Eddy Current Nondestructive Testing NATIONAL BUREAU of STANDARDS

Home , Eddy current, Nondestructive testing

A 11 10 3 073222

TECH HAG. NATL INST OF STANDARDS! »

fe* 1 All 103073222

to NBS SPECIAL PUBLICATION 589

U.S. DEPARTMENT OF COMMERCE / National Bureau of Standards

Eddy Current Nondestructive Testing NATIONAL BUREAU OF STANDARDS

The National Bureau of Standards' was established by an act ot Congress on March 3, 1901.

The Bureau's overall goal is to strengthen and advance the Nation's science and technology and facilitate their effective application for public benefit. To this end, the Bureau conducts research and provides: (1) a basis for the Nation's physical measurement system, (2) scientific and technological services for industry and government, (3) a technical basis for equity in trade, and (4) technical services to promote public safety. The Bureau's technical work is performed by the National Measurement Laboratory, the National Engineering Laboratory, and the Institute for Computer Sciences and Technology.

THE NATIONAL MEASUREMENT LABORATORY provides the national system of physical and chemical and materials measurement; coordinates the system with measurement systems of other nations and furnishes essential services leading to accurate and uniform physical and chemical measurement throughout the Nation's scientific community, industry, and commerce; conducts materials research leading to improved methods of measurement, standards, and data on the properties of materials needed by industry, commerce, educational institutions, and Government; provides advisory and research services to other Government agencies; develops, produces, and distributes Standard Reference Materials; and provides calibration services. The Laboratory consists of the following centers:

Absolute Physical Quantities 2 — Radiation Research — Thermodynamics and Molecular Science — Analytical Chemistry — Materials Science.

THE NATIONAL ENGINEERING LABORATORY provides technology and technical services to the public and private sectors to address national needs and to solve national problems; conducts research in engineering and applied science in support of these efforts; builds and maintains competence in the necessary disciplines required to carry out this research and technical service; develops engineering data and measurement capabilities; provides engineering measurement traceability services; develops test methods and proposes engineering standards and code changes; develops and proposes new engineering practices; and develops and improves mechanisms to transfer results of its research to the ultimate user. The Laboratory consists of the following centers:

Applied Mathematics — Electronics and Electrical Engineering 2 — Mechanical Engineering and Process Technology 2 — Building Technology — Fire Research — Consumer Product Technology — Field Methods.

THE INSTITUTE FOR COMPUTER SCIENCES AND TECHNOLOGY conducts research and provides scientific and technical services to aid Federal agencies in the selection, acquisition, application, and use of computer technology to improve effectiveness and economy in Government operations in accordance with Public Law 89-306 (40 U.S.C. 759), relevant Executive Orders, and other directives; carries out this mission by managing the Federal Information Processing Standards Program, developing Federal ADP standards guidelines, and managing Federal participation in ADP voluntary standardization activities; provides scientific and technological advisory services and assistance to Federal agencies; and provides the technical foundation for computer-related policies of the Federal Government. The Institute consists of the following centers:

Programming Science and Technology — Computer Systems Engineering.

'Headquarters and Laboratories at Gaithersburg, MD, unless otherwise noted; mailing address Washington, DC 20234. 2 Some divisions within the center are located at Boulder, CO 80303. National Bureau of Standards

Library, £-01 Admin. Bldg,

FEB 2 7 1981 not o.cc. -£j

C. 3~ Proceedings of the Workshop on Eddy Current Nondestructive Testing, held at the National Bureau of Standards, Gaithersburg, Maryland, on November 3-4, 1977

Edited by:

George M. Free

Center for Absolute Physical Quantities National Measurement Laboratory National Bureau of Standards Washington, DC 20234

\ JCO

iptrml ptiJhhCa.il 0 U.S. DEPARTMENT OF COMMERCE, Philip M. Klutznick, Secretary

Jordan J. Baruch, Assistant Secretary for Productivity, Technology and Innovation j NATIONAL BUREAU OF STANDARDS, Ernest Ambler, Director

Issued January 1981 Library of Congress Catalog Card Number: 80-600172

National Bureau of Standards Special Publication 589

Nat. Bur. Stand. (U.S.), Spec. Publ. 589, 153 pages (Jan. 1981) CODEN. XNBSAV

U.S. GOVERNMENT PRINTING OFFICE

WASHINGTON: 1981

For sale l>y the Superintendent of Documents, U.S. Government Printing Office

Washington. D.C. 20402 - Price $5.50 FOREWORD

Although testing with eddy cur- These Proceedings are a record of that rents is regarded as one of the major Workshop. methods for nondestructive inspection, many people in the industry regard The purposes of the Workshop were to this technique as one that offers much (1) review the current status of eddy cur- greater potential than is presently rent measurement methodology and applica- realized. It is now used primarily tions, (2) define the directions for im- for the sorting of alloys by conduct- proved techniques and applications, and ivity measurements and for the inspec- (3) assess the needs for work on standards tion of relatively thin conducting mate- and underlying science to address present rial; thin-walled tubing constitutes a and future problems. The attendees were major inspection item for eddy current drawn from industry, university, and gov- techniques. ernment. We have thanked them all individually, but it is appropriate here also In the Nondestructive Evaluation to express our appreciation to them again (NDE) Program at the National Bureau of and particularly to the speakers. Standards, we are working to improve the reliability of nondestructive measure- I also wish to express my apprecia- ments. The present effort in eddy cur- tion to the planners of the Workshop, rent testing is directed primarily at Norman Belecki, George Free, and Barry conductivity measurements; a measurement Taylor of the NBS Electricity Division? service and standard reference materials George Birnbaum, of the NDE Program; and are planned to help the industry improve Robert Green of the Johns Hopkins Uni- this type of NDE measurements. Looking versity. I am confident that these Pro- beyond that, however, we at NBS agree ceedings will serve their intended pur- that new ideas and developments can lead poses and help the industry and NBS de- to greater utilization of eddy current fine fruitful areas for additional work methods. One means to examine that to improve eddy current nondestructive potential was a Workshop on Eddy Current testing. Nondestructive Testing; the Workshop was held at NBS on November 3 and 4, 1977, under the joint sponsorship of the NBS Electricity Division and the NDE Program. Harold Berger Program Manager Nondestructive Evaluation February 1978

iii

PREFACE

The intent of these Proceedings is Due to the method of printing the pro- to provide a record of the NBS Workshop ceedings, not all pictures and diagrams on Eddy Current Nondestructive Testing. turned out to be of equal clarity. I apol- With the excpetion of the first paper, ogize beforehand to those authors whose an overview of eddy current testing by pictures or diagrams are not of the excel- Dr. Robert McMaster, each paper present- lent quality which the participants viewed ed was followed by a period of discus- at the workshop. sion. The Proceedings followed the same format. Unfortunately, the comments of participants could not be at- tributed in all cases, but where it is possible the authors of the many comments, questions, and ideas are noted. Some editing of the discussion periods was done, consequently the discussion George M. Free periods are not "verbatim." Editor

TABLE OF CONTENTS PAGE

FOREWORD iii Harold Berger PREFACE v

ABSTRACT vi i i

THE HISTORY, PRESENT STATUS, AND FUTURE DEVELOPMENTS OF EDDY CURRENT TESTS 1

Robert C. McMaster

EDDY CURRENT TESTING: PRESENT AND FUTURE APPLICATIONS IN THE FERROUS METAL INDUSTRY 33

Richard B. Moyer

EDDY CURRENT STANDARDS IN NONFERROUS METALS 39

Carlton E. Burley EDDY CURRENT INSPECTION OF GAS TURBINE BLADES 45

Robert A. Betz EDDY CURRENT EXAMINATION IN THE NUCLEAR INDUSTRY 49

Allen E. Wehrmeister EDDY CURRENT INSPECTION SYSTEMS FOR STREAM GENERATOR TUBING IN NUCLEAR POWER PLANTS 57

Clyde J. Denton USE OF ROUND ROBIN TESTS TO DETERMINE EDDY CURRENT SYSTEM PERFORMANCE 63

E. R. Reinhart

EDDY CURRENT TESTING IN GOVERNMENT 77

Patrick C. McEleney CURRENT STATUS AND FUTURE DIRECTIONS OF EDDY CURRENT INSTRUMENTATION 81

Tracy W. McFarlan MULTIFREQUENCY EDDY CURRENT INSPECTION TECHNIQUES 87

Hugo L. Libby

PULSED EDDY CURRENT TECHNIQUES FOR NONDESTRUCTIVE EVALUATION . 107

D. L. Waidelich THE INTRODUCTION OF SIGNAL PROCESSING TECHNIQUES TO EDDY CURRENT INSPECTION 113

E. E. Weismantel DEVELOPMENT OF NON-FERROUS CONDUCTIVITY STANDARDS AT BOEING 121 Arthur Jones NBS EDDY CURRENT STANDARDS PROGRAM 133 George Free ELECTROMAGNETIC THEORY AND ITS RELATIONSHIP TO STANDARDS 137

Arnold H. Kahn and Richard D. Spal DISCUSSION: NEW SCIENCE DIRECTIONS AND OPPORTUNITIES CONCLUDED FROM WORKSHOP SESSIONS 143

Leader: Robert E. Green DISCUSSION: NEW DIRECTIONS FOR STANDARDS CONCLUDED FROM WORKSHOP SESSIONS 149

Leader: Norman B. Belecki SPECIAL ACKNOWLEDGEMENTS 155 LIST OF WORKSHOP PARTICIPANTS 157

vi i ABSTRACT

The proceedings of the Eddy Current Nondestructive Testing Workshop held at NBS in November 1977 contain papers related to all areas of eddy current testing. A historical overview of the discipline from its inception until the present is given. Other papers discuss the use of eddy current testing in the primary metals industry (both ferrous and nonferrous metals), the use of eddy currents for the sorting of metals and for defect detection, the state-of-the-art in eddy current instrumentation, and the use of signal processing in the analysis of eddy current signals. The development and use of eddy current standards is discussed as well as several of the newer areas of eddy current development, i.e., multi frequency and pulsed eddy current techniques.

Key words: Conductivity; defect detection; eddy current test; mul tifrequency; nondestructive testing.

viii )

National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy Current Nondestructive Testing held at NBS, Gaithersburg, MD, November 3-4, 1977. Issued January 1981.

THE HISTORY, PRESENT STATUS, AND FUTURE DEVELOPMENT OF EDDY CURRENT TESTS

Robert C. McMaster Departments of Electrical Engineering and Welding Engineering The Ohio State University Columbus, OH

1. Historical Development of Eddy Current 2. Oersted's 1820 Discovery of the Magneti Theory and Test Methods (475_478) Field of an Electric Current It is probable that no other form of nondestructive testing has a history of As described by Maxwell, "....Con- illustrious scientific creativity and jectures of various kinds had been made as practical development that compares with to the relation between electricity and the past century and a half of development magnetism, but the laws of these phenomena, of the concepts and applications of elec- and the form of these relations, remained tromagnetic induction and eddy current entirely unknown till Hans Christian testing. James Clerk Maxwell, in his Oersted, at a private lecture to a few remarkable two- volume work A Treatise on advanced students at Copenhagen, observed

Electricity and Magnetism , published in that a wire connecting the ends of a several editions from 1873 to 1891, sum- voltaic battery affected a magnet in its marized the first half century of this vicinity. This discovery he published in a 1 history [l] . In addition, he conceived tract. . .dated July 21, 1820 [2]. and published the comprehensive group of ....Oersted discovered that the current relations known as Maxwell's equations for itself was the cause of the action, and the electromagnetic field, which mathe- that the 'electric conflict acts in a matically represent almost the entire revolving manner', that is, that a magnet present knowledge of this subject. For the placed near a wire transmitting an electric past hundred years, physicists and re- current tends to set itself perpendicular searchers in electricity and magnetism have to the wire, and with- the same end always occupied themselves with numerous appli- pointing forwards as the magnet is moved cations of Maxwell's theory. However, around the wire.... The space in which during this past century, no one has these forces act may therefore be con- conceived any significant new law to be sidered as a magnetic field.... In the added to Maxwell's principles (with the case of an indefinitely long straight wire possible exception of Einstein's theory of carrying an electric current. ... the lines relativity, which extends the theory of the of magnetic force are everywhere at right electromagnetic field to a four- angles to planes drawn through the wire, dimensional framework of three spatial and are therefore circles each in a plane dimensions and a fourth dimension of perpendicular to the wire, which passes time). NOTE: In the following segments through the wire." (Had Oersted been abstracted from Maxwell's treatise, the provided with a much larger current, it is symbol .... indicates omissions. Paren- possible that even a piece of nonmagnetic theses are used to indicate explanatory conducting metal lying adjacent to the words or comments inserted by the author of current-carrying loop would have reacted to this paper. Superscript numbers following sudden application of the current, by eddy headings identify the specific articles of current reaction. Had this accident oc- Maxwell's treatise used as sources. curred, it is possible that the discovery of the effects of eddy currents might possibly have occurred more than 150 years ago.

figures in brackets indicate literature references at the end of this paper.

1 )

3. Ampere's 1820 Discovery of the Mutual 2. The use of comparison-coil arrangements to reduce or eliminate test Interaction of Two Currents^^ signal components related to common properties of two test objects. Maxwell continues "...The action of 3. The use of differential-coil one circuit upon another was originally arrangements to compare local differences investigated in a direct manner by Ampere in properties of adjacent areas of a almost immediately (in 1820) after the single test object. publication of Oersted's discovery. .. .Am- 4. The methods of coupling pere's fundamental experiments are all of magnetizing-coil fields with test ma- examples of.... the null method of them terial surfaces. comparing forces.... In the null method, 5. The use of dual -coil systems to two forces, due to the same source, are balance out external magnetic and elec- made to act simultaneously on a body tric field effects (including terrestrial already in equi 1 ibrium. . . . No effect is magnetism). produced, which shows that these forces are themselves in equilibrium. This method is peculiarly valuable for comparing the effects of the electric current when it passes through circuits of different forms. By connecting all the conductors in one continuous series, we ensure that the strength of the current is the same at every point of its course.... Since the current begins everywhere throughout its course almost at the same instant, we may prove that the forces due to its action on a sus- pended body are in equilibrium by ob- serving that the body is not at all affected by the starting or the stopping of the current. Figure 1. Maxwell's sketch illustrating Faraday's basic test arrangement with "Ampere's balance consists of a astatic balance coil arrangement [1]. light frame capable of revolving about a vertical axis, and carrying a wire which (It would be difficult to estimate how forms two (rectangular loop) circuits of many hundreds of 20th century patents are equal area, in the same plane or in based upon these simple discoveries by parallel planes, in which (loops) current Ampere a hundred years earlier. It would flows in opposite directions. The object also be difficult to estimate the man- of this arrangement is to get rid of the years of effort and dollar costs lost effects of terrestrial magnetism on the during recent developments of eddy cur- conducting wire.... rigidly By connecting rent test systems by workers who were not two circuits of equal area in parallel aware of the full significance of planes, in which equal currents run in Ampere' s 1820 work. opposite directions, a combination is formed which is unaffected by terrestrial 3.1 Ampere's first experiment magnetism. ... (This balance) is therefore 5 ^ called an Astatic Combination (see fig. (Shielding of lead wires)^ 1). It is acted upon, however, by forces arising from currents or magnets which In Maxwell's words, "Ampere's first are so near to it that they act dif- experiment is on the effect of two equal ferently on the two circuits Am- currents close together (flowing) in pere's theory of the mutual action of opposite directions. A wire covered electric currents is founded on four ex- with insulation is doubled upon itself perimental facts and one assumption." and placed near one of the circuits of the astatic balance (See fig. 1). When a Ampere's 1820 experiments provided current is made to pass through the several useful techniques employed in (looped-back) wire and (the compensated present-day eddy current test systems, loops of) the balance, the equilibrium of including: the balance remains undisturbed, showing that two equal currents close together in 1. The methods of shielding lead- opposite directions neutralize each other. wire connections to test coils.

2 If, instead of two wires side by side, a circular flow path of eddy currents in the wire be insulated in the middle of a metal adjacent test material. Small diversions tube, and if the current pass through the and excursions of eddy currents from a wire and back by the tube, the action truly circular path will have very small out'-ide the tube is not only approximately effects upon signal pickup coils coinci- but accurately null. This principle is of dent with the magnetizing coils. Local great importance in the construction of detectors of distortions of the eddy electric apparatus, as it affords the current magnetic field can have far means of conveying the current to and from greater sensitivity to small disconti- any galvanometer or other instrument in nuities than large-area pickup coils.) such a way that no electromagnetic effect is produced by the current on its passage 3.3 Ampere's third and fourth to and from the instrument. In practice, ,. QVnQV on . (507-509,v 520-521)' experiments ' it is generally sufficient to bind the wires together, care being taken that they Ampere's third experiment demon- are kept perfectly insulated from each strated that external currents or magnets other, but where they must pass near any had no tendency to move a straight sensitive part of the apparatus it is current-carrying conductor in the direc- hotter to make one of the conductors a tion of its length (see fig. 2). The tube and the other a wire inside it." fourth experiment showed that the force ( i'hese techniques, including also twisted lead pairs, are commonly used to connect instruments to sensing coils or semiconductor detectors used today to detect eddy current magnetic field test signals. At higher frequencies, shielding by concentric conductors (usually grounded at one end) aids in avoidance of interfering signals from ambient electromagnetic fields or moving ferromagnetic machine parts or test objects.)

3.2 Ampere's second experiment (Effect of (506) crookedA current+ paths)+ k N v '

Maxwell reports: "In Ampere's second experiment one of the wires is bent and crooked with a number of small sinuosi- ties, but so that in every part of its Figure 2. Maxwell's sketch illustrat- course it remains very near the straight ing Faraday's third experiment showing wire. (See fig. 1) A current flowing no force acting along the length of a through the crooked wire and back again current carrying conductor. through the straight wire, is found to be without influence upon the astatic balance. acting between two adjacent current- This proves that the effect of the current carrying loops varies as the square of the running through any crooked part of the distance between the two loops. Analyses wire is equivalent to the same current of these results indicates that the mutual running in the straight line joining its potential M of two closed circuits carry- extremities, provided the crooked line is ing unit current expresses the work done in no part of its course far from the by electromagnetic forces on either straight one. Hence any small element of conducting circuit when it moves parallel a circuit is equivalent to two or more to itself from an infinite distance to its component elements, the relation between actual position. Any alteration of its the component elements and the resultant position, by which M is increased, will be element being the same as that between assisted by the magnetic forces. Even component and resultant displacements or when the motion of the circuit is not velocities." (This basic principle has parallel to itself, the forces acting on been generally ignored with respect to its it are still determined by the variation significance in detection of small dis- of M, the potential of one circuit on the continuities that locally distort eddy other. The force between the circuits is current flow paths. A circular test dM/dx, and thus M, is related to the coil, for example, produces a mirror-image energy of the electromagnetic field of the current-carrying circuits. "

4. Faraday's 1831 Discovery of the Law of 4.1 Faraday's law for induction by variation *• (528-541) Electromagneticci + +• Inductiorrt a of primary current^^'

Maxwell notes that: "The discovery Maxwell advises the reader to read by Oersted of the magnetic action of an Faraday's "Experimental Researches, electric current led by a direct process Series i and ii," and then summarizes of reasoning to that of magnetization by four forms of Faraday's law of induction electric currents, and of the mechanical His description of the first form of action between currents. It was not, Faraday's law follows: however, till 1831 that Faraday, who had been for some time endeavouring to pro- "Let there be two conducting cir- duce electric currents by magnetic or cuits, the Primary and the Secondary electric action, discovered the condi- circuit. The primary circuit is con- tions of magneto-electric induction. The nected with a voltaic battery by which method which Faraday employed in his the primary current may be produced, researches consisted of a constant appeal maintained, stopped, or reversed. The to experiment as a means of testing the secondary circuit includes a galvanometer truth of his ideas, and a constant culti- to indicate any currents which may be vation of ideas under the direct influ- formed in it. This galvanometer is ence of experiment. Faraday. ... shows us placed at such a distance from all parts his unsuccessful as well as his success- of the primary circuit that the primary ful experiments, and his crude ideas as current has no sensible direct influence well as his developed ones. The reader, upon its indications. however inferior to him in inductive power, feels sympathy even more than "Let part of the primary circuit admiration, and is tempted to believe consist of a straight wire, and part of that, if he had the opportunity, he too the secondary circuit of a straight wire would be a discoverer. Every student near and parallel to the first, the other should therefore read Ampere's research parts of the circuits being at a greater as a splendid example of scientific style distance from each other. in the statement of a discovery, but he should also study Faraday for the culti- "It is found that at the instant of vation of a scientific spirit, by means sending a current through the straight of the action and reaction which will wire of the primary circuit the galvano- take place between the newly-discovered meter of the secondary circuit indicates facts as introduced to him by Faraday and a current in the secondary straight wire the nascent ideas of his own mind. in the opposite direction. This is called the induced current. If the "The method of Faraday seems to be primary current is maintained constant, intimately related to the method of the induced current soon disappears, and partial differential equations and in- the primary current appears to produce no tegrations throughout all space.... He effect on the secondary circuit. If now never considers bodies as existing with the primary current is stopped, a secon- nothing between them but their distance, dary current is observed, which is in the and acting upon one another according to same direction as the primary current. some function of that distance. He Every variation of the primary current conceives all space as a field of force, produces electromotive force in the the lines of force being in general secondary circuit. When the primary curved, and those due to any body ex- current increases, the electromotive tending from it on all sides, their force is in the opposite direction to the directions being modified by the presence current. When it diminishes, the elec- of other bodies. He even speaks of the tromotive force is in the same direction lines of force belonging to a body as in as the current. some sense part of itself, so that in its action on distant bodies it cannot be "These effects of induction are in- said to act where it is not. This, creased by bringing the two wires nearer however, is not a dominant idea with together. They are also increased by Faraday. I think he would rather have forming them into two circular or spiral said that the field of space is full of coils placed close together, and still lines of force, whose arrangement depends more by placing an iron rod or a bundle on that of the bodies in the field, and of iron wires inside the coils." that the mechanical and electrical action on each body is determined by the lines which abut on it. "

(This experiment demonstrates the material, under or lagging behind the fundamental principles of the use of moving primary coil. The decay rate of magnetizing coils in eddy current test- dc current measured at a fixed distance ing. The need for a time-varying primary behind the moving primary coil. The current is clearly indicated. The ad- decay rate of dc current measured at a vantage of close coupling or spacing fixed distance behind the moving primary between the magnetizing coil and test coil would contain information similar to metal surface is also shown. This phase and amplitude data obtained by translates into control of lift-off of phase-plane analysis of ac eddy current probe coils, and preference for high test systems in common use today. Of coil-fill factors with encircling-coil course, this type of system would perhaps eddy current tests. The need for pul- best be used with very rapid scanning sating or alternating primary current is over test surfaces.) also now evident. Finally, the advantages of using ferrite or iron cores in 4.3 Faraday's law for induction by eddy current probe coils are suggested. motion of the secondary circuit^*^ Present-day eddy current test systems make full use of each of these prin- Maxwell states also: "If the secon- ciples, enunciated clearly by Faraday in dary circuit be moved, the secondary 1831.) current is opposite to the primary when the secondary wire is approaching the primary wire, and in the same direction 4.2 Faraday's law for induction by motion when it is receding from it. In all of the primary circuit^"^ cases, the direction of the secondary current is such that the mechanical "We have seen that when the primary action between the two conductors is current is maintained constant and at opposite to the direction of motion, rest the secondary current rapidly dis- being a repulsion when the wires are appears. Now, let the primary current be approaching, and an attraction when they maintained constant, but let the primary are receding. This very important fact straight wire be made to approach the was established by Lenz." secondary straight wire. During the approach, there will be a secondary (This example suggests that a current in the opposite direction to the rapidly-moving conducting test material primary. If the primary circuit be moved such as sheet metal in a rolling mill away from the secondary, there will be a could pass by a stationary test coil secondary current in the same direction carrying direct current which induces as the primary. flow of current in material both approaching and leaving the area of this (Two principles are implied by the local magnetization. Detectors of the concept of induction by motion of the eddy current field in either location primary circuit. The first is that po- could respond to local discontinuities or larized and directional secondary cur- variations in material properties which rents can be induced by moving a straight- influence the amplitude and distribution line primary current over a conducting of the eddy currents.) test surface. Secondly, alternating current could be induced in a conducting 4.4 Faraday's law for induction by the secondary circuit or test material when a relative motion of a magnet and the constant-current primary coil is moved secondary circuit^*^ cyclically up and down or side to side over a secondary coil or conducting test Maxwell continues with: "If we surface. Where scanning eddy current substitute for the primary circuit a tests are required, it is possible that a magnetic shell, whose edge coincides with permanent magnet or a direct-current the circuit, whose strength is numerical- magnetizing coil could be used to induce ly equal to that of the current in the eddy currents, without the need for an circuit, and whose austral face corre- electronic oscillator or ac power supply. sponds to the positive face of the cir- An additional concept implied by this cuit, then the phenomena produced by the technique of induction would be that of relative motion of this shell and the using dc magnetic field detectors to secondary circuit are the same as those measure the magnitude of secondary cur- observed in the case of the pri- rent or eddy currents in a conducting mary circuit. " (The coil of the preceding ) " " ) "

examples could be replaced by a permanent where M is the potential of the coupled magnet when relative motion exists between circuits, and I is the current in any coil the magnet and test material in eddy cur- winding. rent tests, providing adequate secondary current magnitude and speed of motion can be attained. 5. Lenz's 1834 Law Showing Effects Opposing Causes in Electromagnetic (542) Induction 4.5 Summary expressions for Faraday's

n * • A 4.- v(531,534,536)' law of -induction ' ' Maxwell's narrative of the development of basic electromagnetic theory Maxwell summarizes the various continues its description of the early statements of Faraday's law with the years of development as follows: "In following statements: "When the number of 1834, Lenz enunciated the following lines of magnetic induction which pass remarkable relation between the phenomena through the secondary circuit in the of mechanical action of electric cur- positive direction is altered, an elec- rents, as defined by Ampere's formula, tromotive force acts round the circuit, and the induction of electric currents by i which is measured by the rate of decrease the relative motion of conductors.... of the magnetic induction through the Lenz's law is as follows: circuit. .. .The intensity of the electromotive force of magneto-electric induction "If a constant current flows in the is entirely independent of the nature of primary circuit A, and if, by the motion the substance of the conductor in which it of A, or of the secondary circuit B, a acts, and also of the nature of the con- current is induced in B, the direction of ductor which carries the inducing cur- this induced current will be such that, rent.... The electromotive force of the by its electromagnetic action on A, it induction of one circuit on another is tends to oppose the relative motion of independent of the area of the section of the circuits. the conductors. .. .The electromotive force produced in a coil of n windings by a cur- (Stated more generally, Lenz's law rent in a coil of m windings is states that the electromagnetic field

. proportional to the product mn. . . will act so as to oppose or resist any effort made to change its intensity or Maxwell finally states the "true law configuration. Where mechanical motion causes the of magneto- induction" in the following change, mechanical force | terms: "The total electromotive force developed within the system will oppose 1 acting around a circuit at any instant is the change. If mechanical motion is measured by the rate of decrease of the absent, electromotive forces will be number of lines of magnetic force which induced which tend to maintain the status pass through it. When integrated with quo, namely to maintain the total fluxj respect to time, this statement becomes: linkages in the system.) The time integral of the total electromotive force acting round any circuit, together with the number of lines of 6. Neumann's 1845 Develpment of magnetic force which pass through the (S Mathematical Theory of Induction v circuit, is a constant quantity. ... This quantity may even be called the funda- Maxwell's history of developments mental quantity in the theory of electro- continues with: "On (Lenz's) law, F. E. magnetism. Faraday. ... recognized in the Neumann founded his mathematical theory, secondary circuit, when in the electro- of induction in which he established the magnetic field, a 'peculiar electrical mathematical laws of the induced currents; condition of matter' to which he gave the

due to motion of the primary or secondary | name of the Electrotonic State." conductor. He showed that the quantity M .... is the same as the electromagnetic (This quantity being defined as of potential of one circuit on the other.... most fundamental nature appears to be We may regard F. E. Neumann, therefore, similar to the concept of 'flux linkages', as having completed for the induction of measured by the product of the number of currents the mathematical treatment which winding turns and the total magnetic flux Ampere had applied to their mechanical enclosed in the winding, N0. This action. quantity is also expressed by the term MI, .

7. Helmholtz 1847 Derivation of Laws of "It appears, therefore, that a (543 system containing an electric current is Induction From Conservation of Energy^ a seat of energy of some kind; and since we can form no conception of an electric In Maxwell's opinion: "A step of current except as a kinetic phenomenon, still greater scientific importance was its energy must be kinetic energy, that soon after made by Helmholtz in his 1 is to say, the energy which a moving body 'Essay on the Conservation of Force, and has by virtue of its motion. by Sir William Thompson, working somewhat later, but independently of Helmholtz. "We have already shown that the They showed that the induction of elec- electricity in the wire cannot be con- tric currents discovered by Faraday could sidered as the moving body in which we be mathematically deduced from the elec- are to find this energy, for the energy tromagnetic actions discovered by Oersted of a moving body does not depend upon and Ampere by the application of the anything external to itself, whereas the principle of Conservation of Energy. presence of other bodies near the current alters its energy.

8. Faraday's Recognition of Electromagnetic

Kinetic Energy and Momentun/^ 9. Influence of Faraday's Research Upon 19th Century Inventors Maxwell reports that: "Faraday showed that (the phenomenon of self- Michael Faraday's two-volume work induction) and other phenomena which he "Experimental Researches in Electricity" describes are due to the same inductive influenced numerous investigators and action which he had already observed the inventors in Europe and the United States current to exert on neighboring conduc- from the 1830' s to the end of the nine- tors. In this case, however, the induc- teenth century. This led many others to tive action is exerted on the same con- experiment with electromagnetic effects ductor which carries the current, and it and to develop many basic inventions such is so much the more powerful as the wire as Morse's telegraph, Bell's telephone, itself is nearer to the different ele- and Edison's many improvements on tele- ments of the current than any other wire graphic, telephonic, fire alarm, and can be. He observes, however, that 'the stock ticker communication systems. first thought that arises in the mind is Faraday in 1831 also showed before the that the electricity circulates with Royal Society a homopolar generator (a something like momentum or inertia in the disc rotating between the poles of a wire. ' Indeed, when we consider one large horseshoe magnet) for converting particular wire only, the phenomena are mechanical energy into electric energy. exactly analogous to those of a pipe full His influence upon inventors with little of water flowing in a continued stream. or no scientific training was very great, If while the stream is flowing we sud- for Faraday's accounts of his experiments denly close the end of the pipe, the did not use any complicated mathematical momentum of the water produces a sudden formulas. A biographer of Thomas Edison pressure, which is much greater than that notes that Faraday appeared to be the due to the head of water and may be Master Experimenter whose laboratory sufficient to burst the pipe. . . notes communicated the highest intellec- tual excitement--and hope as well. "These results show clearly that, if Faraday's explanations were simple, the phenomena are due to momentum, the steeped in the spirit of truthfulness and momentum is certainly not that of the humility before nature. For Faraday, the electricity in the wire, because the same natural laws were revealed through ex- wire, conveying the same current, experiment. To American inventors, Fara- hibits effects which differ according to day, poor and self-educated, indifferent its form; and even when its form remains to money or titles, exemplified the the same, the presence of other bodies ethics of a true man of science, whom such as a piece of iron or a closed they could emulate. Thus, during the metallic circuit, affects the result." period from 1831 to about 1875, the (This latter effect is that involved in inventions made on the basis of Faraday's eddy current testing.) research were often developed by trial and error, empirically, and step-by-step.

7 10. Maxwell's Proposal for Development of simple equations in both integral and

v(552^' differential form were derived by the Theory of the Electromagnetic Field methods of Lagrange, using relationships from the calculus of variations. Based upon the facts previously Solu- tions for alternating fields are also summarized in this introduction, James available for many configurations of the Clerk Maxwell outlines his plan for fields. developing a unified theory of the electromagnetic field, as follows: It is of interest that simpler techniques, using an 'operational "We are therefore led to inquire map' have been devised by the author for whether there may not be some motion presenting these types of equations and going on in the space outside the wire, their derivations in simple form for use which is not occupied by the electric by second-year engineering students. current, but in which the electromagnetic Since the equations are available effects of current are manifested. in nearly all basic textbooks on the electromagnetic field, they will not be "I shall not at present enter on the repeated here. Lord Kelvin devised the reasons for looking in one place rather solutions of Bessel's equation for than another for such motions, or for the cases of probe coils, for example, and regarding these motions as of one kind provided the than another. so-called Kelvin functions from which simple cases can be readily calculated by hand or by digital com- "What I propose now to do is to puters. examine the consequences of the assumption that the phenomena of the electric 11.1 Development of practical electro- current are those of a moving system, the magnetic induction test methods motion being communicated from one part of the system to another by forces, that It has been reported that Hughes nature and laws of which we do not even demonstrated the basic features of eddy attempt to define, because we can elimi- current nondestructive testing in the nate these forces from the equations of 1860's showing that it was possible to motion by the method given by Lagrange differentiate between metallic conducting for any connected system. coins by a simple arrangement of magnetizing coil and induction of eddy currents " 1 propose to deduce the main in the coins. structure of the theory of electricity from a dynamical hypothesis of this kind, 12. Early Tests for Eddy Current and instead of following the path which has Hysteresis Losses in Electrical Steel led Weber and other investigators to many Sheets remarkable discoveries and experiments, and to conceptions, some of which are as Active practical interest in use of beautiful as they are bold. I have electromagnetic means for sorting of chosen this method because I wish to show metals and detection of discontinuities that there are other ways of viewing the did not result in many useful test devices phenomena which appear to me more satis- prior to the beginning of the twentieth factory, and at the same time are more century. However, the numerous develop- consistent with the methods followed in ments including that of alternating cur- the preceding parts of this book than rent electric power systems, and the use those which proceed on the hypothesis of of transformers and other induction direct action at a distance." machines, provided a base of practical design and a need to investigate the losses occurring in magnetic core 11. Maxwell's Equations for Electric materials used in these devices. Much Circuits and for Electromagnetic effort was devoted to reduction of eddy (578-619) Fields current and magnetic hysteresis losses in laminated steel sheets, particularly by Maxwell's remarkable achievement of addition of silicon and other alloying integrating the available knowledge con- elements which lowered their electrical cerning electromagnetic circuits and conductivity and use of purer iron alloys fields provides the basis for analysis of with, in some cases, directional rolling all basic eddy current and electromag- to attain maximum permeability and minimum netic induction problems—and for most of hysteresis losses. modern electromagnetic theory. These To a first approximation, in cores core materials introduced odd harmonics formed of thin magnetic laminations, it into the magnetizing currents or voltages was shown that eddy current losses tended across inductances of their magnetizing to increase in proportion with the square coils (or into unloaded secondary wind- of the frequency, and hysteresis losses ings on the cores), and the high sensi- in accordance with the 1.6th power of the tivity of the harmonic signals to mate- frequency of alternation of the magnetic rial conditions and mechanical stressing field intensity. Numerous laboratories, were known and purposely avoided where including those of electrical equipment possible. manufacturers such as Westinghouse and The General Electric Company, and of These various effects, well-known to manufacturers of electrical steel sheets electrical designers at the turn of the such as Al legheny-Ludlum and Armco Steel century, have since become possible Company, established measurement labora- methods for control or read-out of eddy tories to monitor properties of produc- current nondestructive test signals. tion steel sheets and assure specified (However, in general, the highly- electromagnetic loss factors for elec- permeable electrical steel sheets now trical steel sheets. The well-known commercially-available are not ideal for Epstein test, and many others, were used eddy current tests since their eddy for these material tests. current losses are so very low. For their evaluation, electromagnetic induc- Many improvements resulted, includ- tion tests responsive primarily to hy- ing use of thinner sheets, use of ori- steresis effects, including higher har- ented steel sheets, and use of insulating monic effects, may prove more useful.) coatings between sheets to limit eddy current flow paths. Also discovered during these magnetic core improvements 13. Development of Techniques for Analysis were the undesirable effects of mechan- of Inductive ac Electrical Circuits ical clamping stresses and stresses resulting from punching and shearing of The sinusoidal oscillations of laminations, which tended to increase alternating-current electric power core losses under ac excitation. Hydro- system voltages and currents introduced gen annealing and other techniques, such new complexities in analysis of circuit as those developed by Dr. Trigvie Yensen performance, as compared with analyses of Westinghouse Research Laboratories, for Edison's earlier direct-current led to improved materials such as Hyper- electric power systems. As early as sil, Hypernik, and other magnetic sheet 1893, Professors Crehore and Beddell of alloys with superior properties. Control Cornell Univeristy prepared a textbook of of other alloying elements, additions of analysis of ac electric circuits, in- up to 50% nickel, and orientation of cluding effects of resistive, capacitive, grain structures and magnetic domains and inductive circuit elements. This were used to develop special steels with book was based upon detailed solution of rectangular hysteresis loops which are the differential equations developed by used in magnetic switching of electrical Maxwell, and involved use of calculus in currents, saturable reactors and magnetic each solution. Soon thereafter, Stein- amplifiers, and many novel electromag- metz came to the United States with the netic devices. These developments il- Thomson-Houston Company (later General lustrated the variations in electrical Electric Company) and he developed much conductivity, magnetic permeability, simplified methods of analysis using grain orientation and anisotropy, me- rotating line segments which he called chanical stresses, alloy contents, and "vectors" (now called sinors) to repre- impurity contents, which influenced the sent sinusoidal quantities. As such line electromagnetic response of ferromagnetic segments rotated about one end (at the materials and changed the apparent induc- origin of coordinates), their vertical tance and resistive losses measured by projections mapped out the ordi nates of their magnetizing coils. The use of the sinusoidal waves, when these vertical direct-current bias to adjust the ap- projections were plotted as functions of parent inductance in saturable reactors time. Together with the technique of and transductors for power control pur- representing impedances on a complex poses also illustrated a means for re- plane (with resistance R as a horizontal

ducing magnetic permeability and incre- coordinate, and inductive reactance, X , mental inductance or inductive reactance. as a vertical coordinate), the use of It was also observed that many magnetic these phasor quantities reduced the

9 -

solutions for steady state alternating part, many of these early comparator currents to simple algebra and trigonome- systems were short-lived, and received try, rather than integral calculus. little acceptance in industry. By com- These methods of signal analysis on the parison, a few such developments, spon- complex plane are widely used today in sored by major industries or persistent analysis of eddy current tests, following creative inventors who sought support and their clear enunciation by Dr. Friedrich set up their own companies, survived and Forster of West Germany, following World are used in their modernized form in War II. American industry today.

14.1 1925-1945 American developments of 14. Early Industrial Development of electromagnetic tests for steel products Electromagnetic Induction Comparators Examples of continuing development Numerous electromagnetic induction of electromagnetic induction tests for or eddy current comparators were patented use in inspection of round bars, tubes, in the United States in the period from billets, and products of the steel in- 1925 until the end of World War II in dustry of the United States were those of 1945. Many of these were referenced in Magnetic Analysis Corporation and Repu- 1950 by McMaster and Wenk in an ASTM blic Steel Corporation. Both are based publication, updating a prior (1948) upon the continuing efforts of a few summary of basic nondestructive test dedicated individuals who passed their methods. (See Tables I, II and Appendix skills and enthusiasms along to their I.) Innumerable examples of comparator successors in the same development tests were reported in the literature and organizations. Charles W. Burroughs, in patents. Many provided simple com- Carl Kinsley, and Theodore W. Zuschlag parator coils into which round bars or were among the pioneers of the Magnetic other test objects were placed, producing Analysis Corporation, whose test products simple changes in amplitudes of test are still commercially available in 1978. signals, or unbalancing simple bridge Archibald H. Davis, Horace G. Knerr, and circuits. In nearly all cases, and Alfred R. Sharpies received basic patents particularly where ferromagnetic test for Steel and Tubes, Inc. (now Republic materials were involved, no quantitative Steel Corporation). Their developments analyses of test-object dimensions, were extended and continued in the Elec- properties, or discontinuities were tromechanical Research Laboratory of possible with such instruments. Often, Republic Steel in Cleveland by Cecil difficulties were encountered in repro- Farrow, Archibald W. Black, William C. ducing test results, since some test Harmon, and Joseph Mandula to the large- circuits were adjusted or "balanced" to scale, automated, production-line eddy optimize signal differences between a current test machines for tubes, bars, "known good test object" and a "known and billets in use today. (Other steel defective test object," for each group of companies had early inventors and devel- objects to be tested. Little or no opers of electromagnetic tests but, in correlation could then be obtained be- many cases, their managements did not tween various types of specimens, each support their continuing developments type having been compared to an arbi- over a period long enough to achieve trarily-selected specimen of the same practical applications.) Within the specific type. General Electric Company, an early sequence of inventive development was Many simple comparators operated on pioneered by men like James A. Sams, 60 Hz alternating current from 110 volt Charles D. Moriarity, and H. D. Roop. ac circuits, using conventional instru- Ross Gunn of the U. S. Naval Research ments such as voltmeters, ammeters, Laboratory pioneered a new form of probe- wattmeters and, occasionally phase me- coil magnetizing system with two small ters. Such meters typically absorbed diameter pickup coils displaced symmet- energy from the test circuits, and had rically along a diameter of the magne- typical accuracies and reproducibilities tizing coil. This was an early example often of only 1% or 2% of full-scale of use of one size of coil for magneti- readings. In other cases, well-known zation, and of pickup coils of much Wheatstone bridge circuits were employed different size, in non-concentric posi- to balance out comparison test arrange- tions. (See Tables I and II for details ments, and to provide greater sensitivity of operation of these test systems, and to signal differences. For the most for other examples from this period.)

10 14.2 Post World War II developments 15. Development of Quantitative Eddy in electromagnetic induction tests Current Test Systems By Institut Dr. Forster Rapid technological developments prior to and during World War II (1941- By far the most important factor 1945) in many fields contributed both to contributing to the rapid development and the demand for nondestructive tests and industrial acceptance of electromagnetic to the development of advanced test induction and eddy current tests during methods. Radar and sonar systems made the 1950-1965 period in the United States acceptable the viewing of test data as was the introduction of sophisticated, images on the screens of cathode-ray stable, quantitative test equipment, and tubes or oscilloscopes. Developments in of practical methods for analysis of electronic instrumentation, and in quantitative test signals on the complex magnetic sensors used both for de- plane, by Dr. Friedrich Forster. Dr. gaussing ships and for actuating magnetic Forster is rightly identified as the mines, brought a resurgence of activity. 'father of modern eddy current testing.' After the war ended, developments such as His experience prior to World War II Professor Floyd Firestone's "supersonic included advanced university education in ref lectoscope" for ultrasonic testing, physics and a significant introduction to and Dr. Freidrich Forster' s advanced eddy electromagnetic measurements related to current and magnetometer systems, became the metallurgy and structure of steels available as industrial nondestructive and non-ferrous metals in German research testing systems. These systems offered institutes. During World War II, this new dimensions for nondestructive mea- advanced knowledge was used in naval surement both of material properties and warfare, particularly with respect to of discontinuity locations and relative magnetic mines. At the conclusion of the sizes. The ten-year lag (from 1945 to war, after a period of imprisonment by about 1955) in industrial management's the French, Dr. Forster retrieved his acceptance of novel developments was technical reports and, "with the aid of a uniquely short, in the case of these screwdriver and a technician," began his instruments. Electronic instrumentation further development of electromagnetic based upon vacuum and gas-filled electron test instruments in the upper story of an tubes was approaching the peak of its old inn just a few miles from Reutlingen, development. These developments permit- where he later established his Institut ted easy construction of variable- Dr. Forster. By 1950, he had developed frequency oscillators and power supplies precise theory for many basic types of

- for the magnetizing coils of eddy current eddy current tests , including both abso- test systems. They also permitted minute lute and differential or comparator test voltage or current signals to be ampli- systems, and probe or fork coil systems fied linearly to levels adequate for used with thin sheets and extended sur- display systems, graphic and permanent faces. Painstaking calibration tests recording systems, and for operation of were made with these coil systems and sorting gates, automation of scanning, with mercury models (in which defects and mechanization of materials handling could be simulated by insertion of small during tests. Aerospace and nuclear pieces of insulators). Each test was power industries were developing rapidly, confirmed also by precise solution of and made unique demands for sensitivity Maxwell's differential equations for the and reliability of instruments for ma- various boundary conditions involved with terials evaluation and reliability as- coils and test objects, at least for surance during service. These industries symmetrical cases such as round bars, (and government agencies related to these tubes, and flat sheets where such math- industries) were the primary sponsors of ematical integrations were feasible. research to advance the art of all forms Further studies were made of the non- of nondestructive testing. However, in linear response characteristics of fer- the case of eddy current instrumentation, romagnetic test objects, and methods governmental support was significantly utilizing very low test frequencies (5 less than in other fields of nondestruc- Hz), harmonic signal analysis, compa- tive testing, for two reasons which are rators at various levels of magneti- discussed next. zation, and precise bridge circuits were developed. In most instances, Dr. Forster replaced measurements of the inductance or

11 impedance of test magnetizing coils with However, even more significant has the more precise technique of measuring been the complete transfer of Dr. response with unloaded 'secondary coils' Forster's advanced technology to enter- coupled to the test materials almost iden- prising American firms manufacturing and tically with the magnetizing coils. The distributing nondestructive testing extent and depth of these scientific equipment, since 1952. As many of you may studies were not matched by any laboratory remember, Dr. Forster made his first in the United States, whether under gov- presentation before an ASNT audience early ernment sponsorship or operating indepen- in the 1950's, after learning aboard ship dently. By extensive publications (not about five words of English, namely: initially in the form of U. S. Patents, "Sonny boy" and "I love you." This first but in the open literature), Dr. Fb'rster personal presentation in the United States made the end results of this research was followed by meetings with management available to the world of technical per- of the Magnaflux Corporation, in which the sonnel. His monumental contribution of author served as a technical advisor to almost the entire theory and technology of explain Dr. Forster's designs and electromagnetic induction and eddy cur- discussion. Agreements for licensing rent test techniques to the ASNT Nonde- under Forster patents were later structive Testing Handbook in the 1955- concluded, and the basic Forster 1959 period provided the means for ed- instruments were "Americanized" by use of ucating thousands of other nondestructive U.S. components and electron tubes, for test personnel in the theory, methods, Magnaflux, by the NDT staff at Battel le equipment, and interpretation of eddy cur- Memorial Institute in Columbus, Ohio. rent tests. This integrated presentation (Here, again, the author had an was then used throughout the world to up- opportunity to become aware of the date eddy current test technology. remarkable character of these new instruments.) During the next few years, increasing amounts of technology were 16. Importation of Dr. Fb'rster' s Eddy transferred to Magnaflux, whose staff Current Technology to the United States (under Dr. Glenn L. McClurg) became qualified in the design and production of Dr. The unique developments in Dr. Forster's various instruments, and then Fb'rster 1 s new laboratory in Reutlingen, marketed these electromagnetic induction West Germany, were made known in the test systems throughout the United United States, not only by those capable States. of reading his publications (in German) prior to 1950, but also by missions in which American personnel were sent to Dr. 17. Proliferation of Sources of Eddy Forster's laboratory for education and Current Equipment Derived from experience with these new forms of test Dr. Forster instrumentation. Richard Hochschild, for example, made a visit of perhaps six The collaboration between Dr. months in Reutlingen. Upon his return, he Forster and the Magnaflux Corporation prepared summary reports which were dis- lasted perhaps ten years, during which tributed by the AEC sponsors of his visit. rapid progress was made in both the German Other personnel from private industry and laboratory and in the United States in from other laboratories made visits to advancing the art of eddy current testing. learn of these new techniques. In the Upon completion of the arrangement with United States, numerous facilities began Magnaflux, Dr. Forster marketed his in- research to test these new concepts and struments through the Forster-Hoover instrumentation, including significant organization in Ann Arbor, Michigan. Rudy efforts at Oak Ridge National Hentschel , who was trained in Reutlingen Laboratories, at the Hanford Works, and in at Institut Dr. Forster, provided infor- other facilities. The splendid creative mation transfer to this new organization. work of Mr. Hugo L. Libby at Hanford (More recently, he has developed similar during the past quarter century, and that advanced instruments at his own facility of Robert Oliver, Robert McClung, Caius V. in Ann Arbor.) After a few years, the Dodd, J. A. Deeds, and others at Oak licensing of Forster instruments to Auto- Ridge, which have continued into the mation Industries, Inc. resulted in fur- 1970' s, may well have initially been ther transfer to advanced technology, and inspired (and sponsored in response to) marketing of equipment throughout a new the new work done by Dr. Fb'rster. organization. The most recent arrangement with Krautkramer-Branson has repeated this

12 unique educational process. At present, be subject to rotations and phase shifts, the large organizations manufacturing as well as to attenuation due to dielec- many types of nondestructive testing tric hysteresis losses. In many ways, equipment and marketing their services microwave nondestructive test systems are widely in the United States are presenting analogous in performance applications to updated versions of Dr. Forster' s basic immersion ultrasonic test systems. By test instruments and modifications devel- Maxwell's theory of the electromagnetic oped by their own staffs. Also in the field, microwaves are reflected like light market are the instruments developed by waves by eddy currents induced in the Magnetic Analysis Corporation, those based surface layers of highly-conducting metal- upon Hugo Libby's research at Hanford (by lic materials. Thus, microwaves appear to Nortec), those based upon the Oak Ridge have the capacity to apply high-frequency Laboratory research and developments by eddy current tests to a metallic surface Richard Hochschild and Donald Erdman from a distance, and perhaps to scan such (which have migrated from the originators surfaces to detect discontinuities which through the Budd Company, Automation change the pulse-reflection patterns. Industries, and Tech-Tran in recent years). Basically, in 1977, these various When the Radac eddy current systems brands of conventional eddy current were sold to the Budd Company, Richard instruments are redundant and similar in Hochschild turned his attention to for- nature, having been updated to mation and development of Microwave semiconductor circuit elements and more Instruments Company in Corona del Mar, recently to integrated circuits in some California. Soon a series of instrument cases. With the typical instruments used systems had been developed, and the long to cover various needs and applications, task of educating industrial and scienti- the presently-available instruments fic users in the capabilities and operate with absolute or differential applications of electromagnetic tests had probe coils, encircling coils, internal to be done all over again for these new bobbin coils, and various special coil and higher frequencies. Of course, the theory circuit arrangements, many of which were and design of microwave generators, horns, described in the 1959 ASNT Nondestructive antennas, detectors, and display systems Testing Handbook by Dr. Forster. Self- had been previously developed for balancing or adjusting instruments, long-distance ranging in radar. Many which establish reference points simply by textbooks presented the electromagnetic the placing of probes upon reference test theory of microwaves in terms readily used materials or specimens, are available in by electrical engineers. Microwave system several cases, utilizing developments by components and electron tubes were commer- Hugo Libby and other innovators. Designs cially available. However, these elec- of probes based upon digital computer trical engineers rarely were aware of the analyses of eddy current distributions in needs of nondestructive test engineers, single- or multiple-layer sheet materials and NDT engineers had little familiarity have been made feasible through the with microwaves. In fact, many NDT per- pioneering work at the Oak Ridge National sonnel were still struggling to catch up Laboratory. Special probes with split with the art of eddy current testing at coils, internal magnetic shields, and the lower frequencies, as explained by Dr. other complexities have also been Forster. After several years of diligent developed for crack detection and other development and continued application special applications. Digital displays of research and marketing efforts with the test signals are also being introduced. assistance of Ron Botsco, Microwave Instruments Company was sold and its proprietor moved to greener pastures in 18. Introduction of Microwave the area of medical services. A few other Nondestructive Test and Measuring Systems organizations built simple microwave test systems, but the development of industrial At very high frequencies, electromag- microwave nondestructive testing has been netic fields can be concentrated into languishing during the 1970' s. Limited beams and propagated through space. When research sponsored by ARPA and other such a beam pulse strikes a conducting government agencies has resulted in metallic surface, for example, it is indications of possibilities of crack reflected and may return as an echo to the detection from a distance, since slots and site of the original pulse transmitter, or wires simulating discontinuities in metal- to other detectors, as in radar detection. lic test object surfaces can be detected In dielectric materials, microwaves can under proper conditions of microwave 13 pulse-reflection testing. (The theory of which duplicate phase-plane data consis- microwave antennas and of time-domain re- tently permit a wide range of interpreta- flectometry of microwaves in tubes, pass- tions to be made, depending upon the ing along wires, and reflecting and refrac- strategic test conditions selected. Phase ting in dielectric layers, offer many in- separation of signals to suppress unwanted dications of potentially-valuable nonde- signals and provide desired signals with- structive test applications.) Since mi- out interfering effects are especially crowaves can be focused, microwave sys- valuable where consistency of geometry j tems could also potentially be designed and physical properties of test materials

analogous to optical instruments and test permit their use. The general use of ref- I erence systems, as well as ultrasonic test sys- standards with drilled holes, milled ! tems. However, in 1978, there appears to or EDM slots, stepped wall thicknesses, be no significant commercial development and certain natural defects provides a

or application of microwave nondestructive quick means of assuring proper operation i tests in progress. during testing, or of calibration and adjustment of control settings at the begin- '

On the other hand, a large-scale ex- ning of test sequences on objects of a par- i ample of microwave exploration of objects ticular type or material. These advantages '

at great distances is occurring in radio generally accrue with nonferromagnetic test I astronomy laboratories throughout the materials and symmetrical simple shapes of world. For example, Professor John D. test objects. They cannot always be at- Kraus of The Ohio State University has tained with magnetizable test materials or constructed a large radio telescope in with parts of complex geometry where re- Delaware, Ohio, and is using it continu- producible positioning may not be feasible. ously to map the universe of radio stars and objects which emit microwave signals. By use of magnetic bias (or 'satura- The mapping has progressed to where many tion magnetization'), depths of penetra- radio sources found have been confirmed by tion of eddy currents and a.c. magnetic films from optical telescopes, and others fields into ferromagnetic materials can be have been predicted in location. Possi- greatly increased. Many simple detectors bilities of emissions from galaxies, of surface discontinuities operate quite j 'black holes', and other astronomical effectively during automatic scanning features still exist. Professor Kraus has despite difficulties due to surface rough- recognized this as a form of "nondestruc- ness or varia ions in hardness or magnetic tive testing of outer space" and has permeabilities in test objects. Large- written a delightful biographical book scale through-coil test systems for "The Big Ear," which summarizes a lifetime smaller-diameter rounds and tubes, and

of study and applications of Maxwell's orbiting probe coil systems used with I theory of electromagnetic fields in clear rods, welded tubes, and even rectangular

j and simple words. billets have been developed to a high degree of ruggedness, serviceability, and

reliability for use in steel mills and on i 19. Advantages of Eddy Current Test large-volume inspection applications.

