
Society.
Nominations for the recognition are
made jointly by the technician’s employer
and the local sponsoring section, and
must include supporting documentation
attesting to the fact that the nominee is a
practicing NDT technician. Presentation of
the recognition is made at the ASNT
Annual Conference.
The winner of the 2017 Lou DiValerio
Technician of the Year Recognition is
Michael J. Diaz.
Mentoring Award
The Mentoring Award was established to
recognize those people in the Society
working to encourage others to reach
goals they may have otherwise not sought
and to offer the rest of the membership
an example of what they could be
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N
accomplishing by acting as mentors. With
the understanding that mentoring may
take many forms, the following eligibility
categories have been established:
Individual – Assisting one or more
individuals to reach goals in the NDT
industry that they may have not otherwise
attained.
Group – Providing positive guidance
to a group such that a goal of significant
value to ASNT or to the NDT industry is
achieved by that group.
Indirect – Providing mentoring oppor-
tunities for others by creating an environ-
ment that fosters and encourages such
relationships with positive outcome.
The winners of the 2017 Mentoring
Award are David R. Blosser, David L.
Culbertson, Juan R. Diaz, Jerry Fulin, and
William H. Spaulding.
Outstanding Paper Recognition,
Materials Evaluation
The purpose of the Outstanding Paper
Recognition is to encourage a high degree
of effort toward technical, educational, or
managerial achievement in NDT through
publication in Materials Evaluation.
Recipients of the recognition are selected
on the merit of written contributions
published by the Society during the
previous calendar year. Selection is based
equally on the material’s originality,
usefulness, and clarity, and on the appro-
priateness and accuracy of its supporting
material. The recognition is presented
during the Annual Conference.
The winners of the 2017 Outstanding
Paper Recognition for Materials Evaluation
are Pietro Evola, Ambra Vandone, and
Piervincenzo Rizzo for their paper, titled
“Fractal Analysis Applied to Laser Spot
Thermography,” published in Materials
Evaluation, Vol. 74, No. 3, March 2016,
pp. 409–417.
857
awards & honors
ASNT Award Winners
Outstanding Paper Recognition,
RNDE
The purpose of the Outstanding Paper
Recognition is to encourage a high
degree of effort toward technical,
educational or managerial achievement
in NDT through publication in RNDE.
Recipients of the recognition are
selected on the merit of written contri-
butions published by the Society during
the previous calendar year. Selection is
based equally on the material’s origi-
nality, usefulness, and clarity, and on
the appropriateness and accuracy of its
supporting material. The recognition is
presented during the Annual
Conference.
The winners of the 2017
Outstanding Paper Recognition for
RNDE are Nicole Martino, Ken Maser,
Ralf Birken, and Ming Wang for their
paper, titled “Quantifying Bridge Deck
Corrosion Using Ground Penetrating
Radar,” published in RNDE, Vol. 27,
No. 2, 2016, pp. 112–124.
858 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Philip D. Johnson Honorary
Member Recognition
An honorary member is a person of
acknowledged eminence in the domain
covered by the Society or is one who
has been recognized as a benefactor to
the Society through services such as
being a national officer. In 1990, this
recognition was renamed in honor of
Philip D. Johnson, who co-founded
ASNT and nurtured it through its first
three decades. This recognition is the
highest that the Society bestows.
The winner of the 2017 Philip D.
Johnson Honorary Member Recognition
is Raymond G. Morasse.
Robert C. McMaster Gold Medal
The Robert C. McMaster Gold Medal
was established by ASNT to provide
recognition to a person who has made
an outstanding contribution, or who
has rendered meritorious service, to
any area in the field of nondestructive
testing or to the Society. Beginning in
1990, the recognition has been given in
honor of the eminent ASNT Past
President Robert C. McMaster,
professor emeritus of the Ohio State
University. The recognition is presented
to the recipient during the ASNT Annual
Conference.
The winner of the 2017 Robert C.
McMaster Gold Medal is Shant
Kenderian.
Young NDT Professional
Recognition
The purpose of the Young NDT
Professional Recognition is to recognize
individuals whose initial career contri-
butions exemplify high standards of
excellence in the areas of professional
achievement and meritorious service.
The recognition is given to supervisors,
educators, managers, researchers,
consultants, developers, and others
who are ASNT members with five to ten
years of involvement in the NDT
industry. Recognition applicants must
be sponsored by a local ASNT section,
council or national committee.
The winner of the 2017 Young NDT
Professional Recognition is Ehsan
Dehghan Niri. w x
New ASNT Certificate Holders
Below are personnel who have recently obtained their initial ASNT certifications. This list includes new certificate holders that were
added to the ASNT database through 1 June 2017. Each certificate holder’s current certification information can be found on the
ASNT website at www.asnt.org/certlist.

ACCP Level II Matthew Murphy Marty Anderson Graham C. Lee

Dustin Anderson Alvan V. Peterman Alan Patrick Andrews Abdur Rab Lone
Leo Bissonette III Jeremy Pittman Dhinakaran Ayyalusamy James W. McCoy, Jr.
Dan Brazys Justin Proctor Robert Begeny Tuan Anh Nguyen
David Brooks Narongsak Saeteaw Maxence Benzaria Surendra Kumar Puli
Andrew D. Casto Richard Sells Paul Coggin Vedamoorthy
Claudio Chavarria Mendoza Jeffrey Tyler Kenneth R. Criss Sean Thomas Radigan
David Dade Cory Vidler Justin Curtis Naseer Khaleel Rahman
Daniel Davenport M. Justine Diraviaraj Edward E. Ramirez
Gibbs N. Dike ASNT NDT Level II Meagan Fitzgerald Justin M. Rogers
Nicholas Dinsmore Daniel Stephen Christensen James S. Hall Pierre Servais
Clinton L. Duckworth Jorn Visser George Kenneth Hodges James C. Smith
Ivan Estupinan Payaam Jahanbazi Mustafa Succaria
Randy Finzel ASNT NDT Level III Sahabudeen Jawahar Ryan Wiess
Joshua Hughes Radwa Atef Abdel Wahab Soham Jobanputra Vineet Yadav
Antonio Jaroso David W. Anderson Rajasekar Ranjit Joshi

860 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
BEYON
 x ME FEATURE
w

D NDT Opportunities for

Nondestructive Evaluation:
Quantitative
Characterization
by Eric A. Lindgren

I
magine finding a piece of material while out on a hike with family or
friends in a local park. The material is unlike anything you have seen
before and appears to be unique in the way it looks and feels. Your
curiosity is triggered; what is this? Is it something of value that was lost
by a previous hiker? Is it a small piece of a meteorite? Is it something a little
more nefarious that requires alerting law enforcement? With a portable probe
you are carrying in your pack, you touch the material and immediately a
wealth of information becomes available: material type, composition,
microstructure, properties, processing, history, and past use. Each value,
including uncertainty bounds, provided by the portable instrument enables
you to understand what you found. This approach to characterization was the
basis for a line of questioning used by Robert Green of John Hopkins
University (now retired) during oral qualifying exams for his doctoral students
and illustrates the potential of using nondestructive testing-based (NDT)
methods for quantitative characterization.
When considering the above future scenario, it is important to ponder for
a moment what is meant by quantitative characterization. As the topic of a
Photo credit: Eric B. Shell, KBRwyle

special issue of Materials Evaluation, a number of readers may question if this

is new and whether current practice already enables quantitative information
to be obtained using NDT methods. Nondestructive testing has a rich history
of use across multiple industries. It is commonly used to assure manufac-
turing quality and is used extensively in the sustainment of deployed assets,
including aerospace, power, oil and gas, and transportation. In addition,
inspections testing is common practice and provides a key input in the safe

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 863
ME FEATURE w
x opportunities for nde
management of assets across all industries. For the
US Air Force, nondestructive inspection capability for
safety-of-flight critical locations is one of approxi-
mately seven inputs required for risk management
(safety) of the structural integrity of aircraft managed
by damage tolerance (Lindgren, 2007). With the
broad-based use of NDT, a reasonable question to ask
is why there is the emphasis on characterization in
this issue. A direct answer is there is a growing need
across multiple industries to move beyond detection,
and better support both manufacturing quality
assurance and sustainment with quantitative evalua-
tion of the material state.
Therefore, the scope of this article is to review
some definitions of terminology as the same word(s)
can often be used to define different things. For
example, industrial users of NDT and academic
researchers in the same field can infer different
meanings from the same words. This will be followed
by a summary of some of the main technical chal-
lenges to realize quantitative characterization from the
author’s perspective. This list may not be comprehen-
sive and any feedback on these challenges is always
welcomed. The final section of this introductory article
will identify possible opportunities for research, devel-
opment, and engineering to realize the desired capa-
bility. Again, this list is not to be construed as
comprehensive, but reflects the opinion of the author
and areas that are identified as having significant
impact on addressing the identified challenges.
Definitions
To ensure there is a common set of meanings to the
words commonly used when addressing quantitative
characterization via NDT-based methods, the following
terms are discussed in greater detail. As a definition,
inversion refers to “the process whereby one starts
with measured data and then estimates the size (or
other characteristics) of a discontinuity” (Thompson,
2008). This is the process by which an NDT signal is
converted to information regarding a discontinuity or a
material property, and is what makes quantitative
characterization possible.
This capability has led to a new term in the field of
NDT called materials state awareness (MSA). A recent
workshop was held under the auspices of the Defense
Materials Manufacturing and Infrastructure Standing
Committee of the National Materials and
Manufacturing Board in the Division on Engineering
and Physical Sciences of the National Academies of
Sciences. The rapporteur’s summary of the workshop
was published by the National Academies Press
(Katt, 2016), and MSA was defined in its announce-
ment of the workshop as “digitally enabled reliable
864 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
nondestructive quantitative materials/damage charac-
terization regardless of scale.”
From this author’s perspective, each of the terms
has a specific function in this definition. Digitally
enabled refers to the need for all the possible data
from an NDT instrument to enable the inversion to
occur. This includes the use of raw data, whether it is
radio frequency waveforms for ultrasound, impedance
planes for eddy current, or thermal diffusion values for
infrared imaging. Digital measurements enable the
“use of advanced data processing and analytics in the
analysis of the data to assist in determining the char-
acteristics of the discontinuity or material properties
(states) of interest” (Katt, 2016).
Reliable refers to the use of statistical metrics to
measure accuracy. Even if the characterization process
resorts to the use of pixel counting or other simplified
methods to size a discontinuity, there needs to be a
metric that provides the end user with an estimate of
the accuracy of the result. For detection there is the
well-known probability of detection (POD) evaluation
as noted in other work (US Department of Defense,
2009). POD evaluation has become the accepted
means to determine the statistical capability of any
NDT procedure when the objective is to detect a
discontinuity. Similar metrics using designed experi-
ments and incorporating best practices from statistics
have also been defined for nondestructive evaluation
(NDE) characterization (Medina et al., 2011).
The next few words in the MSA definition are more
familiar to the NDT community. Nondestructive is self-
evident to our community, but needs to be reinforced
with those who would use the outcome of the charac-
terization process. Quantitative refers to the numerical
nature of an output that is more than an indication of
something being present or not present.
Materials/damage indicates that the discontinuity
being characterized can be more than a typical discon-
tinuity found using NDT methods. NDT measurements
have the power to perform materials level characteri-
zation. Representative examples include the measure-
ment of elastic moduli from ultrasound and
conductivity from eddy current. However, research in
the capabilities of NDT to do not only direct measure-
ments, but also more indirect measurements through
advanced interpretation methods, enables features
like measurement of microstructure of metals (Panetta
et al., 2014) and ply-by-ply features in fiber reinforced
composites to be determined (Nelson and Smith,
2014).
The word characterization is the theme of this
special issue and indicates the objective is to expand
the capability of NDT beyond detection of discontinu-
ities to the ultimate capability of accurately identifying
the nature of the discontinuity. This is a large change
in the overall perspective of value that NDT can bring
to new materials development and to the life manage-
ment of structures, increasing the return on the cost of
performing NDT to the end users.
Lastly, the definition includes the phrase
regardless of scale. This is to confirm that NDT
measurements span a broad range of length scales,
such as fourier transform infrared spectroscopy (FTIR),
which can detect changes in materials at an atomic
level (Banks et al., 2016), to visual detection methods
which are used for macroscale features, typically
greater than a few centimeters. Another representative
example of how advanced research can expand
beyond generally accepted capabilities can be found
in ultrasound. It is common to consider that the limits
there is a need to expand research and
development of NDT-based quantitative
characterization to address all possible factors
of ultrasound to detect features are on the order of
“a wavelength or larger” of the ultrasonic signal.
However, developments in signal processing have
shown the capability to surpass this limit (Simonetti,
2006).
Technical Challenges for Quantitative
Characterization
In specialized applications of NDT-based methods,
preliminary measures of quantification can be made.
However, these applications are for parts with special-
ized geometry and/or certain material types. Even in
these scenarios, the ability to place a metric on the
accuracy of the characterization is limited. Thus, when
considering the concept of characterization, there is
need to include not just the parameter being quanti-
fied, such as length and depth in the case of a fatigue
crack in a metallic structure, but also a statistical
metric of accuracy. For NDT, this is true when a signal,
such as a radio frequency waveform in the case of
ultrasound, is used to determine material features,
such as microstructure. In these scenarios, the quanti-
tative characterization extends beyond pixel counting
in an image to additional mathematical processes to
enable the inversion of the NDT signal into information
regarding the material and/or discontinuity.
Therefore, there is a need to expand research and
development of NDT-based quantitative characteriza-
tion to address all possible factors influencing the
response when assessing a feature of interest. Most
previous research efforts performed to explore the
possibility of characterization used a simple experi-
mental approach of testing one factor at a time in
laboratory environments. Typically, these measure-
ments occur by recording signals before and after the
single parameter change, such as a change in a
residual stress state or the insertion of a representa-
tive damage state. In the case of the atomic level
measurements mentioned in the previous section, the
change was limited to thermal exposure for differing
time increments of the same ceramic test coupon. NDT
literature is full of such successful measurements, and
the knowledge they provide toward understanding the
effect of materials change on the NDT signal is
valuable.
However, for most NDT measurements, there are
many factors that affect the response of the NDT signal
in addition to the primary feature of interest. Examples
include representative changes in the measurement
system as applied outside a laboratory environment,
the geometry and assembly of an actual system (as
opposed to a coupon in the laboratory), and the
complexity and variability of the feature of interest.
This latter factor can be especially challenging when
considering the scale of the feature of interest. If it is a
damage state, the variability will be a function of the
operational conditions of the structure in question. For
many applications, especially in military aviation, the
actual use profile can migrate away from the design
requirements for the system. For microscale structural
features, small perturbations in any heat treatment
and/or loading condition can introduce a change in
the response of the NDT signal, confounding the rela-
tionship between measurement and parameter of
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 865
ME FEATURE w
x opportunities for nde
X Y
Layers Z Z
a
Corner crack
Probe
A. NDT technique (measurement system):
1. NDT method
2. Transducer/probe design
3. Contact condition with part (direct, immersion)
4. Scan plan (directions, resolution, orientation)
B. Part geometry, material, and condition: C.
1. Layer material, number, and thickness (shims)
2. Outer layer surface condition (paint, corrosion)
3. Fastener material/type/head condition
4. Hole geometry (oblong, off-angled, surface conditions,
scratches, chatter, tool marks)
5. Fastener hole fit (asymmetric fit, irregular contact
conditions/loading, sealant)
6. Gaps/sealant between layers (aging)
7. Presence of metal shavings
8. Bushings, residual stress around holes
9. Proximity of adjacent fasteners and edges
10. Presence and condition of repairs
Figure 1. Parameters that can influence the nondestructive testing signal from a fatigue crack in a two-layer structure. Labels X, Y, and Z show
orientation; a = crack length; b = crack height.
interest. Figure 1 is an example that shows the range
of factors that can affect a relatively simple ultrasonic
test of a two-layered structure of a representative
aircraft structure (Lindgren, 2007).
For these reasons, inversion from the signal
to quantitative metrics of interest is frequently an
ill-posed inverse problem; more factors affect the
measurement than can be accounted for in the meas-
urement by itself. This is especially true when the
inversion is used for signal responses from complex
structures or complex features of interest. This
challenge opens a whole new area of need for the
future state of NDT research and development, namely
the introduction of mathematical methods to address
uncertainty quantification when determining statistical
accuracy of characterization. It also integrates addi-
tional advanced mathematical methods, such as
sensitivity studies, at a much broader scale than
866 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Fastener site X
θ
Y
Corner crack
Discontinuity characteristics:
1. Discontinuity number (number of cracks per site)
2. Discontinuity type (cracks, electrical discharge
machining notches)
3. Discontinuity location (layer, location in layer–surface,
midbore, eyebrow cracks)
4. Discontinuity orientation (around fastener site, skew
angle from normal)
5. Discontinuity dimensions (length, aspect ratio)
6. Material within discontinuity (sealant, paint, fluids)
7. Discontinuity morphology (regular, irregular)
8. Discontinuity conditions at crack faces (contact
conditions, residual stress)
previously attempted. To optimize these studies,
statistical methods to determine optimal number of
test articles is needed and is known as design of
experiments (DOE). For those immersed in various
statistical methods, such as Six Sigma, this is not new.
However, the use of these approaches is less common
to most NDT research efforts.
When properly designed, the range of experimen-
tation can be extremely extensive. Factors that need to
be addressed include the equipment capabilities,
such as electronics, sensors, and coupling between
the sensor and the material being interrogated. These
are common parameters considered in most labora-
tory experiments as they are easy to understand and
quantify. However, additional parameters must be
addressed to include both features of the item being
inspected and all aspects of the discontinuity of
interest that influence NDT measurements. For many
research and development should be
encouraged to realize the ability to quantify
material properties and/or discontinuities
using NDT-based methods.
research engineers, the full capture of all the key
attributes of the item being assessed is a challenge.
Part-to-part variability due to manufacturing, mainte-
nance, modification, repair, and/or use can have a
significant effect on the response of the NDT signal. In
addition, factors in the discontinuity can introduce a
significant amount of variability into the response,
such as the number, type, orientation, and
morphology.
The magnitude and range of variables makes an
empirical approach, using only experimental data,
quite challenging in terms of generating and assessing
a large number of samples to enable the effects of all
these parameters to be determined, even when using
a proper DOE approach. The emerging modeling capa-
bility for NDT, even at a relatively slow pace, has
improved significantly over the past 10 years to allow
model-based sensitivity studies to be performed to
assess the impact of these variables on the measure-
ment. In some cases, the impact could be relatively
small and the variable can be discounted. In addition,
with this approach, the need for error propagation in
models needs to be carefully captured. Once again,
this is a new mathematical construct to be integrated
as a standard approach when performing these types
of sensitivity studies.
As can be seen from this discussion, the integra-
tion of capabilities from the applied mathematics
community, which includes computational
modeling/simulation, statistical/regression analysis,
plus advanced analytics, represents a new frontier in
NDT. Using a very broad brush, it is the opinion of the
author that previous attempts to realize quantitative
characterization have taken nibbles at this, but that the
full integration of all the capabilities that can be brought
to bear by the use of applied mathematics has been
only initiated in the current research efforts within the
sponsoring agencies. When these mathematical
methods are coupled with the maturing modeling
capabilities, the rate of development of quantitative
characterization with statistical metrics will accelerate.
A simplistic representation of how all these items fit
together is shown in Figure 2 (Lindgren, 2016).
NDT damage/ 3D Signal analysis
materials representation and uncertainty
characterization and validation quantification
Model-driven quantitative representation of
material/damage state with statistical metrics
Figure 2. Simplistic visualization of integrating nondestructive testing with
analytics to realize quantitative characterization of materials/damage regardless
of scale.
Opportunities for Quantitative
Characterization
As mentioned in the introduction, NDT has a rich
history of use across multiple industries. As indicated
in previous sections, the realization of quantitative
characterization is possible with the ability to address
current technical hurdles. However, for this to occur,
there has to be a reason to need quantitative charac-
terization using NDT-based techniques. Several
representative examples that follow illustrate the
need for a quick, effective, and nondestructive
method to characterize the material in question,
not just examine it for the presence of discontinuities.
It is important to note that this list is not inclusive, but
is intended to provide a few representative examples
where this capability would have significant benefit.
This provides additional rationale for why research
and development should be encouraged to realize
the ability to quantify material properties and/or
discontinuities using NDT-based methods.
A megatrend in manufacturing over the past few
years is additive manufacturing. The promise of
additive manufacturing is the ability to build small
batches of critical, high value parts in a fast and cost-
effective manner, plus include the ability to tailor the
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 867
ME FEATURE w
x opportunities for nde
behavior of a subsection of a component to meet
specific performance criteria. This can be as simple as
a localized build-up of material to complex designs as
a tortuous geometry with variation in material proper-
ties within part segments. To realize this potential,
NDT will have to be engaged to avoid current practices
of extensive testing to qualify a material for a specific
application. For smaller lot production of a structural
critical component, the desired properties will have to
be confirmed in addition to using NDT to ensure
discontinuities greater than the rejectable size are not
present, especially if the custom configuration is used
to shave safety margins in the design of the
component.
In current practice for additive manufacturing
parts, much of this is performed via computed tomog-
raphy. However, if specific fatigue parameters are
sought, it is likely that quantitative information that
goes beyond the capability of current computed
tomography systems will be needed. Similar scenarios
exist for all types of specialized processes of material.
For high temperature metals, specific microstructure is
required to ensure performance at elevated tempera-
tures. For fiber reinforced composites, specific configu-
rations of the fibers in the matrix material are part of
the design process to deliver the desired performance
of the component. Thus, there is a need for characteri-
zation to ensure the quality in these high-value, small
lot production runs.
In addition, initiatives such as the digital thread
and digital twin (Kobryn, 2017) are pathfinders for
how systems can be managed in the future. The
digital thread provides a means to track all digital
information regarding the manufacturing and sustain-
ment of a component and system, including material
properties and disposition processes for any variance,
as in a parameter that deviates from the original
design requirement, from the original design parame-
ters. For high-value assets, the review of nonconfor-
mance, such as an NDT indication, occurs on a
periodic basis and if the variance is permitted, current
practice does not capture this variance. Within the US
Air Force, this typically occurs as part of the Materials
Review Board.
In addition, to refine life management of such
assets, there is a growing need to capture the quantita-
tive magnitude of the variance between the design and
the as-manufactured and, ultimately, the as-maintained
state. As this occurs with the article of interest being
placed in service, the preferred method to capture this
variance is NDT. Similarly, the digital twin concept
provides a digital equivalent of a system in its as-is
state and then exercises the physics in the digital twin
model through various use scenarios to determine
868 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
how it will perform and/or forecast possible emerging
maintenance issues. To make this possible, the as-is
material state needs to have quantitative values as
inputs. This includes any damage present. This entire
collection of attributes is called damage state
awareness. Once again, this requires reliable charac-
terization using NDT. Similar capabilities could be
needed to characterize repairs once damage is
detected and to ensure no additional damage occurs
due to the repair process.
Another concept to optimize maintenance actions
on a system is called condition-based maintenance
(CBM). While the digital twin uses the current state to
forecast future states, CBM uses characterization of
the current state to tailor maintenance processes to
address only those issues that require action. As an
analogy, this is comparable to modern automobiles
that have maintenance reminders that include
defining when to change oil as a function of driving
style, temperature, and overall engine loads. With
these monitoring methods, oil changes can now occur
over a broad range of mileage, even in excess of
16 093 km (10 000 miles). This is a significant change
from previous practice, where the recommended
practice was to change oil every 4828 km (3000 miles)
with no consideration on how the car had been used.
While initial CBM practices consider factors such as
loads, usage, and environment, future refinements of
this concept requires NDT to provide quantitative infor-
mation regarding a material parameter and/or an indi-
cation to determine the required disposition process, if
any, based on the condition of the component.
Feasibility and Next Steps for Quantitative
Characterization
There are institutions where development of the capa-
bility to use NDT for characterization with statistical
metrics is underway. The US Air Force Research
Laboratory (AFRL) is supporting research in this area
through in-house and contracted work. Initial efforts in
this area are focused on turbine engine components
as these have less geometric variability than typical
structural components (Aldrin et al., 2016a). This
builds on previous work sponsored by AFRL that
demonstrated the feasibility of using model-based
methods to characterize discontinuities (Aldrin et al.,
2010). In this particular study, the object was to size
the depth of corrosion pits in the second layer of a
representative two-layered structure. The results
indicated that the depth could be determined using a
multifrequency eddy current method to an accuracy of
3.5%. The outcomes of these initial studies indicate
that model-based methods to characterize features in
representative components is possible.
 Next steps in this development are to continue
adding complexity and other confounding factors that
make inversion so challenging, even after decades of
research in this area. The capability to perform
inversion can be facilitated by constraining sources of
confounding factors, such as use of automation to
assist in manual probe placement (OC Robotics,
2017). Another capability assists in the review of large
data sets to minimize inspector fatigue in their review
(Aldrin et al., 2016b). In addition, the use of validated
models, which are becoming more prevalent, and the
integration of advanced mathematical methods will
enable the promise of characterization with statistical
metrics via NDT to become a reality.
Thus, in summary, there is a strong potential for
the use of NDT-based methods to quantitatively char-
acterize materials properties and/or discontinuities in
materials. This has led to a new term in the field of
NDT called materials state awareness and its defini-
tion reflects this potential regardless of the scale of
the feature and/or property of interest. While it is
recognized there are many hurdles that must be
addressed before this capability is realized, there are
multiple examples to justify why this area of research
needs to be pursued. Initial feasibility studies have
shown there is potential to realize this capability.
However, in doing so, it is important to recall that
multiple aspects of NDT technology should be brought
to bear on the problem to assist in the constraint or
even the elimination of confounding factors that could
inhibit the use of NDT to quantify materials and/or
discontinuities. w x
AUTHOR
Eric A. Lindgren: M.S., Ph.D., (937) 255-6994; e-mail
Eric.Lindgren@us.af.mil.
REFERENCES
Aldrin, J.C., H.A. Sabbagh, R.K. Murphy, E.H. Sabbagh, and
J.S. Knopp, “Sensitivity Analysis of Inverse Methods in Eddy
Current Pit Characterization,” Review of Progress in QNDE,
Vol. 29, AIP, 2010, pp. 771–718.
Aldrin, John C., Eric B. Shell, Erin K. Oneida, Harold A.
Sabbagh, Elias Sabbagh, R. Kim Murphy, Siamack
Mazdiyasni, and Eric A. Lindgren, “Model-based Inverse
Methods for Sizing Surface-breaking Discontinuities with
Eddy Current Probe Variability,” 42nd Annual Review of
Progress in QNDE: Incorporating the 6th European-
American Workshop on Reliability of NDE, Vol. 1706,
published online 2016a.
Aldrin, John C., David S. Forsyth, John T. and Welter, “Design
and Demonstration of Automated Data Analysis Algorithms
for Ultrasonic Inspection of Complex Composite Panels with
Bonds,” 42nd Annual Review of Progress in Quantitative
Nondestructive Evaluation: Incorporating the 6th European-
American Workshop on Reliability of NDE, Minneapolis,
Minnesota, 26–31 July 2015, Vol. 1706, AIP Conference
Proceedings, published online 2016b.
Banks, H.T., Jared Catenacci, and Amanda Criner, “Quanti-
fying the Degradation in Thermally Treated Ceramic Matrix
Composites,” International Journal of Applied Electromag-
netics and Mechanics, Vol. 52, No. 1–2, 2016, pp. 3–24.
Katt, Robert J., National Academies of Sciences, Engi-
neering, and Medicine, Applying Materials State Awareness
to Condition-based Maintenance and System Life Cycle
Management, National Academies Press, Washington, D.C.,
2016.
Kobryn, Pamela A. “Digital Thread and Sustainment,” avail-
able at http://meetingdata.utcdayton.com/agenda/viewab-
stract.asp?aid=41933&ID=airworthiness201776816699.
Lindgren, E.A., J.S. Knopp, J.C. Aldrin, G.J. Steffes, and C.F.
Buynak, “Aging Aircraft NDE: Capabilities, Challenges, And
Opportunities,” Review of Progress in QNDE, Vol. 26, AIP,
2007, pp. 1731–1738.
Lindgren, Eric A., “US Air Force Perspectives on Validated
NDE: Past, Present, and Future,” 42nd Annual Review of
Progress in Quantitative Nondestructive Evaluation: Incor-
porating the 6th European-American Workshop on Relia-
bility of NDE, Minneapolis, Minnesota, 26–31 July 2015,
AIP Conference Proceedings, Vol. 1706, published online
2016.
Medina, E. A., Aldrin, J. C., Santiago, J., Lindgren, E. A.,
Buynak, C. F., Knopp, J. S., “Demonstration for Reliability
Assessment of Structural Health Monitoring Systems Incor-
porating Model-assisted Probability of Detection
Approach,” Proceedings of the 7th International Workshop
on Structural Health Monitoring, Ed. F.-K. Chang, Stanford,
California, September 13-15, 2011, p 2460.
Nelson, Luke J. and Robert A. Smith, “Three-dimensional
Fibre-Orientation Characterisation in Monolithic Carbon-
Fibre Composites,” 11th European Conference on Non-
Destructive Testing (ECNDT 2014), Prague, Czech Republic,
6–10 October 2014, published on Ndt.net.
OC Robotics, “RANDE - Non-destructive Inspection,”
accessed on 10 May 2017 from
www.ocrobotics.com/rande—nondestructive-inspection/.
Panetta, Paul, Leslie Bland, Maureen Tracy, and Waled
Hassan, “Ultrasonic Backscattering Measurements of Grain
Size in Metal Alloys,” TMS2014 Annual Meeting Supple-
mental Proceedings, TMS (The Minerals, Metals & Materials
Society), 2014.
Simonetti, F., “Multiple Scattering: The Key to Unravel the
Subwavelength World from the Far-Field Pattern of a Scat-
tered Wave,” Physical Review E, Vol. 73, No. 3, 2006, p.
036619.
Thompson, R. Bruce, “Back to Basics: The Role of Model
Based Inversion,” Materials Evaluation, Vol. 66, No. 7,
2008, pp. 707–712.
US Department of Defense, Handbook, Nondestructive Eval-
uation System Reliability Assessment, MIL-HDBK-1823A,
April 7 2009.
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 869
 ME FEATURE w
x

CE
by John C. Aldrin and David S. Forsyth

T
he focus of this article is how to properly
evaluate the capability of nondestructive
evaluation (NDE) techniques for materials
characterization. This article addresses the
evaluation of characterization error (CE) for the proper-
ties under test, such as crack dimensions. The article
highlights the unique challenges in evaluating CE with
respect to best practices for probability of detection
(POD) evaluation, and examples are presented
demonstrating the process.

Background
As maintenance of many structural components
moves from time-based maintenance to condition-
based maintenance, there is a need for innovative

NDE methods to not simply detect damage but to charac-

terize completely the part state. Other authors have
provided excellent overviews for the need and value

Characterization of NDE characterization (Lindgren, 2007; Thompson,

2008a). Reliable and emerging techniques include
the sizing of cracks in welds using time-of-flight

Capability diffraction (TOFD) methods, the use of model-based

inversion with eddy current nondestructive testing
(NDT) for sizing cracks in aircraft propulsion compo-

Evaluation nents, the localization of impact damage sites

through guided wave ultrasound and/or acoustic
emission sensors, the evaluation of the material
state of turbine blades using resonance ultrasound
spectroscopy (RUS), and the assessment of environ-
mental degradation of composite materials using
fourier transform infrared spectroscopy (FTIR). Sev-
eral of these characterization techniques are dis-
cussed in greater detail in the technical articles in
this special Materials Evaluation Technical Focus Issue.

870 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Evaluation
To test the reliability of NDE characterization capa-
bilities, validation procedures are needed to ensure
the accuracy of material property characterization, and
damage sizing and localization techniques.
Comprehensive POD evaluation procedures (US
Department of Defense, 2009; Annis and Gandossi,
2010; ASTM, 2015) have been developed to validate
the reliability of NDE techniques and are used by the
US Air Force (USAF) in support of Aircraft Structural
Integrity Programs. Although standard practices exist
for the evaluation of measurement system capability
(ASTM, 2011), the application tends to be more
complex for assessing NDE characterization perform-
ance. The challenge is due to the often-complex
underlying relationships associated with characteriza-
tion and the need to simultaneously address
parameter variability and uncertainty present in relia-
bility studies. Past programs and reports have intro-
duced procedures for evaluating NDE sizing capability,
specifically addressing discontinuities in welds
(Nordtest, 1998; Lozev et al., 2005; Spencer et al.,
2011; Ducharme et al., 2012; Todorov et al., 2014)
and corrosion in aircraft structures (Hoppe et al.,
2002; Forsyth and Lepine, 2002; Bode et al., 2002;
Liao et al., 2003). However, there are some
outstanding issues with the current practice for the
quantitative evaluation of sizing capability. One
frequently cited metric is the calculation of the 95%
safety limit against undersizing bound for quantifying
sizing performance (Nordtest, 1998). Important
assumptions like linearity in the response and
constant variance with changes in discontinuity size
must be verified before such a metric can be
generated. In addition, the simplistic character of the
bound from an ordinary least-squares fit does not
adequately address the true variation of the bound for
non-ideal distributions of discontinuities and limited
sample numbers. A more comprehensive process is
needed to ensure that the bounds on sizing perform-
ance being reported from such studies are valid.
The goal of this article is to provide guidance on
how properly to perform an NDE CE evaluation. An
overview of NDE CE evaluation is first presented that
follows certain practices of the â-versus-a model eval-
uation for POD assessment (US Department of
Defense, 2009; Aldrin et al., 2015; Aldrin et al.,
2016). This paper will also cover the unique chal-
lenges in evaluating NDE CE with respect to best
practices for POD evaluation. Several examples are
presented demonstrating the process, and supple-
mental resources are included throughout the paper
to provide guidance for future applications.
Overview
CE Evaluation Follows Practices Found in â-Versus-a
POD Analysis
From the perspective of quantifying the reliability of
NDE characterization techniques, there is the need to
evaluate the accuracy in estimating the damage or
^
material state ( ) with respect to the known condition
() under test. (The Greek character, , will be used
throughout this paper instead of a, to avoid confusion.
The parameter â, associated with NDE measurements
in POD studies, is rarely considered a direct estimate
of the discontinuity size.) Thus, the evaluation of the
CE must be performed for all critical characteristics
under test. This evaluation can be represented by the
following expression, considering N characteristics:
(1) (
CEi = θˆ i − θ i )
where
i = 1,…,N.
These characteristics may include, for example,
(1) indication location for a structural health moni-
toring system (SHM), (2) crack dimensions, and/or
(3) the material state (or properties). Damage and
material state estimates with proper error bounds are
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 871
ME FEATURE w
x nde characterization capability
needed for condition-based maintenance and
prognosis programs to help evaluate remaining life
and determine necessary maintenance actions. This
problem of evaluating CE with the appropriate bounds
as a function of a critical parameter, such as disconti-
nuity size, is shown in Figure 1. Now, the design and
statistical evaluation for such a study is generally
similar to the â-versus-a procedure for the evaluation
of the relationship between the NDE measurement (â)
and the critical discontinuity size (a). Because CE eval-
uation is analogous to â-versus-a POD analysis, this
enables our community to leverage decades of experi-
ence in designing POD studies for NDE characteriza-
tion capability evaluation. Other research has provided
excellent resources for performing such studies (US
Department of Defense, 2009; Annis and Gandossi,
2010; ASTM, 2015).
Address Complex Multi-Dimensional Sizing Model
There are several unique challenges concerning NDE
characterization evaluation. First, sizing often includes
more than a single characteristic of the discontinuity.
NDE and SHM systems may measure multiple quanti-
ties that are related to damage size and location eval-
uation. As well, for sizing of surface-breaking cracks
and test notches in nickel-based super alloys, the
crack depth, length, and opening (width) parameters
will all play a significant role on the eddy current
response, influencing sizing performance.
Second, from a statistical perspective, the
challenge here is to develop a multidimensional
CE model with confidence bounds, CEi(j), where
i = 1,…,N characteristics being estimated and
j = 1,…,M the key factors that control these measure-
ments. Depending on the characterization application,
Characterization error (CE)
Key characteristic (θ)
^
Figure 1. The evaluation of characterization error (CE =  – ) with respect to a
key characteristic () is analogous to evaluating the relationship between an NDE
measurement (â) and the critical discontinuity size (a).
872 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
sometimes multiple dimensions are needed. NDT
inspectors and engineering expertise of the measure-
ment system are critical to defining a quality study
that addresses all key factors including pertinent
parameter interactions. At the same time, the study
must be reasonable in scope, such that it can be
performed using a practical number of specimens and
test conditions.
Need for Full Knowledge of Discontinuity or Material
State
One critical point for NDE characterization performance
studies is that the discontinuity or material property
states must be very well understood. Uncertainty of the
characteristics under test must be much less than the
expected performance of the NDE characterization
technique. This may seem to be obvious. But, recent
experience indicates that many existing crack POD sets
may not be adequate for sizing demonstrations. For POD
sets with semielliptical surface-breaking cracks, crack
length is assessed using microscopy, while crack depth
is typically calculated based on an empirical relationship
between length and depth evaluated from a limited set
of break-open specimens. Historical POD data for eddy
current inspections were analyzed by another author
(Hoppe, 2010) using alternative statistical models incor-
porating both crack length and depth, indicating the
ability to better separate system and crack factors influ-
encing POD. Crack depth was found to be a large source
of the variability in â-versus-a POD model fits that only
consider crack length. Another recent study (Shell et al.,
2015) investigated eddy current model-based inversion
predictions for several POD sets of cracks in a nickel-
based alloy. While the length estimates were found to
be fairly accurate, the depth estimates were found to
have significant scatter compared to the assumed
depths of the crack set. Even if the mean relationship
between crack length and depth was perfectly under-
stood, natural manufacturing variability in the depths of
real cracks of similar lengths has been observed in
these sets. While variability and uncertainty in the
discontinuity depth dimensions does impact POD evalu-
ation, such uncertain values in depth cannot be used as
known values for crack sizing demonstrations. Even for
the best specimen sets with aspect ratios (length to
depth) well controlled at 2.5  0.25, this set would still
result in uncertainty in depths of 10% which would
clearly be unacceptable for use as known values if one
is attempting to demonstrate sizing of cracks at the
level of 10% or even 20% error. Here, finer control of
specimen development and some post-study destruc-
tive characterization work will likely be necessary to
properly evaluate crack sizing performance.
Verify Statistical Assumptions
All statistical analyses depend on the assumptions
concerning the relationship between reality (the meas-
urements) and the process being modeled. First and
foremost, for any linear regression performed, one
must verify that the model looks like the data. Plotting
the data and evaluating the quality of the fit is a
necessary step. While this may be self-evident,
verifying that the assumptions hold true is unfortu-
nately neglected in many engineering evaluations. For
evaluating sizing over several parameters, multiple
plots will be required. A test case example of eddy
current sizing of crack length and depth is presented
in Figure 2.
For linear regression, there are five implicit
assumptions that must be satisfied for the resulting
parameter estimates to be accurate:
l The response must be continuous and observable.
l The model must be linear in the parameters.
l The variance must be homoscedastic (having
uniform variance).
l The observations must be uncorrelated (for
example, with respect to time).
of sizing performance for this range could be quite
limited (beyond simply binning such indications into a
small discontinuity category). The goal of this interme-
diate step is to assess where good characterization
results can be achieved, and then perform the evalua-
tion of accuracy within that range. A case study is
introduced here based on the estimation of the
length, depth, and width of surface-breaking cracks
and notches using bolt hole eddy current (BHEC) NDE
(Aldrin et al., 2015). These inversion results were
generated using a synthetic set of NDE response data,
created using monte carlo simulation over a distribu-
tion of the three crack dimensions, with a volume-
integral method model in conjunction with a limited
set of empirical results.
First, weak signals from small discontinuities are
often masked by measurement noise. In a POD evalu-
ation, this characteristic in the data is called left
7
6
A
Estimated length (mm)

l The errors must be normal.

C
If any of these assumptions are not met, the 5
analysis will be incorrect and a more appropriate
4
model is needed. For sizing multiple dimensions with B
potential parameter interactions, covariance error 3
models must be evaluated as well, including the test
for normality across multiple properties. More back- 2
ground on verifying these assumptions in NDE relia-
1
bility studies can be found in other scholarship (Annis
et al., 2015).
0 1 2 3 4 5 6 7
Consider Limits of Characterization Performance (a) Length, known (mm)
During Model Build 3
Before fitting a statistical model for the data, an
important intermediate step is to evaluate the 2.5
Estimated depth (mm)

presence and frequency of several possible classes of C

poor characterization results. There are several 2
reasons for potentially poor characterization results:
l weak signals from small discontinuities masked by 1.5
measurement noise,
l saturated signals or conditions exceeding the 1
inversion algorithm design,
l ill-posed inversion problems leading to clustering in 0.5
local minima, missing the global (correct) solution, A,B
l random poor characterization performance due to a
0 1 2 3 4 5
process failure independent of discontinuity size.
(b) Depth, known (mm)
Even for the best sizing techniques, some poor sizing
results can occur. Clearly, if an NDT technique is not
able to distinguish certain smaller discontinuities from Figure 2. Sizing results for bolt hole error characterization test data for (a) crack
other possible false call indications, then the reliability length and (b) crack depth.

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 873
ME FEATURE w
x nde characterization capability

censoring. In a proper POD evaluation, these data are
retained and statistical methods that accommodate
left-censored data are used (US Department of
Defense, 2009; ASTM, 2015). If detected, such NDE
measurements will undoubtedly be difficult to size.
However, in practice, operators typically do not require
the sizing of discontinuities with weak indications well
under the detection threshold. So, in a CE evaluation,
0.8
D
0.6
0.4
CE, depth (mm)
such data can be practically separated from the
primary error analysis. However, any data removed
must be clearly reported. For this case study, inversion
results for clusters A and B, as shown in Figures 2a
and 2b, are examples of poor sizing results associated
with their small discontinuity size. The specific
problem here is that by attempting to include such
small discontinuity results in a combined error
analysis, the statistical model will likely be poor and
could violate certain assumptions (like variance being
homoscedastic, or error being normal). For this case
study, crack sizes with lengths below 1.27 mm or
depths below 0.46 mm were removed from the
primary error evaluation and clearly documented

0.2 (Aldrin et al., 2015).

0 Second, strong signals from very large discontinu-
–0.2 ities can also present several issues for NDE sizing
techniques. For example, eddy current NDE equipment
–0.4
often has limits on the dynamic range and thus large
–0.6 discontinuity responses can sometimes exceed the
E
–0.8 measurement system and become saturated. Such
0 0.5 1 1.5 2 2.5
saturated signals can lead to underestimates of the
(a) Depth, known (mm)
true discontinuity size. Similarly, the inversion
0.5 algorithm may not be designed to invert discontinu-
0.4 ities beyond a certain expected size. This case is
0.3 demonstrated for the BHEC crack inversion study
shown in Figure 2b (see cluster C), where the crack
CE, length (mm)

0.2
0.1 depth inversion algorithm was constrained to a
0 maximum length of 6.35 mm and depth of 2.54 mm.
–0.1 In a POD evaluation, the presence of saturated data in
–0.2 the evaluation is called right censoring. Statistical
–0.3 methods that accommodate right-censored data are
–0.4 used (US Department of Defense, 2009; ASTM, 2015).
–0.5 In practice, it is possible that operators do not require
1 2 3 4 5 6 7 exact sizing when certain discontinuities become
(b) Length, known (mm) exceptionally large, only that the discontinuity satisfies
x 10–3 some large discontinuity classification. For example, if
6
a crack above a certain large size is found, the part
4 will simply be replaced. So, in a CE evaluation, such
clusters of poor classification results due to large
CE, width (mm)

2 discontinuity size can be separated from the primary

error evaluation and whichever data are removed
0
should be clearly reported.
At this stage, the primary CE results can be
–2
evaluated with left and right limits as shown in Figures
3a to 3c. For the BHEC inversion case study problem,
known crack sizes with lengths below 1.27 mm and
–4
0 0.005 0.01 0.015 0.02 0.025 0.03 above 6.35 mm or depths below 0.46 mm and crack
(c) Width, known (mm) depths above 2.54 mm were separated from the
primary error analysis. Next, it is important to check
the data and evaluate if and where any secondary
Figure 3. Characterization error of inversion results for (a) crack depth, (b) crack
length, and (c) crack width. Plots include a linear model fit (green line) with clusters or trends in the data are present. Certain chal-
confidence bounds (blue lines) and corresponding prediction bounds (red lines). lenging inverse problems are known to have issues
with being ill-posed, where the results are prone to

874 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
For the characterization error results,
it is important to look at the data
and identify any potential outliers
clustering at a poor solution due to getting caught in
local minima. Case B in Figure 2 is an example where
signals near the noise threshold are difficult to size,
underestimating depth while overestimating length. If
such data does not fall into the category of the left or
right limits, then more sophisticated statistical models
may be needed to address such results.
Lastly, for some NDE inspections that are highly
dependent upon human factors or possibly subject to
rare events, random poor characterization perform-
ance can sometimes arise. Such instances are
assumed to be due to a process failure that is consid-
ered independent of discontinuity size. An analogous
condition in POD evaluation is referred to as the
random missed call rate. Certain statistical methods
have been developed to evaluate the rate of random
missed calls during a POD study (Moore and Spencer,
2001). Note, properly addressing a random missed
call rate can be difficult and will often require a large
quantity of data in order to achieve a quality evalua-
tion. For the CE results, it is important to look at the
data and identify any potential outliers that could be
associated with such events, possibly independent of
discontinuity size. Possible indications for crack depth
sizing are shown in Figure 3a. There are a few indica-
tions, D and E, that appear to be outside the main
scatter of data. Ideally, further analysis is needed to
determine if exceptionally poor data is the source of
these results, or if it is simply due to having limited
data samples and should be included in the primary
statistical model fit. For this example, indications
D and E were included in the primary statistical
model fit.
Evaluate Characterization Error Model with Confi-
dence and Prediction Bounds
Once the primary data set is selected for the CE model
evaluation, the appropriate statistical models can be
selected and assessed. The CE vector for this example
parameter estimation problem is given by:
⎡CEd ⎤ ⎡θ − θ known _ d ⎤
⎢ ⎥ ˆ ⎢ est _ d ⎥
(2) CE = ⎢CEl ( )
⎥ = θ − θ = ⎢θ est _ l − θ known _ l ⎥
⎢ ⎥
⎣⎢CE ⎥
w⎦ ⎣θ est _ w − θ known _ w ⎦
where
ˆ is the estimated parameter vector,
 is the known parameter vector,
d is the depth,
l is the length,
and w is the width of the crack.
This fit was performed using maximum likelihood
estimation (MLE) in an open source package for statis-
tical computing. The CE model plots in Figure 3
include two sets of bounds, 95% confidence bounds
and 95% prediction bounds. The inner-most bounds
are the confidence bounds on the mean linear model
fit. These bounds are expected to contain the true
relationship in 95 of 100 nominally identical experi-
ments. The outer bounds are prediction bounds on
the individual estimates. Here, the next single obser-
vation is expected to fall within the prediction bounds
95% of the time. In general, the linear model fit
appears to be adequate for the data set presented
here. However, even with this generally well-behaved
data set, there is some observed change in the lower
prediction bounds in Figure 3a and 3b as depth and
length change.
Problem with Single Value Performance Metric
One metric frequently cited is the calculation of the
95% safety limit against undersizing bound (Nordtest,
1998). While there is nothing wrong with working to
evaluate this bound, assuming that a single value
represents this bound can often be in error. There are
several important assumptions like linearity in the
response and constant variance in sizing error with
changes in discontinuity size that must be verified
before using this measure. In addition, the simplistic
character of the bound does not adequately address
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 875
ME FEATURE w
x nde characterization capability

the true variation of the bounds, especially when
discontinuities are not well distributed over the
parameter space (for example, over varying depth) or
there are limited number of samples. A sizing example
using eddy current inspection data with inversion
results for surface cracks in metallic components is
presented in Figure 4a. It is always important to look
at the data, and, clearly, the sizing error is dependent
on discontinuity size. A scatter plot of the CE as a
function of crack depth is show in Figure 4b. Here,
the CE was evaluated with a linear model using MLE
when sizing the crack depth. The CE results shown in
Figure 4b include a linear model fit (blue line) with
corresponding 95% prediction bounds (green line). In
general, the linear model fit appears to be adequate
6 terms of the base unit of measure, while in other
5 This will depend on the customer of the data and how
Estimated depth (mm)
for the limited set of depths and addresses how the
CE is dependent on depth. Upper and lower 95%
bounds are included (purple lines) if one simply
assumes that error does not vary with crack depth.
Since there is a large change in the CE lower bound
with varying crack depth, this simple bound is just not
appropriate. This example demonstrates the need to
take care when attempting to report a single value
that defines the entire lower bound for the safety limit
against undersizing. Operators should not use such a
simple metric for sizing accuracy when it is not appro-
priate.
Evaluating Characteristic Error Bounds with
Non-Constant Variance
One consideration is how to best report the CE results
while also dealing with error variation as a function of
size. There are some cases where error is reported in
instances error might be evaluated on a percent basis.

they plan to use it. One frequent challenge with evalu-

4 ating error on an absolute basis is that variability in
the error of the measurements is likely to grow with
3 the increasing characteristic. Without a natural way of
normalizing the results, for example, on a percent
2 basis, sizing error is more likely to be a function of
size. An example data set for fitting regression error is
presented in Figure 5a, estimating aspect ratio (length
1
to depth) for a set of 15 notches evaluated at both
longitudinal and transverse orientations (see Oneida
0 1 2 3 4 5 et al., 2017).
(a) Depth, known (mm) The goal of performing a regression fit and esti-
3 mating prediction bounds is hindered here by the
violation of the assumption of homoscedasticity, a
nonuniformity of variance in the residuals with varying
2 crack aspect ratio. The assumption of homoscedas-
ticity simplifies the computational treatment, but
CE, depth (mm)

1 serious violations in homoscedasticity will produce

prediction bounds that are incorrect. Instances of non-
constant variance can often be addressed through
0
transformation approaches. Frequently, a log transfor-
mation is used to improve homoscedasticity and
–1 maintain a linear relationship between x and y. As a
general approach, a box cox transform (Box and Cox,
1964), given by the following expression:
–2
1.5 2 2.5 3 3.5 4 4.5
(b)
⎧ y( λ ) − 1
Depth, known (mm) ⎪ i if λ ≠ 0,
(3) y(i λ ) =⎨ λ
⎪ln( y ) if λ = 0,
⎩ i
Figure 4. Sizing example: (a) Eddy current sizing results for crack depth with Y=X
reference line (green line); and b) characterization error linear fit for varying crack
depth (blue line) with prediction bounds (green lines). Horizontal lines represent
can be used through adjusting  to normalize the error
the mean (red line) and bounds on the fit (purple lines), assuming that sizing and generate proper bounds. Characterization error
error does not vary with crack depth. results evaluated using MLE are presented in Figure 5b

876 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
more care, likely associated with more expensive
experimental studies, will be necessary to
properly evaluate characterization capability
for this data set. For the normal (linear scale) data
evaluation, some non-constant variance is observed in
Figure 5a and 5b, increasing with higher aspect ratios.
Prediction bounds using both a linear scale (red lines)
and square root transformation (blue lines) are
included. By addressing the non-constant variance
using the square root transformation, proper predic-
tion bounds are provided that better represent the
sampled data of the experiment.
However, there are some inherent challenges to
using the model-assisted POD evaluation process for
characterization evaluation to reduce sample require-
ments. Unfortunately, there is not a model for sizing
error, only models representing NDE measurements.
As well, one cannot use the same model as part of the
inversion process and for validation purposes. This is
often referred as an inverse crime (Wirgin, 2004).

