You are on page 1of 447

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011

Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

STP 1077

Acoustic Emission: Current
Practice and Future Directions

Wolfgang Sachse, James Roget, and Kusuo Yamaguchi, editors

ASTM
1916 Race Street
Philadelphia, PA 19103

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

Library of Congress Cataloging-in-Publication Data

Acoustic e m i s s i o n : c u r r e n t practice and future d i r e c t i o n s / W o l f g a n g
Sachse, James Roget, and Kusuo Yamaguchi, editors.
(STP ; 1077)
Papers presented at a symposium on world meeting on acoustic
emission, held in Charlotte, NC, on 20-23 March 1989, and
sponsored by AEWG.
"ASTM publication code number (PCN) 0 4 - 0 1 0 7 7 0 - 2 2 . "
Includes bibliographical references and indexes.
ISBN 0-8031 -1389-7
1. Acoustic emission testing. I. Sachse, Wolfgang, 1942-
II. Roget, James, 1949- III. Yamaguchi, K. (Kusuo). IV. AEWG
(Association). V. Series: ASTM special technical publication: 1077.
TA418.84.A2573 1991
620.2-- dc20 90-25872
CIP

Copyright 9 1991 by the American Society for Testing and Materials. All rights reserved.
No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopy, recording,
or otherwise, without prior written permission of the publisher.

NOTE
The Society is not responsible, as a body,
for the statements and opinions
advanced in this publication.

Peer Review Policy
Each paper published in this volume was evaluated by three peer reviewers. The authors
addressed all of the reviewers' comments to the satisfaction of both the technical editor(s)
and the ASTM Committee on Publications.
The quality of the papers in this publication reflects not only the obvious efforts of the
authors and the technical editor(s), but also the work of these peer reviewers. The ASTM
Committee on Publications acknowledges with appreciation their dedication and contribution
of time and effort on behalf of ASTM.

Printed in Baltimore,MD
February 1991

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

Foreword
This publication, Acoustic Emission: Current Practice and Future Directions, contains
papers presented at the symposium on World Meeting on Acoustic Emission held in Char-
lotte, NC on 20-23 March 1989. The symposium was sponsored by A E W G . Co-sponsoring
groups were ASTM Committee E - 7 on Nondestructive Testing, ASNT, IEEE, and SEM.
Professor Wolfgang Sachse of Cornell University, Dr. James Roget of Nordon and CIE,
and Professor Kusuo Yamaguchi of the University of Tokyo, presided as symposium chair-
man. They are also editors of this publication.

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

Contents
Overview--w. SACHSE, K. YAMAGUCHI, AND J. ROGET 1

A E SENSORS AND SYSTEMS

The General Problems of AE Sensors--v. HIGO AND H. INABA 7

Stress Wave Sensing--Affordable AE for Industry--T. J. HOI,ROYD, T. E. TRACI'Y,
N. RANDALL, AND S. D. KING 25

Monitoring Electron Beam Welding Process Using Electro-Magnetic Acoustic
Transducers ( E M A T ' s ) - - H . A. CROSTACK, H. J. STORP, AND P. BOHM 35

A E SOURCES AND WAVE PROPAGATION

Development and Future Aspects in AE Source Characterization--M. ENOKI AND "1.
KISHI 47

Joule Heating Line and Point AE Sources and the Adhesion of Thin Metal F i l m s - -
K. Y. KIM AND W. SACHSE 67

A Calibration Source for Acoustic Emission Analysis--c. R. HEII'I.E, S. tI.
CARPENTER, AND S. S. CHRISTIANSEN 77

Simultaneous Velocity Tomography and Source Location of Synthetic Acoustic
Emission D a t a - - s . C. MAXWEI,L, R. P. YOUNG, AND D. A. HUTCHINS $6

Acousto-Ultrasonics: An Update--A. VARY 95

Theoretical Basis of the Acousto-Ultrasonic Method--M. T. KIERNAN AND J. C.
DUKE, JR. 105

SIGNAl, PROCESSING APPROACHES

Acoustic Emission Technology Using Multi-Parameter Analysis of Waveform and
Application to GFRP Tensile T e s t s - - K . YAMAGUCHI, H. OYA1ZU, J. JOHKAJI,
AND Y. KOBAYASH1 123

Acoustic Emission Detection of Crack Presence and Crack Advance During F l i g h t
s. L. MCBRIDE, M. D. POLLARD, J. D. MaCPHAII,, P. S. BOWMAN, AND I), T.
PETERS 146

Structural Integrity Evaluation Using AE Tcchuiques--B. R. A. WOOD AND R. W.
HARRIS 156

Solving AE Problems by a Neural N e t w o r k - - l . GRABEC, W. SACIISE, AND E.
GOVEKAR 165

STRUCTURAL MONITORING APPI.ICATIONS

Periodic Inspection of Compressed Gas Cylinders and Transport Vessels by U s i n g
Acoustic Emission Testing--H. M. BARTItIh.~MY 185

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

Detectability of Defects in Reactor Pressure Components by Location and
Interpretation of AE-Sources--c. SKLARCZYKAND E. WASCHKIES 199

DEFORMATION STUDIES

Effect of Pre-Exposure to Water on the Acoustic Emission Behavior of 2091-T3 AI-Li
AIIoy--F. ZEIDES AND I. ROMAN 213

Acoustic Emission During Tensile Deformation and Fracture in Austenitic Alloys--
B. RAJ AND T. JAYAKUMAR 218

Relationship Between Acoustic Emission and Flaw Size in Si3N 4 Ceramics--Y. MORI
AND T. KISHI 242

A Comparison of the Acoustic Emission Generated from the Fracture and
Decohesion of Graphite Nodules with Theoretical Predictions--s. H.
CARPENTER AND Z. ZHU 252

Evaluation of Fatigue Crack Growth Rate of Carburized Gear by Acoustic Emission
Technique--Y. OBATA, H. KOBAYASHI, K. AOKI, "I'. YAMAGUCHI, AND K.
SHIBATA 261

NOVEL APPLICATIONS

Characterisation of Dust Impact at Low Velocity by Acoustic Emission--D. J.
BUTTLE AND C. B. SCRUBY 273

Applications of Acoustic Emission Techniques for Diagnosis of Large Rotating
Machinery and Mass Production Products--I. SATO, T. YONEYAMA, K. SATO,
T. TANAKA, M. YANAGIBASHI, AND K. TAKIKAWA 287

Cavitation Monitoring of Hydroturbines with RMS Acoustic Emission
M e a s u r e m e n t - - o . DERAKHSHAN, J. R. HOUGHTON, R. K. JONES, AND P. A.
MARCH 305

Tool Monitoring by Acoustic Emission--J. ROGET, P. SOUQUET, M. DESCHAMPS,
AND N. GSIB 316

Monitoring of the Machining Process by Means of Acoustic Emission Sensors--D. A.
DORNFELD 328

GEOTECHNICAL APPLICATIONS

Microseismics and Geotechnicai A p p l i c a t i o n s - - M . OHTSU 347

Acoustic Emission/Microseismic Activity at Very Low Strain Levels--B. H.
ARMSTRONG AND C. M. VALDES 358