Systems Commercially Available in 1977-78 Automatic marking of defect locations ! permits salvage by grinding out and The eddy current test systems avail- welding repair (if the latter is needed) able commercially in 1978 have many ad- on production line operations. Detection vantages which justify their present wide of defects in surface layers of steels is usage. One great advantage is the repro- well-developed, but measurement of physi- ducibility of measurements possible with cal or metallurgical properties of steels many well-built instruments and test sys- is generally not feasible by eddy current j, tems. Absolute conductivity meters and tests in the United States. One basic | instruments designed for thickness mea- source of difficulty is the sequential use surements of specific metals and alloys of sheets, tubes, or rounds from different are often quantitative and can have accu- mills or different heats, in rapid succes- j racies of 1% or better. Comparison in- sion on production lines. Although the struments permit unique sensitivities for chemical and physical properties of these i detection of discontinuities and of vari- steels from different sources may meet ations in material geometries or proper- manufacturing requirements adequately, no ties. With stable reference specimens, effort is made to control the magnetic ,1 tests can also be repeated with a high permeability properties of these steels to j degree of confidence. Instruments with any type of calibrated standard. As a I phase and amplitude signal capabilities

14 consequence, random variations in magnetic to measure material properties or dimen- permeability prohibit the development of sions, the fact that the quantitative reproducible correlations between absolute displays of signal amplitudes, component measurements of eddy current test signals values, or phase angles have no direct and the actual physical or metallurgical meaning to the untrained observer acts to structures of the test objects. create doubt. When numbers have to be 'looked up in a book or chart' to find the real meaning of test indications, the op- High sensitivity to material electri- portunity exists for human errors. In cal conductivity (in nonferromagnetic ma- addition, since the same book or chart terials) has been attained with small would not be valid for materials other probe coil test instruments typically oper- than a specific material for which the ating in the range of 64 kHz test fre- chart is designed, untrained observers quencies. Such small coil probes tend to will question the results. If modern eddy be sensitive to lift-off, and 'lift-off current tests provided clear, informative compensation systems' such as those de- images or direct read-outs in numbers of a veloped by Dr. Fb'rster are often used to specific dimension, property, or service correct lift-off effects over a small characteristic (which could be immediately range. Similarly, small differential coil checked on reference samples if needed), or field detector systems provide high their use could be multiplied indefi- sensitivity to surface cracks in both non- nitely. For example, where today x-ray or magnetic and in ferromagnetic materials. ultrasonic tests are specified for control However, in general, for such crack detec- of weldments, no one dares to trust eddy tion, manual positioning and scanning with current measurements of these same welds these fine probes is usually required on for control of welding operators or for nonsymmetrical part surfaces or materials acceptance of the welds for specific in service structures and machines. There service conditions. is no low-cost means for total inspection for cracks on parts with complex surfaces, The second disadvantage of present such as those for which liquid penetrant eddy current test systems is that they are tests (or magnetic particle tests on iron greatly limited by artificial constraints or steel parts) provide overall surface inherent in the thinking of present de- inspection at high speed and low costs. signers, manufacturers, and users of these tests. No one has made any fundamental change from the basic designs which Dr. 20. Limitations and Disadvantages of Forster provided in . 1955, nor in the Presently-Available Eddy Current Test methods for interpreting test signals. Systems Because of these unnecessary constraints adopted by tradition, eddy current tests The primary disadvantage of eddy cur- are far less informative or sensitive than rent test systems available in 1977-78 is they should be. Examples of such mental the fact that their test indications are straight- jackets are cited in a succeeding psychologically-unacceptable. They are paragraph. True advancement to the next far less effective in stimulating man- era of eddy current testing cannot occur agement and worker comprehension and until the responsible and active engi- corrective action than the graphic images neers, management, and test personnel provided by other processes such as liquid- develop systems to utilize the full capa- penetrant, magnetic-particle, or x-ray bilities of the method and use these inspection, for example. These eddy cur- systems for effective control of people, rent tests fail to produce a clear, visi- processes, products, and in-service ma- ble, interpretable image of defects or terials and systems. discontinuities from which an almost- instant recognition of their nature, The third disadvantage of present shape, size, or location is obvious to all eddy current test systems is that they are observers. Thus, where the purpose of limited in penetration depths (often to nondestructive testing is to motivate less than 5 or 10 mm) and in magnetizing personnel to best efforts or to permit coil and detector adaptability to rough or immediate correction or repair of defects, contoured test material surfaces. Few eddy current tests which produce fugitive probes or test coils are designed to fit traces on cathode-ray tube screens or into a sharp inside corner or intimately 'meaningless' movements of the needle of a to the outer edge of a sheet material, for panel instrument, are quite ineffective. example. In general, many probe coils are Secondly, even when these tests are used on rigid forms, and cannot conform to

15 irregular contours on test objects. It is the locus curves of response on the com- also often assumed that smal 1 -diameter plex plane. Here, the signal closely probe coils must be used to measure fine approaches the 'empty-coil signal' in both defects or the properties of small areas amplitude and phase. The small contri- of test objects. Yet, small coils assure bution of the eddy current losses to this lack of deep penetration of the magnetic test signal also imply lack of test field into metals or alloys (since the sensitivity. coil field in air is proportionately The sixth disadvantage of some pre- smal 1 ). sent eddy current test systems is the The fourth disadvantage of present variation of magnetizing current amplitude eddy current test systems is their insen- with test frequency. Higher test fre- sitivity to local conditions or disconti- quencies require higher power supply volt- nuities which produce only small distor- ages to provide a given magnitude of cur- tions in eddy current flow paths. (In rent in the test coils. If variations in this sense, "small" is related to the test frequency result in inverse changes diameter of the test coil.) In general, in magnetizing current, tests may be made present test systems do not detect dis- on ferromagnetic test parts at widely- continuities or defects which lie outside different levels of maximum magnetization, the perimeter of the test coils. They are at different test frequencies. This can also typically insensitive to small de- create difficulties with harmonic signal fects which lie on the centerline of the generation and non-linear response charac- test coils. In fact, existing test coils teristics in eddy current test measure- integrate all magnetic flux lines which ments. Alternatively, if true constant their winding turns enclose. With discon- -current magnetization levels cannot be tinuities small in dimensions compared to provided as frequency varies over a wide the coil diameter, the defect signals are range, the designer may limit the test submerged in a large average coil signal instrument to one or a few discrete test so that highly-sensitive detector circuits frequencies for which constant current are needed to detect the minute changes in levels can be assured. Even when multi- amplitude or phase. Even worse, present frequency tests are made at these few coil -type detectors are insensitive to the frequencies, a loss of information at tilt or angle of magnetic flux lines encir- other intermediate frequencies results. cled by the coils. They simply measure the time rate of change of the total mag- A final limitation of some present netic flux enclosed by the test coil. eddy current test systems is their use of This often loses signal magnitude by sinusoidal continuous ac current exci- ratios as great as 100 to 1. Finally, the tations. A useful signal thus lost is use of coils for detection of signals that of magnetic retentivity, and its limits the most minute area detectable to relation to eddy current pulse decay char- that roughly corresponding to the pick-up acteristics. Square wave or spike excita- coil diameter (or the diameter of a ferro- tion can provide both retentivity signals magnetic core within the pick-up coil.) and decay curves for eddy currents within Modern microelectronics can far exceed the test materials. The use of coil -type these limitations on reducing the size of pickups prevents detection of the dc test area whose electromagnetic test components of test signals which could be signal is detected and displayed. generated with pulse or rectangular waveshapes, since response is zero to The fifth disadvantage of present steady- state magnetic flux conditions. eddy current test systems is their limitation to higher test frequencies and to tests at larger phase angles on the com- 21. Artificial Constraints in Design and plex plane. The voltage signal amplitude Use of Eddy Current Test Systems provided by a pickup coil is proportional to the test frequency. If an effort is Possible present stagnation in deve- made to operate at very low test frequen- lopment of new or unique forms of eddy cies to attain deeper penetration and re- current test systems could result from sponse to 'rear-surface' conditions, the constraints in thinking about novel ap- signal can become too low to detect in the proaches, perhaps because these new con- presence of normal noise signals. Even if cepts are not fully documented in the past amplification can permit signal display, history of eddy current inspection. For the low-frequency test condition leads to example, circular test coils were selected signal points on the upper left portion of for initial investigations because they

16 were easy to build and many test objects by the magnetizing coil. If this detector had circular symmetry. Theory also has array could be interrogated in sequence by been directed to circular test coils since rapid techniques such as used to read the solutions of Maxwell's equations for computer memories or to digital ize images, the electromagnetic field could be at- for example, the resultant multi -channel tained more easily with symmetric circular data could be analyzed by digital boundary conditions (such as can be solved techniques, and displayed in any desired with Bessel's equation and its modifi- image format (including two- or three- cations). Actually, however, test coils dimensional images on a television can be wound around square, triangular, or screen). spherical forms. They could be made highly flexible so that they can be made A particularly desirable change from to conform to surfaces of any shape. In prior art would be to utilize very large all cases, advantages accrue in eddy cur- diameter magnetizing coils closely fitting rent testing if the magnetizing coil lift- test-object contours, to assure deep geo- off can be minimized. Flexible magnetizing metrical penetration of the magnetizing coils with stranded conductors imbedded in field. For example, a 10 in diameter test rubber- like sheets or tubes might offer coil could easily project strong magnetic considerable advantages. Applied under fields 2 or 3 inches in front of the coil pneumatic or other pressure, such flexible face. Used with lower test frequencies, sheet magnetizing coils could be fitted to such a coil might provide penetration

gently curved test parts with essentially through 1 or 2 inches of nonmagnetic test zero lift-off. If the detector coil could material (particularly in the case of ma- also be in intimate contact with the terials with electrical conductivities curved surface of the test object, maximum less than about 10% IACS). With arrays of test sensitivity and elimination of non- semiconductor type magnetic field detec- uniform lift-off conditions could be tors, detail sensitivity to near-surface attained. discontinuities might become sufficient to provide good recognizable images of typ- A further typical constraint lies in ical discontinuities and defects. Alter- the assumption that the magnetizing coil natively, a linear array of magnetic field and the pick-up coil should be either (a) detectors might scan linearly across the one and the same coil, or (2) of identical field, or be rotated to provide a circular diameter and coincident in position. scan of the field within or adjacent to True, the literature describes such simple the magnetizing coil. The instantaneous arrangements redundantly. However, the appearance of a recognizable eddy current pickup coil could be of any diameter image of defects would convert this test (preferably smaller than the magnetizing into a psychologically-acceptable test and coil), and be placed at any angle and in greatly increase demand and use for eddy any desired position with respect to the current tests. The repeatability of such magnetizing coil. For example, the pick- images, as coil and probes are moved over up coil could be located at any point, and test surfaces, or tests are repeated after in any orientation, within or completely a time period, would do much to establish outside the annul us of the magnetizing confidence in the reliability of such coil, or even at a point directly under images. In general, the instantaneous only one point of the magnetizing coil character of eddy current images and the winding. In fact, if the pickup coil is ease with which depth sensitivity could be replaced by a semiconductor magnetic field changed or polarized eddy current flow detector, total freedom exists with established, might compare favorably with respect to the number, positions, and x-ray or ultrasonic test images of welds angulations selected for the individual or with fluorescent penetrant or magnetic semiconductor detector elements. For particle tests of surface cracks or seams example, an array of semiconductor detec- and laps. tors could be placed anywhere within, under, or external to the magnetizing coil Another potentially attractive tech- windings to provide a multiplicity of nique is that of using differential probe input signals with only one magnetizing signal pick-ups (preferably by detecting coil. Ideally, such an array should cover unbalance in a four-detector array analo- the entire area enclosed within the mag- gous to a Wheatstone bridge) which would netizing coil or be extended over an area be a direct map of the flow of eddy cur- much larger than the magnetizing coil to rents below test surfaces. The reality of provide total test information concerning eddy current flow paths and their devia- the entire eddy current test field created tions caused by discontinuities could then

17 be visualized readily. Since local detec- use of the analogue circuits used previ- tors in the vicinity of crack ends, for ously for such purposes. In addition, example, can have surprisingly large test incoming test data could be continu- signals (as compared to those of large- ously compared with prior data (from the area pick-up coils), unique opportunities same or other test objects) to detect and exist for precise measurements of crack define differences resulting from discon- lengths and of crack extension rates. tinuities or changes in material proper- These topics are of special interest where ties. A further operation of data- fracture mechanics analyses are to be made smoothing as point-by-point data are of cracks to determine their capability to entered into memory could add an addi- propagate under service stresses. tional degree of precision. Simple extrap- olation or interpolation estimates could be derived from test data so that changes 22. The Pending Revolution in in trends could be detected rapidly as Microprocessor Control of Eddy Current tests progress. Of course, differential Tests measurements, or comparison measurements, could be made also from absolute input Already upon us in the 1977-78 time signals, thus eliminating the need for period is the explosion of use of micro- several test arrangements (absolute, dif- processors and digital computer techniques ferential, or comparison coils) to attain as integral components of nondestructive full information from eddy current tests. test systems and controls. The costs of these components have become so low that A further natural consequence of use they are now toys for amateurs 1 i ke the of digital techniques in data collection older "ham" radio operators. Home compu- and analysis would be the possibilities ters are available in the corner computer for real-time control systems based upon store which rival large-scale digital eddy current test inputs. Recognition of computers of just a few years ago. Micro- material damage in service, high stress processors have already invaded control of levels, or high temperature effects could ignition and carburetors in automobiles or be used to shut down or control systems to domestic appliances in the house-wives' prevent premature failures. In addition, kitchens. They are urgently needed in in processes such as fusion welding, a eddy current instruments, where EPROMs multiplicity of eddy current detectors (erasable read-only program memories) can could be used to monitor and control the be dedicated to specific purposes, such as welding process. Input signals might be providing direct correlations of eddy cur- derived from changing conditions such as rent test signals with material dimensions material thickness, edge or weld groove or properties. Since small test instru- distance, conductivity of metal, tempera- ments could now be made direct reading for ture of metal , depth of penetration of any valid measurements by eddy currents, fusion, and final weld inspection for root the old business of table look-up or in- defects, undercutting, cracking, lack of telligent interpretation has become obso- penetration at the root, and other unde- lete. The advantages are obvious in that sired conditions. Other similar pro- each purpose could involve a separate low- duction control applications could be cost test instrument, or plug-in PROM's cited. could be used to change the correlation data from one test material to another or Still another advantage could be at- from one test frequency to another. tained in telemetering and storage of eddy current test data. Digital data could be The additional advantages to be at- stored in computer memory, transferred to tained by integration of microprocessors magnetic tapes, floppy discs, or other and computers into eddy current test in- large-scale memories, and used as perma- strumentation are the possibilities for nent inspection records (much as data on much more sophisticated real-time analysis nuclear pressure vessels obtained by ul- of test signals. Positions of signal trasonic tests are digitized and stored points could be determined on the complex today). In addition, digital data storage planes, the directions of signal change may in the future permit direct correla- established in response to each test mate- tion of conditions detected by one type of rial variable, and undesired signal com- test (such as ultrasonics) with another ponents could be eliminated, without the type of test (such as eddy currents).

18 23. Future Development of Direct-Imaging mapped similarly for the specific test Eddy Current Test Systems objects in advance.

As noted earlier, the greatest limi- A further potential advantage of tation of eddy current tests, as compared television screen imaging of eddy current with more popular tests like x-ray, pene- test signals could be the low-cost, high- trant, magnetic particle, and C-scan ul- speed production of permanent video tape trasonic tests, is the lack of interpre- records of all test conditions and table images derived from eddy current results. Such video tapes can be made tests. Of course, there is no reason why today on recorders so low in cost that eddy current test probes could not be they too are becoming toys in the home. scanned over test-object surfaces, just as Such video tapes are easily transported is now done in immersion ultrasonics to and can be played back later and at other establish C-scan (or plan view) images locations for review of test results. showing defect locations on the test sur- Video-taped images of standard reference faces. However, such scanning is slow and specimens and defects might also be used costly which inhibits its use even with during visual evaluation of the eddy cur- ultrasonic testing. If eddy current tests rent test images. Also feasible, in these could provide instant images (somewhat cases, is digital enhancement of image like x-ray fluoroscopy) or permit re- contrast, and display of enhanced images cording of test information so that it in various colors or with various could be displayed on the face of a brightness levels which could be adjusted cathode-ray tube like a television pic- for best discrimination of significant ture, the data could become psycho- discontinuities or defects. logically attractive and understandable to many more observers. In addition, if eddy current test images could be informative 24. Future Development of Deep- of conditions through much greater metal Penetration Eddy Current Test Systems thicknesses so as to compete effectively with x-rays and ultrasonics, their useful- The future should see a huge improve- ness could be increased. ment in the depth capabilities of eddy current test systems. At present, most With semiconductor detectors, such as eddy current tests are used for surface indium-arsenide Hall devices, the detector and near-surface inspection where they size can be made quite small (as compared provide high test sensitivities. to the diameters of large magnetizing Uniquely-good performance can be attained coils). If the semiconductor detectors with thin-wall test objects, namely those were formed into arrays (like checker- whose metal wall thickness is a small frac- boards) of perhaps 100 by 100 elements or tion of the eddy current penetration more, within a magnetizing coil of large depth. (Eddy current penetration depth is diameter, the individual picture elements the metal depth at which the eddy current could then be read-out one-by-one in se- density, J, is reduced to about 37% of its quence, just as microprocessor or digital value at the test material surface closest computer images are recorded and repro- to the magnetizing coil.) At a depth of duced on the X,Y coordinates of a televi- three times this penetration depth, the sion picture tube screen. (The hobby eddy current density is only about 5% of computers are now often equipped with the surface current density. At five facilities and programs (in software or times the penetration depth, the eddy cur- PROM's) for such data display as images in rent density is negligible at less than color.) Such data could be collected from 0.5% of its surface intensity. The stan- the detector array, subjected to inter- dard penetration depth is an inverse func- pretation criteria, and the results dis- tion of the square root of the product of played on a full -screen image in short test frequency, material conductivity, times, such as one or two seconds. They and/or material magnetic permeability. In could then be interpreted from the screen, highly- ferromagnetic test materials, the particularly if the microprocessor also penetration depths are typically reduced presents needed digital data correlations by a factor of 10 to 100, as compared with for test conditions and test-object dimen- a nonmagnetic test material. sions and properties. Such tasks as determining if discontinuities were located Improvements in penetration depth are in critical areas of test parts could also obviously attainable by lowering the test be carried out by the microprocessor if frequency or saturating ferromagnetic critical areas had been identified and materials to lower their relative magnetic

19 permeability. The first technique, of sible to utilize short electromagnetic lowering test frequency, was limited in pulses in through-transmission electro- the past by the difficulty of detecting magnetic testing. For example, Paul Gant low test frequencies with pickup signal of Shell Development Laboratories in coils. With semiconductor detector sys- Emoryville, California, used such a system tems, or by adding integrating operational many years ago to transmit electromagnetic amplifiers to pickup coils so as to inte- pulses along oil well drill pipe and steel grate the test signals, it should be pos- tubes. Encircling coils used as trans- sible to work at much lower test fre- mitters and receivers permitted detection quencies. If, as an example, it were of larger discontinuities and of zones of feasible to lower a conventional 10 kHz reduced wall thickness. Richard Hochschild eddy current test frequency to 1 Hz, the also used radar- type echo ranging with his standard penetration depth should increase microwave test equipment to establish by 100 times. However, at such a low test distances to metallic sheet and other frequency, it would take one second to reflectors. complete one cycle of alternation. With modern electronic integration systems, Time domain ref lectometry and stand- such frequencies are not out of the range ing-wave analyses are widely used in high- of feasible measurements. In fact, low frequency electronic engineering analyses. frequency oscillators and analysis systems Microwave parts and 'plumbing' fixtures should be able to handle frequencies as are available from electronic equipment low as 0.01 Hz. manufacturers, for construction of such systems. TDR plug- in hardware is avail- However, increasing the penetration able for high-quality cathode-ray oscil- depth by lowering of test frequency is of loscopes, which can be used directly for no value if the magnetizing coil diameter time-domain reflection eddy current tests. is such that the magnetizing field in air In such tests, microwave pulses are trans- is reduced geometrically so that very few mitted along metallic or dielectric rods, or no flux lines can reach the new pene- tubes, or sheets. Where impedance mis- tration depth limits. The answer here is match conditions are encountered, reflec- to employ large-diameter magnetizing coils tions occur. These systems are entirely (although the eddy current detectors can analogous to ultrasonic pulse-reflection be as small as desired). An example of a tests. Where the electromagnetic waves large-diameter magnetizing coil in present travel in dielectrics or in air around a use is the metal detector used to inspect metallic conductor, reflection can result air passengers prior to boarding aircraft from liquid or solid dielectrics (such as in the United States. Such large-size ceramics), or from metal surfaces (which test coils should also conform to surface typically act as total reflectors). Such contours of test objects, where feasible, techniques might apply for rapid inspec- and provide adequate levels of low-enough tion of metallic material moving at high test frequencies to meet inspection re- speeds in a rolling mill, or perhaps for quirements. Ideally, where feasible, the insepction of dielectric coatings being eddy current test should also result in applied to wires, tubes, or sheets under interpretable images with good psycholo- fast transport conditions. (The travel gical impact, so that they can influence speed of waves encountered in typical both management and workers to their best electromagnetic time domain ref lectometry efforts. on metallic structures is perhaps two- thirds the speed of light (or about 2 x 108 meters per second.) Thus, echoes 25. Future Development of Time-Domain would return from a reflector one meter Ref lectometry Eddy Current Tests from the source in a time period of 10"8 seconds or 10 nanoseconds. Precision, Time-domain ref lectometry is a well- fast-response, high-resolution electronic known technique for detection of discon- signal detection and analysis equipment, tinuities in high-frequency electromagne- such as a cathode-ray oscilloscope or tic field transmission lines, telephone digital systems, would be needed in most and telegraph lines, and by radar. Sim- cases (except when standing wave resonance ilar time-domain ref lectometry techniques conditions are present). are used in ultrasonic nondestructive testing, and particularly in immersion Further development of the recent testing. Short pulses of high-frequency efforts to use microwave beams to interro- wave trains, or a single step or square gate metals surfaces at a distance, to wave pulse, can be used. It is also pos- detect conditions such as slots or cracks,

20 highly-conducting surface coatings, di- serve the microwave signals from large electric surface coatings, or projections trucks, bridges, and machinery. If the and surface irregularities, is still car radio is tuned between broadcasting needed to permit practical test systems to stations, so as to receive only "static" be developed. In a similar sense, use of noise signals, variations in these signals microwave distance measuring devices to can be quickly found as his car passes detect movements of structures such as large trucks with metallic bodies, or large tanks or bridges during earthquakes drives across older iron or steel bridges or under service loading might also be with loose bolts or connections. If long- feasible. The still higher frequency distance static is screened out (as in a laser beams used to range distances in shielded room), the radio signals from surveying are similar, since their electro- contacts dragging across metal surfaces, magnetic waves are still shorter in wave- or from rotating bearings, or from loose- lengths than microwaves. As optical wave- ly-bolted joints undergoing vibration, can guides and signal transmission systems be heard distinctly. In fact, if while improve, it may be possible that these wearing gloves, one taps a knife and fork will also be used for analysis of elec- together while walking about in the vici- trical and magnetic properties of mate- nity of the radio receiver, he can send rials, and so join the ranks of practical Morse code or any other sequence of sig- nondestructive test systems. nals which can be heard on the radio loud- speaker. When two metals rub together, enormous sounds and screeches can be heard 26. The Ultimate Goal: Intelligent as the metals complain of the damage their Materials with Microwave Trouble Signals surfaces are undergoing. When ball or roller bearings rotate under heavy load or The ultimate goal with all forms of with inadequate lubrication, each nondestructive test system development metal -to-metal contact can be announced by should be that of discovering or develop- clicks and distinct signals. Often the ing 'intelligent engineering materials' same sequence of signals is broadcast with which detect troubles by themselves and each rotation of the shaft or of a ball or transmit suitable alarm signals in time to roller with a damaged surface. In all permit human control to prevent disastrous cases, the intensity of these signals can failures. The presently-available tech- be greatly increased by connecting one of nique of acoustic emission nondestructive the metal surfaces to the antenna lead of testing is an example of transmission of the radio (preferably through a shielded signals from materials under mechanical cable). The other metal surface may be stresses or subject to damaging events grounded or allowed to stand insulated such as stress corrosion or fatique damage from all other surfaces. On the other leading to cracking. Man has not tried hand, short-circuiting the two pieces of very hard to hear the many signals emitted metal together at the point of signal by natural and artificial materials. Re- generation generally extinguishes the cent interest has been directed to earth- radio signals broadcast. quake prediction and prediction of danger- ous storms. Probably many nondestructive The well-known triboelectric effect test engineers have not bothered to "lis- (electrification by friction) known by the ten" to the microwave signals emitted by Greeks 2000 years ago, illustrates the metallic surfaces and structures under basis of such microwave emissions. During stress, vibration, or surface attack. Yet, rubbing, one material steals electrons engineers often have to work very hard to from the other material, particularly when muffle or destroy these signals when they contact is broken. Since the electron tend to interfere with intentional human cloud within conducting metals constitutes microwave or radio transmissions. For a plasma, the sudden removal or injection example, it has long been known that rail- of charge locally may create plasma oscil- way axles rotating in journal bearings lations. If one of the metal pieces is create radio "noise" which is considered insulated from the other, it is possible objectionable. In fact, copper straps are that such oscillations result in electro- applied to short-circuit these emissions magnetic waves traveling through the to assure that they do not interfere with metal. It then serves as an antenna to railway signal systems or other communi- broadcast these waves into the space cations. around it. These weak signals can be easily lost in static conditions. Tests Every citizen who drives a car with a in a shielded room permit their clear radio has also had an opportunity to ob- identification and their correlation with material surface characteristics. The differences in electrical potential, mag- author has found these signals to approxi- netic fields, and heat or temperature gra-

1 mate 'white noise , in that they can be dients. When alternating or varying cur- detected at all frequencies from those of rents are induced in ferromagnetic audio amplifiers, through those of a.m. materials, heat is produced not only by and f.m. radio broadcasting, to frequen- ohmic losses proportional to the square of cies of 100 MHz or higher. This could be the current density, but also by hyster- expected from the short time duration esis losses in the magnetic material. The involved in the robbery of electrons from total "iron" losses, composed of both eddy a metal surface. current losses and hysteresis losses, are sometimes employed to indicate material Thus, the ultimate in-service monitor properties. The pick-up may detect vari- system for metallic systems and machines ations in electrical potential distribu- may well be formed of microwave monitors tion, in magnetic field strengths, in high- of electron emissions. When the electron frequency electromagnetic wave properties, charges are removed from a metal , the eddy in temperature, in mechanical force or current reaction is one of high frequen- torque, or in losses in the material of cies, capable of being transmitted through the test object, or combinations of these the metal antenna and from it to detectors factors. A large number of patents cover at moderate distances. Increased stress- tests of these types (Tables III and IV). ing or rubbing of contacts across contam- inated (oxidized) metal surfaces results Potential Pick-up Methods: in enhanced microwave distress signals. These same signals can be used to create Electromagnetic induction tests with television images of metal surfaces, in- potential contact pick-ups were proposed cluding geometric features such as for testing of lead-sheathed cables and scratches, or chemical features such as other tubular conductors by Atkinson (U.S. corrosion, oxides, contaminants such as Patent Reissue 21,853) and Edgar (U.S. fingerprints, or even the effects of ad- Patent 2,186,826). The method has not sorbed gas layers or amorphous coatings. been too popular, however, because it is With a low-voltage electron beam scanning difficult to screen the pick-up circuit such surfaces (as in a vidicon television from the exciting electromagnetic field camera tube), the surface features are not variations. Both Atkinson and Edgar pro- damaged, and their images can be repro- posed methods of introducing suitable neu- duced faithfully over long periods of tralizing or compensating voltages in the time. In this special case, conditions pick-up circuit, 180 deg. out of phase reflected by eddy current reactions as with the disturbing emfs. Braddon (U.S. electrons transit metal surfaces can be Patent 2,074,739) provided a method of imaged with remarkable clarity. Typical locating the flaws in cable sheaths with a images enlarged 30X show detail approach- suitable indicator. Knerr (U.S. Patent ing a few micrometers in dimensions. 2,124,577) also included the use of potential pick-ups in his method for testing tubes and cylindrical objects. Appendix I Magnetic Field Distortion Pick-up Methods: Electromagnetic Induction Non-Destructive Tests The detection of flaws by measurement of distortion in electromagnetic fields (which would have uniform intensity in the Principle of Operation: absence of flaws) has found wide use. Chappuzeau and Emersleben (U.S. Patent Electromagnetic induction non- 1,782,462) employed a series of coils destructive tests are characterized by the whose turns were parallel to the surface induction of varying electrical currents of test objects, such as tubes, bars, and in the test object by means of repeated rails, to detect deviations from normal variations in an electromagnetic field. magnetic field distributions. They fur- This method contrasts with the electric ther employed tuned output and input cir- current conduction tests in which current cuits to obtain optimum response to flows into the test object through direct harmonic signals. Stein (U.S. Patent electrical contacts from an external 1,992,100) proposed tube- and bar-testing source. No input contacts are required apparatus which consists of main exciting with induction- type tests. The induced coils which set up a normally neutral zone current in the test object produces in the fields between them, and a test

22 coil located in the neutral zone, whose Michel (U.S. Patent 2,489,920) used output actuates an indicator such as a vacuum tube phase discriminator circuits cathode-ray tube. to actuate neon indicators and relays in a metal detector for use in manufacturing Several arrangements of exciting and linoleum. Bovey (U.S. Patent 2,495,627) pick-up coils, including those in which used a rectifier bridge to convert unbal- the exciting coil is placed within the anced field signals to a d-c. meter de- tube and the pick-up coils are placed out- flection in his metal -object sorter. side the tube, were patented by David (U.S. Patent 2,065,118). He also employed Transformer Pick-up Methods: electronic amplification of pick-up signals, with provision for permanent Several tests have been proposed in records or for marking the test object. which test objects form the cores of trans- In U.S. Patent 2,065,119, he points out former arrangements. The primary coils that voltages induced by distortion of the are excited with sinusoidal alternating magnetic fields are exceedingly minute, currents, and the secondary induced volt- often as low as one-millionth of a volt. age magnitudes and wave shapes are exam- Also, a variation of 0.0001 in. in the ined to detect flaws or material prop- position of the detector element from its erties. Kinsley (U.S. Patent 1,813,746) true electrical center in the exciting proposed the use of a magnetic oscillo- field can produce comparable signals. graph to examine such wave shapes, as well Furthermore, as the article being tested as the use of relays operating on the dif- moves through the exciting field, its mo- ference between secondary signals obtained tion tends to deflect the electrical from standard and unknown test objects center of the system in the direction of such as tubes and bars. Bill stein (U.S. its travel. Precision adjustments of coil Patent 1,958,079) proposed a rail tester locations are essential to correct these in which the secondary signal is amplified difficulties. and its magnitude suitably indicated. In U.S. Patent 2,084,274, he claimed improved A simple means for testing interior sensitivity as a result of shunting the surfaces of tubes was proposed by exciting and pick-up coils with suitable Greenslade (U.S. Patent 2,104,646), and capacitors. Hallowell (U.S. Patent consisted of search coils connected in a 2,010,189) employed a cathode-ray oscil- Wheatstone bridge circuit. lograph with the exciting signal applied to one set of deflection plates, and the Hay (U.S. Patent 2,150,922) produced differential output signal (between stan- longitudinal a-c. magnetization in cylin- dard and unknown test objects) applied to drical test objects which were then ro- the second set of deflection plates. Ebel tated past a fixed pick-up coil to reveal (U.S. Patent 2,111,210) used concentric flaws. The depth of flaw penetration was exciting and pick-up coils of pancake estimated by varying the exciting frequen- form, located in a plane parallel to the cy or by providing d-c. saturation of surface of the cable sheaths under test. ferromagnetic materials, so as to vary the A similar arrangement was patented by depth of penetration of eddy currents. A Loewenstein (U.S. Patent 2,116,119), the cathode-ray tube was used as an indicator. inner pancake coil diameter being designed to intercept a component of flux greatly In U.S. Patent 2,162,710, Gunn showed dependent upon the thickness of the sheet a small probe containing exciting coils or tube being tested. and pick-up coils located so as to be sensitive only to distortions in eddy current To detect small flaws by eddy current flow in the test object. The pick-up sig- flow in magnetic tubes and bars, Knerr nal was synchronously rectified so that it (U.S. Patent 2,124,577) rendered the mater- might actuate a sensitive d-c. galvano- ial substantially nonmagnetic by subjec- meter. An automobile tire nail detector ting it to a high degree of magnetic satu- patented by Wages (U.S. Patent 2,502,626) ration with a saturating d-c. coil or employed a vacuum tube amplifier and place strong permanent magnet. A transformer- circuit meter to detect magnetic field type exciting and pick-up circuit arrange- distortion. ment was used to compare standard and unknown test objects. In U.S. Patent A third harmonic in the pick-up sig- 2,124,579, he indicates that pick-up coils nal was found responsive to flaws in test should have small dimensions comparable to objects excited uniformly with a 60-cycle those of flaws to be detected, for optimum magnetic field, by Sams and Moriarty (U.S. sensitivity. A plurality of such small Patent 2,007,772). coils, disposed over the surface of the Schlesman (U.S. Patent 2,491,418) test object, may be required for complete proposed that the standard and unknown coverage. test objects be placed successively within or across the opening of a high-frequency Zuschlag arrives at similar conclu- cavity resonator. Changes in conditions sions in U.S. Patents 2,353,211 and of cavity resonance would detect test ob- 2,398,488, in which he proposes the use of jects differing from the standard. small pick-up coils near the surface of a rotating specimen, and suggests several Magnetic Loss Pick-up Methods: arrangements of circuit and detector. Canfield (U.S. Patent 2,245,568) also pro- Burrows (U.S. Patents 1,676,632 and poses to detect flaws by detecting vari- 1,686,679) proposed the use of pick-up ations in eddy current flow, but he uses coils adjacent to tubes and bars excited the quadrature component of pick-up flux by an alternating electromagnetic field. as a sensitive measure of changes in eddy The pick-up coils were connected to the current resistance. This component gives moving coils of a dynamometer relay or an indication which is relatively insensi- meter whose fixed coils were connected in tive to changes in permeability of the series with the exciting current, to ob- article being examined. tain a measure of hysteresis losses in the test specimen. The signal was obtained Irwin (U.S. Patent 2,290,330) devel- from series-opposed pick-up coils, one oped equipment for the simultaneous inde- coil being used with a standard specimen, pendent measurement and recording of a and the second coil being used with the magnitude related to the phase angle be- unknown specimen. The "duroscope," in- tween excitation and pick-up waves, and a vented by Sams and Shaw (U.S. Patent

second magnitude characteristic of the 1 ,789,196-Reissue 18,889), provided mag- pick-up a-c. signal. Variation of leakage netizing and pick-up coils in a single flux, as through a shunt transformer path, probe, and used a similar wattmeter ar- is employed by Thorne (U.S. Patent rangement to measure the iron losses in 2,311,715) to detect flaws which influence the area of cutting tools under the probe. the permeability of rails. DeForest (U.S. Patents 1,897,634 and DeLanty (U.S. Patent 2,315,943) pro- 1,906,551) provided means for measuring posed to concentrate the flux in tubular magnetic losses in sheets and tubes under test objects by introducing low-resistance different testing conditions such that the inserts within the tube at the point of indication was influenced first by the testing. These high-conductivity inserts combined electrical and magnetic pro- have induced in them large eddy currents perties of the material, and second, by a which oppose the entry of magnetic flux change in, and characteristic of, but one into the insert, and presumably con- of the unobserved properties--; for exam- centrate the flux in the tube wall under ple, the magnetic one. These measurements test. were correlated with stresses in the material, in U.S. Patent 1,906,551. High-Frequency Electromagnetic Wave Pick-up Methods: Electromagnetic induction tests in which plates and tubes were excited by a Recent developments in ultra-high high-frequency field coil connected to a frequency sources, wave guides, oscilla- vacuum tube oscillator were employed by tors, and detectors, particularly in war- Kranz (U.S. Patent 1,815,717) and by Mudge time radar developments, have contributed and Bieber (U.S. Patent 1,934,619). The new techniques to non-destructive testing. reaction of the test object (presumably Because of the normal time lag before issu- the magnetic losses) was detected by ance of patents, only a limited number of changes in the amplitude of the exciting such tests have been revealed. Two typi- oscillations in the vacuum tube circuit. cal examples are given below: Eddy current losses in the test Larrick (U.S. Patent 2,489,092) pro- object were employed to detune a high- posed the use of a high-frequency source frequency oscillator, one of whose har- and open-ended wave guide against which monics was heterodyned with a different the surface of the test object is placed. harmonic of a standard frequency signal to Surface resistance and the thickness of provide beat signals, in a rod tester de- nonconducting coatings are evaluated in scribed by Dana (U.S. Patent 1,984,465). terms of the resonant frequency of the Roop (U.S. Patent 2,055,672) placed stan- wave-guide system. -

dard and test bars in opposite sides of an uniformly along the tube wall, the heavy inductance- type bridge circuit, whose un- currents, upon striking a high- resistance balanced output signal was amplified and sand hole or pocket hidden from the sur- applied to a thyratron relay which oper- face by a layer of homogeneous metal, ated suitable markers to indicate the lo- caused a burn-out resulting from fusion at cation of defects on the test object. the point, thus breaking down the wall of the tubular object and revealing the Zuschlag (U.S. Patent 2,077,161) re- hidden defect. veals the difficulties of compensating loss testers for variations (other than Mechanical Pick-up Methods: flaws) in test objects and standards, and for electromagnetic interference from ex- A novel method of detection was pro- traneous sources, and proposes circuit posed by Burrows (U.S. Patent 1,599,645) improvements to reduce their undesirable in which a magnetizable test object was effects. In U.S. Patent 2,098,991, he placed on a rotatable spindle in a three- proposes an artificial standard circuit phase rotating magnetic field. The torque which introduces into the pick-up circuit developed in the object (which acted some- signals corresponding to the indications what like the rotor of an induction motor) of a standard test object, with which the was measured by the displacement of the indications of an unknown test object are supporting spindle against a restraining spring, and was assumed to measure signif- ! automatically compared during testing. Further improvements in the detection cir- icant physical characteristics of the test I cuit are shown in U.S. Patents 2,208,145, object. 2,329,810, and 2,329,811. Technical Literature References Kinsley (U.S. Patent 2,101,780) de- on Electromagnetic Induction signed exciting coil assemblies which pro- Non-Destructive Tests duced uniform flux densities over the test I area in sheets, bars, and strips, so that eddy current and hysteresis losses (of (1) Konig, "Nondestructive Testing of importance in sheet materials for trans- Railway Car Wheels and Axles for formers, dynamos, and other electrical Surface Cracks," Oregon Fortschr. equipment) might be measured in a manner Eisenbahnw. , Vol. 91, December 15, comparable to standard Epstein tests of 1936, pp. 504-509. specially cut samples. (2) L. H. Ford and E. E. Webb, "Nonde- Fermier (U.S. Patent 2,389,190) destructive Testing of Welds," The vised equipment whereby tubes and bars Engineer, Vol. 165, April 8, 1938, could be subjected to a series of exciting pp. 400-401. frequencies, each producing a different penetration of eddy currents. The voltage (3) Anonymous, "Nondestructive Test (Pro- drop across the exciting coil (in the tank duction) for Steel Tubing," Steel, circuit of a vacuum tube oscillator) was Vol. 107, October 21, 1940, pp. registered, for each frequency, by a point 38-40, 75. on the screen image of a cathode- ray oscilloscope, so that the response to several (4) Ross Gunn, "Eddy-Current Method for test frequencies could be observed simul- Flaw Detection in Nonmagnetic taneously. The indications have been cor- Metals," Journal of Applied Me- related with properties, such as carbon chanics, Vol. 8, No. 1, March, 1941, content of material in the surface layers. pp. A22-26.

Thermal Pick-up Methods: (5) W. Jellinghaus and F. Stablein, "Nondestructive Testing to Detect DeForest (U.S. Patent 1,869,336) des- Internal Seams in Sheets," Technische cribed several arrangements of coils for Mitteilungen Krupp, Ausgabe A. Fors- induction heating of tubes and welds, chungsber, Vol. 4, April, 1941, pp. 31-36. whose temperature distribution was i ndi | i cated by isothermal s delineated by melting of stearin or other suitable temperature (6) A. Trost, "Testing Non-Ferrous Pipes, indicators. Somes (U.S. Patent 2,340,150) Bars and Shapes with Eddy Currents," proposed the use of high-frequency induc- Metal lwirtschaft, Vol. 20, pp. tion heating with heavy induced currents. 697-699 (1941); also Chemische As the induction-heated zone progressed Zentralbl, Vol. 1, p. 801 (1942).

25 (7) W. A. Knopp, Jr. , "Rapid Nondestruc- (20) R. L. Cavanagh, "Nondestructive Test- tive Testing with Cathode Ray Oscil- ing of Metal Parts," Steel Proces- loscope," Instruments, Vol. 16, sing, Vol. 32, No. 7, July, 1946, pp. January, 1943, pp. 14-15. 436-440.

(8) F. Forster and H. Breitfeld, "Nonde- (21) P. E. Cavanagh, "A Method for Predic- structive Test by an Electrical ting Failure of Metals," ASTM Bulle- Method," Aluminium, Vol. 25, March, tin, No. 143, December, 1946, p. 30. 1943, p. 130. (22) R. L. Cavanagh, "Nondestructive Test- (9) James G. Clarke and Charles F. ing of Drill Pipe," Oil Weekly, Vol. Spitzer, "Electronic Locator for 125, March 10, 1947, pp. 42-44. Salvaging Trolley Rails," Elec- tronics, Vol. 17, January, 1944, (23) Howard C. Roberts, "Trends in Elec- p. 129. trical Gaging Methods," Instruments, Vol. 20, April, 1947, pp. 326-330. (10) F. Forster and H. Breitfield, "Non- destructive Testing of Light Metals (24) C. H. Hastings, "Recording Magnetic Using A Testing Coil," Light Metals Detector Locates Flaws in Ferrous Bulletin, Vol. 7, April 28, 1944, pp. Metals," Product Engineering, Vol. 442-443. 18, April, 1947, pp. 110-112.

(11) Anonymous, "Nondestructive Testing," (25) Anonymous, "Electronic Comparators," Automobile Engineering, Vol. 34, May, Automobile Engineer, Vol. 37, July, 1944, p. 181. 1947, pp. 271-272.