Discussion on Synthetic Discontinuities 8

To ensure that discontinuity or material property states
are very well understood, more care, likely associated 7
with more expensive experimental studies, will be
6
Estimated aspect ratio

necessary to properly evaluate characterization capa-

bility. Going forward, this might include more specimens 5
being destructively characterized, more expensive char-
4
acteristic verification work through nondestructive
testing (for example, using computed tomography), or 3
manufacturing much higher quality discontinuities with
2
very fine control of the dimensions. For crack sizing
studies, mixtures of cracks that are destructively charac- 1
terized with larger sets of synthetic discontinuities, like
electrical discharge machine or focused ion beam 0 2 4 6 8
notches, are proposed. However, the assumption that Aspect ratio, known
such notches can be surrogates for real cracks in sizing (a)
studies must be properly demonstrated. 1

Discussion on Model-assisted Approaches to 0.5

Support Characterization Studies
The build-up of specimens and experimental testing
CE, aspect ratio

are frequently considered a significant burden in the 0

validation of new inspection techniques. The prepara-
tion of POD samples with fatigue cracks in real parts –0.5
and the process to acquire a statistically significant Prediction bds.,
number of measurements are often very time linear model
consuming and expensive. A model-assisted proba- –1
Prediction bds.,
bility of detection approach for the design and λ = 0.5
execution of POD studies has been proposed to help –1.5
0 2 4 6 8
mitigate validation study costs and to also improve
the POD evaluation by addressing a wider array of (b) Aspect ratio, known
inspection variables (Thompson, 2008b; US
Department of Defense, 2009). Several quality demon- Figure 5. Electrical discharge machine notch aspect ratio estimation results:
strations can be found in prior work (Thompson, (a) with respect to known aspect ratios; (b) with prediction bound using linear
2008b; Aldrin et al., 2009; Dominguez et al., 2012). model (red line) and square root transformation (blue line).

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 877
ME FEATURE w
x nde characterization capability
However, all hope is not lost. For example, since a
physics-based model is often at the heart of a model-
based inversion process, it is expected that the sizing
results should be quite repeatable and may not require
as much data for statistical validation. If the model
based inversion process can compensate for probe vari-
ability and estimate the crack state in terms of all the
key dimension factors that drive the response (Oneida
et al., 2017), the inversion process could theoretically
require less data in estimating the statistical model fit.
The goal here is that the unmodeled characteristics will
be very small when all key factors on the response are
addressed through inversion. Future work is needed to
study different characterization problems and evaluate
the role of sample number on the bounds of NDE CE.
Recommendations and Resources
In this article, an overview of NDE CE evaluation was
presented that follows the framework of the â-versus-a
model evaluation process for probability of detection
(POD) assessment. A CE evaluation should be
performed for all critical characteristics under test.
Because experimental studies and statistical model
building must address the key drivers for variation
with the technique under evaluation, it is very
important to have the appropriate engineering and
statistical expertise involved to ensure all key effects
and interactions are properly addressed.
Tools are available within the NDE community to
help perform these evaluations and verify the model
assumptions. As well, an open-source package for
statistical computing has been used by the authors for
past work on sizing error evaluations (R Core Team,
2017). However, if one finds that the trends in the evalu-
ation are not represented well by conventional linear
regression models, please seek help from key references
on advanced statistics, for example on censored regres-
sion (US Department of Defense, 2009) and nonlinear
regression (Bates and Watts, 1988), or consult a statisti-
cian that practices in the NDE community. w x
AUTHORS
John C. Aldrin: Computational Tools, Gurnee, Illinois,
60031; e-mail aldrin@computationaltools.com.
David S. Forsyth: TRI/Austin, Austin, Texas, 78746; e-mail
dforsyth@tri-austin.com.
ACKNOWLEDGMENTS
The authors would like to acknowledge support for portions
of this work from the Air Force Research Lab (AFRL) under a
SBIR Phase II Contract FA8650-13-C-5180 and under
Research Initiatives for Materials State Sensing (RIMSS). We
are grateful for discussions on this topic with Charles Annis
of Statistical Engineering, Erin Oneida and Eric Shell of
KBRwyle, Jennifer Brown, and Eric Lindgren of the Air Force
Research Laboratory (AFRL/RXCA) Wright-Patterson AFB. R,
an open-source package for statistical computing (R Core
878 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Team, 2017), was used for all of the statistical analyses in
the paper. The proprietary volume-integral software used in
the this project is Victor Technologies VIC-3D eddy current
testing software. Tools to perform various evauations and
verify model assumptions can be found at Eclipse Scientific.
REFERENCES
Aldrin, J.C., C. Annis, H.A. Sabbagh, J. Knopp, and E.A.
Lindgren, “Assessing the Reliability of Nondestructive Eval-
uation Methods for Damage Characterization,” Review of
Progress in QNDE, Vol. 33, AIP, 2014, pp. 1784–1791,
published online 2015.
Aldrin, J.C., C. Annis, H.A. Sabbagh, and E.A. Lindgren,
“Best Practices for Evaluating the Capability of Nondestruc-
tive Evaluation (NDE) and Structural Health Monitoring
(SHM) Techniques for Damage Characterization,” 42nd
Annual Review of Progress in Quantitative Nondestructive
Evaluation: Incorporating the 6th European-American
Workshop on Reliability of NDE, Minneapolis, Minnesota,
26–31 July 2015, AIP Conference Proceedings,
Vol. 1706, published online 2016.
Aldrin, J.C., J.S. Knopp, E.A. Lindgren, and K.V. Jata, “Model-
assisted Probability of Detection (MAPOD) Evaluation for
Eddy Current Inspection of Fastener Sites,” Review of
Progress in QNDE, AIP, Vol. 28, pp. 1784–1791, 2009.
Annis, C. and L. Gandossi, ENIQ TGR Technical Document:
Probability of Detection Curves: Statistical Best Practices,
ENIQ Report No. 41, EUR 24429 EN, JRC Scientific and
Technical Reports, 2010.
Annis, Charles, John C. Aldrin, and Harold A. Sabbagh,
“What’s Missing in Nondestructive Testing Capability
Evaluation?” Materials Evaluation, Vol. 73, No. 1, 2015,
pp. 38–42.
ASTM, ASTM E3023-15: Standard Practice for Probability of
Detection Analysis for â Versus a Data, ASTM International,
West Conshohocken, Pennsylvania, 2015.
ASTM, ASTM E2782-11, Standard Guide for Measurement
Systems Analysis (MSA), ASTM International, West
Conshohocken, Pennsylvania, 2011.
Bates, Douglas M. and Donald G. Watts, Nonlinear Regres-
sion Analysis and Its Applications, Wiley, Hoboken, New
Jersey, 2007.
Bode, M.D., D.M. Ashbaugh, K.L. Boyce, and F.W. Spencer,
“Corrosion Structured Experiment,” Review of Progress in
QNDE, Vol 21, AIP, 2002, pp. 1779–1786.
Box, G.E.P. and D.R. Cox, “An Analysis of Transformations,”
Journal of the Royal Statistical Society, Series B. Vol. 26, No.
2, 1964, pp. 211–252. JSTOR 2984418.
Dominguez, N., V. Feuillard, F. Jenson, and P. Willaume,
“Simulation Assisted POD of a Phased Array Ultrasonic
Inspection in Manufacturing,” Review of Progress in QNDE,
Vol. 31, AIP, 2012, pp. 1765–1772.
Ducharme, P., S. Rigault, I. Strijdonk, N. Feuilly, O. Diligent,
P. Piché, and F. Jacques, “Automated Ultrasonic Phased
Array Inspection of Fatigue Sensitive Riser Girth Welds with
a Weld Overlay Layer of Corrosive Resistant Alloy (CRA),”
NDT.net, September 2012.
Forsyth, D.S., and B.A. Lepine, “Development and Verifica-
tion of NDI for Corrosion Detection and Quantification in
Airframe Structures,” Review of Progress in QNDE, Vol. 21,
AIP, 2002, pp. 1787–1791.
Hoppe, W.C., “A Parametric Study of Eddy Current Response
for Probability of Detection Estimation,” Review of Progress
in QNDE, Vol. 29, AIP, 2010, 1895–1902.
Hoppe, W., J. Pierce, and O. Scott, Automated Corrosion
Detection Program Final Report for 07 April 1997 – 06
October 2001, United States Air Force Report AFRL-MP-WP-
TR-2001-4162, 2002.
Liao, M., D.S. Forsyth, J.P. Komorowski, M. Safizadeh, Z. Liu,
and N. Bellinger, “Risk Analysis of Corrosion Maintenance
Actions in Aircraft Structures,” ICAF 22, 2003—Fatigue of
Aeronautical Structures as an Engineering Challenge, Proceed-
ings of the 22nd Symposium of the International Committee
on Aeronautical Fatigue, Lucern, Switzerland, 5–9 May 2003.
Lindgren, E.A., J.S. Knopp, J.C. Aldrin, G.J. Steffes, C.F.
Buynak, “Aging Aircraft NDE: Capabilities, Challenges, And
Opportunities,” Review of Progress in QNDE, Vol. 26, AIP,
2007, pp. 1731–1738.
Lozev, M.G., D. Hodgkinson, R.L. Spencer, and B.B.
Grimmett, “Validation of Current Approaches for Pipeline
Girth Weld Discontinuity Sizing Accuracy,” Materials
Evaluation, Vol. 63, No.5, 2005, pp 505-510.
Moore, David G. and Floyd W. Spencer, “Interlayer Crack
Detection Results Using Sliding Probe Eddy Current Proce-
dures,” 10th Asia-Pacific Conference on Non-Destructive
Testing, Brisbane, Australia, 17–21 September
2001,published online at Ndt.net.
Nordtest, “Guidelines for NDE Reliability and Descriptions,”
NT TR 394, Approved April 1998.
Oneida, E., E.B. Shell, J.C. Aldrin, H.A. Sabbagh, E.H.
Sabbagh, R.K. Murphy, “Characterizing Surface-Breaking
Cracks through Eddy Current NDE and Model-Based Inver-
sion,” Materials Evaluation, (submitted, 2017).
R Core Team, R: A Language and Environment for Statistical
Computing. R Foundation for Statistical Computing, Vienna,
Austria. URL http://www.R-project.org/, ver. 3.3.3, 2017.
Shell, E.B., J.C. Aldrin, H.A. Sabbagh, E.H. Sabbagh, R.K.
Murphy, S. Mazdiyasni, and E.A. Lindgren, “Demonstration
of Model-Based Inversion of Electromagnetic Signals for
Crack Characterization,” Review of Progress in QNDE, Vol.
34, AIP, 2015, pp. 484–493.
Spencer, R, E. Todorov, P. White, N. Porter, and M. Lozev,
“Advanced Technologies and Methodology for Automated
Ultrasonic Testing Systems Quantification,” DOT DTPH56-
07-T-000002, http://primis.phmsa.dot.gov/matrix/Final
Reports.rdm, EWI Project 50454GTH, 2011.
Thompson, R. Bruce, “A Unified Approach to the Model-
assisted Determination of Probability of Detection,”
Materials Evaluation, Vol. 66, No. 6, pp. 667-673, 2008b.
Thompson, R. Bruce, “Back to Basics: The Role of Model
Based Inversion,” Materials Evaluation, Vol. 66, No. 7,
2008a, pp. 707–712.
Todorov, E., R. Spencer, and M. Lozev, “Methodologies for
Automated Ultrasonic Testing Performance Quantification,”
Materials Evaluation, Vol. 72, No. 76, 2014, pp. 812–824.
US Department of Defense, Handbook, Nondestructive
Evaluation System Reliability Assessment, MIL-HDBK-1823A,
April 7 2009.
Wirgin, Armand, “The Inverse Crime,” arXiv:math–
ph/0401050, submitted on 28 January 2004.
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 879
PRODUCT
Red to Near-IR Enhanced Surface-
Mount Photodiode
Opto Diode Corporation, an ITW company,
introduces the NXIR-5C, a red to near-
infrared (NIR) enhanced photodiode with a
882 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
gallery
circular active area of 5 mm2. With a
spectral response from 320 nm to
1100 nm, the rugged photodiode is
housed in a custom 4.7 mm  4.9 mm
ceramic carrier surface-mount device
(SMD) package. The new SMD joins the
company’s popular NXIR family of products
that is optimized for the near-infrared
spectrum. The NXIR-5C provides superior
responsivity with low reverse bias and high
sensitivity at 0.62 A/W at 850 nm and
0.35 A/W at 1064 nm. It has low dark
current at 1 nA, low capacitance of
5 pico-farads (pF) at 10 volts, and high
shunt resistance greater than 100 M.
The NIR-enhanced detector is specially
designed for laser monitoring, medical
diagnostic equipment, industrial automa-
tion, scientific measurement, and military
applications which require a wide
operating temperature range from –40 °C
to +125 °C. The robust package features
an AR-coated window that provides greater
than 98% transmission. The SMD is
compliant with RoHS and REACH standards
and is available on tape and reel.
Opto Diode Corporation, Camarillo,
California
www.optodiode.com
SPOTLIGHT w
x Radiographic Testing

Imaging Software
Yxlon now provides the ASTM E 2973 digital reference images for aluminum and
magnesium die castings in its imaging software, Image2500 and Image3500, of
X-ray inspection systems including Y.MU2000-D and MU60 AE. The reference
images, which so far have been available only as an image catalog for
radiographic inspections with film, can now be deployed in digital radioscopy for
discontinuity evaluation of aluminum and magnesium alloys in die castings via a
second review monitor. An important enhancement has now been realized with
ASTM E 2973 for aluminum and magnesium die castings to support discontinuity
classifications for these parts and document inspection decisions according to
industry specifications.
Yxlon International, Hamburg, Germany
www.yxlon.com

X-ray Imagers
Detection Technology, Inc., has taken tire inspection intelligence to the next level
by launching the X-Scan U series. The detector series based on the novel digital
platform improves image quality, increases maximum scanning speed, and saves
the overall system cost of industrial in-line and off-line nondestructive testing of
tires. The X-Scan U series is an enhanced product family of the U-shaped X-ray line
cameras developed and optimized specially for high-speed digital tire inspection
utilizing panoramic X-ray sources. The series enables the image quality to meet the
tightest quality requirements by providing robust low noise digital signal,
homogenous data over the detector active area, and the best signal-to-noise ratio
for X-ray imaging. The series has first-class radiation hardness, increasing the
lifetime of the detector up to two times longer. Combined with the excellent
reliability and remote firmware upgrades and diagnostics, the series improves
production efficiency even more and minimizes lifetime cost of X-ray systems.
Detection Technology, Inc., Billerica, Massachusetts
www.deetee.com

Flat Panel Extension for 3D X-ray Microscopes

The new ZEISS FPX flat panel extension for the ZEISS Xradia Versa 500-series of
3D X-ray microscopes delivers large sample, high-throughput scanning with
best in class image quality. Combined with the high resolution of ZEISS Xradia
Versa X-ray microscopes (XRM), the new ZEISS FPX enhances imaging flexibility
and creates workflow efficiencies with an all in one system for industrial
development and academic research. ZEISS Xradia Versa with FPX enables
engineering, development, and researchers to scout large samples 2 to 5 faster
to identify a region of interest (ROI), and then zoom to image areas at high
resolution with the exclusive ZEISS Xradia Versa RaaD dual magnification
microscope objectives that enable resolution at a distance. ZEISS FPX extends
the ability for ZEISS Xradia Versa XRM to achieve full field of view, whole-sample
imaging, up to 12.7 cm (5 in.) in diameter for samples 10 greater in volume
with higher throughput. With ZEISS FPX, fine details are imaged seamlessly from
system to components with an all in one tool to enhance efficiency.
ZEISS International, Oberkochen, Germany
www.zeiss.com

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 883
Contact ASNT
The ASNT International Service Center is open from 8:30 a.m. to 5:00 p.m. Eastern time, Monday
through Friday. Voicemail messages can be left 24 hours a day by following the recorded prompts. In
the U.S. and Canada, call toll free (800) 222-2768 or (614) 274-6003; fax (614) 274-6899. E-mail
addresses for individual staff members are given below. If you prefer, write ASNT, 1711 Arlingate Lane,
P.O. Box 28518, Columbus, OH 43228-0518. ASNT’s website is available at www.asnt.org.
AREA OF INQUIRY CONTACT (EXTENSION) E-MAIL
Executive Offices
Executive Director Arny Bereson (201) abereson@asnt.org
Executive Assistant Michelle Thomas (223) mthomas@asnt.org
Accounting Department
Chief Financial Officer Mary Potter (203) mpotter@asnt.org
Account balance inquiries Angie Guzzo (228) aguzzo@asnt.org
Credit and collections Trina Coakley (220) tcoakley@asnt.org
Dues payment inquiries Margaret Leonard (229) mleonard@asnt.org
Book Department
Book and catalog orders Curtis Smith (214) csmith@asnt.org
Customer service supervisor Trina Coakley (220) tcoakley@asnt.org
Certification and Technical Services Department
Chief Technical Officer James Bennett (250) jbennett@asnt.org Digital IC Amplifier
Application requests Tricia Davis (219) tdavis@asnt.org
ASNT NDT Level III examinations (International) Tricia Davis (219) tdavis@asnt.org Several new features have been added to
ASNT NDT Level III examinations (U.S.) Lisa Law (226) llaw@asnt.org Mahr Federal’s popular Millimar C1200
ASNT NDT Level III recertification Tricia Davis (219) tdavis@asnt.org
Level III refresher courses Kimberly Donaldson (242) kdonaldson@asnt.org Digital IC amplifier, increasing its applica-
Certification Specialist Kimberly Donaldson (242) kdonaldson@asnt.org tion range and user security. The new
General inquiries Lisa Law (226) llaw@asnt.org
IRRSP/radiation safety Jennifer Harris (237) jharris@asnt.org functions include dynamic measurement
Technical Services Supervisor Charles Longo (241) clongo@asnt.org capability, enhanced display tolerance
NDT Technical Specialist/Certification Manager Donny Didion (240) ddidion@asnt.org
Instructional Designer Bob Conklin (614) 766-7715 bconklin@asnt.org viewing, and password protection for the
Conference Department setup menu. The Millimar C1200 Digital IC
Senior Manager of Conferences Christine Schnitzer (202) cschnitzer@asnt.org amplifier is a low cost, easy to view and
Conference registration Angie Guzzo (228) aguzzo@asnt.org
Exhibit and event coordination Ruth Staat (227) rstaat@asnt.org use readout. It is designed to replace
CEU program Angie Guzzo (228) aguzzo@asnt.org analog meters, as it offers analog-like
Program coordination Alicia LeMasters (213) alemasters@asnt.org
Internet
display performance with very fast
ASNT website Stephen Schaefer (222) sschaefer@asnt.org response technology. The high-resolution
Advertising Diane Oen (209) doen@asnt.org display provides clear digital and analog
Marketing Communications Department readings with selectable resolutions/
Senior Manager of Marketing Communications Garra Liming (211) gliming@asnt.org
Advertising Supervisor Diane Oen (209) doen@asnt.org scales. The dynamic measurement capa-
Public Relations and Brand Manager Dana Sims (244) dsims@asnt.org bility added to the C1200 allows users to
Corporate design services Paul Conley (232) pconley@asnt.org
Member Relations and Services Department
capture maximum, minimum, or
Senior Manager of Member Relations and Services Heather Cowles (216) hcowles@asnt.org maximum-minimum (TIR) values. During
Awards Jessica Ames (233) awards@asnt.org measurement, the digital value is held
Sections Coordinator Debbie Segor (235) dsegor@asnt.org
Program Coordinator Jessica Ames (233) james@asnt.org while the analog position is marked by
Committee participation Jessica Ames (233) james@asnt.org blue lines on the scale. The second new
Member benefits Pat White (217) pwhite@asnt.org
President’s Award points Pat White (217) presaward@asnt.org feature allows asymmetric tolerance
Publications Department markers to be displayed on the scale.
Senior Manager of Publications Tim Jones (204) tjones@asnt.org The display can be set to center on the
Library Haley Cowans (245) hcowans@asnt.org
Materials Evaluation
tolerances rather than the nominal size.
Advertising Diane Oen (209) doen@asnt.org Finally, security is enhanced with password
Articles Karen Balkin (207) kbalkin@asnt.org protection for the setup menu, which
Buyers Guide Diane Oen (209) doen@asnt.org
Calendar Haley Cowans (245) hcowans@asnt.org can now be locked and accessed using a
Employment Service Haley Cowans (245) hcowans@asnt.org 4-digit PIN number. When enabled, a
Ready Reference Guide Karen Balkin (207) kbalkin@asnt.org
Reprints Haley Cowans (245) hcowans@asnt.org prompt appears when accessing the setup
Section News Pat White (217) presaward@asnt.org menu. Arrow keys are used to enter the
NDT Handbook inquiries Karen Balkin (207) kbalkin@asnt.org password. Operating features for the
NDTMarketplace inquiries Haley Cowans (245) hcowans@asnt.org
Advertising Diane Oen (209) doen@asnt.org Millimar C1200 Digital amplifier are set up
Educational Materials inquiries Cynthia Leeman (225) cleeman@asnt.org using the 5-button keypad and visual
Toni Kervina (205) tkervina@asnt.org menu options.
RNDE inquiries Hollis Humphries (206) hhumphries@asnt.org
The NDT Technician (TNT) inquiries Haley Cowans (245) hcowans@asnt.org Mahr Federal Inc., Providence, Rhode Island
If you are having trouble locating who should handle your inquiry, please ask the operator at extension www.mahrexactly.com w x
200 to direct your call to the appropriate department personnel.

884 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
 NDT
Web Sightings
The following are online resources of interest to the NDT community. Each listing includes a description of the website and a photo of the site’s
home page. Published four times a year, NDT Web Sightings is a great way to bring traffic to your site.
www.4nsi.com
INTRODUCING THE X3000 DR & CT X-RAY SYSTEM
A Compact System with Huge Capabilities
North Star Imaging’s X3000 answers the need for a compact
system with the unique capabilities generally available on our
larger DR and CT X-ray systems.
Start browsing today at www.4nsi.com.
www.universityofultrasonics.com
Broaden your ultrasonic testing (UT) spectrum with advanced
ultrasonics skills training for tomorrow’s UT, phased array, and
time of flight diffraction inspectors. University of Ultrasonics’
(UUT’s) qualified instructors deliver field-oriented training methods
in a high-level learning environment. With a curriculum structured
to maximize each student’s learning potential, UUT classes go
beyond that which is traditionally taught in basic skills training.
Register online for your class today.
www.physicalacoustics.com
Physical Acoustics, the renowned acoustic emission testing (AE)
leader worldwide, has launched its website featuring its entire line
of sensors, systems, software, and monitoring solutions.
Check out the new Physical Acoustics website for all your AE needs!
Start browsing today at www.physicalacoustics.com
www.asnt.org
ASNT is very excited about its new website. With the new website,
a digital version of Materials Evaluation is available. For a nominal
fee, advertisers can take advantage of making their e-mail and web
address hot links, or even customize the ads so their products or
services link to the company’s website. ASNT will also have
sponsorship opportunities on a first come, first served basis.
Please contact the advertising supervisor for more information.
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 885
886 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7

Carestream Non-Destructive Testing
Announces Appointments
Carestream Non-Destructive Testing is
pleased to announce the appointment of
Laurent Rapon as digital sales manager for
European, African, and Middle Eastern
Region (EAMER). Laurent brings with him a
wealth of imaging knowledge, strong
management experience, excellent rela-
tionship skills, and an unwavering focus
on customer satisfaction. His career
extends over 20 years with a variety of
major imaging companies where he held
roles of increasing responsibility. Laurent’s
new role begins with immediate effect.
Carestream Non-Destructive Testing is
also pleased to announce the appoint-
ment of Fabrizio Benigni as sales manager
for the UK, Ireland, Nordic regions,
Benelux, France, Algeria, and Morocco.
Fabrizio brings with him a wealth of sales
experience, relationship skills, and a drive
to succeed. His career extends over 15
years with major imaging companies
where he held a variety of key customer
focused roles. Fabrizio’s new role also
begins with immediate effect.
Carestream Non-Destructive Testing’s
solutions address the unique needs of
aerospace, petrochemical, and other
industrial testing with innovative research,
products, and services.
Element Appoints Experienced
Alcoa Senior Executive to Board
Element Materials Technology has
announced the appointment of highly
experienced senior executive Olivier
Jarrault as a non-executive member to the
group’s Board.
Jarrault has 30 years of experience
serving in various senior executive roles
for global companies in the aerospace,
commercial transportation, industrial gas
turbine, nonresidential construction, and
industrial markets. Most recently, he
INDUSTRY
served as Alcoa Executive Vice President
and Group President for Alcoa Engineered
Products and Solutions (EPS), where he
oversaw the transformation of EPS into a
global aerospace solutions leader. In his
five year tenure as Alcoa EVP and Group
President of EPS, he was responsible for
five business and 28 000 employees,
serving six markets across 19 countries
globally.
Jarrault joined Alcoa in 2002 as
President of Alcoa Fastening Systems and
progressed to the role of Alcoa EVP and
Group President of Alcoa Engineered
Products and Solutions. He has held
several senior executive roles with global
organizations such as The Fairchild
Corporation and LISI. Jarrault holds a B.S.
in higher mathematics from Ecole Sainte
Genevieve in France and a M.S. in
mechanical engineering from Centre
d’Évaluation de la Sécurité des
Technologies de l’Information in France. He
later graduated with a M.S. in mechanical
engineering from the California Institute of
Technology and holds an M.B.A. from
University of California, Los Angeles.
NDT Products of Canada Acquires
Air Dynamics
Non-Destructive Testing (NDT) Products
Limited, a major reseller and distributor of
NDT equipment and supplies, announced
on 16 March 2017 the successful acquisi-
tion of Air Dynamics Company Limited, a
leading supplier in the Canadian
aerospace industry. The transaction is an
integral part of the NDT Products strategic
growth plan and continued expansion into
Eastern Canada.
Air Dynamics represents major brands
serving the aerospace market including
3M, PPG, Elixair, and Marvel Aero, among
others. Air Dynamics’ product mix includes
aerospace sealants and adhesives, power
supplies and accessories, marking
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N
news
equipment, and tooling and ground
service equipment. Air Dynamics is a
Transport Canada licensed facility and is
an approved supplier for major airlines,
military and defense establishments, and
aircraft manufacturers and original
equipment manufacturers.
Air Dynamics’ facility will remain in
Baie‐d’Urfé, Quebec, and NDT Products
Limited will be managing and overseeing
the company’s sales and operations under
the leadership of M. Sean McKernan,
president and CEO of NDT Products
Limited.
“The time has come to ensure
continued growth and development for the
company,” said Bruce Truesdale, former
president of Air Dynamics Company
Limited. “Sean and I go back more than 25
years. I hand over Air Dynamics with pride
and confidence that Sean, his team, and
the Air Dynamic staff, will continue to keep
Air Dynamics growing and evolving.”
“We look forward to welcoming Air
Dynamics to the NDT Products family,”
said McKernan. “Air Dynamics is well-posi-
tioned in the aerospace industry and has a
significant presence in Quebec and the
Maritime provinces. With this transaction,
we will leverage on both companies’
strengths, utilize operational synergies and
share best practices. Our team is looking
forward to meeting Air Dynamics
customers and welcoming them into the
NDT Products family.” w x
Write Us
Industry News
Materials Evaluation publishes information
on the activities of businesses in the NDT
field of a noncommercial nature. Send
press releases concerning Industry News
to Haley Cowans at hcowans@asnt.org.
887

Heat Exchange Institute Updates
Standards for Air Cooled
Condensers with 2nd Edition
The Heat Exchange Institute (HEI) has
announced the release of Standards for
Air Cooled Condensers, 2nd edition.
NEW
The Standards for Air Cooled Condensers
(ACCs) covers specification and design
recommendations, along with performance
and operational issues associated with air
cooled condensers for power plant appli-
cations. The standard includes typical
purchaser requirements and manufac-
turers’ experience. It also highlights the
important design criteria for air cooled
condensers.
HEI’s Air Cooled Condenser Section
develops this standard and updates it as
needed. The new edition includes relevant
updates, revisions, and clarifications to
most sections. Metric units have been
added, including metric units for all three
tables on exhaust casing.
There are many different types of ACCs
designed for various services. This
standard applies only to two-stage vacuum
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N
media
steam condensers that are predominately
utilized in power plant applications.
The publication of the new ACC
Standard represents another step in HEI’s
continuing program to provide standards
that represent the latest technological
advancements in heat exchange
equipment. HEI is a non-profit trade asso-
ciation concentrating its efforts on the
manufacturing and engineering aspects of
steam surface condensers, closed
feedwater heaters, power plant heat
exchangers, liquid ring vacuum pumps,
steam jet ejectors, deaerators, and air
cooled condensers.
Suggestions for improvement of the
standard are welcome, and should be
directed to HEI.
www.heatexchange.org w x
889
STANDARDS
The Standards Update department focuses on standards, codes, recommended
practices and other documents used by NDT professionals. This department reports
every few months on new publications, draft work currently in process, and the status
of standards development. This month’s installment focuses on current work in
American National Standards and ISO Draft International Standards.
American National Draft Standards
Project Initiation
ANSI procedures require notification by
ANSI-accredited standards developers
of the initiation and scope of activities
expected to result in new or revised
American National Standards. The following
is a list of proposed actions and new
standards that have been received recently
from accredited standards developers. To
view information about additional standards
for which a project initiation notification has
been submitted, and to search approved
American National Standards, please visit
www.nssn.org, which is a database of
standards information. Note that this
database is not exhaustive.
l BSR/AWS B2.1/B2.1M:2014-AMD1-201x,
Specification for Welding Procedure
and Performance Qualification. This
is an addenda to ANSI/AWS B2.1/
B2.1M:2014-AMD1-2015. This specifica-
tion provides the requirements for
qualification of welding procedure specifi-
cations, welders, and welding operators
for manual, semiautomatic, mechanized,
and automatic welding. The welding
processes included are electrogas
welding, electron beam welding, elec-
troslag welding, flux-cored arc welding,
gas-metal arc welding, gas-tungsten arc
welding, laser-beam welding, oxyfuel gas
welding, plasma-arc welding, shielded-
metal arc welding, stud arc welding, and
submerged arc welding. Base metals,
filler metals, qualification variables,
welding designs, and testing require-
ments are also included.
l BSR N14.1-201x, Uranium Hexafluoride -
Packagings for Transport. This is a revision
of ANSI N14.1-2012. This standard
provides criteria for packaging used for
transport of uranium hexafluoride (UF6).
update
It includes specific information on design
and fabrication requirements for the
procurement of new UF6 packaging for
transportation of 0.1 kg (0.2205 lb) or
more of UF6. This standard also defines
the requirements for in-service inspec-
tions, cleanliness, and maintenance for
packaging in service. Also included are
cylinder loadings; shipping requirements;
and requirements for valves, plugs, and
valve protectors.
l BSR/AWS B1.10M/B1.10-201x, Guide
for the Nondestructive Examination of
Welds. This is a revision of ANSI/AWS
B1.10M/B1.10-2016. This guide
acquaints the user with the nondestruc-
tive examination methods commonly
used to examine weldments. The
standard also addresses which method
best detects various types of discontinu-
ities. The methods included are visual,
liquid penetrant, magnetic particle, radi-
ographic, ultrasonic, electromagnetic
(eddy current), and leak testing.
l BSR/AWS B1.11M/B1.11-201x, Guide
for the Visual Examination of Welds. This
is a revision of ANSI/AWS B1.11M/B1.11-
2014. This guide contains information to
assist in the visual examination of welds.
Included are sections on fundamentals,
surface conditions, and equipment.
Sketches and full-color photographs illus-
trate discontinuities commonly found in
welds.
Call for Comment on Proposals Listed
The public comment period has passed for
the following draft American National
Standards, which are currently in review.
l BSR/ICEA S-75-381-201x/NEMA WC 58-
201x, Portable and Power Feeder Cables for
Use in Mines and Similar Applications. This
is a revision and redesignation of ANSI
ICEA S-75-381-2008/NEMA WC 58-2008.
These standards apply to materials,
construction, and testing of insulated
cables used for the utilization of elec-
trical energy in surface and under-
ground mines and similar applications.
Included are portable cables for use in
mining machines, dredges, shovels,
and similar equipment, and mine power
cables for use as connections between
units of mine distribution systems, and
remote control and drill cords for mining
and similar applications.
l BSR/AWS B5.2-201x, Specification for
the Training, Qualification, and Company
Certification of Welding Inspector
Specialists and Welding Inspector
Assistants. This is a new standard. This
specification defines the requirements
and program for an employer (company)
to train, qualify, and company certify
Welding Inspector Specialists and
Welding Inspector Assistants to contract
or industry-specific inspector standards.
The program is developed as a written
practice and controlled by an employer.
The employer is responsible for the
testing, examination, and qualification of
their inspectors.
l BSR/AWS D14.3/D14.3M-201x,
Specification for Welding Earthmoving,
Construction, Agricultural, and Ground-
Based Material Handling Equipment.
This is a revision of ANSI/AWS
D14.3/D14.3M-2010.
This provides standards for producing
structural welds used in the manufac-
ture and repair of earthmoving,
construction, agricultural, and ground-
based material handling equipment.
Such equipment is defined as self-
propelled, on- and off-highway
machinery and associated implements.
Manufacturer’s responsibilities are
presented as they relate to the welding
practices that have been proven
successful within the industry in the
production of weldments on this
equipment. Requirements for basic
weld details, base material, filler
material, processes, welding procedure
qualification and documentation,