Acoustic Emission Monitoring and Analysis Procedures Utilized During Deformation
Studies on Geologic M a t e r i a l s - - x . SUN, H. R. HARDY, JR., AND M. V. M. S.
RAO 365

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

Acoustic Emission Analysis and Ultrasonic Velocity Imaging in the Study of Rock
F a i l u r e - - s . D. FALLS, T. CHOW, R. P. YOUNG, AND D. A. HUTCttlNS 381

APPLICATIONS TO COMPOSITE MATERIALS

Fracture Mechanism Studies of a Carbon Fiber-Peek Composite by Acoustic
Emission--K. O N O , J. S. J E N G , A N D J. M. Y A N G 395

On the Correlation Between Acoustic Emission and Progression of Matrix Splitting
in a Unidirectional Graphite/Epoxy Composite--s. OHAFFARI AND J.
AWERBUCH 404

Identification of Fatigue Failure Modes in Carbon Fibre Reinforced Composites with
the Energy Discriminating Acoustic Emission Method--M. WEVERS, I.
V E R P O E S T , P. D e M E E S T E R , A N D E. A E R N O U D T 416

Detection of Impact Damage in Composite Bi-Axial Test Specimens by Use of
Thermally-Activated Acoustic Emission--J. w. WHITTAKER AND W. D.
BROSEY 424

Acoustic Emission Monitoring of Contact Drying of Southern Pine Veneer --F. c.
BEALL 435

Author Index 445

Subject Index 447

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

STP 1077-EB/Feb. 1991

INTRODUCTION

Acoustic emission (AE) is the phenomenon in which elastic or
stress waves are emitted from a rapid, localized change of strain
energy in a material. AE as a technology has rapidly become accepted
as a non-destructive methodology. It has become in recent years the
basis of a number of recommended practices and inspection codes of
several societies. The applications of AE which involves the detection
of AE signals and possibly their characterization are diverse. Most
commonly, they include the monitoring of manufacturing and other
dynamical processes, the integrity of structural components as well as
fundamental investigations of failure processes of engineering as well
as geological materials.

In the last decade the science, technology and applications of
AE progressed significantly. In order to provide a forum for reporting
important, recent developments and to provide an opportunity to
critically review the directions in which this field is moving, the
Acoustic Emission Working Group with the endorsement of other technical
societies, including ASTM Committee E-7, ASNT, IEEE, and SEM, organized
the World Meeting on Acoustic Emission which was held 20-23 March 1989
in Charlotte, North Carolina. Eighty-seven papers from nineteen
countries were presented at the conference. The thirty-four comprising
this ASTM Special Technical Publication (STP) volume were selected for
their topical content and international appeal.

The first section of this book focuses on AE sensors and systems.
The calibration of AE sensors and AE systems using the pencil break and
a reciprocity technique is discussed by Higo and Inaba. The
development of an integrated AE sensor suitable for use in harsh,
industrial environments and its use in diverse process monitoring
applications is reported by Holroyd et al. The application of non-
contact, electro-magnetic acoustic sensors (EMAT's) in an AE weld
monitoring application is described by Crostack et al.

The second section of the book deals with fundamental
investigations of AE sources and the propagation of simulated AE
signals through a structure for materials characterization
applications. The case of point sources modeling the formation of
microcracks in brittle solids is reviewed by Enoki and Kishi while the
signals from line sources are described in the paper by Kim and Sachse.
Heiple et al. describes a study of the AE accompanying the fracture of
boron particles in an aluminum matrix which may serve as an AE system
calibration signal, The use of a point source generating broadband
ultrasonic signals in a large number of directions in a material, forms
the basis of tomographic technique described by Maxwell et al. to
determine the velocity structure of a specimen which may find
application in the future to image the distribution of stresses or
cracks in materials, Another application utilizing simulated AE
signals is in the so-called acousto-ultrasonic, or AU technique, which

I
C o p y r i g h t 9 1991 b y ASTM I n t e r n a t i o n a l ~ee~.astm.org

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

2 ACOUSTICEMISSION was developed by Vary. The last paper in this section by Obata et al. No further reproductions authorized. A study of the ~E from the fracture and decohesion of graphite nodules in ductile cast iron is reported by Carpenter and Zhu. A theoretical foundation of the A U technique based on Lamb plate modes Js described by Kiernan and Duke. reports the development of a neural-like processing procedure for extracting the location and characteristics of an AE source from the signals detected at a number of sensors. Included is a paper by Barth61~my describing an AE-based inspection procedure for evaluating the integrity of compressed gas cylinders used in transportation systems. The final two papers of this section deal with the application of AE to monitor metal machining processes. He explores the use of adaptive and neural processing Copyright by ASTM Int'l (all rights reserved). The use of AE to investigate the effect of flaw size on the fracture of Si3N4 ceramics is reported by Hori and Kishi. The results of an investigation of the effect of pre- exposure to water on the AE behavior of an aluminum-lithium alloy are summarized in the paper by Zeides and Roman. but based on an envelope processing of the AE signals. Roget et al. Description of a high-performance AE system capable of rapidly extracting a number of waveform parameters from the detected signals is given by Yamaguchi et al. Derakhshan et al. . provide a means for delineating between growing and non-growing defects in reactor pressure components. describes the successful application of AE to monitor the growth of a fatigue crack in a carburized gear. report the use of rms AE measurements to monitor cavitation-generated pressure pulses in a hydroturbine. The next section of the book contains a number of papers reporting novel applications of AE measurements. Dornfeld reviews the generation of AE and its use as a monitoring procedure during a metal cutting process. The application of expert systems to assist in interpreting AE data is considered by Wood and Harris. Five papers comprise the section focusing on AE used in deformation studies and in investigations of environmental and cyclic loading effects. The second paper in this section is by Sklarczyk and Waschkies who demonstrate that AE signal parameters such as risetime. The next section contains four of the papers at the conference which dealt with new signal processing approaches for AE signals. Included here is the paper by Buttle and Scruby who apply quantitative AE techniques to determine the impact source and hence the sizes of small particles striking a plate. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. address the problems related to the application of AE measurements for delineating between tool breakage detection and tool wear monitoring. is the basis of an in-flight AE system reported by McBride et al. The use of AE in structural monitoring applications is the focus of the following section. in this volume he reviews recent developments and considers its further potential and some of its limitations. A similar approach. The development of a digital AE-based system for machinery diagnostics applications is described in the paper by Sat et al. The last paper in this section by Grabec et al. A study of AE generated during tensile deformation and fracture in austenitic alloys is described by Raj and Jayakumar.