(12) J. Albin, "Salvaging and Process (26) P. E. Cavanagh, "Some Changes in Phy- Control with the Cyclograph," The sical Properties of Steels and Wire Iron Age, Vol. 155, May 17, 1945, pp. Rope During Fatique Failure," Trans- 62-64. actions, Canadian Inst. Mining and Metallurgy, July, 1947, pp. 401-411. (13) John H. Jupe, "Crack Detector for Production Testing," Electronics, (27) G. B. Bowman, "Measurement of Thick- Vol. 18, No. 10, October, 1945, pp. ness of Copper and Nickel Plate, 114-115. Monthly Review, Vol. 34, October, 1947, pp. 1149-1151. (14) H. L. Edsall, "Magnetic Analysis In- spection of Metals," Materials & Me- (28) R. S. Segsworth, "Uses of the DuMont thods, Vol. 22, December, 1945, pp. Cyclograph for Testing of Metals," 1731-1735. Electronic Methods of Inspection of Metals (Am. Soc. Metals), pp. 54-70 (15) H. Mader, "Magneto-Inductive Test- (1947). ing," Metal Industry, Vol. 68, January 18, 1946, pp. 46-48. (29) Charles M. Lichy, "Determination of Seams in Steel by Magnetic Analysis," (16) P. E. Cavanagh, E. R. Mann, and R. T. Electronic Methods of Inspection of Cavanagh, "Magnetic Testing of Met- Metals (Am. Soc. Metals), pp. 97-106 als," Electronics, Vol. 19, August, (1947). 1946, pp. 114-121. (30) P. E. Cavanagh, "The Progress of Fail- (17) Vin Zeluff, "Electronic Inspection," ure in Metals as Traced by Changes in Scientific American, Vol. 174, No. 2, Magnetic and Electrical Properties," February, 1946, pp. 59-61. Proceedings, Am. Soc. Testing Mats., Vol. 47, p. 639 (1947). (18) J. E. Carside, "Metallic Materials Inspection," Metal Treatment, Vol. (31) P. E. Cavanagh and R. S. Segsworth, 13, Spring, 1946, pp. 3-18. "Nondestructive Inspection of Mine Hoist Cable, " Transactions, Am. Soc. (19) G. R. Polgreen and G. M. Tomlin, Metals, Vol. 38, pp. 517-545, dis- "Electrical Nondestructive Testing of cussion, pp. 545-550 (1947). Materials," Electronic Engineering, London, Vol. 18, No. 218, April, (32) Carlton H. Hastings, "A New Type of 1946, pp. 100-105. Magnetic Flaw Detector," Proceedings,

26 -

Am. Soc. Testing Mats., Vol. 47, (43) Theodore Zuschlag, "Magnetic Analysis p. 651 (1947). Inspection in the Steel Industry," Symposium on Magnetic Testing, pp. (33) Abner Brenner and Eugenia Kellogg, 113-122, Am. Soc. Testing Mats., "Magnetic Measurement of the Thick- (1949). (Symposium issued as ness of Composite Copper and Nickel separate publication STP No. 85.) Coatings on Steel," Journal of Re- search, Nat. Bureau of Standards, Vol. 40, No. 4, April, 1948, pp. 295-299.

(34) A. M. Armour, "Eddy Current and Elec- trical Method of Crack Detection," Journal of Scientific Instruments and of Physics in Industry, Vol. 25, June, 1948, pp. 209-210.

(35) Kurt Matthaes, "Magnetinduktive Sta- hlprufung," (Magneto-Inductive Test- ing of Steel) Zeitschrift fiir Metal lkunde, Vol. 39, September, 1948, pp. 257-272.

(36) James Gee, "Testing and Inspection of Wire Ropes," Mine and Quarry Engi- neering, Vol. 14, December, 1948, p. 375.

(37) I. R. Robinson, "Magnetic and Induc- tive Nondestructive Testing of Met- als," Metal Treatment and Drop Forging, Vol. 16, Spring, 1949, pp. 12-24.

(38) Abner Brenner and Eugenia Kellogg, "An Electric Gage for Measuring the Inside Diameter of Tubes," Journal of Research, Nat. Bureau of Standards, Vol. 42, No. 5, May, 1949, pp. 461-464. (RP 1986).

(39) George A. Nelson, "The Probolog, for Inspecting Nonmagnetic Tubing," Metal Progress, Vol. 56, July, 1949, pp. 81-85.

(40) P. Zijlstra, "An Apparatus for Detec- ting Superficial Cracks in Wires," Philips Technical Review, Vol. 11, July, 1949, pp. 12-15.

(41) G. H. Vosskuhler, "Nondestructive Testing of the Al-Mg-Zn Alloy Hy-43 by a Magneto- Inductive Method," Metal 1, Vol. 3, August-September, 1949, pp. 247-251 and pp. 292-295.

(42) P. Schneider, "Measuring the Wall Thickness of Light-Metal Cast Parts With Dr. Forster's 'Sondenkawi- meter'," Metal 1, Vol. 3, October, 1949, pp. 321-324.

27 A

Symposium on Non-Destructive Testing

table iii.—patents on electromagnetic induction non-destructive tests.

Patent Patent No. Inventor Assignee Title Date

1,599,645 1/26/24 Charles W. Burrows Burrows Magnetic Method of Testing Magne- Equipment Co. tizable Objects 1,676,632 7/10/28 Charles W. Burrows Magnetic Analysis Method of and Apparatus Corp. for Testing Magnetizable Objects 1,686,679 10/ 9/28 Charles W. Burrows Magnetic Analysis Apparatus for Testing Mag- Corp. netizable Objects 1,782,462 11/25/30 Helmut Chappuzeau Neufeldt und Kuhnke Arrangement for Testing and Otto Emersle- Betriebsgesellschaft Magnetizable Objects ben 1,813,746 6/ 7/31 Carl Kinsley Magnetic Analysis Method of and Apparatus 8 No Corp. for Magnetic Testing 1,815,717 7/21/31 Hermann E. Kranz Western Electric Co. Apparatus for Measuring 3 No Variations in Thickness of Metallic Bodies 1,869,336 7/26/32 Alfred V. de Forest Thermal Method of Testing 23 No Metallic Bodies

1 , OV / , Oj'l LJ 1^/ oo Alfred V. de Forest American Chain Co. Method of and Apparatus 30 No for Electromagnetic Test- ing 1,906,551 5/ 2/33 Alfred V. de Forest Magnetic Testing Method 12 No and Means T A QotTrtc on/1 \/ii"tril XVC. 10,007 / / -r/ JJ J. /I. OdlUS tlUU virgn General Electric Co. Apparatus for Testing Metals 10 No F. Shaw 1,943,619 1/16/34 Wm. A. Mudge and Method and Apparatus for 2 No Clarence G. Bieber Testing Materials 1,958,079 5/ 8/34 Arthur E. F. Bilstein The Pennsylvania Rail- Method and Apparatus for 21 No road Co. Testing for Internal Flaws 1,984,465 12/18/34 David W. Dana General Electric Va- Method of and Apparatus 15 No por Lamp Co. for Detecting Structural Defects in Materials 1,992,100 2/19/35 Wilhelm Stein Testing Flaws and the Like 8 No in Working Materials 2,007,772 7/ 9/35 J. A. Sams and Charles General Electric Co. Magnetic Testing Apparatus 6 No D. Moriarty 2,010,189 8/ 6/35 Howard T. Hallowell, Standard Pressed Steel Means for Testing Metal 2 No jr.T, Co. 2,055,672 9/29/36 Harold D. Roop Metal Testing Device 7 No

I , uOO ,118 Li./ LLf JO Archibald H. Davis, Steel and Tubes, Inc. Method and Apparatus for 9 No Jr. Testing Metals for De- fects

2, Uuj, 1 iy llj ILJ oO Archibald H. Davis, Steel and Tubes, Inc Flaw Detection 11 No Jr. 0 720 of2 /9lo/2 /27Of Fred D. Braddon Sperry Products, Inc. Indicating Device for Flaw 5 No Detector 1 A77 1A1 A /I 3/3.7 Analysis z , u/ / , loi *t/ lo/ oi Theo. Zuschlag Magnetic Analysis Magnetic Method g No Corp. 2,084,274 6/15/37 Arthur E. Billstein The Pennsylvania Rail- Electrical Tester 64 No road Co. 2,098,991 11/16/37 Theo. Zuschlag Magnetic Analysis Co. Magnetic Analysis 8 No 2,101,780 12/ 7/37 Carl K. Westfield U. S. Steel Corp. Electromagnetic Testing of 18 No Materials 2,104,646 1/ 4/38 Grover R. Greenslade Pittsburgh Dry Sten- Means for Testing 3 No cil Co. 1 ill 0 1 1/1C/70 z, ill, ziu o/ Lo/oo Lawrence C. Ebel Anaconda Wire and Apparatus for Determining 7 No Cable Co. Wall Thickness 2,116,119 5/ 3/38 Alfred Loewenstein System for Electrically Meas- 13 No uring the Thickness of Metallic Walls, Sheets, and the Like 2,124,577 7/26/38 Horace C. Knerr and Steel and Tubes, Inc. Method and Apparatus for 9 No Alfred R. Sharpies Testing Metal Articles 1'\ 1,1 T II,,,, 2, 15U,y2z uonaiu Li. nay Apparatus and Method for 9 No Detecting Defects in Elec- trically Conductive Ob- jects 2, 162, 710 6/20/39 Ross Gunn Apparatus and Method for 23 No Detecting Defects in Me- tallic Objects 2,186,826 1/ 9/40 Robert F. Edgar General Electric Co. Eccentricity Indicator 4 No 2,208,145 7/16/40 Theo. Zuschlag Magnetic Analysis Co. Magnetic Analysis 18 No Re. 21,853 7/15/41 Ralph W. Atkinson General Cable Co. Method and Apparatus for 36 No Measuring Eccentricity 2,245,568 7/17/41 Robert H. Canfield Method of and Apparatus 12 No for Examining Ferromag- netic Articles

28 McMaster and Wenk on a Basic Guide

TABLE III—PATENTS ON ELECTROMAGNETIC INDUCTION NON-DESTRUCTIVE TESTS (Continued).

of Patent Ex- Patent No. Inventor Assignee litlc Date Claims pired Number

2,290,330 7/21/42 Emmett M. Irwin Magnetest Corp. Method of and Apparatus No for Testing Properties of Materials 2,311,715 2/23/43 Harold C. Ghorne Apparatus for and Method 26 No of Detecting Haws in Rails and Other Objects 2,315,943 4/ 6/43 Loren J. DeLanty Sperry Products, Inc. Means for lesting lubes 2 No 2,329,810 9/21/43 Theo. Zuschlag Magnetic Analysis Electromagnetic Inspection 19 No Corp. 2,329,811 9/21/43 Theo. Zuschlag Magnetic Analysis Electromagnetic Inspection 19 No Corp. 2,334,393 11/16/43 Lyle Dillon Union Oil Co. Determination of Magnetic 10 No and Electrical Anisotropy of formation Core Sam- ples 2,340,150 1/25/44 Howard E. Somes Budd Induction Heat- fault- lesting Articles of 3 No ing, Inc. Electrically Conductive Material 2,353,211 7/11/44J Theo. Zuschlag Magnetic Analysis Electrical Analysis 3 JNo Corp. 2,389,190 11/20/45 George F. Fermier Reed Roller Bit Co. Testing Means 4 No 2,398,488 4/16/46 Theo. Zuschlag Magnetic Analysis Magnetic Analysis 4 No Corp. 2,489,092 11/22/49 C. V. Larrick General Electric Co. High-Frequency Surface Test- 6 No ing Instrument 2,489,920 11/29/49 F. C. Michel General Electric Co. Metal Detector 5 No 2,490,554 10/15/46 Harcourt C. Drake Sperry Products, Inc. Flaw Detector for Tubing 4 No 2,491,418 12/13/49 C. H. Schlesman Socony-VacuumOilCo., Automatic Inspection De- 2 No Inc. vice 2,495,627 1/24/50 D. E. Bovey General Electric Co. Method for Sorting Metallic 1 No Articles 2,502,626 4/ 4/50 Morris L. Mages Magnaflux Corp. Electronic Metal Detector 4 No

29 I s

> 43 43 £8 .13 -0 H «3 s 3 2 — S -a — o ol 43 2 43 E .a « _ ,Q mo .a 9 a. a e ^ *3 2^ = (i(J« ol 0) m Jl (J — - — 3 01 3 (A 3 i H to u a ex H

g oj? cr? g oji? 5 J? g ° c c c ° „ 0/ 4) O 03 O 43 O 43 TJ ajTS ^"0 V "V .2 .2 -o OS S '3 S2'S.'252S V < 3 "> 3 3 01 3 EC =1 3 3 o o o o o o o O O H o o o o o A O o o o o o c oj T3 -a -a c ui ' X3 t3 4) tn C 4J w CO - ** 4» w >> « - ft- « "43 "43 -° "a> 3 „ us us — US US .a 3 c.a ca e.a x.a .a e .a c u O 41 43 3 • r: 43 • — 43 • — 43 43 U - 43 QJ " a c o >.t! ctct|c~c e to ^ *j -a o m o to o to-a m to o ° « of a ft. w a ssS a a a a a" W W W U

>> Ji 2 ?•> — 1H ^ " 1- 1 OS 3 «03 O u •3 "5-5 (- w ^ 3 >. a 3 03 3 E E E u E o o a g g 2 e_a-3 ° e.a "i i ._ °E 13 4) E 1 2 S3 o E 0.2 o3^^-^ E u 3 O - E Q S3 > u > cj a. <

•a a I 43 c oj ri aE'= E'f E S S o " oj-5 V C 3 U3 *-* C *J 4) e-2 1) Ul U OJ 03*3^3 I 3g, Is E-g^ E 3 = o o _ ° i U | - — ' e i- c i 2! 3 S 2 3 = 2 § c B C C E £ "5 P i- 5 »- ~ « jl; . 3 4) t t *S £ S *J § ,.S- c a~ 3^ c-'^: u 3 M u - - — .ol 01 rt ol —^ ol — oj a H 0 HOC a Q q'"> ^ ^ > > > £ > >

coi coi coi coi coi coi coi coi ^ -a ^ :mf. :mf. :mf. :mf. :mf. E E E •jui: E E 43 o Tt _™ 43 ced ced ced ced ced ced ced ced S U Zl 3 ai a ai V 3 3 o o " n o'o o'o = T3 T3 apu ndu npu] npu] npu] npu] npu Indi - u-O C o S u j £55 H H ft. H H «

St o tic a a a a a a a a a a a a a

Cj _3 a a a ;

V ,~ 43.3 43. SI 43. W W W W W W 3 w 5 w

O O O O u o o o tn u u u u O Ut VI IH f >~ i- I- t' 0 o o O O H O J. ;St2SSi;3i! 43 *j 4» 3 u a 5 S2 S2 *J 03 *J 43 *J 43 O 01 (S ol £- :gE2E2E2E-32 E 2 E S £ 2 E 2£ 3 2 j 2 £ 2 £ 2 £ 2 g 2 g 2 £ « :a3«4J»43343t;*-'43 £ 43 Jr 43 t 43 ' 3^2 fi£ - c o g 5 c S u 43 if, 43y34«ui43(flC Lj 43 43 tA 03 > ol

o -# ~* ° *2

CJO

^ ^ ^ cs cn es

30 .

-3-2 tn S3 Jo ^> Z U g £ j_, O

a ui 3 3 Ul o o o o o O O O O O w O O o

o c g —_o _o _o u 60 „, 60 g To-3 a 3 a ^ a rt > = ,.- 3 5 3 g 3 e 3 C 3 c 3 (J -rt *-> *j +J *j

—; c ^— c r« «j u C c* bo.O — •OOJ M S 3 t'~ T-j C rt rt rt _£S 3 rt.O. w S S W W W W W S H w w

T3 if O 0) *-»

3 1 2 c o OJ G E c z-a ' I:r u^ _ o '-> !fl U) c sS§i.&-oi a i rt *i ° g - c 5 3.1.* as 8 J S ^— 3X1 o .2 « rt — ni — P < I O o o O O U

*j "O -S 6

- H 3— ya > "C is e 3 ,h 3 <".- j; ' c p rt c o

> o o > > > > ffi > pq o

^? « tc^, 3 S E E S o . w O 'C 3 w U « o > g-c^i « a §3 81 > w E *J CJ 3 g UTJ o H few en

s a §

to c 60 C M c 60 C 60 C 60 C 60 d 60 C 60 c 60 C to C 6C C W)C bcC bDC fc£C 60 C 60 C rt O O C3 O O rt O rt O rt O rt O rt O rt O rt o rt o rt o rtOrtOrtOrtO rt O rt O £•= B-C cf r-g E~ 6*5 ! i'i ill! O U I! 3 tJ-O M

o o rt o o o o o t. ° ii ^ rt £" ^ « g- g 2 Ei 2_ S £ 2 E 2 E 2 « 2 6 2 E 2 2 E o . k °. Ir! E . S Cj >^ (U to J2 0 «s c ° c^. c 2 o rt c£ rt u Oy^ tj ** u 1 3 | rt rt rt > > > is >

O *-< O o> lO « c> «h cs «h

cs" CM

31 -

_ o

60, ^ fcfl 4J 3 f" w ^ 3.2 .2 to g.5 o H a

41 O. 3"0

5." c 01 V o 41 O a> 41

< 3 01 3 01 3 01 O O O3 O

.2 c c *J o -MO - a > rt M O M M u o > S SoJS

C 4) M 2 E £ ,;o e « c 2 ° 41 «J J™ E si it I S3 55

D.O aj 41 41 II §2 3 II 3 O •is!! 3.2 o-2« b rt S « 3 » H > H « >

41 C 3 o a c o •a3 C to c > a &E rt 3 C 41 £ S3 £ O

41 ^ O 3 3 O g4i « C 4i bo o cO U bp • S S d »+-. 4i — c — 3 41 -P„ O g . to

S3 is § S3

a c« '-3 2 41 C 4i 2 3 3 bo C S O" c c W 3 o* g to a bo C 8 n bo c u e ^ a 41 E « o fl O ei O O "is if U > rj = £ 3 >.J=-a_ „ S 3 8 t-TTT.- bo— .5fT5 S w.S 4i. S 2--2* S3 W 5 o S3

U h 2 C OJ! £ .3 fro 41 'tin 3J2 C « c 3 " bo ° 41 41 C 2 >, c tS 3Jl i! c

32 National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy Current Nondestructive Testing held at NBS, Gaithersburg, MD, November 3-4, 1977 '"IssuedUCu January 1981.

EDDY CURRENT TESTING: PRESENT AND FUTURE APPLICATIONS IN THE FERROUS METALS INDUSTRY

Richard Moyer Carpenter Technology Corporation Reading, PA 19603

The scope of these remarks is to Almost twelve years ago, the newly present the past, present, and future formed Institute committee surveyed the applications of eddy current testing in steel industry NDT practices through a the ferrous metals industry, or more series of questionnaires. The companies specifically, in the basic steel producing reporting provided information on a total industry. The source of the data for the of 313 NDT inspection systems. These review of the past is a survey conducted involved four product types: bar, plate, about ten years ago by the American Iron semi-finished, and tubular products. The and Steel Institute. The current and distribution of these among the major NDT future information originates from dis- disciplines is shown in the first table. cussions and correspondence with members of the AISI Technical Committee on Non- An assessment of equipment reli- destructive Testing and Inspection Sys- ability was reported in the survey. Re- tems. spondents were asked to judge the reliability of a system as excellent, good, I am grateful for the opportunity to fair, or poor. To these, numerical values review for this particular audience where 4 through 1, respectively, were assigned. we have been, where we are, and where we The resulting average ratings for the would like to go. It is apparent that the various types and methods are shown in the participants of this workshop and the second table. National Bureau of Standards will have a strong impact on these future directions. The prominent feature of these data It is forums like these that will pilot is that eddy current testing had appeared advancements into useful and practical to reach a maturity as long as ten years channels to the benefit of us all, the ago from both an application and a re- steel industry included. liability standpoint. It certainly was one of the big five of NDT.

Table 1. Number of Systems

Product Eddy Liquid Magnetic Radio- Ultra- Form Current Penetrant Particle logical sonic

Bar 31 16 17 Plate 5 18 Semi -Finished 17 38 Tubular 48 17 26 22 58

Table 2. System Reliability

Product Eddy Liquid Magnetic Radio- Ultra- Form Current Penetrant Particle logical sonic

Bar 3.2 3.0 3.4 Plate 3.4 Semi-Finished 3.1 Tubular 3.2 2.4 3.1 3.1 3.2

33 .

But what about today? Has eddy the apparatus, the longer the downtime. current testing extended any horizons? This is not as serious a disadvantage for Has it fulfilled any new application in-process testers as for separate test

needs? Is it more sensitive? More stations, for presumably, adjustments to i More reliable? Do the develop- the tester could be accurate? made while the pro- I ments of the last ten years signify a cessor is changed. truly expanding technology? I Convenience of calibration is another

It is unfortunate that a similar operational problem. This may be more I statistical survey reflecting today's difficult on an in-process tester than on i usages is not available to give quan- a test station, especially if a reference j

titative answers to these questions. A calibration piece with real or artificial \ survey is not needed to fill in one of the defects is involved. blank spaces of Table 1. Eddy current inspection of semi -finished billets, both Another operational difficulty is round and square, has become an accepted maintaining the proper transducer (coil) to technique, routine in some mills. product spacing and alignment despite the influence of temperature extremes, me-

The topic of semi-finished product chanical handling system irregularities, I

testing leads directly to an expanding crooked ends, etc. The latter could be so I related area of application; in process severe as to cause damage or excessive <

testing. This is simultaneous ECT while wear, thus shortening the preventative i another phase of the manufacturing cycle maintenance cycle. Ease of maintenance is being performed. Utilization of the is, of course, another operational con- technique has been made in conjunction sideration. with bar straighteners , wire drawing blocks, and more recently, hot rolling In considering performance problems, mills for both tubular and solid products. relability must be listed prominently. Is it a certainty that, when set to detect In addition to inspecting its own .008 inch seams, an eddy current tester products, the steel industry has found ECT will not accept a bar with one .012 inch [ useful in defect detection in items of its deep? Or, will a bar with a harmless processing equipment. Rolling mill rolls scratch of .004 inch be rejected? Also, and crane hooks are notable examples. in sorting mixed steel, is the separation absolutely correct? The steel industry is The steel industry has found eddy not sure. current testing techniques applicable for I uses other than flaw detection. The more Another performance problem is

J prominent of these are coating thickness accuracy of calibration. There are many

j measurement and sorting. The thickness types of artificial defects used for and uniformity of copper plating on cold calibration or set-up purposes. The true

heading coil stock is reliably determined correspondence of the eddy current re- I natural defects is by this technique. Sorting for separation sponses to them and to |j

of grade, hardness, size, or other feature either unknown or something vastly' dif- ! by eddy currents is a field equally as ferent from one to one. This lack of large and as important as flaw detection. correlation is also influenced by the way the artificial defect was made--electric j The problems associated with today's discharge machining, mechanical metal t ECT applications in the steel industry can removal, manual filing, etc. The shape of be grouped into two categories: opera- the calibration defect is not always rep- !, tional and performance. "Operational" is resentative of the true defect it mea- concerned with the ease or convenience of sures, as with a hole drilled through the employing the method, while "performance" wal 1 of a tube. concerns the sensitivity, accuracy, reliability, and/or effectiveness of the Another performance deficiency is the test. inability of encircling differential coil systems to reliably detect and accurately One of the principal operational evaluate continuous defects. This need is difficulties is the loss of productive severely felt on installations where time during set-up and calibration for rotating probes are unsuitable, such as on i stock size change. This problem exists a hot rol 1 ing mi 1 1 for both encircling coil and rotating probe machines—obviously the more complex

34 .

Test sensitivity and its relation to Discussion inherent noise creates a performance problem. Often a realistic accept/reject Question (Mr. Weismantel, General Electric level cannot be achieved with eddy current Co.): I have two questions, both of them test equipment because that level is involve opinion more than anything else. submerged in background noise. Unfor- Number one, as you know, we looked at tunately, phase adjustments do not always quite a few different mills with respect provide a differentiation to solve the to the NOT capability. One of the things problem. we noticed, especially in the eddy current area more than the ultrasonic area, was A performance feature not yet ad- the lack of standardization between mills equately addressed by equipment designers within the same industry as far as how the is the lack of clear relation between the process was applied and how it was con- instrument's readout and the true nature, trolled. That is one of the reasons that size, orientation, and location of the we, as a purchaser, come out with spec- anomaly disclosed. ifications which you might think are overly demanding or are different than A performance consideration needing someone else would expect. Who do you attention concerns the inspection of shapes feel should try to get standardization other than rounds. There is a need to within the steel industry for a product of inspect the corners of square billets as this sort? critically as the faces. Conversely, there is an equal need to inspect the Answer (Mr. Moyer): I would answer that faces of hexagonal bars to the same degree by saying that the customer pushes the as the corners. producer into standardization. I say that because the status of standardization in Obviously each of the problem areas ultrasonics, for example, is much further described earlier suggests a future need. advanced; in other words, the steel Accordingly, only those having a pressing industry performs inspections to cus- urgency will be repeated in this section tomers' specifications at least 100 times devoted to future directions. The great- more often in ultrasonics than we do in est need is a reliable eddy current eddy current. That is because of customer instrument that will reliably detect and insistence. I must confess that this push accurately evaluate continuous, as well as for standardization has been at the intermittent defects, in hot rolling mill customer's impetus. Although eddy current product which is at 2000 °F as it leaves testing in our mills predates ultrasonics, the mil 1 personnel qualification, standardization, and purchasing specifications to quan- Farther, toward the horizon, I titative levels in eddy currents are visualize the utilization of the miracles lagging behind ultrasonics. Perhaps, we of microelectronics in signal processing take the course of least resistance; if and pattern recognition to lead us out of our customer says, you have got to do it the realm of unreliability and lack of this way or we will not buy from you, we sensitivity. Other solutions to these tend to do it. In-house, we prefer to use problems might be found in pulse techniques eddy currents because of its economic or more sophisticated phase modulation aspects. We need not be quite as rigorous and/or frequency analysis. Perhaps with standardization if we are satisfying computer techniques will provide a hard ourselves, compared to what we would be if copy printout of an eddy current test of a we were satisfying the customer. billet showing exactly where each flaw is located and how severe it is. Or, perhaps Comment (Mr. Weismantel): It would appear instead of the print-out, the computer that there is no attempt to establish a will provide guidance to the grinder so stable process between different mills. that it may remove each flaw. If we had found more standardization between the different producers, we would In summary, to answer the questions have more guidance as to how we establish posed earlier, the science of eddy current our specifications. testing has made progress in the last ten years, but there is a much longer road Comment (Mr. Moyer): It goes a little ahead. deeper than that. There is a question in

35 my mind, perhaps it does not exist in Dr. Question (Mr. Berger): You indicated one McMaster's, but I am not sure what eddy of the big new uses for eddy current currents are really sensitive to. I am testing was in sorting materials, yet one not even sure whether they respond to the of the problem areas you mentioned was the absence of metal or whether they respond difficulty in sorting materials. Could to the work hardening around an artificial you expand upon that? Are the things you

j notch. Ultrasonics is a little different. measure too close in electrical properties Maybe I am gilding the lily a bit, but at for you to make an adequate separation? least ultrasonics response has a stronger sensitivity to the reflecting area of a Answer (Mr. Moyer): Sometimes, that is calibration notch providing it is properly unfortunately true. oriented. We have a greater confidence in quantitative ultrasonic results than we do Question (Mr. Berger): Could you give us in eddy current results. an idea of what level of repeatability or accuracy or sensitivity you are looking to Comment (Mr. Booth, Bethlehem Steel Corp.): achieve the sorting specifications you I would like to respond to that. I think need? customers bring part of the trouble on themselves in that they buy from several Answer (Mr. Moyer): Well, my particular mills in irregular sequence at the lowest company makes a variety of steels, it price. Whatever is coming down the numbers in the hundreds of grades, and production line, you may have steel from some of those defeat any comparator when different mills and several companies you try to separate them; for example, going to production at such a rate there type 316 stainless and 316 low carbon is no way of correlating results of tests. stainless. There are some disastrous And, of course, AISI does not put any mixes for the automotive industry, for specifications on the magnetic properties example, and I feel it is essential that of materials at all. Consequently, unlike the mix be separated with absolute cer- Forster, who initially encouraged his tainty. customers to buy an entire melt or enough steel for a year or two's products and Comment (Mr. Hentschel): There is no calibrate the hell out of it, with this reason to have disastrous mixes any more. random material coming down the line there We manufacture a microprocessor control is no possibility of realistic testing. that will sort through frequencies and so So a part of the problem is the customer's forth on the signature of those steels. fault; he does not demand a whole lot. The question comes down to the grosser differences and not the disastrous ones. Comment (Mr. Weismantel): I have a second comment. You brought up the problem of Comment (Mr. Moyer): This is why I am how eddy current response relates to an EDM delighted this forum is assembled. These notch versus a response from a flaw of any things are being brought to light. particular nature. I look at the calibration of an instrument as a control to Comment (Mr. Bugden): I think the point attain uniform inspection sensitivity. It Bob McMaster made is pertinent to this, as is our responsibility as a purchaser to far as all the variations, not only in try to determine what that response level chemistry but in processing of various or rejection level should be, relative to steels. Certainly since we are talking the types of flaws we think are most about calibration, I think we can see how damaging to us. difficult it is. I would say that it is possible to sort mixes, but it is difficult Comment (Mr. Moyer): I am delighted you in stainless steel or alloys to calibrate are assuming that responsibility. samples, and carry them over.

Comment (Mr. Weismantel): The point I am Comment (Mr. Hentschel): I want to add to | trying to make is we do like to see the point that Dr. McMaster made. In standardization in the calibration of the Europe, they did respond in the manu- equipment. If there is one area the steel facturing processes; they would be willing industry has a lot of standardization in, to change the process to allow an optimum it is in their magnetic analysis equipment. set-up and rearrange the manufacturing to That is one thing I have found true across facilitate testing. When you try to the field; but beyond that, standard- suggest it here, they think you are crazy. ization stops.

36 Comment (Mr. Moyer): Sometimes the suggestions that are made are very ex- pensive. For example, a customer suggested that testing occur at a given intermediate size; that means we would have to interrupt a hot rolling cycle to get it at that size, provide a surface sufficient to accept the test, test it, and then re- introduce it to the hot mill.

Comment (Mr. Hentschel): In the automotive industry, there are examples where specs for the part manufactured do take testing into account. It is beginning to get better.

Question (Dr. Taylor): The Japanese are supposedly rather advanced in automatic steel production, at least that is what we read in the paper. Have they generally used eddy current testing in their steel mills?

Answer (Mr. Moyer): I cannot answer, I am sorry. I am not familiar with their techniques.

Question (Mr. Weismantel): With regard to the accuracy of calibration, you mention the lack of correlation between artificial defects and real defects. This is quite understandable. But how well does one do when comparing the response from two or more artificial defects, presumably made identically. Are these reproducible?

Answer (Mr. Moyer): We have not found them to be as reproducible as we would like. Unfortunately, the most recent reproducible artificial defect that we have found has been a hole completely through the wall of a tube, which pys- chologically is very unacceptable to a customer. But a lot of it has to do with what we said earlier, what are eddy currents sensitive to? It depends on how you manufacture the defect and how you standardize those processes, really, to make artificial defects reproducible. r

P e

i t e a .

National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy

Current Nondestructive Testing held at NBS, Gaithersburg , MD, November 3-4, 1977. Issued January 1981

EDDY CURRENT STANDARDS IN NONFERROUS METALS

Carlton E. Burley Reynolds Metals Company Richmond, VA 23261

This presentation is not intended to Let us examine how we use electrical represent the aluminum industry or any conductivity and how standards are related specific company, but to give some to this parameter. We use quantitative, personal thoughts based on 20 years of as well as qualitative, measurements of experience in NDT activities. conductivity.

When I first started my involvement Figure 1 shows the range of in NDT, being trained in physics and elec- electrical conductivity for some cast tronics, one thing that puzzled me was the aluminum alloys. In all cases, the emphasis on ultrasonics and radiography conductivity extends over a considerable and the exclusion of eddy currents for range. The as-cast condition is shown as defect identification. Soon I learned AC, annealed material as 0- temper and, in that it is difficult to convince metal- some cases, intermediate tempers are lurgists and quality controllers that an given. Notice that there is usually an eddy current "trace" has meaning. overlap among alloys.

In many cases, eddy current defect inspection has been oversold. It is not difficult to find surface cracks and surface scratches, but too often such imperfections mask the more serious problems of internal discontinuities and lead to excessive rejection rates. To avoid this, the sensitivity is reduced and everything passes inspection.

Eddy current inspection for material defects, in my opinion, requires techniques and operator competences that are COHDUCTIVIIY (X IACS usually not available in the typical plant. Figure 1. Conductivity of several aluminum casting alloys. Most of the processes that we use today have developed from laboratory and A similar situation is shown in research investigations. This is, again, figure 2. These are conductivity ranges one of the characteristics of eddy current for several wrought aluminum alloys. technology: the people who best understand electromagnetics and the behavior of This is one factor to be considered materials are in the research laborato- in standards for sorting; we can use eddy ries. With such expertise, we are able to currents for sorting only if we know do much more with eddy currents in the something more than the conductivity. laboratory than can be practically translated to our plants. A further limitation is necessary for the heat treatable alloys. Figure 3 gives Most of the applications of eddy cur- conductivity versus strength values which rents that are used are those involving are typical of a 7XXX or 2XXX alloy. the measurement of electrical conduc- Notice that we have a cycle. We can start tivity. These tests are usually per- with the quenched material, proceed to formed manually. naturally aged, then to artif ical ly aged

39 and overaged and come back to an annealed resistance or conductivity standard. The condition. standards generally used are aluminum bars or rods measured with a Kelvin bridge. Originally, when we set up our electrical standards lab, we furnished NBS with several samples for measuremnt and have sub- 5056 sequently used them as standards. These -H3S < standards have been checked periodically

20 it by NBS or another qualified laboratory.

-T i Such calibration has been a direct current measurement. Thus, when we meet a cus-

-B18 ( tomer's specification for percent IACS, we are certifying it against a DC standard since we calibrate our eddy current in- 1 struments to such DC standards .

Our lab and plant standards are cut

CONDUCTIVITY (I IACS) from Kelvin bridge measured specimens; these are used to certify working standards which are placed on each test Figure 2. Conductivity of several wrought instrument. aluminum al loys. In many cases, we are not concerned with the absolute conductivity. In sorting, for example, only a conductivity difference may be needed. But if you need to

•*SKT measure a sample precisely, a question is: NATURALLY AGED how accurately can you measure on an eddy current instrument?

It has been our experience that an eddy current technique is accurate to 1-2 percent of the reading. If a customer requires more precise conductivity certi- CONDUCTIVITY fication (such as ofter required for an electrical conductor alloy), we would use a Kelvin bridge or equivalent DC Figure 3. Conductivity versus strength measurement. cycle of heat treatable aluminum al loys. Errors can also creep into a calibrated eddy current conductivity meter. There are many double points here. For example, assume you are using stand- For example, if the conductivity is about ards of 35 percent IACS and 50 percent midway or somewhat less than the peak at IACS for a two-point calibration. If then the artificial aging position, on which you want to measure material having conduc- side of the peak are you? Are you heading tivity of about 25 percent, it is easy to for the overaged condition or are you on have an unknown error since you have not the natural aging side? This is important verified the instrument linearity below 35 if the conductivity is to be a criterion percent. In conductivity measurements, for corrosion properties. one calibration standard should be below the lowest value you wish to read and the Thus, strength or some other mechan- second should have higher conductivity ical properties are also required in than any specimen you wish to read. For order to apply eddy currents to identify best accuracy, we recommend using two temper for heat treatable materials. standards which are relatively close together; e.g., 30 and 40 percent should be These measurements, however, are a used if you are measuring material in the good example of an area in which conduc- range of 33-36 percent conductivity. The tivity standards have application. To closer the standards bracket the sample, measure electrical conductivity, you need a the greater is the accuracy obtainable.

l ASTM B193

40 NBS should continue to provide pri- tive test; a cladding variation of 0.0001 mary reference measurements for DC conduc- inch can readily be measured. tivity and resistivity. As an expansion, a facility to provide comparison measure- As an example, figure 5 shows that by ments at 60 kHz and 100 kHz (common fre- using a higher frequency, full-scale cali- quencies used for eddy current con- bration has been reduced to three mils. ductivity meters) would be a valuable aid The calibration is very close to being to the NDT community. truly linear.

Another measurement that is related STANDARD CALIBRATION ALCLAD 7075-T6 to conductivity is the cladding thickness Machined Sample

i measurement, for example, of alclad al-

I loys. Normally, specimens are cut from the Phasemaster B 3.5 corners of plate and sheet and measured Frequency: 500 KHz optically to provide verification of cladding thickness. This is a slow process for large amounts of material and also does not provide a way to monitor or measure cladding thickness over an entire plate. x 1.5 a Eddy current phase relationships can <

| be used to measure cladding thickness on 0.5

aluminum alloys. Special probes had to be 1 1 1 I \ 2 developed . We found that with fre- Meter Reading quencies of 50 to 500 kHz we can accurately determine cladding thickness by Figure 5. Calibration curve for alclad an eddy current conductivity measurement. 7075-T6.

Figure 4 shows a calibration curve Where do you obtain standards for for alclad 2024-T3. We are using a 200 this type of measurement? Figure 6 shows kHz frequency; notice that we can spread alclad aluminum which has been machined to out a five to ten mil cladding thickness provide a calibrated step block. over the entire range of the meter. This is a zero to 50 division meter.

STANDARD CALIBRATION ALCLAD 2024-T3 Machined Sample

11.0 Phasemaster B Frequency: 200 KHz

5.0 \#

i i i i \ 0 10 20 30 10 50 Meter Reading

Figure 4. Calibration curve for alclad 2024-T3.

This range could be further extended. One feature about an instrument of this type is that you can, by zero suppression and range changing, develop a very sensi- Figure 6. Machined step block for clad ding thickness standard.

2 Dr. C. Dodd of ORNL provided valuable guidance and assistance.

41 First, it is necessary to have a rep- thickness changes. Some of the foil/paper resentative section. Initially, the mate- laminates were brought to the NBS Dimen- rial was scanned to be sure that the clad- sional Technology Section where the thick- ding was uniform. Then metal lographic ness variations were confirmed with a sectioning was used to check the inner- laser interferometer. We are confident metallic layer for the presence of dif- that our system can measure foil thickness fusion or any other metallurgical effects variations to one microinch. The several

j that could invalidate the conductivity samples that were measured at NBS are used measurement. in our laboratory and our plants for internal standards. - Once assured of a representative • sample, a series of steps of roughly one The above discussion has primarily mil each were machined from the cladding been based on measurements related to con- side. Then, a narrow slice of material was ductivity changes in aluminum and aluminum cut out of the center of the machined alloys. The eddy current instrumentation

sample and the cladding thickness measured is either amplitude or phase sensitive i metallographical ly at each step. The edge using a single or double probe config- pieces provided two standard step blocks uration. Standards are, basically, speci- j from each machined specimen. mens having known conductivities and/or known thicknesses. The above is an example of developing an in-house standard for a specific appli- The next area to be discussed is use cation. The development of universal of eddy current techniques to locate and standards for all alclad products would be define surface and internal discontinu- a major undertaking and one that we have ities. Standards required and used in not recommended. A procedure for develop- this area raise somewhat different probing such standards would probably be the lems than previously discussed. most useful activity for NBS. One area for using eddy currents is The above techniques have also been location of edge laminations in plate. As used to sort clad and unclad material such aluminum is rolled to plate gauges from as sheet, plate, and tubing. the original ingot thickness, edge laminations or roll-over can develop; even Lift-off techniques are used in the though edge trimming occurs, an edge crack evaluation of products coated with non- may be present in the final product. conductive films. For these measurements, While ultrasonic techniques are frequently standards have been produced in-house (or used, these require the plates to be re- by the equipment supplier) using optical moved from stacks and individually mea- thickness or film weight measurements. In sured. Using a small flat eddy current the range of very thin base materials, probe, cut edges of stacked plates can be i such as coated aluminum foil, the lift-off readily scanned for cracks. Standards are method gives erratic readings. The mea- required to set test sensitivity levels; surement problems are compounded by the plates with known laminations are used. need to use very high frequencies to avoid measuring thickness variations of the Another application that should be foil. more widely used is inspection of tubing

and pipe. While an ASTM procedure for j Thickness measurements of aluminum this method has been published, we find foil can be readily accomplished with eddy that the use of notches and drilled holes currents. For example, 0.0005 inch foil is often inadequate for the specific ex- is laminated to 0.0035 inch paper stock. aminations required. More frequently, Variations in the thickness of the foil standards used are materials with typical can be a cause for rejection of this prod- production defects—inadequate welds, ID uct. If there is a uniform or gradual and 0D voids , etc. change in thickness of the foil, the product is acceptable. But if there are peri- At present, there is not a large odic thickness variations of the order of amount of eddy current defect inspection 15 micro-inches or greater, the material done in the primary aluminum industry. can be rejected. One reason it is not used is the unfortunate oversell of equipment which, upon We have found that a 500 kHz eddy cur- full evaluation, proves to be not designed rent test using a probe with a small diam- or able to meet requirements. If it works eter flat coil can readily measure these with steel or copper, it will not necessarily work with aluminum.

42 An area in which I would like to see structure and chemistry are significantly increased effort is measurement of mate- different from volumetric properties. rials at high temperatures. Continuous Samples which display such properties are casting processes for both sheet and rod not used for our standards. are becoming common; we are going to need better techniques for monitoring such Question : What order of magnitude of products. Eddy current techniques have variation are you seeing in your so much potential, at least theoretically, standards, say, for the worst case? that improved detection procedures and data processing methods should have a good Answer (Mr. Burley): One or two percent. change of commercial success.

Question : If you take a general piece of Part of the problem is that we try to plate or sheet and measure conductivity use conventional techniques since we know variation over the sheet that is supposed we are limited in the amount of money we to be homogeneous, how large a variation can spend on fundamental research. Fre- do you get? quently, the choice is to work with instrument manufacturers, which requires Answer (Mr. Jones): Several percent. We full cooperation and interchange between would not be surprised with two percent. the producer and vendor. This is not When you approach the butt end or head end always possible because of proprietary of the original ingot, you are quite requirements. An active program at NBS likely to find a larger variation. should help the development of eddy current inspection devices. Question : What sort of variation do you get in the middle away from the ends? In conclusion, seminars dedicated to free exchange of information among users, Answer (Mr. Burley): Not more than 1-2 potential users, and vendors, such as dis- percent when you get to the final rolled played at this workshop, should be a good product, say quarter-inch thick plate. If stimulus to better understanding and larger variations are found, they will be utilization of eddy current techniques. due to changes in chemistry, differences in cold working, or differences in thermal Discussion treatment. My earlier figures showed these variations may be several percent

Question : You mentioned you had Kelvin IACS. bridge samples checked by NBS and that, subsequently, these were rechecked. Over Question : On production line, how do you a period of time, did the conductivity control temperature so that it does not change? produce errors far greater?

Answer (Mr. Burley): In some cases, yes. Answer (Mr. Burley): For reporting or One of the problems is that to cover the certification purposes, we measure samples complete range of conductivities, stable in the laboratory and allow them to come alloys are not always available. However, to the same temperature as the standards. most of the standards have remained con- When measurements are made in the plant, stant for many years. Wear and scratches samples may not be at the same temperature are usually the prime cause for as standards; only qualitative values can replacement. be obtained. But in most cases, you are sorting and you are not too concerned since

Question : The Kelvin bridge is a DC de- all readings are being shifted in the same termination of an AC quantity. Is there a direction. significant variation between bridge samples?

Answer (Mr. Burley): For the purposes of certification, most requirements refer to ASTM B193, which is a DC technique. Our philosophy has been to calibrate our eddy current meters against DC resistance standards. Since eddy currents measure only near surface conductivity, while DC measurements are volumetric, there could be considerable difference if surface 43

National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy Current Nondestructive Testing held at NBS, Gaithersburg, MD, November 3-4, 1977. Issued January 1981.

EDDY CURRENT INSPECTION OF GAS TURBINE ENGINES

Robert A. Betz NDT Development Pratt & Whitney Aircraft Group United Technologies Corporation East Hartford, CT 06108

All of the many capabilities of the particular part, the thinnest, and eddy current inspection method are used therefore the most critical, area is at in the development, manufacture, and the trailing edge cavity as indicated. maintenance of gas turbine engines. It The shape of the part is that of an is used for flaw detection, material and airfoil so that it has a nonconstant coating thickness measurements, material sorting and identification, metallurgical condition monitoring, and electrical conductivity measurements. It is used to inspect raw materials, parts during manufacture, and as a service routine, some are unusual; many are common to all users of the method and some are peculiar to the industry. Anything approaching a complete discussion of its applications would fill a good sized book. For my purposes here, a few examples of the kinds of appli- Figure 1. Cross section of a portion of an cation that it finds in the field of gas air-cooled turbine blade. turbines may serve to illustrate its usage. geometry on each side and from end to end. The minimum wall can occur on The performance of a gas turbine either side and at any location along engine improves as the temperature of the length of the blade, thus making an the exhaust gases increases. The ultrasonic measurement impractical. The maximum operating temperature, however, problem of testing blades was solved is limited by the turbine parts, partic- using a phase-sensitive, eddy-current ularly the first turbine blade. There instrument of the type developed by Dodd are, of course, limits to the temper- at the Oak Ridge National Laboratory. ature increases that are possible The use of a purely phase-sensitive through the development of improved instrument virtually eliminates the materials. An alternative approach, air lift-off problems that arise because of cooling, has therefore been extensively the geometry and makes it possible both developed over the past ten to fifteen to locate the minimum wall and to years. Here the blades are made with measure its thickness. The technique internal passages through which rela- has proven to be rapid, reliable, and tively cool air is circulated. Such capable of measuring the thickness schemes require that the wall thickness within ± 0.002 inches. be measured because the blade wall should be as thin as possible for most Another application is the measure- efficient cooling, but, since the blade ment of a multilayer coating. An experi- is highly stressed, too thin a wall can mental coating system using three layers lead to failure. is shown in figure 2. The problem was to determine if each layer fell within The cross-section of an air-cooled the prescribed thickness range. The blade is shown in figure 1. In this first layer, applied directly to a nonmagnetic nickel-based alloy, is very high in nickel and therefore magnetic. occur. This overheating results in The middle layer, being a mixture of the material anomalies. Depending on the nonconductive top layer and the magnetic conditions that generated the anomaly, a first layer, is less strongly magnetic. number of undesirable metallurgical The properties of the various layers changes can occur. In the worst case, suggested that an eddy current test the area includes both retempered and might be used to make the measurements. rehardened material while in the simplest Because the coating system is a complex case only a residual stress field one, an impedance analysis type of results. Because these bearings operate instrument with a storage oscilloscope at very high stress levels, any of these readout was used to study the problem. conditions can result in a premature Figure 3 shows the impedance failure. Fortunately, all of changes in metallurgical structure result in a THIRD LAYER- CERAMIC local change in the permeability of the (NONCONOUCTIVE) material so they can all be detected

MIXTURE OF 1ST I with an eddy current test. Further, 3RD LATER MATERIAL'. (MAGNETIC) each condition has its own characteristic FIRST LATER HIGH NICKEL MATERIAL response by which the eddy current test (MAGNETIC) can identify the condition that is present. Figure 2. Three layer coating. Up to this point, we have been discussing manufacturing inspections. However, eddy current methods are also widely used for service inspections. In FIRST LAYEU fact, the majority of the flaw detection WAX THCC/JESS BASE (MATERIAL applications are in this area.

Gas turbine engines, as with any rotating machine, are subject to vi-

FlfST LAYER IMA*. bration and the resulting fatigue damage. For most parts, fluorescent penetrant inspection is used to detect nevr l THIRD LflYEftS lY»ft-»t this damage, but there are some cases SECOwD L Ay Eft miu. where this method is not satisfactory.

FlSiX, SECOND! One of these is in the root of fan THICO LAYERS blades where vibration gives rise to

-FIBS I f SECOMD LAYEti MliO. THIRD LAYER MO*. both fatigue damage and a mild surface BOUVDED AREAS A6E ACCEPTABLE ZONEC galling. The latter condition interferes FOR THE. COATIUC COmEluATIOWS INDICATED. with penetrant inspection because it tends to close the surface opening of the damage. An eddy current test is therefore used on these parts. Specially Figure 3. Impedance plane response for contoured probes are used to maintain multilayer coating. coil position and alignment because the area of interest is adjacent to a fillet plane relationships of various thickness radius as shown in figure 4. combinations of the three layers. The areas defined by these maximum and A major advantage of the eddy current minimum points represent an acceptable method in service inspection is its adapt- thickness for each layer. Since the ability to remote inspections. By using areas do not overlap, the acceptability some ingenuity, it is often possible to of the thickness of any layer can be make special probes which can be used to determined, provided that the lower perform internal inspections without layers are within their required thick- engine disassembly. This kind of applica- ness range. tion does not always allow maximum sensitivity to be obtained, but where Gas turbine engines use high qual- adequate sensitivity can be obtained, it ity bearings which are required to have saves the considerable cost of teardown a long life; any possible premature and rebuilding. failure is cause for concern. During grinding of the bearing races, localized As the examples cited may have overheating of the surface can sometimes indicated, there are a wide variety of

46 .