890 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
 welding personnel qualification, weld
quality, inspection, and repair are
included with consideration given to
factors that affect weldability.
l BSR/ASME AG-1-201X, Code on Nuclear
Air and Gas Treatment. This is a revision
of ANSI/ASME AG-1-2015. This Code
provides requirements for the perform-
ance, design, fabrication, installation,
inspection, acceptance testing, and
quality assurance of equipment used in
air and gas treatment systems in
nuclear facilities.
l BSR/SAIA A92.20-201x, Design,
Calculations, Safety Requirements and
Test Methods for Mobile Elevating Work
Platforms (MEWPs). This is a new
standard. This Standard is intended to be
used in conjunction BSR/SAIA A92.22,
Safe Use of MEWPs and BSR/SAIA
A92.24, Training Requirements for
Operators of MEWPs. This American
National Standard specifies safety
requirements and preventive measures,
and the means for their verification, for
certain types and sizes of mobile
elevating work platforms (MEWPs)
intended to position personnel, along
with their necessary tools and materials,
at work locations. It contains the struc-
tural design calculations and stability
criteria, construction, safety examina-
tions, and tests that shall be applied
before a MEWP is first put into service.
l BSR/SAIA A92.22-201x, Safe Use of
Mobile Elevating Work Platforms
(MEWPs). This is a new standard.
This Standard is intended to be used in
conjunction with BSR/SAIA A92.20,
Design calculations, safety requirements
and test methods for Mobile Elevating
Work Platforms (MEWPs) and BSR/SAIA
A92.24, Training Requirements for
Operators of Mobile Elevating Work
Platforms (MEWPs). This Standard
specifies requirements for application,
inspection, training, maintenance,
repair, and safe operation of Mobile
Elevating Work Platforms (MEWPs).
It applies to all types and sizes of
MEWPs as specified in BSR/SAIA
A92.20 that are intended to position
personnel, along with their necessary
tools and materials, at work locations.
l BSR/SAIA A92.24-201x, Training
Requirements for the Use, Operation,
Inspection, Testing and Maintenance of
Mobile Elevating Work Platforms (MEWPs).
This is a new standard. This Standard is
intended to be used in conjunction with
BSR/SAIA A92.20, Design, Calculations,
Safety Requirements and Test Methods for
Mobile Elevating Work Platforms (MEWPs),
and BSR/SAIA A92.22, Safe Use of Mobile
Elevating Work Platforms (MEWPs). This
standard provides methods and guide-
lines to prepare MEWP training materials,
defines administrative criteria, and
delivers elements required for proper
training and familiarization. It applies to
all types and sizes of MEWPs defined in
BSR/SAIA A92.20 that are intended to
position personnel, along with their
necessary tools and materials, at work
locations.
l BSR/HI 4.1-4.6-201x, Sealless,
Magnetically Driven Rotary Pumps for
Nomenclature, Definitions, Application,
Operation, and Test. This is a revision of
ANSI/HI 4.1-4.6-2010. This standard
covers the unique features of sealless,
magnetically driven rotary pumps and
includes sections on types and nomen-
clature; definitions; design and applica-
tions; installation, operation, and
maintenance; and test. Because of the
variety of rotary pump configurations
available and the broad range of appli-
cations, familiarization with Hydraulic
Institute Standards ANSI/HI 3.1 - 3.5,
Rotary Pumps for Nomenclature,
Definitions, Application and Operation;
and ANSI/HI 3.6, Rotary Pump Tests, is
recommended.
l BSR/AWS F1.6-201X, Guide for Estimating
Welding Emissions for EPA and
Ventilation Permit Reporting. This is a
new standard. This document assists
companies in estimating emissions from
welding processes for EPA reporting
purposes by choosing the simplest appli-
cable method and following its steps.
Example calculations are included.
l BSR/AWS A5.6/A5.6M-2008 (R201x),
Specification for Copper and Copper- Alloy
Electrodes for Shielded Metal Arc Welding.
This is a reaffirmation of ANSI/AWS
A5.6/A5.6M-2008. This Specification
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N
prescribes the requirements for classifica-
tions of copper and copper-alloy elec-
trodes for shielded metal arc welding.
Classifications are based on chemical
composition, mechanical properties, and
usability of the electrodes.
l BSR/AWS D1.3/D1.3M-201x, Structural
Welding Code-Sheet Steel. This is a
revision of ANSI/AWS D1.3/D1.3M-
2007. This code contains the require-
ments for arc welding of structural
sheet/strip steels, including cold-
formed members, collectively referred to
as “sheet steel,” which are equal to or
less than 3/16 in (0.188 in) [4.8 mm] in
nominal thickness. When this code is
stipulated in contract documents,
conformance with all its provisions shall
be required, except for those provisions
that the engineer or contract documents
specifically modifies or exempts.
l BSR/IIAR 6-201x, Standard for
Inspection, Testing, and Maintenance of
Safe Closed-Circuit Ammonia
Refrigeration Systems. This is a new
standard. This safety standard specifies
minimum requirements for inspection,
testing, and maintenance of closed-
circuit ammonia refrigeration systems.
This standard is intended to assist indi-
viduals responsible for developing and
implementing inspection, testing, and
maintenance programs for facilities with
stationary closed-circuit ammonia refrig-
eration systems using recognized and
generally accepted good engineering
practices (RAGAGEP).
l BSR/ASSE Z359.18-201X, Safety
Requirements for Anchorage Connectors
for Active Fall Protection Systems. This is
a new standard. This Standard estab-
lishes requirements for the performance,
design, testing, marking, and instructions
for use of anchorage connectors in travel
restraint, fall arrest, rescue, work position,
rope access, and suspended
component/tie-back line systems only.
l BSR/ASSE Z359.12-201X, Connecting
Components for Personal Fall Arrest
Systems. This is a revision of ANSI/ASSE
Z359.12-2009. This standard establishes
requirements for the performance, design,
marking, qualification, test methods, and
removal from service of connectors.
891
ISO Draft International Standards
The following are standards that the
International Organization for
Standardization (ISO) is considering for
approval. The proposals have received
substantial support within the technical
committee that developed them and are
now being circulated to ISO members for
comment and vote. Readers interested in
reviewing or commenting on these
standards should order copies from ANSI.
l ISO/DIS 9869-2, Thermal insulation -
Building elements - In-situ measure-
ment of thermal resistance and thermal
transmittance - Part 2: Infrared method
for frame structure dwelling.
l ISO/DIS 17677-1, Resistance welding -
Vocabulary - Part 1: Spot, projection
and seam welding.
l ISO/DIS 11699-2, Non-destructive
testing - Industrial radiographic films -
892 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Part 2: Control of film processing by
means of reference values.
l ISO/DIS 19232-5, Non-destructive
testing - Image quality of radiographs -
Part 5: Determination of the image
unsharpness and basic spatial resolu-
tion value using duplex wire-type image
quality indicators.
l ISO/DIS 11666, Non-destructive testing
of welds - Ultrasonic testing -
Acceptance levels.
l ISO/DIS 10804, Restrained joint systems
for ductile iron pipelines - Design rules
and type testing.
l ISO/DIS 12107, Metallic materials -
Fatigue testing - Statistical planning and
analysis of data.
l ISO/DIS 16063-33, Methods for the
calibration of vibration and shock
transducers - Part 33: Testing of
magnetic field sensitivity.
l ISO/DIS 16063-33, Methods for the cali-
bration of vibration and shock trans-
ducers - Part 33: Testing of magnetic
field sensitivity.
l ISO/DIS 13259, Thermoplastics piping
systems for underground non- pressure
applications - Test method for leaktight-
ness of elastomeric sealing ring type
joints.
l ISO 10816-8/DAmd1, Mechanical
vibration - Evaluation of machine
vibration by measurements on non-
rotating parts - Part 8: Reciprocating
compressor systems - Amendment 1.
l ISO/DIS 18095, Condition monitoring
and diagnostics of power transformers.
l ISO/DIS 29821, Condition monitoring
and diagnostics of machines - Ultrasound
- General guidelines, procedures and
validation. wx
NEW
ROBERT E. SHANNON
Associate Technical Editor
patents
US 9594061
Methods and apparatus for detecting defects
in an object of interest
(J. K. Hartman and L. H. Pearson)
Conventional nondestructive evaluation
techniques have been used in solid rocket
motor programs, including NASA’s Space
Shuttle, to screen for defects such as
cracks, debond, voids in the assembly of
the solid rocket motor consisting of outer
casing, insulation, and internal propellant.
Uncertainty in the size, location, and orien-
tation of defects may result in uncertainty
in the analytical models designed to
assess structural allowable stresses and
strains for the propellant. Defects that
occur in propellant-liner interface may
cause hot gas to be present near the wall
of the rocket motor case. In addition, if a
defect near the liner extends further into
the propellant, the propellant may become
detached from the bonding surface of the
liner. Debonding may cause further
defects, for example, cracks, in the propel-
lant, which may result in augmented and
accelerated burning of the propellant,
including near the wall of the case, as well
as concerns regarding the structural
impact of the decreased bonding with the
liner. In addition, environmental factors,
such as moisture, may weaken the
adhesion strength of the propellant binder
to the surface of the reinforcing and
combustible fillers in the solid rocket
motor over time, which can result in
reduced load bearing capability of the
propellant. Because there is often uncer-
tainty associated with the constitutive
properties of the these polymeric systems,
especially when exposed to environmental
aging, the term “health” of solid rocket
motor is sometimes used to classify the
launch readiness of the solid rocket motor
and the propellant’s ability to withstand
damage during the dynamic launch event.
Conventional methods for screening the
health of a solid rocket motor include radi-
ographic inspection methods for verifying the
health and quality of the propellant, liner,
894 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
and insulation of a solid rocket motor. These
typically require the use of one X-ray device
oriented such that the X-ray is transmitted
substantially orthogonally to the solid rocket
motor to obtain an X-ray image through a
thickness of the solid rocket motor and into
the propellant. As a result, defects in the
propellant may be detected. Then a second
X-ray device must be oriented at an angle
such that it obtains a tangential image of the
solid rocket motor to better detect defects in
the additional insulation layers that largely go
undetected by the first X-ray device. Because
of the size of solid rocket motors, this inspec-
tion method may require an undesirably large
number of man-hours to obtain the images.
In particular, the tangential image may be a
very small field of view relative to the entire
solid rocket motor being imaged. As a result,
after each tangential image is obtained, the
solid rocket motor must be rotated to a new
position to obtain another tangential image.
Each tangential image requires a substantial
number of man-hours to rotate the solid
rocket motor between each image, in
addition to the time that is needed to expose
the solid rocket motor to an X-ray. Thus, while
conventional inspection methods may have
resulted in the detection of defects, the
conventional inspection methods may be
time consuming and costly. Some additional
conventional inspection approaches have
attempted to detect voids using ultrasonic
methods. These conventional ultrasonic
methods have used either a repetitive high
frequency (for example, 1 MHz to 10 MHz)
square or instantaneous pulse for the ultra-
sonic signal propagating into a material, and
measuring and quantifying discreet reflec-
tions off of internal interfaces or potential
defects. Images may be generated based on
the peak amplitude responses from the ultra-
sonic signal. The conventional inspection
methods, however, have not adequately
provided the necessary depth to obtain
accurate information any significant distance
past the inside wall of case. Therefore,
defects in the insulation materials may be
undetected.
This patent presents a new ultrasonic
approach for detecting defects in a struc-
tures comprised of multiple material layers
where each layer may have significantly
different acoustic impedance. Although the
ultrasonic approach described in this patent
ion refers to a solid rocket motor and its
components, this is not limited to solid
rocket motors, but could more generally
apply to any object (for example, space
launch systems, tanks, and armor) that
includes multiple materials with different
acoustic impedances—particularly if there is
an acoustic impedance mismatch between
an outer layer having a higher acoustic
impedance relative to the inner layer(s). For
example, a steel case has a high acoustic
impedance relative to rubber liners. The
greater the impedance mismatch there is
between two materials, the worse conven-
tional inspection methods tend to perform
in detecting defects associated with the
inner materials because more acoustic
energy stays trapped in the outer layer
and/or the acoustic energy is quickly attenu-
ated in the inner layers. Examples were
given where the case may be formed from
steel or other appropriate materials, which
may include metals, metal-alloys, and
carbon or other fiber reinforced plastic
materials. The insulation material may
include Nitrile Butadiene Rubber (NBR),
Room Temperature Vulcanized (RTV) Rubber,
Silicone Rubber, ethylene propylene diene
Monomer (EPDM) Rubber, Urethane Rubber,
Hydroxy Terminated Polybutidiene (HTPB)
Rubber, Inorganic Phosphazene Rubbers,
Natural Rubber, or other suitable insulation
materials. The liner may be formed from
materials such as a polyurethane adhesive,
and the like. Propellants of solid rocket
motors may include combustive and particu-
late materials mixed within an elastomeric
binder material (for example, HTPB,
Polybutadiene/Acrylonitrile [PBAN], or
Nitrate ester/polyester [NEPE] based
polymers).
In operation, the ultrasonic sensor may
interrogate the solid rocket motor to detect
defects by applying an ultrasonic signal to
the solid rocket motor by energizing the
ultrasonic sensor to generate a tone burst
having a predetermined frequency that is
tuned to create a standing wave (as in,
destructive interferences) in the reflections
within the case. With a tone burst, the ultra-
sonic signal is driven for a predetermined
number of cycles (for example, five cycles)
and then the ultrasonic signal is shut off.
Within the tone burst, the ultrasonic signal
may be a sine wave, square wave, or other
periodic signal having the predetermined
frequency band to create a standing wave in
the case. With the reflections in the case
destructively interfering with themselves, the
smaller energy reflections from the deeper
internal layers (for example, insulation, liner,
or propellant) may have a higher signal-to-
noise ratio (SNR) enabling their detection.
The predetermined frequency may be
approximately the resonant frequency for the
case that is based, at least in part, on the
material of the case and the thickness of the
case. For example, for a steel case that is
approximately 0.5 in. thick, the predeter-
mined frequency may be tuned to be approx-
imately 250 kHz, which is approximately the
frequency at which the wavelength is about
twice the thickness of the case. As a result,
the effects of the reflections within the case
may be reduced so that the return signal
coming back from the internal layers (for
example, insulation or liner) may be more
detectable. Because the predetermined
frequency of the ultrasonic signal may be
tuned to the resonant frequency of the case
(or other outer material), which resonant
frequency is dependent on the thickness of
the material used, having a substantially
uniform thickness for the case 304 may
improve results. In some embodiments, the
thickness of the other materials (such as
insulation or liner) may vary without
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N
significantly affecting the results. In
operation the return signals from the
different reflections are received after the
tone burst is shut off, there may be a greater
opportunity to detect the reflections from
the deeper internal layers because the reflec-
tions within the case of the outer material
destructively interfere due to the predeter-
mined frequency being tuned to the
resonance of the outer material. In addition,
because lower frequencies (such as those in
the kHz range) may be used in implementa-
tion of embodiments of the disclosure in
comparison to those employed in conven-
tional methods, more energy may pass
through deeper into the internal layers of the
solid rocket motor such that a deeper image
may be obtained. As a result, the ultrasonic
signal may experience less attenuation in the
internal layers so that information about the
conditions of defects deeper into the solid
rocket motor. w x
895
CORPORATE
3angles, Inc. (Albany, NY)
3D Engineering Solutions
(Cincinnati, OH)
3E NDT, LLC (La Porte, TX)
4A Inspection, LLC (Houston, TX)
A Coast to Coast Inspection Services,
Abdallah I AlTamimi Industrial
Services (Al Khobar, Saudi Arabia)
ABM Franchising Group
(Canonsburg, PA)
Access Plug Flange, Inc. (Houston, TX)
ACML 2i2S (Alexandria, Egypt)
ACNDT, Inc. (Middleburg, FL)
Acoustic Technology Group
(Grandville, MI)
ACS-Solutions GmbH (Saarbrucken,
Germany)
Acuren Inspection, Inc. (Edmonton,
Canada)
Adaptive Energy (University Place, WA)
Ademinsa (Lima, Peru)
Advanced Energy (Fort Collins, CO)
Advanced Inspection Services Co.
(Jubail Industrial City, Saudi Arabia)
Advanced Inspection Technologies
(Melbourne, FL)
Advanced Material Solutions
(Phoenix, AZ)
Advanced OEM Solutions
(Cincinnati, OH)
Advanced Test Equipment Rentals
(San Diego, CA)
AEIS (Rahway, NJ)
Aero Norway AS (Sola, Norway)
Aerofab NDT, LLC (Kent, WA)
Aerotest Operations, Inc. (San
Ramon, CA)
AES Destructive & NDT, Ltd. (Kwai
Chung, Hong Kong)
AES NDT (Las Vegas, NV)
African NDT Centre Pty., Ltd.
(Centurion, South Africa)
AGD Inspection Services, LLC
(Stafford, TX)
AGR Inspection, Inc. (Burleson, TX)
AIP (Houston, TX)
AIP Global Strategies (Pelham, NH)
Air Services (Middleburg Heights, OH)
Aircraft Inspection Services (Grand
Rapids, MI)
Aircraft X-Ray Labs, Inc. (Huntington
Park, CA)
Airon Academy (Trivandrum, India)
Aker Solutions Saudi Arabia Co.,
Ltd. (Al Khobar, Saudi Arabia)
Akroscan End Chile, SA (Santiago,
Chile)
Akura Bina Citra (Bekasi, Indonesia)
Al Mansoori Inspection Services
(Abu Dhabi, United Arab Emirates)
Al Othman Trading & Contracting
Co. (Dammam, Saudi Arabia)
Al Rookal for Engineering
Inspection, Ltd. (Baghdad, Iraq)
Allis Engineering (Lake Elsinore, CA)
Alpha NDT (Ho Chi Minh City, Vietnam)
Alta Vista Solutions (Richmond, CA)
ALX Industries (Pasadena, TX)
AM Technical Solutions, Inc. (Austin, TX)
Amerapex Corp. (Houston, TX)
896 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Thank You
ASNT is proud to present these NDT manufacturers, users
and suppliers who support the Society. This list is current
as of 1 June 2017.
American Aerospace Technical
Academy (Los Angeles, CA)
American Inspection Services, Inc.
(Grand Bay, AL)
American Institute of Nondestructive
Testing (Baxter, MN)
American Marine Corp. (Anchorage, AK)
American Piping Inspection (Tulsa, OK)
American Testing Services
(Miamisburg, OH)
Amo & Partners Engineering Co. (Al
Khobar, Saudi Arabia)
Amosco (Eastleigh, United Kingdom)
AMS Store and Shred, LLC (Lake In
The Hills, IL)
Aolong Group (Dandong City, China)
Aplus+ NDT (Coimbatore, India)
Applied Technical Services
(Marietta, GA)
Applus RTD (Edmonton, Canada)
Aqua Communications, Inc.
(Waltham, MA)
Arad Paya Quality Engineering &
Inspection Co. (Tabriz, Iran)
Aral General Trading, LLC (Port
Saeed, United Arab Emirates)
Arcadia Aerospace Industries (Punta
Gorda, FL)
Arcmart Indonesia (Bandung,
Indonesia)
Armstech Engineers Private Ltd.
(Kerala, India)
Arrow-Tech, Inc. (Rolla, ND)
Arya Fould Gharn (Ahwaz, Iran)
ASG Inspection, Ltd. (Aberdeen,
United Kingdom)
AssetSmart (Calabasas, CA)
Associated X-Ray Corp. (East
Haven, CT)
ATEC Training & Certification
Services (Houston, TX)
Atlas Inspection Technologies
(Seattle, WA)
Atout Technical Supports
(Port-Gentil, Gabon)
Aurora Institute & Inspection
Services (Coimbatore, India)
AUT Solutions (Fulshear, TX)
Automated Inspection Systems
(Martinez, CA)
Avonix Imaging (Maple Grove, MN)
Aycan Data Management
(Rochester, NY)
AZTech Training & Consultancy
(Dubai, United Arab Emirates)
B Centura X-Ray NDT (Cleveland, OH)
Baker Testing Services, Inc.
(Rockland, MA)
Balteau NDT (Hermalle sous
Argenteau, Belgium)
Base Line Data, Inc. (Portland, TX)
Best NDT (Springfield, VA)
partners
BETZ Engineering & Technology
Zone (Chennai, India)
BFW Engineering & Testing
(Paducah, KY)
BG Detection Services/ LA X-Ray
(Sun Valley, CA)
Bighorn Inspection, Inc. (Laurel, MT)
BKS Consulting & Training Institute
(Tehran, Iran)
Blastline Institute of Surface
Preparation & Painting (Kochi,
India)
Blatek, Inc. (State College, PA)
Blitz Academy (Cochin, India)
BlueScope Steel Research
(Wollongong, Australia)
Boeing (Seattle, WA)
Bossier Parish Community College
(Bossier City, LA)
Branch Radiographic Labs, Inc.
(Cranford, NJ)
Bridgsite Nigeria, Ltd. (Port
Harcourt, Nigeria)
BRL Consultants, Inc. (San Antonio, TX)
Bruker (Kennewick, WA)
BTEC LLC (Pueblo, CO)
C D
Cadillac Casting, Inc. (Cadillac, MI)
Cadorath Aerospace (Broussard, LA)
Cairo Inspection Co. (Nasr City Cairo,
Egypt)
Calaya Engineering (Port Harcourt,
Nigeria)
Callington Haven Pty., Ltd.
(Rydalmere, Australia)
CAM (Doha, Qatar)
Canadian Engineering & Inspection,
Ltd. (Edmonton, Canada)
Carbon Steel Inspection, Inc.
(Pittsburgh, PA)
Carestream NDT (Rochester, NY)
Carl Zeiss Industrial Metrology
(Maple Grove, MN)
Caterpillar, Inc. (Peoria, IL)
CATSI, Inc. (Valparaiso, IN)
CD International Technology, Inc.
(Livermore, CA)
CDA Technical Institute (Jacksonville,
FL)
Cenergy International Services, LLC
(Houston, TX)
Central Flying Service (Little Rock,
AR)
CentroTEST Asia, Inc. (Mandaluyong
City, Philippines)
Chemetall US, Inc. (New Providence, NJ)
Chesapeake Testing (Belcamp, MD)
Churchill Steel Plate, Ltd.
(Twinsburg, OH)
Cifra Industrial Services Corp.
(Makati City, Philippines)
Circle Systems, Inc. (Hinckley, IL)
Citadel Engineering Co. (Sarasota, FL)
Clover Park Technical College
(Lakewood, WA)
CNI Pacific Co., Ltd. (Banchang,
Thailand)
Inc. (Portland, OR)
Cokebusters USA (Houston, TX)
College of the North Atlantic
(Stephenville, Canada)
Comet Technologies USA, Inc.
(Shelton, CT)
Comibassal (Alexandria, Egypt)
Commercial Divers International,
Inc. (Goodyear, AZ)
Commodity Resource &
Environmental, Inc. (Burbank, CA)
Connect NDT, Ltd. (Aberdeenshire,
United Kingdom)
Cooperheat Saudi Arabia Co., Ltd.
(Jubail, Saudi Arabia)
CoreStar International Corp. (Irwin, PA)
Cotech IRM Services, Inc. (Houston, TX)
Creaform, Inc. (Levis, Canada)
Curtis Industries, Inc. (Cowansville, PA)
Curtiss Wright Anatec-LMT (Irvine, CA)
Cutech Group (Singapore,
Singapore)
Cuyahoga Community College
(Cleveland, OH)
CXR Corp. (Kure City, Japan)
Cygnus Instruments (Annapolis, MD)
Cylinder Testing Solutions (Denver, CO)
DARLSCO Inspection Services
(Dubai, United Arab Emirates)
Dakota Ultrasonics (Scotts Valley, CA)
Danatronics (Danvers, MA)
Danco Inspection Service, Inc.
(Oklahoma City, OK)
Dantec Dynamics, GmbH (Ulm,
Germany)
Dantec Dynamics, Inc. (Holtsville, NY)
Dares Srls. (Casamarciano, Italy)
DBI, Inc. (Lincoln, NE)
Decibel NDE Inspections & Training
Institute (Kerala, India)
DECTRIS, Ltd. (Baden-Daettwil,
Switzerland)
DEKRA Industrial (Ankara, Turkey)
Delphi Precision Imaging
(Redmond, WA)
Detagulf Global Services, Ltd. (Port
Harcourt, Nigeria)
Detection Technology, Inc.
(Billerica, MA)
DETEK, Inc. (Temple Hills, MD)
Diakont Advanced Technologies
(San Diego, CA)
Diamond Technical Services, Inc.
(Blairsville, PA)
Dixon Hard Chrome (Sun Valley, CA)
DJA Inspection Services, Inc. (Reno,
PA)
DK Shah NDT Training Institute
(Vadodara, India)
DolphiTech (Raufoss, Norway)
Dominion NDT Services, Inc.
(Richmond, VA)
Doncasters (Groton, CT)
Draka Cableteq, Inc. (Bridgewater, NJ)
Dürr NDT, GmbH & Co. KG
(Bietigheim-Bissingen, Germany)
Dynamic Software Solutions
(Niceville, FL)
E Harcourt, Nigeria)
Echo Maintenance (Nederland, TX)
Echo Ultrasonics (Bellingham, WA)
Eclipse Scientific (Waterloo, Canada)
ECS, Inc. (Kennesaw, GA)
Eddyfi (Quebec, Canada)
EddySonix (Orbe, Switzerland)
Edison Welding Institute
(Columbus, OH)
Eishin Kagaku Co., Ltd. (Minato-Ku,
Japan)
ElectroMagneticWorks, Inc. (LaSalle,
Canada)
Enerfab, Inc. (Cincinnati, OH)
Energyray Integrated Services, Ltd.
(Port Harcourt, Nigeria)
Envite Nigeria, Ltd. (Port Harcourt,
Nigeria)
Epica Applied Technologies (San
Clemente, CA)
ETher NDE, Ltd. (St. Albans, United
Kingdom)
ETM, Inc. (Newark, CA)
Euroteck Systems UK, Ltd.
(Tamworth, United Kingdom)
Exova (Linkoping, Sweden)
Extende (Ballston Spa, NY)
ExxonMobil (Baytown, TX)
F Gulf Quality Control Co., Ltd.
Fangzheng Valve Group Co., Ltd.
(Wenzhou, China)
First Alert Sling Testing, LLC
(Lafayette, LA)
Fish & Associates, Inc. (Middleton, WI)
Flathead Valley Community College
(Kalispell, MT)
FlawSpec Manufacturing, Inc.
(Edmonton, Canada)
FlawTech (Concord, NC)
Foerster Instruments, Inc.
(Pittsburgh, PA)
Force Technology (Broendby, Denmark)
Formweld Fitting, Inc. (Milton, FL)
Foster Tecnica Solutions (Kerala,
India)
Fujifilm North America – NDT
Materials and Equipment
(Valhalla, NY)
G United Kingdom)
G&G Technical Services, Ltd.
(London, United Kingdom)
Gamesa Innovation & Technology
(Sarriguren, Spain)
Gamma Petroleum Services (Basra,
Iraq)
Gamma Rad (Tehran, Iran)
GB Inspection Systems, Ltd.
(Cannock, United Kingdom)
GE Measurement & Control
(Lewistown, PA)
GE Power Generation Services
(Houston, TX)
General Dynamics NASSCO Norfolk
(Norfolk, VA)
Genesis Systems Group
(Davenport, IA)
George Consulting Services, Inc.
(Monaca, PA)
GES for Petroleum Inspection &
Training Services (Menofeyah,
Egypt)
Gladd Solutions (Plymouth, MI)
Glajoe Multi Services, Ltd. (Port
Global Academy for Inspection and
Training (Pandalam, India)
Global Diving & Salvage, Inc.
(Anchorage, AK)
Global Engineering Documents (IHS)
(Englewood, CO)
Global Lifting Services, Nigeria, Ltd.
(Port Harcourt, Nigeria)
Global Oil Inspectindo (Indramayu,
Indonesia)
Glomacs Fz, LLC (Dubai, United Arab
Emirates)
GMTech Services Corp. (Anaheim, CA)
Golden Engineering, Inc.
(Centerville, IN)
Goolsby Testing Laboratories
(Humble, TX)
Gravitech Inspection Services
(Ernakulam, India)
Guangdong Goworld Co., Ltd.
(Shantou, China)
Guangzhou Doppler Electronic
Technologies Co., Ltd. (Guangzhou
City, China)
Guided Ultrasonics, Ltd. (Brentford,
United Kingdom)
Gulf Energy International (Dammam,
Saudi Arabia)
(Al-Khobar, Saudi Arabia)
Gulf X-Ray Services, Inc. (Gretna, LA)
Gulmay (Suwanee, GA)
H (Sherwood Park, Canada)
HAECO Americas (Lake City, FL)
Hakem Group (Lyttelton, South Africa)
Haks Engineers Architects & Land
Surveyors PC (New York, NY)
Halifax International, Fze. (Erbil, Iraq)
Hamamatsu Corp. (Bridgewater, NJ)
Helium Leak Testing, Inc. (Simi
Valley, CA)
Hellier (Houston, TX)
Herzog Services, Inc. (St. Joseph, MO)
High Technology Sources, Ltd.
(Didcot, United Kingdom)
Highland Oilfield Services Group,
Ltd. (Aberdeen, United Kingdom)
Hi-Spec Systems, Ltd. (Nantwich,
Hobart Institute of Welding
Technology (Troy, OH)
Hocker, Inc. (Houston, TX)
Hodges Transportation, Inc. (Carson
City, NV)
Honeywell Aerospace de Mexico
(Chihuahua, Mexico)
Honeywell Federal Manufacturing &
Technologies (Kansas City, MO)
HSI Group, Inc. (Torrance, CA)
I (Edmonton, Canada)
I & T Nardoni Institute, Srl. (Brescia,
Italy)
Idaho National Laboratory (Idaho
Falls, ID)
ImechE Engineering Training
Solutions (Sheffield, United
Kingdom)
Imperium, Inc. (Beltsville, MD)
Industrial Inspection Systems, Ltd.
(Vaughan, Canada)
Industrial Testing Laboratory
Services, LLC (Pittsburgh, PA)
Inmedin Ingenieria (Vina del Mar,
Chile)
Innerspec Technologies, Inc.
(Forest, VA)
Innodes (Quito, Ecuador)
Insight KK (Tokyo, Japan)
Inspec Testing, Inc. (National City, CA)
Inspectest Pvt., Ltd. (Lahore Punjab,
Pakistan)
Inspection Plug Strategies, LLC
(Houston, TX)
Inspection Point Seals, LLC
(Prairieville, LA)
Inspection Quality International
(Bangalore, India)
Inspection Technologies, Inc.
(Pomona, CA)
Inspection Technology, WLL (Doha,
Qatar)
Inspectron, Inc. (Novi, MI)
Institute of Industrial Quality
Management (Shoranur, India)
Institute of Nondestructive Testing
and Training (Mumbai, India)
Institute of Reliability Centred
Maintenance (Lahore, Pakistan)
Instrument Technology, Inc.
(Westfield, MA)
Integrated Quality Services
(Ontario, CA)
Integrity & NDT Solutions
(Cundinamarca, Colombia)
Integrity Products & Supplies, Inc.
Integrity Scientific Laboratory
(Dubai, United Arab Emirates)
Integrity Smart Services, LLC
(Muscat, Oman)
Integrity Testing & Inspection, LLC
(Onaway, MI)
International Inspection (Santa Fe
Springs, CA)
International Leak Detection, LLC
(Des Plaines, IL)
International Maritime Welding
Society (NCR, Philippines)
International Quality Systems
(Concepcion, Chile)
International Robotic Tank Solutions
(Khobar, Saudi Arabia)
Intertek (Amelia, LA)
Intron Plus (Moscow, Russia)
Inuktun Services, Ltd. (Nanaimo,
Canada)
Ionix Advanced Technologies (Leeds,
United Kingdom)
IPSI (Courbevoie Cedex, France)
Iranian Society of Technical
Inspection & NDT Companies
(Tehran, Iran)
IRED Thermal Group, Ltd.
Irez Academy (Kollam, India)
Iris Inspection Services, Inc.
(Baytown, TX)
IRISNDT (Houston, TX)
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N
IRM Servicos Eireli (Macae, Brazil)
IS Industrie Thailand, Ltd. (Bangkok,
Thailand)
Isoflex Radioactive, LLC (San
Francisco, CA)
Isotopes Arabia Co., Ltd. (Al Khobar,
Saudi Arabia)
IVC Technologies (Lebanon, OH)
IveyCooper Services, LLC (Soddy
Daisy, TN)
J
Jan Kens Co., Inc. (Monrovia, CA)
JANX (Parma, MI)
JB Testing, Inc. (Blaine, MN)
JC International, Ltd. (Port Harcourt,
Nigeria)
Jentek Sensors, Inc. (Waltham, MA)
JES Pipelines, Ltd. (Willemstad
Curacao, Netherlands Antilles)
Jesse Garant Metrology Center
(Windsor, Canada)
JETS, Inc. (Carrollton, TX)
Jireh Industries, Ltd. (Ardrossan,
Canada)
JMD NDT, Inc. (Piscataway, NJ)
Johnghama International Services,
Ltd. (Warri, Nigeria)
Jubail Industrial College (Jubail
Industrial City, Saudi Arabia)
JZ Russell Industries, Inc.
(Nederland, TX)
K
KA Imaging (Kitchener, Canada)
Kakivik Asset Management, LLC
(Anchorage, AK)
Kalva Engineers Pvt., Ltd. (Nagole
Hyderabad, India)
KARL STORZ Industrial Group (El
Segundo, CA)
Keltron Kerala State Electronics
Devp. Corp., Ltd. (Trivandrum,
India)
Keville Enterprises, Inc. (Boston, MA)
Kimtron, Inc. (Oxford, CT)
Kinetic Solutions, LLC (Fort Ripley, MN)
Koslow Scientific Co. (Englewood, NJ)
KTA Tator, Inc. (Pittsburgh, PA)
Kuwait Pipe Industries & Oil
Services Co. (Kuwait City, Kuwait)
L
Labino AB (Solna, Sweden)
Laboratory Testing, Inc. (Hatfield, PA)
Laser Technology (Norristown, PA)
Laurus Systems, Inc. (Ellicott City, MD)
Lavender International NDT
Consultants (Sheffield, United
Kingdom)
Librestream Technologies
(Winnipeg, Canada)
Lickenbrock Technologies, LLC (Saint
Louis, MO)
Lion Inspection Services, Inc.
(Houston, TX)
Lonestar PMI (League City, TX)
Lucid Software, Ltd. (Chennai, India)
M
M2M (San Francisco, CA)
MA Al Mutlaq Sons Co. (Dammam,
Saudi Arabia)
MAC NDT Services, LLC
(Montgomery, TX)
897
CORPORATE
Magnaflux (Glenview, IL)
Magnetic Analysis Corp. (Elmsford, NY)
Maintenance & Inspection Services,
Inc. (Morganton, NC)
Manadher Al Sahra Trading (Ruwi,
Oman)
Mandina’s Inspection Services, Inc.
(Belle Chasse, LA)
Mansol Technical Training Services
Pvt., Ltd. (Lahore, Pakistan)
Marietta Nondestructive Testing,
Inc. (Marietta, GA)
Martin Testing Laboratories, Inc.
(McClellan, CA)
Matec Instrument Cos., Inc.
(Northborough, MA)
McNDT Pipeline, Ltd. (Channahon, IL)
Medical Intubation Technology Corp.
(Taoyuan City, Taiwan)
Merrill Technologies Group
(Saginaw, MI)
Metafix (Lachine, Canada)
Metal Fatigue Solutions (Las Vegas, NV)
Metals Testing Co. (South Windsor, CT)
Metalscan Inspection Services
(Chennai, India)
Met-L-Chek (Santa Monica, CA)
Meyer Tool, Inc. (Cincinnati, OH)
MFE Enterprises, Inc. (Dripping
Springs, TX)
MFE Rentals (Pasadena, TX)
Miami NDT, Inc. (Doral, FL)
Milan Tool Corp (Cleveland, OH)
Minton Treharne & Davies, Ltd.
(Cardiff, United Kingdom)
MIR Engineering (Tangerang
Selatan, Indonesia)
MISTRAS Group, Inc. (Princeton
Junction, NJ)
Mitchell Laboratories (Pico Rivera, CA)
Modal Shop, Inc. (Cincinnati, OH)
Moraine Valley Community College
(Palos Hills, IL)
Morex 71, Ltd. (Even Yehuda, Israel)
Motabaqah Brand of Saudi
Specialized Laboratories Co
(Riyadh, Saudi Arabia)
Mountain Pressure Testing
(Longview, TX)
moviTherm (Irvine, CA)
Mozzat Enterprise Sdn. Bhd. (Kuala
Belait, Brunei)
MPM Products, Inc. (Ontario, CA)
MQ TEC (Hyderabad, India)
MR Chemie, GmbH (Unna, Germany)
MSPEC (Mussafah, United Arab
Emirates)
MTEC Mechanical Testing Services,
Inc. (Houston, TX)
Mudiame Welding Institute, Ltd.
(Port Harcourt, Nigeria)
Multicare Safety & Industrial
Inspections LLC (Dubai, United
Arab Emirates)
Mytechnic Mro (Istanbul, Turkey)
N Omni Metal Finishing, Inc. (Fountain
Nafto Serv (Istanbul, Turkey)
Nanjing Tycho Information Technology
Co., Ltd. (Beijing, China)
National Oilwell Varco (Singapore,
Singapore)
898 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
partners
National University Polytechnic
Institute (San Diego, CA)
Naya Engineering Services
(Basra, Iraq)
NDE Professionals, Inc. (Portland, OR)
NDE Solutions, LLC (Bryan, TX)
NDT & Corrosion Control Services
(Dammam, Saudi Arabia)
NDT Classroom, Inc. (Buffalo, NY)
NDT Italiana, Srl. (Concorezzo, Italy)
NDT Phils Inspection Services Corp.
(Paranaque, Philippines)
NDT Seals, Inc. (Houston, TX)
NDT Solutions (New Richmond, WI)
NDT Testing, Srl. (Odobesti,
Romania)
NDT Training & Testing Center
(Houston, TX)
NDT-PRO Services (Houston, TX)
NdtXducer, LLC (Northborough, MA)
New Tech Systems (Mansfield, TX)
Newco, Inc. (Florence, SC)
Newport News Shipbuilding
(Newport News, VA)
Nexco-West USA (Vienna, VA)
NextNDT (Redwood City, CA)
Nikon Metrology (Brighton, MI)
Nondestructive Inspection Service,
Inc. (Hurricane, WV)
Norfolk Naval Shipyard
(Portsmouth, VA)
North Idaho College / Aerospace
Technology (Hayden, ID)
North Star Imaging, Inc. (Rogers, MN)
North Texas Calibration (Fort Worth, TX)
Northbridge Energy, Ltd. (Igando,
Nigeria)
Northeast Testing & Mfg, LLC
(Beverly, MA)
NOVO DR, Ltd. (Yehud, Israel)
NQS Inspection, Ltd. (Corpus
Christi, TX)
Nucleom, Inc. (Quebec, Canada)
NXT NDT, Inc. (North Adams, MA)
O Pragma (Quebec, Canada)
Obelix Maintenance & Inspection
Services SAC, (Arequipa, Peru)
Ocean Corp. (Houston, TX)
Oceaneering (Panama City, FL)
Oceaneering International Asset
Integrity (Houston, TX)
Oceanscan USA (Stafford, TX)
OcerT Training & Consulting, Ltd.
(Jianding New City, China)
Ogden Weber Applied Technology
College (Ogden, UT)
OGTC Pvt., Ltd. (Islamabad, Pakistan)
Ohio CAT (Broadview Heights, OH)
Ohmstede, Ltd. (Corpus Christi, TX)
Oil & Gas Educational and Training
Center (Kayamkulam, India)
Oilfield Equipment Marketing, Inc.
(San Antonio, TX)
OKOS Solutions, LLC (Manassas, VA)
Olympus America, Inc. (Waltham, MA)
Valley, CA)
Opgal (Karmiel, Israel)
Optim, LLC (Sturbridge, MA)
Oracle, Ltd. (Apapa Lagos, Nigeria)
Orbit Industries, Inc. (Cleveland, OH)
OSG Testing Pty., Ltd. (Alberton
Johannesburg, South Africa)
P QSA Global, Inc. (Baton Rouge, LA)
Paragon Industries, Inc. (Sapulpa, OK)
Parker Research Corp. (Dunedin, FL)
Patterson Tubular Services
(Channelview, TX)
PCA Engineering, Inc. (Pompton
Lakes, NJ)
PdM Consultores Internacional, Srl.
(Cartago, Costa Rica)
Peak NDT (Derby, United Kingdom)
Peltec Services, Inc. (Darlington, SC)
Pemo Nigeria, Ltd. (Warri, Nigeria)
Performance Review Institute
(Warrendale, PA)
Petra Engineering for Inspection and
Services (Amman, Jordan)
Petro Base, Ltd. (Richmond, TX)
Petro Service International (Cairo,
Egypt)
PetroKnowledge (Masdar City,
United Arab Emirates)
PetroScanalog International, Ltd.
(Port Harcourt, Nigeria)
Pfinder KG (Boeblingen, Germany)
PFL Engineering Services, Ltd.
(Lekki, Nigeria)
PH Tool Reference Standards
(Pipersville, PA)
Phateco Technical Services Joint
Stock Co. (Hai Phong City, Vietnam)
Phoenix Inspection Systems, Ltd.
(Warrington, United Kingdom)
Phoenix Nuclear Labs (Monona, WI)
Pine (Windsor, NJ)
Pinnacle X-Ray Solutions
(Suwanee, GA)
Pipe & Well O&M Services Est
(Dammam, Saudi Arabia)
Plant Integrity, Ltd. (Cambridge,
United Kingdom)
PM Testing Laboratory, Inc. (Fife, WA)
Poco Graphite (Decatur, TX)
Portsmouth Naval Shipyard
(Portsmouth, NH)
PQNDT, Inc. (Arlington, MA)
Precision Flange & Machine, Inc.
(Houston, TX)
Premier NDT Services, Inc.
(Farmington, NM)
Premier Tubular Inspection Services
Pte., Ltd. (Karachi, Pakistan)
Premium Inspection & Testing
(Houston, TX)
Premium Inspection Co.
(Bakersfield, CA)
Prime NDT Services, Inc.
(Schnecksville, PA)
PRL Industries, Inc. (Cornwall, PA)
Pro Mag Inspection, LLC (Gray, LA)
PSSI NDT (League City, TX)
PT Karsa Kencana Indonesia
(Tangerang, Indonesia)
PT Radiant Utama Interinsco Tbk.
(Jakarta Selatan, Indonesia)
Pueblo Community College
(Pueblo, CO)
Q Lake City, UT)
QA Systems Pte., Ltd. (Singapore,
Singapore)
Qass Technologies (Marietta, GA)
QinetiQ NDT Pty., Ltd. (South
Melbourne, Australia)
Q-Sea Corporation (Tampa, FL)
Qualitas Material Testing Laboratories
(United Arab Emirates)
Qualitek, LLC (Houston, TX)
Quality Control Co. (Cairo, Egypt)
Quality Control Council US (Kansas
City, KS)
Quality Control Services Co., Ltd. (Al
Khobar, Saudi Arabia)
Quality Equipment Distributors, Inc.
(Orchard Park, NY)
Quality NDE, Ltd. (Mercier, Canada)
Quality Systems International, Inc.
(Russellville, AR)
Quality Testing Services, Inc.
(Linden, NJ)
Quality Testing Services, Inc.
(Maryland Hts, MO)
Qualtech NDE (Karachi, Pakistan)
R
RADAC -Radiographic Accessories,
Ltd. (Stockton On Tees, United
Kingdom)
RadiaBeam Technologies (Santa
Monica, CA)
Radiago Work Solutions Pvt. Ltd.
(Navi Mumbai, India)
Ram Designs (Broussard, LA)
Rayan Petro Azmoon (Tehran, Iran)
Ray-Check MFG, Inc. (Clovis, CA)
R-CON NDT, Inc. (Menomonie, WI)
Redi Inspection Services
(Evanston, WY)
Regional Utility Services
(Spartanburg, SC)
Reinhart & Associates, Inc.
(Austin, TX)
Rel Inc. (Calumet, MI)
RF System Lab (Traverse City, MI)
Ridgewater College (Hutchinson, MN)
Rig Fab Energy Services (Baton
Rouge, LA)
Rig Solution Engineering (Cairo,
Egypt)
Ritec, Inc. (Warwick, RI)
Rockwood Service Corp. (Danbury, CT)
Rohmann Eddy Current Instruments
& Systems (Spartanburg, SC)
Rokaysan Engineering Ltd., Co.
(Bursa, Turkey)
Rolls-Royce Nuclear Field Services
(Williamson, NY)
Rosen (Shah Alam, Malaysia)
RTW Roentgen-Technik
(Neuenhagen, Germany)
Russell NDE Systems, Inc.
(Edmonton, Canada)
RusselSmith Nigeria, Ltd. (Lagos,
Nigeria)
S
Safe Ocean Service, Inc. (Houston, TX)
Safe Systems For Advanced
Technologies (Jeddah, Saudi Arabia)
SAI Global (Paramus, NJ)
Salt Lake Community College (Salt
SAMES KREMLIN (Streamwood, IL)
Sarl 3MECS Engineering &
Consulting Services (Laghouat,
Algeria)
ScanMaster IRT, Inc. (Greenville, SC)

ScanTech Instruments, Inc.
(Longview, TX)
SCC Inspection Services (Great Falls,
MT)
School of Applied Non Destructive
Examination (Boksburg, South
Africa)
SCI Control & Inspeccion (Ajalvir,
Spain)
Score QC (Trivandrum, India)
SE International, Inc.
(Summertown, TN)
SEAL Aviation (Fort Lauderdale, FL)
Sense of Siam International Trading
Co., Ltd. (Sattahip, Thailand)
Sensima Inspection (Gland,
Switzerland)
Sensor Networks (Boalsburg, PA)
Sensors & Software, Inc.
(Mississauga, Canada)
Setcore Arabia Petroleum Services
(Dammam, Saudi Arabia)
SGS Transportation (Cincinnati, OH)
Shanghai CHiNDT Systems and
Services (Shanghai, China)
Shanghai Puxian Mechanical
Technology Co., Ltd. (Shanghai,
China)
Shanghai QiJi Inspection Technology
Co. (Shanghai, China)
Sherwin, Inc. (South Gate, CA)
Siemens Energy, Inc. (Pittsburgh, PA)
Sigma NDT Services Pvt., Ltd.
(Chennai, India)
Signature TechnicAir (Greensboro, NC)
Silean (Tremonton, UT)
Silverwing (Swansea, United
Kingdom)
SIUI (Shantou, China)
Sky Prime Aviation Services (Riyadh,
Saudi Arabia)
SME (Plymouth, MI)
Snell Group (Barre, VT)
SOLEND (Cochabamba, Bolivia)
Son Set Consultants, LLC (Owasso, OK)
Sonartech (Kempton Park, South
Africa)
Sonaspection International, Inc.
(Concord, NC)
Sonatest, Ltd. (San Antonio, TX)
Sonic Systems International
(Houston, TX)
Sonotec (Central Islip, NY)
Source Production & Equipment Co.,
Inc. (St. Rose, LA)
Southern Inspection Services
(Chennai, India)
Southwest Research Institute (San
Antonio, TX)
Sowsco Inspection Services, Ltd.
(Port Harcourt, Nigeria)
Sparrows (Bridge of Don Aberdeen,
United Kingdom)
Spartan College of Aeronautics &
Technology (Tulsa, OK)
Special Oilfield Services Co., LLC
(Ruwi, Oman)
Specialized Services & Supplies
General Trading & Contracting Co.
(North Shuwaikh, Kuwait)
Specpro (Santiago, Chile)
Spectronics Corp. (Westbury, NY)
Spellman High Voltage Electronics
Corp. (Hauppauge, NY)
Join Us
Being a part of the Society links your business to the worldwide
NDT community and puts your business on the front lines of the
industry. To learn more about becoming a Corporate Partner, see
the Membership section of the ASNT website at www.asnt.org.
ST Aerospace Engineering Pte., Ltd.
(Singapore, Singapore)
St. Johns NDT Training & Services
(Pathanamthitta, India)
Stalion-Primi (Port Harcourt, Nigeria)
Standard Testing and Inspection
Services, Ltd. (Port Harcourt,
Nigeria)
Stanley Inspection (Houston, TX)
Star Pipe Service, Inc. (Moore, OK)
Steinol Solutions Pvt., Ltd.
(Islamabad, Pakistan)
Stowen (Katy, TX)
Structural Integrity Associates
(Huntersville, NC)
Sullivan & Associates, Inc.
(Ladson, SC)
Superior Inspection Services, LLC
(Broussard, LA)
Supervisor of Shipbuilding
Conversion and Repair (Bath, ME)
System One (Cheswick, PA)
T Tulsa Welding School (Tulsa, OK)
Target NDT (Mangaf, Kuwait)
TCR Arabia Co., Ltd. (Dammam,
Saudi Arabia)
Team Industrial Services (Alvin, TX)
Team QualSpec (Torrance, CA)
Tech Service Products, Inc.
(Harahan, LA)
Tech4Imaging (Columbus, OH)
Techinco (Tehran, Iran)
TechKnowServ Corp. (State College,
PA)
Techna NDT (Kent, WA)
Technical Loadarm, Ltd. (Guelph,
Canada)
Technisonic Research, Inc. (Fairfield,
CT)
Technology Design Ltd. (Winsford,
United Kingdom)
Technoscan Inspection Services
(Pathanamthitta, India)
Techshore Inspection Services
(Cochin, India)
Techstreet (Ann Arbor, MI)
Tecnatom, S.A. (Madrid, Spain)
Tecnoplet, S.A. (Francisco de
Orellana, Ecuador)
Teledyne DALSA (Santa Clara, CA)
Teledyne ICM (Andrimont, Belgium)
Tesco Corp. (Kanagawa, Japan)
Test Equipment Distributors, LLC
(Troy, MI)
Test NDT, LLC (Brea, CA)
Testex, Inc. (Pittsburgh, PA)
Testia, Ltd. (Newport, United
Kingdom)
Testing Equipment Specialist Team
Co. (Abqaiq, Saudi Arabia)
Testing Service Group SAC (Lima,
Peru)
Texas Research International
(Austin, TX)
Thermal Wave Imaging, Inc.
(Ferndale, MI)
ThermalStar Training Center
(Phoenix, AZ)
Thermographie GG, Inc. (Granby,
Canada)
TIBA Oil Tools (Cairo, Egypt)
TMAC Technologies (Mansfield, TX)
TP Group, S.A. (Bogota, Colombia)
Trainee World Institute (Baghdad,
Iraq)
Tribogenics (Los Angeles, CA)
Trident Refit Facility (Kings Bay, GA)
Trinity NDT Engineers (Bangalore,
India)
Tru Amp Corp. (Jackson, MS)
TSC Inspection Systems
(Buckinghamshire, United
Kingdom)
TTAsia Co., Ltd. (Ho Chi Minh City,
Vietnam)
Turbo Nondestructive Testing, Inc.
(Kemah, TX)
Turnco, LLC (Houston, TX)
Turner Specialty Services, LLC
(Pasadena, TX)
TWI, Ltd. (Cambridge, United
Kingdom)
U Weldtest (Bir Khadem, Algeria)
Ultrasonic Sciences, Ltd. (Aldershot,
United Kingdom)
Uniclimb Services Pte., Ltd.
(Singapore, Singapore)
Unitek Engineering (Gangnam-gu,
South Korea)
University of Alaska Anchorage
(Anchorage, AK)
UniWest (Pasco, WA)
US Army Yuma Proving Ground
(Yuma, AZ)
US Photon Service (Hayward, CA)
USPS NCED (Norman, OK)
UT Quality (Edmonton, Canada)
Utex Scientific Instruments, Inc.
(Mississauga, Canada)
UTX, Inc. (Holmes, NY)
V Yxlon (Hudson, OH)
VAAL University of Technology
(Vanderbijlpark, South Africa)
Valley Inspection Service, Inc.
(Allentown, PA)
Vandergriff Technologies NDT
Services (Haltom City, TX)
Varex Imaging (Salt Lake City, UT)
Vector TUB, GmbH (Hattingen,
Germany)
Velosi, Sdn. Bhd. (Kuala Belait,
Brunei)
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N
Venture Inspection, Ltd. (Derby,
United Kingdom)
Verasonics, Inc. (Kirkland, WA)
Verichek Technical Services, Inc.
(Bethel Park, PA)
Versa Integrity Group (Sulphur, LA)
Vibranalysis, Inc. (Trujillo Alto,
Puerto Rico)
Vibspectrum International, LLC
Trading & Electromechanical
(Dubai, United Arab Emirates)
Vidisco, Ltd. (Or Yehuda, Israel)
Vincotte International Algeria
(Mahelma, Algeria)
Vincotte International Middle East,
LLC (Abu Dhabi, United Arab
Emirates)
Virgin Galactic (Mojave, CA)
Virtual Media Integration
(Pensacola, FL)
VisiConsult X-Ray Systems &
Solutions, GmbH (Stockelsdorf,
Germany)
Vision Financial Group, Inc.
(Pittsburgh, PA)
Vizaar Industrial Imaging
(Gibsonia, PA)
VJ Technologies, Inc. (Bohemia, NY)
Volume Graphics, Inc. (Charlotte, NC)
Volunteer NDT Corp. (Chattanooga, TN)
W
Walashek Industrial & Marine, Inc.
(National City, CA)
Warren Associates (Pittsburgh, PA)
Wasatch Photonics (Logan, UT)
Washita Valley Enterprises, Inc.
(Oklahoma City, OK)
Welding and Control (Oran, Algeria)
Welding Inspection (Cologne, NJ)
Welding Technology & NDT Research
Application Center (Ankara, Turkey)
WENS Quality Assurance Pvt., Ltd.
(Singapore, Singapore)
WesDyne Amdata (Windsor, CT)
West Penn Testing Group (New
Kensington, PA)
Westminster Education Circle
(Ernakulam, India)
Williams Bridge Co. (Richmond, VA)
Willick Engineering Co., Inc. (Santa
Fe Springs, CA)
World Testing, Inc. (Mt. Juliet, TN)
Wyle (Dayton, OH)
X
X-Ray Associates, LLC (San Dimas, CA)
X-Ray Industries, Inc. (Troy, MI)
Y
Z
Zamil Lifting & Industrial Supports
(Dammam, Saudi Arabia)
Zetec, Inc. (Snoqualmie, WA)
Zuuk International, Inc. (
Charleston, SC) wx
899
 Correction
Correction
On page 502 of the April 2017 issue of Materials Evaluation,
Figure 1 in “A New Detection Sensor for Wire Rope Based on
Open Magnetization Method” was incorrectly captioned. The
correction appears below.
Materials Evaluation regrets this error.

Ferrous yoke
Magnetic-yoke Permanant
testing package magnet
stack II

Permanant
magnet
stack I
Magnetic
sensing unit
Defect

Tested wire rope

Figure 1. Testing principles based on yoke magnetizing

method for wire rope.