OVERVIEW 3 procedures for analyzing the detected AE signals. If applied properly. ~Journal of Acoustic Emission (1990). The last paper. written by Beall. Wevers et al. Whittaker and Brosey describe the use of a cyclic thermal loadings to generate AE by which impact damage in KevlarR-wound aluminum spheres can be detected. No further reproductions authorized." which was led by A. describes the novel combination of AE source studies and ultrasonic tomographic imaging to investigate the failure mechanisms in rocks. The editors express their deep appreciation to the more than fifty reviewers who so carefully read all the manuscripts and provided critical reviews of them. The final paper in this section by Falls et al. A panel discussion was organized which was chaired by D. such multi- measurement techniques can yield significantly more information about a process than when used individually. Other members of the panel included Y. Pollock I (Physical Acoustics Corp. J. describe the use of an energy-related measure of the AE signals to monitor the damage development in a fatigue-loaded carbon fiber/epoxy laminate. Ghaffari and Awerbuch describe the correlation they establish between AE and the initiation.) and "Critical Instrumentation Issues. and tensile tests of a thermoplastic carbon fiber-PEEK composite are reported by Ono et al. Each panel member presented an overview of recent developments and trends in his country. The use of AE to detect microstrains in the earth prior to an earthquake is considered by Armstrong and Valdes. Roget (France) and P. It was agreed that a continuing exchange of information about the development of AE-related codes and standards among AE groups would be desirable and an important undertaking. Beattie (Sandia National Laboratories). Eitzen (USA) and which was used to exchange information about the status of AE-related codes and standards in several countries. There were two sessions at the conference for which no papers are included in the book. reports the use of AE to monitor the contact drying process of a wood veneer. Tscheliesnig (Austria). In Press. G. accumulation and progression of matrix splitting in unidirectional graphite/epoxy specimens. flexure. . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The AE parameters that can be used to identify the failure mechanisms initiated by bending. A survey of AE source characterization studies yielding the moment tensor components and the ability of characterize the crack type and its orientation in geological specimens is described by Ohtsu. Higo (Japan). Unique to this conference was an evening discussion session focusing on the topics "Critical AE Problems for the Researcher. The final section of the book focuses on the application of AE measurements for investigation failure processes in composite materials." which was led by A. The next paper in this section reports on the use of a novel amplitude analysis procedure to delineate between several deformation mechanisms in geological materials is described by Sun et al. Three papers comprise the section dealing with geotechnical applications of AE. Copyright by ASTM Int'l (all rights reserved).

Investigation of large amplitude AE burst-type emission before and during yield of some materials. Japan James Roget Nordon & Cie Nancy. (3) The development of realistic artificial sources for AE system calibration application. The efforts of Barbara Stafford. Application of chaos theory to AE. These included: Study of AE signal cascades. 4 ACOUSTICEMISSION The discussors identified the following issues for further investigation: (I) The development of a coherent. Quantitative AE from frictional sources. . NY USA Kusuo Yamaguchi University of Tokyo Tokyo. both by improved detection methods and by new or novel methods of signal processing. Therese Pravitz. Absolute acoustic measurements. unified theory of acoustic emission to explain phenomena such as the Kaiser and Felicity effects for materials subjected to repeated loadings or load-hold tests. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. and Kathy Greene deserve special recognition. Additional investigation of elastic waves in plates and shells for source location and characterization applications. No further reproductions authorized. New and imaginative AE applications. Exploration of AE in conjunction with other NDT techniques. A note of thanks must be extended to the editorial staff of ASTM without whom this project could not have been undertaken. France Copyright by ASTM Int'l (all rights reserved). (2) The development of rational guidelines for the realistic stimulation of AE in structures which may be subjected to multi- dimensional states of thermal or mechanical stress in service. Wolfgang Sachse Cornell University Ithaca. A number of additional topics were identified by members of the audience for future investigation. and (4) Ways of obtaining more information from the detected acoustic signal. Application of neural networks to analyze AE signals.

No further reproductions authorized. AE Sensors and Systems Copyright by ASTM Int'l (all rights reserved). . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.

227. Sannomiya.. both in the time and frequency domain. W. A B S T R A C T : The general problems of AE sensors are overviewed. Mr.org Copyright by ASTM Int'l (all rights reserved). Then the sensitivity measurement methods are compared. the AE analysis methods. have progressed and are widely applied not only to the fundamental research but also to the evaluation of the actual structure and equipments. Tokyo Institute of Technology. ASTM STP 1077.. 7 Copyright9 1991 by ASTM International www. pencil lead The technology and equipment of AE have been progressed significantly in the past ten years. Y. The results of a sensor's sensitivity. No further reproductions authorized. and Hidehiro Inaba THE GENERAL PROBLEMS OF AE SENSORS REFERENCE: Higo. the Research Laboratory of Precision Machinery and Electronics. "The General Problems o f AK Sensors. The convenient calibration method for sensor sensitivities. Nagatsuta. American Society for Testing and Materials. H. J. The characteristics of the lead is quite similar to the lead produced in 1975. Especially. 418-01.. was demonstrated. Yamaguchi. Inaba is a research scientist of the Fuji Ceramics Co. Eds. especially the "acoustic pressure method" and the "pencil lead fracture method" are discuss ed. acoustic pressure method. The characteristics of lead. obtained by NBS and by the Hatano method show that there is very little difference between them.astm. Then sensor sensitivities were obtained by the standard lead and are compared with the results of the reciprocal method. specially standard pencil lead from JAEWG which produce very stable reproducible AE signals. Hujinomiya-shi. Yakichi Higo. Higo is an Associate Professor of Materials Science Division. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Roger. . sensor sensitivity. Yokohama. Japan. sensor mounting condition. KEYWORDS: AE. Japan. Midoriku. Philadelphia 1991. Sachse. Ltd. The effects of mounting conditions of sensors on the sensitivity are discussed.. Dr." Acoustic Rm4~sion: G~rrent Practice and Future Directions. and K. and I n a b a .

The general problems of the sensor are. Some phenomenon occurs in the specimen and it emits AE. This leads to difficulty in exchanging and comparing actual AE signal data among different research groups using different equipments and sensors. Thus. No further reproductions authorized.Then an AE wave propagates through the specimen and is detected by the AE sensor. I Equipment I 3) Amplifier . Therefore. .Basic AE measurement system The basic AE measurement system consists of an AE sensor. 8 ACOUSTICEMISSION 1) M o u n t i n g Condition Coup iant Pressure I 4) AE ] l~.1 -. a) Effect of mounting condition (couplant. a) Effect of mounting condition on sensitivity. amplifier and AE signal analyzing equipment. even if the same sensor and A E equipment are used under the same measurement conditions. even if sensor sensitivities have been calibrated. mounting pressure and so on) on sensitivity (amplitude and phase components). the following subjects are mentioned and discussed. where the AE changes from an elastic wave to an electrical signal. However. the AE sensor has many kinds of unclarified problems to be solved. Then. the AE signals are processed by using electronic devices and technology. These factors mentioned above affect detected A E signals. Specimen Fig. Copyright by ASTM Int'l (all rights reserved). b) Sensor sensitivity. c) Degradation of the sensitivity and convenient methods for its calibration. the results of A E event count m a y be changed depending on the mounting condition. In this paper. The AE sensor is the most important part of theAE equipment.'~Band pass filter L~nearity alzc . and change the peak voltage. as schematically shown in Figure 1. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. it is quite important to find some solution to make it possible to compare AE data and for producing an AE data base in the near future. b) Sensor sensitivity and its measurement methods. c) Degradation of the sensitivity and its method of evaluation. duration time and so on.