The checking of aluminum alloys for hardness requires good conductivity standards. This is an area that is currently being studied and that is to be presented in more detail by others. There seems to be little reason why such a program should not eventually be successful

How one establishes uniform reference standards for other types of tests is not so apparent. Consider coating FATIGUE CRACK thickness measurements as an example. LOCATION If only relatively few (say a couple of dozen) combinations of base material and coating need to be considered, the problem would probably be manageable. Figure 4. Fatigue crack inspection of fan Unfortunately, this is not the case. In blade roots. a large company, there may be as many as 30 or 40 base material -coating com- eddy current applications in the industry, binations that could require measurement, and to cover this range, a considerable exclusive of nonconducti ve coatings. diversity of equipment is required. Throughout the country, the number of While it would be nice to have one piece combinations must be gigantic. Then, of equipment that would be all things to there are always the unusual cases to be all tests, this does not appear to be considered, such as the three layer very likely. For the most part, there coating discussed earlier. does seem to be commercially available equipment that can solve those problems Even more difficult from the standards that lend themselves to the eddy current point of view are tests such as the approach; it is just that the greater blade wall thickness measurement that the number of jobs to be done, the has been discussed. Here, the only larger the number of different instru- practical calibration is an indirect one ments one must have available. The fact because one cannot make a direct mechanical that so many jobs can be done should not measurement of the reference masters. be taken as meaning that new equipment To do so requires removal of the opposite developments are not needed. With new wall which changes the eddy current or improved equipment and techniques, it response. To calibrate this inspection, may be possible to find more appli- three blades were chosen that gave high, cations or to significantly improve low, and midrange response. Using these those that are already in use. Improved to establish a uniform instrument set- sensitivity to subsurface flaws, for up, enough parts were sectioned and example, would be a welcome improvement mechanically measured to establish a in certain applications. For improved calibration curve that related actual capabilities, however, most users must wall thickness and eddy current instrument depend on the equipment manufacturers; meter reading. few users are in a position to develop new instrumentation. Even when a user Reference standards for flaw detec- does have the capability to make new tion could well be a fertile field of equipment developments, the chances of investigation. Over the years, users such equipment being used outside his and manufacturers of eddy current own facility are slim unless commercial equipment have come up with a wide exploitation follows. variety of methods for calibrating equipment to do flaw detection. These The area of reference standards is have included round file notches, drilled an extremely large one. The desirability holes, electrical-discharge machined of having traceable standards is quite (EDM) notches, machined rectangular obvious, but the problems associated notches, and machined "V" notches. with such a task appear formidable. The While any of these can be used to estab- way appears relatively easy in only a lish a repeatable machine set-up level, few cases. Let us consider the reference there would seem to be considerable standards required for a few typical question about how they relate to one

appl i cations. another and to real flaws. In physical 47 appearance, an EDM notch, especially a electrode would be curved just like the narrow one, superficially appears to be tube, and then he moved it over and made a better simulator of a natural crack the notch. He said, "I do that all the than does a drilled hole. But how time when I am making ultrasonic notches, significant is a physical similarity? A and it does not cause them any trouble." question that has been raised and never I found tremendous signals from places fully answered is whether an artificial where he had burnished the electrode. If

standard can truly simulate a real you are not aware of this effect you may I crack. Cracks are the result of stresses think the signals are coming from the within the material, and there is notch of the standard; they are not, usually some residual stress field they are really coming from the burnished remaining that can modify the eddy place. If you use EDM notches as standards current response. How true is this for small notches, be careful when they idea, and is it true for all types of do not appear to be consistent. It cracks in all materials? These are only apparently has something to do with the a few of the dozens of questions that conditions under which the notches were come to mind when considering this made, the oil, temperature, etc., and do particular aspect of eddy current not let them burnish the electrodes. standards.

Question : How do you validate your The eddy current inspection method measurements for the case of the multi- has become a major tool for the resolution layer laminate you talked about? of problems not amenable to solution by other nondestructive testing methods. Answer (Mr. Betz): As I said, we did it Its application in research and develop- as a feasibility study. What we did to ment programs and in manufacturing and get the data that we obtained was have service applications is essential for our people make us a maximum and minimum any well-rounded nondestructive testing coating on the base material, and then a program. Advancing gas turbine technology maximum and minimum second layer to get will require continuing development of the proper combinations, and then we inspection techniques, equipment, and went to a nonconducti ve shim stock for standards. the third layer.

Comment (Mr. Weismantel): I think one of our problems in the NDE area is that we keep trying to make notches to represent the flaw we are trying to find. The purpose of an artificial defect is to make a reproducible condition so some facility on the other side of the country can set up that particular condition, and essentially work to the same sensitivity or a similar sensitivity. I do not think that we will ever get to the point where we will be able to use a notch to represent the flaw you are looking at, because flaws vary so much. The purpose of the standard is not to represent the flaw, but to bring you to a point where you can find the flaws that you have shown. This you can do under certain conditions.

Comment (Mr. Brown): I am not sure whether to make these comments now or later when we get into the nuclear area.

But, I wanted to pass on some experi- ences I have had with EDM notches. They are not all alike. I had some EDM notches made and the person who was making them brought the electrode down on the tube and moved it back and forth a little bit so that the end of the

48 National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy

Current Nondestructive Testing held at NBS, Gai thersburg , MD, November 3-4, 1977. Issued January 1981.

EDDY CURRENT EXAMINATION IN THE NUCLEAR INDUSTRY

Allen E. Wehrmei ster Babcock & Wilcox Company Lynchburg Research Center Lynchburg, VA 24505

This paper discusses eddy current Tubes installed in steam generators testing in the nuclear industry. Almost are subjected to a more complicated test all eddy current inspection at Babcock & environment. Field inspection of these Wilcox (B&W) is for tubing; tubing specifi- tubes is the primary purpose for my cally for nuclear steam generators. In- participation in this NBS workshop. Two spection is performed during tube manu- types of steam generator concepts are facture and after installation in steam widely used, the recirculating steam generators. generator (RSG) or U-bend generator and the B&W designed once- through steam An automatic shop tube inspection generator (OTSG). The OTSG has all station is shown in figure 1. Straight straight tubes, no U-bends. The test lengths of tubing are inspected for problems encountered in these generators anomalies formed during the manufacturing can be similar. I will discuss some of process. The types of tube anomalies are the inspection problems encountered, how predictable and readily detected with the some have been overcome, and others that eddy current method. The consistency of still require a solution. These problems daily shop operation yields a highly have a significant adverse impact on the reliable test system. reliability of eddy current examinations of install pri tuhpc

Figure 1. Automatic eddy current inspection station for steam generator tubes.

49 What is an OTSG test environment? other test frequencies for more informa- The OTSG is about 60 feet high and con- tion. Multi-frequency examination is used tains 15,500 Inconel 600 tubes (see to validate anomalies and perform flaw fig. 2). Superheated water (referred to characterization. Leaks between the pri- as primary side water) enters the top of mary and secondary sides are of primary the generator and exits at the bottom. concern, but eddy current examinations do The secondary side water (on the outside not detect leaks. Leaking tubes are iden- of the tube) enters at the bottom and con- tified with hydrostatic tests. Eddy cur- verts to steam, exiting at the top. There rent examination detects tube anomalies are 15 tube support plates located along which may or may not have leak potential. the length of the generator. The supports are made of 1-1/2 inch thick carbon steel Any phenomenon that interferes with plate. Each tube passes through each the flaw signal shape or orientation support. affects the ability of an analyzer to interpret the data. The support plate produces an eddy current signal pattern like a horizontal figure eight. The tube region within ± 1/2 inch of each edge of a support plate is subject to the possibility of a flaw signal mixing with the tube support signal. For each tube sup- 15 500 lakes port plate, therefore, about 2 inches of 1(4 Kills [lllll lll|lk| tube is masked by an interfering tube sup-

,CI[ll!IC OIKtil port plate signal. That represents 32 IF 2 140.000 Ft. SlrtlM liu il Tikis inches out of 56 feet of tubing; and as it MCE TIMItl STF.il SdEIITII S till ll 1 Flllklll Flllll turns out, these areas are the most critical regions in the generator. The instrument on the right of figure 5 is a computer system that was designed to eliminate the effects of the tube support signals during analysis. This signal processing makes the signal look as though it were from free and clear tubing.

Figure 6 illustrates the computer signal processing concept. When a differential eddy current coil system detects Figure 2. Schematic of Once Through a crack in free and clear tubing, a clas-

Steam Generator (OTSG). Vertical sical flaw signal 1 is generated. When tubes are supported by 15 horizontal one edge of a support plate is detected, plates. Two OTSG's are in each one half of the horizontal figure-eight

nuclear steam system. pattern is generated 2 . When a crack is at the edge of a support plate, a distor- The tube is inspected for general ted tube support signal (or a distorted wall thinning as a probe is driven into flaw signal) is generated 3 . Subtracting the steam generator. Examination for dis- the tube support signal from the distorted crete flaws is made while the probe is signal results in a classical flaw signal being withdrawn. The probe drive and ma- that can be interpreted. nipulator system are sketched in figure 3. The eddy current signals are recorded on Figures 7 and 8 are examples of magnetic tape and on a strip chart for actual inspection signals before and after post analysis. B&W uses a test frequency signal processing. The resultant indi- which produces about one standard depth of cations are classical flaw signals from eddy current penetration in the tube wall. the outer surface of the tube. This anal- Typical signals at this test frequency are ysis is not clear from the distorted sup- displayed as shown in figure 4 (from arti- port plate signal deviation alone. We ficial flaws). A range of signal orien- must analyze that deviation and judge its tations (phase angles) are used to significance. When support plate signals establish flaw through-wall penetration. are distorted, they represent a deviation from normal, something detected. Unless The data are taken to a data analy- the signal deviation is studied and its sis center for post-test review. When cause established, we do not really know "flaw- like" signals are detected, the what has been detected. questionable region is examined again at 50 MASTER

Pivot Locator

Figure 3. Master/Slave probe manipulator concept. Probe position is verified with a television system prior to inserting probe in tube. The Master template and eddy current instruments are remote from the OTSG.

63.5% 10.8% t

VERTICAL HORIZONTAL

STRIP CHART

Figure 4. Oscilloscope and strip chart tracings of artificial anomalies 10.8%, 36.5%, 63.5%, and 100% of the tube wall thickness.

51 Figure 5. The first Computer Eddy Current Analyzer (CECA-1) shown during field post-analysis.

CRACK

EDDY CURRENT PROBE

SUPPORT PLATE

CD "N CRACK AT EDGE OF SUPPORT PLATE (D CD CD V Ho \f

Figure 6. Illustration of the computer signal processing concept.

52 IISTIITEI UPPER TIRE SHEET SIGNAL

DISTORTED SUPPORT PLATE SIGNAL

NORMAL UIS SIGNAL (REFERENCE) 1 ANALYZED INDICATION ANALYZED SIGNAL 40-50% O.D.

Figure 7. A distorted tube support plate Figure 8. A distorted upper tube sheet signal and the resultant flaw signal signal, a reference tube sheet signal, after subtracting a good support plate and the processed resultant signal. reference signal. Tracings of actual Tracings of actual field data. field data.

For example, flaws are not always the Figure 10 shows what the effects of cause of distortion or signal deviations. cold working or residual stress have on a The distorted signal in figure 9 produced flaw. Forty percent and sixty percent EDM a "chatter" indication when analyzed with notches were cold worked (rubbed with the the computer system. "Chatter" or ID shaft of a screw driver) in the labora- ripples are produced during tube manu- tory. In each sample, the phase informa- facture. It is not considered detri- tion was distorted, yielding incorrect mental, unless its signals mask all flaw information about flaw depth. The dis- signals. To eliminate ID chatter signals torted signals made the flaws appear is to improve analysis. deeper. A dent and a 100 percent through

i 60% em nitch

V, ii. LING

1 mm cue warn III COLD WORK

I 1st SUPPORT PLATE DISTORTED SI6NAL

40% EDM NOTCH V / %n. LONG ^ f WITH C8L0 WORK / NO COLD WORK

DING X 100% HOLE v 100% HOLE PLUS « DING

LOOKS LIKE 70% FLAW ANALYZED SIGNAL "CHATTER"

Figure 9. A distorted support plate signal and the processed non-flaw Figure 10. Experimental signal tracings, resultant signal. Tracings of actual illustrating the effects of "cold field data. working" on signal shape and orientation.

53 wall hole, however, appeared like a shal- It is not held by the tube support plates, low flaw. These are signal analysis (flaw it is only confined to a region. characterization) problems that develop because of external influences on real Question (Mr. Weismantel): Could you give flaws. What other mechanisms are at work? me some idea of the sizes of the defects you are seeking and what the sensitivity The signal shown in figure 11 was level is? monitored during repeated in-service inspection. Analysis indicated that a flaw Answer (Mr. Wehrmeister): We look for 20 was growing, and that it was deep. When percent throughwall indications in accord- the tube was removed from service, the ance with Reg. Guide 1.83. B&W tubing has eddy current indication was analyzed as a 0.037 inch wall. shallow. Destructive tests confirmed a shallow flaw. The "effect" (stress?) that Question (Mr. Weismantel): What is the caused the distorted information disap- length of that 20 percent throughwall? peared when the tube was removed from service. The cause of the distortion, or the Answer (Mr. Wehrmeister): No length is "effect" producing incorrect analysis, has specified. not been determined. The development of a technique to eliminate the influence of Question (Mr. Weismantel): Regardless of this "effect" is required. whether it is 20 thousandths long or 100 thousandths or ten inches?

Answer (Mr. Wehrmeister): That is right.

Question (Mr. Weismantel): The inter- I.S.t. DATA JUNE. 1977 ference you obtain from support plates, I

DISTORTED 15th S.P. gather you tried higher frequencies to null out that interference?

Answer (Mr. Wehrmeister): Higher frequencies do not null it out; they lower I.S.I. DATA IUNE, 1977 the sensitivity to outer tube surface anomal ies. DISTORTED 15th ANALYZED FLAW SIGNAL SUPPORT PLATE Question (Mr. Weismantel): You would not I EDDY CURRENT SIGNAL see the support if you went to a higher FROM REMOVED TUBE frequency. Would that give you an IN NIT CELL AUGUST. 1977 adequate inspection?

Answer (Mr. Wehrmeister): No, shallow 0D Figure 11. Tracings of actual field data discontinuity signals would be smaller and illustrating the effect of "other approach a horizontal position, thereby factors" on signal shape and making detection difficult. Higher fre- orientation. quency is used when we are looking for phase angle relationships to establish Di scussion depth.

Question (Mr. Ammirato): Are you able to Question (Mr. Weismantel): Are your sup- inspect near the tube sheet? port plates carbon steel or stainless?

Answer (Mr. Wehrmeister): Yes, the tube Answer (Mr. Wehrmeister): Carbon steel. sheet and the tube support edges are similar; you get the same kind of response. Question (Mr. Titland): Do you calibrate Each can be analyzed with the computer your computer on the support plates inside system. the steam generator, or on a model?

Question (Mr. Ammirato): How is the tube Answer (Mr. Wehrmeister): We use the sup- sealed in the support plate, compared to port plates in the generator. the tube sheets? Question (Mr. Brown): Do you use one sup- Answer (Mr. Wehrmeister): The tube is port plate chosen because you like the welded and rolled into the tube sheets. looks of it, or do you take several and average them.

54 Answer (Mr. Wehrmeister): We use those that appear most consistent, we use signals from a previous inspection.

Question (Dr. McMaster): Do you have much evidence of stress corrosion signals in these tests?

Answer (Mr. Wehrmeister): We have not established the cause of some signals, but none to date resemble what you might expect from stress corrosion.

National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy

Current Nondestructive Testing held at NBS, Gaithersburg , MD, November 3-4, 1977. Issued January 1981

EDDY CURRENT INSPECTION SYSTEMS FOR STEAM GENERATOR TUBING IN NUCLEAR POWER PLANTS

Clyde J. Denton

Zetec, Inc. Issaquah, WA 98027

1. Introduction magnitude and phase. The Zetec eddy current system provides a method to Eddy current inspection of steam read out and record the two quadrature generator tubing in commercial nuclear components of the test coil voltage power plants has evolved from a simple vector. manual effort to test two tubes during 1970, to completely automated systems 2. Discussion inspecting thousands of tubes today. The system employed to inspect Although improvements have been steam generators uses eddy currents as made in the recording and interpretation the probing media to measure variations of data, as well as in mechanical fix- in the conductivity of the tube wall turing, the basic eddy current test is being tested. still performed in the same way. An alternating voltage is impressed The following is a brief descrip- across two test coils. The magnetic tion of the eddy current test technique. field developed by current flow in the test coils causes eddy currents to flow An alternating voltage is impressed in the tube wall. The corresponding across two test coils. The magnetic magnetic field caused by eddy current field developed by current flow in the flow in the tube wall is out of phase test coils causes eddy currents to flow with the field developed by the current in the tube wall. The corresponding in the test coils. .Since these fields magnetic field caused by eddy current tend to cancel one another, the coil flow in the tube wall is out of phase voltage is decreased and phase shifted with the field developed by the current in proportion to the magnitude of eddy in the test coil. Since, these fields currents in the test piece; thus, the tend to cancel one another, the coil coil voltage is dependent on the elec- voltage is decreased in proportion to trical properties of the tube being the magnitude of eddy current flow in tested. The electrical properties which the test piece. The magnitude of the affect the flow of eddy currents are eddy currents in the test piece, thus permeability and conductivity. In non- the coil voltage, is dependent on the magnetic materials, such as Inconel and electrical properties of the tube being 300 series stainless steel, conductivity tested. The electrical properties which is usually the only significant vari- affect the flow of eddy currents are able. When the conductivity decreases permeability and conductivity. In non- due to a discontinuity in the tube wall, magnetic materials, such as Inconel and the coil voltage increases and phase 300 series stainless steel, conductivity shifts in direct relationship with the is usually the only significant variable. depth and volume of the conductivity When the effective conductivity de- change. Thus, the amount of increase in creases due to a discontinuity in the the coil voltage and the phase change is tube wall, the coil voltage increases in related to the size of the discontinuity. direct relationship with the effective conductivity change. Thus, the amount The coil voltage is sinusoidal; of increase in coil voltage is related thus, it can be described with a single to the size of the discontinuity. The vector having magnitude and phase. The coil voltage is sinusoidal; thus, it can eddy current test system used in steam be described with a single vector having generator inspection provides a method

57 for reading out the two quadrature tains high and low filters to decrease components of the test coil voltage normal plant noise. vector. The fifth instrument is used to The two test coils are electrically assist in data analysis and will be connected in opposite legs of the discussed at length later in this balancing network in the eddy current presentation. instrument. Thus, the tube is being inspected by the differential technique. The eddy current test system is The differential technique decreases the normally used in conjunction with a effects of probe motion, temperature mechanical system which positions the variations, and geometry differences. probe over the correct tube and then However, changes in nominal wall thick- inserts and withdraws the probe. The ness are not detected. insertion rate is approximately two feet per second and the withdrawal rate is The electronic portion of Zetec's one foot per second. The inspection is eddy current system contains five sep- performed during the retraction of the arate instruments. The main instrument probe. is a Zetec/Automation Industries EM-3300 Eddy Current Tester. The EM-3300 has a When the probe is inserted the continuously variable frequency from 1 proper distance, the tube number is kHz to 2.5 MHz with a digital readout to written on the strip chart and the voice indicate the operating frequency. The entry is made on the magnetic tape, then readout is accomplished on an X-Y memory the probe is retracted while the re- oscilloscope which is an integral part cording systems are operating. of the EM-3300. The instrument has X-Y outputs of plus or minus 8 volts and a When the magnetic tape is com- frequency response of DC to 100 Hz. pleted, the tape and its associated strip chart records are taken to a The output of the EM-3300 is con- remote location where they are analyzed nected to a Zetec FM-2300S Two-Channel by an ASNT-TC-1A Level IIA qualified Magnetic Tape Recorder. The tape re- interpreter. corder also has input and output capabilities of plus or minus 8 volts and DC The equipment used to analyze data to 100 Hz frequency response. In addi- consists of a tape recorder identical to tion to recording the X-Y channels, the the one used to record the data, and a tape recorder has a microphone to allow vector analyzer which more realistically tape recording tube identification and should be called an electronic pro- other pertinent data. The circuits in tractor. The "analyzer" provides a the recorder are designed to allow voice rapid means of measuring the phase angle insertion and retraction without inter- and amplitude of signals. action with the test data. The basis for phase analysis eddy The output of the FM-2300S is con- current testing can be simplified and nected to the input of a Two-Channel explained as follows. Given four Strip Chart Recorder. The strip chart concentric tightly fitting tubes as recorder has a frequency response range shown in figure 1, and starting with the from DC to 100 Hz, and it is capable of displaying a voltage input of plus or minus 8 volts. The strip chart recorder provides two functions. First, it provides a permanent record which can be scanned rapidly for initial inspection results. Secondly, since it monitors the output of the magnetic recording, it I 1 TUBE assures that the recording equipment is / 4 2 TUBES ,' //3 TUBES functioning properly. I//'/"> TUBES

PHASE I AMPLITUDE The fourth instrument is a Zetec Model I Communications Amplifier which allows voice contact between four stations with variable inputs and outputs Figure 1. Phase relationships. for all stations. The amplifier con-

58 .

probe in air, first the air vector is probe was a differential bobbin type and obtained. When the probe is inserted in the two defects not penetrating through the smallest diameter tube, eddy currents the wall are on the outside surface of flow in the tube wall with a resulting the tube. magnetic field. The resultant coil voltage vector is decreased in amplitude Figure 4 is essentially the same as and phase shifted. As the second tube Figure 3 except additional defects are is slipped over the probe area, the shown and the optimum frequency, wall vector amplitude is further decreased thickness, and conductivity are used. and phase shifted. The current flowing in the second tube is a function of the magnetic field from the coil and the magnetic field associated with the current flow in the first tube. This process continues for each tube with the current flow in each tube dependent on the current flow in the adjacent tubes. The eddy currents are not affected (in a nondefective tube) by the laminar type 100 tube to tube interface. Thus, this example can be expanded to include eddy current flow in a solid tube wall. The Figure 3. Signal phase angle comparisons current flowing in any circumferential at three frequencies. tube segment has its own distinctive phase and magnitude. The exact phase and magnitude at any point in the tube wall is dependent on the test frequency and the conductivity of the tube being tested. The eddy current test system's function is to detect and record variations in the magnitude and pattern of eddy current flow in the tube wall.

When a differential probe is passed through a tube with a defect, the signal is formed as in figure 2. Point 1 of figure 2

GOOD TUBE DIFFERENTIAL COILS " 100% DEFECT SIGNAL FORMATION

PHASE ANGLES AT 400KHz 7 >fc OD X .050 INCONEL TUBE DIFFERENTIAL COILS

Figure 2. Signal formation. Figure 4. Actual phase angles at optimum test frequency. is the signal from a good tube, point 2 shows the first coil approaching the Taking the data from figure 3 and defect, point 3 shows the coil directly plotting a calibration curve of percent centered in the defect, point 4 shows penetration of the tube wall versus the first coil leaving the defect and signal phase angle results in the data the second coil entering the defect, and presented in figure 5. point 5 shows the completion of the signal The eddy current test sytem has been shown to exhibit a long term two sigma Figure 3 shows three defects tested measurement error of plus or minus 5 at three different frequencies. The percent under actual field conditions.

59 Plotting this information versus the the tube test length is short. Thus, it calibration curves in figure 5 results is obvious that fixture positioning time in the measurement error curves shown in is relatively short. The complete data figure 6. station and fixture control center can be operated up to 150 feet from the steam generator, although shorter distances are recommended.

Discussion

Question (Dr. Mc Master): You did not mention the Russians. Are they using 40 60 80 100 120 140 160 180 200 220 your services? DEGREES

Answer (Mr. Denton): The reactors that we have been involved with are in Figure 5. Calibration curves for three Finland. What they have done is copied frequencies. the Hanford tube sheets; so it is essentially the same system. The Hanford tube sheet has a dual pipe going in and 20 i 100 KHz out. The inlets are on top, outlets on 15 - 200 KHz i.00 KHz + 10 - the bottom. The Russians merely took

5 - that and made it two different tube

0 - sheets, with an inlet and an outlet, and S the tubes still go both ways. 10

IS - MEASUREMENT ERROR Question : On the B&W steam generator, 20 - do you use a template? 20'fc 50% 100% OF WALL

Answer (Mr. Denton): Yes. There was a Figure 6. Measurement error comparisons template. I do not know if it is NRC or at three frequencies. ASTM code-- somewhere in the system it says you have to positively identify the Note that the test sensitivity tube. That sounds great when you write shown in figure 5 indicates more sen- it, but realistically when you are 100 sitivity at 100 kHz than at 400 kHz, but feet away, to check this thing, you have the measurement error curve shows twice to put on two pairs of coveralls, boots, as much error at 100 kHz. This is one gloves, etc., go inside and say, that is of the considerations which determined the right tube, all right. So the the selection of 400 kHz for flaw template and TV system eliminate that. detection in 7/8 inch and 3/4 inch x Even if you have a system that has dials .050 inch wall Inconel 600. on it and it does not really require a template, you may still put it in just The mechanical portion of the Zetec to satisfy the positive ID of the tube. eddy current test system varies to accommodate the conditions imposed by Question (Dr. Green): Doesn't one the designs of the various steam gen- person often take the data while a erator vendors. second person analyzes it?

Basically, all of the systems Answer (Mr. Denton): Yes. The data is function as follows. A template with stored on magnetic and paper tape and no tube number identification is temporarily analysis is done on the job at all. installed in the steam generator. A There are many reasons why we do it this rotatable circular fixture with a minimum way. of two independent motions is installed

over the template. The fixture operator Question : Are there any changes in the positions the probe guide tube and its characteristics of the probe due to the associated light and TV camera over the radioactive environment? tube to be inspected. The probe/pusher puller mechanism is used to insert and Answer (Mr. Denton): No. retract the probe. Test speeds of over 100 tubes per hour are achievable when

60 Comment (Mr. Wehrmeister): Water in the generator tube also does not affect the test. We inspect generators prior to draining in what is called the critical path. It costs upward of a quarter of a million dollars every day a generator is down; so you want to complete the inspection as quickly as possible. So we do inspect them while they are still full of water. b National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy

Current Nondestructive Testing held at NBS, Gai thersburg , MO, November 3-4, 1977. Issued January 1981.

USE OF ROUND ROBIN TESTS TO DETERMINE EDDY CURRENT SYSTEM PERFORMANCE

E. R. Reinhart 1 In-Service Inspection Incorporated 333 Victory Avenue South San Francisco 94080

1. Introduction

The operational availability of a determine, in subsequent inspections, the number of Pressurized Water Reactors extent of slow flaw growth to the degree (PWRs) has been reduced by the recent dis- necessary for judging the effect of re- covery of deformation and cracking in medial SG activities (change to all vola- steam generator (SG) tubing in several tile treatment, etc.). 2 operating reactors [1,2] . The more severe deformation is known as denting In response to the obvious need for and occurs in the area of the tube improved NDE technology, considerable support plates. In-service inspection, activity is being funded in NDE systems during periods of reactor shutdown, is and development for SG inspection by EPRI, presently used to detect and analyze this government agencies inspection groups, problem. To satisfy regulatory require- nuclear system steam suppliers (NSSS), ments for in-service inspection of steam and foreign groups [6,7,8]. Multifre- generators, the only inspection method quency ET, new ET probes, and ultrasonic presently used and accepted is eddy cur- systems are all in various stages of

rent testing (ET) [3] . For this inspec- development. In light of the present SG tion, differential coil bobbin type eddy problems, the utilities need to sort this current probes are inserted in the inside NDE activity into the categories of ex- diameter (ID) of the primary side of the pected near-term improvement (within six steam generator and drawn through the months) and mid-term improvements (within length of the steam generator. The 12 months). The near- term improvements present eddy current systems and tech- should have the potential to improve in- niques were evolved from technology devel- spection performance for the next series

1 oped during the early 1 960 5 [4]. In- of major in-service inspections (winter service inspection experience (training 1977) and for units that will be cleaned of inspectors, analysis of data, etc.) or where water treatment will be changed. was primarily derived from the involve- The near-term improvements would therefore ment of various groups with the Nuclear reflect system changes that are now Navy. In the past, this test has been ready for field use but require qual- very successful in detecting such pro- ification. In the area of mid-term blems as wastage and corrosion in straight improvements, systems technology should sections of steam generator tubing [5]. be identified that is amenable to However, the recent occurrences of denting accelerated effort for incorporation in the tube support area provide the in- into systems that can be used for fall spector with complex eddy current signals 1978 in-service inspection. In addition, that may mask flaws. Denting and "oval- goals for long-term R&D activity should ization" of tubing also restrict access by be defined. the inspection probe. Questions have also been raised regarding the capability of To address these needs, as well as the existing eddy current methods to to define a baseline for existing SG

formerly with Electric Power Research Institute, Palo Alto, CA 94303. 2 Figures in brackets indicate the literature references at the end of this paper.

63 inspection capability, EPRI recently SG tubing flaws and in-service inspec- initiated a technical round- robin program. tion. Reports that were of particular Conventional NDE methods and advanced value in planning the program are listed multi frequency ET systems were evaluated. as references 12 through 17. These A panel of in-service inspection special- reports gave a fairly good assessment of ists and theoretical NDE consultants the location, nature, and frequency of observed and participated in the round- defects found in present pressurized robin evaluations. water reactor (PWR) SG designs. Many of these reports were obtained from a The results of this program will be literature survey conducted by Battel le used by EPRI in two areas. Columbus Laboratories for this study.

(1) Plan long-range R&D projects Although there was considerable for the EPRI Nuclear Division. This study information on several types of SG will establish the performance level of problems, these reports lacked detailed present inspection systems and point out information on the denting problem. For areas where long-range R&D activities a better definition of this problem, an should be conducted. Most of the short- NDE specialist meeting was therefore held and mid-term development effort in this on February 24, 1977, at the offices of area will be conducted by EPRI 1 s newly EPRI. From the results of this meeting, established Steam Generator Project Office a better idea of the nature of the denting described below. problem was obtained, along with considerable information to aid in planning an (2) Define NDE performance goals NDE performance evaluation study. Selec- for the Steam Generator Project Office. ting the type and nature of the study is The Steam Generator Project Office has discussed in the next step. been established by EPRI and member utilities to rapidly develop technology 2.2 Definition of study. to alleviate serious losses in PWR plant availability caused by the previously- From the results of the NDE spe- mentioned problems associated with steam cialist meeting, the literature survey, generators. The Steam Generator Project and several additional communications, office has identified NDE development an EPRI Technical Planning Study (TPS- effort as a key item in its plan for 77-709) was selected as the vehicle for improved availability; it will therefore conducting further effort in this area. use the results of the technical planning Technical planning studies are conducted study to focus attention on the areas that by EPRI to support research and devel- have the most potential for achieving opment planning for the engineering and near-term improvement. economic feasibility of proposed technological development and/or hardware Details of planning and conducting options. Such studies permit identifica- the study are presented in the following tion of the most promising options and sections. the major technological issues which must be resolved before the initiation of a 2. Planning the Program comprehensive research program. The technical planning study approach was Determining the nature of present also selected since this represents one SG NDE inspection problems, determining of the most expedient EPRI methods the performance of present and devel- (minimal contractor negotiation time, oping NDE systems related to those streamlined review and approval process, problems, and planning remedial action etc.) for responding to studying near- were considered the major objectives of term utility problems. Major objectives initial EPRI activity in this area. From of this study were defined as: a review of past work in studying NDE system performance, conducted by EPRI and (a) First, the overall baseline others, the following steps were taken performance of present NDE systems (includ- in planning an initial study [9,10,11]. ing the operators) in response to a variety of defect types should be determined. 2.1 Definition of problem. This baseline would establish the nature and extent of future R&D activities. This first step in planning the study involved a compilation and study (b) Second, the performance of of available reports on the subjects of several new inspection methods, tech-

64 niques, and equipment, should be evaluated Each nondestructive testing system to their potential j determine for solving was evaluated by this panel in the present NDE problems. Both field proto- following manner: type as well as laboratory methods should be evaluated. I (1) General impressions. Prior to laboratory tests, details of the (c) Third, the study should be system were described by the initiated and completed as soon as possible system suppl ier. in order to transmit the information to the EPRI Steam Generator Project Office (2) Scan of known defects. The

I and other interested EPRI Nuclear Depart- panel was allowed to review the i ments for use in planning comprehensive system in operation and review R&D programs. such details as data analysis, J etc. 2.3 Organization of the study. (3) Scan of unknown defects. Data Since the nature of the inspection were then taken using a mockup problem was recognized as being very com- containing a series of simulated plex, and since EPRI needed to rapidly defective tubing. obtain as much comprehensive information as possible, a technical round-robin (4) Summary of results. Based on program, aided by theoretical and applied the results of (1), (2), and (3) NDE specialists, was selected as the above, each panel member basis for the study. It was felt that submitted his conclusions to the data from simulated in-service in- EPRI regarding the performance spections, when combined with the analysis of the NDE system under evalua- and observations of an expert review tion. panel, would provide considerable insight into the various parameters affecting A mockup containing examples of de- inspection system performance. fective tubing was essential to conducting the study and is described in the follow- 2.4 Details of the study. ing section.

This study incorporated the following B. A key element in any study of detai Is: in-service inspection performance is simulation of the inspection environment that A. NDE Evaluation Panel. Under the the NDE system "sees." In this respect, a direction of an EPRI Project Manager, a realistic mockup is essential. Since this six-man NDE technology evaluation panel study was aimed at determing current SG was used to access performance of the inspection performance as used in the various NDE systems. The panel was com- nuclear industry, all three NSSS SG posed of one NDE inspection specialist designs were considered. Looking at the from the following Nuclear Steam System present three SG designs depicted in Suppliers: Babcock & Wilcox, Combustion figures 1, 2, and 3, the task of designing Engineering, and Westinghouse. The re- a realistic mockup initially appears monu- maining three members were selected from mental. This would be true unless one the following independent groups: Battel le considers that the inspection systems to Columbus Laboratories, EPRI, and Southwest be evaluated in this study only have Research Institute. access to the tube sheet and inside surface of the heat transfer tubing. With Battel le and EPRI are independent respect to possible mockup configurations,

research laboratories whereas Southwest table 1 presents various configurations Research represents an independent in- that could be used for NDE performance service inspection group. The above team, studies or systems development. The composed of both NSSS suppliers and inde- mockup design selected for this study was pendent laboratory representatives, was configuration 3. An air transportable formed to lend credibility and objectivity mockup was designed since several of the to the project results. The above groups systems that were to be evaluated were in also supplied examples of defective tubing the laboratory or prototype stage of devel- and aided in developing a realistic test opment, and transport of these systems program. from the laboratory was not feasible. Transporting the mockup to the various NDE development laboratories was also optimun from a scheduling and economics standpoint. 65 36 in. (~91.4 cm) Primary Inlet Nozzle

Auxiliary Feedwater Inlet

Tubes

24 in. <~61.0 cm) Steam Outlet Nozzle (2)

Feedwater Spray Nozzles (32)

Figure 1. Example of U-bend steam generator design (similar to Calvert Cliffs 28 in. (~71.1 cm) Primary Outlet # 1). Nozzle (2)

Steam Outlet to Turbine Generator

Figure 3. Once-through steam generator design (similar to OCONEE #1).

Essential features of the mockup are shown in figures 4, 5, 6, and 7. The mockup was designed and built by Battelle Columbus Laboratories within one week of Upper Shell contract initiation. The mockup proved to be easy to transport and assemble at the test site, and could be used to evaluate a large number of SG tubing configurations, including tube supports and U- Tube Bundle bends. The next section describes the Tube Support samples used with the mockup to test the various systems.

Lower Shell C. Test Samples. Although the study addressed NDE problems associated with all existing NSSS steam generator designs, one Feedwater tubing size for all the test specimens was Inlet selected to simplify the logistics of the program. For the same reason, samples were all 7/8 inch nominal OD (.050 wall) Tube Plate Inconel 600 steam generator tubing of a Primary Coolant configuration typical to several SG Outlet designs, including the Westinghouse series Channel Head 51 PWR steam generators. The tubing Primary Coolant samples were either supplied by members Inlet of the NDE evaluation panel from existing Figure 2. Second example of U-bend steam test samples or fabricated specifically generator design (similar to Surry # 1). for the EPRI study. The samples included

66 .

Table 1. Steam Generator Mockup Configurations.

Mockup Configuration Purpose Key Elements in Design

1 Develop and evaluate Complex access must be simulated NDE for inspection with simulated tube bundles, of the secondary outside surface of SG, tube side of the SG (OD supports, etc. This is probably inspection). a fixed site design. The space required depends on the SG design; however, it could be of limited height for inspection at only one level

2 Develop and evaluate To evaluate total NDE system, total NDE system for including remote positioners,

inspection of SG etc. , total simulation of tube tubing from the sheet geometry and distances is primary side. required. Realistic tubing lengths and U-bend geometry is also required. This design is probably fixed site and the overall height could be considerable (>50ft). (see Reference 5).

3 Develop and evaluate Inspection environment could be basic inspection simulated with single or limited system probe and number of tubes. Tube supports instrumentation and vertical tube sheets are performance under simulated with sleeves over the simulated dynamic tubing. This design can be fixed inspection. or air transportable conditions.

4 Develop and evaluate Simple tubing holder with probe NDE system or motion controlled by simple probe

components unoei Drive mecnan i sm. i uoe u i stances laboratory can be as small as 6 inches (see conditions, with Reference 10). laboratory controlled probe motion.

specimens loaned to the program for an ongoing Battelle Col umbus/Brookhaven National Laboratory program through the courtesy of Dr. John Weeks of Brookhaven. The nature of the various samples were:

Notched Samples . These samples used electrodi scharge-machined (EDM) notches to simulate narrow crack-like defects (fig. 8). EDM notches ranged in width from 0.005 in to 0.009 in. The flaws were machined at various depths, lengths, and orientations (axial, circumferential, and at 45° to the tube axis). Two samples were also machined with the same flaw, one with work hardening, one without.

Figure 4. Air transportable steam generator mockup as shipped. 67 .

Figure 5. Mockup during assembly.

Figure 7. Mockup as seen by inspectors during simulated SG inspection, tubes

on opposite side. I

Tapered EDM CD. Notch

Max. Depth—60% of Wall

Section A-A 23"

Figure 6. Mockup during simulated SG inspection as viewed by evaluation panel Figure 8. Typical configuration of axial notch specimen.

68 .

Wastage Samples . Samples representing wastage type defects were obtained by grinding metal from the outside surface to simulate large- volume wastage type flaws (low depth, large surface area). Compound wastage, which is a large- volume (low depth, large surface area) flaw, combined with a low- volume (large depth, small surface area) flaw, was also simulated in several specimens since this condition has been seen in service (fig. 9).

Dented Samples . The dented samples consisted of the following configurations:

Minor dent . These samples contained tubing with circumferential dents ranging in diametrical restriction from 0.002 to 0.005 inch. EDM notches of various depths and length and at various locations (center and edge of dent) were machined in these samples to study the capability of NDE for detecting and sizing flaws in the presence of dents. In all these specimens, a carbon steel sleeve, simulating a tube support, was placed over the dented section. Magne- tite was also packed on the outside of the tube in the crevice Figure 9. Specimen containing wastage between the dented tube and the defect. tube support. Plastic end caps the Plastic End were glued to each end of Closures simulated tube support to retain Carbon Steel, the packed magnetite (figs. 10 Simulated Tube and 11). Support and Packed FE3O4

Major dent . These samples were 7/8 Diameter Inconel similar to the minor dent speci- 600 Tubing mens with the diametrical re- strictions increased to 0.050 inch. Figure 10. Dent specimen, supplied by BCL. Major dent with "oval ization "

The specimens listed in table Plastic Collar 2 represent the dented tube con- Joining Test Specimens figurations with the added complexity of diametrical "oval ization" in the region of the dent (figs. 12 and 13). Since the "ovalized" tube no longer permits simple slip on carbon steel

tube supports, the dented re- Aluminum Bracket gions in these samples were — Holding Specimens wrapped with slit sections of to Mockup carbon steel (figs. 14 and 15). In these specimens, EDM notches Figure 11. Dent specimen mounted in test block.

69 were placed at the beginning Table 2. Dented and Oval i zed Test Samples (one flaw), center (two flaws) (Dimensions in Inches) and end (one flaw) of the dent 0D I.D. section (figs. 13 and 14). N0°' SIZE SHOULDER CENTER CENTER MINOR MAJOR MINOR MAJOR MINOR MAJOR

(A) (B) (CJ (0) (E) ( rFl All of these samples were sup- I J

plied by Zetec, Inc. Zetec also 0.624 1.069 0.618 1.047 0.500 0.937 supplied a calibration standard 2 0.010 0.744 0.989 0.7Z4 0.968 0.620 0.862 3 0.010 0.859 0.891 0.039 0.871 0.735 in 0.766 of the type presently used 4 0.015 0.634 1.063 0.618 1.031 0.500 0.921 the nuclear industry. 5 0.015 0.7S4 0.931 0.724 0.951 0.620 0.845 6 0.015 0.359 0.391 0.839 0.851 0.735 0.747 7 0.020 0.644 1.057 0.670 1.015 0.500 0.905 Pitting . These specimens contained 8 0.020 0.764 0.973 0.724 0.933 0.620 0.828 machined conical defects designed to 9 0.020 0.375 0.875 0.839 0.839 0.735 0.735 simulate localized low volume pitting of various depths and at various locations.

Corrosion Samples . These samples contained laboratory- induced inter- granular cracks to simulate the corrosion cracks occasionally reported in operating steam generators. Section X-X

U-bend Samples . The U-bend samples contained defects all starting at the inside surface of the tubing. All defects were EDM notches, and these were located at the tangent and apex areas of the tube, at the introdose and extradose. To facilitate Section Y-Y fabrication, the EDM notches in these samples were placed in the tubes before the tubes were bent to their Figure 12. Configuration of dented and final configuration. The inner rows ovalized test samples. of a series 51 SG were the only U-bends simulated since the sharp radius of curvature of the rows represented the most difficult access 270° 0° 90° problems for U-tube inspection of this SG design. The outer rows of U- tubes, having a more gradual radius of curvature, were considered to represent an inspection situation similar to a straight section of tubing and were therefore not used in Defect this study. No. 1 ^ Defect - No. 2 X Tube Supports Drilled carbon steel plates were slipped over the Inconel tubing to simulate the influence of tube supports on the eddy current Raws inspection (fig. 16). The influence of the tube sheet was not simulated in this study since problems in this area did not appear as severe as the defect situations described above. There was also insufficient detailed information regarding problems in the tube sheet area to allow simulation. If warranted, this area may be addressed in future studies. Figure 13. Location of defects in dented and ovalized test samples correlates with figure 12. 70 (1) Zetec Incorporated. The basic Zetec single frequency system represents present industry state-of-the-art equipment and techniques. The system uses a bobbin type differential coil inspection probe. The system is rugged and simple but is highly operator dependent. A prototype system using a rotating ET probe was also evaluated. Tests of these systems were performed at the Zetec laboratories in Issaquah, Washington.

(2) Holosonics/Intercontrole. This system represents state-of-the-art French field inspection technology and utilizes a multi frequency eddy current approach to improve detection and analysis of flaws in the presence of extraneous signals (tube supports, etc.). Final data analysis is manual. Other components of the system are also significantly different from

Figure 14. Tube specimen with oval i zed present U.S. field equipment. This system dent and flaws, and split carbon steel was evaluated at the offices of Holosonics/ Intercontrole Richland, Washington. simulated tube support. ,

(3) Battelle Northwest Laboratories. This is a multi-frequency system developed from EPRI funding. The system utilizes a modified Zetec ET probe combined with a instrumentation system that acquires four frequency data during inspection and automatically analyzes data from two of the frequencies to eliminate extraneous signals from probe wobble, tube supports, etc. This system was also evaluated at the Zetec laboratory in Issaquah, Washington. This system was in the prototype development stage and this study was the first evaluation of the system Figure 15. Oval i zed tube specimen with simulated field inspection simulated tube support as tested. under conditions.

Each of the above nondestructive testing systems was evaluated by the NDE panel in the following manner:

(1) General Impressions. Prior to laboratory tests, details of the system were described by the system

suppl i er.

(2) Scan of Known Defects. The panel was allowed to review the system in operation and review such details as

data analysi s , etc.

Figure 16. Tube specimen with slip-on (3) Scan of Unknown Defects. Data were simulated tube support. then taken using a mockup containing a series of simulated defective 3. Tests of the NDE Systems tubi ng.

The following three SG NDE systems were evaluated:

71 (4) Summary of Results. Based on the results of 1, 2, and 3 above, each panel member submitted his conclusions to EPRI regarding the performance of the NDE system under evalu- U-Bend Test ation. Specimen

A mockup containing examples of defec- Straight tive tubing was essential to conducting Section the study and was previously described. Without Defect

In general, the tests associated with scanning the unknown defects progressed Specimen from simple straight tubing configurations With Dent & Flaws in to progressively more complex tubing and Tube Support flaw geometries. Straight sections of tubing containing notches, pits, and wast- Mockup age type flaws, and without tube supports, were tested first. Several of these tests Straight were then repeated with tube supports Section With Notch added to the test specimens. These sup- Defects ports were located near, at the edge, and directly over the flaws. Placement was Specimen With Dent usually based on field experience with & Flaws in real flaws. Tube Support

After the straight sections, the var- Specimen ious U-bend configurations were tested. A With IG Cracks typical test configuration is shown in figure 17. Following these tests, the Straight systems were evaluated using dented tubing Section of various configurations. It should be With Notch Defects noted that the Zetec rotating probe system was the only system capable of testing Calibration moderate or extensive dents or dents with Standard an "ovalized" configuration. Zetec was also the only system possessing a probe capable of testing U-tubes after passing through a moderate or severe dent. For this reason, fewer tests were run on the other systems.

A number of tests were also conducted to study the influence of probe design, fill factor, test frequency, and gain on Figure 17. Typical test configuration. basic single frequency system performance. aspects of flaw characteristics are not The preliminary results of these considered in this analysis. Also, the tests are discussed in a later section. ratio of incorrect defect calls versus correct defect calls are not considered at 4. Data Analysis this time. The probability of detecting a flaw, shown as the ordinate of the graph A number of approaches can be taken in figure 18 is simply the ratio of the to analyze the data. The method presented number of defects reported divided by the here is to consider two aspects of inspec- total number of defects present in the tion system performance, the probability specimen, i.e., probability of detection that a flaw of a specific through-wall Pd, at specific defect depth is: penetration will be detected, and the accuracy of sizing through-wall penetration flaws reported ^ once a flaw is detected. These are the flaws scanned basic results available from existing eddy current inspection systems. Inferences For the analysis of data in this from the inspectors, regarding the nature study, correct defect detection required of the flaws, their length, and/or other 72 the flaw to be reported in proper sequence ABC-1 ABC-4 A-3 and approximate location in the tube. In 100 - —o—o——o— -a B-4 figure 18, probability of detection is BC-2 / ABC-3 C-3 plotted as a function of percent of through-wall flaw penetration, computed as 80 the maximum depth of the flaw into the tubing wall divided by the total wall WASTAGE .2 60 thickness.