900 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
meetings
Meetings are events at which
paper and/or poster presentations
are made and recent developments
in technology, research and devel-
opment are discussed by those in
attendance. These are generally
sponsored by academic or profes-
sional technical associations. The
sponsor is the same as the contact
except where noted.
For ASNT meetings and events
(highlighted in red) contact the
ASNT Conference Department,
1711 Arlingate Lane, P.O. Box
28518, Columbus, OH 43228-
0518; (800) 222-2768 or (614)
274-6003; fax (614) 274-6899;
e-mail conferences@asnt.org.
2017
16–21 JUL
44th Annual Review of
Progress in Quantitative
Nondestructive Evaluation,
Utah Valley Convention Center,
Provo, Utah. Contact: Kay
Williams; (515) 294-9749; fax
(515) 294-7771; e-mail registra-
tions@iastate.edu; website
www.qndeprograms.org.
5–7 SEP
56th Annual Conference of the
British Institute of Non-
Destructive Testing, The
International Centre, Telford,
UK. Contact: Karen Cambridge;
44 01604 438300; fax 44
01604 438301; e-mail
conf@bindt.org; website
www.bindt.org.
5–7 SEP
Materials Testing 2017, The
International Centre, Telford,
UK. Contact: Karen Cambridge;
44 01604 438300; fax 44
01604 438300; e-mail
mt2017@bindt.org; website
www.materialstesting.org.
PLEASE NOTE: Materials Evaluation’s Calendar department is
The following courses are listed
without necessarily giving their full
titles.
derived from information sent to our offices by the sponsoring
organizations. ASNT staff is not responsible for collecting or
verifying the information contained herein: for more information Acoustic Emission Testing
on meetings or courses, please contact the sponsoring organi-
12–14 SEP
zation. The Calendar copy deadline is the first of the month, two
months prior to the issue date: for example, 1 August for the For Scientists and Engineers,
October journal. Send your organization’s information by e-mail, Princeton Junction, New Jersey.
fax, or mail to the Associate Editor, Materials Evaluation, 1711 Mistras.
Arlingate Lane, P.O. Box 28518, Columbus, OH 43228-0518; fax 9–13 OCT
(614) 274-6899; e-mail hcowans@asnt.org. Information in the Level I, Princeton Junction, New
Calendar runs for four months at a time. ASNT reserves the right Jersey. Mistras.
to reject event listings for any reason. Listings will be edited to
conform to ASNT’s editorial style. Electromagnetic Testing
29 JUN–3 JUL
25–29 SEP 2018 Level I/II (SNT-TC-1A),
19th Schweissen & Schneiden, Bangalore, India. Trinity.
Düsseldorf Exhibition Ground, 28–31 OCT
1–8 JUL
Düsseldorf, Germany. Contact: ASNT Annual Conference,
George R. Brown Convention Level I/II, Trivandrum, India.
Messe Essen; 49 0 201 724
Center, Houston, Texas. Decibel Remote.
529/648; fax 49 0 201 72 44
Contact: ASNT. 3–15 JUL
448; e-mail team@schweissen-
Level I/II, Kerala, India.
schneiden.com; website
2019 Decibel.
www.schweissen-
schneiden.com. 10–14 JUL
10–14 NOV
10–11 OCT ASNT Annual Conference, Level I, Atlanta, Georgia. ATS.
Offshore Energy Exhibition & Westgate Resort & Casino, Las 10–22 JUL
Conference, Amsterdam RAI, Vegas, Nevada. Contact: ASNT. Level I/II, Kerala, India.
Amsterdam, the Netherlands.
Contact: Navingo BV; 31 010 courses Decibel.
17–21 JUL
20 92 600; fax 31 010 43 68 Courses are events where partici-
pants are instructed in the tech- Level II, Atlanta, Georgia. ATS.
134; website www.offshore-
nologies and methodologies of a
energy.biz. 17–29 JUL
particular technical area and which
10–13 OCT generally conclude with the Level I/II, Kerala, India.
World Conference on Acoustic student being evaluated to Decibel.
determine the student's retention
Emission 2017, Xi’an, Shanxi, of the material presented. These 27–31 JUL
China. Contact: Zhanwen Wu; events often offer some form of Level I/II (SNT-TC-1A),
86 1 59068313; fax 86 10 course credit or continuing Bangalore, India. Trinity.
59068023; e-mail education units to those partici-
pants successfully completing the 24–28 JUL
wcae2017@163.com; website
course. For ASNT refresher Eddy Current Level I, St. Louis,
www.wcacousticemission.org.
courses, see page 911. Missouri. Quality.
30 OCT–2 NOV ASNT neither approves nor
disapproves of any program or 31 JUL–4 AUG
ASNT Annual Conference,
Gaylord Opryland Resort and training course claiming to meet Eddy Current Level I, St. Louis,
Convention Center, Nashville,
the recommendations of ASNT’s Missouri. Quality.
Recommended Practice No.
Tennessee. Contact: ASNT. SNT-TC-1A. The following are 1–8 AUG
contacts for only those organiza- Level I/II, Trivandrum, India.
tions that offer public courses Decibel Remote.
listed in this month’s Calendar.
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 901
calendar
Course Contacts
The following are contacts for only those organizations that offer public courses listed in this month’s Calendar.
Atlantic: Atlantic NDT Training; Gary L.
Chapman; 24 Flat Rock Rd., Branford, CT
06405; (203) 481-4041; website
www.atlanticndttraining.com.
ATS: Applied Technical Services; Lisa
Henry; 1049 Triad Ct., Marietta, GA
30062; (888) 287-5227; (678) 444-
2897; fax (770) 514-3299; e-mail
lhenry@atslab.com; website www
.atslab.com/training.
BRL: BRL Consultants, Inc.; 219 W.
Rhapsody Dr., San Antonio, TX 78216;
(210) 341-3442; fax (210) 341-2844;
e-mail info@brlconsultants.com;
website www.brlconsultants.com.
Decibel: Decibel NDE Training Institute;
1st Floor Plainfield, Pattambi, Palakkad,
Kerala, India 679303; 91 9387 674
153, 91 466 2214 333, or 91 8089 000
250; e-mail info@decibelnde.com;
website www.decibelnde.com.
Decibel Remote: Decibel Remote
Training Center; TC No. 1/1374(12),
2nd Floor, Kottakath Bldg., Poonthi Rd.,
Kumarapuram, Trivandrum, India;
91 81 29508881; e-mail decibeltvm
@decibelnde.com; website www
.decibelnde.com.
ETS: ETS Sistemi Industriali, Srl.; Alberto
Monici; Via S. Francesco 323, 20861
Brugherio (MB), Italy; 39 039877790;
e-mail a.monici@etssistemi.it; website
www.etssistemi.it.
Extende: Extende, Inc.; P.O. Box 461,
Ballston Spa, NY 12020; (518) 490-
2376; fax (518) 602-1367; e-mail
contactus@extende.com; website
www.extende.com.
Extende France: Extende; Le Bergson,
15 Ave. Emile Baudot, 91300 Massy,
France; 33 1 78 90 02 21; fax 33 09 72
13 42 68; e-mail contact@extende.com;
website www.extende.com.
GINDT: Gavade Institute of
Nondestructive Testing & Training;
Sunil Baburao Gavade; Plot No. 29/B,
Swatantravir Sawarkar Marg, Opposite
Gopinath Patil Parsik Janatha Sahakari
Bank, Sector 1, Airoli, Navi Mumbai
400 708, Maharashtra State, India; 091
99 67581383; fax 091 22 27793512;
e-mail infogindt@gmail.com; website
www.gindt.co.in.
902 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Kraft: Kraft Technology Resources; Karl
E. Kraft; 1377 Timshel St., Dayton, OH
45440; (405) 819-7786; fax (405) 691-
4342; e-mail kraftndt@aol.com; website
www.ndtbootcamp.com.
LTS: Leak Testing Specialists, Inc.;
Cyndi Reid; 5776 Hoffner Ave., Ste. 304,
Orlando, FL 32822; (407) 737-6415;
fax (407) 737-6416; e-mail cyndi.reid
@leaktestingspec.com; website
www.leaktestingspec.com.
MFE: MFE Enterprises, Inc., 150 Holder
Ln., Dripping Springs, TX 7862; (281)
441-8284; e-mail bduke@mferentals
.com; website mferentals.com/training.
Mistras: Mistras Group, Inc.; Christina
Librandy; 195 Clarksville Rd., Princeton
Junction, NJ 08550; (609) 716-4020;
fax (609) 716-0706; e-mail christina
.librandy@mistrasgroup.com; website
www.mistrasgroup.com.
Moraine: Moraine Valley Community
College; 9000 W. College Pkwy., Palos
Hills, IL 60465, M150; (708) 974-5735;
e-mail ccce@morainevalley.edu; website
www.morainevalley.edu/ccce/contin-
uing/nondestructive-testing.
MPM: MPM Products, Inc.; Jeri Matza;
1718 East Grevillea Ct., Ontario, CA
91761; (918) 740-0290 or (800) 429-
0128; fax (909)-947-3257; e-mail
jerimatza@mpmproducts.com; website
www.mpmproducts.com.
Odyssey: Odyssey Technology Corp.;
Carol Sansieri; 3000 Village Run Rd.,
Unit 103, #149, Wexford, PA 15090;
(843) 817-3475; e-mail carols@odyssey
test.com.
PQT: PQT Services (Plumstead Training);
Kim Rosa; 806 Botany Rd., Greenville,
SC 29615; (864) 292-1115; fax (770)
514-3299; e-mail krosa@atslab.com;
www.pqt.net.
QCTL: QCTL, Inc.; Rod Reinholdt or
David White; 21112 Scott Park Rd.,
Davenport, IA 52804; (800) 391-8500;
fax (563) 391-0112; e-mail testlab1
@att.net; website www.testlab1.com.
Quality: Quality Testing Services;
Melissa Rankin; 2305 Millpark Dr.,
Maryland Heights, MO 63043; (314)
770-0607; (888) 770-0607; fax (314)
770-0103; e-mail training@quality
testing.com; website www.quality
testing.com.
Snell: The Snell Group; 322 N. Main St.,
Ste. 8, Barre, VT 05641; (802) 479-
7100; fax (802) 479-7171; e-mail
info@thesnellgroup.com; website
www.thesnellgroup.com.
Test: Test NDT; Richard Harrison; 193
Viking Ave., Brea, CA 92821; (714) 255-
1500; fax (714) 255-1580; e-mail
ndttrain@aol.com; website
www.testndt.com.
Trinity: Trinity Institute of NDT
Technology; Ravi Kumar T. or Shiva
Kumar R.; Plot No. V-22a, 2nd Stage,
Peenya Industrial Estate, Bangalore,
India 560058; 91 99009 29439 or
91 98441 29439; e-mail training
@trinityndt.com; website www
.trinityndt.com.
Ultrasonics: University of Ultrasonics
/NDEU; Ron Kent; P.O. Box 328 Elkton,
FL 32033; (205) 822-5203; e-mail
info@nde.university; website
www.nde.university.
WTTI: Welder Training and Testing
Institute; Tracy Wiswesser; 1144 N.
Graham St., Allentown, PA 18109;
(800) 223-9884; e-mail tracy
@welderinstitute.com; website
www.wtti.edu.
XRI: X-Ray Industries; Bob Henchar;
1961 Thunderbird Dr., Troy, MI
48084; (248) 244-1582; e-mail
bob.henchar@xrayindustries.com;
website www.xrayindustries.com.
Electromagnetic Testing, cont. 16–20 OCT
Eddy Current Level II,
1–12 AUG Pittsburgh, Pennsylvania.
Level I/II, Kerala, India. Odyssey.
Decibel.
16–28 OCT
7–11 AUG Level I/II, Kerala, India.
Level I, Atlanta, Georgia. ATS. Decibel.
7–19 AUG 23–27 OCT
Level I/II, Kerala, India. Eddy Current Level I, San
Decibel. Antonio, Texas. BRL.
14–18 AUG Eddy Current Level I, St. Louis,
Level II, Atlanta, Georgia. ATS. Missouri. Quality.

14–26 AUG 27–31 OCT

Level I/II, Kerala, India. Level I/II (SNT-TC-1A),
Decibel. Bangalore, India. Trinity.

15–17 AUG 30 OCT–3 NOV

Eddy Current Recurrent San Eddy Current Level II, San
Antonio, Texas. BRL. Antonio, Texas. BRL.
Eddy Current Level II, St. Louis,
1–5 SEP Missouri. Quality.
Level I/II (SNT-TC-1A),
Bangalore, India. Trinity. Infrared and Thermal
1–8 SEP Testing
Level I/II, Trivandrum, India.
Decibel Remote. 10–13 JUL
Level I Buildings, Toronto,
4–16 SEP Ontario, Canada. Snell.
Level I/II, Kerala, India.
Decibel. 10–14 JUL
Level I, Barre, Vermont. Snell.
11–23 SEP
Level I/II, Kerala, India. 17–21 JUL
Decibel. Level I, Burlington, New Jersey.
Infraspection.
12–14 SEP Level I/II, Kerala, India.
Eddy Current Recurrent, San Decibel.
Antonio, Texas. BRL. Level II, Barre, Vermont. Snell.
18–30 SEP 24–28 JUL
Level I/II, Kerala, India. Level I/II, Kerala, India.
Decibel. Decibel.
28 SEP–2 OCT NDT Level II, Detroit, Michigan.
Level I/II (SNT-TC-1A), Snell.
Bangalore, India. Trinity. 25–27 JUL
1–8 OCT For Rail Industry, St. Louis,
Level I/II, Trivandrum, India. Missouri. Quality.
Decibel Remote. 7–16 AUG
2–6 OCT Level I/II, Trivandrum, India.
Level I, Atlanta, Georgia. ATS. Decibel Remote.
2–14 OCT 8–11 AUG
Level I/II, Kerala, India. Level I, St. Louis, Missouri.
Decibel. Quality.
9–13 OCT 14–18 AUG
Level II, Atlanta, Georgia. ATS. Level I, St. Louis, Missouri.
Quality.
9–21 OCT Level I, Toronto, Ontario,
Level I/II, Kerala, India. Canada. Snell.
Decibel. Level I/II, Kerala, India.
Decibel.

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 903
calendar
Infrared and Thermal Testing, cont. Level I, Toronto, Ontario, Liquid Penetrant Testing Level I/II, Allentown,
Canada. Snell. Pennsylvania. WTTI.
Level I/II, Kerala, India. 3–7 JUL
21–25 AUG 7–11 AUG
Decibel. Level I/II, Kerala, India.
Level I, Indianapolis, Indiana. Level I/II, Kerala, India.
Decibel.
Snell. 18–19 OCT Decibel.
Level I/II, Kerala, India. Mechanical Specialty Course, 7–12 JUL
Decibel. Level I/II, Trivandrum, India. 7–12 AUG
Charlotte, North Carolina. Level I/II, Trivandrum, India.
Level II, Indianapolis, Indiana. Snell. Decibel Remote.
Snell. Decibel Remote.
23–27 OCT 10–11 JUL
11–14 SEP Level I/II, Greenville, South 14 AUG
Level I, Tampa, Florida. Snell.
Level I, Atlanta, Georgia. ATS. Carolina. PQT. Recurrent, San Antonio, Texas.
Level I/II, Kerala, India. BRL.
11–15 SEP Decibel. 10–14 JUL
Level I, Burlington, New Jersey. Level II, Montreal, Quebec, Level I/II, Kerala, India. 14–18 AUG
Infraspection. Canada. Snell. Decibel. Level I/II, Kerala, India.
Level I, Denver, Colorado. Level I/II (NAS-410), Branford, Decibel.
30 OCT–3 NOV
Snell. Level I, Burlington, New Jersey. Connecticut. Atlantic. 19–24 AUG
Level I, Minneapolis, Infraspection. 13–14 JUL Level I/II, Trivandrum, India.
Minnesota. Snell. Level II, Tampa, Florida. Snell. Level II (SNT-TC-1A), Brea, Decibel Remote.
Level I, Montreal, Quebec, California. Test. 21–22 AUG
Canada. Snell.
Leak Testing 17–18 JUL Level I/II, Atlanta, Georgia.
18–22 SEP Level I/II, Palos Hills, Illinois. ATS.
17–21 JUL
Level I, Chicago, Illinois. Snell. Moraine. Level I/II, Jacksonville, Florida.
Level I, Seattle, Washington. Level I/II, Kerala, India.
Decibel. Level I/II, Jacksonville, Florida. PQT.
Snell. PQT. 21–25 AUG
Level I/II, Kerala, India. 24–30 JUL
Decibel. 17–21 JUL Level I/II, Kerala, India.
Level I/II, Trivandrum, India.
Level II, Atlanta, Georgia. ATS. Level I/II, Kerala, India. Decibel.
Decibel Remote.
Level II, Burlington, New Decibel. 23–24 AUG
14–18 AUG
Jersey. Infraspection. 18–21 JUL Level I/II (SNT-TC-1A),
Level I/II, Kerala, India.
Level II, Seattle, Washington. Level I/II, St. Louis, Missouri. Bangalore, India. Trinity.
Decibel.
Snell. Quality. 25–30 AUG
21–25 AUG
25–29 SEP 19–20 JUL Level I/II, Trivandrum, India.
Mass Spectrometer Level I/II,
Level I, Calgary, Alberta, Level I/II (SNT-TC-1A), Decibel Remote.
Orlando, Florida. LTS.
Canada. Snell. Bangalore, India. Trinity. 4–8 SEP
Level I, Little Rock, Arkansas. 28 AUG–1 SEP
19–24 JUL Level I/II, Kerala, India.
Snell. Pressure Change Measurement
Level I/II, Trivandrum, India. Decibel.
Level I/II, Kerala, India. Level I/II, Bubble Level I/II, &
Decibel. Level III Prep, Orlando, Florida. Decibel Remote. 5–8 SEP
LTS. 24–25 JUL Level I/II, St. Louis, Missouri.
2–6 OCT
1 SEP Level I/II, Atlanta, Georgia. Quality.
Level I, Cincinnati, Ohio. Snell.
Level II, Pittsburgh, Bubble Level I/II, St. Louis, ATS. 7–12 SEP
Pennsylvania. Snell. Missouri. Quality. 24–28 JUL Level I/II, Trivandrum, India.
11–15 SEP Level I/II, Kerala, India. Decibel Remote.
7–16 OCT
Level I/II, Trivandrum, India. Pressure Change Measurement Decibel. 11 SEP
Decibel Remote. Level I/II, Orlando, Florida. LTS. 25–30 JUL Level I, Davenport, Iowa. QCTL.
18–22 SEP Level I/II, Trivandrum, India. Recurrent, San Antonio, Texas.
10–13 OCT
Level I/II, Kerala, India. Decibel Remote. BRL.
Level I, St. Louis, Missouri.
Quality. Decibel. 31 JUL–3 AUG 11–15 SEP
Level II, Heath, Ohio. Mistras. Level I/II, Kerala, India.
16–17 OCT 24–30 SEP
Level I/II, Trivandrum, India. Decibel.
Electrical Specialty Course, 1–5 AUG
Charlotte, North Carolina. Decibel Remote. Level I/II, Kerala, India. 12 SEP
Snell. Decibel. Level II, Davenport, Iowa.
25–27 SEP
For engineers, Orlando, QCTL.
16–20 OCT 7–8 AUG
Level I, St. Louis, Missouri. Florida. LTS. Level I/II, San Antonio, Texas. 12–15 SEP
Quality. BRL. Level I/II (NAS-410), Troy,
16–20 OCT
Level I, Omaha, Nebraska. Level I/II, Kerala, India. Level I/II, Greenville, South Michigan. XRI.
Snell. Decibel. Carolina. PQT.

904 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
14–15 SEP 16–20 OCT 25–29 SEP 26–28 JUL
Level II (SNT-TC-1A), Brea, Level I/II, Kerala, India. Level I/II (SNT-TC-1A), Troy, Level I/II, Atlanta, Georgia.
California. Test. Decibel. Michigan. XRI. ATS.
18–19 SEP 18–19 OCT 9–13 OCT 1–5 AUG
Level I/II, Atlanta, Georgia. Level I/II (SNT-TC-1A), Level I/II, Allentown, Level I/II, Kerala, India.
ATS. Bangalore, India. Trinity. Pennsylvania. WTTI. Decibel.
Level I/II, Greenville, South 19–24 OCT 16–20 OCT 1–6 AUG
Carolina. PQT. Level I/II, Trivandrum, India. Level I/II, Branford, Level I/II, Trivandrum, India.
Level I/II, Jacksonville, Florida. Decibel Remote. Connecticut. Atlantic. Decibel Remote.
PQT.
23–27 OCT 23–30 OCT 7–11 AUG
18–22 SEP Practical, Kerala, India.
Level I/II, Kerala, India. Level I/II, Kerala, India.
Level I/II, Kerala, India. Decibel.
Decibel. Decibel.
Decibel.
25–30 OCT 9–11 AUG
20–21 SEP
Level I/II, Trivandrum, India. Magnetic Particle Testing Level I/II, San Antonio, Texas.
Level I/II (SNT-TC-1A),
Decibel Remote. 1–6 JUL BRL.
Bangalore, India. Trinity.
Level I/II, Trivandrum, India. Level I/II, Greenville, South
19–24 SEP Magnetic Flux Leakage Decibel Remote. Carolina. PQT.
Level I/II, Trivandrum, India. Testing Level I/II, Allentown,
Decibel Remote. 3–7 JUL Pennsylvania. WTTI.
17–29 JUL Level I/II, Kerala, India.
25–28 SEP Decibel. 13–18 AUG
Level I/II, Kerala, India.
Level I/II, Portland, Oregon. Level I/II, Trivandrum, India.
Decibel. 10–12 JUL
NPI. Decibel Remote.
7–11 AUG Level II (SNT-TC-1A), Brea,
25–29 SEP California. Test. 14–18 AUG
Level I/II, Dripping Springs,
Level I/II, Kerala, India. Level I/II, Kerala, India.
Texas. MFE. 10–15 JUL
Decibel. Decibel.
13–22 AUG Level I/II, Kerala, India. Level I/II (NAS-410), Branford,
25–30 SEP Decibel.
Level I/II, Trivandrum, India. Connecticut. Atlantic.
Level I/II, Trivandrum, India.
Decibel Remote. 11–14 JUL
Decibel Remote. 21–22 AUG
14–26 AUG Level I/II, St. Louis, Missouri. Level I/II (SNT-TC-1A),
2–6 OCT Quality.
Level I/II, Kerala, India. Bangalore, India. Trinity.
Level I/II, Kerala, India.
Decibel. 12–14 JUL
Decibel. 21–25 AUG
18–30 SEP Level I/II, Greenville, South Level I/II, Kerala, India.
7–12 OCT Carolina. PQT.
Level I/II, Kerala, India. Decibel.
Level I/II, Trivandrum, India.
Decibel. 13–18 JUL
Decibel Remote. 22–25 AUG
13–22 OCT Level I/II, Trivandrum, India. Level I/II, St. Louis, Missouri.
9–10 OCT Decibel Remote.
Level I/II, Trivandrum, India. Quality.
Level I/II, San Antonio, Texas.
Decibel Remote. 17–18 JUL
BRL. 23–25 AUG
Level I/II, Greenville, South 16–28 OCT Level I/II (SNT-TC-1A), Level I/II, Atlanta, Georgia.
Carolina. PQT. Level I/II, Kerala, India. Bangalore, India. Trinity. ATS.
Level I/II, Allentown, Decibel. 17–21 JUL Level I/II, Jacksonville, Florida.
Pennsylvania. WTTI. Level I/II, Kerala, India. PQT.
9–13 OCT Magnetic Particle and Decibel. 25–30 AUG
Level I/II, Kerala, India. Liquid Penetrant Testing 19–21 JUL Level I/II, Trivandrum, India.
Decibel. 21–30 JUL Level I/II, Palos Hills, Illinois. Decibel Remote.
10–13 OCT Practical, Kerala, India. Moraine. 1–6 SEP
Level I/II, St. Louis, Missouri. Decibel. Level I/II, Jacksonville, Florida. Level I/II, Trivandrum, India.
Quality. PQT. Decibel Remote.
7–11 AUG
16–17 OCT Level I/II, Allentown, 24–28 JUL 4–8 SEP
Level I/II, Atlanta, Georgia. Pennsylvania. WTTI. Level I/II, Kerala, India. Level I/II, Kerala, India.
ATS. Decibel. Decibel.
23–30 AUG
Level I/II, Jacksonville, Florida. Practical, Kerala, India. 25–28 JUL 11–13 SEP
PQT. Decibel. Level II, Heath, Ohio. Mistras. Level II (SNT-TC-1A), Brea,
16–18 OCT 21–30 SEP 25–30 JUL California. Test.
Level I/II (SNT-TC-1A), Practical, Kerala, India. Level I/II, Trivandrum, India. 11–15 SEP
Branford, Connecticut. Atlantic. Decibel. Decibel Remote. Level I/II, Kerala, India.
Decibel.

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 905
Magnetic Particle Testing, cont. 13–18 OCT 13–23 JUL Level II Non-film (NAS-410),
Level I/II, Trivandrum, India. Film Interpretation Level II, Greenville, South Carolina.
13–14 SEP
Decibel Remote. Trivandrum, India. Decibel PQT.
Level I, Davenport, Iowa. QCTL. 16–17 OCT
Remote. 24 JUL–1 AUG
13–18 SEP
Level I/II (SNT-TC-1A), 17–21 JUL Level II Film/Non-film
Level I/II, Trivandrum, India. Bangalore, India. Trinity. Computed Radiography Level I, (NAS-410), Greenville, South
Decibel Remote. 16–20 OCT
Greenville, South Carolina. Carolina. PQT.
Level I/II, Kerala, India. PQT. 24 JUL–2 AUG
15 SEP Digital Radiography Level I,
Level II, Davenport, Iowa. Decibel. Computed Radiography Level I,
Greenville, South Carolina. Atlanta, Georgia. ATS.
QCTL. 18–20 OCT PQT.
Level I/II, Atlanta, Georgia. Digital Radiography Level I,
18–19 SEP Level I, Greenville, South Atlanta, Georgia. ATS.
Level I/II (SNT-TC-1A), ATS. Carolina. PQT.
Level I/II, Jacksonville, Florida. Level I Film, Atlanta, Georgia.
Bangalore, India. Trinity. Level I, Allentown, ATS.
PQT. Pennsylvania. WTTI.
18–21 SEP Level I/II (SNT-TC-1A), Level I Film/Non-film
Level I, Troy, Michigan. XRI. (NAS-410), Atlanta, Georgia.
Level I/II, Portland, Oregon. Branford, Connecticut. Atlantic. Level I Film, Greenville, South
NPI. ATS.
23–27 OCT Carolina. PQT. Level I Non-film (NAS-410),
18–22 SEP Level I/II, Kerala, India. Level I Film to Non-film Atlanta, Georgia. ATS.
Level I/II, Kerala, India. Decibel. Transition (NAS-410),
Decibel. Greenville, South Carolina. 31 JUL–2 AUG
25–30 OCT PQT. Level I Non-film Transition
19–22 SEP Level I/II, Trivandrum, India. (NAS-410), Atlanta, Georgia.
Level I Non-film (NAS-410),
Level I/II (NAS-410), Troy, Decibel Remote. ATS.
Greenville, South Carolina.
Michigan. XRI.
30 OCT–1 NOV PQT. 31 JUL–4 AUG
20–22 SEP Level II (SNT-TC-1A), Brea, Level II, San Antonio, Texas. Computed Radiography
Level I/II, Atlanta, Georgia. California. Test. BRL. Level II, Atlanta, Georgia. ATS.
ATS. Level II, St. Louis, Missouri. Digital Detector Array
31 OCT–3 NOV
Level I/II, Greenville, South Quality. Radiography Level II, Troy,
Carolina. PQT. Level II, Heath, Ohio. Mistras.
17–25 JUL Michigan. XRI.
Level I/II, Jacksonville, Florida. Digital Radiography Level II,
Radiographic Testing Level I Film/Non-film
PQT. Atlanta, Georgia. ATS.
(NAS-410), Greenville, South
25–29 SEP 1–8 JUL Carolina. PQT. Level II, Atlanta, Georgia. ATS.
Level I/II, Kerala, India. Practical, Kerala, India. Radiation Safety, Palos Hills,
17–29 JUL
Decibel. Decibel. Illinois. Moraine.
Film Interpretation, Kerala,
25–30 SEP 1–10 JUL India. Decibel. 31 JUL–9 AUG
Level I/II, Trivandrum, India. Film Interpretation Level II, Level I/II, Kerala, India. Level II Film/Non-film
Decibel Remote. Trivandrum, India. Decibel Decibel. (NAS-410), Atlanta, Georgia.
Remote. ATS.
26–29 SEP 21–23 JUL
Level I/II, St. Louis, Missouri. 1–12 JUL Level I/II (SNT-TC-1A), 1–6 AUG
Quality. Level I/II, Trivandrum, India. Bangalore, India. Trinity. X-ray, Trivandrum, India.
Decibel Remote. Decibel Remote.
1–6 OCT 24–28 JUL
Level I/II, Trivandrum, India. 3–15 JUL Computed Radiography 1–12 AUG
Decibel Remote. Film Interpretation, Kerala, Level II, Greenville, South Film Interpretation, Kerala,
India. Decibel. Carolina. PQT. India. Decibel.
2–6 OCT
Level I/II, Kerala, India. Digital Radiography Level II, Level I/II, Kerala, India.
Level I/II, Kerala, India.
Decibel. Greenville, South Carolina. Decibel.
Decibel.
10–14 JUL PQT. Level I/II, Trivandrum, India.
8–10 OCT
Level I, San Antonio, Texas. Interpretation, San Antonio, Decibel Remote.
Level I/II, Greenville, South Texas. BRL. Practical, Kerala, India.
BRL.
Carolina. PQT. Level II, Greenville, South Decibel.
Level I, St. Louis, Missouri.
9–13 OCT Quality. Carolina. PQT. 7–9 AUG
Level I/II, Kerala, India. Radiation Safety, Palos Hills, Level II Film, Greenville, South Level II Non-film Transition
Decibel. Illinois. Moraine. Carolina. PQT. (NAS-410), Atlanta, Georgia.
Level II Film to Non-film ATS.
11–13 OCT 10–22 JUL Transition (NAS-410),
Level I/II, San Antonio, Texas. Film Interpretation, Kerala, Greenville, South Carolina. 7–11 AUG
BRL. India. Decibel. PQT. Level I, Davenport, Iowa. QCTL.
Level I/II, Allentown, Level I/II, Kerala, India. Level II, Troy, Michigan. XRI.
Pennsylvania. WTTI. Decibel.

906 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
7–16 AUG
Film Interpretation Level II,
Trivandrum, India. Decibel
Remote.
7–19 AUG
Film Interpretation, Kerala,
India. Decibel.
Level I/II, Kerala, India.
Decibel.
14–18 AUG
Computed Radiography Level I,
Greenville, South Carolina.
PQT.
Digital Radiography Level I,
Greenville, South Carolina.
PQT.
Level I, Greenville, South
Carolina. PQT.
Level I Film, Greenville, South
Carolina. PQT.
Level I Film to Non-film
Transition (NAS-410),
Greenville, South Carolina.
PQT.
Level I Non-film (NAS-410),
Greenville, South Carolina.
PQT.
14–19 AUG
Radiation Safety, Kent,
Washington. Mistras.
14–22 AUG
Level I Film/Non-film
(NAS-410), Greenville, South
Carolina. PQT.
14–26 AUG
Film Interpretation, Kerala,
India. Decibel.
Level I/II, Kerala, India.
Decibel.
21–25 AUG
Computed Radiography Level I,
Atlanta, Georgia. ATS.
Digital Radiography Level I,
Atlanta, Georgia. ATS.
Level I, Brea, California. Test.
Level I Film, Atlanta, Georgia.
ATS.
Level I Non-film (NAS-410),
Atlanta, Georgia. ATS.
Radiation Safety for
Radiographers, Greenville,
South Carolina. PQT.
21–30 AUG
Level I Film/Non-film
(NAS-410), Atlanta, Georgia.
ATS.
22–24 AUG
Level II Image Interpretation,
Troy, Michigan. XRI.
24–28 AUG
Radiation Safety (IRRSP Prep),
Brea, California. Test.
26–28 AUG
Level I/II (SNT-TC-1A),
Bangalore, India. Trinity.
28–29 AUG
IRRSP Refresher, Atlanta,
Georgia. ATS.
28–30 AUG
Level I Non-film Transition
(NAS-410), Atlanta, Georgia.
ATS.
28 AUG–1 SEP
Computed Radiography
Level II, Atlanta, Georgia. ATS.
Computed Radiography
Level II, Greenville, South
Carolina. PQT.
Digital Radiography Level II,
Atlanta, Georgia. ATS.
Digital Radiography Level II,
Greenville, South Carolina.
PQT.
Level II, Atlanta, Georgia. ATS.
Level II, Greenville, South
Carolina. PQT.
Level II, Brea, California. Test.
Level II Film, Greenville, South
Carolina. PQT.
Level II Film to Non-film
Transition (NAS-410),
Greenville, South Carolina.
PQT.
Level II Non-film (NAS-410),
Greenville, South Carolina.
PQT.
Radiation Safety, St. Louis,
Missouri. Quality.
28 AUG–5 SEP
Level II Film/Non-film
(NAS-410), Greenville, South
Carolina. PQT.
1–10 SEP
Film Interpretation Level II,
Trivandrum, India. Decibel
Remote.
1–12 SEP
Level I/II, Trivandrum, India.
Decibel Remote.
Practical, Kerala, India.
Decibel.
4–16 SEP
Film Interpretation, Kerala,
India. Decibel.
Level I/II, Kerala, India.
Decibel.
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 907
calendar
Radiographic Testing, cont.
11–15 SEP
Level I, Heath, Ohio. Mistras.
Level I, Kent, Washington.
Mistras.
Level I, St. Louis, Missouri.
Quality.
Radiation Safety, Tulatin,
Oregon. MPM.
Radiation Safety, Troy,
Michigan. XRI.
11–23 SEP
Film Interpretation, Kerala,
India. Decibel.
Level I/II, Kerala, India.
Decibel.
13–23 SEP
Film Interpretation Level II,
Trivandrum, India. Decibel
Remote.
18–22 SEP
Computed Radiography Level I,
Atlanta, Georgia. ATS.
Digital Radiography Level I,
Atlanta, Georgia. ATS.
Level I Film, Atlanta, Georgia.
ATS.
Level I Non-film (NAS-410),
Atlanta, Georgia. ATS.
Level II, Heath, Ohio. Mistras.
Level II, St. Louis, Missouri.
Quality.
18–27 SEP
Level I Film/Non-film
(NAS-410), Atlanta, Georgia.
ATS.
18–30 SEP
Film Interpretation, Kerala,
India. Decibel.
Level I/II, Kerala, India.
Decibel.
22–24 SEP
Level I/II (SNT-TC-1A),
Bangalore, India. Trinity.
25–27 SEP
Level I Non-film Transition
(NAS-410), Atlanta, Georgia.
ATS.
Radiation Safety for
Radiographers, Jacksonville,
Florida. PQT.
25–29 SEP
Computed Radiography Level I,
Greenville, South Carolina.
PQT.
Computed Radiography
Level II, Atlanta, Georgia. ATS.
908 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Digital Radiography Level I,
Greenville, South Carolina.
PQT.
Digital Radiography Level II,
Atlanta, Georgia. ATS.
Level I, Greenville, South
Carolina. PQT.
Level I Film, Greenville, South
Carolina. PQT.
Level I Film to Non-film
Transition (NAS-410),
Greenville, South Carolina.
PQT.
Level I Non-film (NAS-410),
Greenville, South Carolina.
PQT.
Level II, Atlanta, Georgia. ATS.
25 SEP–3 OCT
Level I Film/Non-film (NAS-
410), Greenville, South
Carolina. PQT.
25 SEP–4 OCT
Level II Film/Non-film (NAS-
410), Atlanta, Georgia. ATS.
1–6 OCT
X-ray, Trivandrum, India.
Decibel Remote.
1–12 OCT
Level I/II, Trivandrum, India.
Decibel Remote.
Practical, Kerala, India.
Decibel.
2–4 OCT
Level II Non-film Transition
(NAS-410), Atlanta, Georgia.
ATS.
2–6 OCT
Computed Radiography Level
II, Greenville, South Carolina.
PQT.
Computed Radiography/Digital
Radiography Level I, Arlington,
Texas. MPM.
Digital Radiography Level II,
Greenville, South Carolina.
PQT.
Level I, Troy, Michigan. XRI.
Level II, Greenville, South
Carolina. PQT.
Level II Film, Greenville, South
Carolina. PQT.
Level II Film to Non-film
Transition (NAS-410),
Greenville, South Carolina.
PQT.
Level II Non-film (NAS-410),
Greenville, South Carolina.
PQT.
Radiation Safety, Davenport,
Iowa. QCTL.
2–10 OCT
Level II Film/Non-film
(NAS-410), Greenville, South
Carolina. PQT.
2–14 OCT
Film Interpretation, Kerala,
India. Decibel.
Level I/II, Kerala, India.
Decibel.
7–16 OCT
Film Interpretation Level II,
Trivandrum, India. Decibel
Remote.
9–11 OCT
Radiation Safety for
Radiographers, Jacksonville,
Florida. PQT.
9–13 OCT
Computed Radiography/Digital
Radiography Level II,
Arlington, Texas. MPM.
Computed Radiography
Level II, Troy, Michigan. XRI.
Level II, Kent, Washington.
Mistras.
9–21 OCT
Film Interpretation, Kerala,
India. Decibel.
Level I/II, Kerala, India.
Decibel.
16–20 OCT
Computed Radiography Level I,
Atlanta, Georgia. ATS.
Digital Radiography Level I,
Atlanta, Georgia. ATS.
Film Interpretation, Davenport,
Iowa. QCTL.
Level I Film, Atlanta, Georgia.
ATS.
Level I Non-film (NAS-410),
Atlanta, Georgia. ATS.
Radiation Safety for
Radiographers, Greenville,
South Carolina. PQT.
16–25 OCT
Level I Film/Non-film
(NAS-410), Atlanta, Georgia.
ATS.
16–28 OCT
Film Interpretation, Kerala,
India. Decibel.
Level I/II, Kerala, India.
Decibel.
21–23 OCT
Level I/II (SNT-TC-1A),
Bangalore, India. Trinity.
23–25 OCT
Level I Non-film Transition
(NAS-410), Atlanta, Georgia.
ATS.
23–27 OCT
Computed Radiography Level
II, Atlanta, Georgia. ATS.
Digital Radiography Level II,
Atlanta, Georgia. ATS.
Level I, Portland, Oregon. NPI.
Level II, Atlanta, Georgia. ATS.
Level II, Allentown,
Pennsylvania. WTTI.
23 OCT–1 NOV
Level II Film/Non-film (NAS-
410), Atlanta, Georgia. ATS.
30 OCT–1 NOV
Level II Non-film Transition
(NAS-410), Atlanta, Georgia.
ATS.
30 OCT–3 NOV
Computed Radiography Level I,
Greenville, South Carolina.
PQT.
Digital Detector Array
Radiography Level II, Troy,
Michigan. XRI.
Digital Radiography Level I,
Greenville, South Carolina.
PQT.
Level I, Greenville, South
Carolina. PQT.
Level I Film, Greenville, South
Carolina. PQT.
Level I Film to Non-film
Transition (NAS-410),
Greenville, South Carolina.
PQT.
Level I Non-film (NAS-410),
Greenville, South Carolina.
PQT.
Level II, Portland, Oregon. NPI.
30 OCT–7 NOV
Level I Film/Non-film
(NAS-410), Greenville,
South Carolina. PQT.
Ultrasonic Testing
1–12 JUL
Phased Array Level II,
Trivandrum, India. Decibel
Remote.
1–15 JUL
Practical, Trivandrum, India.
Decibel Remote.
3–15 JUL
Level I/II, Kerala, India.
Decibel.
Phased Array Level II, Kerala,
India. Decibel.
Time of Flight Diffraction
Level II, Kerala, India. Decibel.
3–22 JUL
Practical, Kerala, India.
Decibel.
9–20 JUL
Level I/II, Trivandrum, India.
Decibel Remote.
Practical, Kerala, India.
Decibel.
10–14 JUL
Level I, Atlanta, Georgia. ATS.
Level I, Heath, Ohio. Mistras.
Level I, Concord, North
Carolina. Sonaspection.
Level II, Davenport, Iowa.
QCTL.
10–22 JUL
Auto Data Interpreter, Kerala,
India. Decibel.
Level I/II, Kerala, India.
Decibel.
Phased Array Level II, Kerala,
India. Decibel.
Time of Flight Diffraction
Level II, Kerala, India. Decibel.
10–29 JUL
Practical, Kerala, India.
Decibel.
12–14 JUL
A-scan Thickness, Allentown,
Pennsylvania. WTTI.
12–15 JUL
Level II, Concord, North
Carolina. Sonaspection.
12–16 JUL
Level I/II (SNT-TC-1A),
Bangalore, India. Trinity.
13–24 JUL
Level I/II, Trivandrum, India.
Decibel Remote.
Time of Flight Diffraction
Level II, Trivandrum, India.
Decibel Remote.
16–30 JUL
Practical, Trivandrum, India.
Decibel Remote.
17–18 JUL
Introduction to Phased Array,
Brea, California. Test.
17–21 JUL
Level I, Portland, Oregon. NPI.
Level I, Greenville, South
Carolina. PQT.
Level II, Atlanta, Georgia. ATS.
Level II, Heath, Ohio. Mistras.
Level II, Concord, North
Carolina. Sonaspection.
Phased Array Level I
(SNT-TC-1A), Brea, California.
Test.
Thickness, Greenville, South
Carolina. PQT.
17–29 JUL
Level I/II, Kerala, India.
Decibel.
Phased Array Level II, Kerala,
India. Decibel.
Time of Flight Diffraction
Level II, Kerala, India. Decibel.
24–28 JUL
Level I, Palos Hills, Illinois.
Moraine.
Level II, Portland, Oregon. NPI.
Level II, Greenville, South
Carolina. PQT.
Phased Array Level II
(SNT-TC-1A), Brea, California.
Test.
31 JUL–4 AUG
Level I, St. Louis, Missouri.
Quality.
Level I, Concord, North
Carolina. Sonaspection.
Level I, Brea, California. Test.
1–12 AUG
Automated, Trivandrum, India.
Decibel Remote.
Level I/II, Kerala, India.
Decibel.
Phased Array Level II, Kerala,
India. Decibel.
Time of Flight Diffraction
Level II, Kerala, India. Decibel.
1–21 AUG
Practical, Kerala, India.
Decibel.
7–11 AUG
Level I, Atlanta, Georgia. ATS.
Level II, St. Louis, Missouri.
Quality.
Level II, Brea, California. Test.
7–19 AUG
Auto Data Interpreter, Kerala,
India. Decibel.
Level I/II, Kerala, India.
Decibel.
Phased Array Level II, Kerala,
India. Decibel.
Time of Flight Diffraction
Level II, Kerala, India. Decibel.
8–29 AUG
Practical, Kerala, India.
Decibel.
9–20 AUG 1–12 SEP
Level I/II, Trivandrum, India. Phased Array Level II,
Decibel Remote. Trivandrum, India. Decibel
13–22 AUG
Remote.
Practical, Kerala, India. 1–15 SEP
Practical, Trivandrum, India.
Decibel.
Decibel Remote.
13–24 AUG
Level I/II, Trivandrum, India. 4–16 SEP
Decibel Remote. Level I/II, Kerala, India.
14–18 AUG
Decibel.
Level II, Atlanta, Georgia. ATS. Phased Array Level II, Kerala,
India. Decibel.
14–26 AUG
Time of Flight Diffraction
Level I/II, Kerala, India.
Level II, Kerala, India. Decibel.
Decibel.
4–23 SEP
Phased Array Level II, Kerala,
Practical, Kerala, India.
India. Decibel.
Decibel.
Time of Flight Diffraction
Level II, Kerala, India. Decibel. 9–20 SEP
Level I/II, Trivandrum, India.
15–17 AUG
Thickness, St. Louis, Missouri. Decibel Remote.
Quality. 11–15 SEP
Level I, Atlanta, Georgia. ATS.
16–20 AUG
Level I/II (SNT-TC-1A), Level I, Greenville, South
Carolina. PQT.
Bangalore, India. Trinity.
Level I, Concord, North
16–30 AUG Carolina. Sonaspection.
Practical, Trivandrum, India. Level II, Allentown,
Decibel Remote. Pennsylvania. WTTI.
17–30 AUG Thickness, Greenville, South
Phased Array Level II, Carolina. PQT.
Trivandrum, India. Decibel 11–23 SEP
Remote. Auto Data Interpreter, Kerala,
21–25 AUG India. Decibel.
Level I, San Antonio, Texas. Level I/II, Kerala, India.
BRL. Decibel.
Level I, Greenville, South Phased Array Level II, Kerala,
Carolina. PQT. India. Decibel.
Level I, Davenport, Iowa. QCTL. Time of Flight Diffraction
Level II, Kerala, India. Decibel.
Phased Array Week 1, St.
Louis, Missouri. Quality. 11–30 SEP
Thickness, Greenville, South Practical, Kerala, India.
Carolina. PQT. Decibel.
23–25 AUG 13–17 SEP
Annex Q, Allentown,
Level I/II (SNT-TC-1A),
Pennsylvania. WTTI.
Bangalore, India. Trinity.
23–30 AUG 13–20 SEP
Auto Data Interpreter, Practical, Kerala, India.
Trivandrum, India. Decibel Decibel.
Remote.
13–24 SEP
28 AUG–1 SEP Level I/II, Trivandrum, India.
Level II, San Antonio, Texas. Decibel Remote.
BRL. Time of Flight Diffraction
Level II, Greenville, South Level II, Trivandrum, India.
Carolina. PQT. Decibel Remote.
Phased Array Week 2, St.
16–30 SEP
Louis, Missouri. Quality.
Practical, Trivandrum, India.
Decibel Remote.
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 909
calendar
Ultrasonic Testing, cont. Level I/II, Kerala, India. 23–30 OCT 15–17 AUG
Decibel. Auto Data Interpreter, Weld Inspection, Heath, Ohio.
18–22 SEP Phased Array Level II, Kerala, Trivandrum, India. Decibel Mistras.
Level I, Pittsburgh, India. Decibel. Remote. 21–25 AUG
Pennsylvania. Odyssey. Time of Flight Diffraction 30 OCT–3 NOV Level I/II, Kerala, India.
Level I, St. Louis, Missouri. Level II, Kerala, India. Decibel. Level I, Atlanta, Georgia. ATS. Decibel.
Quality. 9–28 OCT Phased Array Week 2, St. 29–31 AUG
Level I, Brea, California. Test. Practical, Kerala, India. Louis, Missouri. Quality. Level I/II (SNT-TC-1A),
Level II, Greenville, South Decibel. Bangalore, India. Trinity.
Carolina. PQT. Visual Testing
11–15 OCT
Level II, Concord, North 6–8 SEP
Level I/II (SNT-TC-1A), 5–7 JUL
Carolina. Sonaspection. Level I/II, Atlanta, Georgia.
Bangalore, India. Trinity. Level I/II, St. Louis, Missouri.
Phased Array (first 40 h), ATS.
San Antonio, Texas. BRL. 13–22 OCT Quality.
11–15 SEP
Practical, Kerala, India. 10–14 JUL
18–30 SEP Level I/II, Kerala, India.
Decibel. Level I/II, Kerala, India.
Level I/II, Kerala, India. Decibel.
Decibel. 13–24 OCT Decibel.
18–22 SEP
Phased Array Level II, Kerala, Level I/II, Trivandrum, India. 11–13 JUL Level I/II, Kerala, India.
India. Decibel. Decibel Remote. Level I/II, Portland, Oregon. Decibel.
Time of Flight Diffraction 16–20 OCT NPI.
Level II, Kerala, India. Decibel. 25–27 SEP
Level I, Heath, Ohio. Mistras. 12–14 JUL Level I/II (SNT-TC-1A),
25–29 SEP Level I, Greenville, South Level I/II, Jacksonville, Florida. Bangalore, India. Trinity.
Level II, St. Louis, Missouri. Carolina. PQT. PQT.
Quality. Level II, Portland, Oregon. NPI. 25–29 SEP
Level II, Brea, California. Test. Thickness, Greenville, South 17–21 JUL Level I/II, Kerala, India.
Phased Array (second 40 h), Carolina. PQT. Level I/II, Kerala, India. Decibel.
San Antonio, Texas. BRL. Decibel.
16–28 OCT 25–30 SEP
26–28 SEP Level I/II, Kerala, India. 24–26 JUL Level II, Trivandrum, India.
Thickness Davenport, Iowa. Decibel. Level I/II (SNT-TC-1A), Decibel Remote.
QCTL. Phased Array Level II, Kerala, Bangalore, India. Trinity.
1–6 OCT
India. Decibel. 24–28 JUL Level II, Trivandrum, India.
1–12 OCT
Phased Array Level II, Concord, Level I/II, Kerala, India. Decibel Remote.
Automated, Trivandrum, India.
North Carolina. Sonaspection. Decibel.
Decibel Remote. 3–5 OCT
Time of Flight Diffraction
2–6 OCT 25 JUL Level I/II, Portland, Oregon.
Level II, Kerala, India. Decibel.
Level I, Atlanta, Georgia. ATS. Level I, Davenport, Iowa. QCTL. NPI.
16–30 OCT Level I/II, St. Louis, Missouri.
2–14 OCT 25–30 JUL
Practical, Trivandrum, India. Quality.
Level I/II, Kerala, India. Level II, Trivandrum, India.
Decibel Remote.
Decibel. Decibel Remote. 9–13 OCT
17–30 OCT Level I/II, Kerala, India.
Phased Array Level II, Kerala, 26–27 JUL
Phased Array Level II, Decibel.
India. Decibel. Level II, Davenport, Iowa.
Trivandrum, India. Decibel
Time of Flight Diffraction QCTL. 16–18 OCT
Remote.
Level II, Kerala, India. Decibel. Level I/II, San Antonio, Texas.
1–3 AUG
18–20 OCT BRL.
2–21 OCT Level I/II, Greenville, South
D1.1 and D1.5, Allentown,
Practical, Kerala, India. Carolina. PQT. 16–20 OCT
Pennsylvania. WTTI.
Decibel. Level I/II, Kerala, India.
1–6 AUG
19 OCT Decibel.
9–13 OCT Level II, Trivandrum, India.
Digital Thickness Level II,
Level I, Portland, Oregon. NPI. Decibel Remote. 23–27 OCT
San Antonio, Texas. BRL.
Level II, Atlanta, Georgia. ATS. Level I/II, Kerala, India.
2–4 AUG
Phased Array Level I, Concord, 23–27 OCT Decibel.
Level I/II (SNT-TC-1A), Troy,
North Carolina. Sonaspection. Level II, Heath, Ohio. Mistras.
Michigan. XRI. 24–26 OCT
Level II, Greenville, South
9–20 OCT Level I/II, Jacksonville, Florida.
Carolina. PQT. 7–11 AUG
Level I/II, Trivandrum, India. PQT.
Phased Array Week 1, St. Level I/II, Kerala, India.
Decibel Remote. Level I/II (SNT-TC-1A),
Louis, Missouri. Quality. Decibel.
9–21 OCT Time of Flight Diffraction, Bangalore, India. Trinity.
14–18 AUG
Auto Data Interpreter, Kerala, Concord, North Carolina. 31 OCT–2 NOV
Level I/II, Kerala, India.
India. Decibel. Sonaspection. Level I/II, Greenville, South
Decibel.
Carolina. PQT.