2) mounting pressure (mechanical force to hold the sensor against the structure or the specimen) and 3) surface condition of the specimen. PA= FHA. fit).2(b) is expressed as follows. The transfer function through the whole system. HA. The phase component directly affects to the receiving AE wave form. MH. No further reproductions authorized. g^(t) is the output signal of the receiving sensor T H. arrival time and so on. Mori [3] and Breckenridge et al.The system in Fig. [4]. this condition affects not only the peak voltage but also the number of event counts. the voltage sensitivity of receiving sensor. HIGO AND INABA ON PROBLEMSOF SENSORS 9 MOUNTING CONDITION OF AE SENSORS The factors of mounting condition which strongly relate to the sensitivity are thought to be 1) the couplant material between the sensor and AE wave propagation medium. gA(t) is the output signal of the receiving sensor TII [1. . GA = HA = S: [~IhA c~:M:: (:) F where F and G^ are the Fourier transform off(t) and gA(t) respectively.the schematic diagram showing the transmitting input signal to the transmitting sensor TI. to the transmitting sensor T:. Therefore. M~I and hA are the transfer function of the transmitting and the receiving sensor and the wave propagation medium respectively.. SI.. Figure 2 is a schematic diagram showing the transmitting input signal. in Fig. Because. These factors affect not only the resonance frequency but also phase component of the AE sensor's sensitivity. the most important thing is to know the characteristics and the reproducibility of the mounting condition.l--" gA(t) (13) I HA . Measuring method f(t) 4TII A IT.= GA aIM:: a: MH (2) Copyright by ASTM Int'l (all rights reserved). Firstly.2]. A. . Then the acoustic pressure spectral density function at the surface. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.2(b) according to their components.2(a) can be described using the transfer functions in frequency domain as shown in Fig. ~ and cq are the transfer functions representing the coupling condition between the sensor and the propagation medium. F hA Fig.. is obtained by the reciprocity calibration procedure developed by Hatano. when AE waves are analyzed in time domain.2 -. is obtained as follows.

Gn _ 0~n GA aA (3) A a~ includes not only amplitude but also phase component. 2SO L!N FREO r IO{} Fig. .qM. which wa: synthesized by a computer using the equation (4) and (5) [5.V o T I HE (z) 100 (b) -3o dBv C~o ~BH~ S 10dB/ --120 ~BV I I 1 0.4 / 0)kAW [ X . and Ok is random n u m b e r between 0 and 2. 0~k=2rCk IF((0k~2 -_1_! .h. . 7 . m -44.:4'~l!l l).7 ]..6.~l't)~ttll)'li~' . time domain (a) and frequency spectrum (b).)'t)1.r k) m A T < t < ( m + I ) A T (4) k=1 "3072 ' where T is the repeat period ( T = 3072 AT ).. hm = a A(6h ' k + 3072n + 6h... when the same sensors anc propagation medium are used and the mounting condition is changed. the difference o mounting condition. M l 1 1 1 ~ Xo-~ ~nnlliU. where Fig. The signal used for the measurement was periodic pseudo-random noise.~ shows an example of both f(t) and F(co).. .~)!)llj[ri'~IT.~. Act.. m is an integer f r o m I to 3072.. 10 ACOUSTICEMISSION If the signal F has good reproducibility.m& ~ .. . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. A(Zn-H~ .t~oJI~IUF~'. . then it is possible to identify PA. The Fourier transform of f(t) is F(a)~). Therefore. is expressed as follows. No further reproductions authorized.3 -.Pseudo-random noise . _k + 3072n) ' t (5) 2 / (a) ~44o7 ~v mmTl~m~l~ m~immlmmm?:ll. Copyright by ASTM Int'l (all rights reserved). 512 fit)= ~ A c o s ( 2 m k . and if the same transfer system is used excep' for MH and a l ..

When using W-40s grease (mentioned later).o~ Noise wave Propagating ~ XSO0. The block diagram of the measuring system is shown in Fig.~--~ 'Transient ' Recorder I Ocncmtor System I V I Model 204-A I ] I 11 PC-8801 I { GP-B 'I1 MS-430A I Fig. Effect of mounting pressure In order to obtain good coupling conditions for detecting AE. Sampling clock i peu. -60 I I I I I 0 5 10 15 20 25 Mounting p r e s s u r e ( K g / c m =) Fig./-~--~'k-~--A------A A.~-~ -[] .~010 ~ 0 ~ 0 0 -42 -44 / -46 -9 .O. In addition of these characteristics..4 -..7MPa for a dry contact according to ASTM E650-85....0 .. mechanical force is applied to hold the sensor against the structure orthe specimen. HIGO AND INABA ON PROBLEMS OF SENSORS 11 The characteristics of the signal are that the power is flat in both the time an frequency domain. 0-1MHz -56 -56. and the power and phase of each frequency are completely identified [6].Q-o.. The sensitivity increases with increasing pressure._. __.--~. -40 0 0 . I -52 re @._. 2-3MHz ~_A_~-A__. the relationship between receiving sensitivity of a 20rome flat type sensor (frequency range was 100kHz to 5MHz) and the pressure is shown in Fig.. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement..EO_O_EO~ . 4-5MHz A m -48 .. No further reproductions authorized.5 -. -O .. 0 . 3-4MHz -50 /O0... O0 ~0 -. The power spectrum of receiving signals are indicated for each 1MHz range ( shown b3 different symbols in the figure). Copyright by ASTM Int'l (all rights reserved).____.-. O..4. D4:~.O~D. The power spectrum of receiving signals are indicated for each 1MHz range ( shown by different symbols in the figure).Block diagram of the measuring system.~=~O 0 O -9 ..5. This pressure should greater than 0. .Relationship between receiving sensitivity of a 20mm~ flat type sensor (frequency range was 100kHz to 5MHz) and the pressure. 1-2MHz -54 -A .. the important thing is its very precise reproducibility... ..

The effect of couplant on sensor sensitivity obtained by the reciprocal method. to detect over 4MHz. However. P w >.1 o -"s o ~do ebo ldoo le~ frequency (KHz) (a) I :~ xl 0. more than 10Kg/mm (1MPa) is required [8]. ~do 8bo 12'00 16~ frequency (KHz) (b) Fig.03MPa) of applied force is sufficient. Effect of couplant on sensitiVity 20. No further reproductions authorized. 12 ACOUSTICEMISSION When the detecting frequency range is up to 2MHz. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Copyright by ASTM Int'l (all rights reserved). . > 0. 300g/mm (0. (a) is for a poorly coupled sensor with air babbles in the couplant and (b) is carefully mounted using cyanoacrylate.6 -.