Although the number of test samples 40 per data point was in some cases relatively small, particularly when the de- SYSTEMS 20 fects are categorized by particular geom- A c SF B& C <- MF-2f, 3f etries (pits, notches, etc.), each system scanned similar defects approximately the A-2 same number of times. Resultant trends in 20 40 60 80 100 relative system performance were therefore Defect Depth, (percent of wall) considered val id. Figure 18. Systems A, B, and C detection This points to one of the difficul- of wastage. ties in establishing system performance curves for any one type of flaw geometry. In addition to detection probability, If three data points are taken at each 10 the flaw sizing capability of an NDE test percent defect depth, 30 data points are needs to be determined. One approach is needed. When additional flaw geometries to show the mean error in percent of tube are added, or when more data points are wall thickness with plus and minus two desired, the resultant number of tests and standard deviations, versus defect depth. required data analysis increases rapidly This approach is being used to analyze the to the point where a one-week test pro- data in this program and the final results gram, as conducted for each system in this will be presented in future papers and in study, becomes impractical. Future a special EPRI report. studies of this nature must therefore consider improved methods of rapidly scanning Preliminary test results from eval- and analyzing a large number of specimens uating the three systems are now consid- in a relatively short period of time, or ered in the next section. concentrate on a limited number of defect types. 5. Preliminary Results and Discussion

The above analysis has one obvious Analysis of the considerable test drawback. By presenting detection proba- data generated in this study is incomplete bility as a function of percent of at this time; however, the following pre- through-wall penetration, the influence of liminary results do indicate several in- flaw volume upon inspection results is not teresting trends regarding defect readily apparent. In this case, a very detection under a variety of test narrow axial 60 percent through-wall flaw conditions. could produce the same detection probability as a 30 percent deep wastage type 5. 1 Wastage flaw covering a large volume of the tubing wall. Since both of these defects can Figure 18 indicates the detection have a different effect on tube integrity, performance of the three systems when used the practice of reporting defect detection to inspect steam generator tubing for or sizing accuracy as a function of flaw wastage- type defects. These flaws are in depth alone could be misleading in judging straight sections of the tubing and not in the real performance of an NDE system for tube support or tube sheet areas. The some applications. Although this is the systems referred to in figure 18 and all major analysis approach followed at the subsequent figures are: present time, more comprehensive methods - of judging inspection system performance System A Zetec Inc. , single fre- are being considered and may be used in quency (SF), conventional push-pull future analysis of data. drive unit, differential coil probe (set at code sensitivity for these tests)

73 o

System B - Holosonics/Intercontrole, differences since the single frequency multiple frequency (MF), conventional results were also good. It is interesting push-pull drive unit, differential to note that the multifrequency systems coil probe did not miss one wastage flaw in all the tests. Detection of wastage- type flaws System C - Battelle Northwest Lab- represents one of the optimum uses of eddy oratories, multiple frequency (MF), current testing and its performance in conventional push-pull drive unit, this capacity has been very successful differential coil probe. over the last several years in both commercial and military applications. The On figures 18 and 19, the number af- results of figure 18 are therefore not ter the system designation (A-3) indicates unexpected and tend to confirm the valid- the number of independent tests conducted ity of the round robin test program. on a flaw of a specific defect depth. The data point presented in the figure is the 5.2 Single Frequency Parameter Study average of the set. The sensitivity used by the single Since the wastage defects present frequency system throughout the initial rather large volume flaws, the detection phase of this test program was established probability was expected to be good for using ASME Section XI guidelines and is all three systems, and it was. Detection representative of existing ISI test sensi- performance does not drop until the defect tivity. It is important to remember that depth drops below 20 percent of through- the single frequency results, presented wall penetration for the single frequency previously, were not conducted at maximum system. equipment sensitivities. To investigate the full capability of the single fre- ABC-5 ABC-5 quency system, a special series of tests 100 — were conducted in which the effects of instrument sensitivity, inspection fre- s so quency and probe fill factor were considered. The probe fill factor is defined as the ratio of the diameter of the ET probe 2 60 Single F versus Multi-F squared to the inside diameter of the tube AXIAL NOTCHES squared. SYSTEMS 40 o 225 kHz. 750 LC Probe A series of axial notches, 20, 40, 3f (100, 240, & 400 kHz), Standard Probe and 60 percent deep, were scanned by the 20 2f (200 & 400 kHz), System A team members using various test Standard Probe parameters. These narrow axial notches ABC-5 | (.005 in. width) are used to represent the 0 20 40 60 80 crack- like flaws parallel to the tube Defect Depth, (percent of wall) axis. Although these flaws are perpendicular to the flow of eddy current in the Figure 19. Maximum single frequency test tube and, therefore, in a favorable orien- sensitivity vs. standard multifrequency tation for detection, they are very low sensitivity for detection of axial volume defects and produce less response notches. than a wastage- type flaw of equal depth. Since these defects are difficult to Since the multifrequency 3 systems detect, they are ideal for system sen- produce roughly twice the amount of in- sitivity studies. spection information and at two different test frequencies per flaw per inspection After the above study, System A was scan than does the single frequency (at retested against unknown defects using 225 400 kHz), the detection probability is kHz with a special probe having an outside expected to be improved. As shown in the diameter of 0.750 inches. For this series figure, the multifrequency systems do in- of tests, axial notches of 20, 40, and 60 dicate a slightly improved detection capa- percent were used. The test results are bility for very small wastage flaws. It shown in figure 19. is difficult to ascertain system performance

3 Multi-frequency in this sense refers to a simultaneous coil excitation as distinguished from sequential tests of a single frequency system at more than one frequency.

74 The two multifrequency systems, pretation of data, and analysis of system i.e., Systems B and C, scanned the same performance. series of notches used to establish the results in figure 19 and the subsequent References probability of detection curves are also shown in figure 19. As shown, the detection probability for all three [1] Van Rooyen, D. , "PWR Steam Generator systems are identical. From this it Tubing: Corrosion Problems," Inter- appears that the single frequency system, national Symposium on Application of if operated using the appropriate test Reliability Technology to Nuclear conditions, can approach the detection Power Plants, IAEA- SM-21 8/26, Vienna, capability of the multi frequency systems Austria, October 10-14, 1977. for axial notches in straight sections of tubing remote from tube supports. [2] Steam generators to be replaced at Surry, Turkey Point Nuclear Units, Since past information indicates that Nucleonics Week, ]8, [22] (June 2, flaw sizing accuracy falls with decreased 1977). frequency, simply dropping the test frequency of a single frequency test to in- [3] U.S. Atomic Energy Commission Regu- crease flaw detection probability is not latory Guide 1.83, "Inservice In- always the solution to improved overall spection of Pressurized Water Reactor inspection performance [5]. The type of Steam Generator Tubes," June 1974. flaw expected, sizing accuracy requirements, and field experience, must all be [4] Libby, H. L. , An improved eddy cur- considered before the frequency and other rent tubing test, Report No. test parameters are selected. HW-81780, GE/AEC contract No.-AT(45 -1)-1350, Hanford Atomic Products 6. Future Effort Operation, Richland, Washington (May 7, 1964). Analysis of the considerable data generated in this study will continue. [5] Sandona, E. and Denton, C. J., Eddy- The first published report of the detailed current examination of Beznau steam results will be presented at the Second generator tubing, in Nuclear Energy

International Conference on Nondestructive Maturity . (Pergamon Press, Oxford Evaluation in the Nuclear Industry, and New York, 1975) pp. 361-376. February 13-15, 1978 (Session III, Prob- lems Areas in NDE - Steam Generator, [6] EPRI project RP403, Multi frequency February 13, 1978). eddy current system for steam generator tubing inspection, Research and As a follow-on to this study, EPRI Development Projects, EPRI Report, 67 has initiated the project RP1172, "Eval- (May 5, 1977). uation, Quantification, and Qualification

, of Steam Generator NDE Technology." This [7] Beiers, T. S. , Ivey, J. S. and - project will continue the SG NDE perfor- Romey, W. M. , "Phase I Report In- mance studies using an NDE evaluation vestigation of Parameters Influencing panel and air transportable mockup. This the Accuracy of Eddy Current Test 12-month project is expected to begin in Predictions for Nuclear Steam Gen- December 1977. erator Tubing O.D. Imperfections,"

MML-75-41 , Report for Empire State Electric Energy Res. Corp. (Project No. July 1975. , EP4-2), 18, This study would not have been possible without the excellent support of the [8] Becker, R. and Holler, P., "Results following members of the NDE evaluation of the Application of the Eddy Cur- panel: A. Wehrmeister and H. Whaley, rent Testing with a Multi frequency Babcock and Wilcox; S. Brown, Battelle Method," Paper No. lb, Eighth World Columbus Laboratories; J. Lareau, South- Conference on Nondestructive Testing, west Research Institute; H. Houserman and Cannes, France, Sept. 7, 1976. A. Sagar, Westinghouse. [9] Buchanan, R. A. and Talbot, T. F. These panel members provided exten- "Analysis of the Nondestructive Exam- sive support in the areas of program plan- ination of PVRC Plate-Weld Specimen ning, test conduction, reduction and inter- 2512J." A Report for the PVRC Sub- committee on Nondestructive Examina- Discussion tion of Materials for Pressure Compo- nents, Pressure Vessel Research Com- Question (Mr. Endler): Is carbon steel mittee, Welding Research Council, May still being used in support plates? 21, 1973. Answer (Mr. Reinhardt): I think there are

j [10] Lockheed-Georgia Company, "Technical some vendors here. They might answer Manual, Nondestructive Inspection," that, regarding their new designs. AFLC program to Determine the Reli- ability of Nondestructive Inspection Comment (Mr. Houserman): Some of those Under Actual Field and Depot Condi- are being changed. There has been a lot of tions, F41608-73-D-2850-0038, May 26, study, not only on the material, but the 1975. configurations.

[11] Reinhart, E. R. , "A Study of In- Question (Mr. Mester): You mentioned Service Ultrasonic Inspection Prac- multifrequency equipment did not do as tice for BWR Piping Welds," EPRI Spe- well in some areas, or did better in cial Report, NO-436 SR, August 1977. others. Was this the type of equipment that Hugo Libby was describing?

[12] Flora J. H. , and Brown, S. D. , "Eval- uation of the Eddy-Current Method of Answer (Mr. Reinhardt): We tested the Inspection Steam Generator Tubing," Intercontrole pulsonic system from France. Report No. BNL-NUREG-50512R, Battelle They came over and tested the mockup we Columbus Laboratory/Brookhaven Na- had. That was a commercial system, what I tional Laboratory Contract No. call a field test system. EY-76-C-02-0016, September 30, 1976. We tested the system that is being

[13] Weeks, J. R. , Materials performance developed at Battelle Northwest which I in operating pressurized water re- consider, a development- type system. It actor steam generators, Nuclear is more refined in technology than the Technology, 28, 348-355 (March 1976). intercontrole system. It attempts to do automatic analysis which is rather

[14] Fletcher W. D. , and Malinowski, D. signi f icant.

D. , "Operating experience with Westinghouse steam generators," The Intercontrole system relies Nuclear Technology, 28, 356-373, heavily on a lot of manual analysis, but (March 1976). it has been in the field several times. So we had, a field-ready multi-frequency

[15] Hare, M. G. , Steam generator tube prototype system and we also tested a failures: world experience in water- laboratory prototype multi -frequency cooled nuclear power reactors in system. 1974, Nuclear Safety, 17, [2], 231-242 (March - April 1976). But our goals in testing the two systems were different. One, to see if the

[16] Weeks, J. R. , Corrosion of steam gen- field system could be taken into the erator tubing in operating pressur- field, to solve some current problems. Our ized water reactors, in Corrosion role in the Battelle system was to help Problems in Energy Conversion and direct their R&D in further development.

Generation , C. S. Tedmon, Jr., ed. , This is the first time that they had (The Electrochemical Society, Prince- really interfaced with the new problems, ton, New Jersey 1974), pp. 322-345. and these problems are new. That is one of the reasons we conducted the round

[17] Birkle, A. J. , PWR steam generator robin for them. So there were different inservice inspections, in Third Con- goals in the two programs. ference on Periodic Inspection of

Pressurized Components . (I Mech E Conference Publications 1976-10, London, England, September 20-22, 1976) pp. 61-65.

76 . 1

National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy Current Nondestructive Testing held at NBS, Gaithersburg, MD, November 3-4, 1977. Issued January 1981

EDDY CURRENT TESTING IN GOVERNMENT

Patrick C. McEleney

U.S. Army Materials and Mechanics Research Center Arsenal Street Watertown, MA 02172

In preparing this presentation, the current method is not a major factor in the first reference I reviewed was AMC P 702- design and procurement stages of the 11, "Guide to Specifying NDT in Materiel government materials life cycle. Where the Life Cycle Applications." This handbook is eddy current method is most important is intended to serve as a guide to managers in during the service life of systems, i.e., incorporating NDT in the management of in detecting defects that have arisen in materiel at the different stages of the service. The standards used in most cases life cycle. The document serves as a basic are the parts themselves with artificial or reference to facilitate the planning, natural defects. They are used to help selection, and application of NDT for define the condition of the parts and the ensurance of satisfactory performance at end of service life so that the system can reasonable cost. This document is cur- be withdrawn from service before failure. rently being revised and will be submitted for DOD approval in the near future. The majority of papers presented at the Defense Conference dealt with main- I turned to Chapter VI of the revised tainability (in-service) inspection. One, document (draft) and under "Electromag- however, (S. Friedman, NSRDC - 1974) dealt netic (Eddy Current) Testing" found only with calibration standards and specifi- ten ASTM documents referenced. Looking cations for eddy current crack detection. further, under "General," I found two He noted an increased use of eddy current documents which pertain to eddy currents: instrumentation for' the detection and MIL-I-6870B Inspection Requirements, Non- characterization of cracks in structural destructive for Aircraft Materials and weldments. He also reported an apparent Parts; and Air Force TO-00-25-224, Welding dearth of adequate standards aimed explic- High Pressure and Cryogenic Systems (Sec- itly at structural weld crack detection tion 4). Subsequently, I found two other effectiveness. Mr. Friedman concluded that documents which should have been listed based on analysis of the experimental under Eddy Current Testing: MIL-T-15005E results, it would appear that current for 70-30 and 90-10 Copper Nickel Alloy standards and practices in eddy current Condenser and Heat Exchanger Tubing, July inspection for cracks in structural weld- 1962; and T.0. IF-111A-36. ments are generally adequate. However, some easily implemented measures should be Proceeding, I reviewed the records of taken in order to ensure greater effec- the Defense Conference on Nondestructive tiveness without any increase in false Testing, which originated in 1951 and will alarm rate. These are: in a few weeks hold its 26th conference. I noted it was not until 1967 that they 1. Calibration blocks should recognized eddy current testing as a conform as closely as possible to significant method and appointed an EC the metal under test in terms of solution of consultant to assist in the electrical conductivity; problems before the group. Prior to that, it had been included in a battery of other 2. Instrument sensitivity should be minor methods. There have been ten papers checked on a relatively wide on at the various eddy currents presented simulated crack in addition to sessions of conference. the checking it on a slit-saw cut or other, still tighter, simulated Several indicators which I wi 1 other crack; not detail here demonstrate that the eddy 77 3. Instrument sensitivity should be of calibration data during or between set with an insulating shim be- inspections. tween crack and coil to simulate maximum expected lift-off; and Further, dynamic signal injection is particularly versatile in that virtually 4. Lift-off and instrument zero any signal pattern can be generated by should be checked and reset, if properly programming the semiconductor necessary, on sound base metal memories. Once the memories are pro- structure prior to the inspec- grammed, the information is stored indef- tion. initely or until intentionally erased. Complete libraries of program data or The shortcomings of calibration of eddy memory devices can be accumulated to ac- current OD tubing inspection systems by commodate particular test conditions or passing a tube containing fabricated flaws test criteria, such as tube material, size, through the inspection coil was noted in nominal wall thickness, and flaw types. MIL-T-15005E (Copper Nickel Alloy Tubing). The memories are easily duplicated with It is difficult to obtain accurate and conventional PROM programmers at a small reproducible fabricated flaws in the cost in comparison to fabrication of reference tubing. machined flaw standards.

In recognition of this problem, the In addition, metallic tabs and passive Naval Ships Engineering Center has spon- loading coils are also attractive alterna- sored a program at Battel le-Northwest to tives to machined flaws. Metal tabs are investigate alternate means of calibrating easier to fabricate and less costly than eddy current inspection systems applicable machined flaws and can provide a good to 1 inch diameter, 0.070 inch wall copper- variety of signal patterns for calibration nickel tubing. Two alternate calibration purposes. Loading coils and tabs can be approaches have been investigated: mounted on nonmetallic forms to permit ease of handling and use. The signal pattern 1. the injection of reference amplitude and phase angle control that is signals into the electromagnetic possible with loading coils using passive field surrounding the eddy electrical components presents some in- current inspection probe by teresting possibilities for switching means of special coils and elec- arrangements to generate complete sets of tronically developed signals calibration signal patterns. Neither tabs representing flaw conditions; nor loading coils contain active circuitry and or require reference signals from the test instrument which makes them more adaptable 2. the production of reference to a variety of instrument designs. This signals by translating specially is in contrast to the signal injection prepared metallic tabs and circuitry which must be tailored to a electrically loaded coils past specific instrument design, although the the eddy current inspection readjustments necessary to accommodate probe. most instrument designs are relatively minor. They concluded that artificially generated signal patterns provide an There are a few other points which I alternative to fabricated flaws for pro- might briefly mention which are of consid- ducing eddy current calibration signals. erable interest to the Government, although The signal injection and passive signal not strictly of interest at this time. generation techniques described for OD Evolution of a defect characterization tubing inspection can provide the variety scheme for eddy current inspection has been of signal patterns necessary to confirm impeded by the lack of an adequate model of proper operation and calibration of the the magnetic field defect interaction eddy current instrument under all anti- common to all the magnetic methods of cipated inspection conditions. Signal nondestructive testing. The major ac- injection techniques can duplicate actual complishment to date has been to show that flaw patterns, and passive loading coils or the magnetic field/defect interaction can metallic tabs can closely simulate many be modeled by finite element analysis typical flaw patterns. Signal injection techniques including material nonlin- coils can be incorporated into inspection earities and complex defect geometries. It coil assemblies to permit periodic recall remains to verify the results experimentally and to examine the feasibility of

78 applying the modeling technique to include residual magnetism effects (for magnetic particle testing) and alternating current conditions (for eddy current testing).

One further point has come to my attention in the last several months. Some of my colleagues were preparing questions for the DARCOM Eddy Current Level III Certification, and in reviewing the many references, a plethora of terminology became readily apparent. As one might imagine, this creates much confusion. Although much of this is what you run into in physics texts, it was noted that some authors active in the eddy current field have their own unique terminology. It is hoped that those taking the exam have all read the appropriate reference material; if not, they are definitely in trouble.

Discussion

Question (Mr. Berger): You mentioned the biggest use of eddy current testing is related to in-service inspection problems in the field, and that the standards tend to be actual parts which are defective. Doesn't that make it very difficult to compare the calibration procedures of one maintenance depot with another?

Answer (Mr. McEleney): It doesn't help, but the results have been satisfactory.

Question (Mr. Berger): Does one person develop standards for the whole country?

Answer (Mr. McEleney): For example, government personnel examined the whole inventory of M-39, 20 mm gun tubes. There were mixed lots of improperly heat treated steels in the in-service pieces. These barrels were distributed all over the world, so they were examined in entirety, using good and bad barrels as standards. These standards were shipped around to various stations but all standards came out of one place, Watertown Arsenal. There are several other instances that are similar. Usually, there is one source of standards for a particular application.

79 ir

li I National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy Current Nondestructive Testing held at NBS, Gaithersburg, MD, November 3-4, 1977. Issued January 1981.

CURRENT STATUS AND FUTURE DIRECTIONS OF EDDY CURRENT INSTRUMENTATION

Tracy W. McFarlan Ultrasonic & Electromagnetic Equipment Magnaflux Corporation Chicago, IL 60690

During the early 1950' s, eddy current Today, the testing was introduced in this country to technology of eddy current testing has improved to satisfy market requirements for a line of such an extent we electronic instruments that could be used that can test ferrous tubing

from the I. D. , automatically test steel i for surface crack detection, sorting of billets for longitudinal seams, critical materials according to alloy detect content, hardness, geometry variations, corrosion between first and second members on aircraft structures, test I and direct conductivity determinations. wire at mill speeds, There was also a requirement for practi- test rods at elevated temperatures, automatically sort cal means of automatically testing parts large quantities of parts in automotive at production line speeds with automatic plants, find microscopic cracks on readout of test results. Out of this complicated shaped aircraft structures emerged several vacuum tube instruments that cannot be found with other tech- I with limited frequency ranges utilizing niques, and accurately readout conduc-

; meter, cathode ray tube, and recorder tivity of materials in percent IACS. readouts, that were used for manual and automatic testing of critical components. Practical test specifications and These instruments were operated on 110 V recommended practice documents have been lines, and they were considered portable developed by several industrial compa- if they could be carried around by one or nies, trade societies, and government two men. Operation required a skilled agencies that provide good test guide- man who understood the basic principals lines for many critical applications. of eddy current testing and electronics This work continues and, for the most and had the ability to interpret test part, these documents are abreast of data. events as they occur in the marketplace.

ASTM established a subcommittee of E- Personnel technical training facili- 7 to develop a glossary and to write ties are available in many sections of recommended practice documents covering our industry to educate people who are critical industrial applications, and ASNT involved with eddy current testing from provided good educational material in the the operator in the field to the super-

I first edition of the NDT handbook. visor who is ultimately responsible for establishing set-up of the test and In the 1960's, solid state circuits action based on test results. became available in the electronics industry and this caused considerable As for the future, all of us can changes in the methods of testing with look into the crystal ball and see eddy current equipment. The instruments different things. It is apparent to us were made much smaller and many of them that the most important development must could be powered with batteries. Opera- be the instrument's ultimate ability to tion of the instruments was simplified, make more and more decisions on its own. automatic gates were developed, variable The instrument will have the ability of frequency operation over much larger collecting sizeable quantities of test ranges appeared, and readouts were im- data, and, properly programmed, it proved considerably. Allied with these should have the capability of digesting developments were greatly improved me- the data and making an accept/reject chanical handling techniques that per- decision. Thus, pattern recognition, mitted reliable testing of materials of accurate mechanical control of the probe/ many different sizes and shapes at high coil with relationship to the test material, speeds. .

storage of information and accurate cali- ment calculation. If the ROI is satis- bration techniques will play an important factory, we will design, produce and part. The computer and microprocessor market standard instruments. Now what will have a considerable impact on future do we do with the instruments that are instruments' designs because of their special? Or, what should we do with control and decision-making capabilities. instrument requirements where there is We anticipate that the instrument will not a big market; a production of five or have fewer controls, readouts will be much ten instruments? To establish a product easier to interpret, and the mechanical for this market is difficult. Economi- portion of the system will become more cally, we cannot afford to develop a sophisticated. Also, operation of the standard product line. Instead, we equipment at the extreme limits of the design something on special order. But frequency range will lead to solutions of the sales price will be higher than a test problems that are unresolved today. standard product. This problem arises because most companies who have special The future needs for improved eddy requirements do not want to invest larger current testing in industry are numerous. amounts in special equipment. So you Time permits mention of only a few here. negotiate back and forth. Sometimes Testing at elevated temperatures has companies will build their own equipment always been a problem because of the and other times they are willing to go difficult requirements for cooling the ahead with the purchase of specialized probe or coil. A breakthrough in terms of types of equipment from the supplier. a new coolant or material used to make the detecting element, that will withstand Question (Mr. Brown): Do you think the high temperatures would be a big help. microprocessor will make it possible to Precision mechanical devices that can make fewer instruments that can be tai- accurately move the probe or part through lored in a wide variety of ways? the eddy current system to improve test results are badly needed. Calibration of Answer (Mr. McFarlan): Absolutely, I the eddy current testing system in terms think the microprocessors will find their of actual testing conditions must be way into the "manually operated instru- improved if we expect this method to ments" area that we have been working become a more valuable and reliable test with for crack detection. I can see tool microprocessors helping us to interpret data. Bob McMaster put his finger on Calibration techniques that approach it- -interpretation is killing us. The the actual test conditions are highly microprocessor is going to be one of the desirable. Of great importance is the ways of obtaining better interpretation. improvement of test specifications and codes. Industry technical societies and Comment (Mr. Brown): You should consider some government agencies have done a the fact that you may be able to make one commendable job to date. This work must instrument for both eddy currents and continue with greater emphasis on the ultrasonics. practical application of the eddy current system in the field. Answer (Mr. McFarlan): Right, we can design a combination system. Di scussion Question (Mr. Taylor): I was just won- Comment (Mr. Moyer): No criticism in- dering whether you might make some pre- tended, but you have your rose-colored dictions on what role the Bureau of glasses on when you say the suppliers will Standards might play in the future? come up with our needs, especially when our needs are very specialized. You will Answer (Mr. McFarlan): The NBS Conduc- come up with the needs that will guarantee tivity Program is excellent. There has Magnaflux or Magnetic Analysis or whoever been a dire need for the program for a is the designer, the maximum dollar. If long time. Beyond that program, if the we could guarantee to buy enough equip- government gets too involved in the area ment, you would come up with that need. of NDT standards, it could represent a problem. I have talked to people about Answer (Mr. McFarlan): In our company, this, and it is something we ought to after we have analyzed the marketplace bring out and talk about. needs, all of the economic factors that are involved, we make a return on invest-

82 In industry we have a competitive This is the name of the game, and we situation, and we like the idea of run- know it. There is really not much we do ning our own show. This is typically about it. If Carpenter and Magnaflux American. There is fear in our minds could enter into working relationships that the government will get too involved. for example, we are not spending a lot of They are going to control the industry, money for nothing; more could be accom- distort the standards, and try and tell pl ished. us how to do things.

Comment : I think the bulk of us could do The Bureau's contributions have been with very simple instrumentation. Now very good. They have worked well within that we are developing complex dual

i that organization. If that is an example systems, I am all for them. But, let us of how they are going to handle them- put some software in them so that a high selves in the future—great. We need school grad today can be trained to run them. it. I am not saying that is always necessary, but it should be true for the Comment (Dr. Green): I would like to majority of tests. make a comment. I know from my own

experience in working with the National Comment : I agree with an earlier comment j Bureau of Standards that everywhere they about the willingness to invest large

I go they are viewed with awe and fear in amounts of money to get something that the factories. People thought NBS was will do the job. Unfortunately, I too many going to regulate them. That is the people rely on market analysis relative least of the Bureau's intent. I think to what the worth is of developing a new that is the least intent from the present type of equipment for a new type of program. Of course, something develops application. At the start when one or and someone else takes over and the two people have an idea to go some place, intent can change. You cannot guarantee the market does not look very big. I am I that present policy will continue. I sure Foerster did not know what his total know at the present time, the Bureau does market was, other than the fact he knew not plan on being a regulatory agency. there were different applications. The problem is we sometimes defeat ourselves Answer (Mr. McFarlan): Another area we by market analysis. Once new equipment ought to talk about is the bill before becomes available, it is amazing how much Congress that nobody understands, that the market grows and becomes greater than might involve NDT. people first visualized.

Comment : I do not know what bill it is. Question (Mr. Weismantel): There is a great need in microprocessors; is Magna- Question (Mr. McFarlan): If you people flux pursuing this area? know something about it and could en-

lighten us, please do so. Answer : Yes. In the computer area I can tell you as of Monday of last week, we

Answer : My last information was that committed ourselves to an engineering bill was not coming out of committee. program to get into computerized NDT. Yes, we are in it. We have recognized Comment (Mr. McFarlan): I think we ought the possibilities for some time, but the to know what is get some back- opportunity was not right, the timing was | it and ground on it. It is conceivable that if not right until now. Now, we think it is it dies in committee now, it may show up right. Look at the trade shows and see in the near future. what is happening, the ASNT show in Detroit. It is quite obvious what is

Comment : A comment on speciality sys- happening. Computerized systems were tems. We are involved primarily in there. That kind of acitivity spurs us special systems as a supplier, not only on and motivates people to act.

! instrumentation, but material handling equipment. Oftentimes, we will develop Question (Mr. Berger): I find the | 1 the proposal stages in very complicated economics discussion interesting, cer- highly engineered special systems, and we tainly a driving factor; but I would like

| will go to, in our case, steel companies to get to some other aspects of the

I with all of this work. They will take instrumentation problem. Our problem is the information, go out for bids on it, measuring certain characteristics of the and give it to the low bidder. instrumentation. We had a meeting on

83 I ultrasonics here a few weeks ago, as you for flaws, there are problems. Standard- know, and one of the points made was that izing the sorts of things we are talking there would be no attempt by anybody to about is going to create more problems standardize how everybody's pulser should for industry than it is going to solve. work, but there should be agreement on Maybe because it is my business and I am how to measure those pulses in terms of particularly sensitive to it, but I am rise time, shape, whatever. Are there running up against ASME codes, the NRC, similar problems in regard to eddy cur- etc. It is a fact of industry. We are rent instrumentation? meeting paperwork requirements. And even though we have a better test, a more Answer (Mr. Hentschel): First, one would rapid test that gives more results, if it have to clean up the terminology. Every- does not meet these paperwork require- body uses terminology that is diverse. ments, it is not accepted. The standardization of these kinds of documenta- Question (Mr. Berger): You are saying tion hinders what we are all trying to do. terminology is now in such a bad state that we cannot agree? As an example, consider the ASME code followed by the nuclear industry. Answer (Mr. Hentschel): Yes. The termi- By the time something technical has been nology has to be agreed upon. developed and is accepted, published, and accepted by NRC, there is a period of Comment (Dr. McMaster): May I comment on about seven years. I do not want to see that? Twenty years ago when we brought us put in this position by NBS. In seven out the first edition of the handbook, we years trying to live up to standards we were voted down by two-thirds of every set today, we have got to be very careful group and they said the terminology was of what we are documenting. obscure, too theoretical, and too impractical to ever make it in the field. It Question (Mr. Berger): I think you are is better to make the word probe coil right. I do not have any problem with standard. The handbook terminology that. I go back to what I said at the should not be frozen but, should in a beginning of the meeting. The kind of paragraph say, probe coil, and then give standards or procedures we really like to seven other different ways to describe develop, beyond the conductivity measure- the same thing by inference. Then, ments, are those that contribute somehow regardless of what literature a person to better reliability or better measure- reads ten years in the future, there is a ments, or better assurance that what you chance he will be able to recognize the are measuring is there. It could be I words. I detest frozen words of the type various aspects of the instrumentation you need to have in specifications. are not important, and measurements of | the coil may not be important to a par- Comment (Mr. Brown): I hate to disagree, ticular objective. But, are there mea- but there are some words like sensitiv- surements, calibration procedures, what- ity, resolution, phase angle, that are ever, that can be made in the eddy things to be measured with numbers; and current field which would be helpful? they should have generally agreed-upon For example, the ultrasonic test blocks, j

definitions. We are way behind in this is there anything comparable in the eddy l field. And, when we get them, it will be current field?

interesting to see what they describe. i Answer (Mr. Denton): Right now, we have Comment (Mr. Richardson): I would like what we call a master standard and we to comment. I believe the only thing you have a mag tape on that standard, so we ! can describe is the linearity of the re- make ten more standards. Right now the sponse of the instrument. You can do standards are traceable to my desk or that with signal injection. your desk. If it is traceable to NBS,

j it is a lot more acceptable.

Comment : I am in the business of provid-

: problem we have is sorting ing inspection services to the nuclear Comment One | industry. It would appear that NBS is steel, grades of steel, due to slight developing instruments that are going to changes in chemistry, heat treats,

measure conductivity, that is all right, various things. If you could come up ! you should standard or some mea- have some conductivity stan- with some sort of | dards to verify the performance of surement, whether it is permeability, 1 tests. In the business of testing tubing saturation effects, whatever, to help us 1 84 sort grades of steel, bar stock, plate stock, Bethlehem, J & L, etc. , which was made today or in the middle of the winter.

Comment (Mr. Wehrmeister): To set a little parallel when you measure conductivity, as in sorting, you are looking for a quantitive number to describe what you are measuring. The standard that developed through the years is a standard of what it is you are measuring, regardless of what instrument you use, so you can calibrate it to the standard. This is where you have to draw the line. When you get defect detection, cracks, things of that sort to use parallel structure, you would be standardizing cracks. And,

I have not found one yet. I think that is where it ends. Standards end with flaw detection, but it begins in sorting appl i cations.

Comment (Mr. Brown): I know Mr. Denton makes good standards. But there is a good probability he makes them the same way each time. What if somebody at the other end of the country looks at the boiler code and starts making standards? How will they compare with the records in your desk drawer? I have an uneasy feeling about holes drilled at different speeds, different temperatures, voltages, etc. I think you could do a lot of good by surveying the range of variation that exists due to different techniques of making holes. Do it over a wide variety of conditions, fast, slow, EDM. Then the areas in which standardization is needed might be clearer.

Comment (Mr. Berger): We have done something very similar with ultrasonic reference blocks by borrowing all the types of blocks we could find. The variation was found to be about 40 percent. At least, we know what the number is.

85 Ha

t.e

* :

5 .

National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy

Current Nondestructive Testing held at NBS, Gai thersburg , MD, November 3-4, 1977. Issued January 1981

MULTIFREQUENCY EDDY CURRENT INSPECTION TECHNIQUES

Hugo L. Libby 2302 Frankfort Street Richland, WA 99352

1. Introduction Significant work in the multi frequency eddy current techniques has been Eddy current nondestructive testing done by Mr. Bob Meister and the staff at methods are electromagnetic field methods Battelle Columbus [7] using a different in which eddy currents are induced in the approach than that reported in the body of test specimen by alternating currents this paper. Their approach, a post- flowing in inspecting coils adjacent to or analysis procedure, uses an eddy current surrounding the test specimen. Test system designed around a PDP 11/40 mini- specimen conditions are monitored by computer and involves nonlinear transfor- measuring the impedances of the test coils mation of measurements, application of the (or currents and voltages of the coils) as transformed measurements to a decision they are affected by eddy current flow algorithm, and the display of results. within the specimen. The methods have been quite highly developed and they are In contrast, the technique described used widely in the metals industry for in the body of this paper is a real time the inspection of electrically conducting is 1 method. Much of this paper based upon materials and parts [I] . work performed by Battelle Northwest [8] sponsored by the Electric Power Research Most eddy current inspections are Institute. made using single frequency excitation, with the equipment permitting a selection Important aspects of the eddy current of any one of several different fre- inspection method are design and construc- quencies. However, it has been found that tion of the eddy current inspection coils, use of two or more test frequencies simul- handling or transporting of the test spec- taneously, resulting in a large number of imens, selection of test frequency or fre- degrees of freedom in the signal, can give quencies, adjustment of instrument sensi- a larger amount of information about the tivity, selection and use of test cali- test specimen than obtained using can be bration specimens or standards, choice of a single frequency Other [2,3,4,5]. signal filtering means, test specimen workers are active in multi frequency eddy speed of translation, setting of any auto- current applications, but publications are matic alarm indicators, and interpretation difficult to find. Halmshaw in an [6] of test data when required. article on potential developments in nondestructive work of R. testing mentions Especially important in the design of Becker and P. Holler in Germany and refers eddy current inspection systems is the to various papers on the subject presented size, shape, and configuration of the in- in Session of International III the 7th spection coils and the selection of in- NDT at Conference in Warsaw in 1973 and spection frequencies. Of the essence here the NDT Materials at Nijmegen Conference are the flow patterns of the eddy currents in 1974. which are affected by the factors mentioned, as well as by the presence of A multi frequency eddy current tube irregularities within the test specimens inspection developed by system has been which affect the electromagnetic proper- Intercontrole, France, but Gif-sur-Yvette, ties of the specimens. the writer has found no technical articles describing this work.

figures in brackets indicate literature references at the end of this paper.

87 The electromagnetic skin effect is 2.1 Single frequency method important in any eddy current system and especially so in the multi frequency A single frequency eddy current method. It is the variation of the skin inspection device is depicted in figure effect with frequency and the resulting 1. The single frequency generator A differences in the flow pattern of eddy currents that make it possible for the multi frequency method to produce more information about the test specimen than does the single frequency method [4].

2. Multi frequency Eddy Current Principles

Significant multi frequency eddy current inspection principles are:

a. Two or more excitation frequencies are applied simultaneously to the inspection coil assumbly. Figure 1. Single frequency eddy current b. The filtered, demodulated outputs tubing inspection device. representing the response of the system to the different excitation carrier signals supplies excitation currents to the can carry independent information as a differential internal eddy current probe result of the eddy current skin effect coils. The amplified bridge output which varies with frequency. signal is applied to an amplitude-phase detector which produces demodulated in- c. The principles of superposition phase (0°) and quadrature-phase (90°)

apply to the effects of the excitation signal outputs C x and C 2 , respectively. currents for inspection coils adjacent to Two signals are shown in the output signal

nonmagnetic (nonferrous) test specimens. plane C x versus C 2 caused as the inspection probe assembly is caused to trans- 1 d. The principles of small signal verse past two hypothetical defects pi analysis are usually applied in idealized and p 2 in sequence. Null bridge balance analyses because of the resulting simpli- conditions are assumed except when the j fications. probe coils are near the defects. It is j also assumed that small signal conditions e. The inspection of magnetic test exist and that the Lissajous figures 1 specimens is beyond the scope of the formed are straight lines in the Cj versus

present discussion because of extreme C 2 signal plane. nonl inear effects. The following equations can be

The principles of the multi frequency written to describe the two signals C x and (multiparameter) method of eddy current C 2 when parameters p! and p 2 appear at inspection have been described [3,4] from different times: three viewpoints:

auPi = Cx (1) i a. Generalization of the single fre- 1 quency phase discrimination technique. a2iPi = c 2 (2)

b. Algebraic solution of a set of a 12 p 2 (3) simultaneous equations. a22 p 2 = c 2 (4) c. Geometrical approach involving

vectors and signal space. where Cj and C 2 can vary as p x and p 2 a fluctuate. The coefficients a lx , a 21 , 12 , The interrelationships and common and a 22 are constants associated with the basis of these three viewpoints will inspection coil system, test specimen, and become apparent as the discussion general electronic signal circuits of the proceeds. instrument.

88 Equations (1) and (2) are written 2.2 Multi frequency theory assuming parameter p 2 is zero. Similarly, parameter Pi is zero for eqs. (3) and (4). It has been shown in a previous Now, assuming that the principles of paper [3] that the required independence linear algebra apply, a consequence of of signals may be obtained by adding the small signal assumption, we can com- excitation frequencies and analyzing bine these equations into two equations circuits. Simple theory indicates that as follows: for each additional frequency applied, two additional variables may be solved. How- a nPi + a i2P2 _ Ci (5) ever, more advanced signal theory indicates that, assuming the test specimen + = a 2lPl a 22P2 C 2 . (6) variables signal effects are of the minimum phase type, the number of additional This system of two simultaneous equa- variables accommodated will approach one tions has two variables or parameters per additional frequency as the number of

Pi and p 2 . The values of the coefficients frequencies are increased. a Ul . . . a 2 2 can be determined by varying p x and p 2 individually and measur- Assuming that we are considering a ing their effect upon the values of C x modest number of excitation frequencies, and C 2 , which can be observed. Equations the eq. (7) which was developed for a (5) and (6) can be expressed in matrix single frequency system can now be gen- form as eralized to handle additional frequencies by simply increasing the number of rows [A] P] = C] (7) and columns in [A] and the number of rows in P] and C]. where The equation system in the form of eqs. (5) and (6) may be solved using the a ll a 12 rules of algebra. The matrix equation [A] = (8) a21 a 22 [A] P] = C] (7)

may be solved using the rules of matrix Pi algebra: P] = (9) P2 P] = [A]"1 C] (11)

-1 and where [A] is the inverse of [A].

The phase discrimination technique is used widely in the single frequency C] = (10) system to discriminate against a single c 2 variable signal such as the one caused by p 2 in figure 1 by rotating the signal pattern by varying the reference phase The matrix [A] is considered to be a adjustment until the signal lies along the modulation relating the test specimen C 1 axis. In this position, it has no parameter P] to the demodulated output component in the C 2 axis direction (line signal C]. Excitation amplitude and Ci). Similarly, the parameter or variable relative phase, amplifier gain and p x can be discriminated against by instrument phase shifts are assumed to rotating the pattern so that its signal remain constant. The system of two eq- has no component in the C 2 direction. It uations, with known values of the co- is noted that with the circuit shown the efficients a., and the quantities c x two parameters cannot be separated at the and c 2 , can be^solved for variables p x and same time. This limitation is not one of p 2 . A third signal caused by a third principle, and can be overcome by the use variable cannot be accommodated in this of additional circuits. Also, this limi- system of equations, as it could be tation is not inherent in eqs. (5), (6), expressed as a linear combination of the (7), and (11). two existing signals and thus is not independent. The output of the single The foregoing concepts are exempli- frequency system thusly can be analyzed to fied and expanded in the diagram in figure yield only two variables or parameters. 2. The inspection coil assembly is

89 i \ — ii

PARAMETER SPACE P SIGNAL SPACE S

Figure 2. Multi frequency concept. excited by a multiple frequency generator, and the receiver has several amplitude-phase detectors whose outputs Figure 3. Multi frequency eddy current are Cj. This is followed by a tubing inspection device. transformation section which performs the inverse function indicated by eq. (11). C x to C 4 of the receiver-detector Figure 2 also applies the signal theory system can be considered to have come concepts of parameter space, signal space, from two receivers, each operating at a and estimated parameter space to the eddy different frequency. The output of the current problem. first is Cx and C 2 , and the output of the

second is C 3 and C4 . In the application of these concepts in a practical instrument, the circuits The transformation unit or section within the transformation section can be can have any one of many different forms. adjusted manually to produce the required The one shown here in a line diagram is a parameter separating function. In the direct instrumentation of the adjoint of foregoing discussion, the need to instru- the matrix [A] previously discussed. The ment the inverse of [A] is inferred. In line diagram indicates that each output fact, it is only necessarjy to instrument Pi is in general supplied by some com- adjoint [A] as it is the adjoint which bination of the inputs C-j. Provision must contains the variable separating capabil- be made to sum these input quantities in ity or decoupling capability. This is various amounts of either sign including stated algebraically as: the null quantity. The requirement here for separation of variables is that the summing circuits feeding any p. output -l _ Adjoint [A] adj [A] r/n = (12) line be adjusted so that the effect of all " |A| the remaining variables be discriminated against at that particular output line.

where I A I is the determinant of [A]. Thus, in the example shown for four variables, the summing circuits feeding It is noted that |A| is simply an output line must be adjusted so that amplitude factor. In a practical example, p x the effect of variables and are it is often desired to adjust the output p 2 , P3, p 4 eliminated at output line p 1 . Similarly, amplitudes at . . . individually. p x p the summing circuits feeding output line

p 2 must be adjusted so that the effects of

variables , and are minimized at 2.3 Multi frequency eddy current device. p 1} p 3 p 4 output line p 2 and in a likewise manner for the remaining lines. Reaching the A multi frequency eddy current desired adjustments is difficult as it is inspection device is shown by the block desirable to present to the inspection diagram in figure 3. It is shown to coil assembly the several specific emphasize the nature of a two-frequency variables, whose effects are to be mini- system, but the extension in principle to a mized, in sequence for convenience in greater number of frequencies is shown by each case. the dotted lines. The diagram has been drawn to emphasize that the signal outputs

90 .

It is emphasized that a fifth inde- ing inner wall defects, and summing pendent variable or parameter cannot be circuits adjusted at f and g to give separated using this four-parameter optimum separation of the respective sig-

; system. The system must be expanded by nals at 1 ines 10 and 1 1 adding another excitation frequency, a receiver channel, and additional trans- The circuit shown in figure 5 uses former circuits to accommodate an addi- Cartesian coordinate transformation tional parameter. devices to perform the summing functions [9]. The first parameter is discriminated against following the same procedure as 2.4 Transformation circuits described in figure 4, except that rotation of the signal pattern rotator's

Two other forms of transformation shafts , and are adjusted (<)>!, 2 3 ) sections are shown in the next two instead of potentiometers. Rotators 4

figures. Figure 4 shows one based upon and (|> 5 are next adjusted to minimize signals on lines 8 and 9 caused by two Cj 1 other parameters. Next, 4> 6 is adjusted to minimize signals caused by the fourth parameter on line 11. Although the calibration (adjustment) of the rotators for optimizing the discrimination between variables is usually done experimentally by the operator, the explanation of their function can be clarified by following signals through the circuit by an analyti- Figure 4. Parameter discrimination or cal method. This is done in the next transformation section. section. As an introduction to the analytical approach, the equations des- the method of Gauss for eliminating one cribing the operation of the rotator variable at a time [3]. This particular transformation unit are now given. circuit can discriminate against signals

caused by two variables (p x and p 2 ) and has two readout channels, one for p 3 and

one for p 4 . This circuit, in contrast to the more general arrangement of the transformation circuit in figure 3, may require interchanging of the input leads depending upon the distribution of signal components between the leads. This need for increased flexibility arises because the laws of linear algebra must be satisfied. In operation the circuit is adjusted (calibrated) to perform the desired discrimination between signals. The circuit in figure 4 is a four-parameter circuit and has the capability of Figure 5. Parameter discrimination discriminating against two parameters and section using plural transformation of separating signals from two other rotars. parameters at the two output channels 10 and 11. The signals caused by the first The transformation equations for the in figure parameter p 1 are minimized on lines 5, 6, transformation rotators shown and 7 by adjusting in sequence the summ- 5 are: ing circuits at summing junctions a, b, - = <|> sin <]) and c, while the signals caused by p x are x' x cos y (13) being applied to the inputs at lines 1, y' = + 2, 3, and 4. Next, a second parameter x sin y cos inspection probe is now caused to traverse Equations (13) and (14) are related to past two selected defects, one represent- those used in analytic geometry to des- ing outer wall defects, and one represent- cribe transformation of Cartesian coordi-

91 —

nates. However, eqs. (13) and (14) differ We find the value of to give a slightly in that they have been modified 5 null or near null output c for selected to describe the transformation of signals 2 signal S in the two-dimensional subspace flowing through the circuits rather than by equating e = 0 in eq. (20) and deter- rotation of coordinates. Given a 2 the mine that vector input signal with components x and the shaft angle which will produce a y, 0 + (dj sin <|> 4 d 2 cos <|) 4 ) null (or near null) signal at the output x 1 is obtained by equating the right hand side of eq. (13) to zero and solving for

sin 5

= tan (|) (21) COS 5 - = 5 x cos <|> y sin <|> 0 or SiS-J = * = tan (15) cos +Y (J) y <(> + d cos _^ d x sin 4 2 ^ d) = tan x ) = tan • (22) (16) 5

Finally, the output of lines 10 and Equation (16) can be applied se- 11 can be obtained by another application quentially to rotators, 1, 2, and 3 in of eqs. (13) and (14). The value of 6 figure 5 to determine the values of required to suppress the signals at the <(>!, , and to minimize the effect of 2 3 output line 11 caused by signals occupy- the first variable on the signals dj, ing the two-dimensional subspace is d , and d on lines 5, 6, and 7. Next, 2 3 determined by application of eq. (16): eqs. (13), (14), and (16) can be used to

values of and to determine the 4 5 discriminate against two more variables. These two variables produce signals which = 1 <|) 6 tan I (23) may occupy a two-dimensional subspace of the main four-dimensional signal space pq

which applies in this example. When ty lt

and are determined, the signals When the rotator <|) is set at the (t>2, 3 6 of given in eq. the vari- dj, d 2 , and d3 can be found for any input value $ (23),

signals to rotators , and . able or parameter can be observed (when ty lt 2 3 present) on line 11 (signal f 2 ). In signal The signal at e x is addition, under the proper input conditions, other signals not discrimi-

= <$> + nated against will also give indications e x d x cos 4 d 2 sin ty 4 (17) at line 11. Results of the signals caused by the three variables discrimi- and to minimize the signal at ej for a nated against will also be observed at specific input signal S (in the two- line 11. Amplitudes of these residues dimensional subspace) which has compo- depend upon the success of the dis- nents d and d we find from d lf 2 , 3 , crimination adjustments. eq. (16): d -1 l (18) = tan gl - 4 3. Application of the Theory Four-Parameter Example The signal at line a is, from eq. (14): The application of the foregoing principles is illustrated in this section

a = sin + d cos by a combined approach using eddy current di 4 2 4 (19) instrument measurements of the responses flaws and analyses The output signal at e 2 is to fabricated tubing showing the projected performance of a these e = (d sin + d cos <)> transformation section with 2 x 4 2 4 ) responses as inputs. The following

cos 4> - d sin . four-parameter example illustrates 5 3 5 (20) the application of the multi frequency principles described in the foregoing section.

92 .

3.1 Data acquisition

This example uses signal data scaled MULTIFHEQUENCY GENERATOR from Lissajous patterns obtained using a lOOKIIt 300KHi h/ test specimen of Inconel 600, 7/8 in. diameter steam generator tubing containing fabricated calibration holes. The wall thickness T is approximately 0.050 in. (1.27 mm). The calibration regions and other conditions used in this report are:

0 one 100 percent T drilled hole, 0.067 in. (1.7 mm) diameter (T Figure 6. Equipment for obtaining data equals wall thickness) for algebraic solution of two-frequency inspection method. ° one 80 percent T drilled hole,

0.078 in. (1.98 mm) diameter 100 kHz having detector outputs C x and

C 2 , and one operating at 300 kHz having ° four 20 percent T drilled holes, detector outputs C 3 and C 4 . 0.1875 in. (4.76 mm) diameter spaced at 90 degree intervals The patterns obtained by displaying around the circumference of the Ci versus C2, and C 3 versus C 4 by use of tube the multiplexer were separated on the oscilloscope screen by adjustment of dc 0 a simulated tube support made of offset controls in the respective mild steel 3/4 in. thick detector circuits. surrounding the tube. Tracings of several specific points Lissajous figures were generated for on the Lissajous pattern photographs for these tube inspection conditions using the several inspection conditions are shown equipment arrangement shown in figure 6. in figure 7. Signal components for these (Signals observed of other calibration conditions as scaled from the photographs holes in this tube section were present, are given in table 1 but results are not presented here in the ^WOBBLE interest of brevity.) The equipment 200mv comprised: (1) a two- frequency laboratory eddy differentially current system with a 4 20% T a connected internal inspection coil probe 1300KH ,>-!>' operating , simultaneously at 100 kHz and 80% T-<. 0 300 kHz; (2) a multiplexing electronic 3 C3 *• switch; and (3) a cathode ray / 4 100% T —V.'' oscilloscope. The in-phase (0°) and 3 >6 quadrature (90°) demodulated (detected) V outputs of the two carrier frequencies, 100 kHz and 300 kHz, were applied to the multiplexing switch. The purpose of the multiplexer is to time-multiplex the four Figure 7. Signal loci for five test output signal channels of the detectors so conditions at 100 kHz and 300 kHz that two Lissajous patterns, one from each obtained with multiplex system. of the two carrier frequencies, could be displayed nearly simultaneously on the cathode ray oscilloscope screen. Dwell 3.2 Objectives of calculations times of the electronic switches were about 2.5 ms, and the repetition rate was The functioning of the transformation about 10 ms, giving a display rate of unit will be shown by using the measured about 100 points per second for each instrument outputs as inputs to the Lissajous pattern. transformation unit and by calculating the settings of the rotators and the The measurement system used is equi- resulting signals in the transformation valent to that of two single frequency section. More specifically, we desire to inspection devices, one operating at calculate:

93 e .