910 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Short Courses/Topical
Seminars
1–13 JUL
Welding Inspection and
Quality Control Level I,
Trivandrum, India. Decibel
Remote.
4–5 JUL
CIVA NDE Simulation Software:
Intro and Applications (RT, CT),
Massy, France. Extende France.
13–24 JUL
Welding Inspection and
Quality Control Level II,
Trivandrum, India. Decibel
Remote.
21–30 JUL
Introduction to NDT,
Trivandrum, India. Decibel
Remote.
1–12 AUG
Welding Inspection and
Quality Control Level II,
Trivandrum, India. Decibel
Remote.
7–11 AUG
MFL per API 653 Annex G,
Dripping Springs, Texas. MFE.
13–24 AUG
Welding Inspection and
Quality Control Level II,
Trivandrum, India. Decibel
Remote.
18 AUG
NDT 101, St. Louis, Missouri.
Quality.
21–30 AUG
Introduction to NDT,
Trivandrum, India. Decibel
Remote.
29–31 AUG
Basic Metallurgy for NDT,
Heath, Ohio. Mistras.
1–13 SEP
Welding Inspection and
Quality Control Level I,
Trivandrum, India. Decibel
Remote.
13–24 SEP
Welding Inspection and
Quality Control Level II,
Trivandrum, India. Decibel
Remote.
21–30 SEP
Introduction to NDT,
Trivandrum, India. Decibel
Remote.
25–29 SEP
CIVA NDE Simulation Software:
Intro and Applications (UT,
GWT), Massy, France. Extende
France.
1–12 OCT
Welding Inspection and
Quality Control Level I,
Trivandrum, India. Decibel
Remote.
3–5 OCT
Training Seminar for NDE
Users, Wheat Ridge, Colorado.
Olson.
10–11 OCT
CIVA NDE Simulation Software:
Intro and Applications (ET),
Massy, France. Extende France.
13–24 OCT
Welding Inspection and
Quality Control Level II,
Trivandrum, India. Decibel
Remote.
21–30 OCT
Introduction to NDT,
Trivandrum, India. Decibel
Remote.
Level III Examination
Preparation/Refreshers
1–6 JUL
UT Level III Prep, Trivandrum,
India. Decibel Remote.
3–7 JUL
Basic Level III, Kerala, India.
Decibel.
MT Level III, Kerala, India.
Decibel.
7–12 JUL
RT Level III Prep, Trivandrum,
India. Decibel Remote.
10–14 JUL
Basic Level III, Houston, Texas.
Kraft.
PT Level III, Kerala, India.
Decibel.
RT Level III, Kerala, India.
Decibel.
12–14 JUL
Basic Level III, Greenville,
South Carolina. PQT.
MT Level III, Greenville, South
Carolina. PQT.
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 911
calendar
Level III Examination 7–11 AUG MT Level III, Greenville, South 4–6 OCT
Preparation/Refreshers, cont. PT Level III, Kerala, India. Carolina. PQT. Basic Level III, Greenville,
Decibel. PT Level III, Greenville, South South Carolina. PQT.
PT Level III, Greenville, South
RT Level III, Kerala, India. Carolina. PQT. MT Level III, Greenville, South
Carolina. PQT.
Decibel. RT Level III, Greenville, South Carolina. PQT.
RT Level III, Greenville, South
Carolina. PQT. 7–12 AUG Carolina. PQT. PT Level III, Greenville, South
UT Level III, Greenville, South RT Level III Prep, Trivandrum, UT Level III, Greenville, South Carolina. PQT.
Carolina. PQT. India. Decibel Remote. Carolina. PQT. RT Level III, Greenville, South
VT Level III, Greenville, South VT Level III, Greenville, South Carolina. PQT.
8–10 AUG Carolina. PQT. UT Level III, Greenville, South
Carolina. PQT. PT Level III, St. Louis, Missouri. VT Level III, St. Louis, Missouri. Carolina. PQT.
13–16 JUL Quality. Quality. VT Level III, Greenville, South
MT Level III Prep, Trivandrum, 13–16 AUG Carolina. PQT.
7–12 SEP
India. Decibel Remote. MT Level III Prep, Trivandrum, RT Level III Prep, Trivandrum, 7–12 OCT
17–18 JUL India. Decibel Remote. India. Decibel Remote. RT Level III Prep, Trivandrum,
PT Level III, Branford, 14–18 AUG India. Decibel Remote.
11–15 SEP
Connecticut. Atlantic. ET Level III, Houston, Texas. Basic Level III, St. Louis, 9–13 OCT
17–19 JUL Kraft. Missouri. Quality. PT Level III, Kerala, India.
MT Level III, Houston, Texas. 16–18 AUG IR Level III, Barre, Vermont. Decibel.
Kraft. Basic Level III, Greenville, Snell. RT Level III, Kerala, India.
17–20 JUL South Carolina. PQT. PT Level III, Kerala, India. Decibel.
PT Level III Prep, Trivandrum, MT Level III, Greenville, South Decibel. UT Level III, Brea, California.
India. Decibel Remote. Carolina. PQT. RT Level III, Kerala, India. Test.
PT Level III, Greenville, South Decibel. 13–16 OCT
18–19 JUL
Carolina. PQT. 13–16 SEP MT Level III Prep, Trivandrum,
MT Level III, Branford,
RT Level III, Greenville, South MT Level III Prep, Trivandrum, India. Decibel Remote.
Connecticut. Atlantic.
Carolina. PQT. India. Decibel Remote.
19–21 JUL UT Level III, Greenville, South 16–19 OCT
PT Level III, Houston, Texas. Carolina. PQT. 17–20 SEP Eddy Current Level III, Brea,
Kraft. VT Level III, Greenville, South PT Level III Prep, Trivandrum, California. Test.
Carolina. PQT. India. Decibel Remote. 16–20 OCT
20–21 JUL
Basic Level III, Branford, 17–20 AUG 21–25 SEP RT Level III, St. Louis, Missouri.
Connecticut. Atlantic. PT Level III Prep, Trivandrum, Basic Level III Prep, Quality.
India. Decibel Remote. Trivandrum, India. Decibel 17–20 OCT
21–25 JUL Remote.
Basic Level III Prep, 21–25 AUG PT Level III Prep, Trivandrum,
Trivandrum, India. Decibel Basic Level III Prep, 25–27 SEP India. Decibel Remote.
Remote. Trivandrum, India. Decibel IR Level III, Burlington, New 19–20 OCT
Remote. Jersey. Infraspection. MT Level III, Brea, California.
24–28 JUL
ET Level III, Kerala, India. RT Level III, Houston, Texas. 25–29 SEP Test.
Decibel. Kraft. UT Level III, Kerala, India. 21–25 OCT
UT Level III, Kerala, India. UT Level III, Kerala, India. Decibel. Basic Level III Prep,
Decibel. Decibel.
1–6 OCT Trivandrum, India. Decibel
UT Level III, Houston, Texas. VT Level III, Kerala, India.
UT Level III Prep, Trivandrum, Remote.
Kraft. Decibel.
India. Decibel Remote. 23–24 OCT
25–27 JUL 1–6 SEP
2–6 OCT PT Level III, Brea, California.
MT Level III, St. Louis, UT Level III Prep, Trivandrum,
Basic Level III, Kerala, India. Test.
Missouri. Quality. India. Decibel Remote.
Decibel. 25–27 OCT
1–5 AUG 4–8 SEP Basic Level III, Brea, California. RT Level III, Brea, California.
Basic Level III, Kerala, India. Basic Level III, Kerala, India. Test. Test.
Decibel. Decibel. MT Level III, Kerala, India. UT Level III, Kerala, India.
MT Level III, Kerala, India. MT Level III, Kerala, India. Decibel. Decibel.
Decibel. Decibel. UT Level III, St. Louis,
Missouri. Quality. 30–31 OCT
1–6 AUG 6–8 SEP
VT Level III, Brea, California.
Basic Level III, Greenville,
UT Level III Prep, Trivandrum, Test. w
x
India. Decibel Remote. South Carolina. PQT.

912 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7

Characterizing Surface-breaking Cracks through Eddy
Current NDE and Model-based Inversion
by Erin K. Oneida*, Eric B. Shell†, John C. Aldrin‡, Harold A. Sabbagh§, Elias H. Sabbagh¶, and R. Kim Murphy**
ABSTRACT
Initial results for model-based inversion of eddy
current flaw response data are presented as an
alternative or supplement to amplitude-based
sizing. The approach provides an estimation of the
flaw dimensions (length, depth, and width) and
orientation that are independent of amplitudes
and assumptions from a probability of detection
(POD) analysis. The effects of probe construction
characteristics on eddy current signals are consid-
ered. The inversion method performs well over a
range of intermediate flaw sizes and orientations,
representing an improvement for flaws that are
dissimilar from those used in an amplitude-based
POD model. Future research directions to improve
inversion estimates associated with greater uncer-
tainty are discussed.
KEYWORDS: eddy current, inversion modeling, flaw
characterization.
* KBRwyle, 2700 Indian Ripple Rd., Dayton, Ohio 45440; e-mail
erin.oneida@wyle.com
† Ph.D., KBRwyle, 2700 Indian Ripple Rd., Dayton, Ohio 45440; e-mail
eric.shell@wyle.com
‡ Ph.D., Computational Tools, 4275 Chatham Ave., Gurnee, Illinois 60031;
e-mail aldrin@computationaltools.com
§ Ph.D., Victor Technologies LLC, P.O. Box 7706, Bloomington, Indiana;
47407-7706; e-mail has@sabbagh.com
¶ Victor Technologies LLC, P.O. Box 7706, Bloomington, Indiana 47401;
e-mail ehs@sabbagh.com
** Ph.D., Victor Technologies LLC, P.O. Box 7706, Bloomington, Indiana
47401; e-mail rkmurphy@att.net
ME TECHNICAL PAPER w
x
Introduction
Nondestructive testing methods are critical to ensuring the
structural integrity of aircraft propulsion components. As the
average age of the US Air Force (USAF) fleet increases, life
extension practices that rely on accurate flaw characterization,
such as condition-based maintenance, are required to ensure
flight safety (Lindgren, 2016). Eddy current inspection tech-
niques are frequently used to inspect engine components for
cracks with lengths between 254 and 1524 μm (10 and
60 mils). The eddy current response amplitude can be used
with an â versus a probability of detection statistical analysis
to determine if a flaw length (or depth) likely exceeds a speci-
fied critical value (US Department of Defense, 2009). While
amplitude-based methods provide acceptable sizing estimates
for current part inspections, the associated assumptions and
uncertainty affect the accuracy of any reported flaw dimen-
sions and may result in the unnecessary removal of parts from
service.
Amplitude-based flaw sizing methods rely on an assump-
tion that there is a common length-to-depth aspect ratio
(for example, 2:1) among the flaws located in inspected parts
and those in the POD specimens. Looking at amplitude
versus flaw planar area on a log-log scale, such as in Figure 1a,
a linear relationship is apparent, even for flaws with varying
aspect ratios. POD-derived parameters are dependent on an
assumed linear relationship between response amplitudes
and crack lengths (or depths), and, to achieve this linearity,
a constant aspect ratio is required. Looking at the data in
Figure 1b, it is clear that an amplitude based approach relying
on linear fit parameters derived for 2:1 flaws would provide
poor size estimates for flaws with varying aspect ratios.
Following the methodology of MIL-HDBK-1823A, eddy
current response data is collected using multiple probes and
specimens to capture some of the amplitude variations that
can be expected during eddy current testing. The residuals
between the data and linear fit lead to an uncertainty param-
eter, which is then used to determine a critical amplitude that
ensures, with a predetermined degree of certainty, that a criti-
cally sized flaw will be detected as such. Noisier data leads to
a lower critical amplitude, which results in an increased
number of parts failing inspections. While there are many
factors that can cause noise during POD data collection and
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 915
ME TECHNICAL PAPER w
x characterizing surface-breaking cracks

tions due in part to a lack of robustness when applied to data

exhibiting variability associated with depot nondestructive
3.5
evaluation (NDE) measurements, parts, and flaw conditions.
Recent work has investigated the use of model-based inver-
sion methods that rely on simulated eddy current response
log(Amplitude)

signals to size uncharacterized flaws in bolt holes (Aldrin et

3
al., 2014) and on flat surfaces (Shell et al., 2015; Aldrin et al.,
2016; Aldrin et al., 2017). This work involves using the full
eddy current flaw response together with computer-simulated
2.5
eddy current responses (Sabbagh et al., 2013) to estimate a
flaw’s length, depth, width (crack/notch opening), and
angular orientation with respect to the inspection scan direc-
2 tion. The accuracy of the simulated responses is dependent on
4.5 5 5.5 6
log(Known area) properly modeling the inspection process and the inspection
(a)
probes. While the simulated probes are typically based on
ideal dimensions and coil placement, actual inspection probes
exhibit variations, such as the coil being rotated or tilted about
3.5
an axis, that affect eddy current response signals (Mooers et
al., 2014; Aldrin et al., 2016; Aldrin et al., 2017). Some probe
log(Amplitude)

variations have been represented in simulations (Mooers et

3 al., 2015; Aldrin et al., 2016; Aldrin et al., 2017), and ongoing
2.5
2 effort (Aldrin et al., 2014; Shell et al., 2015; Aldrin et al.,
2.4 2.6 2.8
(b) log(Known length)
Aspect ratio varies (Plate #1)
Aspect ratio varies (Plate #2)
Aspect ratio = 2:1 (Plate #3)
Figure 1. Eddy current response amplitudes for transversely oriented
notches in a nickel-based superalloy plotted against (a) measured
notch profile areas (μm2) and (b) measured notch lengths (μm).
actual inspections, for example, probe characteristics, some of
this variation is due to uncertainty of the crack depths in POD
specimens. To better estimate flaw dimensions and determine
if parts should be removed from service, new sizing
approaches that are not dependent on fixed aspect ratio
assumptions are under development.
Significant research investigating the use of inverse
methods for flaw characterization using eddy current data has
been performed over the past few decades (Sabbagh and
Sabbagh, 1988; Norton and Bowler, 1993; Bowler, 1997;
Auld and Moulder, 1999; Yusa et al., 2007; Liu et al., 2008;
Sabbagh et al., 2008; Yusa, 2009; Sabbagh et al., 2013). Many
of these efforts have focused on the sizing of surface-breaking
cracks or corrosion pits. However, there has been limited
progress in transitioning these techniques to practical applica-
work is investigating the development of inversion methods
to characterize the probe properties and improve flaw dimen-
sion estimates for data collected using probes with typical
variations (Aldrin et al., 2016; Aldrin et al., 2017).
The work presented in this paper is part of an existing
3 3.2
2016), and it includes characterization results obtained using
updated algorithms and improved software-generated forward
model libraries (FMLs). Flaw dimensions and orientations
are estimated for notches and cracks with aspect ratios
ranging from 1:1 to 7.5:1 and for notches with semielliptical
or rectangular profiles. The flaws are embedded in a titanium
alloy or nickel superalloy, and, during data collection, the
probes were carefully positioned to acquire data representa-
tive of probes with different properties or positioning relative
to the part surface. Validation of the dimensions and orienta-
tions estimated using the inversion method is accomplished
by comparing estimated values to either nondestructively
measured flaw characteristics or dimensions obtained by
breaking open cracked specimens.
Experimental Data Collection
To ensure the robustness of the developed inversion method,
there was a need for a large set of eddy current response
signals representing the expected variation observed during
depot inspections. Factors considered include specimen
material type; flaw type (crack versus notch), geometric
planar profile (rectangular versus semielliptical), aspect ratio
(length:depth), dimensions (length, depth, and width), and
orientation relative to scan direction; and probe characteris-
tics and positioning relative to the specimen surface and scan
direction. To date, more than 5100 two-dimensional eddy
current flaw response datasets have been collected.

916 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
 Transverse data collection
Scan direction
Index direction
Figure 2. Experimental data collection details.
General Inspection Characteristics
To create a vast library of eddy current responses, data was
collected using seven different split-D differential reflection
probes, which have 1 mm (40 mil) nominal diameter ferrite
cores and are operated at a 2 MHz inspection frequency
(Shell et al., 2015). A single piece of ~69 μm (2.7 mils) thick
tape was affixed to the probe shoe across the coil. Eight-axis,
automated eddy current inspection stations (ECIS) were used
to collect data following a raster scan pattern with the sample
spacing and inspection parameters listed in Figure 2 (Shell
et al., 2015). All probes were calibrated on a 383  208  76 μm
(15.09  8.18  3 mil) (length  depth  width) electrical
discharge machined (EDM) notch in a reference bar
(nickel-based superalloy) to achieve a desired peak-to-peak
amplitude equal to approximately 400 digitizer counts (488
mV) with a vertical phase orientation. A high-pass filter was
not used during collection, and the sample spacing was
refined compared to typical inspections. These collection
techniques led to high-quality eddy current response data that
was well-balanced for probes exhibiting proper coil place-
ment.
Notched/Cracked Specimens and Flaw Orientation
Eddy current response data was collected for 54 different
EDM notches and 169 different cracks, which were in either a
nickel-based superalloy or titanium alloy. Known values for
the length, depth, and width of each EDM notch were
obtained by using reported values or measuring them using a
replication approach. Notch length-to-depth aspect ratios
varied from approximately 1:1 to 7.5:1, and, with the excep-
tion of eleven rectangular notches, the planar geometric
profile was semielliptical. The inspected cracks were part of
POD sets that had reported lengths and depths, where the
lengths were measured optically and the depths were esti-
mated by destructively characterizing a few specimens to
estimate likely aspect ratios. The flaws were scanned in both
transverse and longitudinal orientations, and several flaws
were also scanned at 27 different angles. Eddy current
response data for a 495  251  38 μm (19.5  9.9  1.5 mil)
Calibration Inspection
Index distance (μm) 38.1 50.8
Scan sampling
21.2 42.3
distance (μm)
Scan speed (cm/s) 2.12 4.23
Sample rate (pts/s) 1000 1000
notch scanned at 4 different angles are shown in Figure 3.
This paper focuses on the application of the inversion
method to 38 notches in four specimens (Table 1) and to
63 cracks. Previous work has also relied on the experimental
data collected using Plates #1 and #4 (Shell et al., 2015;
Aldrin et al., 2016; Oneida et al., 2016; Aldrin et al., 2017).
Recently, a new nickel plate was constructed to further test
the inversion method’s robustness when applied to notches
with varying aspect ratios. Plate #2 was carefully designed so
that for each rectangular notch (#1–5), there is a semiellip-
tical notch with approximately the same length and depth
(#6–10). Notches 1–5 all have similar planar areas, ranging
from 0.1138 to 0.1284 mm2 (176.4 to 199 mils2), while
notches 6–10 have areas that range from 0.0888 to 0.0981 mm2
(137.6 to 152.1 mils2). The remaining five notches (#11–15)
have depths that correspond to notches 6–10 and their areas
range from 0.0502 to 0.0585 mm2 (77.8 to 90.7 mils2). Data
from Plate #2 is used to investigate the ability of the inversion
method to estimate lengths and depths when the planar areas
are similar, but the aspect ratios are variable. Inability to
estimate lengths and depths for varying aspect ratios is a
major limitation of existing amplitude-based, POD-dependent
methods. The eddy current data for this plate also enables
investigation of the inversion method capabilities for notches
with rectangular profiles.
The current inversion method was applied to previously
studied data (Shell et al., 2015), obtained from a set of cracked
nickel POD specimens with approximately 2.5:1 aspect ratios
and lengths ranging from 432 to 991 μm (17 to 39 mils).
Since crack depths could not be readily measured, the known
values were previously estimated by using aspect ratios
measured from a few destructively characterized specimens
(Shell et al., 2015). In the current work, the inversion method
is also applied to eddy current data collected for five cracks in
nickel specimens. After experimental data collection, these
specimens were destructively characterized, allowing for
comparison of the estimated dimensions to measured crack
depths and lengths.
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 917
ME TECHNICAL PAPER w
x characterizing surface-breaking cracks

Flaws
Coil axial view

Scan Length
direction

Depth
0° 30° 60° 90° Semielliptical

300

200

100

–100

–200

–300
0° (Longitudinal) 30° 60° 90° (Transverse)

Figure 3. Eddy current responses for a 495  251  38 μm (19.5  9.9  1.5 mil) semielliptical notch positioned at four different angles relative
to the scan direction.

Probe Characteristics
Probes are constructed according to a specific design;
however, acceptable variations in coil placement can lead to
non-uniformity in the eddy current response footprint across
nominally identical probes. Four realistic coil positioning vari-
ations were considered during experimental collection efforts:
setback (Z-direction), rotation (about Z-axis), and tilt about
either the X- or Y-axis (Figure 4). Non-ideal probe posi-
tioning relative to the specimen surface, which occurs due to
allowable part and ECIS alignment tolerances, can cause eddy
current responses similar to those associated with non-ideal
coil placement.

TABLE 1
Dimensions (μm) for notches in the three nickel plates and one titanium plate studied in this paper. Eleven of the notches had rectangular
profiles, while the remaining had semielliptical profiles. The seven crossed-out notches on Plate #1 are not studied in this paper.
Notch Nickel plate #1 Nickel plate #2 Nickel plate #3 Titanium plate #4
ID Variable aspect ratios Variable aspect ratios 2:1 Aspect ratios 2:1 Aspect ratios
Length Depth Width Length Depth Width Length Depth Width Length Depth Width
1 1011 503 43 940* 122 38 470 236 89 267* 140 69
2 155 76 28 759* 150 41 399 198 86 391* 201 69
3 127 124 23 638* 191 38 574 292 89 521* 264 66
4 3073 1384 71 566* 218 36 704 361 89 658* 333 69
5 503 503 53 505* 254 36 325 163 89 777* 378 69
6 3043 505 43 958 127 38 856 429 86 1031* 503 69
7 2294 391 61 754 150 36 1057 518 86
8 765 124 28 635 193 38 264 135 81
9 460 79 28 559 224 38 1295 640 89
10 74 76 25 495 251 38
11 249 122 28 508 127 36
12 249 241 30 437 163 36
13 530 251 30 386 193 33
14 1529 254 28 312 221 38
15 2550 249 30 251 254 38
* Notch has rectangular profile.

918 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7

rotation about Z
Rotation ΘZ
rotation about Y
Tilt ΘY
Y –200
X
rotation about X
Tilt ΘX
–10°
Figure 4. Eddy current responses for a transversely oriented flaw scanned at three different rotation or tilt angles.
To collect data representative of non-ideal coil placement
or probe position, controlled variations in setback, tilt, and
rotation were included in the experimental design. To re-
create the effects of coil setback, between 1 and 5 additional
layers of tape were placed between the probe shoe and the
specimen. Coil rotation was accomplished by automatically
rotating the probe about its Z-axis using the ECIS. Tilt was
approximated by using a probe with a shoe that had a reduced
contact area, which allowed the probe to be more easily tilted
using the rotary axes of the inspection station. For the
rotation and tilt cases, the probes were first calibrated at the 0°
orientation, and then the probes were positioned at five
different angles (–10°, –5°, 0°, 5°, and 10°). Figure 4 illustrates
the changes in eddy current responses that occur when the
coil is not properly positioned.
Inversion Procedure for Flaw Characterization
The basic concept underlying the developed inversion
method involves comparing experimental eddy current
responses associated with uncharacterized flaws to
300
200
100
0
–100
–200
–300
200
100
–100
200
100
–100
–200
0° 10°
software-generated eddy current responses that were simu-
lated for numerous flaw sizes and orientations. The major
components of this method comprise 1) developing a forward
model library; 2) using the experimental calibration response
to calibrate the forward model library; and 3) iteratively
comparing the uncharacterized flaw to the calibrated forward
models while varying the model parameters until the best fit is
reached. Details of these steps are presented in this section.
Forward Model Libraries
Semielliptical and rectangular flaws located in nickel or
titanium were modeled using software that relies on volume-
integral method calculations to simulate eddy current inspections
(Sabbagh et al., 2013), where nickel was assigned a conductivity
of  = 0.782  106 S/m (1.34 %IACS) and titanium had  =
0.580  106 S/m (1.00 %IACS). For each material and
geometry combination, flaws with the lengths, depths, and
widths indicated in Table 2 were modeled (Shell et al., 2015).
A split-D differential probe was simulated with geometric and
material properties representative of the probes used during
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 919
ME TECHNICAL PAPER w
x characterizing surface-breaking cracks

TABLE 2 later created by using cubic spline interpolation. To include

Dimensions and angles included in software models, except where
noted.
Depth (μm) 63.5, 127, 254, 508, 762, 1016
Length (μm) 254, 508, 762*, 1016, 1524*,
2032, 3048
Width (μm) 0.0254**, 25.4, 76.2, 127
Angle 0°, 30°, 60°, 90°
*Data for these lengths generated using cubic spline interpolation.
**Data for this width generated using linear extrapolation.
experimental data collection (Shell et al., 2015). To generate
data for four flaw orientations (Table 2), a transversely
oriented flaw was included in the model, and the probe was
), and an improved width. The
rotated to different angles. Data was simulated every 127 μm
by the vertical and hori-
(5 mils) along the length of each flaw and every 50.8 μm
scales the
(2 mils) in the perpendicular direction. The resulting data
was then stored in a forward model library. Prior to using the
simulated data during the inversion process, it is rotated to
the correct angle and then re-sampled.
The generated signals all include horizontal and vertical
components, which were gathered into four separate forward
model libraries according to material and geometry type.
Signals for two additional lengths (762 and 1524 μm) were
data representative of a crack, linear extrapolation was used to
create eddy current responses for a width of 0.0254 μm
(0.001 mils). During the inversion process, eddy current
responses were generated for flaws with dimensions and
angles not included in Table 2 by using N-dimensional cubic
spline interpolation with the existing FML response data.
Calibrating Forward Model Library
The first step of the inversion process involves calibrating the
forward model library following the outline in Figure 5. First,
the experimental calibration flaw response is centered and
re-sampled. This response is then compared to the FML
response associated with the known calibration flaw dimen-
sions, and an iterative scheme is used to find the flaw angle
( ), calibration parameter (
FML is calibrated by multiplying
zontal components for each flaw in the FML.
generated data and rotates the phase of the impedance plane
response. The iterative method seeks to minimize the differ-
ence between the experimental response signal for the calibra-
tion flaw and the corresponding calibrated FML flaw.
Inversion Method to Characterize the Unknown Flaw
Flaw dimensions and orientations are estimated by comparing
the unknown flaw signals to simulated flaw responses

Centered and interpolated

calibration EC response
Ex and Ey Use
(Mx  iMy) to recenter
Get forward model experimental response and
using new and width resampled to get new
Ex and Ey
Get forward model
Mx and My, with dimensions
383  208  76 μm Calculate angle
= 90° Calculate new

NO Repeat 10
Calculate width

Calculate Repeated 3

=
1 + i
2

YES

Use
to adjust all
flaws in forward
model library

Figure 5. Outline of inversion calibration procedure.

, , width, and the new center are calculated by minimizing the difference between
Ex + iEy and the corresponding
(Mx + iMy).

920 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
 Experimental Compare to
data:calibration corresponding
flaw centered forward model to
and resampled calculate β
Vertical
β = 18643 e2.445i
Horizontal
Uncalibrated
Measured
forward model
(383 × 208 × 76 μm)
(383 × 208 × 76 μm)
Figure 6. Outline of inversion method starting with calibration data and leading to estimated flaw dimensions.
obtained from the calibrated FML. An outline of the inversion
method is shown in Figure 6; for more information, see other
scholarship (Shell et al., 2015; Sabbagh et al., 2013). This
procedure involves using a general nonlinear least squares
algorithm together with custom algorithms that compare the
experimental and generated flaw responses. Random initial
guesses of the desired parameters are supplied, and the algo-
rithms search for the combination of length, depth, width, and
orientation that minimizes the difference between the experi-
mental and simulated data. When comparing the experi-
mental and calibrated FML responses, data is compared along
nine horizontal lines that are spaced 304.8 μm (12 mils) apart in
the index direction. A comparison of the experimental data
and the calibrated forward model identified as the best match
by the inversion algorithms is shown in Figure 7.
Application to Experimental Data
To investigate the capabilities of the developed model-based
inversion method, it was used to estimate the length, depth,
width, and orientation of notches and cracks whose data
exhibited many different characteristics that might be
observed during depot inspections. First, relatively well-
balanced eddy current responses were characterized for
notches that were in nickel and titanium alloys; exhibited a
large range of length, depth, and width combinations; had
rectangular or semielliptical profiles; and were inspected at
Experimental data: Calibrated forward
Use β to scale uncharacterized model library used
forward models flaw centered to estimate
and resampled flaw size
Matching calibrated
Calibrated forward model Measured dimensions
forward model
(383 × 208 × 76 μm) (754 × 150 × 36 μm)
(732 × 158 × 25 μm)
many different orientations relative to the scan direction.
Next, the inverse method was applied to non-ideal notch
responses, which were obtained by including known amounts
of setback, rotation, and tilt during the experimental collec-
tion effort. Finally, the inverse method was applied to cracked
specimen response data to investigate characterization capa-
bilities for flaws resembling those present in actual engine
parts.
Case 1: Variations in Dimensions, Orientation, Geometric Profile,
and Material
The inversion method was applied to the experimental data
collected using a well-balanced probe for the three nickel
plates in Table 1. Notches on Plate #1 with planar areas
smaller than 0.0129 mm2 (20 mils2) are not part of this study,
because the background noise interferes with the inversion
method algorithms and leads to poor sizing estimates.
Additionally, four notches with lengths larger than 2286 μm
(90 mils) are not studied, because the inversion algorithms
have not been optimized for flaws of that size. The semiellip-
tical nickel forward model library was used for the inversion
procedure for all notches except for the five rectangular
notches on Plate #2, which used the rectangular nickel FML.
Initially, only the eddy current response data for notches
scanned in the transverse or longitudinal orientation was char-
acterized, and the estimated lengths, depths, widths, and
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 921
ME TECHNICAL PAPER w
x characterizing surface-breaking cracks

Horizontal Vertical

30
–1.27 –1.27 200
20
Index (mm)

Index (mm)
10 100

0 0 0
0
–10
–100
–20
1.27 1.27
–30 –200
(a)
–1.27 0 1.27 –1.27 0 1.27
Scan (mm) Scan (mm)

30
200
–1.27 20 –1.27

100
Index (mm)

Index (mm)

10
0 0 0 0
–10
–100
1.27 –20 1.27
–30 –200
(b)
–1.27 0 1.27 –1.27 0 1.27
Scan (mm) Scan (mm)

40 300
Experimental Experimental
Eddy current response
Eddy current response

Forward model 200 Forward model

20
100

0 0

–100
–20
–200

(c)–40 –300
–2.54 –1.27 0 1.27 2.54 –2.54 –1.27 0 1.27 2.54
Scan (mm) Scan (mm)

Figure 7. Comparison of: (a) an experimental response (notch size: 754  150  36 μm, = 90°); and (b) the matching calibrated forward
model (notch size: 726  150 x 36 μm, = 91.8°) identified by the inversion method, which (c) compares experimental and simulated
responses along nine horizontal lines

aspect ratios are shown in Figure 8, along with the differences the transverse and longitudinal orientations were considered.
between the estimated lengths and depths and their corre- There is excellent agreement between the estimated and
sponding measured values. known orientations.
Next, inversion was performed on data from notches on The inversion method was also applied to eddy current
Plate #2 that were oriented at 27 different angles relative to data for the notched titanium specimen (Plate #4, 2:1 aspect
the scan direction. The resulting dimension estimates ratios), which was collected using a well-balanced probe.
(Figure 9) exhibit more spread than was observed when only The estimated lengths and depths are shown in Figure 10.

922 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
 1.6 10

Estimated length (mm)

0.7

Estimated aspect ratio

Estimated depth (mm)

1.4
0.6 8
1.2
0.5 1 6
0.4 0.8
0.3 0.6 4
0.2 0.4
2
0.1 0.2
0 0 0
0 0.2 0.4 0.6 0 0.5 1 1.6 0 1 2 3 4 5 6 7 8 9 10
Known depth (mm) Known length (mm) Known aspect ratio
Plate #1 Plate #2 Plate #3 x=y Plate #1 Plate #2 Plate #3 x=y Plate #1 Plate #2 Plate #3 x=y
0.08 0.2 0.1

Estimated width (mm)

0.06 known length (mm) 0.15
known depth (mm)

Estimated minus
Estimated minus

0.08
0.04 0.1
0.02 0.05 0.06
0 0
–0.02 –0.05 0.04
–0.04 –0.1
0.02
–0.06 –0.15
–0.08 –0.2 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.5 1 1.5 0 0.02 0.04 0.06 0.08 0.1
Known depth (mm) Known length (mm) Known width (mm)
Semielliptical (L) Rectangular (L) ± 50.8 μm Semielliptical (L) Rectangular (L) ± 50.8 μm Plate #1(L) Plate #2 (L) Plate #3 (L) x=y
Semielliptical (T) Rectangular (T) Semielliptical (T) Rectangular (T) Plate #1(T) Plate #2 (T) Plate #3 (T)

Figure 8. Estimated depths, lengths, widths, aspect ratios, and differences between estimated and measured dimensions for notches in
nickel Plates #1-3 that were scanned in the transverse (T) and longitudinal (L) orientations.

1 300
Estimated orientation (°)
Estimated depth (mm)

0.3
Estimated length (mm)

Semielliptical Semielliptical Semielliptical

Rectangular Rectangular 250 Rectangular
0.25 0.8
x=y x=y x=y
200
0.2 0.6
0.15 150
0.4
0.1 100
0.2
0.05 50
0 0 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0 0.2 0.4 0.6 0.8 1 0 50 100 150 200 250 300
Known depth, mm Known length (mm) Known orientation (°)
(a) (b) (c)

Figure 9. Estimated (a) depths; (b) lengths; and (c) orientations for notches on Plate #2 that were positioned at 27 different orientations relative
to the scan direction.

To determine
, the semielliptical nickel FML was used with (coil setback), rotation, and tilt in two directions. For the
the experimental calibration data. This
was used to calibrate rotation and tilt cases, the experimental calibrations were
the rectangular titanium FML, which was then used to charac- performed with the probe rotation and tilt nominally set to 0°.
terize the titanium notches. The titanium grain noise repre- During an actual part inspection, if a probe had a tilted or
sented in the eddy current responses (Figure 10) can distort rotated coil, this characteristic would hold true during calibra-
the flaw response signals and is responsible for some of the tion. To account for this inconsistency between calibration and
error in the sizing estimates. inspection state, a 530  251  30 μm (20.85  9.9  1.2 mil)
notch on the plate was specified as the calibration notch for the
Case 2: Non-ideal Probe Properties or Surface Positioning inversion procedure, and the method used the notch-specific
The inversion method was applied to eddy current data dimensions to determine
. Since the 530  251 μm notch
collected on Plate #1 under conditions simulating probe liftoff was used for calibration, inversion estimates of its dimensions

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 923
ME TECHNICAL PAPER w
x characterizing surface-breaking cracks

0.6 1.2
Longitudinal Longitudinal
Estimated depth (mm)

Estimated length (mm)

0.5 Transverse 1 Transverse
x=y x=y
0.4 0.8

0.3 0.6

0.2 0.4

0.1 0.2

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.2 0.4 0.6 0.8 1 1.2
(a) Known depth (mm) (b) Known length (mm)

–1.27 –1.27
Vertical

Vertical
0 0

1.27 1.27

–1.27 0 1.27 –1.27 0 1.27

–1.27 –1.27
Horizontal

Horizontal

0 0

1.27 1.27

–1.27 0 1.27 –1.27 0 1.27

(c) Longitudinal (d) Transverse

Figure 10. Estimated: (a) depths; and (b) lengths for 2:1 notches in a titanium plate (Plate #4) that were scanned; (c) and (d) show the effects
of grain noise on the eddy current response for Notch #1.

0.8 1.8 0.1

Estimated depth (mm)

Estimated width (mm)

Estimated length (mm)

0.7 1.4
0.8
0.6 1.2
0.5 1 0.6
0.4 0.8
0.3 0.6 0.4
0.2 0.4
0.2
0.1 0.2
0 0 0
0 0.2 0.4 0.6 0.8 0 0.5 1 1.5 0 0.02 0.04 0.06 0.08 0.1
(a) Known depth (mm) (b) Known length (mm) (c) Known width (mm)
69 μm 137 μm 206 μm x=y 69 μm 137 μm 206 μm x=y 69 μm (L) 137 μm (L) 206 μm (L)
69 μm (T) 137 μm (T) 206 μm (T)
x=y

Figure 11. Estimated: (a) depths; (b) lengths; and (c) widths for notches in nickel (Plate #1) that were scanned in the transverse (T) and
longitudinal (L) orientations with controlled amounts of setback.

924 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
and orientations are not included in the results presented in
this section.
Experimental data was collected in a way that enabled a
controlled amount of additional liftoff between ~69 and
343 μm (~2.7 and 13.5 mils). In actual practice, coil setback
variations are not expected to exceed 76 μm (3 mils), and the
data presented in this section includes additional liftoff
ranging from 69 to 206 μm (2.7 to 8.1 mils). For the notches
on Plate #1, the inversion estimates for the lengths, depths,
and widths are shown in Figure 11. The estimated dimensions

0.6 1.6
Θx Θx
1.4

Estimated length (mm)

Estimated depth (mm)

0.5
1.2
0.4
1
0.3 0.8
0.6
0.2
0.4
0.1
0.2
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.5 1 1.5
Known depth (mm) Known length (mm)
0.8
Θy 1.6
0.7 Θy
1.4
Estimated length (mm)
Estimated depth (mm)

0.6
1.2
0.5
1
0.4
0.8
0.3
0.6
0.2 0.4
0.1 0.2
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.5 1 1.5
Known depth (mm) Known length (mm)
0.6
Θz 1.6
0.5 Θz
1.4
Estimated length (mm)
Estimated depth (mm)

0.4 1.2
1
0.3
0.8

0.2 0.6
0.4
0.1
0.2
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.5 1 1.5
(a) Known depth (mm) (b) Known length (mm)

Θ = –10° or 10° Θ = –5° or 5° Θ = 0° x=y

Figure 12. Estimated: (a) depths; and (b) lengths for notches with varying aspect ratios in nickel (Plate #1) that were scanned in the transverse
and longitudinal orientations. Experimental data was collected by carefully controlling coil rotation and tilt in two directions to represent angles
ranging from -10° to 10°.

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 925
ME TECHNICAL PAPER w
x characterizing surface-breaking cracks

show a much greater spread than was observed in Figure 8.

The inversion method appears to have largely overestimated
the widths of the transverse notches and underestimated the 0.5
widths of the longitudinal notches. The effects that additional 0.45 Longitudinal
Transverse
liftoff has on the eddy current response signal are not suffi- 0.4 x=y

Estimate depth (mm)

ciently accounted for through model calibration. 0.35
The rotation (Z) and tilt (X, Y) angles considered in 0.3
this paper are –10°, –5°, 0°, 5°, and 10°. A well-balanced probe 0.25
was used for the rotation cases, and the minimal contact area
0.2
probe, which exhibited more signal asymmetry even when
0.15
not tilted, was used for the tilt cases. Data for both transverse
and longitudinal flaw orientations were characterized using 0.1
the inversion algorithm (Figure 12). Estimates for three eddy 0.05
current datasets for the 460  79  28 μm notch (X = 10°, 0
0 0.1 0.2 0.3 0.4 0.5
transverse; Y = –10°, transverse and longitudinal) and (a) Assumed depth (mm)
one eddy current dataset for the 249  122  28 μm notch
(Y = –10°, longitudinal) are not included in Figure 12 due to 1.1
very inaccurate inversion results. The poor length  depth 1 Longitudinal
Transverse
estimates for the 460  79 μm notch are 2458  15, 21  0.9 x=y

Estimate length (mm)

1016, and 3048  21 μm, and for the 249  122 μm notch, 0.8
the estimate is 3048  13 μm. 0.7
0.6
Case 3: Application to Cracked Specimens 0.5
The inversion algorithms were used with updated forward 0.4
model libraries to estimate lengths and depths for a set of 0.3
eddy current responses for cracks in nickel. The estimated 0.2
depths and lengths are shown in Figure 13 for cracks with 0.1
measured lengths larger than 406 μm (16 mils). The assumed 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
depths in Figure 13 were determined by using measured (b) Known length (mm)
lengths with an assumed aspect ratio. Estimates for smaller
cracks are not included, because, for many of these crack
responses, the background noise was large relative to the flaw Figure 13. Estimated: (a) depths; and (b) lengths for cracks in nickel
response, and the noise resulted in poor centering leading to with measured lengths greater than 406 μm (16 mils) that were
inaccurate inversion estimates. scanned in the transverse and longitudinal orientations.

0.4 1.2 8
Estimated length (mm)
Estimated depth (mm)

Longitudinal Longitudinal Longitudinal

Estimated aspect ratio

0.35 7
Transverse 1 Transverse Transverse
0.3 x=y x=y 6 x=y
0.25 0.8 5
0.2 0.6 4
0.15 3
0.4
0.1 2
0.2
0.05 1
0 0 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0 0.2 0.4 0.6 0.8 1 1.2 0 1 2 3 4 5 6 7 8
(a) Known depth (mm) (b) Known length (mm) (c) Known aspect ratio

Figure 14. Estimated: (a) depths; (b) lengths; and (c) aspect ratios for cracks in nickel specimens scanned in the (T) transverse and (L)
longitudinal orientations. Specimens were broken open to obtain known depths and lengths.