Copyright by ASTM Int'l (all rights reserved).7 -. ~.d and e) The couplants used were silicone oil. The resonant frequency of each sensor is almost the same. the sensor was rightly attached to the AE propagation medium with cyanoacrylate.q -|1 1 FREOUENCY (tIHz) (e) Fig. When the couplant is looser because of included bubbles and so on. mixture of silicone grease and oil.8~ o~ -o .. and is about 350KHz [1]. Figure 6(b) demonstrates the effect of a loose couplant condition [1].q -tBe B I 2 ~ I 2 FREGUENCY (HHz} FREgUENCu (MHz) (a) (b) a-l: 0-15 .r e :e - 1 FREQUENCY (HHz) FREOUENCY (HHz) (c) (d) o-g o-l~ |8o . the output of the receiving sensor will be changed even if the acoustic pressure spectral density function and the voltage sensitivity of the sensor itself are exactly the same. pine resin base wax and W-400 respectively. . . (a. then the sensitivity was measured (Fig.6a). . There is a tremendous difference between the two mounting conditions. In this study.6(b) was obtained for a poorly coupled sensor. However.b. We assumed that the an and ai are same.c. The result show n in Fig. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.The effect of couplant material on the reproducibility of phase component of sensitivity. HIGO AND INABA ON PROBLEMS OF SENSORS 13 Figure 6(a) shows the~voltage sensitivity for longitudinal wave of a sensor obtained by the reciprocity calibration procedure using three similar sensors. Ac~. . No further reproductions authorized. silicon grease. m . the sensors were carefully bonded to the propagation medium. . The sensor was carefully mounted. it is not always to maintain the couplant condition between the sensor and the propagation medium. . For comparison.

6(a). No further reproductions authorized. the reproducibility of amplitude component oJ sensitivity is not so bad. Solid or higher viscosity couplants such as pine-resin. Thus. This range most affects the detected AE wave form. The best results a room temperature were obtained with W-400 couplant [2]. 14 ACOUSTICEMISSION _Effect of couplant material on phase component of sensitivit_y When a sensor was carefully mounted with various couplants and the sensitivity measured and compared with Fig. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. the analyzed results of AE parameter might be changed. Ac~ of equation (3) [2]. However.The power spectrum of receiving signals detected using silicon grease (a) and W-400 couplant (b). glue or silicon grease gave bac results. Figure 7 shows the effect o: couplant material on the reproducibility of phase component. the reproducibility of the phase component of th~ sensitivity is strongly related to the material of the couplant. the 0 0-7 -100 0 1 FREQUENCY (MHz) (a) 0-17 W Z -100 0 i 2 FREQUENCY (MHz) (b) Fig. especially in low frequency range. However. . Copyright by ASTM Int'l (all rights reserved).8 -.

HIGO AND INABA ON PROBLEMS OF SENSORS 15

powerspectrum of receiving signals detected using silicon grease and W-400 couplant
were almost same, as shown in Fig.8 [2] . Therefore, in this study, W-400 was the best.
However, couplant materials select carefully for AE wave analysis in time domain.

SENSOR SENSITIVITIES AND ITS MEASUREMENT METHODS

Many methods to measure the sensor sensitivities have been proposed in the last twenty
years. Typical methods to obtain the quantitative sensitivity are two, proposed by NBS
[4,9] and Hatano et al [3].

V/m/s Longitudinal wave
I I I i i

3
NAIS F-2215H No 610
s NSC

"~ 2000

T

0 I I I I~r
,-'--Y'~"'~t"-'~'~
0 1.0
Frequency ( NHz )

Fig.9 -- Sensor sensitivities obtained by NBS and Nippon Steel (Hatano's
Method).

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

16 ACOUSTICEMISSION

In 1982, the results of both methods were compared using same sensors [4]. Figure c
shows one of the results. At that time, it was concluded that there was little difference
between them, ~ in details, such as the peak height of the resonant frequency are
slightly different. The reason for the difference was mainly thought to be the mounting

dB(OdB=lV/m/s) Long;tudlnal W~ve
70
(a)

"-X
5O

._>
--,...
\
c'-

30

10

O0 500 000 500 2000
Frequency (KHz)

dB(OdB=lV/m/s) Longitud;nal W~ve
70
(b)

50

U)
3o
09
A,
,o 'f1
100 500 1000 500 2000
Frequency (KHz)
Fig.10 -- (a) is the sensitivity of the sensor before applying thermal cycles and
(b) is after the cycles.

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

HIGO AND INABA ON PROBLEMS OF SENSORS 17

and the surface condition of the AE wave propagating medium. However, There are
some limitation on both methods, caused by
1) size and the alignment of AE sensor with the specimen,
2) material characteristics (elastic constant etc.) on which sensor mounted,
3) accuracy of the distances between AE sensor, reference AE sensor and quasi AE
source on NBS method,
4) transmitting characteristics of AE sensor on Hatano's method.

DEGRADATION OF THE SENSITIVITY
AND A METHOD FOR ITS CONVENIENT CALIBRATION

We quite often experience a degradation of sensor sensitivity, especially when the AE
method is applied to measure the characteristics of super conducting materials or
monitor the super conducting magnet at liquid He temperature. The AE sensor
experiences thermal cycles. During the cycle, degradation of sensitivity occurs because
of the thermal expansion coefficient of sensor component materials are different. If the
sensor is significantly damaged, it is very easy to find the degradation. However, when
the degradation is not so obvious, it is quite difficult to find it. Figure 10 shows the former
case [10]. Before the thermal cycles, the sensitivity was measured (Fig.10(a)). Then
thermal cycles were applied five times, and the sensitivity was measured (Fig.10(b)).
The sensitivity decreased about 30dB. However, for the latter case, we need some simple,
quick and convenient calibration method to know the characteristics of degradation.

Convenient calibration method (Acoustic pressure method)

As mentioned in a previous section, we have a method for mounting the sensor to
obtain good reproducibility. Using a periodic pseudo-random noise, sensor sensitivity,
Mn, is derived from equation (2) as follows [1],

Mn = p ~ (6)

The sensitivity difference between before, M,b, and after thermal cycles, M,,, is

AMnba = -Mn--~-a (7)
Mnb

Figure 11 demonstrates the small degradation o f sensitivity after 5 thermal cycles,
obtained by the method. Figure ll(a) is the sensitivity of the sensor before applying
thermal cycles, (b) is after the cycles and (c) is the difference between them.

C o n v e n i e n t calibration method (Pencil lead fracture method)

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

18 ACOUSTICEMISSION

48(O49=IV/m/s) LongiLud;n~| W~ve
70

c~

5O

(a) > 30
.:.
a)
c
CD
09 ~0

100 500 ~000 1500 2000
Frequency (KHz)
dB(OdB=IV/=Is| Lo~tu4~l W~ve
70

50

(b) >, 3O

10

O9
IO 0 500 1000 1500 2000
Frequency (KHz)

(gB,'
I0

.0 0
, , l
<~ -10
>,
(C) "~ -20

u) -30
c
~D
09 -~0
00 500 1000 i500 2000

Frequency (KHz)

Fig.11 -- (a) is the sensitivity of the sensor before applying thermal cycles, (b)
is after the cycles and (c) is the difference between them.