Table 1. Components of Signals in Figure 7. Relative Signals at Detector Output Existing Si gnal

Condition Symbol C x (bj C 2 (b 2 ) C 3 (b 3 ) C 4 (b 4 )

Probe $12 17. 9 48. 3 - 6. 0 229. 8 wobbl

Support $23 358. 7 -793. 3 -262. 6 -607. 6

$24 521 . 1 -576. 0 -112. 8 -715. 7 $2 5 513. 9 -179 7 50. 7 -564. 7

100% T $32 -227 6 -101 9 -380 6 -231 1 hole $33 -246 4 -166 0 -425 3 -449. 0 .10/1 Q ^>34 1 04 U 1 CO U O 606. I

80% T §41 - 97 4 19 8 -116 0 100. 0 hole $42 -202 3 - 41 4 -367 5 30. 3 $43 - 87 2 - 33 0 - 98 9 - 13 5

4-20% T $61 -110 0 90 5 - 42 3 164. 8 4 holes $62 -232 0 79 1 -218 8 314 0 $63 -111 2 27 6 -121 4 155 3

produce minimum probe wobble signal (while a. The settings of rotators, ty lt the probe is caused to wobble) on lines 5, ! for line 5, <|) 2 for line

6, and <() for line 7. It is found in the b. The output of rotators 3 ty l 3 2 , example that the support signal is and 3 for flaw signals caused by the effectively a two-dimensional signal, 100 percent T, 80 percent T and the 4-20 occupying a two-dimensional subspace. percent T fabricated defects in the This signal is discriminated against by standard tube, given the settings of adjusting iteratively the rotators <)> 4 and rotators , and <)> for 2 3 discriminating 5 until the Lissajous pattern of this against probe wobble. signal when viewing the signals on line 9 versus those on line 10 (or lines 11 and c. The settings of rotators 4 , collapses to as nearly a straight line and to discriminate against 12) 4>5, 6 the signal as possible. Final discrimination support signals. is obtained by rotating rotator (e versus 6 x deflection) when viewing signals on line e 2 ) caused by the defects 80 percent T 9 versus those on line 10. and 4-20 percent T when and are (t> 4 5 adjusted to produce a null signal for the In this example, these adjustments signal caused by the 100 percent defect will all be calculated using the measured (wobble discriminated against, but no instrument response given in table 1. discrimination against the support signal s). We do not have the same flexibility in the calculations as we would have e. The relative amplitudes of the operating the rotators manually because peak output signals of the support (S 23 , of the large number of computations S 24 , and S 25 ) and the 100 percent T, 80 required to simulate the iterative manual percent T, and 4-20 percent T signals, all adjustments. at output 1 ine 11.

Discrimination of signals will be 3.3 Probe wobble discrimination illustrated by use of the plural signal rotator transformation unit shown in Equation (16) applies for discrimi- figure 5. In operation, the rotators nation against probe wobble: , and <|> 01, 2 3 are adjusted manually to 1 = 0.00058 tan" S (16) y Probe Wobble S 12 d 2 = -0.0035 d 3 = 0.0021 where x and y are the input signals to a specific rotator and § is the rotator di = 612.02 angle setting to discriminate against the >23 d 2 = -651.34 signal vector x, y. Using the rotator d 3 = -278.38 designations given in figure 5 and noting from table 1 that the probe variable di = 688.80 signal vector is Support '24 d 2 = -416.45 d 3 = -131.45

di = 544.33 - S 12 = 17.9 48.3 6.0 229.8 '25 d 2 = 59.70 (29) d 3 = 35.94 we have di = -173.36 - 100% T Hole >33 d 2 = 70.09 -1 >! = tan = 0.3706 = 20.34° d = -436.88 jJ448. J fa 3 (24) di = -175.31 - 80% T Hole '42 d 2 = 46.75 -1 d = -366.58 = tan - 0.2102 =11-87° 3 fa 229.8 fa (25) di = -234.61 4-20% T Holes S 62 d 2 = 12.82 d 3 = -210.53. = -1 = -1.50°. fa tan ^y.« = 0.02611 fa (26) The signals d 1} d 2 , and d3 for vectors S S S S S and 23 , 24 , 25 , 33 , 42 , S 62 are shown in figure 8 in three 2-space d The outputs of the first three rotators, projections, x versus d 2 , d 1 versus d 3 the signals d 1( d 2 , and d 3 on lines 5, 6, and d 2 versus d3 . It is noted that these and 7 in figure 5, are expressed by eq. signals have been transformed from the (13) expanded here: original values on lines 1, 2, and 3 to new values because of the discrimination - - effect of probe wobble. d 1 b x cos <(>! b 2 sin 4>i against the

- d 2 = b 2 cos 4> 2 b 3 sin 0 2 (27) SnUBCE - d = . 3 b 3 cos fa b 4 sin (|) 3

Substituting the values of cos § and sin

<|> for , and (|) in eqs. 0X, 2 3 (27) produces

d = 0.9377 b - 0.3475 b ] ] 2

d 0 = 0.9786 b„ - 0.2057 b„ (28)

0.9997 b. 0.0261 b . 4

Next, b b substituting the values of 1} 2 , b , and b of S S 3 4 the peak signals 12 , 23 , S S S S S eqs. 24 , 25 , 33 , 42 , and 62 in (28) gives the d following values for 1 , d 2 , and d 3 , the signals on lines 5, 6, and 7 in figure 5.

Figure 8. Three 2-space views of signals in 3-space.

95 3.4 Null display of signal S e = 22.37 42 ]

It is informative to determine the e = -21.28 2 display of e x versus e 2 (fig. 5), the = input to the final rotator <}> 6 when S e 99,82 62 l rotators and have been adjusted to 4 5 present a null view of the vector signal e = -112.70 2 S 33 , the 100 percent T signal. This view will show the relative disposition of the These and corresponding projections of

S S , and S which are caused the d d , and d axes are shown in vectors 33 , 42 62 l5 2 3 by the three flaw conditions 100 percent figure 9. T, 80 percent T, and 4-20 percent T, respectively. It will also serve to 3.5 Discrimination against clarify further solutions used for support signals determinations of signals at e x versus e 2 . It has been observed that the support signals can represent a two-

The values of <^ 4 and <|> 5 to result in dimensional subspace in the 3-space re- the null view of S 33 at e 1 and e 2 are maining after discrimination against the determined from eqs. (18) and (22) with effects of probe wobble. This aspect of substitution of the appropriate component the support signals can be illustrated = values of the vector signal S 33 (d x by taking the cross products of the

-173.36, d = -70.09, d = -436.88), vectors S S , and S (all 2 and 3 23 , 24 25 at lines and the previously determined values of 5, 6, and 7) and then determine the an- sin and cos gles between these cross products to 4 find how closely the three vectors lie in a plane in the 3-space. d -1 l d = tan gi (18) 4

+ d x sin 4 d 2 cos 4 =

It is found that and may have 4 5 the following pairs of values:

Pair 1 67. 99° -23.17°

Pair 2 67. 99° 156.83°

Pair 3 247. 99° -23.17°

Pair 4 247. 99° 203.17°

Figure 9. Calculated display showing The last pair, Pair gives the 4, peak signals. proper aspect of the vectors S 33 (null), S 42 , and S 62 when S 33 is chosen as the null vector and the direction of view Calculating: is from the head of the vector S 33 to the origin. The signal values at e x and = u S23 X s 24 e 2 (fig. 5) for these conditions are: V = S24 X S25 S e, = 0.00079 33 w = S23 X $25 e = -0.00396 2

96 and noting that

U • V cos y uv

U • W cos y vw IVI fwl

U • W cos *uw " TUTTwr we find that

= 2 85° YJ uv Figure 10. Input to rotator with 6 = 47° 1 null vector normal to S . YJ vw 25

= 1 88° A second choice was made for the null YJ uw vector at ex versus e 2 , based upon a small Next, an average of three vectors U, vector model of the signals at lines 5, 6, V, and W is formed by adjusting their and 7. Inspection of this model indicated component values to that corresponding to that a more reasonable choice of null the average length of U, V, and W. The vector would be one that is normal (or resulting average is denoted vector D and nearly normal) to vector S 24 and, of the relative value of its components are: course, occupying the approximate plane D = -3.1365, -12.6942, 23.2449. The containing the three support signal vector D is normal to the approximate vectors. plane containing the three support signal vectors S 2 3, S 24 , and S 25 on lines 5, 6, The vector D defines the subspace and 7 in figure 5. (plane) containing the support signals in an average kind of manner. The vector S 24 We will now find two vectors, both is determined by solving the equation normal to vector D and normal to each other to be used to calculate the settings S 24 • D = 0, (30) of rotators <|> 4 and

An attempt was made to build up the S 24 = 688.80, -416.45, -131.45 (31) first vector S 25 by assuming it to be near

S . The rotator angles and were We then assume that 25 4 (t> 5 then calculated to present a null view of S = 688.80, -131.45. the vector S 25 x D at ej versus e 2 , as we 24 q, (32) did for vector S 33 in figure 9. When this was carried through to completion, the Solving eq. (30) gives results shown in figure 10 were obtained. (Calculations of this kind are discussed q = -410.89. (33) in more detail in the next few paragraphs.) The support vectors in The vector S 24 is the first vector figure 10 have been transformed as desired needed which is normal to D. The second and they appear in a line, but the vector normal to D, and the one to be used projections of the 100 percent T, 80 per- as a null vector in the display which cent T, and 4-20 percent T signals are discriminates against the support signals separated by an angle of only 7° or less. is denoted S' and is equal to the vector This is insufficient angle difference for product: practical use. The resulting display of these flaw signals also has a reverse So 4 x D = S' phase angle aspect, that is, flaws farther from the metal surface produce signals having increasing leading phase angles. 97 It is found to be equal to

1 S = 11,219.79, 15,598.86, 10,032.38 .

We find and by again applying eqs. 5 = = (18) and (22) with d x 11.21979, d2 15.59886, and d 3 = 10.03238 to give:

° 4> 4 = 35.73° or 215.73 -800 ^00 ^So O S, °. 4)5 =-62.43° or 117.57

The underlined values are chosen as they result in the desired aspect when viewing Figure 11. Input to rotator with null the vectors in the e x versus e 2 display. vector normal to S 24 .

Again, the values of the other sig- It remains, to obtain the nals at e 1 and e 2 (fig. 5) given by eqs. discrimination against the support (17) and (20) are: signals, to rotate the signal patterns in figure 11 so that the support signals S23 ei -877 17 produce a minimum signal in the ordinate

e 2 167 44 direction, or the signal h on line 11, figure 5. This is done by adjusting

s 24 ei -802 35 rotator 4> 6 . We note that the greatest e 2 146. 20 angle spread of the support signals is 0.60°. between S 23 and S 25 , and it is The S25 ei -476. 76 angle of S 25 is e 2 92 82

S33 ei 99. 80 1 625 = tan" = 168.98 . e 2 314 09

s 42 ei 115. 02 e 2 260. 01 Assuming we need to approximately equal- ize the maximum positive and negative S62 ei 197. 95 swings of the support signal, we should e 2 128. 04 rotate the pattern in the counter- clockwise (positive) direction 0 degrees These results are shown in figure 11. The where: results can be seen in better perspective by now referring back to figure 9 wherein 6 = 180 - 168.98° - 0.4325 = 10.588°. the 100 percent T signal is viewed

"end-on," and the projections of the 80 The output of rotator 4> 6 is obtained percent T and 4-20 percent T signals are by using eqs. (13) and (14) which are seen in accordance with their particular rewritten here: orientations. Now, referring to figure - 11, we can see that we = cos <|> e sin are viewing the 100 g >! 6 2 6 percent T signal from some other = viewpoint, the one determined by the h e x sin 4> 6 + e 2 cos 4> 6 . vector S' , which presents us with one = 10.59° "edge-on" view of the support signal When <|> 6 we find that plane. The support signals are thus viewed as being very nearly in line. It S23 g -893.00 is now clear that the choice of vectors h 3.41 within the support signal plane other than S' as null vectors results in a rotation $24 g -815.56 of all signals around a line perpendicular h - 3.71 (normal) to the support plane and passing through the origin. When this pattern is s 2 5 g -485.7 rotated thusly, the projection aspects of h 3.64 the signals in the e x versus e 2 plane change greatly. The visualization of this S33 g 40.39 is aided by again referring to figure 9. h 327.08 98 S42 g = 65.29 h = 276.72

S 62 g = 171.05 1 00%T S33 h = 162.23 . 80%T S42 These values of g and h are used to produce figure 12, and it is observed that the support signals now have small components in the h (ordinate) direction, and the remaining signals have a practical orientation having the required phase 4-20%T angle direction. S„

The h signal components are:

$33 h = 327 08

S42 h = 276. 72 s„ s. $62 h = 162. 23

= SUPPORT FABRICATED FLAWS $23 h 3. 41 3 PROBE POSITIONS ONE PROBE POSITION EACH

S 24 h 3.71 Figure 13. Calculated relative values of $25 h 3.64 . signals showing discrimination against tube support signals.

Another way to explain this effect is to relate it to the theory of the multifrequency method. In this example, we have a four-variable or four-parameter system. Three variables have been "eliminated," one for probe wobble and two for the support. This leaves only one to be indicated at the output h. Of course, -800 -600 -400 -200 200 400 other variables not provided for by the system will also produce deflections in the ordinate direction.

Figure 12. Output to rotator <|> 6 with null vector normal to S24. 3.6 Simulated cross-section display of tubing [3] These h components are plotted in figure 13 at a larger scale to show the large The multif requency method promises to amount of discrimination more clearly. make it possible to perform many The amount of discrimination in this inspections not feasible with a single example is good. Although the three flaw frequency (one frequency at a time) signals in figure 12 appear to carry phase system. An example of this is shown in information, it should be emphasized that figure 14 in which a four-parameter system when the support signal occurs simultane- was used to discriminate against probe ously with a flaw signal, the two are wobble signals and to separate inner wall additive, and the horizontal (g component) and outer wall simulated flaws. Tubing of the support signal is added to the g radial position signals obtained from a component of the flaw signal. Phase angle resolver produced a simulated pattern of information is retained when the flaw the tubing specimen on an X-Y cathode ray signals appear remote from the location oscilloscope provided with capability of z of the support. axis modulation. The output of the four-parameter tubing inspection device modulated the pattern to show the relative location and relative severity of flaws opening on the outer and inner walls.

99 metal saws or by filing or machining, Grooves and other irregularities can be made by machining, A much used method for producing notches is the electric discharge machining process. Notches measuring a few thousandths of an inch wide to simulate cracks can be made by this method.

Drill holes do not represent exactly the common types of defect found in metals, but rather serve as convenient alternates by which the sensitivity and TEST SPECIMEN TYPE 304 STAINLESS general performance of the inspection STEEL 2-'< IN O 0 '* IN. WALL device can be set or measured. It is ARTIFICIAL FLAWS '.IN DIAMETER holes drilled HOLES IN OEPTHS difficult to produce SHOWN IN INCHES partially through the wall from the inside of a tube. Notches are more costly to Figure 14. Simulated display of tubing produce, but they can be made on the cross section. inside of tubing and give a much better simulation of cracks.

4. Calibration Methods and Standards The nature of the calibration problem is complex and certainly cannot The same general principles of cali- be treated adequately in this paper. bration and use of standard specimens Continuing research is being done in which apply to the single frequency eddy this area, and much more is needed. The current technique also apply to multi- development of standards for the eddy frequency techniques. However, with the current tests have lagged behind that for more general method there are added com- the other major nondestructive tests. plications resulting from the greater This must be at least in part caused by number of degrees of freedom and the the abstract nature of electromagnetics associated larger number of specimen and the associated problems in inter- variables processed. preting the wide range of signal effects observed. These, in turn, are a result The most highly developed standards of the indirect nature of the eddy current for eddy current application are conduc- inspection. In many examples, the con- tivity specimens for calibrating elec- ditions which give rise to signal changes trical conductivity meters. In these are only indirectly related to material applications, two standards are sometimes or structural content. used for each meter scale, one for establishing a calibration point near the upper The relationships between the cal- end of the scale, and a second for a point ibration problems for the single frequency near the lower end of the scale. method and those for the multi frequency method can be clarified by referring to Dimensional specimens are usually the section on principles. Let us first made by the laboratory or manufacturer examine the algebra describing the single interested in the application. frequency technique. This is exemplified by eqs. (5) and (6): An area of increasing interest is that + a = c 5 of standards for tubing inspection. a n Pi 12 P2 i ( ) Two conflicting needs are experienced: + = (1) the desire to produce effective stan- a 21 Pi a 22 P2 c 2- (6) dard irregularities which will cause signals similar to those from specified These are simplified equations based mill run flaws or to give other specified upon small signal (linear) theory. How- response, and (2) ease of fabrication. ever, they can be informative even though they do not describe the behavior of Drill holes are often used because of large eddy current signals. The co- the ease of fabrication and reproduc- a a and a each may efficients an, 12 , 2 i, 22 ibility of results. Holes are drilled depend upon all of the inspection through the wall or partially through the variables. For our present purpose, we wall. Notches are sometimes made using divide these inspection variables into

100 .

two groups. Group A includes the factors areas. Firstly, the number of algebraic which can remain essentially constant equations (two for the single frequency during an inspection period, and Group B approach) is increased by two for each new includes the factors which usually change frequency added. Secondly, the outputs of during an inspection. the multiplicity of channels are further processed through additional circuits Inspection Variables which are called transformation circuits in this paper. These additional circuits Group A - Essentially Constant provide an involved mixing of the detector outputs for producing the desired Probe assembly excitation separation of signals. The adjustment of these circuits is done in the calibration Instrument AC bridge adjustments procedures. The equations applicable here for the two-frequency system indicate the Instrument sensitivity increased complexity over that of the single frequency example are obtained by Instrument signal phase adjust expansion of eq. (7).

+ + + = Instrument output channel sensi- a n Pi a i2 ^2 a i3 P3 a i4 P4 Ci tivity control = a 21 Pi + a22 P2 + a 23 P3 + a24 P4 ^2 Group B - Usually Expected to Vary = a31 Pi + a 32 P2 + a33 P3 + a 34 P4 ^2 Probe wobble effect + + + = a 41 Pi a 42 P2 a43 P3 a 44 P4 C3 . Effect of test specimen (34) Electrical conductivity Again, as in eqs. (6) and (7), Cj... C Magnetic effects (if any) n in eqs. (34) are the various detector outputs. The next equation relates these Specimen and coil detector outputs to the final outputs. temperature changes (if + any) t>n Ci b 12 C 2 + b 13 C 3 + b 14 C 4 = Pi

+ + = Presence of flaws and other b 21 C 1 + b 22 C 2 b 23 C 3 b 24 C 4 P 2 irregularities within the + + + C = P test specimen. b 31 C x b 32 C 2 b 33 C 3 b 34 4 3

+ + + b C = P Because of our assumptions regarding b 41 C x b 42 C 2 b 43 C 3 44 4 4 the fixed aspects of the coefficients a., in eqs. (6) and (7), we can consider them (35) to comprise a modulation matrix which varies during an inspection mainly as a where the primed P 1 quantities indicate function of the inspection specimen. the final outputs, the estimated values During a calibration period, the elements of the parameters, and the b xl ...b 44 of this matrix are changed to new values, quantities represent the expansion of the -1 then being functions of the calibration [A] matrix in eq. (11). controls such as gain or phase reference settings. The single frequency inspection Further insight into the effect of output circuit has either one or two the multi frequency system on calibration output channels. With two channels, the can be seen graphically by referring to main calibration effects are changed in figures 10 and 11. The transition from the gain of either or both channels and a figure 10 to figure 11 is the result of rotating the vector signal pattern in 3- rotation of the pattern in the C x versus space around an axis normal to the sup- C 2 display plane. Individual control of port signals and passing through the the gain of the C x and C 2 channels cause distortion of the signal pattern. origin. The effect of this rotation is to greatly change the relative angle sep- In contrast to the single frequency aration between the three flaw vector inspection technique the multi frequency signals. In contrast, variations in the technique is more complicated in two main

101 reference phase adjustment in the single [6] Halmshaw, R. , Potential Developments frequency system cause just a simple in NDT, The British Journal of Non- pattern rotations. destructive Testing, 19, [1], 21 (January 1977)

5. Future Possibilities [7] Multi frequency Eddy Current Inspec- tion for Cracks Under Fasteners, The future of the multi frequency eddy AFML-TR-76-209, Battelle Columbus current technique is very promising. It Laboratories, December 1976. will produce results not possible with the single frequency method. Multi frequency [8] Advanced Multi frequency Eddy Current equipment can be made to operate in System for Steam Generator Tubing several different modes, including single Inspection, Report to Electric Power frequency modes. In its present stage of Research Institute by Battelle, development, multi frequency devices are Pacific Northwest Laboratories, more complicated and more difficult to Richland, WA, 99352 (to be adjust and operate than single (one at publ ished). time) frequency instruments. The development of automatic calibration means is [9] Libby, H. L. and Wandling, C. R. , needed. Transformation (Analyzer) Device Using Post Detector Signal Pattern Rotators for Multiparameter Eddy Current Tester, BNWL-1469, Battelle The permission of Battelle-Northwest Northwest, Richland, WA, September and the Electric Power Research Institute 1970. to use figures 4 through 13 and the basic data given in table 1 is gratefully acknowledged. Also acknowledged are the Di scussion helpful suggestions of G. J. Posakony, Manager, Nondestructive Testing, Battelle- Question (Dr. Birnbaum): I would like to Northwest. get back to your comment about nonlinear effects, even though your talk was not directed at that. Is there information References there that can be used, for example, to deliberately try to look at the nonlinear [1] Mc Master, R. C. , ed. , Nondestructive effect as a flaw detection method? Testing Handbook , Vol. II, Ronald Press, New York (1959), Sections 36, Answer (Mr. Libby): Yes there is. I am 37, and 38. limited here with the two frequencies, in this example. But the more frequencies we Libby, H. L. , U. S. Patent 3,229,198 [2] would include, and using Taylor's approx- January 1 , 1966. 1 imation, you can say a curve is a sum total of a lot of segments. So even though you [3] Libby, H. L. , Broadband Electro- have nonlinear effects, if you have enough magnetic Testing Methods, Part IV, information from the signal, then you can Multiparameter Principles, BNWL-953, work with this. I think it is straight- Battel le Northwest, Richland, WA, forward mathematically, All that I have January 1969. described here, I have done by hand — by pocket calculator, but this can all be done [4] Libby, H. L. Multiparameter Eddy , using algebra or through the computer. You Current Concepts, in Research Techni - can use all of the regression techniques, ques in Nondestructive Testing , R. S. and nonlinear approximations. It is just

Sharpe, ed. , Academic Press, London more difficult to do. 345-382. (1970) , pp.

- I have tried the method described with [5] Libby, H. L. , Introduction to Elec magnetic materials with just a few frequen- tromagnetic Nondestructive Test cies and the results were discouraging. Methods , Wi ley-Interscience, New York But I think only because I did not have (1971) . enough information, enough different frequencies, to describe all of these curves.

Question (Dr. Birnbaum): I was thinking more, for example, of using two frequencies 102 Ui and U 2 , and looking at the sum frequency Question (Mr. Blew): On your example, which is created by a nonlinear inter- there was a scaled amplitude of 200 j

I . action. millivolts. What is the actual amplitude of the smallest flaw that you can handle Answer (Mr. Libby): This is done now. with a signal-to-noise ratio that would The 60-cycle testing of magnetic materials allow you to detect it accurately? over the years has made use of these harmonics, and they are present there. Answer (Mr. Libby): Well, this varies a lot. It would vary with application. This is a relationship that can be It depends on the material and the further explored. I feel this is just a particular test, and the noise level. start. We have just scratched the surface. There are all kinds of possi- Question (Mr. Blew): What would you think bilities and this is where the micro- the practical limitation would be? processor will help to handle the greater amount of information that can Answer (Mr. Libby): Well, I do not know be made available. what the limitation is, actually. We have worked with flaws a few mils -- in Question (Mr. Wehrmeister): Was this depth. But now if we are discriminating system designed specifically for de- against a support on the outer wall, the

' tection, or do you receive from it tests are more insensitive to small information that provides defect anal- signals on the outer wall than in the ysis in terms of its depth? regular test.

Answer (Mr. Libby): So far, we use Question (Mr. Blew): Would this have mainly the amplitude of the signal to been in about a mil? determine the severity of the condition. If I have several flaw conditions, more Answer (Mr. Libby): I just cannot tell than is accounted for by the number of you what the limitation would be. variables that I can handle, then any of these flaw conditions appearing in- Question (Mr. Blew): With practical dividually will show up with a phase experience are you getting down into the angle difference, like the 100, the 80 millivolt region of signal? and the 20 percent flaws. Answer (Mr. Libby): Well I think this Now, that phase angle difference is just relative as' far as the milli- will show up on the screen as long as I volts. That depends on how much you are am translating the coil past those driving the coils, what the instrument flaws. But now, if I put the support gain is. I had those units on the example right on top of the flaw> the support to help me in my calculations, so I car- signal is in the horizontal direction. ried through all those relative amplitudes from the start, starting with the basic Consider the flaw signals now. data. Even if they occur right at that same point as the support signals it will not Question (Mr. Brown): From the standards change their vertical contribution at and calibration point of view, is it all. But, in this case I have used up true to say that you have to have a all the degrees of freedom I am entitled sample of each of the parameters that to. I have used one parameter for you are juggling? If there are one or

wobble, two for the support, and I have two things you want to get rid of, you got one left to read out one additional have to have a sample of them; and if parameter. there are one or two things you want to measure out, you have to have a sample

But if I have three parameters, I of them. So, the standard for multipara- say, there are the 100 percent, the 80 meter testing might very well have percent, and the 20 percent flaws. There several different kinds of parameters on is some limitation here. I have to the standards. throw out the other two. Generally, you do not have flaws appearing under the Answer (Mr. Libby): And there is this support, so you can use the angle problem. If you are using a system like information. But when they occur under the one I described where I am manually could be the support, then I can use only the adjusting, the electronics amplitude information. arranged so that it will aid this Comment (Dr. Mc Master): If I may use the adjustment. That is something for the board a moment, I want to mention one future, where the processor, micro- little thing I found that helped me on I

processor would come in. You need to wobble with probe coils and could be used i present these parameters in fairly rapid here to get that one nasty variable out. sequence if you are doing it manually, It applies to other coils as well. because you have several things to minimize. I have got to go across one, To take a very simple case, our two, three parameters, and if there is one magnetizing coil might provide a certain in there that I do not want to minimize, number of ampere turns or magnetizing then I have to remember that. And I must force or flux density or signal in the minimize wobble at the same time. Of vertical direction. And if we put in course you can do the wobble separately. ferromagnetic materials, we will increase But, in the first generalized adjustment the flux, so that you get a larger where there are three knobs to adjust, resonance curve. three or four parameters, then you have got to wobble the probe at the same time. If, with respect to, say, a probe You could do the wobbling, and then as you coil and the surface, you have a liftoff, pass the different parameter signals, S, and if you were to wiggle the probe up different flaws that you are calibrating and down, say, through a modestly adequate against, then minimize them that way. In range to be greater than any effects of some cases, you can do them one at a time, surface displacement, it is possible to but you have to be careful. arrive at a very interesting situation. If this represents the 100 percent signal This is a more costly system, and it in air, in the absence of the test object, is more complicated to adjust, more com- it is something you can easily calibrate plicated to operate. But as it becomes an instrument to. more automated, we can make automatic cal ibrations. If this is your curve with the ferromagnetic material present, then this Question (Mr. Berger): I think you really point also represents a vector of 100 answered the question I was going to percent magnitude. Notice, the phase has raise. Because your original answer to changed. But you can wiggle from here to his question implied that you needed a here with negligible change in the overall physical standard in order to calibrate signal. It sits there at 100 percent all the distance. I was going to question the time. So all you do is tune the that. I think you could store in computer oscillator with the ferromagnetic object memory what the signals would look like. in place such that when you wiggle this up and down there is no visible effect on Answer (Mr. Libby): Like the system that your signals, and then read out a was described; yes, an approach like that frequency of balance, if you will, could be used. Yet, there are some subtle whatever you want to call it, a frequency things here. I do not want to over- which restores 100 percent signal, which simplify it in a few slides like this, but often can be read out rather accurately. it represents many years of effort. And there were a lot of difficulties along the I find very frequently when you read way. out frequency instead of other parameters, you get about five figures of stable You must be careful. For example in indication. So I have often thought that the wobble adjustment, I kind of glossed in cases where wobble is a problem, if you over that. It was just stated that there took it out at the probe by the selection is wobble adjustment, which gets rid of of the frequency which is automatically the wobble. And I said it just like that, self cancelling for liftoff of huge and it comes out beautifully on the slide. amounts, and then went into what you are But the output signals, especially the doing, it seems it would be helpful. phase angles between the final output signals that I showed for those three Question (Mr. Bugden): If you use two flaws, are fairly sensitive to the wobble frequencies, is the relationship of the adjustment, because you are dealing with two frequencies to each other, of great all these different dimensions. Once you consequence? adjust for it, then it can hold. But you do not want to change your mind after you Answer (Mr. Libby): I like to work with have it calibrated. two to one or three to one, but we have

104 proved mathematically that as long as you take different frequencies, no matter how close together they are, the independence of information exists. The signal-to-noise ratio may or may not improve depending on the choice or frequencies. No matter how close together you get these frequencies, theoretically, there is some difference, but when the skin effect becomes more equivalent for the different frequencies, you have less difference in signals to work with.

Question (Mr. Bugden): Do you choose the frequencies for convenience?

Answer (Mr. Libby): Yes.

Question (Mr. Mester): Have you determined the amount of liftoff that you can compensate for?

Answer (Mr. Libby): I do not have an exact figure, but it will compensate for a rather large amount.

Question (Mr. Mester): Has this work been confined to ID coils or have you worked with surface coils?

Answer (Mr. Libby): Well, I have used surface coils for conductivity variation, and variation in thickness, conductivity being one variable, thickness the other. We have used encircling coils, and small probe coils. I said it was not good for magnetic materials, but if, for example, you have magnetic inclusions, just below the surface of the material, it is very good for detecting that; it eliminates the wobble effects and some other variables and detects the magnetic inclusions. Or, if you wanted to cancel that out, and read some flaw signals that occur near it, this could be done. But this is when you have a small amount of magnetic material.

Question (Mr. Mester): Is this system described in a document that is available?

Answer (Mr. Libby): Yes, and there are references in it to some of the previous work. There is also a reference to a new report that will be made available to the public through EPRI, which includes this work, plus additional information, but that has not been issued yet.

105 Cur

Jai National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy Current Nondestructive Testing held at NBS, Gaithersburg , MD, November 3-4, 1977 Issued January 1981.

PULSED EDDY CURRENT TECHNIQUES FOR NONDESTRUCTIVE EVALUATION

D. L. Waidelich University of Missouri Columbia, MO 65201

1. Introduction A block diagram of a typical system is shown in figure 1. The first systems The pulsed eddy current system has employed had a pulse generator driving a been used for the nondestructive evalua- probe coil which launched the electro- tion of materials since the early 1950' s. magnetic waves in the specimen of the Some of its advantages are much less metal to be tested. The pickup coil re- thermal drift and much greater resolution sponded to the waves issuing from the than for systems using continuous sinu- metal and containing the information con- soidal waves. A typical system is de- cerning the defects in the metal and the scribed and some of the waveforms are properties of the metal. The output of presented. Also, it is shown that the the pickup coil was observed on an oscil- problem of lift-off may be overcome by loscope. Later, the output was filtered employing the idea of crossing points. in various ways, and an electronic gate An equation involving the pulse length, was used to pick out a particular portion the material constants and the depth of or portions of the output wave for re- is developed. Some results penetration cording. Recording may be done as a con- when testing non-metallic mate- obtained tinuous trace on paper or as a digital given and also some recent rials are readout printed periodically. The output experiments with thick metallic slabs are could also be entered into a computer for Finally, various problems are discussed. further processing, such as digital fil- presented and some suggestions are made. tering, the employment of a decision process, or an adaptive method involving 2. Previous Work storage. Various forms of alarms or marking devices could also be actuated. Much of the early work on the pulsed eddy current method was summarized in a 1 recent reference [l] . In the early 1950' eddy current systems used single PULSE DRIVEN s, METAL frequency sinusoidal sources, and the GENERATOR PROBE subsequent heating of the probe coils led SPECIMEN to thermal drifting which, in turn, caused errors in locating defects. It PICKUP was decided to use a pulse generator to OSCILLOSCOPE COIL drive the probe coil, and the thermal problem disappeared immediately. There

was a new problem, however, in trying to FILTER interpret the results as viewed on the RECORDER AND GATE screen of a cathode-ray oscilloscope. It was found quickly that the defects near the surface of the metal would show up in the first part or head end of the pulse, while those deeper in the metal would Figure 1. Pulsed eddy-current system. affect the tail of the pulse.

figures in brackets indicate the literature references at the end of this paper.

107 .

The most usual circuit supplying energy to the driven probe consists of a capacitor D which is charged slowly through a resistor R from a dc power supply as shown in figure 2. The capacitor is then discharged suddenly by a

R SCR 0 c or POWER THYRATRON ~] SUPPLY ORIVEN C PROBE COIL

Figure 2. Circuit supplying the driven probe. Figure 4. The voltage across the pickup silicon-controlled rectifier (SCR) or a coil thyratron through the driven probe coil. The pulse waveform is very nearly a half- curve and finally an exponentially decay- wave sinusoidal loop as shown in figure ing voltage. Defects and changes in the 3, and the length of the pulse is deter- material properties change the shape of mined primarily by the quantity -/EC where the voltage from the pickup coil. These changes carry the information about the defect or the material property, and this information may be abstracted by the proper method.

One of the problems is the effect of "lift-off." This may be overcome by employing the idea [2] of the "crossing point." If the pickup is placed directly upon a metal specimen, part of the pickup coil voltage of figure 4 is shown as the curve AA' of figure 5. As the coil is

Figure 3. The current through the driven probe.

L is the inductance of the driven probe coil and C is the capacitance of the capacitor. The pulse shown is about two microseconds long and has a peak value of approximately 12 amperes. There is a little tail to the pulse which is caused by the deionization of the discharge device. The shape of the current pulse is not very important because the metal acts as a low pass filter. Consequently, the higher harmonics contained in a rectangu- Figure 5. Crossing point. lar pulse, for example, will be attenu- ated rapidly and the result in the pickup moved away or lifted from the metal sur- coil will be nearly the same as if the face, the trace of the pickup coil volt-

1 1 driving pulse were a half-wave sinusoidal age successively moves from AA to BB ,

1 loop. then to CC, and finally to DD . The crossing point 0, however, is not affect- The pickup coil voltage is shown in ed by lift-off, and so if the pickup coil figure 4. Note that there is an initial voltage is sampled electronically at the jump in voltage, then a sinusoidal shaped crossing point 0, the output of the sampler will be completely independent of

108 .

lift-off. The position of the point 0 the field is detected on the surface of will depend upon the presence of a defect the steel by the small magnetic probe or upon the properties of the metal spec- coils. The presence of a defect in the imen, so its motion may be used to locate steel is indicated by aberrations that a defect or to determine metal properties occur in the detected field. Some work without worrying about the effects of has also been done in detecting defects lift-off. in composites such as those made of graphite. 3. Recent Work

DRIVEN COIL One important relationship in pulse work is that between the length in time of the pulse, the constants of the materials being tested, and the depth of pene- V tration of the electromagnetic waves into METAL -VN CRACK the materials. In the Appendix, it is demonstrated that PICKUP

2 T = au D (1) Figure 6. System employed for ferrous material where T = length in seconds of the pulse; a = electrical conductivity of the mater- 4. Questions and Suggestions ial in mhos per meter; u = magnetic permeability of the material in henries per It appears as if further knowledge meter; and D = depth of penetration into is needed in the direction of what are the material in meters. In some recent the limits as to the thickness of mater- work in aluminum and using pulses about a ials that may be traversed by the electro- millisecond long, this equation has been magnetic waves. Equation (1) may be modi- found to be useful in predicting the fied by the state of the art, for example, depth of penetration. by the sensitivity of the detectors available and, undoubtedly, by the noise pre- Some work on testing poor electrical sent. The question also arises, does conductors, such as plastics, indicated this equation or a similar one apply to that the above magnetic probes were not poor conductors, .semi-conductors, and very successful. The material seemed to insulating material? There is another have relatively little effect upon the problem that occurs when defects are de- magnetic flux lines emanating from the tected from one side of a material as probes. It was thought that this type of compared to through-transmi ssions , and material might react more on the electric this seems to indicate that the C of the flux lines, so capacitive probes were Appendix should be greatly reduced when fashioned and simulated defects in plas- through-transmission is employed. There tic materials were detected by the use of is also a considerable amount of work pulsed waves launched by the capacitive that should be done on the probes em- probes [3]. Accidently, it was found that ployed. Two problems especially would be these probes were also extremely sensi- important in this direction. The first tive in picking up and locating bits of would be the development of better probes metal in the plastics. for the poorer conductors and better insulators, and the second would be the inves- Lately, experiments have been made tigation of the masks or shields for use aimed at transmitting the waves through with longer pulses to provide better an inch or more of aluminum and in detect- resolution. Further work is needed in ing defects in a second metal layer the use of electronic gates, amplifiers, through about a quarter inch of aluminum. and the devices that record the informa- Also defects in steel have been detected tion. Also methods of decreasing the through a quarter inch of metal by using noise are needed, and these would include the set-up shown in figure 6. The mag- correlation methods and all types of fil- netic field is generated in the steel by tering. A number of theoretical studies using a coil wound on the elongated C- would help greatly in understanding the laminations. When the current in the processes and in determining optimum coil is cut off, the field collapses, and operation of the equipment.

109 ,

References introduced in (A-3). For example, it has been found in nonmagnetic stainless steel

[1] Waidelich, D. L. , Pulsed Eddy Cur- that pulses 2 microseconds long may be rents in Research Techniques in employed to reach depths of 40 mils (1 = Nondestructive Testing , R. S. Sharpe, millimeter) in the steel. Now if a 6 ecT (Academic Press, London, 1970), 1.1 x 10" mhos per meter and p = pp. 383-416. 4n x 10" 7 henries per meter, then C = 1.447. The actual pulse length is not

[2] Waidelich, D. L. and Huang, S. C. , at all critical, so for most work C is The Use of Crossing Points in Pulsed put equal to unity in (A-3). The actual Eddy Current Testing, Materials value of C depends somewhat on the state Evaluation, 30, 20-24 (January 1972). of the art in that if more sensitive detectors are employed, D would increase [3] Decker, W. A. and Waidelich, D. L. for a given T, and thus C would have to Nondestructive Testing Using an be decreased. Also, the effect of the Electric Field Probe, in Proceedings noise present would change C. The above of the Seventh International Confer - value of C was obtained in detecting de- ence on Nondestructive Testing fects from one side of the material. , If Warsaw, Poland, June 1-8, 1973, through-transmission is employed, it Paper D-02. appears from some tests as if the value of C should be reduced to about 0.05.

Appendix Discussion Assume that the surface of the material is the x-y plane and the positive z- Question (Mr. Wehrmeister): In pulse axis extends into the material. The eddy current work, what are the effects conductivity of the material is a mhos from acoustic energy generation in the per meter and the magnetic permeability transmission of the pulse? Are some of is u henries per meter. The vector Helm- the time delays that you refer to the holtz equation for the magnetic field acoustic energy being transferred to your intensity H is assumed independent of x pickup coil, as opposed to the electro- and y. In addition, the Laplace trans- magnetic energy being transferred to the form is employed to introduce the complex pickup coil, especially in magnetic variable s in place of the t in seconds. material?

Then Answer (Mr. Waidelich): In all tests, there was no actual contact with the material, so it would be rather difficult to get very much acoustic ultrasonic type ^-4 = ausH (A-l) transmission across the air between the dz^ pickup coil and the material itself. Un- doubtedly, there is some motion in the specimen, the metal specimen itself. But and the solution is I have seen relatively little effect that can be noticed. Z^CT|JS H = (Ho/s)e~ (A-2) Question (Mr. Blew): In your work, when there was a conducting film in the air or when there were films on a conducting where Ho is the initial magnetic field base metal, what would be the minimum intensity at the surface (z = o) of the thickness that could be resolved, and material. Now the exponent of (A-2) must how close were the respective conductivi- be dimensionless , and since s has the ties to each other? dimension of the reciprocal of the time T

Answer (Mr. Waidelich): The example, if 2 I remember, was zirconium on uranium. I T = Cap D (A-3) do not remember the conductivities too well, but both are relatively poor conductors. The closer the conductivities where the depth D is used in place of Z, become, the more difficulty you have in and C is a dimensionless constant. To separating them. determine the approximate value of C, known values of T, a, u, and D may be

110 Question (Mr. Blew): And what were the Question (Mr. Mester): You mentioned a relative thicknesses? thermal problem.

Answer (Mr. Waidelich): The thickness of Answer (Mr. Wadelich): The heating that the cladding was about 30 mils, something exists when using sinusoidal currents of that order. causes a lot of drift. The drift caused difficulty in getting everything nulled Question (Mr. Blew): And the base? out. But, for one pulse the thermal effect is relatively small. You can put a Answer (Mr. Waidelich): The base was large current in this one pulse and get quite thick. I would say easily a quarter quite a strong response. The only trouble inch. in doing this is the problem of trying to pick up the information afterwards. You Question (Mr. Mester): In the example do not have the advantage of using all where you have a quarter inch of steel as the sinusoidal methods. the limitation of what you have penetra- ted, was DC saturation used during the test?

Answer (Mr. Waidelich): No, we had not used that in that particular example. We were going to do that in one of the future experiments.

Question (Mr. Mester): What power levels are involved with your inputs to your driven coil?

Answer (Mr. Waidelich): We were using approximately a 100 volt pulse. I cannot tell you what the current was. We did not measure the current.

Question (Dr. McMaster): How big were the capacitors?

Answer (Mr. Waidelich): We used a number of sizes. That is something like .01,

.05, . 1 , .5 uF.

Question (Mr. Bugden): Is there any particular ratio between the duration of the pulse and the whole cycle that you find beneficial?

Answer (Mr. Wadelich): It depends on how thick a metal is tested. The electromagnetic waves penetrate and are reflected from the back surface. The deeper a metal, the more time it takes for this process to occur. This is indicated in this equation to some extent. If you put a particular pulse into a thick metal, oftentimes the information you are looking for comes a long distance after the input pulse is stopped. You might find that you get something like this in your pickup coil --that is, a long tail. This tail can be quite long. The surface information is prior to this, and the deep information is way down here some place.

Ill

National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy Current Nondestructive Testing held at NBS, Gaithersburg, MD, November 3-4, 1977. Issued January 1981

THE INTRODUCTION OF SIGNAL PROCESSING TECHNIQUES TO EDDY CURRENT INSPECTION

E. E. Weismantel Quality Measurement Systems The Aircraft Engine Group General Electric Company Cincinnati, OH 45215

1. Introduction 2. Current Trends

The use of eddy current techniques It has only been in the last five to for the nondestructive interrogation of ten years that advances in the process materials is not a new consideration technology itself, coupled with the since the process has been in use to some availability of a new generation of degree for this purpose since the Second electronic equipment concepts, has mar- World War. As with any advancing tech- kedly broadened the potential of the nology, the increased application of the process. process results in a firmer definition of its best uses as well as its limitations. This recently available equipment The basic process itself offers a very allows the use of the impedance plane for high sensitivity for finding small mate- the analysis of eddy current signal rial flaws in the near surface region. response. A typical impedance plane However, because the process is very presentation is illustrated in figure 1. sensitive, it also responds to other non- This illustration shows the position of flaw type conditions that may exist various materials on the impedance plane during the normal application of the relative to their reactance and resis- process. Typical major influences are: tance effects on the' coil. Differences localized changes in conductivity due to in the electrical conductivities between alloy segregation, thermal effects, materials are illustrated on the result- or residual strain patterns within the ing plots. The occurrence of flaws material, as well as factors that would within an alloy generally results in a also affect the reactance of the system small change along this curve. Coil -to- such as coil to metal intimacy, part material spacing on the other hand, configuration, etc. Such factors have assumes a vector direction as shown perhaps inhibited the broad application toward the "Air" termination point on the of the process more than anything else conductivity curve. Thus, with the use since, except for specific applications of the impedance plane, one gains the where the effect of these other influ- ability to observe whether the coil's ences could be minimized or where the reaction is due to a change in conduc- flaws being sought were large enough as tivity or due to coil to material spac- to override the effects of these other ing, probe wobble, etc. factors, the eddy current process sometimes developed suspicions as to its A typical commercial instrument is reliability. The early uses of the shown in the next illustration, figure 2. process were further hampered by the This instrument has an added feature in characteristics of the early instru- the rotational knob shown on the upper mentation that was available for its left hand portion of its front panel. application since the meter display of This control provides a control over the this vintage integrated the effect of all display which allows the rotation of the of these influences into a single meter impedance plane so that the lift-off readout. effect, for example, can be made to occur in a specific coordinate direction. Without the rotation feature, effective use of the impedance plane relies on the observational skills and attenti veness of 113 .

the output along the conductivity curve 1000 consists of both horizontal and vertical movement. This is diagrammatical ly shown 900 in figure 3. If a two channel recorder such as that shown in figure 4 is adapted .i£n MAGNETIC MATERIALS 800 .OlO"" to the recording of output data from the rotated impedance plane, the one channel

700 can be arranged to contain the vertical movement along the conductivity curve while the second channel contains a com- 600 posite of the movement in the lift-off

\.006 , '> NON-MAGNETIC MATERIALS direction and the horizontal movement as- .OlO"\.O03" 500 .006" sociated with conductivity change effects. .003- A typical presentation of this is shown in 100 figure 5.

200 LIFT-OFF FROM STEEL, TI 6-li, LEAD & 20U VERTICAL DEFLECTION •VECTOR NUMBERS INDICATE COIL TO METAL SPACING RFSISTANCE

300 100 500 600 700 800 900 HORIZONTAL COMPONENT OF CONDUCTIVITY RESPONSE Figure 1. Impedance Plane.

CONDUCTIVITY ChANGE OR FLAW RESPONSE CURVE

PURE LIFT-OFF RESPONSE

HORIZONTAL DEFLECTION

Figure 3. Representation of typical lift- off and conductivity change response on the rotated impedance plane - lift-off horizontal

Figure 2. Typical commercially available eddy current instrumentation. the operator since both the lift-off vector and the conductivity effects contain both vertical and horizontal components, and the plotting of horizontal and vertical output data on normal recording instrumentation would be of little value for most applications. However, the rotational feature of present day instrumentation becomes of significant importance to the process. With this feature, the vector representing lift-off conditions can be brought to react in an Figure 4. Dual chart recorder used for almost entirely horizontal direction while data gathering.