926 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
 Eddy current response data was collected for five cracked
specimens which were subsequently destructively character-
ized to measure the depths. The inversion method was
applied to this eddy current data and inversion estimates of
the depths and lengths for the largest three flaws are shown in
Figure 14. The smallest two flaws had measured dimensions
equal to 356  102 μm (14  4 mils) and 422  74 μm
(16.6  2.9 mils), and the inversion method provided poor
size estimates for these cracks. The background noise was
large relative to the signal and negatively affected the flaw
centering algorithms. Additional inversion algorithm develop-
ment and experimental data preprocessing could be
performed to improve the estimates for these small flaws.
Discussion
The developed model-based inversion method provided
very good estimates of depths and lengths for transverse and
longitudinal notches in nickel that were inspected using a
well-balanced probe. Of the 64 characterized eddy current
responses (Figure 8), only one depth estimate differed from
its known value by more than 50.8 μm (2 mils). Four length
estimates differed from their measured values by 50.8 to
76.2 μm (2 to 3 mils), and four estimates were greater than
76.2 μm (3 mils). The flaw responses leading to length differ-
ences exceeding 76.2 μm (3 mils) were all associated with
notches that were either 958 or 1529 μm (37.7 or 60.2 mils)
long. The forward model library has more simulated data for
smaller lengths (254, 508, 1016 μm), than larger lengths
(2032, 3048 μm). When using cubic spline interpolation to fit
flaws in the 1016 to 1524 μm range, the interpolation relies on
data for significantly larger flaws, which could be contributing
to the inaccurate length estimates. All depth estimates for the
rectangular nickel flaws were within 25.4 μm (1 mil) of the
known values and all length estimates were within 35.6 μm
(1.4 mils) of the known values. The estimated widths for the
transverse notches were closer to the measured widths than those
estimated for the longitudinal notches. The widths for longitu-
dinal notches may be less accurate, in part, due to the algorithms
using seven horizontal lines, spaced 304.8 μm (12 mils) apart, to
compare the experimental and simulated data. For transverse
notches, the points compared are spaced 25.4 μm (1 mil) apart
in the flaw width direction, while they are spaced by 304.8 μm
(12 mils) across the width for longitudinal notches.
The inversion method estimated aspect ratios ranging
from 1:1 to 7.5:1 and accurately identified flaw orientations
relative to the scan direction. These capabilities represent an
important improvement over existing amplitude-based
methods that are dependent on an assumed aspect ratio and
flaw orientation. Existing methods would estimate very similar
lengths and depths for each group of five flaws on Plate #2
that had similar planar areas, and are incapable of determining
flaw orientation. While excellent orientation estimates were
obtained, the accuracy of the depth and length estimates was
inconsistent across the different angles.
The probe inventory used for depot inspections of USAF
engine parts contains probes with variations in the coil posi-
tioning, which can result in the non-ideal eddy current
responses investigated in the second case study. To use the
inversion method to characterize flaws detected during depot
inspections, it must consistently provide acceptable sizing
estimates for data collected with coils exhibiting variations in
setback, rotation, or tilt. For probe rotation angles (
Z)
ranging from –10° to +10°, the mean and standard deviation
of the absolute differences between the estimated and known
depths are 27.4 and 25.6 μm (1.08 and 1.01 mils), respec-
tively, while the mean and standard deviation for the absolute
length differences are 44.2 and 40.6 μm (1.74 and 1.60 mils),
respectively. Greater deviations between the estimated and
known dimensions were observed for the tilt cases; however,
many of the larger deviations are associated with ±10° tilt
angles, which are approaching the bounds of what is accept-
able for existing probes. The current work does not address
combinations of challenging scenarios such as simultaneous
tilt and rotation, nor does it investigate the ability to handle
many different flaw orientations when non-ideal probe char-
acteristics are present.
The existing forward model libraries can be expanded to
increase robustness and likely enable better characterization
of cracks and larger flaws. Forward models for cracks (very
small widths) are currently obtained by extrapolating from
forward models with widths of 25.4 and 76.2 μm (1 and 3 mils).
Flaws with lengths greater than 1016 μm (40 mils) use cubic
spline interpolation that depends on generated models with
large variations in lengths (the only simulated lengths above
508 μm are 1016, 2032, and 3048 μm). Future work can
involve expanding the number of lengths, depths, widths,
and angles in Table 2; replacing the interpolated data for the
762 and 1524 μm (30 and 60 mil) lengths with simulated
eddy current responses; and refining the sample spacing in
the simulated inspections, which is currently 127 μm (5 mils)
along the flaw length and 50.8 μm (2 mils) across the width of
the flaw. Additionally, ongoing related efforts are investigating
expanding the forward model libraries to include generated
eddy current signals that are representative of non-ideal probe
characteristics or probe positioning relative to the part surface
(Aldrin et al., 2016; Aldrin et al., 2017).
For relatively small flaws, the current inversion algorithms
can produce poor estimates due to problematic background
noise effects. Small flaws in titanium alloys are more chal-
lenging to size than those in the nickel alloys studied due to
the distortion of the eddy current response caused by grain
noise. Recent work has studied this challenge (McMahan
et al., 2017). Future work could investigate signal prepro-
cessing improvements to reduce detrimental noise effects and
consider a grain noise model as part of the inversion process.
The experimental data used for this investigation was
collected without a high pass filter, had relatively refined
sample spacing, and was collected using one inspection
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 927
ME TECHNICAL PAPER w
x characterizing surface-breaking cracks
technique on a flat surface. Typical surface data collected
during part inspections are collected using much coarser
sample spacing and with a high-pass filter that alters the flaw
signal. A previous study (Oneida et al., 2016) investigated the
possibility of converting the filtered, coarse data into the
format used in this paper. It involved resampling the data and
applying Wiener deconvolution to effectively remove the
effects the filter has on the data. After processing the filtered
data and applying the inversion method, estimates of the flaw
dimensions were in good agreement with those estimated
using data collected with refined sample spacing and without
a high pass filter. To use the inversion method as a general
tool for characterizing flaws located during depot inspections,
which include different inspection techniques, inspection
frequencies, filters, and probes, additional investigation and
development would be required.
Potential Applications
The primary application of this capability is for use as either a
full replacement for, or supplement to, amplitude-based sizing
for USAF depot inspections. Amplitude-based sizing is robust
for detection of potential defects. However, improvements
could be made to the characterization, classification, and
sizing capabilities of indicated defects. Amplitude-based
inspections blindly assume that all signals that make it
through signal processing are due to cracks. Model-based
inversion or filtering algorithms based on model-based inver-
sion could be used to classify between material noise,
geometry noise, material defects, and crack defects. As an
example, future work will focus on the characterization of
nonmetallic inclusions (NMIs) and classifying between NMIs
and cracks. This may allow parts with benign, but high ampli-
tude, response defects to remain in use through a secondary
analysis. A similar approach may help classify between
material noise caused by anisotropic conductivity in titanium
grain clusters and responses from defects.
Another important potential application of this inversion
method involves using it to estimate crack depths in POD
specimens, independent of the assumed aspect ratio of the set.
Due to the typical manufacturing process of cracks for POD
studies, there are occasionally shallow cracks that do not
match the assumed aspect ratio of the set. Currently, these
result in low responses and scatter in the data that are statisti-
cally interpreted as variability in the inspection technique
instead of inaccuracies in the estimated size of the crack.
The resulting inspection is penalized by the statistics of the
POD analysis, resulting in unjustified low thresholds and
more potential false indications and part dispositions. If there
is reason to suspect that a crack does not have the assumed
depth, model-based inversion could be used to estimate the
depth, and possibly justifiable removal of that specimen from
the POD dataset.
928 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Conclusions
A model-based inversion method was developed to use the
full eddy current response to estimate the length, depth,
width, and orientation of a flaw. To assess the accuracy of the
method, it was applied to a large number of experimental
eddy current response datasets, which represent the variety of
response signals encountered during typical depot inspec-
tions. For many notches scanned in the transverse or longitu-
dinal orientations using the well-balanced probe, there was
good agreement between the estimated lengths and depths
and their corresponding measured values. Additionally, the
inversion method very accurately estimated the notch orienta-
tion for all angles investigated. Less accurate dimension esti-
mates were obtained for the eddy current data associated with
the more extreme probe tilt/rotation angles and setback
distances. Smaller flaws, especially those embedded in
titanium, were more likely to be associated with poor inver-
sion estimates due to background noise interference. Future
work will focus on expanding the forward model libraries to
contain more simulated data, improving size estimates for
cracks and small notches, and more consideration for probe
construction variations. Overall, this inversion method shows
great promise for sizing flaws exhibiting a range of aspect
ratios and flaw orientations, which is a limitation of current
amplitude-based sizing methods.
ACKNOWLEDGEMENTS
The authors wish to acknowledge the technical interaction and support by
Siamack Mazdiyasni of AFRL/RXCA. Five cracked specimens with charac-
terization results were provided by Waled Hassan of Rolls Royce. This
work is supported by the US Air Force Research Laboratory (AFRL)
through Research Initiatives for Materials State Sensing (RIMSS) program
through UTC, Contract No: FA8650-10-D-5210, Agreement No. 12-
S7114-03-C1, and RIMSS II program through UTC, Contract No:
F48650-15-D-5231, Agreement No. 17-S8205-01-C1.
REFERENCES
Aldrin, John C., Harold A. Sabbagh, Elias Sabbagh, R. Kim Murphy, Mark
Keiser, David S. Forsyth, and Eric A. Lindgren, “Model Inverse Methods for
Bolt-Hole Eddy Current (BHEC) Inspections,” 40th Annual Review of
Progress in Quantitative Nondestructive Evaluation, Baltimore, Maryland,
21–26 July 2013, AIP Conference Proceedings, Vol. 1581, published online 2014.
Aldrin, J.C., E.B. Shell, E.K. Oneida, H.A. Sabbagh, E. Sabbagh, R.K.
Murphy, S. Mazdiyasni, and E.A. Lindgren, “Model-Based Inverse
Methods for Sizing Surface-Breaking Discontinuities with Eddy Current
Probe Variability,” 42nd Annual Review of Progress in Quantitative
Nondestructive Evaluation: Incorporating the 6th European-American
Workshop on Reliability of NDE, Minneapolis, Minnesota, 26–31 July
2015, AIP Conference Proceedings, Vol. 1706, published online 2016.
Aldrin, John C., Erin K. Oneida, Eric B. Shell, Harold A. Sabbagh, Elias
Sabbagh, R. Kim Murphy, Siamack Mazdiyasni, Eric A. Lindgren, and Ryan
D. Mooers, “Model-Based Probe State Estimation and Crack Inverse
Methods Addressing Eddy Current Probe Variability,” 43rd Annual Review
of Progress in Quantitative Nondestructive Evaluation, Vol. 36, Atlanta,
Georgia, 17–22 July 2016, AIP Conference Proceedings, Vol. 1806, published
online 2017.
Auld, B.A. and J.C. Moulder, “Review of Advances in Quantitative Eddy
Current Nondestructive Evaluation,” Journal of Nondestructive Evaluation,
Vol. 18, No. 1, 1999, pp. 3–36.
Bowler, John, “Review of Eddy Current Inversion with Application to
Nondestructive Evaluation,” International Journal of Applied Electromag-
netics and Mechanics, Vol. 8, No. 1, 1997, pp. 3–16.
Lindgren, Eric A., “US Air Force Perspective on Validated NDE-Past,
Present, and Future,” 42nd Annual Review of Progress in Quantitative
Nondestructive Evaluation: Incorporating the 6th European-American
Workshop on Reliability of NDE, Minneapolis, Minnesota, 26 –31 July
2015, AIP Conference Proceedings, Vol. 1706, published online 2016.
Liu, Xin, Yiming Deng, Zhiwei Zeng, Lalita Udpa, and Jeremy S. Knopp,
“Model Based Inversion Using the Element-Free Galerkin Method,”
Materials Evaluation, Vol. 66, No. 7, 2008, pp. 740–746.
McMahan, Jerry A., Jr., John C. Aldrin, Eric B. Shell, and Erin K. Oneida,
“Bayesian Flaw Characterization from Eddy Current Measurements with
Grain Noise,” 43rd Annual Review of Progress in Quantitative Nonde-
structive Evaluation, Vol. 36, Atlanta, Georgia, 17–22 July 2016, AIP
Conference Proceedings, Vol. 1806, published online 2017.
Mooers, Ryan D., John C. Aldrin, and Jeremy S. Knopp, “Realistic Split D
Differential Probe Model Validation,” 41st Annual Review of Progress in
Quantitative Nondestructive Evaluation: Vol. 34, Boise, Idaho, 20–25 July
2014, AIP Conference Proceedings, Vol. 1650, published online 2015.
Mooers, R.D., J.S. Knopp, J.C. Aldrin, and S. Sathish, “Simulated Para-
metric Study Based on a Representative Split-D Differential Eddy Current
Probe,” 40th Annual Review of Progress in Quantitative Nondestructive
Evaluation: Incorporating the 10th International Conference on
Barkhausen Noise and Micromagnetic Testing, Baltimore, Maryland,
21–26 July 2013, AIP Conference Proceedings, Vol. 1581, published
online 2014.
Norton, Stephen J. and John R. Bowler, “Theory of Eddy Current Inver-
sion,” Journal of Applied Physics, Vol. 73, No. 2, 1993, pp. 501-512.
Oneida, Erin K., Eric B. Shell, John C. Aldrin, Harold A. Sabbagh, Elias H.
Sabbagh, R. Kim Murphy, Siamack Mazdiyasni, and Eric A. Lindgren,
“Flaw Characterization using Inversion of Eddy Current Response and the
Effect of Filters and Scan Resolution,” 42nd Annual Review of Progress in
Quantitative Nondestructive Evaluation: Incorporating the 6th European-
American Workshop on Reliability of NDE, Minneapolis, Minnesota, 26–
31 July 2015, AIP Conference Proceedings, Vol. 1706, published online 2016.
Sabbagh, Harold A., R. Kim Murphy, Elias H. Sabbagh, and John C. Aldrin,
“Application of Model Based Inversion to Eddy Current NDT of
Heat Exchanger Tubing”, Materials Evaluation, Vol. 66, No. 7, 2008,
pp. 764–774.
Sabbagh, Harold A., R. Kim Murphy, Elias H. Sabbagh, John C. Aldrin,
and Jeremy S. Knopp, Computational Electromagnetics and Model-Based
Inversion: A Modern Paradigm for Eddy-Current Nondestructive Evaluation,
Scientific Computation, Springer-Verlag New York, New York City,
New York, 2013.
Sabbagh, L. David and Harold A. Sabbagh, “Eddy-Current Modeling and
Flaw Reconstruction,” Journal of Nondestructive Evaluation, Vol. 7, No. 1,
1988, pp. 95–110.
Shell, Eric B., John C. Aldrin, Harold A. Sabbagh, Elias Sabbagh, R. Kim
Murphy, Siamack Mazdiyasni, and Eric A. Lindgren, “Demonstration of
Model-Based Inversion of Electromagnetic Signals for Crack Characteriza-
tion,” 41st Annual Review of Progress in Quantitative Nondestructive
Evaluation: Vol. 34, Boise, Idaho, 20–25 July 2014, AIP Conference Proceed-
ings, Vol. 1650, published online 2015.
US Department of Defense, Handbook, Nondestructive Evaluation System
Reliability Assessment, MIL-HDBK-1823A, April 7 2009.
Yusa, Noritaka, “Development of Computational Inversion Techniques
to Size Cracks from Eddy Current Signals,” Nondestructive Testing and
Evaluation, Vol. 24, Nos. 1–2, 2009, pp. 39–52.
Yusa, Noritaka, Haoyu Huang, and Kenzo Miya, “Numerical Evaluation of
the Ill-Posedness of Eddy Current Problems to Size Real Cracks,” NDT &
E International, Vol. 40, No. 3, 2007, pp. 185–191.
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 929
ME TECHNICAL PAPER w
x
Toward Characterization of Single Crystal Elastic
Properties in Polycrystalline Materials using
Resonant Ultrasound Spectroscopy
by T.J. Lesthaeghe*, R.A. Adebisi†, S. Sathish†, M.R. Cherry‡, and P.A. Shade§
ABSTRACT
Resonance ultrasound spectroscopy (RUS) is a
nondestructive technique that exploits the natural
resonance behavior of a material to characterize its
elastic properties. The traditional RUS approach
utilizes an analytic approximation to determine
resonance behavior given a guess set of elastic
moduli. An optimization process is then used to
fit elastic properties to experimentally measured
resonance frequencies. This approach generally
requires certain limiting assumptions to be made
with respect to sample geometry, crystallographic
orientation, and requires single crystal samples
to obtain single crystal elastic properties. Toward
the goal of developing a measurement process
to obtain single crystal elastic properties on
polycrystalline aerospace alloys without the
aforementioned limitations, a framework has been
developed to enable this process on samples with
multiple grains, arbitrary sample shape, and
without restrictions to crystallographic orientation.
This framework utilizes off-the-shelf finite element
method software and optimization routines,
enabling rapid modeling and easy adaptation to
different scenarios. Testing of this framework was
performed on an amorphous specimen, a single
crystal specimen, and preliminary results show
* Center for Nondestructive Evaluation, Iowa State University, Ames, Iowa
50011; email tylerl@iastate.edu.
† Ph.D.; University of Dayton Research Institute, Dayton, Ohio 45469
‡ Air Force Research Laboratory, WPAFB, Ohio 45433
§ Ph.D.; Air Force Research Laboratory, WPAFB, Ohio 45433
930 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
promise on a nickel aluminide bicrystal, provided
that the sample geometry, density, and crystallo-
graphic orientations are adequately characterized.
KEYWORDS: resonant ultrasound spectroscopy,
finite element, elastic moduli, nonlinear optimization.
Introduction
Resonance ultrasound spectroscopy is a nondestructive char-
acterization technique that can be used to measure the elastic
moduli of a material by exploiting its natural resonance
behavior. The resonance frequencies of a material are
predominantly determined by geometry, density, grain struc-
ture, and elastic properties. If the former three properties are
adequately characterized, the natural resonance frequencies of
a sample can be experimentally measured and can solve the
inverse problem to determine elastic properties. The develop-
ment and history of RUS is described in outside works
(Maynard, 1996; Migliori and Sarrao, 1996).
Solution of the forward problem in conventional RUS
measurements, that is, computation of expected resonance
behavior given a guess set of properties, is performed as
detailed in a seminal paper on the topic (Visscher et al.,
1991). However, this technique generally requires that single
crystal elastic properties be obtained from precisely prepared
single crystal samples. The sample must have a specific
geometry (to be detailed momentarily) with crystallographic
axes appropriately aligned to the sample axes and the sample
mounted and excited during experimental measurement of
resonance frequencies as to assume free boundary conditions
(Migliori et al., 1993).
Solution of the forward problem when any of the assump-
tions above are invalid can be approached by utilizing finite
element method (FEM) software. However, this presents
challenges with respect to solving the inverse problem due to
the increased computational requirements. Limited work has
been done in this area; however, several approaches to
handling computational constraints have been proposed,
including a fixed-point iteration technique and an approach
utilizing a derivative matrix to determine optimal guess input
parameters (Liu and Maynard, 2012; Plesek et al., 2004).
Additionally, some recent work utilized genetic algorithms
with FEM to perform RUS on materials with high damping
and arbitrary geometry (Remmillieux et al., 2015).
This work details the development of a framework that
seeks to enable RUS on samples with arbitrary geometry,
multiple grains with varying crystallographic orientations, and
boundaries that may not be able to be assumed stress free.
Toward that end, the framework presented was tested on two
single crystal samples with simple geometries for comparison
to conventional RUS. Additionally, preliminary results on a
nickel aluminide (NiAl) bicrystal with dimensions on the
millimeter scale showed promising results, provided adequate
characterization of geometry, density, and crystallographic
orientation.
The work presented here is simply one step in the process
toward the development of techniques to determine single
crystal elastic properties in fine-grained aerospace alloys
without the need to grow sufficiently large single crystals. In
particular, this work provides the computational component
to a micro-RUS technique presently being studied by the
authors in which several grains may be involved in the RUS
process. Furthermore, this work seeks to enable computa-
tional study of the limitations of RUS measurement to obtain
single crystal properties in samples where multiple grains are
present.
Conventional Experimental Measurement Process
The experimental measurement process used in most typical
RUS measurements is well-detailed in literature (Maynard,
1996; Migliori et al., 1993). Briefly, the typical process
involves an apparatus designed to suspend a specimen of
interest on its corners with the minimum possible loading,
thus enabling the assumption of free boundary conditions.
The sample is suspended between two transducers in a pitch-
catch configuration. The sample is excited and a frequency
spectrum is collected. A schematic of this process can be seen
in Figure 1.
Excitation Receiving
transducer transducer
Function Oscilloscope
generator Specimen Amplifier or DAC
Figure 1. Schematic of typical resonance ultrasound spectroscopy
measurement setup.
In the conventional process, sample preparation is highly
critical. Determination of single crystal elastic properties
typically requires growing a single crystal of the material of
interest and then orienting so that major crystallographic axes
align with the sample axes. The sample is then polished using
precision procedures, such as those described elsewhere for
the rectangular parallelepiped geometry (Migliori et al.,
1993). Other canonical geometries can be used provided
that an appropriate solution exists for the forward problem
described in the next section. Such solutions have been
developed for geometries including, but not limited to,
rectangular parallelepipeds, cylinders, spheres, shells, and
cones (Visscher, 1991).
Conventional Forward and Inverse Problem Solution
The forward problem is conventionally solved by seeking
stationary points for the Lagrangian using formulations origi-
nally proposed by researchers and expanded on by others
and later refined to a general form (Anderson et al., 1973;
Demarest, 1971; Holland, 1968; Ohno, 1976; Visscher, 1991).
To briefly summarize, the Lagrangian L is defined as
L = V(KE – PE)dV, where KE is kinetic energy and PE is
potential energy. Kinetic energy and potential energy can
be defined in terms of the displacement field ui(x,y,z,t) =

u i(x,y,z)ejt, which is assumed to have a harmonic time depend-
ence t with angular frequency ω. Thus, the Lagrangian in
einstein notation can be written as:
⎛1 1 ⎞
(1) L = ∫ ⎜ ρω 2 ui ui − c ijkl ui , j uk ,l ⎟ dV
V⎝2 2 ⎠
where
ρ is density,
cijkl is the linear elastic stiffness tensor from Hooke’s law.
A solution with the rayleigh-ritz technique approximates
the displacement vector by an expansion of some set of basis
functions, which must meet boundary conditions and be valid
on the entire domain, given by:
N
(2) ui ( x , y , z) ≈ ∑ aipΦ p ( x , y , z)
p =1
where
aip are expansion coefficients,
p are the basis functions.
Substituting Equation 2 into Equation 1 produces:
1 2 1
(3) L≈ ω Mipjq aip a jq − Kipjq aip a jq
2 2
where
the mass matrix, M, is defined as follows.
(4) Mipjq = δ ij ∫ ρΦ pΦ q dV
V
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 931
ME TECHNICAL PAPER w
x characterization using rus
The stiffness matrix, K, is defined as follows.
(5) K ipjq = ∫ c ikjl Φ p , k Φ q ,l dV
V
The basis functions proposed by another author
(Visscher, 1991) are of the form in Equation 6.
(6) Φ p ( x , y , z) = x l ym zn
These are subjected to the truncation condition l + m + n
 N and the order of the corresponding mass and stiffness
matrices are thus on the order of R = 3(N + 1)(N + 2)(N + 3)
/ 6 (Visscher, 1991). The solution converges toward the exact
value as N approaches infinity; however, N must be limited
for computational concerns. In practice, N is typically chosen
such that 12  N  18 (Liu and Maynard, 2012).
Finally, applying Hamilton’s principle, and using r to
represent all combinations of ip and s for the combinations of
jq, the eigenvalue problem is established as Equation 7.
(7) ω 2 Mrs as = Krs as
This can be solved to determine the natural angular
resonant frequencies (eigenvalues) and the corresponding
mode shapes (eigenvectors).
The associated inverse problem is traditionally solved
using the levenberg-marquardt algorithm, which interpolates
between the gauss-newton technique and the gradient descent
technique (Marquardt, 1963). To use this algorithm, a target
function to be minimized is defined as:
( )
S 2
(8) F = ∑ wi ficalc − flmeas
i =1
where
ficalc and fimeas are sorted lists of length S respectively
containing calculated and experimentally measured
resonance frequencies,
wi[1,0] is a weighting function used to remove modes
poorly matched or missing/weak resonances.
Details on implementation are readily found in prior work
(Migliori and Sarrao, 1996).
Framework Development
As previously mentioned, the conventional RUS approach to
computing resonance frequencies based on the rayleigh-ritz
technique is limited to simple geometries. Additionally, while
crystal orientation and multiple grains could be incorporated
into the rayleigh-ritz approach, it comes with added
complexity. This work endeavors to develop an approach to
handling scenarios with arbitrary geometry and multiple
grains with varying orientation. Additionally, while the test
cases presented later all work with the assumption of free
932 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
boundary conditions, it is desirable to have flexibility with
respect to this limitation as well. These requirements motivate
the usage of FEM to calculate resonance frequencies.
Computation of Resonance Frequencies with Finite Element
Method
The forward problem is readily solved utilizing commercial or
homebrewed FEM codes. In this work, a commercial FEM
package was utilized. Models were constructed initially using
graphical user interface tools fully parameterized and without
any explicitly defined numeric values, and then exported for
usage in a scientific scripting language. This approach allows
rapid reuse of existing models where only modifications to
parameters are needed and also allows these models to be
incorporated into target functions for off-the-shelf optimizers.
Mesh sizes are set at the beginning of the optimization
process and are configured based on a common rule that
there should be 6 to 10 mesh elements per smallest wave-
length of interest (Marburg, 2008). A mesh refinement study
further validated that mesh error at frequencies of interest in
this work is at or below 0.3% of sufficiently converged
frequencies. Mesh error is at least an order of magnitude
smaller than optimization residual error. Typical forward
model computations on simple geometries with millimeter-
scale dimensions are on the order of 10 s. While this is several
orders of magnitude longer than comparable computations
with the rayleigh-ritz approach, it is still within the realm of
reasonable for a laboratory measurement technique being
performed on a desktop computer with reasonable consumer
hardware. Obviously, computing requirements increase as the
model becomes more complex.
Solving the Inverse Problem
As mentioned previously, the levenberg-marquardt algorithm
is generally used in traditional RUS to handle the inverse
problem. This requires the computation of the full gradient
of the target function with respect to all input parameters.
Additionally, the levenberg-marquardt algorithm does not
guarantee convergence to a global minimum. Initial attempts
to use this algorithm required many repeated runs of the opti-
mization process with varied initial conditions and bounds to
obtain good results. Given the possibility for meshing related
numerical error in the context of FEM, particularly if
attempting to fit geometric model parameters in addition to
elastic properties, the authors chose to look at other
approaches.
The open-source toolkit Implicit Filtering was chosen
instead (Kelly, 2011). The specifics of the algorithms used
can be found elsewhere, but briefly, it is a hybrid of a gauss-
newton technique and grid search (Kelly, 2011). Additionally,
matters such as scaling are handled automatically. It was
found to be highly efficient provided that the problem space is
reasonably well bounded.
 This algorithm performs the nonlinear least squares opti-
mization given by:
⎛ ⎡ meas
( ) (
T
⎜ ⎣ fi − f icalc ⎡⎣ p⎤⎦ ⎤⎦ ⎣⎡ fimeas − ficalc ⎣⎡ p⎤⎦ ⎦⎤ ⎟
(9) min ⎜ wi
p ∈η ⎜ 2 ⎟
⎝ ⎠
where
 = {pRO|Lj  pj  Uj},
p is a vector of length O containing input parameters to the
computational model,
Lj and Uj are lower and upper bounds for the jth input
parameter,
T indicates the vector transpose,
wi[1,0] is a weighting parameter used to remove poorly
matched modes and missing/weak resonances.
Additionally, for reporting purposes, two metrics were
used to evaluate error. A simple sum squared residual error is
given by Equation 10.
2 following equations (Siebörger et al., 2001)
( )
S
(10) essq = ∑ w i f imeas − f icalc ⎡⎣ p⎤⎦
i=1
Equation 10 provides a simple metric to compare error
between forward model runs, where fimeas is the same for all
values being compared. Alternatively, an average percent
difference metric is given by Equation 11.
1 S fimeas − ficalc ( p)
(11) e avdiff = ∑ wi meas
S i =1 ⎛ f i + f icalc ( p) ⎞
⎜ ⎟
⎜ 2 ⎟
⎝ ⎠
Equation 11 provides a comparable metric of performance
without consideration to the number of frequencies utilized;
however, it does introduce a frequency dependent bias.
It is also worth briefly discussing the number of frequen-
cies to utilize in the fitting process. Some resonance modes
are impacted by the same parameter pj in approximately the
same or similar fashion requiring the utilization of many more
than O resonances to obtain a successful fit (Migliori and
Sarrao, 1996). However, care and some manual manipulation
are generally required to ensure that modes are matched
correctly, as missing modes will cause frequencies above that
frequency to be fit incorrectly. The weighting parameter wi
can be used to provide a space for missing modes. Taking
repeated measurements in differing orientations can help to
fill in some of these gaps; however, this process may require
some trial and error. The test cases presented in the following
seek to obtain as many modes from experimental measure-
ment as reasonably possible.
Additionally, bounding and initial conditions are impor-
tant. Ideally, one would want to perform repeat optimization
processes with varied bounds and initial conditions to test
robustness of the fit. In the test cases presented as follows,
convergence studies with varied bounds and initial conditions
showed that bounds within ±10% of the minimum values
converged to the minimum in the first optimization run. In
the test case as follows for the fused silica cylinder, bounds
⎞ within ±50% would converge within two optimization runs;
)
however, this became much less true as the number of
⎟
parameters increased. Regardless, testing of robustness is
encouraged.
Furthermore, this process can be greatly aided by repara-
meterization of the independent elastic moduli and capital-
izing on known symmetries. In many specimens of interest,
including those examined in this study, there is either
isotropic or cubic crystal symmetry. In these cases, it is only
necessary to solve for two or three independent elastic
moduli, respectively. In the cubic case, for example, conven-
tional RUS code typically performs optimization in terms of
the independent elastic moduli c11, c12, and c44. However,
reparameterizing these into the directional Young’s modulus
E<001>, directional Poisson’s ratio v<001><>, and the zener
anisotropy ratio A was observed to improve performance of
the optimization process. These parameters are defined in the
2
c11 + c11c12 − 2c12
2
(12) E 001 =
c11 + c12
c12
v 001 =
(13) ⊥
c11 + c12
2c 44
A=
(14) c11 − c12
The effect of this reparameterization is that all resonance
frequency residuals (that is, all elements in the vector fimeas –
ficalc) are sensitive to changes in E<001>. The degree of sensi-
tivity varies depending on the resonance mode and degree of
misalignment between crystallographic and sample axes. In
the test scenarios presented as follows, the parameter E<001>
was observed to be quickly adjustable to dramatically reduce
residual error.
The code responsible for the optimization process was
segmented into three distinct parts so that starting the opti-
mization process requires providing one of each of three types
of snippets. The first snippet contains definitions of the neces-
sary model parameters, such as geometric dimensions and
density, experimentally measured resonance frequencies fimeas
and corresponding weight function wi, and initial conditions
and bounds on parameters to be optimized. The second
snippet is the objective function to be optimized, which takes
vector p with O input parameters and outputs a length S
vector containing wi(fimeas – ficalc[p]). This snippet contains
the necessary code to update the model on each iteration with
new input parameters, perform the model run, and return
results—a process seen outlined in the schematic in Figure 2.
The third snippet contains the code necessary to dynamically
build the forward model within the FEM software using the
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 933
ME TECHNICAL PAPER w
x characterization using rus

TABLE 1
Critical properties of specimens used in amorphous and single
crystal test studies
Constant model
parameters Property Fused silica Nickel superalloy
Type Amorphous Single crystal
Shape Cylinder Parallelepiped
Length (mm) – 2.86
Nonlinear Width (mm) – 3.89
Finite element Height (mm) 12.71 3.96
least forward model Implicit
squares filtering Diameter (mm) 25.98 –
(target function)
algorithm Chamfer (mm) 0.9 –
Density (g/cm3) 2.19 8.66

Error metric However, a hardware lock-in amplifier was utilized to provide

Elastic properties
and other input
a direct output of the frequency spectrum.
parameters
Tests on Amorphous and Single Crystal Samples
To ensure proper operation of the framework on cases easily
validated independently, experiments were performed with
two single crystal specimens: a fused silica cylinder and a
Initial conditions Termination
for input condition nickel superalloy parallelepiped. Table 1 presents the
parameters reached geometry and other necessary parameters for the two speci-
mens. Table 2 presents the results of the optimization process
on the fused silica cylinder, along with comparative results for

Figure 2. Nonlinear least squares fitting process for elastic properties

TABLE 2
with the finite element method.
Fit elastic properties for the fused silica cylinder used in the single
crystal test study*

scripting interface. Each of these three snippets can be inter- Technique c11 c44 Residual
changed independently, allowing the same code base to be (GPa) (GPa)
quickly adapted to handle new problems and configurations. Ultrasonic testing wave speed 76.2 30.9 –
Rayleigh-ritz 78.8 31.3 12.0
Framework Testing FEM without chamfer 78.9 31.5 9.71
FEM with chamfer 77.7 31.2 0.94
A photograph of the experimental measurement apparatus
utilized in test experiments is shown in Figure 3. The process *Residual error is as defined in Equation 10 with frequencies in MHz;
FEM = finite element method.
utilized here is similar to those described previously.

Figure 3. Experimental resonance measurement setup with nickel

aluminide bicrystal. Figure 4. Sketch of fused silica cylinder with chamfer.

934 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
TABLE 3
Fit elastic properties for nickel superalloy parallelepiped used in the single crystal test study*
Technique c11 (GPa) c12 (GPa) c44 (GPa) Residual
Rayleigh-ritz 254.1 162.6 130.2 262.9
Finite element model 255.1 162.8 131.6 34.69
* Residual error is as defined in Equation 10 with frequencies in MHz.

elastic moduli obtained using results obtained using the TABLE 4

rayleigh-ritz technique, ultrasonic wave speed measurement, Critical properties of single crystal sections from NiAl bicrystal
and results obtained using the FEM approach with two including crystallographic orientation measured using EBSD
different variants of the geometry. One FEM geometric defined with euler angles in passive bunge notation
variant assumes a perfect cylinder, as does the rayleigh-ritz Property Left section Right section
technique. However, the physical test specimen has a small Length (mm) 2.89 3.01
chamfer on each end, illustrated by the sketch in Figure 4. Width (mm) 3.08 4.54
This feature was readily incorporated into the FEM model Height (mm) 3.67 3.05
and results are included in Table 2. Density (g/cm3) 5.88 5.90
A quantity of 31 experimentally measured resonance Z = α (°) 36.0 43.2
modes were utilized in the fitting process for the rayleigh-ritz X′ = β (°) 57.3 91.8
Z′ = γ (°) 350.0 179.0
technique and FEM, with one additional mode being
weighted out since it was missing from the experimental
data but identified by both forward models.
As can be seen, the various techniques all produce compa-
rable results; however, the FEM-based inversion process of the bicrystal. To briefly review the theory, crystallographic
produced results with the lowest residual error. The presence orientation was introduced into computations of resonance
of the chamfer in the FEM model did not substantially change by applying a coordinate transformation to the elastic stiffness
the optimized elastic moduli, but it did lower the residual matrix cij. The coordinate transformation is given by the
error by a factor of approximately 10. expression:
Table 3 displays results of optimization on the nickel
superalloy parallelepiped. This is compared against the values
obtained utilizing the rayleigh-ritz technique. A quantity of (15) c′ij = Rc ij RT
35 experimentally measured resonance modes were used in
the fitting processes. The results show good agreement
between the two approaches; however, again, residual error where R is the 6  6 matrix defined as:
was lower utilizing the FEM approach.
⎛ ⎞
⎜ T112 T122 T132 ⎟
Tests on Single Crystal Sections of Nickel Aluminide Bicrystal
⎜ T212 T222 T232 ⎟
The NiAl bicrystal utilized in this effort was originally manu- ⎜ ⎟
factured by diffusion bonding together two pieces of NiAl ⎜ T312 T322 T332 ⎟
⎜ ⎟
single crystal sections cut into different orientations (Miracle, ⎜ T21T31 T22T32 T23T33 ⎟
1990). Since it was received in this form, sections were taken ⎜ ⎟
⎜ T T
31 11 T T
32 12 T33T13 ⎟
from each end to perform initial testing and to obtain better
⎜ T11T21 T12T22 T13T23 ⎟
estimates of the expected elastic properties prior to work with
(16) R = ⎜ ⎟
the bicrystal. The dimensions and other critical properties are ⎜ ⎟
displayed in Table 4. This table also displays crystallographic ⎜ 2T12T13 2T13T11 2T11T12 ⎟
⎜ ⎟
orientations as measured utilizing electron backscatter diffrac- ⎜ 2T22T23 2T23T21 2T21T22 ⎟
tion (EBSD). These measurements were taken with a ⎜ 2T12T33 2T33T31 2T31T32 ⎟
⎜ ⎟
commercial scanning electron microscope operating at 20 kV. ⎜ T22 2T33 + T23T32 T21 2T33 + T23T31 T22 2T31 + T21T32 ⎟
Maps across the entire surface were taken with a 2 μm step ⎜ ⎟
size. ⎜ T12 2T33 + T13T32 T13 2T31 + T11T33 T11 2T32 + T12T31 ⎟
⎜ T 2T + T T ⎟
A challenge that arose during this testing step came as the ⎝ 12 23 13 22 T13 2T21 + T11T23 T11 2T22 + T12T21 ⎠
result of miscommunication of reference frames between
EBSD data and the FEM forward model for the right section

JUNE 2017 • MATERIALS EVALUATION 935

ME TECHNICAL PAPER w
x characterization using rus

TABLE 5
Preliminary optimizations performed on right single crystal section of the NiAl bicrystal to resolve error in crystallographic orientation
reference frame. Permutation 3 resulted in lower residual error and it was later confirmed with electron backscatter diffraction that this
was the correct orientation*
Property 1 2 3 4 5 6
Length (mm) 3.05 3.01 3.01 3.05 4.54 4.54
Width (mm) 3.01 3.05 4.54 4.54 3.05 3.01
Height (mm) 4.54 4.54 3.05 3.01 3.01 3.05
c11 (GPa) 156.0 155.9 203.9 204.6 197.4 190.4
c12 (GPa) 79.2 79.0 137.6 138.6 132.5 123.6
c44 (GPa) 96.2 96.3 110.9 111.2 111.7 109.2
Residual 27 077.4 27 280.5 2217.1 2365.5 3200.7 3897.4
* Residual error is as defined in Equation 10 with frequencies in MHz.

T is the 3  3 transformation matrix given by: correct reference frame. It is worth noting the similarity
between residual error between several of these data points.
⎛ cos α cos γ − sin α cos β sin γ ⎞ This is readily explained by the dimensions of two sides of the
⎜ ⎟ parallelepiped being similar. It is also important to note that
⎜ sin α cos γ + cos α cos β sin γ ⎟
⎜ ⎟ this scenario only worked because the nature of the error was
⎜ sin β sin γ ⎟ well understood. In general, it is necessary to expect crystallo-
⎜ ⎟ graphic orientation to be known parameters (with perhaps
⎜ − cos α sin γ − sin α cos β sin γ ⎟
⎜ ⎟ some known measurement uncertainty) to be successful at
(17) T =⎜ − sin α sin γ + cos α cos β cos γ ⎟ obtaining fits.
⎜ sin β cos γ ⎟
⎜ ⎟ The resulting optimizations for elastic properties on both
⎜ ⎟ single crystal sections from the NiAl bicrystal are presented in
⎜ sin α sin β ⎟ Table 6. These fits utilized 42 and 40 experimentally
⎜ ⎟
⎜ − cos α sin β ⎟ measured resonance modes for the left and right sections,
⎜ cos β ⎟ respectively. Also presented are optimized values for crystallo-
⎝ ⎠
graphic orientation which align within expected EBSD meas-
where urement error (expected to be on the order of 1° due mostly

, , and  correspond to euler angles in passive bunge to alignment error) on the right section; however, optimized
notation (Auld, 1973). angles varied slightly more than expected on the left section.
It is probable that these parameters may have been capturing
Passive bunge notation describes a rotation of the sample error from elsewhere in the process—for example, errors in
coordinate frame to the crystal coordinate frame in which a dimensional properties or the sample not being a perfect
rotation
was first performed about the Z axis, then a parallelepiped. However, the resulting optimized elastic prop-
rotation  about the newly transformed X’ axis, and finally a erties still otherwise agreed with the values obtained from the
rotation  about the transformed Z’ axis.
Returning to the optimization challenges on the right side
of the bicrystal, it was known that the sample coordinate TABLE 6

frame was aligned in an orthogonal fashion with respect to the Fit elastic properties and crystallographic orientations for single
crystal sections from the NiAl bicrystal*
EBSD measurement frame; however, the face of the sample
measured was not the face expected. Given that this sample Property Left section Right section
was a rectangular parallelepiped with sample symmetry, this c11 (GPa) 202.4 203.0
specific case made it possible to check the six possible orienta- c12 (GPa) 134.5 135.4
tions using the FEM optimization process with the expecta- c44 (GPa) 114.4 114.8
Z = α (°) 33.2 43.3
tion that one of the orientations would result in lower residual
X′ = β (°) 60.9 92.3
error. This is effectively performed by permuting the dimen- Z′ = γ (°) 349.5 178.8
sions of the sample in the optimization code. This process was Residual 206.3 66.28
readily automated utilizing the framework. The results are Average percent difference 0.22% 0.15%
shown in Table 5. As can be seen, one permutation has
* Residual error is as defined in Equation 10 with frequencies in MHz, and
lower residual error than the others. It was later verified by average percent difference is as defined in Equation 11.
repeating EBSD measurements properly that this was the

936 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
right section—0.3% difference in c11, 0.7% difference in c12,
and 0.3% difference in c44 between the two sides. Addition-
ally, while expected values from literature varied depending
on exact atomic composition and temperature, the presented
results fell within at least several percent of literature estimates
(Miracle, 1993).
Tests on Nickel Aluminide Bicrystal
Initial attempts at fitting elastic properties on the NiAl
bicrystal produced results that were dramatically different
than what was to be expected. It became quickly apparent that
some of this was the result of similar issues related to crystal-
lographic orientation as previously described. Additionally,
some preliminary analysis of the sample under an optical
(a)
(c)
Figure 5. Micrographs of nickel aluminide bicrystal showing grain boundary (in μm) on all four sides: (a) front; (b) back; (c) top; and
(d) bottom.
microscope indicated that a more rigorous characterization of
the sample was going to be required to have a chance at
successfully fitting elastic properties.
The sample underwent a series of polishing operations to
enable optical microscope images to be captured indicating
the position and shape of the grain boundary on all sides. A
sample of these images is displayed in Figure 5. As can be
seen, the grain boundary is somewhat jagged at the
microscale; however, it is relatively straight macroscopically.
The assumption of a straight grain boundary will be kept in
place; however, it is of interest to explore sensitivity to this
feature.
Additionally, these micrographs reveal the presence of a
potential feature along the grain boundary. EBSD maps were
(b)
500 μm
(d)
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 937
ME TECHNICAL PAPER w
x characterization using rus
6 Critical properties of the NiAl bicrystal assuming a perfect
5
Misorientation from mean (°)
4
3
1 Since there was some uncertainty in density within reason-
1 mm
0
Figure 6. Stitched electron backscatter diffraction map from the
nickel aluminide bicrystal face shown in Figure 5a.
taken across the entire surface of the specimen, displayed in
Figure 6, that indicate the presence of small regions along the
grain boundary with misorientation of approximately 5 to 6°
from the mean orientation. These features appear to be
limited to within 100 μm of the grain boundary. It is not
believed that these will cause dramatic issues; however, sensi-
tivity to this feature is of interest for future examination.
It is also worth noting from the optical micrograph that
the sample is not a perfect parallelepiped. An advantage of the
FEM approach is the ability to easily incorporate 3D solid
model representations of a specimen, as opposed to using
canonical geometries. The dimensions for each face were
measured using the optical micrographs and were fit together
into a 3D model that attempts to capture some of these
geometric inaccuracies. The impact of this approach on fitting
elastic properties is to be studied further.
However, more importantly, this error can have a substan-
tial impact on calculations of density. Density was determined
by measuring mass and dividing by volume. Assuming a
perfect parallelepiped and using the averaged dimensions
displayed in Table 7 to compute density, a value of 5482 kg/m3
was obtained versus a value of 5534 kg/m3 obtained using the
volume computed from the 3D model. This is approximately
a 1% difference, which, since wave speed is proportional to
the relevant elastic modulus divided by density, results in a 1%
difference in the elastic modulus, depending on the value
used. It is also worth noting that these densities are almost 7%
different than the density values previously measured from
larger portions of this sample before sectioning and the single
crystal sections. The reason behind this discrepancy is not
known at this time.
938 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
TABLE 7
parallelepiped
Property Left grain Right grain
Length (mm) 1.35 2.13
Width (mm) 4.39 4.39
Height (mm) 1.75 1.75
Density (g/cm3) 5.53 5.53
Z = α (°) 184.11 288.46
X′ = β (°) 43.87 34.28
Z′ = γ (°) 86.00 286.31
able bounds, it would be useful to fit this as a parameter
during the optimization process. However, attempting to fit
both E and  at the same time would result in a situation
where increasing one parameter would cause an increase in
the other parameter and vice versa. Since there is some
numerical error present, this makes it easy for the optimiza-
tion process to find a local minima and give the appearance of
obtaining a proper fit if one is not aware of this issue.
However, an attempt can be made to identify this ratio, while
also simultaneously testing the robustness of fit on other
parameters of interest, by simultaneously fitting E and 
repeatedly with different initial conditions and bounds. In this
process, the ratio over a number of optimization processes
can be evaluated and can also see how well other parameters
consistently fit to the same values.
Additionally, to further simplify the process of performing
optimizations on this sample, it is recognized that the avail-
ability of whole surface EBSD maps provides a very accurate
representation of crystallographic orientation with respect to
the angle between the two grains. Absolute measurement
error for this equipment is less than 0.1°. Much of the error
in orientation will be the result of the sample axes not being
perfectly orthogonal to the EBSD reference frame, mostly due
to rotation about the out-of-plane axis in the EBSD frame.
This simplification can be utilized to avoid optimizing all
six euler angles and only optimize one parameter 1 which
rotates both sets of euler angles together about the Z axis.
This is not presented as an alternative to properly fitting all
values to obtain final results; however, it reduces the number
of parameters to speed up the following experiment.
With these considerations, a five-parameter optimization
process can be attempted on E, , A, , and 1. As mentioned,
this process was repeated many times with varied initial condi-
tions and bounds within 10% of the values from the single
crystal sections, yielding different results as the fit finds minima
along the line (E / ). The results are displayed in Table 8. A
quantity of 30 experimentally measured resonance modes were
included. It was readily observed that (E / ) = 4083.30 m/s.
TABLE 8
Results of optimization of the parameters E, ν, A, ϕ1, and ρ for varied initial conditions and bounds to fit the ratio (E / ρ)*
c11 c12 c44 E ν A ϕ1(°) ρ Residual Average (E / ρ)
(GPa) (GPa) (GPa) (GPa) (kg/m3) % difference (m/s)
209.83 142.32 114.83 94.7923 0.4041 3.4018 0.0962 5672.07 189.56 0.2700 4088.05
206.31 139.59 114.35 93.6518 0.4035 3.4274 0.1987 5618.46 189.28 0.2679 4082.72
210.15 141.99 116.79 95.6460 0.4032 3.4267 0.1826 5737.98 189.33 0.2677 4082.76
211.43 143.09 117.13 95.9246 0.4036 3.4277 0.1738 5754.79 189.18 0.2678 4082.73
201.33 136.23 111.86 91.3764 0.4036 3.4365 0.2289 5486.91 189.43 0.2671 4080.87
206.10 139.08 114.66 94.0231 0.4029 3.4217 0.1597 5637.51 189.31 0.2671 4083.89
212.38 144.23 116.11 95.7102 0.4045 3.4076 0.1319 5730.72 189.42 0.2711 4086.72
208.71 141.13 116.10 94.8478 0.4034 3.4357 0.2169 5694.66 189.28 0.2666 4081.12
207.68 140.57 115.15 94.1978 0.4036 3.4318 0.2312 5653.44 189.34 0.2670 4081.92
206.11 139.78 113.62 93.1315 0.4041 3.4261 0.1558 5586.50 189.26 0.2697 4082.99
199.53 134.64 111.19 91.0291 0.4029 3.4272 0.2283 5460.50 189.58 0.2653 4082.95
198.57 134.05 110.51 90.5252 0.4030 3.4256 0.1566 5429.83 189.31 0.2669 4083.11
206.03 139.65 113.69 93.1963 0.4040 3.4255 0.1891 5590.00 189.11 0.2689 4083.13
Mean 0.4036 3.4247 0.1807 189.34 0.2679 4083.30
Standard deviation 0.0005 0.0099 0.0409 0.13 0.0016 2.01
95% 0.40 3.42 0.16 189.26 0.2670 4082.09
Confidence 0.40 3.43 0.21 189.42 0.2689 4084.52
* It can be seen that the parameters ν, A, and ϕ1 are nominally constant. Residual error is as defined in Equation 10 with frequencies in MHz and average percent
difference is as defined in Equation 11.

TABLE 9
Estimated values of elastic properties computed using average (E / ρ), ν, and A as displayed in Table 8 for densities computed assuming
a perfect parallelepiped, the 3D model, and an average of previously measured densities for the NiAl bicrystal
Property Averaged dimensions 3D model Average of previous
Density (kg/m3) 5482 5534 5888
c11 (GPa) 201.44 203.35 216.36
c12 (GPa) 136.32 137.61 146.42
c44 (GPa) 111.51 112.57 119.77

Additionally, it can be seen that there is very little variance
in the resulting values of the parameters , A, and 1.
Accepting average values for those parameters, values of c11,
c12, and c44 can be computed utilizing density values for the
case with averaged dimensions, from the 3D solid model,
and using an average of previously measured densities from
this material. It can be seen in Table 9 that the density value
from the 3D model produces elastic properties that are very
similar to those from the single crystal section. Furthermore,
almost identical results were obtained by performing a
four-parameter optimization holding density constant at the
values shown in Table 9. It is clear that there is still error that
remains to be investigated. However, residual error was gener-
ally obtained that is similar to that from the single crystal
sections and obtaining estimated elastic properties that are
within approximately 1% of values from the single crystal
sections reported, as reported in Table 6.
Conclusions and Future Work
It has been shown that the framework developed in this work
is functioning as expected and generally producing acceptable
fits of elastic parameters. However, it is clear that adequate
characterization of specimens is critical to obtaining accurate
results. This becomes particularly true as sample sizes become
smaller, with geometry and material properties becoming
more difficult to measure, and when more than one grain is
incorporated.
Additionally, while the FEM approach does increase
computation time over the traditional rayleigh-ritz approach,
as a laboratory research technique, it is still well within the
realm of feasibility and is a viable approach to performing
inversions on RUS data from specimens with arbitrary
geometry and multiple grains with varying orientations. Addi-
tionally, the framework developed makes adapting to new
scenarios relatively quick and painless.