The two methods mentioned above, NBS and NSC, are suitable for measuring th(
precise sensitivities, but they are not convenient for general u s e , because the system i~
heavy in weight and also very costly in price. Thus some alternative convenient
methods have been proposed, such as, the helium gas jet or pencil lead fracture method.
The pencil lead fracture method (PLF) has been proposed by Dr. Hsu (NBS) in 1975. This
method is popular, ASTM has adopted it as a standard method [11]. PLF used to us(
Pentel 2H 0.5mm lead as a standard. However, the size of the lead changed to 0.3mm

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

HIGO AND INABA ON PROBLEMS OF SENSORS 19

I00 (a)

_c
0
0 0

c- 0

I

50 o
o
o 0 o

>
I 1 I
< 0 , , , , , , , , , , ,

75 80 85
Products year
0.5r 2 H

~. I O 0
(b)
v

c-

c-
o
o
0
o
.~ 50 o
o

o 0 o

, 1 , , , , I , l , I I , ,
< 0
75 8O 85
products year
0.5~ H B

Fig.12 -- Average strength (g) of the lead produced each year from 1974 to 1978,
measured by three points bending ( span was 40 mm)o

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

20 ACOUSTICEMISSION

[12]. Figure 12 shows the historical change of strength of Pentel 0.5mm 2H and HB [13].
The characteristics and minufacturing method of lead has been changed since 1980.
During the period, the strength of the lead was improved and doubled. Therefore the
characteristics of AE, emitted by the fracture of it, were also changed. The most
important characteristics of the lead are the reproducibility of fracture strength (Weibull
Coefficient) and of the AE wave. A measurement system for AE waves emitted by pencil
lead fracture is schematically shown in Fig.13 [13]. Figure 14 shows the peak voltage ot
AE emitted by the pencil lead manufactured from 1974 to 1988, corresponding to Fig.12.
Peak voltage showed much more scatter with 0.5mm 2H.

Pente 1
nc i I Pre-amp
~3~ SteeI Gain 4 0 d 8 Oigital
storage
v oscilloscope
/ Ibl~ \ Sensor ~>~5 usec/Oiq
/Guide rinq \flat typeiO45s [y. 0 2 V/Oiv ]
\ No.8055 - " "

Fig.13 -- Measurement system for AE wave emitted due to the pencil lead
fracture.

The pencil lead which is sold at ordinary stationary shops have a Weibull Coefficient
(WC) of about 10 and the reproducibility of AE is quite bad. The WC in 1975 was about 20.
JAEWG tried producing and supplying standard pencil lead which has very similar
characteristics of strength and AE signals as the lead produced in 1975.
The WC of standard lead is above 40 and it gives extremely high reproducibility of AE
waves [13,14]. Figure 15 shows the signal emitted due to the lead fracture and the
sensitivity of the sensor used for the signal measurement.
When the pencil lead is fractured, the lead is gripped in the mechanical pencil with
teflon guide ring at the tip. The ring is very important factor for obtaining good
reproducibility of AE waves. Figure 16(a) shows the shape of the most widely used ring.
However, if the collar of the ring is very weak, it deforms very easily when the lead is
fractured, and causes a decrease in reproducibility of AE. Figure 16(b) shows th~
modified guide ring with more stiffness atthe collar. Also a special mechanical pencil
for AE was manufactured to avoid scratching the pencil lead surface. The
reproducibility of the signal was so good that the sensor sensitivity can be obtained using
the signal as follows(6)
Figure 17 demonstrates the sensor sensitivity obtained by the standard pencil lead
fracture method (Fig.17(a)), and by the reciprocal method (Fig.17(b)). Both results
agree.

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

HIGO AND INABA ON PROBLEMS OF SENSORS 21

(a)

J..O

o0.5

IKJ

O-

I I I I I
0 , , , , , ~ , , ' ' , , , ,

70 75 B0 85
Products year
0,51 2 H

1.0 (b)

v

+t
~

0 I J , J ~ I , I J f ,[ I i , J I , , ,

70 75 80 B5
ProducEs year
0.51 H B
Fig.14 -- Relation between the peak voltage of AE emitted due to the pencil
lead fracture of 0.5mm 2H (a) and 0.5mm HB (b), and manufactured year,
corresponding to Fig. 12. Scatter bands of peak voltage are shown by bars with
the mean value by open circle.

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

22 ACOUSTIC EMISSION

(a)

(1)
o')

0
>

c)..
"5
0

Time
M.~,X=-35.6542 d8
-30

(b)
-40

~U
"EJ -50

Q. -60
E
< -7o
-80 ~uA^~i,Nt~,.,A~
IvV ' ^k/~ ~,~
-go
0 2000 4000 5000
Frequency (KHz)
dB(OdB=IV/m/s) Long~tud~n~l Wave
70
~)
50

c}.
30
E
<
~0!

i
100 500 1000 1500 2000
Frequency (KHz)

Fig.15 -- The signal emitted from a standard pencil lead fracture. (a) is in
time domain (Vertical axis; 0.2V/Div, Horizontal axis; 5 ~m sec/Div), (b) is
power spectrum of (a), and (c) is the sensor sensitivity for the signal
measurement.

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

Iid~ Denci:l.-V / / (b) Fig. . db(Odb= tvlm/s) Longitudinal Wave db(Oclb=lv/m/s) Longitudinal Wave 70 to <E <E 1o Io ItT.The sensor sensitivity obtained by the standard pencil lead fracture method (a). / . ~ " fl 2H Pente).~'~-V / / (a) Pent_e :1. /' . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. t ~ ~oo Frequency (KHz) Frequency (KHz) (a) by Pencil (b) by Reciprocal Method Fig. oia. HIGO AND INABA ON PROBLEMS OF SENSORS 23 Pente 1 Teflon guide pent•I\ ~ / r~ng ~ ~ / .'~ o .-" Teflon n. 5mm oia. Copyright by ASTM Int'l (all rights reserved)..16 -...~-. / Lead: ""fi~'%Y o. /:.~. and by the reciprocal method (b). .Traditional shape of a teflon guide ring (a) and a modified one (b). No further reproductions authorized..~'~::~.~. ~ / Lead: ] \Xr . 17 -.:~%.5.

[11] ASTM standard E1067-85.. pp. Proceeding of the 2nd Symposium on Nondestructive Evaluation for New Materials. 13. [13] Higo. Ono. pp. (in Japanese).. 1. Then sensor sensitivities obtained by the standard lead and the reciprocal method were compared. Nunomura. 1986 pp. [2] Higo. H. and Inaba.s24 . JSNDI. progress in Acoustic Emission~ vol.27.. pp. [5] Hushimi. [3] Hatano. The characteristics of lead. vol.. E976-84 and others. Y.169. 103 . A.. F. The characteristics of the lead are quite similar to that produced in 1975. Journal of Acoustic Society America. [14] Higo. et al. [7] Higo. vol.. pp.131-137. 18. H.114. [4] Breckenridge. especially the "acoustic pressure method" and the "pencil lead fracture method" were discussed. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. vol... K. M. Y. vol. obtained by the NBS and by the Hatano method showed that they yielded very similar results. 1988. E. 1989. Journal for Acoustic Emission.343 . Progress in Acoustic Emission. [12] ASME Boiler and Pressure Vessel Code (1983) and others. [6] Nakamura.230 - 238. Progress in Acoustic Emission. pp. S.164 . A. pp. Y. 1978. [8] Higo. 1989.407 . 1980. and Hatano.. 1984. specially standard pencil lead of JAEWG which result in stable reproducibility of generating AE signals. T. 1976. pp.7-141.448 . Higo. Japan Journal of Applied Physic_s. Y. 24 ACOUSTIC EMISSION CONCLUSION The general problems of AE sensors were reviewed. Progress in Acoustic Emission. 4. Takashima. pp. H... A.344 - 349. pp... Y. vol. The effects of mounting condition of the sensor on the sensitivity were discussed and the sensitivity measurement methods were compared. and Kazama. and Inaba. and Wada. S. H. . vol. vol. [10] Ninomiya.59. was demonstrated. H.. Nakamura. S. Higo. Tokyo. Applied Physics. and Wada. M. 1976.685 . Watanabe. and Wada.. pp. Y.451. Proceeding of Japan Electric Society Spring Meeting. and Nunomura. 1988. H. The convenient calibration methods for sensor sensitivities. 1982. Proceeding of 5th International Acoustic Emission Symposium.. [9] ASTM standard El106-86. No further reproductions authorized. et al.691.. 3.350.458. 2. Y. A.. Y. Copyright by ASTM Int'l (all rights reserved). and Mori. The results of sensor's sensitivity. 16. REFERENCES [1] Ono. et al. Japan.