114 1

LIFT-OFF REACTION ESTABLISHED AS PURE HORIZONTAL MOVEMENT

CHANNEL 2

VERTICAL COMPONENT OF CONDUCTIVITY CHANGE - CHANNEL 1

0.070' LONG CRACK (NATURAL FLAW)

0.006" DIAMETER x 0.080" LONG HOLES - 0.003" BEL0H THE INSPECTION SURFACE

CHANNEL #1 CHANNEL #2

Figure 5. Illustration of typical record-

ed eddy current response. FOUR EDM NOTCHES - 0.020" x 0.010"

3. Other Application Influences Figure 6. Illustration of typical eddy Thus far, we have done little more current response on turbine blade edges than to manipulate the impedance plane, without signal processing. but from these illustrations the ease with which we can derive more meaningful data 4. Overcoming Geometry and from the eddy current response should be Conductivity Effects evident. Unfortunately, to this point, we have been using idealized illustrations to Sometime back when General Electric show the significance of recent advances first undertook the task of applying eddy in eddy current instrumentation and appli- current inspection in the production en- cation methods with regard to the poten- vironment, we recognized that the vari- tial of the process. In facing real world abilities we have discussed thus far might situations, even though these advances are affect the reliable use of the process, certainly significant, a number of other and we, in fact, delayed the use of eddy factors are encountered that further cloud current inspection by production operators the interpretation of signal response re- until a more interpretable condition could sulting from the use of the process. Fig- be established. With the cooperation of ure 6 shows a strip chart recording of the the equipment vendor, we evolved into the typical eddy current response observed in use of signal processing as a useful tool the inspection of turbine blade edges to further enhance implementation of the using all of the innovations we have dis- process. Naturally, the first use of sig-

I cussed to date. Certainly, even though a nal processing involved a black box which flaw signal is evident on this recording, is illustrated in figure 7. This addition the chance for operator misinterpretation to the normal eddy current inspection remains high due to the confusing response equipment already described yielded a dra- associated with the blade edge inspection. matic increase in the interpretability of The changes evident during this inspection the eddy current signal as is evidenced in occur due to changes in electrical con- figure 8. The upper portion of this il- ductivity along the cast airfoil as lustration shows the signal response nor- well as changes in geometric configuration mally resulting from an airfoil blade edge from platform to tip. The eddy current containing known flaws. Although the re- process is sensitive to both of these ef- sponse of the flaws is evident to a train- ] fects as the recording shows. ed operator for most of the airfoil, certainly as flaw size and thus the response gets smaller, the chance for a miss on the part of the operator increases. The lower portion of this diagram shows the inspection of the same airfoil edge using the

115 processed signal. Clearly, the effects of current inspection of a series of grooved conductivity and geometry changes have blocks — representative of a fillet con- been all but eliminated, the flaw response dition. Three different alloys were used originates from a constant baseline, al- in the experiment. Each groove contained lowing a better judge of relative signal up to three fatigue cracks in a size range amplitude, and the signals resulting from varying from 0.010 to 0.250 inches. A flaws are clearly evident. total of 131 cracks was used in the evaluation. The result of this study is presented in figure 9. The superior flaw detection capability of the process seems to have been unaffected by the signal processing used for these tests. In fact, in the small crack size end of the curve, the detection efficiency of the process seems to have been improved probably due to enhanced operator interpretabil ity. With this background, we introduced the use of processed eddy current signals to the production inspection of blade edges more than a year ago. Increased benefits of further signal processing developments appear to be obtainable.

WITHOUT SIGNAL PROCESSING

KITH SIGNALSIGN PROCESSING

Figure 7. Signal processor used with eddy current instrumentation.

INCREASING CRACK LENGTH ->~

H0H-PR0CESSED SIGNAL Figure 9. Relative improvement in flaw detection performance resulting from signal processing.

PMXESSED SIGNAL Until now we have discussed things that have already happened specifically with regard to signal processing. These improvements have resulted in the ability Figure 8. Direct comparison of processed to get some intelligence out of the eddy vs. non-processed eddy current signal. current inspection data and have opened the way for the further adapting of the This very elementary approach to sig- process to the use of the computer for the nal processing consisted only of the re- further advancement of application tech- moving of the gradually changing responses nology. Typically, today's inspection of due to geometry and conductivity while cast airfoils uses a varying accept/reject allowing the more discrete and abrupt level depending upon where along the air- changes due to flaw conditions to remain. foil a response must be considered. The Because we were concerned that these at- most critical portion near the airfoil tempts to improve operator interpret- fillet allows a maximum response amplitude abil ity might degrade the very excellent of 10 percent as shown in figure 10. flaw detection capability of the eddy cur- Higher allowable amplitudes exist as one rent process, a statistically designed goes outward from the platform due to the experiment was undertaken to assess the lower stresses and lesser critical ity in detection efficiency of the process both these areas. Currently, effort is under- with and without the use of signal pro- way on a semi -automated inspection system cessing. The test consisted of the eddy which uses microprocessors to control the

116 .

movement of the inspection probes as well process. Even though current efforts to as the acceptability of the signal re- extend the technology should and will con- sponse observed along the airfoil edges tinue, it is only through the enhancement relative to the probe's position at the of our theoretical understanding of the time the response is observed. Many other process that the value of multif requency possibilities exist for the application of testing and other advanced methods can further signal-processing techniques to really reach their full potential. the process as we move to the future, but one major unknown feature must yet be recognized. 6. Summary

1. The introduction of the impedance plane to practical use has opened many avenues for improving the interpretation of eddy current signal response.

2. The rotation of the impedance plane to differentiate the characteristics of the response further enhances the application of the process.

3. Signal processing methods can be applied to improve the interpretability of the response.

4. With these accomplishments, the use of eddy current inspection can be tied to microprocessors and computer control

5. Although marked advancements have been made in the application of the process much theoretical work remains to be done to gain its full value and potential.

Figure 10. Inspection criteria for typi- Discussion cal turbine blade edge. Question (Mr. Judd): The signal pro- 5. Looking Into the Future cessing device that was shown on one of your slides, is this commercially avail- Certainly the use and value of eddy able? current technology has increased markedly in the recent past. However, as we look Answer (Mr. Weismantel): I expect it into the future we must recognize how much probably is commercially available today. more we have to know if we are to use sig- It was developed on this program with the nal processing of eddy current responses cooperation of the equipment manufacturer. to its fullest advantages. Currently, the He has a unique advantage in the fact that application of the process is guided by a it is applicable to his equipment but not few basic laws of physics and electronics to some of his competitors' equipment supported by an amount of empirical deri- without going into the internals of the vations and technical logic. Few of these competitors' equipment. really define the conditions we experience in the testing of the complex metallurgical Question (Mr. Judd): Is this device es- alloys which forms our every day work and sentially a wave shaper? to which the process is to be applied. Therefore, we have to work towards the Answer (Mr. Weismantel): I did not fully development of a theoretical understanding describe the total application of the of the process used in these situations signal processor. It has several dif- supported by computer modeling to help ferent functions. The one that I was predict the needs and performance of the describing involved the filtering out of lower frequency occurrences that are Question (Mr. Lagin): The signals that associated with geometry and with chem- you showed, were typical signals that I istry changes. These occur rather grad- would see on a strip chart recording. ually, whereas a flaw response is a very You say they were on an oscilloscope? sharp occurrence. And, that is about the most simple approach that you can take to Answer (Mr. Weismantel): Those were signal processing. The processor also strip chart recordings. But, that was removes electrical noise from the signal because the signal had already been as well as performing a number of other processed and had been leveled out. functions. Comment (Mr. Lagin): As the strip chart Question (Mr. Lagin): Did you try any recording is progressing in time, you pattern recognition schemes for detection obtain a very sharp signal over a small of dings, cracks, or grooves? amount of time on a defect. For a groove, it would be a slower type signal Answer (Mr. Weismantel): Yes, although we and you could perform image processing are not applying these. Again, it is a techniques on a signal like that. situation where the people in the laboratory have developed some proficiency, Answer (Mr. Weismantel): Yes. I am not and are trying to transmit some of that saying you cannot do that. The problem knowledge to people who might not be as you have is that the response from some flexible as the laboratory people. We are flaws can resemble the response of a moving in that direction, and it seems to groove or notch or ding, except that they be entirely possible, although what I say might move in one direction or the other has not really been totally proven. as far as their first movement is concerned. Not all responses start out in a Question (Mr. Lagin): The signal which positive direction, but the motion seems you actually process, is it like a signal to be related to the character of what off a strip chart recorder? you are encountering.

Answer (Mr. Weismantel): No. It is the Question (Mr. Houserman): The graphs you signal sensed by the probe that is passed presented on detectabi 1 ity, were the through the flaw detector and then pro- statistics gathered from a production cessed before it goes to both the os- type operation? cilloscope and the strip chart recorder. Answer (Mr. Weismantel): Yes. Question (Mr. Lagin): But, it is quite possible to use the signal processing Question (Mr. Houserman): With the scheme on a signal in the recorder? people that typically do the measurement?

Answer (Mr. Weismantel): Yes. I do not Answer (Mr. Weismantel): Yes. think you have as big an advantage in doing that, but yes, you could do that. Question (Mr. Houserman): Secondly, We also have a switch which allows the would the data show that with signal signal processor to be switched in and out processing you are getting more false of the system. This is especially alarms? valuable if you are doing any analysis work where the operator can still use the Answer (Mr. Weismantel): No, actually we oscilloscope to a large advantage. are getting less. We delayed the introduction of the process to production The production system we have, until we had the signal processor, be- incidentally, inspects two blade edges at cause during the time we were initially a time. Both the leading and trailing looking at this problem, we recognized blade edges are tested simultaneously. It the very high rate of false encounters really has two parallel systems, two that we were having. At one time without signal processors, two flaw detectors, but the signal processor, false signal alarms it only has one CRT since the CRT is were encountered on approximately 10 needed only for setup and for problem percent of the parts processed. With the

analysis. I do not know what advantage signal processor it is about 1 percent, you would have in trying to process the and we think we have a better product. signal after it came out of the flaw detector and oscilloscope but before going Question (Mr. Denton): Looking at the into the strip chart. data on this strip chart, it appears your

118 signal processor is really a resistor and capacitor differentiator. Is that true?

Answer (Mr. Weismantel): Not entirely.

Question (Mr. Brown): The curve that you showed flattened off on the right side. Did you increase the size of the crack? If your crack was several inches longer, a foot longer, does it drop off?

Answer (Mr. Weismantel): Not that I know of. But, since we have not had any cracks that are that long, I do not have data to show that.

119

National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy Current Nondestructive Testing held at NBS, Gaithersburg, MD, November 3-4, 1977. Issued January 1 981

DEVELOPMENT OF NON-FERROUS CONDUCTIVITY STANDARDS AT BOEING

Art Jones Boeing Aerospace Company Seattle, WA 98124

1. General

Figure 2 shows the chain of The need to verify the accuracy of traceability to NBS from dimensional, eddy current meter readings which are used resistance, and temperature to determine the physical characteristics standards. of non-ferrous alloys by measuring their electrical conductivity is fully accepted by industry, both manufacturers and users. The eddy current meters, therefore, must CONDUCTIVITY Sl*NOA»D! TBACEAgUITY TO NBS be accuracy certified by means of NBS 1 traceable conductivity standards for the readings to be both reliable and repeatable. Figure 1 shows a typical graph of tensile strength vs. %IACS (conductivity) for aluminum. To calibrate and accurately

AVERAGE VALUE

, 95% CONFIDENCE LIMITS

Figure 2.

29.5 30.0 30.5 31.0 31.5 32.0 2. Definition of %IACS

CONDUCTIVITY (% IACS)

tensile strength and eddy The accepted definition of commer- Fi qure 1 general relation between ' CURRENT CONDUCTIVITY MEASUREMENTS OF 2024-T4 ALUMINUM cially pure copper as stated in NBS Copper Wire Tables #31 is: a copper rod, one certify these eddy current meters, some- meter long having a uniform cross-section times called conductivity meters, the of one sq. mm and a resistance of 1/58 ohm Boeing Company embarked upon a research at 20 °C is 100 percent International and development program in 1966 to produce Annealed Copper Standard, or 100 %IACS. their own NBS traceable non-ferrous con- Using this as a reference then, all other ductivity standards, since none were com- nonferrous metals can have their conduc- mercially available at that time. Also, tivities determined relative to the hypo- NBS was engaged in providing only commer- thetical 100 %IACS value. By using the cial copper conductivity standards at that formula time so it was necessary for Boeing to obtain indirect traceability by means of dimensional, resistance, and temperature = RA P standards. All of these NBS traceable L standards were available at the Boeing Me- trology Laboratory (BML) in Seattle, WA. where: p volume resistivity, in ohm cm 2 per cm,

x The National Bureau of Standards, U.S. Department of Commerce. 121 ,

R = resistance in ohms at 20 °C figure 3 for a typical indirect reading of a particular length of type eddy current meter. Traceability to uniformly dimensioned non- ferrous metal

A = the cross-sectional area in square centimeters, and

L = the length being measured, for the resistance R, in centimeters we can solve for volume resistivity. Con- verting the area and length to microhm centimeters, and using the definition for 100 %IACS above, we get,

Figure 3. 172.41 (a constant) Indirect reading type eddy %IACS meter. unk volume resistivity (in micro- current ohm centimeters) NBS or any other primary standard was not available at this time. This is the general equation for finding Direct reading eddy current meters were employed later on the relative conductivity in %IACS for all non-ferrous metals from their dimensions (except on "B nuts") when conductivity standards covering the %IACS span of and resistance at 20 °C. interest, usually aluminum, were made available. Calibration of these meters with only "end of scale" conductivity 3. Historical standards can result in large mid-scale The first use of an A-C probe coil errors. The two standards usually pro- current method to measure the electrical conduc- vided on direct reading eddy meters claim neither accuracy nor trace- tivity of non-ferrous metals was made in ability. See figure 4 for a direct read- 1939 by German industry. Subsequent im- ing type eddy current meter and a set of provements in lift-off and sensitivity increased both repeatability and accuracy so that the eddy current meter finally came into its own as an important nondestructive testing tool in the early

1 1 960 s . The sorting of known non-ferrous alloys, mostly aluminum, and the verification of their proper heat treatment was now possible with speed and with moderate accuracy. Unknown alloys might have to be verified by spectroscopic means because of the overlap in conductivity values between heat-treated alloys of one composi- - OB J, tion and nonheat- treated alloys of a different composition. Once the alloy was properly identified, eddy current testing could take over the task of determining the correctness of heat treatment. Figure 4. Direct reading type eddy Boeing first used eddy current meters current meter. for crack detection in the late 1950' s. secondary conductivity standards. In late However, it was not until heat-treat I letter from Hawker- identification of special alloy hydraulic 1974, received a Si ddeley Aviation Ltd. of England indicat- fittings, called "B nuts", was urgently 2 percent IACS difference between required in 1962 because of stress corro- ing a U.S. and French Aerospatiale standards of sion problems that we began to use eddy the national scene, I current meters for heat-treat identifi- conductivity. On witnessed differences of nearly 1 cation of aluminum alloys. One such meter have percent IACS between Boeing secondary was a Magnaflux ED-500 which required the other U.S. manu- use of curves for conductivity values. See standards and those of 122 facturers. Both of these types of discrep- current and potential clamps, was made ancies are intolerable because of the pos- with an estimated accuracy of about 2 sibility of allowing improperly heat percent of reading at the normal lab tem- treated metals to be used in aircraft perature of 23 °C ± 1°. Dimensional area structures with the possibility of dan- measurements and the spacing between the gerous or fatal consequences. Figure 5 inner potential clamps were made by the shows three sets of Boeing non-ferrous Physical-Mechanical Section of BML A conductivity standards in carrying cases laser interferometer was used for mea- for 2,3, or 8 standards. suring the distance between the two inner clamp marks. The indeterminate position of these marks is why the accuracy was relatively poor. Later, length measurements made were far more accurate and used a laser to measure the distance between two very thin scratch marks on a soft aluminum bar. All later dimensional measurements were made at 20 °C ± 0.5 °C. Figure 6 shows the method used for thickness measurement, and figure 7 shows the laser interferometer method for determining the effective length.

Figure 5. Three sets of Boeing non- ferrous secondary conductivity standards.

4. Initial Standards Requirements

The need to make rapid non-destructive tests on incoming non-ferrous material has become increasingly important because incorrectly heat-treated alloys can fail in service and have even been suspect in some collapsed aircraft nose Figure 6. Thickness measurement. wheel accidents in the late 1950' s. In order to insure the accuracy necessary to properly categorize both the raw stock and finished material, BML was assigned the task of producing accurate non-ferrous conductivity standards which had traceability to NBS.

5. Preliminary Steps

The first involvement with eddy current conductivity standards at Boeing Metrology Labs came in 1966, when personnel of the Boeing Airplane Division (now the Boeing Commercial Airplane Com-

pany) brought some 1 in x 44 in x 1/8 in aluminum bars of various alloys to the Figure 7. Laser interferometer length primary standards laboratory to be mea- measurement. sured. These bars were produced as a result of the requirements of the The resistance measurement was made MIL-A-22771B government specifications on using an L&N six dial double ratio set, an aluminum forgings. A somewhat crude adjustable reversible, direct current measurement, using L&N List No. 4308 source, a sensitive null detector and any 123 one of three NBS traceable shunts: 0.01, 6. Primary Standard Bars 0.001, or 0.0001 ohm as required. Figure 8 shows these shunts in their oil bath. A To provide the required accuracy for separate, stirred, temperature-controlled working standards of conductivity, it was shunt oil bath, a double ratio set, and a first necessary to fabricate and certify specially designed primary standards primary bars from the best possible conductivity bar oil bath were used for commercially available materials using resistance measurement. The latter oil both known and newly developed techniques. bath was developed at a later date Copper, aluminum, bronze and titanium (approximately 1967) as part of the sheet stock 0.25 in thick approximately 2.0 in wide and 60 in long was cut and careful ly fabricated using as reference

Figure 8. Precision shunts in stirred oil bath with 6 dial double ratio set. Figure 10. Current and potential connection details. overall research and development program to produce traceable primary the ASTM Description B193 method for standards reference conductivity bars. volume resistivity. This is an absolute Figure 9 shows the conductivity bar method utilizing dimensions, resistance, and temperature for determining volume resistivity. Figure 11 shows 18 of the 19 primary standards conductivity bars in the shallow transfer oil bath. The lid is removed to show the bars in place. This bath is used for 100 kHz calibration of secondary standards which is described later on in this paper.

Figure 9. D-C primary bar calibration facility oil bath.

DC calibration fixture on a rack above its oil bath, ready for a primary bar to be inserted in place. Figure 10 shows current and potential connection details Figure 11. Primary standards conductivity with a primary bar in normal position. bars in shallow oil bath.

124 7. Primary Conductivity Bar Stability The following table 1 shows the changes in the dimensional and electrical When the eight primary conductivity values of the original eight primary bars bars were constructed in 1966, it was and the change in the individual certified envisioned at that time that there could values over an eight year period from 1967 be small but perhaps significant to 1975. The %IACS changes in the table alterations in the initial values due to are derived from differences between the dimensional variations caused by solid original 1967 DC values of conductivity state changes. Additionally, the (using resistance in ohms, length in resistivity of some of the alloys could centimeters and area in square also vary due to microstructural changes. centimeters) and the 1975 values of Errors such as grain direction and conductivity which include grain direction stratification, which was discussed in ISA and stratification corrections. Thus it paper 70-613, "Error Analysis of is an overall view of changes in certified Non-Ferrous Conductivity Standards," could values. Other comparisons are made later change in value with time and also cause on using uncorrected data. some change in the certified bar values. These latter two errors were not reverified since the original data was 8. Analysis of Changes taken, but care was taken in the selection and fabrication of the newest bars to If we discount the uncertainties for minimize these effects. In order to the moment and try to determine the com- determine what parameters had changed in bined effects of area and resistance the bars themselves and to decide how much changes on conductivity in %IACS, we can effect these changes had on the certified see that if conductivity values, it was necessary to 172.41 L measure all of the bars again both %IACS = RA ' dimensional ly and electrically under the same tightly controlled conditions as in the resulting conductivity values should the original calibration. change inversely as the area and

Table 1

8 Yr. Certified Value Material (alloy) Area Change Res. Change (DC) %IACS Changes

Copper +0. 127% -0. 055% -0 080

-0. -0. -0 Al. HOOF 023 4 098 003 -0 -0 Al. 6061 -0. 029 3 106 032 -0. -0. Al. 5052-0 +0. 005 o 044 012 -0. Al . 2024T4 +0. 094o 0. 000 149 a -0. Al. 2024 -0. 005 o +0. 105 072 -0. +0. Yel. Brass +0. 017 4 103 018 Titanium'3

a 2024T351 b See 75-17L for titanium. Original bar retired-too thin. NOTE: Area change values have an uncertainty of ± 0.08%, the resistance change values have an uncertainty of <±0.01% and the %IACS value changes, are calculated from both of the above plus temperature uncertainty and have an uncertainty of <±0.1%.

125 6

resistance. Or, stated another way, as 10. Comparison of Original and Latest Data the area or resistance increases, the %IACS should get smaller, all other terms The matching of magnitudes and di- remaining constant. If the value of rection now indicates the actual changes either A or R increases at the same rate occurring in the primary bars. A that the other decreases, the effect of comparison of the uncorrected bar changes both tend to cancel. An analysis of each are the changes due only to the area and bar follows in table 2. resistance changes follows in table 4.

It is significant that the errors due 9. Grain Direction and Stratification to grain direction and stratification have Effect changed originally assigned values by as much as 0.12 %IACS. These corrections One reason for the above discrepancy were not obtainable at the start of the is that a change was made in the aluminum program, but are now figured into the set values to correct for grain direction values of the primary bars adding to the and stratification. Since the original credibility of the certified secondary bars were cut with the grain direction (or standard values. More will be said about direction of roll), the %I ACS values tend secondary standard certification methods to be higher than actual because of the later in this paper. lower resistance values when measured along that axis. Tests were performed to determine the magnitude of this effect. 11. Additional Primary Bars (75-17 L) The corrections were calculated to be half of the difference between "with grain" and In 1968, a review of the Boeing "across grain" values for eddy current Company's requirements resulted in adding measurement purposes. Since the original eight additional primary standard con- values of conductivity were too high, the ductivity bars to the original eight bars. corrective action was to recertify the (The original thin titanium bar was not %IACS values to a proper lower figure. retired until 1970.) Most of these new Thus, all of the five aluminum alloys had bars were in the lower end of the con- from -0.02 percent to -0.12 percent IACS ductivity spectrum because of the large change added algebraically to the most amount of research being devoted to ti- recently calculated %IACS figures tanium fabrication in the SST (supersonic depending on the particular alloy. transport) project at that time. Three

Stratification corrections were generally bronze bars nominally 6.8 9 , 8.7 4 , and about one-third the value of the grain 16. 4 %IACS plus five titanium bars nominally 0.97 l-05 and direction corrections and, except in the 3 , 1,00a, 6 , 1.23s, case of aluminum HOOF, had the opposite 3.62 4 %IACS were processed and calibrated sign, tending to make %IACS figures higher as described in ISA paper 68-550. A in conductivity values. Taking all of similar error analysis was performed on these factors into consideration, we then the 75-17L bars and showed a maximum A% find the following result in table 3a and of %IACS value of -1.14a. 3b.

Table 2

Material (alloy) Area. Res. Calc. Change 8 Yr. Cert. Value Change

Copper Lower Lower 3 Al. HOOF Higher Lower 3 Al. 6061 Higher Lower 3 Al. 5052-0 Higher Lower

Al . 2024T4 Lower Lower

Al. 2024 351 Lower Lower Yel. Brass Higher Higher

3 0pposite effect than predicted. 126 )

Table 3a

Pri. Bar %IACS %IACS %IACS a Material AC corrections DC(new)

Copper 101.2 0(est. 101.2 Al. HOOF 60.16 -0.06. 60.23 8 0 Al 6061 41.57 4 -0.06c 41.64, Al 5052-0 35.52, -0.02- 35.54~

Al 2024-T4 30.60^ -o.nj 30.72 0 Al 2024T351 29.83 t -0.03- 29.87, Yel. Brass 27.24, 0(est.) 27.24,

^IACS corrected values are certified from 60 kHz to 200 kHz after adjusting for grain direction and stratification effects.

Table 3b

Pri. Bar Material %IACS DC (Orig) %IACS DC Diff.

Copper 101. -0.08 2g Al. HOOF 60.17 +0.06 ] Al 6061 41. 60, +0.04

Al 5052-0 35. 53 +0.02 ] Al 2024-T4 30.75, -0.03.

Al 2024T351 29.88 -0.01 f Yel. Brass 27.23, +0.01

Table 4

(Table 3) (Table 1) a Pri. Bar Material Uncorr. Changes Area Res. Changes Diff. %IACS

Copper -0. 08%IACS +0 07%IACS 0 01

Al. HOOF +0. 06 -o. 12 0. 06

Al. 6061 +0. 04 -0. 13 0. 09

Al. 5052-0 +0. 02 -o. 04 0. 02

Al. 2024 T4 -0. 03 +0. 09 0. 06

Al. 2024T351 -0. 01 +0. 10 0. 09

Yel. Brass +0. 01 -o. 09 0. 08 a Evidence of the repeatability of readings taken with the facility.

127 .

12. Filling in Some Gaps in %IACS from DC to AC values. The fourth area Standards Values which required at least one additional primary standards bar was between the 60.2 Even though the low end of the con- and 101.1 %IACS standards. A conducting ductivity spectrum seems to be quite bronze bar was fabricated in the same complete, especially in the titanium manner as previously described in the range, it was necessary to add a new bar references, and now gives an additional so that eddy current meters having a range certified conductivity value in that area of 0 to 3.5 %IACS could have a meaningful at 85.39 %IACS. None of the 4 new bars calibration. A review of the titanium were tested for permeability, based on the standards shows the highest value to the results of the original tests on the first 3.6 %IACS which is off scale, and the next 16 bars, which showed negligible effects lower value is 1.2 %IACS which is too far from permeability, except copper nickel down scale to be significant when used by bronze. itself or with lower values in %IACS. A new bar having a value of 3.3 7 %IACS has The results of the calibration of the been fabricated and put into the set of 19 four new non-ferrous conductivity bars are total located in the transfer oil bath as follows in table 5. previously shown in figure 11.

The portion of the aluminum con- 13. Improved Accuracy Transfer to ductivity spectrum existing between the Secondary Standards 41.6 %IACS and 60.2 %IACS standards required at least 2 more standards to The original method developed to produce better results for eddy current determine values of secondary conductivity meter calibration and for curve-fitting standards utilized an a-c impedance the data better as explained later on in bridge, modified to include a guard this paper. Unfortunately, although two circuit, lift-off compensation and various different alloys of aluminum, 6061-0 and probe compensation capacitor settings. The 2024-0 were processed, the values turned frequency was fixed at 100 kHz ± 1.0 Hz out to be within about 0.2 %IACS of each and power entered the bridge circuit at a other at about 47.5 %IACS rather than the voltage somewhat less than 10 volts rms. nominal 5 percent difference expected from An a-c null detector indicated the bridge table of physical properties used for the balance condition and a Wagner ground was selection process. included for precise and repeatable detector balance. See figure 12 for the As a result of previous experience schematic of the bridge circuit. and analysis, both aluminum bars were selected in the -0 condition for least The initial method of experimenting stratification and were then cut at 45° to with slope curves in pFAIACS values vs. the direction of roll, effectively negating %IACS, with changing probe compensation the effect of grain direction. This means capacitance values in segmented areas and that the AC values assigned to each bar with weighted averages for determination will be the same as the DC values, and no of secondary standard conductivity values corrections are required for these 2 was adequate when comparing like or very potential sources of error in the transfer close values of conductivity. It failed

Table 5

a Material (alloy) Conductivity Temp. Coeff of Resis/°C

Conducting Bronze 85.39 %IACS 0 003 2 Al. 6061-0 47.64 0. 003 3 2

Al. 2024-0 47. 46-, 0. 003 3 Titanium 55 3.37 0. 003 6 6

The temperature coefficient of resistivity per °C is accurate to ±0.0002 between 15 and 35 °C also called TCF or temperature correction factor.

128 to deliver the expected accuracy, however, only on the pF values at balance. Only when the known values were over two one fixed probe compensation setting was percent different in %IACS from the used. Corrections to each segment, de- reference standard values. Deviations pended upon how close to the primary were discovered when one primary standard standards values the unknown standard was bar was used to verify another primary and corrections were automatically standard bar several %IACS different from applied. The results were well within our the first. The ensuing investigation anticipated values of accuracy when showed that the entire premise on which calibrating one primary standard against transfer of primary bar accuracy to any other primary standard in a particular secondary standards was based, with segment. allowance for bridge errors, was somewhat less accurate when the two standards, The six segments and the method of reference and unknown, differed by more curve fitting applied is shown below. than 2 percent. With some of the secondary standard nominal values several 15. Development of Curve Fitting percent away from closest reference Method Details standard, certifications within ±0.35 %IACS of the stated values required Interpolation formulas were derived re-examination. using curve fitting techniques based, in

part, on programs in the Stat. Pac 1 of the Hewlett Packard HP-65 programmable calculator. Since it had been previously observed that if the bridge readings in pF were plotted with the %IACS on log- log graph paper, the resulting points were nearly in a straight line; logarithms of the bridge readings and %IACS values were used in the curve fitting process. An attempt to determine a single overall formula for the entire range from 0.96 percent to 101.2 %IACS showed that errors would be too large. Therefore, the %IACS CONDUCT' /IT* SPiOCt values were divided into segments as fol lows:

Figure 12. (1) 0.96% to 1.23%

(2) 1.2% to 3.6%

(3) 6.8% to 27.2% 14. Curve Fitting Method (4) 27.2% to 41.6%

After analyzing several different (5) 35.6% to 47% approaches, it was decided to segment the (6) 41.6% to 101.2% conductivity spectrum into six parts and use curve fitting techniques, with an H-P 65 to determine the unknown values based

Segment No. Conductivity Limits %IACS Curve Fitting Method

1 0.96 to 1.23 Mod. Pwr. Curve

2 1.2 to 3.6 Mod. Pwr. Curve

3 6.8 to 27.2 Linear Regression 4 27.2 to 41.6 Mod. Pwr. Curve

5 35.6 to 47 Mod. Pwr. Curve

6 41.6 to 101.2 Mod. Pwr. Curve

129 !

A different formula was derived for 17. Secondary Conductivity Standard each segment. Linear regression proved to Calibration Improvements give an acceptable fit for the data for segment (3); the other segments were curve A new oscillator with improved output fitted using a modification of the power voltage and better stability char- curve. These formulas, as applied here, acteristics which can now deliver a steady are as fol lows: 10 volts rms to the bridge circuit at 100

kHz ± 1 Hz has replaced the original oscillator. The results have been much Linear regression: more consistent readings from the primary standards bar as well as a slight improve- log %IACS = A + A log pF ment in detector sensitivity due to Q ] somewhat higher voltage. See figure 13 Power curve (modified): for a view of the probe and the bridge circuit control console.

log pF 1 log %IACS - where A A a, b are 0 , 1( and coefficients determined by the curve fitting process and "pF" is the corrected bridge reading in pF.

16. Programming the Calculator

These coefficients are then incorporated into HP 65 programs--one for each segment—which are used to determine conductivity of standards tested. In some cases, additional corrections are incor- Figure 13. 100 kHz conductivity transfer porated to reduce errors at points on the bridge and probe. interpolation curve close to the conductivities of the standard bars. When coupled with the new curve Since the %I ACS is determined by the fitting techniques, the accuracy improve- formulas in terms of only one variable, ment in actual bridge reading values the bridge reading in pF, no "standard reduces the uncertainty of %IACS accuracy reading", as such, is necessary. In limits from an estimated +0.15 percent to practice, however, a bridge balance is approximately +0.07 %IACS. Adding this made for a standard close (in %IACS) to uncertainty to the +0.2 %IACS, which is the value of the "unknown" to account for now thought to be quite conservative based any drift in the capacitance elements in on historical performance evaluation, we the bridge. The difference between the feel confident of the stated ±0.35 %IACS, standard reading taken at that time and or 1 percent of reading accuracies, the standard reading used to determine the whichever is less, assigned to secondary interpolation curve is shift then used to non-ferrous conductivity standards the unknown reading by a like amount produced by the Boeing Metrology before applying the interpolation formula. Laboratories. Figure 5 showed some types of secondary standards in their containers The HP-65 curve fitting programs used after being calibrated against the primary also provide an additional parameter r 2 , bars in the transfer oil bath. or "goodness of fit." It was found that this number must be quite close to 1; 0.999 or greater in most cases to give 18. How Good Are Eddy Current satisfactory results. This is probably due Meter Standards? mostly to the scale compressing effects of logarithms. From what has been discussed previously, it can be seen that unless some

130 )

unknown factor has not been considered, the Using these secondary conductivity Boeing secondary non-ferrous conductivity standards serves to guarantee the accuracy standards which are categorized to be of direct reading eddy current meters by working standards for conductivity, are verifying the scale tracking in the area of well within the ±0.35 %IACS or 1 percent of interest. Although, for example, several value uncertainty assigned to them. scale points in the most frequently used Periodic recal ibration and recertifi cation aluminum conductivity range, 28 to 60 of the Boeing primary standard bars and the %IACS, are compared to the standards in annual in-oil recertif ication of the that range, this does not assure the secondary standards using the 100 kHz accuracy of scale indication outside of conductivity bridge transfer oil bath shown that range. If the entire scale is to be in figure 13 will keep the secondary utilized from the low or titanium range up standards within the assigned accuracy through copper, certified secondary con-

1 imits. ductivity standards covering the entire

range from 1 to 100 %IACS should be used. It has been Boeing policy to resurface Without such standards the indicated secondary standards which are received for values of either direct or indirect eddy recal ibration in such condition that lift- current meters are questionable. off errors can exist as a result of excessive wear. Some slight drift character- Several manufacturers are now in the istics have been observed in secondary business of providing certified conduct- standards over the 10 year period, but ivity standards for use with eddy current recerti f ication keeps them well within meters. A few of the types produced by their uncertainty limits for the one the Boeing Company were shown in figure calendar year cycle assigned to them in- 5. Several other configurations have been house. Commercial customers usually made to satisfy internal requirements, but observe longer cycles to suit their needs basically the accuracies and non-ferrous or internal cycle periods. materials are the same as with the standards forms.

Error Analysis

Primary Standard Bars (used to calibrate secondary coupons)

Component Max Expected Error Remarks

DC Ref. Resistor ±0. 005% Traceable to NBS Thomas 1 ohm Double Ratio Set ±0. 001% Galvo Readability ±0. 01% At maximum sensitivity Thermometer ±0. 012% 19° to 21 °C Thermocouples in bath ±0 0016% Galvo reads opposing T.C. DC Sys. Instability ±0 04% Temperature of bath oil Conductivity Material ±0 054% Non-homogeneity Stratification Error NA Corrections Made Grain Direction Error NA Corrections Made

Repeatabi 1 ity <±0 1% Total RSS error = ±0.12% of reading in ohm cm @ 20 °C (cert to ±0.20 % IACS or 0.5% of reading, whichever is less at 20 C)

Secondary Standard Coupons

Component Max Expected Error

Primary bar basic error ±0.12% of reading 100 kHz bridge error <±0. 10% of reading Probe error (position) <±0. 10% of reading Curve fitting error ±0.05% of reading Aging and wear (1 yr. cycle) ±0. 10% of reading (max. Temperature uncertainty ±0.05% of reading = ±0.22% of reading Total RSS error . C) (Cert, to ±0.35 %IACS or 1% of reading, whichever is less at 20

131 19. Conclusions [3] Jones, A. R. , Error Analysis of Non- Ferrous Conductivity Standards, pre- The preceding error analysis indi- sented at the ISA International Con- cates that the certified accuracies ference and Exhibit, Philadelphia, assigned to the primary standard Pennsylvania, October 1970 (70-613). conductivity bars for a 2 year re-certification cycle and the secondary [4] The Boeing Company D6-4813, Material/ standards for a 1 year cycle are Temper Identification - Aluminum reasonable. Coupled with this, our Al loys. experience has shown minimal differences in the primary bar values due to changes in solid state structure and other aging [5] ASTM Designation B193-65, Resistivity factors. of Electrical Conductor Materials (Standard Method of Test). The fact that all the measurements are basic parameters which have been in [6] NBS Copper Wire Table, No. 31, p. 75. existence for decades gives us confidence in our ability to provide indirect trace- [7] Libby, H. L. , Introduction to Electro- ability to NBS basic standards of magnetic Nondestructive Test Methods. resistance, length and temperature.

Adherence to good machine shop practices, [8] Jones, A. R. , Non-Ferrous Conduc- careful screening of materials, tivity Standards - A Ten Year Review, utilization of the latest measurement presented at ISA International Con- techniques and, above all, careful ference and Exhibit, Houston, Texas, attention to minute details in both October, 1976. constructing and measuring these standards has incorporated a high order of built-in [9] Jones, A. R. , Eddy Current Meter accuracy. Standards, How Good Are They?, presented at ASNT Spring Conference, At present, NBS in Gaithersburg, Phoenix, Arizona, March, 1977. Maryland, is preparing to construct non-ferrous conductivity standards. They Discussion indicate that comparisons with these standards are still about a year away. Question (Dr. Mc Master): I have found Discussion with NBS personnel indicate there are measurable eddy currents up to that an approach similar to the one used three times the diameter of the coil. at Boeing will be used by them both in Every time you measure something smaller construction and in measurement of their than that, you are clipping the corners. conductivity standards. Have you found much error due to the ratio of coil diameters, or have all coil The only additional research proposed diameters you have tested been so small in this field at Boeing is to build about that you are nowhere near the sample edge? four more primary standards bars this year to increase the accuracy of the curve (Mr. Jones): You are talking about edge fitting technique in some of the segments effects? where the spread between any two of the primary bars is larger than deemed suit- (Dr. Mc Master): Yes. able. Answer (Mr. Jones): We measure in the center of the secondary standards with a small coil and try to get somewhere near References the middle of the primary bar. Also, we always measure on the same spot on the

[1] Jones, A. R. , Bulk Conductivity Mea- primary bar. surement Standards, presented at ASTM

Annual Meeting, Boston, Mass. , June 1967.

[2] Jones, A. R. , Non-Ferrous Conduc- tivity Standards, presented at ISA Annual Meeting, New York City, October, 1968 (68-550).

132 .

National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy Current Nondestructive Testing held at NBS, Gaithersburg, MD, November 3-4, 1977. Issued January 1981.

NBS EDDY CURRENT STANDARDS PROGRAM

George M. Free Electrical Measurements and Standards Division National Bureau of Standards Washington, DC 20234

1. Introduction From the relation J = aE in which J is the current density, o the electrical conduc- The present goals of the NBS eddy tivity and E the electric field, the con- current program are twofold: the creation ductivity of a uniform bar may be written of an electrical conductivity calibration service for non-ferrous metal standards, L A = cross sectional area of a = and the development of non-ferrous metal RA material Standard Reference Materials to be issued L = distance between potential as conductivity standards. The calibration contacts service will provide measurements of the R = resistance electrical conductivity of standards sent to NBS by industry and issue reports on if R and A are considered to be average these standards. The Standard Reference values over the length L. If the bar is Material Program will make available to only slightly irregular, this relation industry electrical conductivity standards will hold if a, dR/dx, and A are consid- suitable for use in calibrating eddy ered to be slowly varying functions over current instrumentation. the length of the bar.

In order to establish the NBS 1/A = 1 a = electrical conductivity calibration dR dR 1 service instrumentation and methodology dx dx 1 must be developed in several areas [I] . Primary conductivity standards consisting where A is the cross-sectional area at a of long metal bars are being developed specific point on the bar and dR/dx is the which will be measured using dc techniques slope of the resistance vs. length curve to determine the conductivity of the at a specific point. In approaching the material. Using eddy-current techniques, determination of conductivity in this customer conductivity samples will then be manner, two sets of information are nec- compared with the primary standards to essary. A mapping of the cross-sectional arrive at the conductivity of the sample. area along the length of the bar and a mapping of the resistance of the bar. The Work has been developing along two two sets of data will then appear as in

lines, design and construction of dc figure 1 measurement apparatus, and design and construction of ac measurement apparatus. To determine a, the slope of the re- These two measurement systems will be sistance curve at a point will be calcu- discussed in more detail. lated and the area of the bar at that point will be determined. Dimensional measurements of the bar will be done by 2. The DC System another group within NBS. The dc measurement of resistance will be done using a Since eddy current measurements are current comparator as described below. made at a point, or over a small area, the NBS dc measurement is designed to The operation of the current compar- determine conductivity at a specific point ator is shown in figure 2. A current is

1 Figures in brackets indicate the literature references at the end of this paper.

133 "

of the bar. The separation distance between potential points will be measured by a laser interferometer. Thus, a series of measurements of resistance can be made along the bar with a minimum distance of

separation of 1 cm. The total length of the bar will be approximately 1.5m with a cross-section of 50 mm x 6 mm.

The measurement will be done in an oil bath which is stable to about ±3 x 10 3 °C. The range of temperature possi- ALONG BAR ble in the bath is 18 °C to 30 °C. With Figure 1. Determination of (dR/dx) and temperature, dR/dx, and cross-sectional (A) at a specific cross-section of metal area known, the conductivity can be cal- bar. culated. The bars will then become the primary standards of conductivity at NBS.

CIRCUIT 2 CIRCUIT 1 3. AC System u UTAL will be BAR The eddy current measurements

D,C. supply done on a bridge as seen in figure 4. The D.C. SLAVE 1 „ 1 major components of the bridge are the three inductive voltage dividers, a variable frequency power supply and a phase sensitive detector. The bridge is de- nm 1r rnV> uuuj| signed for use in three general ways. In the positions of the two test coils on the bridge diagram, the following may be Figure 2. Current comparator schematic placed: with metal bar in place. (a) Coil and capacitor. Used to cali- established in circuit (1) by the main brate inductance in terms of capac- power supply. The primary and secondary itance standards. currents separately pass through a number (b) Two coils used to make the measure- of turns wrapped on a common core, the ments in an absolute sense. One coil turns being wound in opposition; a flux will be in air while the other coil sensor detects any residual flux in the will be placed on the metal sample. core and adjusts the slave power supply so (c) Two coils used to make the measure- that zero flux exists in the steady state. ments in a relative sense. Both

The I N = I N . The turns ratio N coils will be placed on the same X X s s s can also be varied to achieve a null at the piece of metal for a zero reading, detector. At the point of balance I R = then one coil will be placed on the

I R or R = R M /N . Then tlfe fin- unknown sample. The change in con- 5 s known resistance i s l&ow^ in terms of a ductivity will be determined from the standard and a turns ratio. change in the bridge balance conditions. During the resistance measurement, TEST COILS the metal bar will rest on a modified optical bench, as in figure 3. Current will be injected at (A) and leave at (B); the potential contacts are at (1) and (2). Potential contact (1) will remain fixed while (2) will be varied along the length

POTENTIAL CONTACTS

m (2) C) <2">

I PETAL BAR \ [ I » I 1 III , , _J ^ '"

I I II I I

/ OPTICAL BENCH _^

NETWORK Figure 3. Motion of potential contacts along length of metal bar. Figure 4. Eddy current bridge 134 The actual conductivity of an unknown 5. Other Areas of Study will be obtained by interpolating the experimental results for measurements on two Coils are being constructed using the primary standards and the unknown, using relations derived by Dodd et al . [2]. primary standards having conductivities Values for the impedance of a coil on a above and below the unknown. conducting flat surface have been calculated. These results are being used to 4. NBS Services achieve optimum performance in the coils constructed, i.e., maximum sensitivity Reports on the calibration of a cus- to change in conductivity and minimum tomer's sample are envisioned as follows. change in impedance for lift-off. The Besides a written report, a graph will be relations will also be used to establish given as seen in figure 5. Conductivity the theoretical shape of the impedance will be shown as a function of curve between points of known dc con- frequency and as a function of temperature ductivity, i.e., those points for which for a specific range of frequencies. The primary standards are available. An total uncertainty assigned to the test attempt will be made to correlate will be a function of the metal sample measured coil parameters with the theo- that is being calibrated, i.e., its retical values. If the agreement uniformity. For a uniform metal sample the between actual measured values and theo- total uncertainty should be about 0.1 retical calculations is sufficiently percent IACS. close, a quasi-absolute determination of conductivity can be made.

A second experiment that has been briefly investigated but which will be

IACS pursued further is based on the theorem of Van der Pauw [3]. This relation is commonly used in semiconductor technology to determine the dc resistivity of a sample of uniform thickness.

In this theorem, the following relation is derived:

^nRid/p -'nR 2 d/p _ + e 1

1 1 -i 1- -i 1 h FREQUENCY Hz Resistivity becomes a function of two re-

sistance measurements, R x and R 2 , and a measurement of d the thickness of the sample. The advantages are obvious. Figure 5. Information on conductivity sample to be included in test report. All of the above work will be directed towards the goals stated at the Although plans for constructing the beginning of the paper, the establishment Standard Reference Materials are not yet of precise standards of electrical con- finalized, there are several directions ductivity which are usable by industry. that could possibly be followed. First is The present goal of the program is to the use of powder metallurgy or some other have the electrical conductivity cali- metallurgical technique to guarantee that bration service available in the Fall of the sample is indeed uniform. A cali- 1978. The first SRM's will be issued bration report as above would be issued in 1979. with the sample. Another possibility is to use the most common metals and alloys References for standards. If common metals are used, the need for temperature corrections would [1] Jones, Arthur R, "Electrical Conduc- be eliminated since the metal to be tested tivity Measurement Standards," Pre- would be at the same temperature as the sented at ISA Conference, Boston, standards used to calibrate the eddy 1970. current meter.

135 [2] Dodd, C. V., Deeds, W. E. , Luquire, Question (Mr. Wehrmeister): What type of W. Integral solutions to some you J. , geometry in test samples will be ac- eddy current problems, International cepting. Will they be flat, or can I send Journal of Nondestructive Testing, 2> you a piece of tubing? 29-90 (1969). Answer (Mr. Free): The samples, at least [3] Van der Pauw, J. J., A method of to start out with, will be flat, geomet- measuring specific resistivity and rically a flat sample. Hall effect of discs of arbitrary shape, Philips Research Reports, 1_3, Question (Mr. Jones): When you are making

No. 1 (February 1958). your resistance measurements, what type of spacing will you have between the potential probes? Discussion In other words, say you start near Question (Mr. Wehrmeister): I am curious the top, do you take ten readings and the to know what sort of tolerances you are move a quarter of an inch, and then take putting on your readings in terms of ten more readings to get the uniformity percent IACS? And what tolerance is across the width of your standard bar industry working towards? which you said was two inches?

Answer (Mr. Free): In this experiment, we Question (Mr. Free): Do you mean to get are hoping to achieve an uncertainty of .1 the mapping of the cross-sectional area, percent IACS. By tolerance, do you mean or the mapping of the resistance? the uncertainty in measuring conductivity in a sorting operation? Answer (Mr. Jones): The mapping of the resi stance. Question (Mr. Wehrmeister): How accurately do they have to obtain a reading or a Answer (Mr. Free): Initially, there will measurement? be two measurements.

Answer (Mr. Free): As I understand it, Question (Mr. Jones): Somewhere near the when running a characteristics test on middle? incoming metal, it is around 4 or 5 percent. Answer (Mr. Free): Two points, somewhat off center, but on either side of the Comment (Mr. Jones): Usually, the pre- center. cision should be plus or minus .5 percent, IACS. For example, if you have an allowed band of 31-1/2 percent to 33-1/2 percent, you would have to hold that to 32 percent or 33 percent, but 15 percent would be used up in your measurement uncertainty, considering the conductivity standards being used.

On the primary bars, we try and get to .2 percent plus or minus half percent reading, whichever is less.

Comment (Mr. Lagin): One thing which I think industry would like to see come out of NBS in the future is a standard representing various radii of curvature. Usually we have to take conductivity measurements on curved surfaces, and since the probe has finite dimensions this introduces error. There is a large percentage of error when the parts have a radius of curvature of less than half an inch. The conductivity measured could be in error by as much as 10 percent.

136 National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy Current Nondestructive Testing held at NBS, Gaithersburg, MD, November 3-4, 1977 Issueducu January 1981. '

ELECTROMAGNETIC THEORY AND ITS RELATIONSHIP TO STANDARDS

Arnold H. Kahn and Richard D. Spal 1 Center for Materials Science National Measurement Laboratory National Bureau of Standards Washington, DC 20234

as the dimensions of the obstacle. In 1. Introduction this contribution to the Workshop, we report on calculations of the eddy cur- The interpretation and derivation of rent and impedance changes associated the maximum information from electro- with a surface crack in a plane slab and magnetic testing and the optimum design on a cylinder. In the case of the cyl- of testing methods depends on principles inder, impedance diagrams are given, and found in classical electromagnetic theo- the modifications due to the crack are ry. The value of an approach based, in demonstrated. part, on a theoretical grounding is well demonstrated in the discussion by Libby 2 [l] . In the development of standards, 2. Crack on a Plane Slab the application of theoretical analysis can be used to base the standards, at Our model for a crack [4] assumes least in part, on physical laws. We hope that there is a long, thin gap in the that discussions at this meeting will conducting material. We treat the gap as suggest areas where theoretical efforts preventing the flow of any normal com- may be profitably undertaken. ponent of electric current density. The magnetic field for which we perform the As one of the NBS activities in calculations is applied tangential to the electromagnetic NDE, we have performed surface and parallel to the crack. The analyses of the nature current of eddy geometric configuration is shown in distributions in the vicinity of a crack figure 1. In the figure, the width of in conducting material. The purpose of the crack is exaggerated so as to show these investigations is to provide pre- the unrestricted penetration of the dictions of the changes of signals in 1U)t applied a.c. magnetic field, H e testing apparatus due to cracks, or Q , alternatively to aid in the characteri- into the crack. The lines with arrow zation of defects. heads indicate schematically the direction and path of the current at some instant Literature searches on this subject of time. The solution to this problem is have yielded only a few theoretical moderately difficult and much can be treatments of the modification of eddy learned by breaking it into two component currents due to the presence of defects. problems, the corner and the tip, as The treatment of Burrows [2] and Dodd, et shown in figure 2. These can be solved al., [3] considers the perturbation of exactly. The results may then be com- the eddy current pattern by an ellip- bined to give the results for a crack soidal inclusion. However, in this provided its depth is greater than four treatment the inclusion had dimensions skin depths. which were small with respect to the electromagnetic skin depth. However, for The equation to be solved is [1,4] greatest sensitivity of detection the (V2 +k 2 )H = 0 (2. frequencies must be such that the skin , 1) depth is of the same order of magnitude

department of Physics, University of Pennsylvania, Philadelphia, PA 19104.