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 939
ME TECHNICAL PAPER w
x characterization using rus
Moving forward, sensitivity issues will be examined on the
NiAl bicrystal sample to gain a better feel for how the kinds of
uncertainty identified during the course of this study may
impact the ability to obtain accurate results with more compli-
cated scenarios.
It is also of interest to consider how the impact of addi-
tional grains may influence the ability to obtain accurate
single crystal properties. As the number of grains increases,
elastic property fits performed with RUS will lose sensitivity
to individual grains and will approach homogeneous bulk
elastic properties of the polycrystalline aggregate. Work by
others has demonstrated that this limit appears as the largest
grains approach 15% of the sample’s smallest dimensions in
pure copper samples (Kaplan et al., 2009). The framework
developed during this effort could be used to study this effect
from a purely simulation perspective. Additionally, efforts are
underway to enable this framework to be compatible with
data obtained from synthetic microstructure generation tools
that could facilitate such a study (Groeber and Jackson, 2014).
ACKNOWLEDGMENTS
The authors wish to thank Daniel Miracle, from the Air Force Research
Laboratory, for providing the NiAl specimen being examined in this
research effort. Additionally, the authors are grateful to John Aldrin, of
Computational Tools, for support provided during this effort. This work
was performed at Iowa State University and the Air Force Research Labo-
ratory. The authors acknowledge support from the Air Force Research
Laboratory under contract #FA8650-14-D-5224, task order #0004. The
authors also acknowledge support from the Materials & Manufacturing
Directorate of the Air Force Research Laboratory and the Air Force Office
of Scientific Research (program managers David Stargel and James
Fillerup).
REFERENCES
Anderson, O.L., E. Schreiber, and N. Soga, Elastic Constants and their
Measurements, McGraw-Hill, New York, NY, 1973.
Auld, B.A., Acoustic Fields and Waves in Solids: Vol. 1, Wiley, New York, NY,
1973, p. 74.
Demarest, H.H., “Cube-resonance Method to Determine the Elastic
Constants of Solids,” Journal of the Acoustical Society of America, Vol. 49,
No. 3B, 1971, pp. 768–775.
Groeber, M.A., and M.A. Jackson, “DREAM.3D: A Digital Representation
Environment for the Analysis of Microstructure in 3D,” Imaging Materials
and Manufacturing Innovation, Vol. 3, No. 5, 2014, pp. 1–17.
940 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Holland, R., “Resonant Properties of Piezoelectric Ceramic Rectangular
Parallelepipeds,” Journal of the Acoustical Society of America, Vol. 43, No. 5,
1968, pp. 988–997.
Kaplan, G., T.W. Darling, and K.R. McCall, “Resonant Ultrasound Spec-
troscopy and Homogeneity in Polycrystals,” Ultrasonics, Vol. 49, No. 1,
2009, pp. 139–142.
Kelly, C.T., Implicit Filtering, Society for Industrial and Applied
Mathematics, Philadelphia, PA, 2011.
Liu, G., and J.D. Maynard, “Measuring Elastic Constants of Arbitrarily
Shaped Samples using Resonant Ultrasound Spectroscopy,” Journal of the
Acoustical Society of America, Vol. 131, No. 3, 2012, pp. 2068–2078.
Marquardt, D., “An Algorithm for Least-squares Estimation of Nonlinear
Parameters,” Journal of the Society for Industrial and Applied Mathematics,
Vol. 11, No. 2, 1963, pp. 431–441.
Marburg, S., “Discretization Requirements: How many Elements per
Wavelength are Necessary?” Computational Acoustics of Noise Propagation
in Fluids – Finite and Boundary Elements Methods, edited by S. Marburg and
B. Nolte, Springer, Berlin, Germany, 2008, pp. 309–332.
Maynard, J., “Resonant Ultrasound Spectroscopy,” Physics Today, Vol. 49,
No. 1, 1996, pp. 26–31.
Migliori, A., and J.L. Sarrao, Resonant Ultrasound Spectroscopy, Wiley, New
York, NY, 1996.
Migliori, A., J.L. Sarrao, W.M. Visscher, T.M. Bell, M. Lei, Z. Fisk, and R.G.
Leisure, “Resonant Ultrasound Spectroscopic Techniques for Measure-
ment of the Elastic Moduli of Solids,” Physica B, Vol. 183, Nos. 1–2, 1993,
pp. 1–24.
Miracle, D., “The Deformation of NiAl Bicrystals,” Technical Report, WL-
TR-92-411, Wright Laboratory, Dayton, OH, 1990.
Miracle, D., “Physical and Mechanical Properties of NiAl,” Acta Metallur-
gica et Materialia, Vol. 41, No. 3, 1993, pp. 649–684.
Ohno, I., “Free Vibration of a Rectangular Parallelepiped Crystal and its
Application to Determination of Elastic Constants of Orthorhombic
Crystals,” Journal of Physics of the Earth, Vol. 24, No. 4, 1976, pp. 355–379.
Plesek, J., R. Kolman, and M. Landa, “Using Finite Element Method for the
Determination of Elastic Moduli by Resonant Ultrasound Spectroscopy,”
Journal of the Acoustical Society of America, Vol. 116, No. 1, 2004, pp. 282–287.
Remmillieux M.C., T.J. Ulrich, C. Payan, J. Riviere, C.R. Lake, and P.-Y. Le
Bas, “Resonant Ultrasound Spectroscopy for Materials with High Damping
and Samples with Arbitrary Geometry,” Journal of Geophysical Research:
Solid Earth, Vol. 120, 2015, pp. 4898–4916.
Siebörger, D, H. Knake, and U. Glatzel, “Temperature Dependence of the
Elastic Moduli of the Nickel-base Auperalloy CMSX-4 and its Isolated
Phases,” Materials Science and Engineering: A, Vol. 298, No. 1, 2001,
pp. 26–33.
Visscher, W.M., A. Migliori, T.M. Bell, and R.A. Reinert, “On the Normal
Modes of Free Vibration of Inhomogeneous and Anisotropic Elastic
Objects,” Journal of the Acoustical Society of America, Vol. 90, No. 4, 1991,
pp. 2154–2162.

Process Compensated Resonance Testing Models for
Quantification of Creep Damage in Single Crystal
Nickel-based Superalloys
Julieanne V. Heffernan*, Leanne Jauriqui†, Eric Biedermann‡, Alexander Mayes§, Richard Livings¶, Brent Goodlet**,
Siamack Mazdiyasni††
ABSTRACT
Process compensated resonant testing (PCRT) is a
full body nondestructive evaluation (NDE) method
that measures the resonance frequencies of a part
and correlates them to the part’s material state,
structural integrity, or damage state. This paper
describes the quantification of creep damage in a
virtual part population via the correlation of PCRT
parameters to creep strain using inversion
methods and vibrational pattern recognition (VIPR)
analysis. Modeled populations were created using
the finite element method (FEM) for single crystal
(SX) nickel-based superalloy dogbone and turbine
engine airfoil geometries. The modeled popula-
tions include nominal variation in crystallographic
orientation, geometric dimensions, and material
properties. Modeled populations also include parts
with variable levels of creep strain, allowing for
NDE sensitivity studies. FEM model inversion tools
quantified creep strain and distinguished it from
other variations in the part populations. Resonant
modes that were found to be particularly sensitive to
creep strain were evaluated using VIPR algorithms to
* Vibrant Corporation, 8330 Washington Pl. N.E., Albuquerque, New
Mexico, 87113-1674; (575) 418-0890; jheffernan@vibrantndt.com
† Vibrant Corporation, 8330 Washington Pl. N.E., Albuquerque, New
Mexico, 87113-1674; (575) 418-0890; ljauriqui@vibrantndt.com
‡ Vibrant Corporation, 8330 Washington Pl. N.E., Albuquerque, New
Mexico, 87113-1674; (575) 418-0890; ebiedermann@vibrantndt.com
§ Vibrant Corporation, 8330 Washington Pl. N.E., Albuquerque, New
Mexico, 87113-1674; (575) 418-0890; amayes@vibrantndt.com
¶ Vibrant Corporation, 8330 Washington Pl. N.E., Albuquerque, New
Mexico, 87113-1674; (575) 418-0890; amayes@vibrantndt.com
** University of California, Santa Barbara, Materials, E-II Bldg. 503, Rm.
1355, Santa Barbara, California; (720) 291-5613;
brent.goodlet@gmail.com
††Air Force Research Laboratory (AFRL/RXCA), Bldg. 655, Rm. 18,
Wright-Patterson AFB, Ohio; siamack.mazdiyasni@us.af.mil
ME TECHNICAL PAPER w
x
correlate and quantify creep strain with PCRT
metrics. The results for PCRT forward models,
model inversion, and VIPR correlations were
verified with experimental creep strain measure-
ments made for dogbone specimens. This verifica-
tion demonstrated that PCRT inspections can be
trained through forward models to detect and
quantify creep damage in a part.
KEYWORDS: creep, PCRT, resonance, forward
modeling, inversion, NDE, nickel-based superalloy.
Introduction
Process compensated resonant testing is a full body nonde-
structive evaluation method that uses multiple resonance
frequencies of a part to evaluate its structural integrity and
material states (ASTM, 2016). PCRT combines resonant
ultrasound spectroscopy (RUS) (Migliori et al., 1993),
pattern recognition analysis, and statistical scoring of
frequency data to perform pass/fail (P/F) NDE, process
monitoring, life monitoring, and material characterization for
commercial and scientific applications (Schwarz et al., 2005).
One of the most critical applications is the evaluation of gas
turbine engine airfoils made from nickel-based superalloys
(Piotrowski et al., 2008). Prior work on nickel-based super-
alloy gas turbine airfoils has shown PCRT capable of meas-
uring shifts in resonance frequency peaks which are correlated
with high temperature exposure (Piotrowski et al., 2008; FAA,
2010). Additionally, the NDE potential gained from coupling
forward finite element method (FEM) models of incremen-
tally crept dogbones to the PCRT measured changes in reso-
nance resulting from creep strain was explored for
polycrystalline nickel-based superalloy dogbones (Goodlet
et al., 2017; Biedermann et al., 2016).
PCRT applications use a training set consisting of compo-
nents that represent the range of acceptable process varia-
tion—one or more defective conditions—as well as the range
of severity in those defective conditions. The VIPR algorithms
use this training set to identify the most diagnostic resonances
for the condition(s) of interest (Jauriqui, 2010). VIPR readily
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 941
ME TECHNICAL PAPER w
x quantification of creep damage
identifies parts with resonance differences due to a defect,
however, it has not frequently been used to characterize or
quantify defects. Additionally, a significant number of targeted
defective samples are often hard to obtain from a service
population or are too costly to create experimentally. Forward
modeling in PCRT has the potential to greatly increase the
method’s material characterization capability while reducing
the need for large training populations of physical samples
(Biedermann et al., 2016).
PCRT inversion tools have been expanded to use the
mechanical resonances of a part to simultaneously estimate
varying material properties and changes in part geometry.
The capability of performing inversion using surrogate
models from FEM simulations has also been developed,
addressing crystal orientation and complex geometries
(Biedermann et al., 2016). Work continues towards the goal
of using inversion tools to identify the type and severity of
damage, on a path toward identifying structural deficiencies
with greater confidence. Further development of these NDE
tools holds the promise of reduced risk of in-service compo-
nent failure, and extension of serviceable component lifetimes
due to reduced uncertainty in life estimates.
This work explores the quantification of creep damage in a
virtual part population via the correlation of PCRT parame-
ters to creep strain using inversion methods and VIPR
analysis. FEM modeled design spaces of single crystal (SX)
nickel-based superalloy samples for dogbone and turbine
blade geometries were created. Design spaces included
models with nominal levels of variation in crystallographic
orientation, geometric dimensions, and material properties.
The design spaces also included models with excessive plastic
deformation to simulate creep strain. FEM inversion tools are
then demonstrated to successfully quantify creep strain from
natural parameter variations in the modeled part population.
With diagnostic modes found to be sensitive to creep strain
using VIPR algorithms, quantification of creep strain is also
demonstrated using traditional PCRT metrics. Finally, inver-
sion methods were verified with experimental data collected
from a SX dogbone crept incrementally. This work demon-
strates that FEM inversion and VIPR algorithms can be
trained through PCRT forward models to invert the quantity
of creep damage in a part.
Methods
Finite-Element Method Forward Modeling of Creep
A series of populations were created using FEM software for
SX superalloy dogbones and turbine blade geometries for the
purpose of developing virtual training sets of acceptable and
nonacceptable parts accounting for expected population vari-
ances of geometric, material, and defect creep elongation
parameters. Forward models were generated through FEM.
Modeling the shape change associated with creep was accom-
plished with a nonlinear plastic analysis, which is generally
942 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
more straightforward than a full creep plasticity analysis and
should yield very similar deformation characteristics. Defor-
mation of SX material required an anisotropic (cubic)
elastic/plastic material model using a Hill stress potential
function as detailed by other scholarship (Ramaglia and
Villari, 2013; Biedermann et al., 2017). After each creep strain
interval was modeled, the deformed shape was imported into
an FEM modal analysis, using a block lanczos eigensolver, to
predict the change in resonance frequencies resulting from
creep strain accumulation.
Forward model parameters of the 10 mm gage length SX
dogbone have been previously reported (Biedermann et al.,
2015), including dimensional variation in gage length, elastic
properties, crystallographic orientation, and accumulated
creep strain. To create a PCRT forward model sensitivity
study of creep for a more complex geometry, a 54 mm long
turbine blade with an internal cooling passage was also repre-
sented using models (Grabcad, 2016). Figure 1 shows the
dogbone and blade models. In the blade geometry the root
volume was fixed and a body force was applied to simulate the
boundary conditions and centripetal loads the blade could
experience in service. The angle () between the (001) crys-
tallographic axis (Z’) and the long axis of the airfoil (Z) was
varied to capture normal variation and potential casting
defects in grain orientation. To simulate the deformed
geometry associated with creep, the load applied to the airfoil
was adjusted to produce varying levels of nonlinear plastic
strain as measured along the trailing edge of the blade.
A series of one-factor-at-a-time (OFAT) studies varied
each geometric, material, and creep parameter individually to
evaluate its effect on resonance variation and establish a
normal distribution of parameters within the ranges specified.
Next, monte carlo (MC) simulations randomly selected
parameter values from within the specified ranges and gener-
ated FEM models to simulate the frequency profiles produced
by the monte carlo parameterization. Then, a selection of
normal design points were modeled with varying amounts of
creep strain. Table 1 describes the parameters and their
ranges chosen for the single crystal blade monte carlo design
spaces.
Previous results have quantified the overall uncertainty in
resonance frequencies resulting from coupled variation in
geometry, material properties, crystallographic orientation,
and creep damage for nickel-based superalloys (Biedermann
et al., 2017). For instance, error can be introduced with inac-
curate mode shape tracking. Each resonance mode has both a
characteristic frequency and mode shape, or pattern, in which
the part oscillates as it resonates. For inversion interpolations
to produce meaningful results, proper mode alignment
between the two FEM model runs is essential. However, this
process is complicated by the fact that changes in any param-
eter value can affect the ordering in which resonance modes
appear in the frequency regime. This mode-swapping requires
a robust mode shape based tracking and matching of modes.
 X Z’
Grip 2

Grip 1
θ
Z
Gage
(a)

X Z’

θ
Z

Core shift
ge
Root Leading ed Core
Suction side

Leading edge Pressure side

(b)
Trailing edge
Trailing edge

Figure 1. Diagrams: (a) dogbone geometry; and (b) and blade geometry. For each geometry, angle () is defined between the (001)
crystallographic axis (Z’) and the long axis of the dogbone or airfoil (Z).

TABLE 1
Coupled parameters for the SX blade forward model monte carlo design space studies*
Parameter description Parameter range
Normal population: geometry Airfoil thickness diameter ±0.02%
Airfoil span ±0.04%
Leading edge inflation/deflation ±0.01%
Trailing edge inflation/deflation ±0.01%
Core thickness diameter ±0.04%
Twist/untwist of airfoil ±2.6°
Core shift ±0.1% leading edge to trailing edge,
±0.1% pressure to suction side
Normal population: material Poisson’s ratio along crystal (001), (ν(001)) ±2%
Young’s modulus along crystal (001), (E(001)) ±2%
Anisotropy ratio, (A) ±3%
Density ±0.2%
Angle of (001) crystal axis rotation, (θ) ±15°
Defect population: geometry Same as normal population Ranges listed above
Defect population: material Added creep strain, as measured along trailing edge 0.5–10%
* Geometric ranges represent estimates of acceptable design tolerance.

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 943
ME TECHNICAL PAPER w
x quantification of creep damage
Mitigating mode matching error can be very difficult and
often involves direct comparison of all the mode shapes in a
design space. Fortunately, while resonance frequencies vary
significantly within the possible parameter space, resonance
mode shapes are much more consistent. Therefore, resonance
modes for each simulation were matched according to a series
of modal assurance criterion correlations, described in detail
in other work (Pastor et al., 2012).
FEM Inversion Method for Creep
Historically, inversion has been used to estimate bulk material
properties of parts with simple geometries (Migliori et al.,
1993; Migliori and Sarrao, 1997). In order to estimate more
complex material states and geometries, such as creep in a
coupon or turbine blade, an FEM model-based inversion
approach was chosen.
The purpose of inversion is to estimate a set of parameters
describing a part, such as creep strain, from a measured set of
the part’s resonant frequencies. This is achieved using an iter-
ative solving method, shown in Figure 2, wherein the user
gives the measured frequencies and an initial guess of the
parameters to an optimization routine. The optimization
routine then passes the guessed parameters to a model, which
outputs what the corresponding resonant frequencies should
be. The modeled frequencies are compared to the measured
sample frequencies and the optimization routine adjusts its
parameter estimates based on the resultant error. This process
Initial parameter
Street guesses
Inversion
Inversion
optimization
loop routine
Finish Best fit parameters
Figure 2. Flow diagram of the resonant ultrasound spectroscopy inversion process.
944 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
is repeated until the resonance mode shape error is below a
convergence threshold, at which point the current parameteri-
zation is returned as the best fit parameters.
The inversion process often requires many iterations to
converge to a solution, and running a full FEM model run at
each iteration could require hours, or even days, to complete a
single inversion analysis. One way to decrease inversion time
is to simulate a series of design points that spans the bounds
of the parameters of interest and then create a metamodel that
quickly interpolates between the generated frequencies.
The metamodeling approach described here is a gridded
design space. It involves creating an N-dimensional evenly
spaced grid of FEM design points and employing cubic spline
interpolation for estimations between design points. Gridded
design spaces offer good frequency estimates that increase in
accuracy with resolution of the design space. However, they
also require an exponential increase in the minimum number
of design points generated with each additional inversion param-
eter, as well as significant increases in interpolation time. This
makes gridded design spaces well suited for inversion problems
with four or fewer parameters, while greater payoffs come from
numerous inversion runs. Ultimately, the decision between
conducting inversions from interpolated gridded design space
data or performing FEM calculations on the fly must consider
the likely number of inversion runs, the number of invertible
parameters, the range of possible parameter values, and the
desired resolution of each parameter being estimated.
Measured resonance
frequencies
Model
Modeled resonance
frequencies
Residual error
calculation
TABLE 2
Dogbone parameter ranges for material properties, dimensions, and creep strain used for inversion studies*
Parameter Range
E001 ±2%
ν001 ±7%
A ±5%
θ 0–10°
Creep 0–8%
Gage length ±1.00%
Gage diameter ±0.75%
Grip lengths ±1.11%
* Creep deformation presented as % change in nominal gage length.
The inversion toolset used a gridded design space gener-
ated from forward models of dogbones that varied five
material properties: elastic modulus (E001), Poisson’s ratio
(001), anisotropy ratio (A), crystal orientation (
grain
angle), and creep strain. In addition to the design space,
several models were created that varied the material proper-
ties as well as the dimensional parameters of gage length, gage
diameter, and grip length. The material and dimensional
varied samples were used to gage the error in the inversion
code caused by extra variation that was not in the gridded
space. Table 2 provides nominal values and bounds for the
gridded design space as well as the dimensional parameter
bonds for the extra models.
VIPR Creep Correlation Method
VIPR is used to identify the resonance modes that are most
sensitive to various conditions of interest, or are most diag-
nostic of defective conditions. Depending on the variation
included in the acceptable and nonacceptable training sets,
the diagnostic pattern will often include peaks that are sensi-
tive to the defect, and some that are not. As such, VIPR
focuses on the relationships and patterns between resonances,
not the absolute frequencies themselves. Statistical scoring
using the mahalanobis-taguchi system (MTS) calculates the
mahalanobis distance which, in a PCRT context, is the
measure of a part’s position relative to the central tendency of
the reference population in N-dimensional frequency space
(Cudney, 2009). VIPR uses taguchi-based signal-to-noise
methods to evaluate thousands of combinations of resonances
from a database of candidates to discover the combination
that best separates acceptable from defective parts using the
MTS score (Taguchi et al., 2001; Schwarz et al., 2005).
While VIPR can easily identify resonance peaks sensitive
to creep (Biedermann et al., 2016), it had not been used as an
inversion tool to quantitatively evaluate the amount a
specimen has been crept. In contrast with the direct inversion
methods discussed in the preceding section, VIPR predictions
require relatively few resonance measurements. In this study,
Minimum Nominal Maximum
125.2 GPa 127.7 GPa 130.3 GPa
0.348 0.374 0.400
2.615 2.753 2.890
0° 0° 10°
0% 0% 8%
9.90 mm 10.0 mm 10.10 mm
3.97 mm 4.00 mm 4.03 mm
4.45 mm 4.50 mm 4.55 mm
forward model design spaces were used to train VIPR to
separate modeled samples with creep strain, from modeled
samples without creep damage. Then VIPR algorithms were
used to identify combinations of diagnostic modes that
produced the best correlations to creep damage. Once these
modes were identified, a VIPR-based correlation tool for
quantifying creep strain was established, and tested against
modeled and measured PCRT data.
Experimental Creep on Single Crystal Dogbones: Inversion
Validation
Iterative creep was performed on a SX dogbone in order to
produce measured PCRT data for validation of forward models
and to test the inversion methods discussed above. The single
crystal dogbone had the same geometry as described in other
works (Biedermann et al., 2016), while the experimental proce-
dures employed were similar to those described in previous
work on creep of polycrystalline superalloy dogbones (Goodlet
et al., 2017). First the dogbone was affixed to the load frame,
gripping the sample by a recessed ledge just beyond the gage
and filleted regions. A clam-shell furnace enclosed the sample
and the specimen grips, and was used to heat the dogbone to
950 °C (1223.15 K). Once the furnace temperatures stabilized
at the target temperature, a uniaxial load of 300 MPa was
applied and extensometer readings began. Two digital exten-
someters attached to a creep-resistant scaffolding aligned to the
dogbone grips recorded the elongation of the dogbone for the
duration of the test. When the desired amount of creep strain
had accumulated, the test stopped by unloading the specimen
and opening the clam-shell furnace for cooling in air. Using this
procedure, a single dogbone was iteratively crept for a total
of approximately 14% strain, as measured from the original
10 mm gage length. After each creep step the dogbone was
cooled to room temperature, measured with calipers to verify
the extensometer readings, and evaluated with PCRT. Both the
VIPR and FEM inversion approaches were then used to
quantify the creep elongation experienced by the dogbone
based on its measured resonance response.
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 945
ME TECHNICAL PAPER w
x quantification of creep damage

Results and Discussion

FEM Forward Modeling of Creep

Over 300 dogbone and 250 blade MC-generated design

points, with and without creep, were simulated. A baseline
model with nominal values and zero creep strain was estab-
lished for each specimen geometry. The first 50 resonance
modes of subsequent model runs were mode matched to align
with the established baseline mode order. Combining these
subsequent model runs, each with unique MC parameteriza-
tions, resulted in a diverse population of modeled results for
inversion estimation and VIPR training.
Figure 4. One-factor-at-a-time results for changes in resonance
The modeling results for SX dogbones were presented by frequencies of a simulated blade with increasing creep strain.
another author (Biedermann et al., 2015). Unlike the uniaxial
plastic strain deformation seen in the creep of the SX dogbone
samples (Biedermann et al., 2016), the blades do not deform
in a purely axial manner, as shown in Figure 3, due to their
complex geometry. However, in both the dogbone and blade
geometries, increasing creep strain often produced a decrease
in frequency for the sensitive resonance modes. Of the modes
identified as diagnostic modes substantially affected by accu-
mulation of creep strain, lower-frequency modes exhibiting
bending motions were most common.

Figure 5. Median variation of model frequencies in the blade monte

carlo populations, with creep (dashed orange line) and without creep
(a) (solid blue line).

Figure 4 shows the OFAT results from iterative creep

(b)
(c)
Figure 3. Sample dimensions depicted before creep strain was
applied (solid), and after (dashed): (a) axial deformation of an SX
dogbone; (b) twisting deformation (
c); and (c) bending deformation
along the long axis (D to D’) in a turbine blade.
modeling of blade geometries. Similar to the OFAT results,
creep populations have a general decrease in frequency varia-
tion within the monte carlo design space population. Figure 5
shows the median frequency variations of the blade monte
carlo design points, in those with and without creep.
The forward modeling results were imported into a virtual
training set database for analysis using PCRT tools. Creation
of virtual spectra (via superposition of lorentzian distribu-
tions) for each simulation also allows for a visual comparison
of the model data and identification of trends. Figure 6 shows
an example of seven virtual spectra created by successively
increasing creep strain on each blade simulation. As creep
strain increases, modes that are more sensitive to creep strain,
such as Mode 15, shift more than others (for example, Mode
16 and 17). The same trend can be seen in the OFAT results
presented in Figure 4.

946 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
 Eight additional models were created to test the inversion
15 16 17 code. The first four FEM design points, A, B, C, and D, varied
0% material properties randomly from within the bounds of the
gridded design space while the dimensions were fixed at
1% nominal values to confirm that the inversion would converge
to the correct values. Then four additional points, W, X, Y,
train
2%
and Z, were generated with random material and dimensional
ep s

3% variations. These later points helped to examine the effects of

g cre

normal dimensional variation on the inversion of creep.

asin

4% Initial guesses of each parameter were fed into the inver-

Incre

5%
6%
29.9 KHz
Figure 6. Virtual resonance spectra of modeled turbine blades with
increasing creep strain.
FEM Grid Inversion
All creep inversion work presented in this section pertains to
the 10 mm gage length SX dogbone as described in other
scholarship (Biedermann et al., 2016). A gridded design space
was created that varied 5 material parameters: elastic modulus
(E001), Poisson’s ratio (001), anisotropy ratio (A), crystal
orientation ( grain angle), and creep strain (Table 2).
sion code for each of the 8 design points. The inversion code
then used between 30 and 35 resonance peaks to estimate
creep strain. Several inversions were performed on each
design point using different starting guesses each time. Table
35.2 KHz
3 shows the known modeled creep strain, the best fit inverted
estimate for creep strain, as well as the difference between
them for each point. The error for samples A through D quan-
tifies typical levels of uncertainty due to interpolation within
the gridded design space and convergence of the inversion
code. The larger error seen in samples W through Z reflects
the additional uncertainty caused by dimensional variation. In
order for the inversion code to estimate creep strain, other
material properties of the samples had to be inverted. Table 4
shows the absolute material property values used in the
modeling as well as the best fit inversion results. Overall, both
the inversion of the creep and inversion of material properties
were well matched to the known (modeled input) values.

TABLE 3
Inversion results of creep for modeled inputs
Sample Modeled creep, % Best fit, inversion creep, % Difference
Material variation A 1.33 1.17 –0.16
Material variation B 2.67 2.55 –0.12
Material variation C 5.33 5.40 0.07
Material variation D 6.67 6.72 0.05
Material and dimensional variation W 3.90 3.08 –0.82
Material and dimensional variation X 1.30 1.42 0.12
Material and dimensional variation Y 4.56 4.64 0.08
Material and dimensional variation Z 6.12 5.70 –0.42

TABLE 4
Inversion results of material properties for modeled inputs
Sample Modeled Inversion best fit
E (GPa) ν A θ° E (GPa) ν A θ°
Material variation A 126.03 0.39 2.80 3.75 125.96 0.39 2.80 3.85
Material variation B 126.88 0.36 2.84 6.25 126.88 0.36 2.84 6.16
Material variation C 128.59 0.37 2.66 8.75 128.60 0.37 2.66 8.74
Material variation D 129.44 0.38 2.71 1.25 129.46 0.38 2.71 1.18
Material and dimensional variation W 125.76 0.36 2.84 0.72 125.74 0.36 2.84 0.00
Material and dimensional variation X 128.91 0.36 2.82 6.80 128.82 0.36 2.83 7.08
Material and dimensional variation Y 127.61 0.38 2.64 8.57 127.60 0.38 2.64 8.46
Material and dimensional variation Z 130.14 0.39 2.69 2.64 130.21 0.39 2.68 2.51

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 947
ME TECHNICAL PAPER w
x quantification of creep damage

VIPR Creep Correlation
The PCRT VIPR toolset generates statistical scores that are
typically used for P/F NDE of components. VIPR scores can
also be correlated to material conditions of interest in an
inversion/characterization capacity. In this study, VIPR’s
ability to quantify the creep severity was explored.
For the dogbone geometries, VIPR used data from both
the monte carlo forward models, as well as the grid-based
models used initially for inversion. The monte carlo forward
model and the grid-based inversion models had different
nominal modeled material properties, as shown in Table 5.
The differences could confound FEM inversion results, as the
absolute frequency residuals are a key part of the convergence.

TABLE 5
Variation in the modeled material properties used for VIPR predictions
E001 range ν001 range A range θ range
Inversion grid 125 to 130 GPa 0.374 to 0.400 2.62 to 2.89 0 to 10°
Monte carlo study 110 to 114 GPa 0.316 to 0.329 3.00 to 3.14 0 to 12°

10 4.0

Difference from predicted creep

3.0
8
B
2.0 A
6 D
1.0
Creep (%)

4 C 0.0

A –1.0
2 C D
B –2.0
R 2 = 0.995
0 –3.0
–20 0 20 40 60 0 5 10
(a) MTS score (b) Creep (%)

Validation Design FEM Inversion points

10 4.0
Difference from predicted creep

R 2 = 0.963
3.0
8 B
A
D 2.0
6 1.0
Creep (%)

B C 0.0
4
A –1.0
2 C D
–2.0
0 –3.0
–20 0 20 40 60 0 5 10
(c) MTS score (D) Creep (%)

Validation (Grid, MC) Design (Grid, MC) FEM Inversion points

Figure 7. Results: (a) vibrational pattern recognition’s correlation of mahalanobis-taguchi system to dogbone creep strain; (b) predictive ability
of vibrational pattern recognition and inversion tools when trained with gridded inversion space data and validated with monte carlo forward
model study data; (c) vibrational pattern recognition’s correlation of to mahalanobis-taguchi system creep severity, and predictive ability;
(d) when trained with a mix of gridded and monte carlo forward model study data.

948 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
However, VIPR finds optimal resonance patterns to diagnose blade population having a more complex geometry and larger
the creep condition. The absolute frequency values are not as geometric tolerances than those of the dog bone populations.
critical as the relational differences between frequencies. A summary of all the VIPR creep predictions are shown in
As an initial test, the gridded data from the FEM inversion Table 6.
were used as the VIPR training set, and the forward monte
carlo models were used as validation samples. The training set
included over 115 undeformed samples and over 230 samples
with creep. The validation set included more than 250 samples, 7
with and without creep strain. In the initial test, VIPR was 6 R2 = 0.771954
restricted to using the lower order modes (mode numbers 5
5 to 15). Modes below mode 5 were excluded because PCRT 4
measurements showed the highest measurement error at the

Creep (%)
3
lowest modes, where the fixture’s boundary conditions can
2
have a measurable effect. Modes above mode 15 were
excluded because mode-swapping in the modeled sets became 1
more prevalent in that range. 0
VIPR solutions developed from the gridded model design –1
set were evaluated and those with the best correlation –2
between MTS score and creep were chosen. For the 0 5 10
dogbones, the best VIPR correlation used 5 resonance modes, (a) MTS Score
as opposed to the 30 to 35 used in the inversion space. These 3
Difference in predicted creep
generated excellent predictions of creep levels for the valida-
2
tion monte carlo samples. The upper portion of Figure 7
shows a sample of these results. Figure 7a and 7b also high- 1
lights the inversion model design points A, B, C, and D.
Figure 7c and 7d shows an alternate solution where parts 0
from each the gridded and monte carlo data sets were used in
–1
the VIPR training set, with 15% of the data points overall
excluded for validation. –2
For VIPR analysis of the blade geometries, only the monte
carlo design spaces were used. The set included 149 samples –3
0 2 4 6 8
without creep, and 43 samples with levels of creep strain
(b) Creep (%)
ranging from 1.0% to 5.7%. Approximately 10% of the
samples were randomly withheld for validation. VIPR solu-
tions trained to the blades did not have as high a correlation Design Validation
between MTS score and creep strain as demonstrated by the
dogbone samples, but the error between the actual creep
Figure 8. Vibrational pattern recognition’s correlation of mahalanobis-
values and the predicted creep values was comparable. taguchi system to: (a) creep severity; and (b) predictive ability, when
Figure 8 shows the results for the blade geometries corre- trained with monte carlo forward model study data for blade
lating MTS score to creep. The lower correlation value, geometry. The dashed line is the proposed mahalanobis-taguchi
compared to the dogbones, may be due to the monte carlo system limit.

TABLE 6
Results of vibrational pattern recognition predictions of creep
Part Geometry Dogbone Dogbone Blade
Training/design data Inversion grid Grid, Monte carlo Monte carlo study
Validation data Monte carlo study Grid, Monte carlo Monte carlo study
Number of resonances in VIPR pattern 5 5 6
MTS-creep correlation 0.99 0.96 0.77
Maximum prediction in creep difference (design set) 0.57 2.67 2.6
Average predication in creep difference (design set) 0.0 0.0 0.0
Maximum prediction in creep difference (variation set) 3.2 2.18 2.7
Average prediction in creep difference (variation set) –0.57 0.03 –0.01

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 949
ME TECHNICAL PAPER w
x quantification of creep damage

TABLE 7
Inversion of material properties from measured data
Sample, measured creep strain, % Inversion best fit
E (GPa) ν A θ°
0 127.32 0.38 2.84 4.08
0.20 127.81 0.38 2.83 4.28
0.60 128.10 0.38 2.82 3.90
3.20 128.36 0.38 2.83 4.10
9.10 128.64 0.39 2.85 2.92

TABLE 8
Inversion and VIPR estimates of creep strain from measured data
Measured creep VIPR estimate VIPR difference Inversion, best fit Inversion error,
strain, % creep strain, % creep strain, % difference
0.0 4.30 4.30 0.00 0.00
0.20 4.60 4.40 0.19 –0.01
0.60 4.80 4.20 0.59 –0.01
3.20 7.10 3.90 2.87 –0.33
9.10 13.60 4.50 9.13 0.03

VIPR sorting scores are also used to set limits for P/F-type
sorting. In such an application, using the blade models as
training data, an MTS cutoff would be selected that correlated
to a worst allowable case level of creep strain (perhaps 2%).
From Figure 8, the MTS cutoff to reject parts with less than
2% creep strain would be MTS = 2 (dashed line). Parts
measured with an MTS score higher than two could be
rejected from an overhaul process stream, or segregated for
further evaluation.
Experimental Verification and Measured Data Inversion
A single crystal dogbone was iteratively crept (as described
earlier) and removed at intervals to make PCRT measure-
ments. The PCRT model inversion methods were applied to
the measured spectra for the crept dogbone. E(001), (001), A,
, and creep strain were inverted using the FEM approach,
while the VIPR approach was used for a creep-only estimate.
The inversion results for the material properties are shown in
Table 7. The error between the measured and inverted values
of creep are shown in Table 8. Figure 9 illustrates the differ-
ences between the known creep values and the inverted creep
estimates for both modeled and measured data inversion
inputs.
The inversion code had excellent estimations of creep
strain in the measured and modeled data. The VIPR method
showed high correlation in modeled data yet showed an offset
between the estimated and measured creep strains. However,
the offset of creep estimation was consistent (approximately
4%) across all creep levels, and the correlation trend agreed
with the trend in the modeled data. VIPR currently examines
the resonance patterns of a part relative to the entire trained
population. Future work will evaluate VIPR results by
comparing a part to itself, before and after service creep, to
evaluate if the correlation between MTS and creep deforma-
tion can be improved. This verification demonstrates that
both FEM and VIPR inversion can be trained through PCRT
forward models to invert the quantity of creep damage in an
SX dogbone.
In addition to estimating creep strain, the inversion code
produced estimates of the material properties for the modeled
and measured data. The true values of the measured data
material properties are unknown and verification is needed to
determine how close the inversion estimations are. To further
evaluate the inversion results, FEM models were created using
the best-fit inversion parameters of creep strain and material
properties. Figure 10 compares the change in frequency from
the undeformed state to 3% creep strain in the modeled and
measured data. While a slight offset exists between the
measured and modeled frequencies, the overall pattern is
consistent because the measured and modeled mode sensitivi-
ties are the same. In fact, the measured creep data include a
global frequency shift of approximately +0.3% after the first
creep increment, which was not anticipated by the models. A
similar anomalous global increase in resonance was reported
after the first creep increment of polycrystalline superalloy
dogbones by other authors, who postulate these observations
may be due to changes in material microstructure or the
residual stress state of the part (Goodlet et al., 2017).

950 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
 10 0.5

% Change in frequency
from undeformed state
Creep (% of nominal

8 0
gage length)

6 –0.5

4
–1

2
–1.5 Measured data
FE Model
0
A B C D –2
(a) Sample 7 17 27
10 Resonance index
Creep (% of nominal

8 Figure 10. Frequency changes for 3% creep strain in measured data

gage length)

creep (solid line) and forward model based on inversion results

6 (dashed line).

2 Conclusions and Future Work

0 This work demonstrates that a series of forward models can
W X Y Z be used to train FEM inversion and VIPR algorithms to
(b) Sample estimate the quantity of creep strain. Validation of forward
14 models with experimental data showed the good model
predictions of the resonance frequency changes in a range of
12
Creep (% of nominal

creep strains. This work also shows that in modeled data, a

gage length)

10 part’s resonant frequencies can be used to invert other

8 material and geometric parameters that describe a part
6 including E(001), (001), A, and grain angle (). Ultimately,
PCRT forward models and inversion can be expanded to
4
include additional geometries, damage mechanisms, and
2 inversion parameters. Future work will create and evaluate an
0 inversion gridded design space for the SX blade geometries.
0 0.2 0.6 3.2 9.1 Monitoring and measuring the change in resonance exhibited
(c) Sample by a single part throughout its service life would allow the
inversion process to focus on characteristics that have
Modeled input creep Bounds of design space
changed, without being confounded by sources of variation
VIPR best fit creep Inversion best fit creep
from the initial part state. For example, the primary grain
Measured creep
angle affects resonance measurements, but does not change
significantly during a service interval. Comparing a part to
itself over time should therefore highlight the effects that
damage accumulation has on resonance, while being immune
Figure 9. Plots of measured creep strain versus inversion estimates to the natural parameter variation observed in a population
from (a) modeled data with material variation; (b) modeled data with of parts.
material and dimensional variation; (c) measured data.
ACKNOWLEDGEMENTS
The authors wish to thank John Aldrin for his support during the course of
this work. This research was supported by the US Air Force Research
Laboratory (AFRL) through a Materials and Manufacturing Directorate
(AFRL/RX) Structural Materials Broad Agency Announcement (BAA)
Contract FA8650-15-M-5208 and an AFRL Small Business Innovation
Research (SBIR) Phase II Contract, FA8650-15-M-5074. This paper
has been cleared for public release by AFRL under case number
88ABW-2017-1323.

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 951
ME TECHNICAL PAPER w
x quantification of creep damage
REFERENCES
ASTM, ASTM E3081-16: Standard Practice for Outlier Screening Using
Process Compensated Resonance Testing via Swept Sine Input for Metallic
and Non-Metallic Parts, ASTM International, West Conshohocken,
Pennsylvania, 2016.
Biedermann, Eric, Leanne Jauriqui, John C. Aldrin, Brent Goodlet, Tresa
Pollock, Chris Torbet, and Siamack Mazdiyasni, “Resonance Ultrasound
Spectroscopy Forward Modeling and Inverse Characterization of Nickel-
Based Superalloys,” Review of Progress in QNDE, Vol. 34, AIP, 2015, pp.
835–844.
Biedermann, Eric, Leanne Jauriqui, John C. Aldrin, Alexander Mayes,
Tom Williams, and Siamack Mazdiyasni, “Uncertainty Quantification in
Modeling and Measuring Components with Resonant Ultrasound Spec-
troscopy,” 42nd Annual Review of Progress in Quantitative Nondestructive
Evaluation:, Incorporating the 6th European-American Workshop on Reli-
ability of NDE, Minneapolis, Minnesota, 26–31 July 2015, AIP Conference
Proceedings, Vol. 1706, published online 2016.
Biedermann Eric, Julieanne Heffernan, Alexander Mayes, Garrett
Gatewood, Leanne Jauriqui, Brent Goodlet, Tresa Pollock, Chris Torbet,
John C. Aldrin, Siamack Mazdiyasni, “Process Compensated Resonance
Testing Modeling for Damage Evolution and Uncertainty Quantification,”
43rd Annual Review of Progress in Quantitative Nondestructive
Evaluation: Vol. 36, Atlanta, Georgia, 17–22 July 2016, AIP Conference
Proceedings, Vol. 1806, published online 2017.
Cudney, Elizabeth A., David Drain, Kioumars Paryani, Naresh Sharma,
“A Comparison of the Mahalanobis-Taguchi System to A Standard
Statistical Method for Defect Detection,” Journal of Industrial and Systems
Engineering, Vol. 2, No. 4, 2009, pp. 250–258.
Federal Aviation Administration (FAA), AC 33.70-1: Guidance Material
For Aircraft Engine Life-Limited Parts Requirements, Advisory Circular
No 33.70-1, Initialed by ANE-111, 2010 US Department of Transporta-
tion, Washington, DC, 2009.
Goodlet, B.R., C.J. Torbet, E.J. Biedermann, L.M. Jauriqui, J.C. Aldrin,
T.M. Pollock, “Forward Models for Extending the Mechanical Damage
Evaluation Capability of Resonant Ultrasound Spectroscopy,” Ultrasonics,
Vol. 77, 2017, pp. 183–196.
952 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Grabcad, “CAD turbine blade model,” accessed 17 May 2017 from
https://grabcad.com/library/turbine-blade—4.
Jauriqui, Leanne and Lem Hunter, “A More Comprehensive NDE: PCRT
for Ceramic Components,” Review of Progress in Quantitative Nonde-
structive Evaluation: Volume 30AB, San Diego, California, 18–23 July
2010, API Conference Proceedings, Vol. 1335, published online 2011.
Migliori, A., J.L. Sarrao, William M. Visscher, T.M. Bell, M. Lei, Z. Fisk,
and R.G. Leisure, ‘‘Resonant Ultrasound Spectroscopy Techniques for
Measurement of the Elastic Moduli of Solids,’’ Physica B: Condensed
Matter, Vol. 183, No. 1–2, 1993, pp. 1–24.
Migliori, Albert and John L. Sarrao, Resonant Ultrasound Spectroscopy:
Applications to Physics, Materials Measurements, and Nondestructive
Evaluation, Wiley, New York City, New York, 1997.
Pastor M., Binda M., Harcarika T., “Modal Assurance Criterion” Procedia
Engineering 48 (2012) 543-548.
Piotrowski, D., L. Hunter, and Trista Sloan, “Process Compensated
Resonance Testing JT8D-219 1st Stage Blades,” ATA NDT Forum 2008,
(September 24, 2008). (http://www.vibrantndt.com/wp-content/
uploads/75_2008_ATA_NDT_Forum-PCRT_of_JT8D-T1_Blades1
.pdf).
Ramaglia, Alessandro D. and Paolo Villari, “Creep and Fatigue of Single
Crystal and Directionally Solidified Nickel-base Blades via a Unified
Approach Based on Hill48 Potential Function: Part 1 Plasticity and
Creep,” ASME Turbo Expo 2013: Turbine Technical Conference and
Exposition, San Antonio, Texas, 3–7 June 2013.
Schwarz, Jim, Jay Saxton, and Leanne Jauriqui, “Process Compensated
Resonant Testing in Manufacturing Process Control,” Material Evaluation,
Vol. 63, No. 7, 2005, pp. 736–739.
Taguchi, Genichi, Subir Chowdhury, and Yuin Wu, The Mahalanobis-
Taguchi System, McGraw-Hill Professional, New York City, New York,
2001.