Wirksworth. Derbyshire. KEYWORDS: integrated sensor.org Copyright by ASTM Int'l (all rights reserved).. Trevor J. ASTM STP 1077. T. The use of signal enveloping techniques in the sensor eases the subsequent handling of its outputs whilst retaining the useful information in the detected stress wave signal. T. D.. Mr Tracey is Analogue Systems Engineer. J. 25 C o p y r i g h t 9 1991 b y ASTM I n t e r n a t i o n a l ~. stress wave sensing INTRODUCTION The industrial requirements of AE hardware have been critically reappraised and a shift in emphasis from a non-destructive testing role to a general sensing role has been identified. The sensor is mounted in a robust and easy to install housing which greatly simplifies the sensing of stress wave activities in the industrial environment. Tracey. American Society for Testing and Materials. Timothy E. Roget and K. Philadelphia 1991. Neil Randall and Simon D King STRESS WAVE SENSING . . S. Holroyd. Mr Randall is Product Development Specialist and Mr King is Applications Specialist. E. thick-film hybrid. No further reproductions authorized. industrial applications. ABSTRACT: A fully integrated stress wave sensor has been created which incorporates the basic functions of an AE system.Acoustic Emission: Current Practice and Future Directions. J. Ravenstor Rd. and King. A number of applications are discussed to illustrate the versatility of the approach described. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. DE4 4FY. Eds. This shift requires 'AE systems' which are radically different to conventional AE systems in terms of their appearance and performance.. The label Stress Wave Sensing has been chosen due to the almost universal use of the term stress waves to describe propagating activity detected Dr Holroyd is Technical Director. Yamaquchi. as well as mode of use. Sachse. UK. Stresswave Technology Limited. W. "Stress Wave Sensing Affordable AE for Industry". Tracey. stress wave sensor.astm. Randall.AFFORDABLE AE FOR INDUSTRY REFERENCE: Holroyd. N. In view of this it is felt appropriate to rename the approach presently being pursued (SWS) 'Stress Wave Sensing' to avoid confusion with conventional (AE) Acoustic Emission.

According to this definition. amplification and signal processing. REALISATION OF THE STRESS WAVE SENSOR At its heart an AE system consists of a transducer. All too often these pockets of research aimed at various process and condition monitoring applications have not been fully developed and applied as a result of the high cost of AE hardware and its all too apparent laboratory origins. In doing this a major breakthrough has been the integration of the transducer into the hybrid circuit in the midst of its conditioning and processing electronics. ruggedness and ease of installation since these all affect both the viability and cost effectiveness of an application. In a stress wave sensor these functions need to be integrated into a common housing. location and characterisation technique which has been its main thrust since the sixties. The transducer has a mechanical resonance at the working frequency and the output from the transducer element is amplified within a narrow bandwidth centred on the transducer resonance. This results in a combined transducer and amplifier response which takes a simple form. therefore. There are also ample opportunities for trading off certain aspects of performance against cost as discussed in detail in [2]. These considerations led to the concept of a fully integrated AE system or stress wave sensor. No further reproductions authorized. . In view of the resonant detection there is little to be gained from analysing the frequency content of the raw signal. For such applications the users primary requirements of the hardware are reliability. Acoustic Emission activity is regarded as a specific category of stress wave activity. Taking these considerations into account the concept of applying AE as a simple sensor was developed in which the measurand was stress wave activity whatever its origin as discussed in [i]. a load sensor senses load and a stress wave sensor senses stress waves). Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. To achieve this in a physically acceptable package size it was necessary to use thick-film hybridisation techniques in addition to component efficient designs. 26 ACOUSTICEMISSION by AE transducers irrespective of the operative causal mechanisms (ie a pressure sensor senses pressure. THE STRESS WAVE SENSOR CONCEPT The potential for monitoring the wider physical world using AE hardware has been illustrated over the years in a number of papers which have broadened the horizons of the technique from the defect detection. Furthermore in view of the random nature of the source processes and the variables and uncertainties in the propagation path there will be no usable Copyright by ASTM Int'l (all rights reserved). Such a design takes advantage of the compatible acoustic properties of the alumina substrate used in the construction of the hybrid circuit and the piezoelectric element which forms the transducer element. as shown in Figs 1 & 2. Such a sensor should be easy to install on industrial machinery and provide a readily usable output.

The resulting hybrid element which forms the heart of the integrated stress wave sensor is shown in Fig 3 and measures only 32mm by 32mm. A photograph of the sensor is presented in Fig 4. This is mounted in a rugged housing to make it easier to handle and in order to protect it from the environment although in principle it could be bonded to any surface to provide an 'AE' monitoring function. with regard to sensitivity. that of the housing. Under these circumstances a small reduction of sensitivity caused by distancing the transducer element from the workpiece is largely irrelevant. The sensor housing is excited at some point or in some region and this causes a complex field of stress waves to evolve which can be described statistically via a measurement at any point on the housing. Since the envelope of the signal is generated by an analogue circuit there is a decision to be made concerning the value of the time constant over which the smoothing process is to be carried out. Hence. electromagnetic shielding and the necessary acoustic properties. A long time constant output (ie a slowly responding voltage signal) is convenient in that it can be easily read from a meter display or plotted on a chart recorder. for example. the stress wave sensor housing as a whole reverberates in response to stress wave excitation. In view of this enveloping has been used as the primary signal processing method. resistance to chemical attack. As well as reducing size the incorporation of the transducer into the electronics gives considerable benefits in terms of ruggedness and cost. the usable part of the signal resides solely in its amplitude (both its absolute level and the nature of its variations). Secondly. In order to satisfy these two conflicting needs the design of the present stress wave sensor incorporates a dual enveloping circuit having fast (i00 ~sec) and slow (i sec) time constants. HOLROYD ET AL. the theoretical loss of temporal resolution through the choice of a 'diffuse-field housing' is also irrelevant provided the enveloping time constant is comparable to. Considering the baseplate of the sensor therefore this means that it is possible to quantitatively relate the activity detected at the sensitive piezoelectric element to that at the point of sensor housing attachment even though it is at a remote (and in principle arbitrary) position. ON STRESS WAVE SENSING 27 information in the phase of the signal. . The casing material is cast stainless steel in order to give mechanical strength. Firstly. In other words the sensor is designed from a diffuse field standpoint. Copyright by ASTM Int'l (all rights reserved). neither of these present a practical problem for the intended use of these sensors. No further reproductions authorized. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. whilst a short time constant output (ie a rapidly responding voltage signal) is convenient for analysing fine structure and timing measurements (eg via an oscilloscope or other instrumentation system). or larger than. Whereas a typical AE transducer has a (quite delicate) face which is coupled to the workpiece of interest. for many industrial applications the limit of detection is set by other interfering stress wave activity not the sensitivity and electronic noise level of the transducer. Whilst in principle this approach has disadvantages in terms of the limit of detection and the temporal resolution of activity. The use of a tab mounting design represents a radical departure from traditional AE design practice.