2 Figures in brackets indicate literature references at the end of this paper.

137 ,

where the complex propagation constant, k, is

' k = (iouiM) . (2.2)

In the above equations, H is the field inside the conductor, a is the electrical conductivity, w is the angular frequency of the applied field, and u is the permeability of the conductor. At the surface of the material, and in the crack gap, the field must take on the value of

10 the applied field, H e . In terms of Q the electromagnetic sKin depth, 6, given by

,/£ = 6 (2/ou)u) , (2.3)

Figure 1. Schematic drawing of eddy currents in the vicinity of a surface the propagation constant may be expressed crack in a slab of conducting material. as The a.c. magnetic field applied at the surface is normal to the figure and uniform in space. k = (l+i)/6 . (2.4)

In order to calculate changes of power a) INFINITE CRACK dissipation and energy stored, it is not Y necessary to map the field; we need only know the current density at the surface, which, in turn, depends on the normal derivative of the field at the surface. The results are conveniently expressed in 7. terms of the complex Poynting vector,

= S l/2(ExH*) ,

b) RIGHT-ANGLED CORNER where E and H are the electric and mag- V netic fields at the surface of the material. The real part of S gives the power dissipation per unit area and the imaginary part gives the energy stored [4,5]. (In the next case, the cylinder with a crack, the Poynting vector leads directly to the complex impedance.)

In figure 3, we show the normal component of the Poynting vector, S, as a function of distance from the tip of a deep crack. The values of S are nor- Figure 2. Two solvable problems related appropriate to malized to S 0 , the value to the crack of figure 1. In the plane surface in the absence of a crack. infinite crack, a magnetic field, We see that the greatest dissipation and 1ujt

H e , is present in the gap as far reactance changes occur near the tip and Q as the tip. In the right-angled corner that at a distance beyond 1.56 from the problem, the field is applied to the tip fields are "back to normal." A boundary, parallel to the corner. similar situation occurs at the corner, as Arrows indicate the current flow. 138 is seen in figure 4. The loss at the tip are greater than those of equivalent corner vanishes, but the "range" of the flat surface, while those at the corner corner is about 2.56. The losses at the are less. In effect, the eddy currents take a short cut at the corners and a long trip around the tip. The net dissipation, per unit length of crack, for a 1 crack of depth d, with d > 46 is

2 (2d-0.786) |H /(2a6) . | (2.5) Q

There is an effective shrinkage of the depth, relative to equivalent flat surface, of approximately 0.396. For cracks more shallow than 46, a direct solution is necessary. This will be done in the fol lowi ng section.

3. Surface Crack on a Long Cylinder

lm S In this section, we present the results of calculations of the eddy currents in a long cylinder with a radial crack at the surface. The calculations 0.0 0.5 15 DISTANCE FROM TIP. */8 were performed for arbitrary depth of crack, and the results were developed in terms of an infinite series of trigono- metric functions and cylindrical Bessel Figure 3. Plot of the normal component of functions. The results can be made the Poynting vector on the surface of an visual izable by means of the impedance infinite crack at a distance, x, from the diagram. tip. In the plot the Poynting vector is In figure 5 we show the familiar normalized to ReS 0 , where S 0 is the Poynting vector for plane surface in the impedance diagram for a solenoid enclos- absence of a crack. Distance is in units ing a cylinder (with a 100 percent fill- of 6, the skin depth. ing factor) [1]. The plot shows the imaginary versus the real parts of the impedance of the coil. A point on the curve corresponds to a particular value of the ratio of cy linder radius to skin depth, a/6(= aVou)u/2).

The effects of cracks of varying depth are now shown in figure 6. Our first observation is that the presence of a crack does not change the shape of the curve; it only shifts the position on the curve of each a/6 point. For an initial attempt to understand the phenomena, we calculated the impedance curves for cracks of depth d = 2a, a, and 0.5a. We have plotted these curves, shifted in space in the figure. Equal values of a/6 on the different curves are connected by dashed lines. It is evident that the greatest shift of impedance caused by the ~ Figure 4. Plot of the normalized Poynting presence of the crack occurs for a/6 vector, S', on the surface of a square 1.7, where the curve has a vertical corner, at a distance, x, from the tangent. corner. Distance is in units of 6, the ^kin depth. Figure 5. Impedance diagram for a long cylindrical solenoid with a conducting Figure 6. Impedance diagrams for a long core which completely fills the solenoid. cylindrical solenoid with a conducting The real and imaginary parts of the core which contains a radial surface impedance are normalized to ujLo, the crack. Impedance curves are plotted, reactance of the empty solenoids. as in figure 5, for three selected Points on the plot correspond to depths of crack. The dashed lines particular values of a/6, the ratio schematically indicate the shifts of of core radius to skin depth. a/6 points as the crack depth, d, is varied in the calculations. These calculations have not yet been carried out for very shallow cracks because of slow convergence of the References series. To cope with this difficulty, we

have developed as an alternative a [1] Libby, H. L. , Introduction to variational approach which should yield Electromagnetic Nondestructive Test greatest accuracy for shallow cracks and Methods (Wiley-Interscience, New which will overlap the region treated by York, 1971). the series method. This work is now in

progress. [2] Burrows, M. L , "A Theory of Eddy Current Flaw Detection," Disserta- A provisional estimate for the tion, University Microfilms, Inc., effect of the crack can be given. If the Ann Arbor, Michigan (1964). frequency is chosen so that a/8 a 1.7

(near the maximum dissipation), then we [3] Dodd, C. V., Deeds, W. E. , and

expect the shift of the a/6 to be ~ -0.4 Luquire, J. W. , Int. J. of Non- times the value of d/a. For these values, destruc. Testing 2, 29 (1969). the impedance change is imaginary and

corresponds a , R. , and Feldman, to shift of the amount [4] Kahn, A. H. , Spal A., J. Appl. Phys. (to be published).

140 .

[5] Collin, R. E. , Field Theory of Comment (Dr. Birnbaum): Well, then I Guided Waves (McGraw-Hill Book Co., agree with Libby, it defies intuition. New York, 1960). One would expect that as you vary the frequency, the skin depth becomes the same order as the crack length, and then Di scussion when it becomes larger you would get some more drastic differences in impedance. Question (Mr. Libby): I understood you to say that you found no difference in Comment (Mr. Kahn): No, I find it just the shape of the curve for these con- moves it on the curve. ditions?

Answer (Mr. Kahn): We found slight differences which we believe are computer error.

Comment (Mr. Libby): I am rather surprised at this. What it indicates, looking at the curve, is that the effect of the crack would be the same effect obtained by lowering the frequency.

Answer (Mr. Kahn): Right.

Comment (Mr. Libby): I would expect when the currents are flowing in the usual way about the crack, there would be some component that tells the coil that we are dealing with a bar having a smaller radius. I am just wondering why this is not the case.

Question (Mr. Kahn): You are saying it is not?

Answer (Mr. Libby): I would expect the effect to show up in the size of the curve. I would expect that where you have a curve like the one shown that with a smaller diameter it would shift down and the curve would be in a different place on the complex plane. This is what I would expect.

Comment (Mr. Kahn): I normalized the curve, it is normalized to the inductance of the empty coi 1

Question (Dr. Birnbaum): Perhaps I did not understand. You based the calculations on what assumption, that the skin depth was much larger than the crack depth?

Answer (Mr. Kahn): No, it is arbitrary. The whole curve marks out an arbitrary range of the skin depth from zero to infinity.

141

National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy Current Nondestructive Testing held at NBS, Gai thersburg , MD, November 3-4, 1977. Issued January 1981.

DISCUSSION NEW SCIENCE DIRECTIONS AND OPPORTUNITIES CONCLUDED FROM WORKSHOP SESSIONS

Robert E. Green, Jr. Johns Hopkins University Department of Mechanics and Materials Science Baltimore, MD 21218

In the first talk of the Workshop, by IV. Environment Dr. Mc Master, the point was made that much of the eddy current instrumentation and Thermal test methods existing today has not Radiation advanced beyond principles established in 1830. Eddy current development work has V. Other Methods used the new technologies available only to refine and improve systems that have Microwave years. New approaches been used for many Triboelectric to basic problems are almost nonexistent. Beyond this, much of the information that Figure 1. General areas for research and is present in an eddy current test is development work. completely ignored.

Although I have not listed the var- Listed in figure 1 are general areas ious items in any special order, we def- in which research and development work initely have to consider the material could be done. These areas are meant to aspects of the part to be inspected. In a be independent They are of standards. research laboratory, we can tailor make areas which could be important for materials so that a- given physical prop- scientific as well as technological erty can be varied in a systematic manner progress. these areas, Your comments on while holding other properties constant. or additions you might make are Thus, in the laboratory, we can use spe- important. cially selected materials to test the sensitivity of eddy current instruments to I. Specimens a variety of material parameters. On the other hand, in field eddy current testing, Ferrous we have to work with materials which have Nonferrous usually been optimally designed to possess Homogeneity desirable mechanical properties, but which Anisotropy may possess extremely complex electrical Geometry and magnetic properties and, therefore, Multilayer present special problems for eddy current testing. Thus, we must consider, as has II. Coils been pointed out several times, whether the material is ferrous or nonferrous Geometry and probably even more critical with Orientation respect to eddy current measurements is Arrays the inhomogeneity of real structural Other Detectors materials. Computer Involvement In my own opinion, a good eddy cur- III. Displays rent tester is a fairly sensitive scientific device, and therefore it measures Go, No Go, or Meter both changes in electrical conductivity, CRT magnetic permeability, and geometrical Other Imaging Devices effects. Alloy type materials which are 143 most commonly used, have more than one what kinds of data we measure. And that, chemical or elemental constituent, and again, would tie in with the array; they these often segregate in different con- could be- used with fast scan systems. figurations. Therefore, you will not get Computer-controlled ultrasonics systems a constant eddy current reading no matter are already making great progress in this what parameter you are measuring on your regard. standard piece, if the eddy current detector is sufficiently sensitive. I am not The display could range from a red really clear in my mind how this is going and green light for some types of to be overcome in general, but I think operators, to a very complicated type of that specimens are an area for consider- display if it was for a scientist in a ation. research lab. I would hope it would not bother him too much how complicated the Multi- layer types of materials are pattern looked. also important. There was some work reported, and composite materials were I can see some interesting work being briefly mentioned. Here, we are dealing done on developing imaging type displays with extreme inhomogeneity of some types to go along with the arrays, or with the of materials to be inspected. If a rapid scanning, where you could display scientist was concerned with materials, I pictures of defects. I would think think this first section would be of great whether we like it or not, most people, interest to him in not only making conduc- non-specialists especially, really like to tivity measurements, permeability measure- see pictures of a defect or flaw and that ments, etc., but perhaps residual stress is the thing that is coming in this field type measurements or other types. as well as other fields.

The geometrical problem was pointed Multi-frequency techniques and the out by several people. People would like pulse techniques are both options that to be able to inspect various configu- people are developing and using, and it rations and parts of present configura- appears that more work can be done there. tions that cannot presently be done. I was especially interested in Professor Waidelich's little equation. When you Most of the other areas I would think get a good equation from experimenting, would appeal more to electrical engineers, one which theoreticians have not derived, electronic engineers, or physicists, or it stimulates theoretical work. From the people with that bent. new theoretical work, in turn, you can expand the experimental investigations. With respect to coils, Professor Mc So, this is a nice area to look into Master brought up a number of the para- almost immediately. meters in his talk the first day. The different geometries of coils, using a Eddy current testing in hostile envi- coil of triangular or ellipsoidal shape, ronments presents special difficulties. might be of interest. The orientation of Several people brought up the problem of the coil with respect to the work piece testing hot items. This also could apply and the applied magnetic field could be to cold items, if you happen to have some studied. And perhaps, arrays of coils or work in Alaska. Either extreme of temper- arrays of other types of detectors may be ature will affect test results. With useful for rapid scanning. Several people respect to hostile radiation type environ- have pointed to the need for that. You ments, there appears to be considerable could also have other detectors, not just progress from the talk given yesterday, the little coils that we normally use. especially the device that Clyde Denton Some other detectors may lead more natu- talked to us about. rally into an array configuration than the coils. I know that a lot of work is going Other methods are very interesting to on in arrays for ultrasonic testing, and I me, such as the microwave technique and do not see why we would be prohibited from the phenomenon that Professor Mc Master doing this in eddy current work. pointed out, triboelectricity, materials emitting electromagnetic waves spontane- We have heard mentioned the use of ously, similar to acoustic emission. computers for signal processing. There is the possibility of using computers to not I would like to have some discussion only analyze the data, but also to control now, and in the discussion, consider your

144 .

own practical needs. Where could some whatever to better understand just what is scientific work be done that would be happening as far as the electromagnetic useful to you? fields are concerned?

Mr. Brown : At the risk of sounding Dr. Mc Master : Could I respond with too simple, I would suggest that a signif- a suggestion for your organization? I icant contribution could be made if you feel one of the problems is that if you figured out how to make a reproducible take all of these problems in one hole in the side of a tube. There are "mishmash," you are going to have to have many types of standards, paper standards a Latin squares statistical experiment to that call for a hole of one form or figure out what did what. If we divide it another, 20 percent or 50 percent through in sequence in terms of fundamental phe- the wall or all the way through. But I am nomena, we can have many different groups not convinced, in fact, I am very insecure working on various aspects. So, let me about the reproducibility of some of these suggest this division of the problem, and holes from one piece to another. The it is only a suggestion. boiler code calls for a flat bottom hole to go part-way through, and some of the The first area of study would be the other standards call for holes all the way magnetizing coils and their fields and through. But many people use holes for anything that has to do with magnetizing various purposes, and I do not think those coils and their fields, including how holes are reproducible enough. I suspect these fields are modified when they are in that NBS could do something that would the presence of a test object. This is a help people manufacture reproducible holes field which could be handled by a group of so the data was more comparable back and people. It involves a portion of these forth across the country. problems, and involves fundamental theory and related analyses.

Dr. Green : I assume you mean by a hole, not only the hole itself, but what A second field of study would be the would happen to the material surrounding eddy currents. It is perfectly possible the hole when the hole is made? in an eddy current instrument to wipe out the entire coil signal with electronics,

Mr. Brown : Right. This would in- and then measure nothing but eddy cur- clude deburrings on the inside of a tube, rents. It seems to me, we should study which could be a real problem. the eddy currents. It is easy to do in the constant current .instruments, if you

Dr. Green : Perhaps even the residual abolish the coil field and forget it from stress? now on and go to the eddy currents and their distribution in the metal, for all Mr. Weismante l: Wouldn't you want to the different shapes and/or frequencies start by putting holes in flat plates that affect distribution and/or defects, first? Eventually, you have to get to a etc. That could be done possibly by radius effect, but you have to start with techniques like we saw demonstrated something basic and then build up the earl ier. mechanism to get into the more involved three-dimensional structures at different A third level is the magnetic field

radi i created by the eddy currents. This is a different problem from the eddy current current magnetic Mr. Wehrmeister : I was just going to distribution. The eddy suggest SDN-243 which is for copper inspec- field is superimposed in space and with tion, primarily copper tubing inspections. nonferromagnetic test objects we have seen It offers some guidance to drilling holes that superposition applies. Thus, we can and filing notches in material, and also analyze it independently of the magne- some tolerances which people in that ti zi ng coi 1 f iel d. particular industry have been living with for a good number of years. And, finally, the detector response to the induced field. Since the detector in a uniform way to its magne- Mr. Berger : I do not want to turn responds of a test off this discussion, but I would like to tizing coil field in the absence ask a question. Is there a need for object, the induced field response can be greater theoretical work to go along with considered independently. Obviously sig- putting holes into flat plates or tubes or nal analysis and interpretation would be another area of endeavor. 145 Now, the only reason I suggest this been some talk about electrically cali- classification is the obvious fact that brating detectors. I think that is okay you can divide the job into four parts or and should be done, but let's not forget five and each could be thoroughly explored about the actual standard. independently of the others. And maybe when you put the pieces together you have Dr. Green : Any other suggestions? a coherent program.

Mr. Brown : This is a negative

Mr. Libby : One comment. I would suggestion. I agree with all these lofty join the first two areas since current ideals and have for years, like everybody distributions are very definitely a else. But I caution NBS not to take on function of the coil geometry. the world. Do not try to do it all. We have all found that it is not practical.

Dr. Mc Master : Yes, there is no Do a good job in the standards field question about that. But if you subdivide first; do not try to expand into and conquer the pieces, you may be able to everything. analyze the relationship between these pieces. If you put them all together, the Mr. Mester : In response to some of problem gets too complex, at least for my Bob Mc Master's suggestions, the idea mind. occurred to me, and other people have mentioned it in some of the previous

Mr. Weismantel : I think the way Dr. discussions, of using taped data. In my Mc Master has organized the problem is a company, we are working to correlate very logical way to go. But, I would like responses from our equipment to to see coming out of this whole thing some information gathered about actual defects computer modeling which would give us a in the material which are determine by much more valuable tool, more powerful other methods. tool for the application of the process. The problem I face is characteristic of What we do is set our instrument many people. A lot of applications of condition to take the run data and running eddy currents that are successfully used the instrument at a particular setting, are volume testing of parts with a similar for instance. This is a one-shot test geometry. since the material is hot, you never get a chance to run it again. It can never be For example, in a turbine engine you run through under the same conditions, but might have 30 different shapes that are the data is now on tape. You go back and- unique to whatever the component is that actually put the tape data back in again, has to be inspected. These vary in alloy, changing time constants and levels, reject and they vary in electrical and thermal levels, to force the conditions to fit conductivity characteristics. We must what the correlations should be. have a way of building up the theory so that we do not have to do everything by What I think may be appropriate when empirical means. We have to know how to considering how my equipment compares with handle a radius, and determine what the the other fellow's equipment, is producing sensitivity is for finding a flaw in that on magnetic tape, for instance, a signal radius relative to the sensitivity on a which is indicative of, or representative flat plate. This requires a further of, some standard defect. Of course, the extension of Maxwell's equation and a information on the tape is determined by development of a scientific approach to environmental conditions, the size of the the subject. coil, the field, and many other things.

Mr. Titland : In the area of dis- But I am saying you have established plays, you mentioned two things, either a at least some reference. This tape can be red and green light, or a picture. But I given to people who would feed it in at think there is a third item that should be some point in their equipment for testing. considered in displays, that is the stand- Maybe their signal has a lot of noise and ard with holes and defects which checks they are trying to determine just how the sensitivity of the instrument. For capable the equipment is of handling that example, the inspector who comes along background noise. This may be helpful when you are testing pressure vessels when everybody is working with the same likes to see that the standard will truly type of signal. They have the signal check the functioning of the instrument. input. What their output is, is The reason I mention this is there has determined by a black box. I do not think 146 you can regulate and say everybody should have the same black box. But if we started with the same hole and the same applied field and we have the same eddy current pattern, we are looking at the same output. The rest of your system I think you have to consider separately, and this may be one way of doing it.

National Bureau of Standards Special Publication 589. Proceedings of the Workshop on Eddy Current Nondestructive Testing held at NBS, Gai thersburg , MD, November 3-4, 1977. Issued January 1981

NEW DIRECTIONS FOR STANDARDS CONCLUDED FROM WORKSHOP SESSIONS

Norman B. Belecki Electrical Measurements and Standards Division National Bureau of Standards Washington, DC 20234

In any industry, the standards that electrical conductivity SRM's important to are used for quality control and for other the eddy current work now being done in purposes may have many forms. If the industry? If this type of standard and standards that are used do not attack the calibration service is important, what are primary measurement problems of a specific the limits, in percent IACS, that should industry, the result is wasted time, money, be offered? At present, we are thinking and unresolved measurement problems. If of working in the area of 3 percent to 100 the standards that are used do not percent IACS. But should we go lower than directly relate to the measurement pro- 3 percent and if so, how low should we go? blems that are faced, there is no reason A final question related to the conduc- for their existence. From the preceding tivity standards is the range of frequen- talks and discussions of the workshop, I cies that are important in the testing of have made a list (fig. 1) of areas in samples. Much of the commercial eddy cur- which standards or calibration services rent equipment functions at 60 kHz. might have a direct impact on industry. Should we limit ourselves to testing at this frequency? In the testing of other

electrical parameters, I have found that (1) Conductivity Standards when one of the parameters is limited in Calibration Service range, there is a tendency to restrict Standard Reference Materials advances which could be made in areas that (2) Terminology are peripheral to this parameter. (3) Performance Standards e.g., Probe Characteristics These are a few of the questions re- (4) Reference Data lated to eddy current standards that I (a) Metals think need answers. Now, we would like to (b) Comparison of Methods open the floor to discussion and questions (c) Geometrical Effects you might have about eddy current stand- (5) Tutorial Material, to bridge gap ards, what standards are needed now, and between theory and practice. what standards will be needed in the (6) Measurement Methods, future. (7) Standards and Recommended Practice for Cladding Measurement. Di scussion (8) Ferromagnetic Standards

(9) Sensitivity Tests for Equipment Mr. Weistmantel : Why limit the con- (10) Round Robin Tests with Standard ductivity standards to 3 percent IACS? Defects. Why not go lower?

Figure 1. Area for standards development. Mr. Belecki : That is what I am trying to get at; how low should we go? That is the question I am raising. Before starting the discussion on

: a standards for eddy current testing, I would Mr. Weismantel Go down to as low like to raise several questions about our conductivity as we have commercially own present program, i.e., conductivity available materials. I do not know how standards. Is an electrical conductivity low that would be. service, either a calibration of customers' conductivity samples or the issuance of . —

Mr. Free : I had a conversation with coatings on metals by impedance mismatch someone from the steel industry several reflection signals and their phase or am- weeks ago, and I do not know whether this plitude. Where it is possible to go up information is accurate or not. He stated through that range of frequencies it would that for many types of your lower con- be very helpful ductivity materials, one percent or below, eddy currents really do not give you much Mr. Blew : What kind of correlation useful information. You do not search would your samples have with what NBS is with them; you cannot tell the difference doing with wafer resistivity samples? between heat treats. It does not give you the needed information. Mr. Belecki : I think we would have to run experiments to be sure that our

Dr. McMaster : I do not agree. I measurements are consistent throughout would hate to take a piece of steel and thei r range. try to analyze its contents, including

residual things like gases with eddy cur- Mr. Free : This brings up a point. I rents. But, when you get down to titanium do not know whether it should be discussed or something like that, you have a chance. here or not. I wanted to raise it this

I would suggest we have missed a very morning, but we did not have time. The valuable area down there, far more than we idea of the percent IACS scale for con- know. Although these are exotic, costly ductivity-is it meaningful when the only five dollars-a-pound materials used in time we use it is when we do a critical applications, where do you have a cal ibration? greater need? I would encourage going all the way. I would like to see you go down Anytime we are doing a real honest-

to 100th of 1 percent with conductivity to-goodness measurement, we are talking standards. about ohms, units of length, etc; we are never talking about percent IACS except as

Mr. Brown : You can sort graphite a mythological beast. that way and it is 1000 micro ohm

centimeters. Dr. McMaster : It is used in the copper industry for electrical copper

Dr. McMaster : And finally, when you only. come to graphite fibers in a matrix, you

are going to establish orientation, pack- Mr. Jones : The aircraft industry

ing density, etc. , by eddy currents. By uses it for sorting incoming material. taking directional magnetic fields, you

can get a pretty good idea of the angle of Dr. McMaster : Only because the orientation. meters have an IACS scale.

I think there will be a great deal of Mr. Jones : It is a workable system. engineering moving out of straight metals We did not like it either when we started into composites and refractory materials, making the standards, but it does work. which by the way become conductive when

they get hot enough. Glass is a beautiful Mr. Endler : There are many com- conductor at higher temperatures. mercial and military specifications about IACS.

Mr. Belecki : I was wondering if

there were any other geometries that would Mr. Belecki : I think that any change be desirable in conductivity standards. of units would probably have to be done If so, what would they be? gradually as these specifications came up for review, to use both, and then even-

Mr. Wehrmeister : I am just guessing, tually remove the IACS. That is the but I would think foil, a very thin mate- normal way of doing things. rial, would be of interest as a conduc-

tivity standard. And that, of course, is Mr. Blew : On the sample geometry going to have a very large effect on the are you going to define criteria there, or range of frequencies that you are talking is it going to be up to this group to set about. the standard geometry, thickness, and so on?

Dr. McMaster : On the other hand, the

upper range of frequencies should be in Mr. Belecki : One of the things that the microwave range and beyond. In the we hope to do when we get around to con- future, we will be interrogating refractory 150 sidering the SRM's and what form they the hard way to go about it. Wouldn't it should have, would be to get together with be far simpler to start with a relatively all the instrument manufacturers and dis- pure material like aluminum, and progres- cuss the standardization possibilities in sively alloy in copper to cover that range that area. I think it is almost crucial from 100 percent down to 60 percent down in order to get the optimum performance to 38 percent or so, and do this in a con- out of these standards, and I think that trolled experiment so you can build them the spill -out of that kind of a meeting to demand? And in fact, if you had a would have an effect, too, on what we method of hot measurement of the conduc- would say for criteria on samples sub- tivity, you might even tell when you are mitted to NBS for calibration. there, very much like you do in steel.

There is an analogy to our regular Mr. Belecki : Yes, we have talked calibration program for standards. It is about a number of such things. clearly written in the SP-250, which is the publication that describes our ser- Dr. McMaster : You could add phos- vices, that we will not accept instruments phorous to copper to get the rest of the for calibration or standards for cali- scale. And then finally, a third sugges- bration if they are not serviceable. The tion would be for those things that are same kind of general statement can serve somewhat different conductivity per unit for conductivity samples. square imitating thin foils. The tin oxide coating, a nonelectrostatic coating, they I think that as time goes by all that are Nisa coatings made by Pittsburgh Plate kind of thing will be ironed out as well. Glass, can be laid down on glass. You can This problem is analogous to that of put it on glass, quartz, or ceramic so you standard resistors. Standard resistors can have a stable base, you spray on tin generally, especially those designed for fluorite, methanol and so on solutions at oil immersion, have mercury contacts and maybe 1100°F, 1200°F or higher tempera- certain dimensional characteristics and tures, and it comes out with a glossy, our equipment is designed to accommodate chemically resistanct, extremely adherent, all of those kinds of variables. I assume and, as far as I know, longtime stable that that kind of accommodation will be coating. This is what is used on the wind- reached. shields of jet aircraft.

Mr. Blew : Do you have an idea of the Mr. Belecki : You spoke of these fee, as compared to your standard things Wednesday night. resistors?

Dr. McMaster : Yes. I would like to

Mr. Belecki : No. The only thing I suggest that artificial standards are far would say is that there is a good like- more easily prepared than finding a lot of lihood that all the measurement equipment aluminum that is consistent in its con- is going to be automated. ductivity, because they do not make aluminum specifically for its electrical con- Congress requires us to recover the ductivity except for certain conductors. cost of carrying out the operation; not If you make your own you could adjust it necessarily of establishing the capabil- and possibly have a feedback control ity, but carrying out the operation. device so that when you approach the desired conductivity, you slow it down or make many of them more Mr. Blew : What would be an ap- stop. You could proximate price? What is a standard precisely, and you could cover everything resistor calibration costing now? in uniform steps instead of having these funny steps that we have now.

Mr. Belecki : It could be anything area I from $100 to $200, someplace in that Mr. Free : This was another range, depending on the work involved. wanted to bring up this morning and did conduc- But I do not think this would cost any- not. Is it better to make ideal thing like that. tivity standards, more of what you are talking about or, is it better to come out which are really the most Dr. McMaster : I get a vague impres- with standards sion that you are expecting to use as common al loys? standards commercially avaliable metals and alloys, materials that suit your Dr. McMaster : The point is these are needs. At first glance, that seems like used to calibrate instruments. If you are

151 calibrating an instrument, you would like Mr. Belecki : I think that is true. to know in what range or percentage of full scale, it is accurate. The standards Mr. Wehrmeister : Especially in cus- are never actually used in a comparison tom-made standards, for example. coil to detect a matching alloy. It is an instrument calibration standard. Mr. Belecki : I think that is surely true. We have done that in a couple of

Mr. Free : Where I would see the hand- cases. In my own field, for example, an iness of one of the aluminum alloys, is electronics company was having problems that it can be carried right to the indus- verifying measurements on the production trial setup, and you do not have to worry line for ladders used in A/D converters, about corrections for temperature. You and we built a special transportable are zeroing in on that alloy, so to speak, standard so that we could directly test and you are comparing it with a lot of the that process with them. same alloy you know is in your warehouse. That brings up another thing that

Dr. McMaster : Those are secondary your talk implied; we have methods of ser- standards. vicing customers in our regular calibration areas via techniques similar to the

Mr. Belecki : The other advantage in round robin method that you talked about. the alloys is if a person has a lot of It might be a very useful approach espe- measurements on exactly the same metal cially in relation to those areas where sample, that will allow him to apply dif- people have a lot of data, but do not know ferential measurement techniques, rather how to correlate it with data that some- than trying to make an absolute conduc- body el se has. tivity measurement, and raise the sensitivity several orders of magnitude because Mr. Reinhardt : There is another role of that. for NBS. I am always looking for uncon- cerned third parties; a team that might

Mr. Richardson : Are you going to have some interest, personal or profes- tell us how to make the measurements on a sional, in looking at a test a certain conductivity standard? way.

Mr. Belecki : I think that would Mr. Belecki : Considering measurement probably be useful. The area of reference methods, one of the things I thought of data is another area that we do not have that would be useful would be a method of any immediate plans in, but it is some- testing the sensitivity of measuring thing that might be useful for various equipment. That seems to be somewhat in people. question in some of these various areas. The work at Battelle Northwest that was

Mr. Brown : Those of us who have described yesterday by Mr. Davis would be looked in 10 to 100 books for the values a possible way to pursue that type of of resistivity, would like to have one test. list with all the materials on that list.

Mr. Brown : Along the lines of a

Mr. Schwarz : How about some measurement method, there is a significant reference data on permeabilities of var- degree of uncertainty in describing the ious steels? many variables in an eddy current test; how it is set up, how stable it is, what

Mr. Weismantel : Reference data does the results are and what do they mean not have to be limited to eddy currents etc.; probabilities and statistics are you know. Velocity of sound is an involved. I recognize there are practical important area. and statistical methods for handling these things; but, from the users' standpoint

Mr. Wehrmeister : It was mentioned they usually are too complicated for prac- yesterday that vendors have to run ROI's, tical application. returns on investments, to see if it is

feasible to do something for general Mr. Belecki : We have the same prob- industry. I do not, somehow, suspect that lems in the standards area. the Bureau runs return-on-investment

studies and it might be possible to supply Mr. Brown : The Bureau of Standards things to industry that could not be could make a great contribution by normally supplied by an industrial vendor. evolving some simple--let me say that 152 three times—simple, simple, simple-- Mr. Weismantel : That is what I am statistical techniques, for evaluating the saying, you can artificially synthesize uncertainties involved. them.

Mr. Belecki : I suspect that there Mr. Ammirato : That depends on what are large numbers of measurements made on object you are after. On the one hand, clad surfaces, and I wondered what kind of you have to calibrate the instrument to standards might be useful in that area? make sure that is working all right. On the other hand, you are trying to get some

Mr. Jones : Mil 1537 covers a little idea of what the actual flaw size is in bit of that. I have been rewriting it. your part. If you have them varied, you We are also putting definitions in, which will not know what size they are. comes under terminology. Maybe NBS, would be interested in seeing what we define as Mr. Weismantel : You can develop a conductivity in primary standards, sec- technique where you get a good prediction ondary standards, etc., so we could get of what the size is. We have gone through some agreement at the beginning, because it, but not being farsighted enough, we it is still being written. cut these things up, something which you would like to avoid so that the test

Mr. Weismantel : Do you really want pieces are preserved. Then you can always to get into the areas you have listed go back and measure relative detection there? In other words, standards for clad- efficiency. The defects have to be in ding, standards for defects? You have large enough quantity to give you some tubings, you have other things. It seems statistical appraisal of what you have to me that there are so many ungodly com- got. binations of things, alternatives.

Mr. Titland : Two comments on this:

Mr. Belecki : I think that the way I First, I think the Welding Institute in would have to answer that would be to say England has developed a technique where a that we would get into an area if it were flaw is placed inside the material by useful to get into, and were within the machining in pieces, and then putting the range of our resources to do so. I under- pieces together in a vacuum. I think this stand what you are saying. We could not technique may be available. Secondly, when possibly supply people with artifacts or we make a standard it could be used not calibration service for standard flaws for only for eddy current, but also for every possible conceivable application. radiographic and ultrasonics.

Mr. Weismantel : A problem that we Mr. Belecki : That is right. have is the fact that there are not any good standards available that can be used Mr. Weismantel : So that you could to measure relative detection efficiency compare the efficiency of the different to determine when there are improvements processes. or lack of improvements in a process due to advancements in the process technology. Mr. Reinhardt : This, to me, is a There is no base to say that one instru- very critical area, this simulation of ment or method is better than another. The real defects and keeping them in an ar- T-crack blocks that I discussed in my talk chival place. It opens up another area is our attempt to do this internally. which is a computer learning method. If They are our own controls to get a measure you have them in some central place where of where we are. But it would be extremely they are never taken apart, this allows useful if somewhere there was a grand people with the difficulty to go there. master series of buried defects. No EDM We have a deficiency of seeing real or notches or anything like that, but sub- simulated flaws in our industry, because surface flaws, that could be used to gauge the metallurgists get there first and take process efficiency. There are techniques them apart. in which you can produce subsurface flaws of different sizes in character and orien- Mr. Weismantel : That is the only way tation. And natural-looking flaws, not they know what size they are. EDM notches and things of this sort.

Mr. Belecki : We only have a couple

Mr. Belecki : Well, don't you think of more minutes. Synthetic standards like there is some possibility of being able to those Tom Davis talked about yesterday synthesize those kinds of things artifi- could be used in a large variety of ways, cial ly? both as ways of testing instrumentation coming down the power line on your mea- and ways of augmenting some of the theo- suring apparatus. Do you have a lot of retical things that we talked about problems in that regard? We have had a earl ier. fair amount of input from companies, especially those making medical instru- One of the things that we were talk- ments, that want some kind of standard- ing about at lunch time, which we thought ization in that area for liability pur- might be useful, is the characterization poses. Right now if they make an instru- of a measurement system component. Perhaps ment that is totally free from that kind it might be worthwhile to have some sort of effect, they cannot be competitive. of performance specifications or something of that order for probes and pickups. Mr. Mester : I was just going to say that that is a very big problem. When you

Mr. Weismantel : The ferrites that go buy equipment in the lab it works fine. into them, start with them because they You put it in the field and you have prob- have a large variability. lems; transducers probably aren't chilled; ground loops exist, you have problems with

Dr. McMaster : And the ceramic bases noise coming in on the line, even if you for the coil forms tend to distort. There have taken pains to separate your lines. should be a ceramic with zero temperature It is a very big problem, and I do not coefficient of expansion, wound with Invar want to go into it a lot. But it is an or something, which has a small tempera- area which has been neglected. ture coefficient.

Mr. Belecki : Any other comments?

Mr. Blew : Not on methods, but on the instrumentation itself as far as the performance is concerned. It seems from what I have gotten, that it is varied as far as the sensitivity and the noise characteristics are concerned. You have been more or less talking about electronics in a black box. I looked at NBS Technical Note 865 from the meeting in 1974 and saw sources and standards, and also on the devices such as A/D D/A, and I imagine stability of analog dividers and so on. I was wondering, has any of this been resolved since the 1974 meeting?

Mr. Belecki : Many of the things that are mentioned in there are being worked on in the program of the Electrical Instru- ments Section that Barry Taylor discussed yesterday. They have done a fair amount in the way of terminology and data conversion work or at least they are right in the middle of doing that. There are, I think, three committees, IEEE committees and ANSI committees, that members of the section are working on. And I think that especially in the data conversion area, and in the area of microprocessor and instrumentation interfaces, there is work being done.

The noise problem has not been addressed yet, and that brings up another problem that we have heard about from the general instrumentation and test equipment industry. I wonder what comments I might get from you about the effects of noise ,

ACKNOWLEDGMENTS

NBS would like to give special thanks We would like to thank ASTM for per- to Mr. J. R. Barton of Southwest Research mission to print the tables which appear who presented a paper on eddy current at the end of Dr. McMaster's paper. These testing in the transportation industry. tables are from ASTM Special Technical This paper was not available at the time Publication No. 112. of publication. Also appreciated were presentations by Mr. Rudy G. Hentschel Our sincere thanks to all who partic- Hentschel Engineering, entitled "Lets Not ipated in the workshop. Forget Standards for Ferrous Metals," and a review of the work done at Battel le Northwest on Simulating Tubing Flaws by Mr. Tom Davis of Battel le. The former was not available at the time of publication and a complete discussion of the latter topic can be found in Materials Evalu- ation, "Calibration of Eddy Current Sys- tems with Simulated Signals," by J. C. Crowe, Sept. 1977, Vol. 35, Number 9, p. 59.

155

LIST OF PARTICIPANTS

Frank Ammirato Clyde Denton General Electric Co. ZETEC, Inc. M&T Lab, Bldg. 28, Rm. 450 P. 0. Box 140

1 River Road Issaquah, WA 98027 Schenectady, NY 12345 Thomas F. Drumwright J. R. Barton Alcoa Technical Center Southwest Research Institute Alcoa Center, PA 15069 P. 0. Drawer 28510 San Antonio, TX 78284 John F. Endler Qual ity Assurance Winthrop A. Baylies The Anaconda Company, Brass Di ADE Corporation P. 0. Box 830 127 Coolidge Hill Road Waterbury, CT 06720 Watertown, MA 02172 Robert H. Foxall Robert Betz Wean United, Inc. Pratt & Whitney Aircraft P. 0. Box 751 Department 983 1985 N. River Road K Building Warren, OH 44482 East Hartford, CT 06108 John Gleim Austin R. Blew Naval Ship Engineering Center Lehighton Electronics, Inc. Code 610D R. D. 1, Box 374 Hyattsville, MD 20782 Lehighton, PA 18235 Robert Green Creighton Booth Johns Hopkins University Bethlehem Steel Corp. Mech. & Matls. Science Dept. Bethlehem, PA 18016 Baltimore, MD 21218

Russell L. Brown Rudy G. Hentschel Westinghouse Hanford Co. Hentschel Instruments, Inc. P. 0. Box 1970 2505 South Industrial Highway Mail Stop W-A132 Ann Arbor, MI 48104 3707 C Building Richland, WA 99352 Howard Houserman Westinghouse NDS Don Bugden RSD Center Bldg. 701 Magnetic Analysis Corp. P. 0. Box 2728 535 S. Fourth Avenue Pittsburgh, PA 15230 Mount Vernon, NY 10550 Art Jones Carl E. Burley Boeing Aerospace Company Reynolds Metals Company Org. 2-4807, MS 2P-10 4th & Canal Streets P. 0. Box 3999 Richmond, VA 23261 Seattle, WA 98124

Caius V. Dodd Tyler W. Judd Oak Ridge National Laboratory Republic Steel Corporation Metals & Ceramics Div. Research Center P. 0. Box X 6801 Breckville Road Oak Ridge, TN 37830 P. 0. Box 7806 Independence, OH 44131 Tom Davis Battel le Northwest John Kol lar PS Bldg. 3000 Area Balteau Electric Corp. Richland, WA 99352 63 Jefferson Street Stamford, CT 06902

157 Lawrence Lagin Allen Sather Grumman Aerospace Corp. Materials Science Division Quality Control Laboratory Argonne National Laboratory Plant 10 Argonne, IL 60439 Bethpage, NY 11714 Ronald Schwarz A. L. Lavery Manufacturing Development Bldg. B DOT Transportation Systems Center GMTC Kendall Square Warren, MI 48090 Cambridge, MA 02142 L. Erik Titland Hugo Libby Baltimore Gas & Electric Co. 2302 Frankfort Street Westport Metallurgical Laboratory Richland, WA 99352 P. 0. Box 1475 Baltimore, MD 21203 Tracy McFarlan Magnaflux Corp. Donald Waidel ich 7300 West Lawrence Avenue University of Missouri Chicago, IL 60656 Dept. of Electrical Engineering Columbia, M0 65201 Patrick C. McEleney U. S. Army Materials & Mechanics A. E. Wehrmeister Research Center Babcock & Wi Icox Co. Arsenal Street Lynchburg Research Center Watertown, MA 02172 P. 0. Box 1260 Lynchburg, VA 24505 Robert C. McMaster 6453 Dublin Road E. E. Weismantel Delaware, OH 43015 Quality Measurements Mail Drop E45 Michael Mester General Electric Co. U. S. Steel Corp. Aircraft Engine Group Research Laboratories Cincinnati, OH 45215 125 Jamison Lane Monroeville, PA 15149 National Bureau of Standards Attendees Richard B. Moyer AISI Committee on Nondestructive Testing Dr. L. H. Bennett and Inspection Materials Mr. H. Berger Carpenter Technology Corp. Dr. G. Birnbaum Research and Development Center Mr. N. Belecki Reading, PA 19603 Mr. G. Free Dr. A. Kahn Noel Poduje ADE Corporation 127 Coolidge Hill Road Watertown, MA 02172

Eugene R. Reinhardt Electric Power Research Institute 3412 Hi 11 view Avenue P. 0. Box 10412 Palo Alto, CA 94303

Guy D. Richardson Detek, Inc. 6807 Cool ridge Drive Camp Springs, MD 20031

Walter Rolfe Magnaflux Corp. 7300 West Lawrence Avenue Chicago, IL 60656

158 NBS-114A IREV- 2-8C) U.S. DEPT. OF COMM. 1. PUBLICATION OR 2. Performing Organ. Report No. 3. Publication Date REPORT NO. BIBLIOGRAPHIC DATA January 1981 SHEET (See instructions) NBS SP 589

4. TITLE AND SUBTITLE Proceedings of the Workshop on Eddy Current Nondestructive Testing held November 3-4, 1977 at the National Bureau of Standards, Gaithersburg, MD.

5. AUTHOR(S)

George M. Free, Editor

6. PERFORMING ORGANIZATION (If joint or other than NBS. see instructions) 7. Contract/Grant No.

NATIONAL BUREAU OF STANDARDS DEPARTMENT OF COMMERCE 8. Type of Report & Period Covered WASHINGTON, D.C. 20234 Final

9. SPONSORING ORGANIZATION NAME AND COMPLETE ADDRESS (Street. City. State. ZIP)

Same as No. 6

10. SUPPLEMENTARY NOTES

Library of Congress Catalog Card Number: 80-600172

JJ Document describes a computer program; SF-185, FIPS Software Summary, is attached.

11. ABSTRACT ( A 200-word or less factual summary of most significant information. If document includes a significant bibliography or literature survey, mention it here)

The proceedings of the Eddy Current Nondestructive Testing Workshop held at NBS in November, 1977 contains papers related to all areas of eddy current testing. A historical overview of the discipline from its inception until the present is given. Other papers discuss the use of eddy current testing in the primary metals industry (both ferrous and nonferrous metals), the use of eddy currents for the sorting of metals and for defect detection, the state-of-the-art in eddy current instrumentation, and the use of signal processing in the analysis of eddy current signals. The development and use of eddy current standards is discussed as well as several of the newer areas of eddy current development, i.e., mul ti frequency and pulsed eddy current

techniques. .

12. KEY WORDS (Six to twelve entries; alphabetical order; capitalize only proper names; and separate key words by semicolon s)

Conductivity; defect detection; eddy current test; multifrequency; nondestructive

[ testing,

|13. AVAILABILITY 14. NO. OF PRINTED PAGES fj^ Unlimited f~J For Official Distribution. Do Not Release to NTIS 153 Order From Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402. 15. Price

S3. 50 Q7J Order From National Technical Information Service (NTIS), Springfield, VA. 22161

USCOMM-DC 6043-P80

NBS TECHNICAL PUBLICATIONS

PERIODICALS NOTE: The principal publication outlet for the foregoing data is the Journal of Physical and Chemical Reference Data (JPCRD) JOURNAL OF RESEARCH—The Journal of Research of the published quarterly for NBS by the American Chemical Society National Bureau of Standards reports NBS research and develop- (ACS) and the American Institute of Physics (AIP). Subscriptions, in those disciplines of the physical and engineering sciences in ment reprints, and supplements available from ACS, 1 155 Sixteenth St., which the Bureau is active. These include physics, chemistry, NW, Washington, DC 20056. engineering, mathematics, and computer sciences. Papers cover a broad range of subjects, with major emphasis on measurement Building Science Series— Disseminates technical information methodology and the basic technology underlying standardization. developed at the Bureau on building materials, components, Also included from time to time are survey articles on topics systems, and whole structures. The series presents research results, closely related to the Bureau's technical and scientific programs. test methods, and performance criteria related to the structural and As a special service to subscribers each issue contains complete environmental functions and the durability and safety characteristics elements systems. citations to all recent Bureau publications in both NBS and non- of building and NBS media. Issued six times a year. Annual subscription: domestic Technical Notes— Studies or reports which are complete in them- $13; foreign $16.2?. Single copy, $3 domestic; $3.75 foreign. selves but restrictive in their treatment of a subject. Analogous to NOTE; The Journal was formerly published in two sections: Sec- monographs but not so comprehensive in scope or definitive in tion "Physics and Chemistry" and Section B "Mathematical A treatment of the subject area. Often serve as a vehicle for final Sciences." reports of work performed at NBS under the sponsorship of other DIMENSIONS/NBS—This monthly magazine is published to in- government agencies. form scientists, engineers, business and industry leaders, teachers, students, and consumers of the latest advances in science and Voluntary Product Standards— Developed under procedures technology, with primary emphasis on work at NBS. The magazine published by the Department of Commerce in Part 10, Title 15, of the highlights and reviews such issues as energy research, fire protec- Code of Federal Regulations. The standards establish tion, building technology, metric conversion, pollution abatement, nationally recognized requirements for products, and provide all health and safety, and consumer product performance. In addi- concerned interests with a basis for common understanding of the characteristics of the products. NBS administers this program as a tion, it reports the results of Bureau programs in measurement standards and techniques, properties of matter and materials, supplement to the activities of the private sector standardizing engineering standards and services, instrumentation, and organizations. automatic data processing. Annual subscription: domestic $11; Consumer Information Series— Practical information, based on foreign $13.75. N BS research and experience, covering areas of interest to the con- NONPERIODICALS sumer. Easily understandable language and illustrations provide useful background knowledge for shopping in today's tech- Monographs— Major contributions to the technical literature on nological marketplace. various subjects related to the Bureau's scientific and technical ac- Order the above NBS publications from: Superintendent oj Docu- tivities. ments. Government Printing Office. Washington. DC 20402. Handbooks— Recommended codes of engineering and industrial Order the following NBS publications— FIPS and NBSIR's—from practice (including safety codes) developed in cooperation with in- the National Technical Information Services, Springfield. VA 22161 terested industries, professional organizations, and regulatory bodies. Federal Information Processing Standards Publications (FIPS Special Publications— Include proceedings of conferences spon- PUB)— Publications in this series collectively constitute the sored by NBS, NBS annual reports, and other special publications Federal Information Processing Standards Register. The Register appropriate to this grouping such as wall charts, pocket cards, and serves as the official source of information in the Federal Govern- bibliographies. ment regarding standards issued by NBS pursuant to the Federal Property Administrative Services Act of 1949 as amended, Applied Mathematics Series Mathematical tables, manuals, and and | — Public 89-306 Stat. and as implemented Ex- studies of special interest to physicists, engineers, chemists, Law (79 1127), by biologists, mathematicians, computer programmers, and others ecutive Order 11717(38 FR 12315, dated May II, 1973) and Part 6 Title of Federal Regulations). engaged in scientific and technical work. of 15 CFR (Code

National Standard Reference Data Series— Provides quantitative NBS Interagency Reports (NBSIR)—A special series of interim or data on the physical and chemical properties of materials, com- final reports on work performed by NBS for outside sponsors piled from the world's literature and critically evaluated. (both government and non-government). In general, initial dis-

Developed under a worldwide program coordinated by NBS under tribution is handled by the sponsor; public distribution is by the the authority of the National Standard Data Act (Public Law National Technical Information Services, Springfield, VA 22161, 90-396). in paper copy or microfiche form. U.S. DEPARTMENT OF COMMERCE National Bureau of Standards Washington. D C. 20234 POSTAGE AND FEES PAID U.S. DEPARTMENT OF COMMERCE OFFICIAL BUSINESS COM-21S

Penalty for Private Use. $300

SPECIAL FOURTH-CLASS RATE BOOK