Handheld Fourier Transform Infrared Spectroscopic
Characterization of Ceramic Matrix Composites
by Amanda K. Criner*, Christine Knott†, Megan Imel‡, Derek King§, and Ming Chen*
ABSTRACT
Surface chemical characterization of ceramic matrix
composites (CMCs) with carbon fiber, silicon
carbide (SiC) matrix, and various particulates
(hafnium diboride and silicon nitride) using
reflectance data obtained from a handheld fourier
transform infrared spectroscopy device is
presented. The models of the assumed physics
(including measurement error processes) are
discussed and then used to analyze data obtained
from the CMCs. The validity of the assumptions
that are implicitly made in the data analysis is
evaluated. Estimated probability distribution
functions of resonant wavenumbers (cm−1)
obtained from the data are discussed with regard
to identifying SiC and the presence of hafnium
diboride. The results presented here and associ-
ated techniques will be used to identify surface
chemical changes in the materials after they have
been exposed to high temperature environments.
KEYWORDS: ceramic matrix composites, fourier
transform infrared spectroscopy.
* Materials and Manufacturing Directorate, Air Force Research Laboratory.
† Materials and Manufacturing Directorate, Air Force Research Laboratory;
email christine.knott.1@us.af.mil
‡ Southwestern Ohio Council for Higher Education.
§ UES, Inc.
ME TECHNICAL PAPER w
x
Introduction
Ceramic matrix composites (CMCs) are a key material
system for improving the thrust-to-weight ratio in high
performance aircraft systems because of their high tempera-
ture performance and strength (Ohnabe et al., 1999). Some
current engines contain CMC turbine shrouds that line the
hottest zone. Engines that are scheduled to be released in
2019 will have two CMC combustor liners, two CMC
nozzles, and one CMC shroud (Levy, 2017). CMCs become
brittle when oxidized from exposure to high temperatures and
loading (Tressler, 1999). The oxidation of CMCs leads to a
catastrophic brittle failure mode rather than a more gradual
ductile failure mode (Wang et al., 1991). This motivates the
development of nondestructive techniques to detect and
quantify oxidation in CMCs. While the oxidation of CMCs
changes the mechanical properties of the material, the oxida-
tion caused by thermal exposure is a chemical process. These
chemical changes correspond to the formation of different
molecules, phases, or structures in unexposed CMCs, which
can be detected and characterized with spectroscopy.
Fourier transform infrared (FTIR) spectroscopy is a
powerful tool for chemical characterization. The development
of handheld FTIR spectroscopy devices was motivated by and
has enabled the nondestructive evaluation of materials for
thermal degradation in polymer matrix composites (Eibl,
2017; Rein and Seelenbinder, 2015). In polymer matrix
composites (PMCs), the absorbance is measured for these
materials because they are absorbing in the relevant infrared
wavenumber of 4000 to 650 cm−1. Absorbance data is linear
with respect to the underlying optical constants, which causes
the data analysis techniques, such as partial least squares
(PLS) and principal components analysis, to be readily inter-
pretable (Frank, 1990; Liu et al., 2008; Rosipal and Krämer,
2006). Reflectance, which is measured for CMCs because
they are not absorbing in the relevant wavenumber regions, is
related to the permittivity through Snell’s law and Fresnel’s
equations. Because the reflectance is nonlinear with respect
to the permittivity, the results obtained by applying PLS are
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 953
ME TECHNICAL PAPER w
x characterization of cmcs

not intuitively related to the quantities of interest. It is also
unreasonable to expect that sufficient a priori information and
a sufficient amount of data can be collected to use techniques
such as artificial neural networks (Durand et al., 2007).
The success of handheld FTIR devices to evaluate thermal
degradation in PMCs suggests the potential for FTIR to char-
acterize thermal degradation in CMCs. Previous work details
the development and application of data analysis techniques
designed for measurements that are nonlinear in interpretable
quantities (such as oxidation product volume percent) to
reflectance data obtained from CMCs (Banks et al., 2016;
Criner et al., 2014; Criner et al., 2015). These results offered
further evidence of the potential of a handheld FTIR device as
a surface chemical characterization technique to evaluate the
thermal degradation of CMCs.
The work here details some of the data analysis techniques
to properly model and analyze reflectance data; however, in-
depth FTIR spectroscopy and data analysis discussion are
beyond the scope of this paper. A thorough baseline of as-
received CMC specimens is presented here. The results and
techniques presented will be used in a future work to evaluate
any surface chemical changes that occur after the specimens
are exposed to high temperature environments, including
formation of oxidation products and other changes in
chemical structure.
Specimen Description
Five unique carbon fiber-based CMCs were fabricated by
impregnating fabric with a slurry of 30% volume powder
particulate and 70% volume liquid polycarbosilane. The deag-
glomerated powders used were hafnium diboride (HfB2),
silicon carbide (SiC), and silicon nitride (Si3N4). The CMC
specimens were manufactured by polymer infiltration and
pyrolysis processing techniques. This process started with
plain weave carbon fiber fabric. The fabric was coated with
pyrolytic carbon before layup. The liquid polycarbosilane was
used as the base for the slurry used during the layup. Unlike
regular organic polymers, the backbone of polycarbosilane
contains silicon and carbon atoms. This is in contrast to
organic polymer backbones, which are strictly carbon atoms.
High temperature pyrolysis, where the polymer is exposed to
a temperature of 1873 K (1600 °C) or greater in an inert
atmosphere, converts this polymer to a crystalline SiC. The
polymer was mixed with 30% volume of particulate powder
by stirring 15 g batches of deagglomerated powder into the
polymer, forming a slurry. The slurry was spread onto
5.08  5.08 cm squares of the pyrolytic carbon coated
fabric, and then heated to B-stage the thermoset, which is a
semicure of the polymer such that it holds the powder in
place. This allowed the specimens to be autoclaved into an
intermediate PMC laminate without the powder redistrib-
uting itself. After B-staging, eight plies of fabric were laid up
for autoclaving, where the intermediate laminate was then
pyrolized at 1873 K (1600 °C), converting the polycarbosi-
lane to crystalline SiC. Table 1 gives a ply-by-ply description
of each fabricated CMC specimen. The nomenclature for
each specimen indicates the order of the polymer slurry in
each layer, where H represents HfB2, S represents SiC, and
SN represents Si3N4. For example, the H-S-SN specimen has
HfB2 on the top, SiC in the middle, and Si3N4 on the bottom,
with composition changing from one ply to the next. A cross-
section of H-S-SN depicting the graded nature of the powder
loading throughout the CMC is presented in Figure 1.
Experimental Setup
Data at normal incidence was collected using a portable
handheld FTIR sensor with the diffuse reflectance attachment.
The wavenumber range of the sensor is 650 to 4000 cm –1
(2.5 to 15.0 μm in wavelength). The FTIR sensor uses a
michelson interferometer that collects double-sided interfero-
grams from both the specimen and a reference. Sixteen
interferograms were collected and averaged by the device for
both the reference and the specimen. The resultant spectra for
the specimen were then divided by the spectra obtained from

TABLE 1
Nomenclature and composition of graded Cf based CMCs*
Ply composition (30% volume total leading)
Specimen Particulate Ply number
1 2 3 4 5 6 7 8
H HfB2 100% 100% 100% 100% 100% 100% 100% 100%
H-S HfB2 100% 85% 70% 55% 45% 30% 15%
SiC 15% 30% 45% 55% 70% 85% 100%
S SiC 100% 100% 100% 100% 100% 100% 100% 100%
S-SN SiC 100% 85% 70% 55% 45% 30% 15%
Si3N4 15% 30% 45% 55% 70% 85% 100%
H-S-SN HfB2 100% 65% 35%
SiC 35% 65% 100% 85% 50% 15%
Si3N4 15% 50% 85% 100%
* Note that the volumes are of the 30% particulate, so 100% corresponds to 30% total volume.

954 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
 HfB2
65 HfB2/SiC
35 HfB2/SiC

SiC
85 Sic/Si3N4
50 SiC/Si3N4
15 SiC/Si3N4
Si3N4 1mm

Figure 1. Scanning electron microscopy image of carbon fiber

silicon carbide (SiC) (H-S-SN). The labels denote the composition
of the powder used in the matrix loading during the ceramic matrix
composite layup. Dotted lines follow the pattern of the tows in
the plain weave and denote the boundaries of the plies. White
contrasted regions contain hafnium diboride (HfB2), light gray
corresponds to SiC or silicon nitride (Si3N4), medium gray regions
are the carbon fiber tows, and the black areas are voids. The
darkest gray are voids that filled with epoxy when the specimen (a)
was mounted for scanning electron microscopy imaging. Note that
the content of HfB2 (white) decreases from the first to third ply.

the reference. The reference used here is a 100 μm diffuse

reflectance reference standard (pictured in Figure 2a). Boxcar
apodization was used along with the mertz phase correction
technique. Specular reflection measurements are effective for
very smooth specimens. Rather than polishing the specimens
to obtain sufficient smoothness for specular measurements,
the diffuse reflectance attachment was used. This attachment
focuses the incident light at normal incidence to the specimen
and collects the diffuse reflectance at many angles, using a
circular ellipsoidal mirror (Seelenbinder, 2011). The diameter
of the spot size for this attachment was 9.5 mm, resulting in a
total circular spot area of 70.88 mm2. The device was placed
on the charging dock and operated in benchtop mode
(Figure 2b). Three locations were sampled on each side for
every specimen.
The first point was taken from the middle, and the (b)
second and third points were taken near the ends. At each
point, 15 data sets were collected in quick succession without
re-referencing so that experimental model calibration could
be accomplished. A new reference was obtained for each Figure 2. Collection set up: (a) the 100 μm diffuse reflectance
reference cap; and (b) handheld fourier transform infrared operating
point. The total data acquisition time for each point was in benchtop mode.
approximately 5 min.

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 955
ME TECHNICAL PAPER w
x characterization of cmcs

1.5
Si3N4 dominant sides H-S-SN Back

Reflectance
1 S-SN Back

0.5

0
600 850 1100 1350 1600 1850 2100 2350 2600 2850 3100 3350 3600 3850 4100

1.5
SiC dominant sides F Front
Reflectance

1 S Back
S-SN Front
0.5 H-S Back

0
600 850 1100 1350 1600 1850 2100 2350 2600 2850 3100 3350 3600 3850 4100

1.5
HfB2 dominant sides H Front
Reflectance

1 H Back
H-S Front
0.5 H-S-SN Front

0
600 850 1100 1350 1600 1850 2100 2350 2600 2850 3100 3350 3600 3850 4100
(a) Wavenumber (cm–1)

1.5
Si3N4 dominant sides
Reflectance

H-S-SN Back
1 S-SN Back

0.5

0
640 690 740 790 840 890 940 990 1040

1.5 S Front
SiC dominant sides S Back
Reflectance

1 S-SN Front
H-S Back
0.5

0
640 690 740 790 840 890 940 990 1040

1.5 H Front
HfB2 dominant sides H Back
Reflectance

H-S Front
1
H-S-SN Front

0.5
(D)
0
640 690 740 790 840 890 940 990 1040
(b) Wavenumber (cm ) –1

Figure 3. Sample reflectance above background with respect to wavenumber: (a) all of the data collected from all of the specimens. Dominant
side refers to the particulate that is on the side of the specimen facing the sensor. Reflectance results are presented for three dominant side
particulate conditions: (top) silicon nitride (Si3N4); (middle) silicon carbide (SiC); and (bottom) hafnium diboride (HfB2); (b) Zoomed in plot of
the smaller wavenumbers for the same specimens: (top) Si3N4; (middle) SiC; and (bottom) HfB2.

956 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Fourier Transform Infrared Spectroscopy
Infrared spectroscopy detects vibrations that cause changes in
the dipole moment of the system (called infrared active
modes) while raman spectroscopy detects vibrations that
cause changes in the polarizability of the system (called raman
active modes). Both spectroscopy techniques provide some
information about the presence, amount, and structure of
some constituents through the observed vibrations in solids
(Solé et al., 2005).
The reflectance spectra from the specimens do not
directly relate to physically intuitive vibrational quantities
because of the nonlinearity of reflectance with respect to
optical constants. They may be compared with other
reflectance spectra in the literature. Because HfB2 is conduc-
tive, it will not produce pronounced peaks in the reflectance
spectra, though it is expected to subtly change the spectra
through the presence of free carriers and changes in the bulk
limit dielectric constants. The molar volume of Si3N4 is much
larger than the molar volume of HfB2 and SiC. In the 30%
volume mixture of Si3N4 with SiC (the maximum amount in
the specimens discussed), only approximately 10% of the
molecules are Si3N4. The data collected from the SN sides
of the specimens, from the S sides of the specimens, and
from the H sides of the specimens (labeled as in Table 1) over
the entire spectrum is depicted in Figure 3a (top), (middle),
and (bottom), respectively. The noisy region centered at
2350 cm–1 corresponds to the CO2 emission band. Although
nitrogen and oxygen are also in the air around the specimen
during collection, they do not have a permanent dipole
moment with which to interact with the infrared energy and
cannot absorb infrared radiation so neither of these species
appears in the signal (Chalmers, 2002).
There is a maxima at approximately 800 cm–1, which is
consistent in all of the datasets. This can be seen in Figure 3b
(top) for the SN sides, in Figure 3b (middle) for the S sides,
and in Figure 3b (bottom) for the H sides. The polycarbosi-
lane base for all of these specimens crystallizes into SiC. A
maximum in the reflectivity at approximately 800 cm–1
(called the reststrahlen region) corresponds to a resonant
wavenumber at approximately 800 cm–1, which is common in
SiC (Chibuye et al., 1994; Engelbrecht et al., 2012; Latte-
mann et al., 2003; Spitzer et al., 1959). The location of the
resonant wavenumber compared to the maxima and the
presence of a minimum (change in sign of the slope) after the
reststrahlen region are dependent on the dampening, the limit
dielectric constant, the resonant wavenumber of the trans-
verse and lateral optical modes, the presence of a damage
layer, and surface roughness, among other factors (Engel-
brecht and Van Rooyen, 2011; Engelbrecht et al., 2012;
Engelbrecht et al., 2014; Engelbrecht et al., 2015). The
magnitude of reflectance varies from one location to another
on the same specimen and from specimen to specimen. This
has been observed previously in CMCs (Banks et al., 2016;
Criner et al., 2015). It is conceivable that the within-specimen
amplitude fluctuations are caused by differing properties
(such as composition and surface roughness) in the
sampling regions, especially when the heterogeneity in the
material in Figure 1 and the device spot size of 70.88 mm2
are considered.
Models and Data Analysis
For spectra that are closely related to the optical constants,
such as absorbance spectra, the infrared active vibrational
modes present can often be inferred directly from the spectra.
Unlike absorbance spectra, reflectance spectra are nonlinear
in their optical parameters, meaning the understanding of the
underlying structure of the material cannot be obtained
directly or through linear data analysis techniques. The
kramers-kronig transform can be used to transform the
reflectance spectrum into the related optical constants,
including absorbance, but these constants do not provide
detailed band parameters nor resolve contributions from
vibrational modes that have overlapping bands. Detailed band
parameters are needed to track subtle changes and features
(Efimov, 1996). Dispersion analysis, another technique for
analyzing vibrational spectra, obtains information about the
resonant wavenumber and other parameters underlying the
optical constants using nonlinear regression analysis.
Nonlinear regression analysis obtains estimates of quantities
of interest (corresponding to the vibrational modes) through
curve fitting to the observed data. The obtained parameters
are the parameters that cause the predicted observed physical
quantities (model of reflectance here) to be sufficiently close
to the experimentally observed physical quantities
(reflectance data). Obtaining these underlying changes will
allow subtle changes in the chemical structure to be observed,
which can correspond to changes in mechanical properties
(Gouadec and Colomban, 2007). This framework can also
lead to actionable information about the material, such as
confidence regarding the presence of oxidation products
(Banks et al., 2015).
Dispersion analysis combined with the assumption that
for glasses the underlying resonance wavenumbers follow
normal distributions is discussed in prior work (Efimov,
1996). Additional works discuss that the form of the analytical
model of the optical constants ~  , which is discussed as
follows), including the assumed distribution of the resonant
wavenumbers, causes this approach to be less general than the
kramers-kronig approach (Efimov, 1996; Efimov, 1999).
The approach presented here takes the dispersion analysis
approach of past work to a further level of generality, in that it
only assumes there is a distribution of resonant wavenumbers
in the lorentz model, and relies on convergence results, which
state that the approximation will get arbitrarily close to the
true probability distribution (Banks et al., 2014; Banks et al.,
2015; Banks et al., 2016; Efimov, 1996).
The experimental data must be modeled in order to
carry out nonlinear regression analysis. The model for the
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 957
ME TECHNICAL PAPER w
x characterization of cmcs

TABLE 2
Notation
Symbol Parameter Value
θ Incident angle of beam 0°
nA Refractive index of surroundings 1
 Wave number Dependent variable, varies from 750 – 1500 cm–1
~
 Dielectric constant or permittivity (complex number) Varies with wave number
ñ Refractive index of material (complex number) Depends on 
~
r Perpendicular polarization reflectivity Depends on ñ, Snell’s law/Fresnel’s equations
r|| Parallel polarization reflectivity Depends on ñ, Snell’s law/Fresnel’s equations
R Total reflectance Average of r and r||

experimental data entails a mathematical model of the physics The classical lorentz oscillator model is given by:
that would be observed without any measurement error,
and a statistical model that describes the random errors Δε ω02
that are associated with the measurement process. The (3) ε (ω) = ε∞ +
dielectric constant, ~  (), is a function of the wavenumber (ω 2
0 )
− ω 2 − iγω
(the independent variable)  cm–1 and the resonant
wavenumber 0 cm–1 , among other parameters. The mathe- where
matical model relates quantities of interest (which are param- 0 is the resonant wavenumber,
eters in the model for the dielectric constant) to the observed  is the broadening,
data (the reflectance at each wavenumber). Throughout this  is the contribution to the limit dielectric constant
description, the notation in Table 2 is used. (Banks et al., 2016; Efimov, 1996).
The model of the reflectance in terms of the dielectric
constant ~ is obtained by using standard optical results that This model is most appropriate for crystalline systems
relate the dielectric constant to the refractive index, then (MacDonald et al., 2000). Recent results demonstrated appli-
relate the refractive index to the two polarizations of reflec- cation of the prokhorov metric result to appromixate an
tivity, and finally relates the polarizations of reflectivity to the underlying probability distribution in FTIR of an amorphous
reflectance (Catenacci, 2016; Griffiths, 1999). The dielectric CMC as-received, after 10 h of heating at 1473 K (1200 °C),
constant squared is the refractive index n~() = ~  2(). The and after 100 h of heating at 1473 K (1200 °C) (Banks et al.,
perpendicular and parallel polarizations of reflectivity are 2016; Catenacci, 2016).
derived using Snell’s Law and the fresnel equation. The resonant wavenumber 0 is modeled by the random
1
variable x that follows the probability distribution, P. The

r⊥ (ω) =
(
nA cos θ I − n 2 ⎡⎣ω⎤⎦ − nA2 sin 2 θI ) 2 model for the dielectric function is then given by Equation 4.
1
nA cos θ I + (n 2⎡
⎣ω⎤⎦ − nA sin
2 2
θ )
I
2
x2
(4) ε (ω) = ε∞ + Δε ∫ dG ( x)
x 2 − ω 2 − iγω
⎛ n2 ⎞1
(1) nA ⎜⎜1 − 2 A sin 2 θI ⎟⎟ 2 − n (ω) cos θ I
⎝ n ⎣⎡ω⎤⎦ ⎠
r (ω) = The success of previous work that detected different levels
⎛ n 2 ⎞1
nA ⎜⎜1 − 2 A sin 2 θ I ⎟⎟ 2 + n (ω) cos θI
of thermal exposure in CMCs using the height and width of
⎝ n ⎡⎣ω⎦⎤ ⎠ the inferred piecewise linear basis functions motivated the
choice of piecewise linear basis functions (Banks et al., 2016).
The underlying probability density function, dP(x), may be
The reflectance is then given by the average: approximated by a sum of piecewise linear basis functions
ℓj(x):
(2) R ( ω) =
1
2 ( 2
r⊥ ⎡⎣ω⎤⎦ + r ⎡⎣ω⎦⎤
2
) M
(5) dG ( x) ≈ ∑ α j  j ( x)
where j =1
|| is the modulus of the complex number (z = x + iy, then
|z| = (x2 + y2).

958 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
 with nodes xj, weights lj(x) given by: The parameter is given by the equation:
⎧ x−x ⎡ k k k ⎤
⎪ j −1 q k = ⎢c k , ∈ k∞ , α 1 , α 2 ,@ , α M ⎥
x j −1 < x < x j ⎣ ⎦
⎪⎪ x j −1 − x j
 j ( x) = ⎨ The model (Equation 9) for the parameter values esti-
(6)
(
⎪ − x − x j +1 ) x j < x < x j +1 mated is not sensitive to the dampening parameter, , in
⎪
⎪⎩ x j − x j −1 Equation 8. Therefore, this parameter was fixed at 2.5,
which is based on initial data analyses and literature
and weights j such that: (Engelbrecht et al., 2012). The data set indices (k) represent
each individual data set of reflectance values at each of the
M 402 wavenumbers (i cm–1) between 750 and 1500 cm–1.
∑ α j ∫ x j +1  j ( x) dx = 1
x
(7) j −1 The vector qk represents the unknown variables that are esti-
j =1
mated. The values –j = j are used in the curve fitting
for j {1, 2, 3, …, M}. Equation 7 is required to ensure so that the constraint (Equation 7) is enforced after the
that the approximation is a probability distribution. The prob- estimation using equations found in an outside work
ability density’s support is (x0, xM+1), so dP(x) = 0 for x ≤ x0 (Catenacci, 2016).
and x  xM+1. Under the prokhorov metric, the dielectric The mathematical model that contains assumptions about
constant (and subsequent calculation of the reflectance) using the physics of the material (Equation 9) must be combined
the approximation (Equation 5), converges to the desired with a statistical model that contains assumptions about the
quantities as M (the number of nodes) approaches . The random error induced by various measurement factors
index j indicates the wavenumber that are being evaluated. including air variations, electronic noise, and their interactions
The dielectric constant is given by Equation 8. with the material. The data analyses used correspond to both
maximum likelihood estimation (MLE) and least squares, but
M x 2 j (x) with different assumptions (Banks et al., 2009). The proce-
ε (ω) = ε∞ + Δε ∑ α j ∫ x j +1
x
(8) dx
j =1
j −1
x 2 − ω 2 − iγω dure for each in this example is the same, given certain
assumptions. The MLE approach depends on the assumption
The integrals were solved using a symbolic computation of a specific distribution while least squares relies upon large
program, and the resultant function was evaluated using a sample size. When normally distributed errors are assumed in
numerical computing program. MLE, the algorithms and estimation results are equivalent.
This model does not include the effect of free carriers Other parameter estimation routines and approaches may be
through a drude term (equivalent to lorentz with 0 = 0), taken, such as bayesian estimation and bootstrapping. An in-
surface roughness, nor a damage layer. The reflectance model depth discussion comparing the results, advantages, and
(Equation 2) does not exceed 1 in value (representing 100% disadvantages of using these approaches in analysis of
reflection), but values in the experimental data do exceed 1. reflectance data using the prokhorov metric are discussed
There are two differences between the physical assumptions elsewhere but are beyond the scope of this work (Catenacci,
in the derivations to obtain the reflectance model (Equation 2016). The independently, identically distributed (IID) non-
2) and the experimental setup that may cause this discrep- parametric model is given by:
ancy. The reflectance model (Equation 2) assumes absolute
measurements of specular reflectance, while the experiments (10) Yik = f (ω i ; qk ) + σ k ε ik
obtained relative measurements of diffuse reflectance. A
multiplicative constant, ck, is added to account for these devia- where
tions of the experimental measurements from the physical f(i; qk) is given by Equation 9,
model assumptions. The mathematical model that is used in 2k is the measurement error variance.
the analyses is given by:
The assumption about the random error variable ik
(9) f ( ω i ; q k ) = c k R (ω i ; q k ) differs between MLE and least squares approaches. In both,
the random variable ik is assumed to be IID with mean 0
where and variance 1. This is the extent of the assumptions in the
i is the wavenumber, ordinary least squares approach. For the MLE approach, the
ck is the multiplicative constant previously discussed, error ik is further assumed to follow a standard normal distri-
R(i; qk) is the reflectance as modeled in Equation 2 with bution (ik ~ N[0, 12]).
dielectric constant given by Equation 8 evaluated at As is depicted in Figure 3a, the amount of variation in the
parameter qk. reflectance data is different across the spectrum. This is even
clearer in Figure 3b; there is more spread in the data from

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 959
ME TECHNICAL PAPER w
x characterization of cmcs
wavenumbers between 650 and 740 cm–1, while there appears
to be less from 990 to 1040 cm–1. This indicates that for the
reflectance data the random error is not IID. Random error
that is not IID can be properly modeled by introducing an
error function g(i; qk) in the statistical model. The data
model is then given by:
(11) Yik = f (ω i ; qk ) + σ k g (ω i ; θ k ) εik
where
k are the parameters for the variance function,
the assumptions regarding ik are as in the ordinary least
squares formulation (Banks et al., 2009; Davidian and
Giltinan, 1995).
In order to analyze data with the model in Equation 11,
iteratively reweighted least squares is used, which adds signifi-
cant computational time to the data analysis (Banks et al.,
2009). If data can be collected that is obtained with no varia-
tion in the parameters characterizing g and f, the variance
function may be estimated pointwise. This can be seen by
considering Yik1 and Yik2, where (qk1, k1) = (qk2, k2).
(12) 1 2 1 ( 1 1 )(
Yik − Yik = σ k g ω i ; θ k εik − εik
2 )
The (ik1 – ik2) are IID with mean 0 and variance 2. The
MLE approach further assumes that this error is normally
distributed.
A pointwise estimate of the g(i; k) may be obtained at
each wavenumber, i, by considering the differences between
data sets in which the parameters have not changed. Fifteen
data sets were collected in succession without re-referencing
or moving the specimen (k{1, 2, …, 15}). The pointwise
estimates of g(i; k), denoted g^ik, are obtained by analyzing
these differences. The differences follow a distribution with
mean 0 and variance k2 g2(i; k). The estimation of the
variance at each wavenumber, i, is then used to obtain the
weights:
1 J
( )
2
(13) Wik = σˆ 2k gˆ ik2 = ∑ ΔYj
J − 1 j =1
where
Yj are the differences between data sets where the param-
eters have not changed (Degroot and Schervish, 2002).
These Wik are then used as weights in the parameter esti-
mation routines. The appropriateness of this assumption for
the data considered here is discussed in the Results section.
The model given in Equation 11 with weights Wik has cost
functional and parameter estimates:
(Yik − f ⎡⎣ωi ; qk ⎦⎤)
2
N
(14) J (q k ) =∑
i=1 Wik
960 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
(15) qˆ k = arg min qk ∈H J (q k )
where
N = 402 is the total number of data points.
The cost function (Equation 14) was minimized using a
trust region reflective algorithm.
When carrying out the nonlinear minimization (Equation 15),
the results are very sensitive to initial guess of qk. Rather than
beginning with a fine mesh of nodes (xj) for the probability
distribution functions (PDFs) in Equations 5, 6 and 8, succes-
sive analyses were carried out with increasing resolution of
nodes xj (by increasing the value of M). Initial analyses were
carried out to select the nodes used in the automated data
analysis of all the data sets. The wavenumber range for the
PDFs (from 700 to 1550 cm–1) was taken to be larger than
the wavenumber range for the data (from 750 to 1500 cm–1).
The nodes were evenly spaced beginning with two intervals
between 700 and 1550 (one node at 1125) and doubling
the number of intervals until there were 256 intervals
(255 nodes). The resultant (PDFs) were inspected for areas
that were zero (nodes above 1500), subsections that needed
more resolution (750 to 975 cm–1) and subsections that
did not need high resolution (700 to 750 cm–1 and 975 to
1500 cm–1). It is not surprising that the interval 700 to
750 cm–1 does not require fine resolution because the
reflectance data analyzed begin at 750 cm–1. A consistent
Obtain 15 data sets from single point
Form weights (Equation 13)
Form cost function (Equation 4)
Use numerical procedure to find PDF (αj,
which is the height of the PDF at node xj),
ε∞, Δε, and c that minimizes (Equation 14) Increase resolution
the cost function (that yields model of nodes xj
reflectance values that are closest to the
experimental reflectance data when weighted)
Estimate resonant wavenumber PDF, estimate
limit dielectric constant and estimate
contribution to the static dielectric constant
(Equation 15)
Interpret infrared active molecules present
including species, amount, and structure of
constituents
Figure 4. Flow chart of the data analysis procedure. PDF = probability
distribution function.
dip in estimated PDF height in initial analyses for the nodes
near 975 cm–1 motivated using 975 cm–1 as an endpoint.
Some mathematical model discrepancy is probably caused by
this choice of nodes. Future work will include improving the
selection of the nodes in the PDFs.
For the automated analysis of all of the data sets, the
nodes at 700, 750, 975, and 1500 were fixed. Four intervals
were placed between 700 and 750 for all analyses at 700,
712.5, 725, 737.5, and 750. The number of intervals between
750 and 975 increased in each iteration, starting at 2 intervals
and doubling each time until 32 intervals were reached (2, 4,
8, 16, 32). The number of intervals between 975 and 1500
started at 1 and also doubled with each iteration, up to
between the side 2 PDF, which attains a value of 0.04 at
712 cm–1 in Figure 6c. The model is not very sensitive to the
nodes below the first data point at 750 cm–1. Improving the
node locations will ameliorate this problem. This peak causes
the height of the PDF to be small at 806 cm–1 because the
PDFs must have area one. The asymmetric peak at 806 cm–1
in the PDFs corresponds to the transverse optical mode in
6

Simulated IID normal data

16 intervals (1, 2, 4, 8, 16). Two initial guesses were used in
4
each successive interval. One initial guess was the previous
estimated values with the new weights for j interpolated 2
from weights with the previous nodes. The other initial guess
0
used the previous estimate but with all weights (j’s) set to
zero. Whichever estimate obtained the lowest cost function –2
(J[qk]) was carried forward (Equation 14). The final number
–4
of estimated weights was 51 (4 + 16 + 32 = 52 intervals have
51 interior endpoints corresponding to the nodes). –6
800 900 1000 1100 1200 1300 1400 1500
A summary of the data analysis procedure is given in a (a) Wavenumbers (cm–1)
flow chart in Figure 4 with references to the relevant equa-
tions for each step. 3
Weighted differences

Results 2

Based on the assumptions in Equation 13, the weighted differ- 1

ence between data sets should appear IID. An example of
0
simulated IID data from a normal distribution is depicted for
comparison in Figure 5a. A representative experimental –1
weighted difference (the difference between data sets that are
–2
assumed to have the equal parameters in Equation 13 divided
by the square root of the estimated weights) is depicted in –3
Figure 5b. If the assumptions that the data sets are from the 800 900 1000 1100 1200 1300 1400 1500
(b) Wavenumbers (cm–1)
same parameter set values and the pointwise weight estimates
sufficiently approximate the variance function g2(i; k) are
correct, the data plotted in Figure 5b should be IID. The 10
values in Figure 5b do appear to be IID. In Figure 5c,
Weighted residuals

weighted residuals from a representative model fit of the data 5

(point 3 of data set 14 from the SiC side of the SiC-Si3N4
specimen) are depicted. In this figure, clusters of extreme 0
values at 800 and 990 cm–1 indicate a mathematical model
discrepancy, which is likely caused by the selection of nodes.
–5
At 800 and 990 cm–1, the distance between each node is 7
and 33 cm–1, respectively. There are likely features present
that would be better captured by the model fit with increased –10
800 900 1000 1100 1200 1300 1400 1500
node sampling in these regions. (c) Wavenumbers (cm–1)
The model fits and data from side 1 and side 2 of the S
(SiC) specimen are depicted in Figure 6a and 6b, respectively.
In Figure 6c, the estimated PDFs for both sides of the S
Figure 5. Data sets: (a) simulated independently, identically
specimen are plotted. There is not very much variation in the distributed data from a normal distribution; (b) example experimental
side 2 data for wavenumbers less than 790 cm–1 in Figure 6b, weighted differences; and (c) weighted residuals from one of the
which does not seem to correspond to the large difference analyses.

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 961
ME TECHNICAL PAPER w
x characterization of cmcs

SiC. In Figure 6d, there is a small divot at 968 cm–1, which
corresponds to the lateral optical mode in SiC (Pitman et al.,
2008). Side 1 and side 2 of the H (HfB2) specimen are
depicted in Figure 7a and 7b, respectively.
These data sets all have a minimum of approximately
980 cm–1. These minima correspond to a minimum at
975 cm–1 in the PDFs in Figure 7c and more clearly in
Figure 7d. This divot at 975 cm–1 is more pronounced in the
sides of the specimens that have HfB2, which may indicate that
there is a change in the structure of the SiC lateral optical modes
on the HfB2 sides. The PDFs, like those obtained from the S
specimens in Figure 6c, have an asymmetrical peak at 806 cm–1
that corresponds to the transverse optical mode of SiC.
The results from the other specimens were similar to
those of the H and S specimen. All estimated PDFs had an
asymmetrical peak at 806 cm–1 corresponding to the SiC
transverse optical mode. The sides that had HfB2 (those in
Figure 7c, and the HfB2 side of the H-S-SN specimen) had
pronounced divots at 975 cm–1, which corresponded to
minima in the reflectance spectra at 980 cm–1 and the SiC
lateral optical mode. The sides that did not have HfB2 had
subtle divots at approximately 970 cm–1.
Conclusions and Future Work
Mathematical and statistical models for FTIR data obtained
from CMCs were presented and used to evaluate CMCs that
are mostly crystalline SiC. The statistical model was efficiently
incorporated by carrying out an experimental model calibra-
tion. The validity of implicit assumptions in the approach was
assessed by examining differences for the accuracy of these

(a) (b)

(c) (d)

Figure 6. Model data: (a) All data fits and data used for front of S specimen (silicon carbide [SiC]); (b) all data fits and data used for back of S
specimen (SiC); (c) estimated probability density function of the resonant wavenumbers for all data sets from the S specimen (SiC on both
sides); and (d) a subset of the probability distribution functions in Figure 6c to highlight the divot at approximately 970 cm–1.

962 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
 (a) (b)

(c) (d)

Figure 7. Model data: (a) all data fits and data used for front of H specimen (hafnium diboride [HfB2]); (b) all data fits and data used for back of
H specimen (HfB2); (c) estimated probability density function of the resonant wavenumbers for all data sets from the H specimen (HfB2 on both
sides); and (d) a subset of the probability distribution functions in Figure 7c to highlight the divot at 975 cm–1.

assumptions. Discrepancies in the mathematical model, which
occur near wavenumbers at approximately 800 and 990 cm–1,
were identified through a similar procedure.
Estimated PDFs demonstrate an asymmetrical SiC peak at
806 cm–1 corresponding to the transverse optical mode in all
of the materials that were analyzed. The estimated PDFs had
a divot at approximately 970 cm–1 corresponding to the
lateral optical mode, which was more pronounced in materials
that contained HfB2.
Further work must be carried out to model the relative
diffuse reflectance in a more sophisticated manner, which will
account for the reflectance values that exceed 1 in a more
physically accurate way (Kortüm, 1969). Other modeling
improvements will include incorporating damage layer
models, surface roughness models, and improving the
selection of nodes in the PDFs. Ill-identifiability concerns will
need to be addressed because many roughness effects have
the same form as heterogeneous damage layer effects.
The results presented here and associated techniques will
be used to identify surface chemical changes in the materials
after they have been exposed to high temperature environ-
ments. Oxidation products will be detected through the
addition of a peak in the estimated PDFs. With the data
analysis framework, hypothesis tests will be developed to
associate confidence with the presence of the oxidation
products. For instance, the formation of silica (SiO2) will
cause the addition of at least one peak at approximately
1080 cm–1 (Criner et al., 2015). With the data analysis frame-
work, hypothesis tests will be developed to associate confidence
with the presence of the oxidation products. The improvements

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 963
ME TECHNICAL PAPER w
x characterization of cmcs
in modeling and data analysis will enable inference of subtle
changes in structure, including changes in organization in
lattice vibrations and changes in mechanical properties.
ACKNOWLEDGMENTS
The authors would like to thank Garth Wilks, Josiah Dierken, Jared Cate-
nacci, H.T. Banks, Matt Cherry, Lenell Kern, Eric Lindgren, Sean Coghlan,
Keith Klug, and Mike McCarty for their discussions, insights, and support.
Some of this work was carried out under AFOSR project number
14RX05COR.
REFERENCES
Banks, H.T., J. Catenacci, and A. Criner, “Quantifying the Degradation in
Thermally Treated Ceramic Matrix Composites,” International Journal of
Applied Electromagnetics and Mechanics, Vol. 52, Nos. 1–2, 2016, pp. 3–24.
Banks, H.T., J. Catenacci, and S. Hu, “Asymptotic Properties of Probability
Measure Estimators in a Nonparametric Model,” SIAM/ASA Journal on
Uncertainty Quantification, Vol. 3, No. 1, 2015, pp. 417–433.
Banks, H.T., M. Davidian, J.R. Samuels, and K.L. Sutton, “An Inverse
Problem Statistical Methodology Summary,” Mathematical and Statistical
Estimation Approaches in Epidemiology, Springer, Dordrecht, Netherlands,
2009, pp. 249–302.
Banks, H.T., S. Hu, and W.C. Thompson, Modeling and Inverse Problems in
the Presence of Uncertainty, CRC Press, Boca Raton, FL, 2014.
Catenacci, J.W., “Quantifying Degradation in Ceramic Matrix Composites
Through Electromagnetic Interrogation and the Related Estimation
Techniques,” Ph.D. thesis, North Carolina State University, 2016.
Chalmers, J.M., “Mid-infrared Spectroscopy: Anomalies, Artifacts And
Common Errors,” Handbook of Vibrational Spectroscopy, Vol. 3, 2002,
pp. 2327–2347, John Wiley and Sons, Ltd., Hoboken, New Jersey.
Chibuye, T., C.G. Ribbing, and E. Wäckelgård, “Reststrahlen Band Studies
of Polycrystalline Beryllium Oxide,” Applied Optics, Vol. 33, No. 25, 1994,
pp. 5975–5981.
Criner, A., A. Cherry, A. Cooney, and T. Katter, “Characterization of
Degradation Using Reflectance Spectroscopy,” AIP Conference Proceedings,
Vol. 1581, No. 1, 2014, pp. 1594–1602.
Criner, A., A. Cherry, A. Cooney, T. Katter, H. Banks, S. Hu, and J. Cate-
nacci, “Identification of Thermal Degradation using Probabilistic Models
in Reflectance Spectroscopy,”AIP Conference Proceedings, Vol. 1650, No. 1,
2015, pp. 1898–1906.
Davidian, M., and D. Giltinan, Nonlinear Models for Repeated Measurement
Data, Chapman & Hall, London, England, 1995.
Degroot, M.H., and M.J. Schervish, Probability and Statistics, Addison
Wesley, Boston, Massachusetts, 2002.
Durand, A., O. Devos, C. Ruckebusch, and J. Huvenne, “Genetic Algorithm
Optimisation Combined with Partial Least Squares Regression and Mutual
Information Variable Selection Procedures in Near-infrared Quantitative
Analysis of Cotton-viscose Textiles,” Analytica Chimica Acta,, Vol. 595, No.
1, 2007, pp. 72–79.
Efimov, A., “Vibrational Spectra, Related Properties, and Structure of Inor-
ganic Glass,” Journal of Non-crystalline Solids, Vol. 253, 1999, pp. 1–24.
Efimov, A., “Quantitative IR Spectroscopy: Applications to Studying Glass
Structure and Properties,” Journal of Non-crystalline Solids, Vol. 203, 1996,
pp. 1–11.
Eibl, S., “Comparison of Surface and Bulk Analytical Techniques for the
Distinct Quantification of a Moderate Thermal Pre-load on a Carbon Fibre
Reinforced Plastic Material,” Polymer Degradation and Stability, Vol. 135,
2017, pp. 31–42.
Engelbrecht, J., E. Janzén, A. Henry, and I. Van Rooyen, “Impact of Dielec-
tric Parameters on the Reflectivity of 3C-Sic Wafers with a Rough Surface
Morphology in the Reststrahlen Region,” Physica B: Condensed Matter,
Vol. 439, 2014, pp. 115–118.
964 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
Engelbrecht, J., I. Van Rooyen, A. Henry, E. Janzén, and B. Sephton,
“Notes on the Plasma Resonance Peak Employed to Determine Doping in
SiC,” Infrared Physics & Technology, Vol. 72, 2015, pp. 95–100.
Engelbrecht, J., I. Van Rooyen, A. Henry, E. Janzén, and E.Olivier, “The
Origin of a Peak in the Reststrahlen Region of SiC,” Physica B: Condensed
Matter, Vol. 407, No. 10, 2012, pp. 1525–1528.
Engelbrecht, J., and I. Van Rooyen, “The Influence of Various Dielectric
Parameters on the Reststrahlen Region of SiC,” Physica B: Condensed
Matter, Vol. 406, No. 3, 2011, pp. 593–596.
Frank, I.E., “A Nonlinear PLS Model,” Chemometrics and Intelligent
Laboratory Systems, Vol. 8, No. 2, 1990, pp. 109–119.
Gouadec, G,. and P. Colomban, “Raman Spectroscopy of Nanomaterials:
How Spectra Relate to Disorder, Particle Size, and Mechanical Properties,”
Progress in Crystal Growth and Characterization of Materials, Vol. 53, No. 1,
2007, pp. 1–56.
Griffths, D., Introduction to Electrodynamics, third edition, Prentice-Hall,
Inc., Upper Saddle River, New Jersey, 1999.
Kortüm, G., Reflectance Spectroscopy: Principles, Methods, Applications,
Springer-Verlag, Heidelberg, Germany, 1969.
Lattemann, M., E. Nold, S. Ulrich, H. Leiste, and H. Holleck, “Investigation
and Characterisation of Silicon Nitride and Silicon Carbide Thin Flms,”
Surface and Coatings Technology, 2003, pp. 174–175.
Levy, D., “Ceramic Matrix Composites Take flight in LEAP Jet Engine,”
<https://phys.org/ news/2017-01-ceramic-matrix-composites-flight-
jet.html>, 4 January 2017.
Liu, F., Y. He,and L. Wang, “Determination of Effective Wavelengths for
Discrimination of Fruit Vinegars using Near Infrared Spectroscopy and
Multivariate Analysis,” Analytica Chimica Acta, Vol. 615, No. 1, 2008,
pp. 10–17.
MacDonald, S.A., C.R. Schardt, D.J. Masiello, and J.H. Simmons, “Disper-
sion Analysis of FTIR Reflection Measurements in Silicate Glasses,”
Journal of Non-crystalline Solids, Vol. 275, No. 1, 2000, pp. 72–82.
Ohnabe, H., S. Masaki, M. Onozuka, K. Miyahara, and T. Sasa, “Potential
Application of Ceramic Matrix Composites to Aero-engine Components,”
Composites Part A: Applied Science and Manufacturing, Vol. 30, No. 4, 1999,
pp. 489–496.
Pitman, K., A. Hofmeister, A. Corman, and A. Speck, “Optical Properties of
Silicon Carbide for Astrophysical Applications I: New Laboratory Infrared
Reflectance Spectra and Optical Constants,” Astronomy & Astrophysics,
Vol. 483, No. 2, 2008, pp. 661–672.
Rein, A., and J. Seelenbinder, “Composite Thermal Damage – Correlation
of Short Beam Shear Data with FTIR Spectroscopy: Portable, Nondestruc-
tive Analysis,” application note, Agilent Technologies, 2015.
Rosipal, R., and N. Krämer, “Overview and Recent Advances in Partial
Least Squares,” Subspace, Latent Structure, and Feature Selection, Springer,
New York, NY, 2006, pp. 34–51.
Seelenbinder, J., “Handheld FTIR Spectroscopy Applications using the
Agilent Exoscan 4100 FTIR with Diffuse Sample Interface,” application
note, Agilent Technologies, 2011.
Solé, J., L. Bausa, and D. Jaque, An Introduction to the Optical Spectroscopy
of Inorganic Solids, John Wiley & Sons, Hoboken, NJ, 2005.
Spitzer, W., D. Kleinman, and D. Walsh, “Infrared Properties of Hexagonal
Silicon Carbide,” Physical Review, Vol. 113, No. 1, 1959, p. 127.
Tressler, R.E., “Recent Developments in fibers and Interphases for High
Temperature Ceramic Matrix Composites,” Composites Part A: Applied
Science and Manufacturing, Vol. 30, No. 4, 1999, pp. 429–437.
Wang, Z., C. Laird, Z. Hashin, B. Rosen, and C.-F. Yen, “Mechanical
Behaviour of a Cross-weave Ceramic Matrix Composite,” Journal of
Materials Science, Vol. 26, No. 17, 1991, pp. 4751–4758.
SERVICE directory

EMPLOYMENT SERVICE
Looking for a job or an employee?
E-mail Associate Editor Haley Cowans at hcowans@asnt.org for information on
listing your qualifications in Materials Evaluation. This is a free service for ASNT
members. Our Employment Service listings can also be found in the Jobs section of
ASNT’s website at www.asnt.org.

POSITIONS WANTED

ASNT Level III (VT, UT, RT, MT and PT) with 30+ years diversified experience. Looking
for a rewarding, challenging position. Open to relocation. Reply to Dept. 06-01-17.

Ultrasound physics certified with ARDMS. Looking for position as NDT assistant in
order to train and become proficient in the industry. Brings diversified skill set and
background, including experience as a police officer, an IT analyst, and manager of
worldwide news company, and has applied for a TWIC. Open to relocation. Reply to
Dept. 06-02-17.

ASNT Level III (basic, MT), CSWIP 3.1, Bgas Grade 2, ASNT Level II (RT, UT, MT, PT).
Fugro USA Land, Inc.
Has 7+ years of experience in oil and gas construction (both onshore and offshore),
10831 Train Court
shop floor works, refinery, and petrochemical fields. Available immediately. Reply to Houston, TX 77041
Dept. 06-03-17.
713.937.0029 www.fugro.com
Physics Engineer having ASNT RT Level II certification with 3 years of international
work experience as QA/QC engineer in an oil pipeline project and as documents
controller on procedures in a power plants project is seeking a position as
Radiographer. Reply to Dept. 07-01-17.

ASNT NDT Level III (RT, PT, MT) and AWS/CWI with forty years field experience in
nuclear, oil and gas, aerospace, castings, structural steel. Positions have included
but are not limited to: extensive radiographic interpretation, QA/QC, NDT, weld
inspection, training, nuclear document review, procure review/writing, auditing.
Experience includes fourteen years experience in Saudi Arabia, Dubai, Egypt. Reply
to Dept. 07-02-17
To reply to Employment Service ads, contact Haley Cowans at hcowans@asnt.org.

966 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 967
SERVICE directory

968 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
NDTMarketplace
Your next opportunity to feature
your new product in ASNT’s
semi-annual product guide,
NDTMarketplace, will be
December 2017. Bring your
products to the attention of the
key decision makers of
the NDT industry with display
advertising opportunities.
Contact Advertising Supervisor
Diane Oen for more information
on getting your products the
attention they deserve!

E-mail: doen@asnt.org

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 969
SERVICE directory

970 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
 ASSOCIATED X-RAY
X-RAY Sales & Service
New & Reconditioned Equipment
• Portable & Stationary
• Real Time & Digital Imaging Systems
• Sale – Rental – Lease
In-House or On-Site Service For:
• Andrex • Philips • Sperry/Stavely
• XMAS • Balteau • Tubes/H.V. Cables
• TFI • Scanray • Seifert/Eresco
• Gemini • ICM • Magnaflux
• Rigaku • Gulmay • Pantak
• Faxitron • Comet • Astrophysics
246 Dodge Ave, E. Haven, CT 06512
Ph: (203) 466-2446, Email: axrcorp@aol.com

J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 971
SERVICE directory
ASSOCIATED X-RAY
X-RAY • NDT
• Real Time & Digital Imaging Systems
• Portable & Stationary Equipment
• Cabinet X-Ray Systems
• Micro focus equipment
• Radiation safe rooms and cabinets
• Safety interlock switches
• Room alarms/portable warning units
• Automatic film processors
• X-Ray film and chemicals
• High intensity illuminators
• Dark room supplies
• US & Export sales and service
246 Dodge Ave, East Haven, CT 06512
Ph: (203) 466-2446, Email: axrcorp@aol.com

972 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 973
SERVICE directory

974 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7
J U LY 2 0 1 7 • M AT E R I A L S E VA L U AT I O N 975
 AD index
July 2017
coming Bolded listings in the ad index below indicate platinum and gold advertisers.

attractıons Advanced OEM Solutions www.aos-ndt.com 892

AINDT www.trainingndt.com 834

CoreStar International www.corestar-corp.com 840

l In August, Materials
Evaluation will feature papers Curtis Industries www.curtis-test.com 895
on Radiographic Testing with
a Product Spotlight on Laser FEI www.fei.com 879

Methods. Make sure your

Fuji www.fujindt.com IFC
company is in the issue;
contact the advertising Hellier www.hellierndt.com 842
supervisor today!
Logos Imaging www.logosimaging.com 882
l The September issue will
MFE Enterprises www.mfescan.com 833
explore NDT Education and
Training, critical topics that are NDT Boot Camp www.ndtbootcamp.com 900
important for NDT professionals
in all areas. This issue will also NDT Classroom www.ndtclassroom.com 859
include the Education Forum, a
North Star Imaging www.4nsi.com 885
special advertising section.
Contact the advertising super- NOVO-DR www.novo-dr.com/ 837
visor to be part of the issue!
Olympus www.olympus-ims.com BC
Looking for a low cost, highly
visible advertising program? Physical Acoustics www.physicalacoustics.com 839, 885

Consider placing a directory ad Quality Testing www.qualitytesting.com IBC

in Materials Evaluation. Build
product or service recognition RF Systems www.rfsysstemlab.us 886
with this popular program.
TecScan www.tecscan.ca 860
Use the Coming Attractions
TesTex www.testex-ndt.com 889
information to help plan your
advertising schedule. Contact University of Ultrasonics www.universityofultrasonics.com 849, 885
the Advertising Supervisor at
Vidisco www.vidisco.com 848
(800) 222-2768 X209 or by
e-mail at doen@asnt.org. Virtual Media Integration www.starrview.com 844

976 M AT E R I A L S E VA L U AT I O N • J U LY 2 0 1 7