5 1. 28 ACOUSTIC EMISSION 60 40 20 o~ d -20 -40 25 50 75 I00 Frequency (kHz) FIG 1 COMBINED RESPONSE OF CRYSTAL & AMPLIFIER (LF) 40 30 o4 8 20 <E -10 0 0.5 Frequency (MHz) FIG 2 COMBINED RESPONSE OF CRYSTAL & AMPLIFIER (HF) FIG 3 HYBRID CIRCUIT USED IN STRESS WAVE S E N S O R Copyright by ASTM Int'l (all rights reserved). No further reproductions authorized. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. .0 1.

'has too much detection sensitivity been sacrificed?'. . The purpose of a valve for example is to restrict the flow of fluids and in doing so it presents an obstruction to the flow which (subject to the flow conditions) is likely to induce flow anomalies such as turbulence and cavitation. 'over what range of applications can enveloping provide useful data?' etc. If. fast response and global monitoring. Such anomalies give rise to energy loss in the flow and part of this loss takes the form of stress wave activity. Caution must be exercised in extrapolating from these findings. Whilst the signal level can readily provide information on the Copyright by ASTM Int'l (all rights reserved). non-invasive detection. a valve is left open at a fixed setting and the flow rate is remotely altered then the energy loss due to the restriction of the valve or any other flow obstruction increases as the flow rate increases and therefore gives rise to an increasing stress wave level as shown in Fig 6 for steam flowing through an orifice plate. In this general purpose role the stress wave sensor has the advantage (in common with all AE measurements) of sensitivity to a variety of source processes. Experience gained in the application of such sensors to a wide range of industrial processes and plant suggests that these concerns are all too easily overstated. If the valve is progressively closed the increasing restriction to the flow increases the detected signal level as shown in Fig 5 for the case of a water valve (ie decreasing flow rate gives increasing stress wave level). ON STRESS WAVE SENSING 29 Of course the great advantage of the bolt-on housing design is the increased protection of the sensitive piezoelectric element from accidental mechanical damage (eg due to rough handling) and the elimination of possible sensor damage due to overtightening during installation or uneven mounting surfaces. For brevity the examples which will be cited to back up this claim are arbitrarily restricted to those concerning fluid related source processes. cause the activity to be broadband. The transient and localised pressure changes which accompany eddies and bubble collapse. HOLROYD ET AL. If a valve is fully open then it presents the minimum restriction to the flow and therefore gives rise to a low signal level. No further reproductions authorized. APPLICATION OF THE STRESS WAVE SENSOR The resulting sensor is aimed at general use and is intended to be applied alongside. for example. on the other hand. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. When a stress wave sensor is used to monitor the activity from a valve then the detected signal level is dependent upon the intensity of the source processes present. Whilst industry in general has shown a ready acceptance of the resulting stress wave sensors it is to be expected that the immediate reaction of an AE practitioner to the above developments may well include some scepticism: 'is tab mounting acceptable?'. a temperature or load sensor for example. Indeed high coupling forces can be positively encouraged in view of the resulting improvement in coupling efficiency and stability which results. and in a similar manner to.

Another area in need of investigation for an accurate flow rate measurement is the long term consistency of the turbulence levels associated with a valve since the possibilities of surface scoring and scale build up need to be borne in mind. material and geometry) it is not reasonable to expect to obtain a calibrated flow meter simply by bolting a stress wave sensor onto a valve in view of the large range of fluid properties and valve designs. being non-invasive. Clearly if extreme sensitivity to these secondary effects was observed this could be of great practical significance from a maintenance viewpoint in certain applications. . As this happens the engine efficiency reduces. To investigate the effects of this on the generated stress wave signal a stress wave sensor was bolted onto a six cylinder diesel engine which was run at a variety of engine speeds. It is therefore critically important that the pump does not dry run since this can rapidly cause serious damage to the stator. The monitoring of stress wave activity provides a more direct indication of the presence and relative extent of cavitation which can be of particular benefit in system components with variable geometry and. This is illustrated in Fig 7 where a characteristic increase in stress wave activity occurs as the (NPSHa) Nett Positive Suction Head available reduces. following prolonged operation of the engine it is likely that the spring force will begin to reduce at an unpredictable rate. Starting from a well-balanced condition the spring return force of one of the injector valves was deliberately weakened by the removal of a shim such that it opened and closed at approximately 230 psi rather than the normal 270 psi. Using the industry standard procedure for interpreting the NPSHa vs head rise curve for this system suggests that significant cavitation should occur when NPSHa <3. In certain types of pump there is a large swept area of rotor/stator contact and this is lubricated by the fluid being pumped. 30 ACOUSTIC EMISSION presence/absence of flow and the consistency of flow (rate. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Furthermore. The sensor used for this trial had a maximum operating temperature of 70~ and therefore a waveguide was used to keep it within its operating range. The width of these pulses for an engine operating at 2400 rpm is plotted in Fig 8 from which it can be seen that a broader injection pulse is associated with the cylinder which had a weakened spring return force. When spring return valves are used it is necessary to balance the different spring return forces by the use of shims. No further reproductions authorized. If the flow at the inlet to a pump is being restricted then cavitation can be induced by the reduced fluid pressure. The variability of the pulse widths from the other cylinders is probably indicative of the difficulty in balancing such injectors. In a diesel engine it is important to its efficient operation that the fuel injectors of the different cylinders are 'balanced' (ie the injection process occurs at the same point relative to the engine cycle in each cylinder). The complex stress wave signal was analysed with respect to the engine cycle in order to isolate the injection pulse associated with each cylinder. in sterile/hygienic systems. It has been found that a stress wave sensor can provide an instant indication of dry running from the reduction in the pumping process Copyright by ASTM Int'l (all rights reserved).

HOLROYD ET AL. 0 20 40 60 BO 100 Flow R a t e ( c u b i c meb'es per tlour) FIG 5 EFFECT ON SW LEVEL OF CLOSING VALVE 600 -~ 400 200 . T T 1 . No further reproductions authorized.5 3 2 1. ON STRESS WAVE SENSING 31 FIG 4.5 0 T I T 1 . STRESS WAVE S E N S O R (SAME SCALE AS FIG 3) 3. . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 200 4OO 600 800 1000 Flow R q t e ( k g / h r ) FIG 6 EFFECT OF FLOW RATE ON S W LEVEL AT ORIFICE PLATE Copyright by ASTM Int'l (all rights reserved). T T r .5 1 0.5 3 2.

a standard control system. In particular the outputs from the stress wave sensor can be fed via an A to D converter into an appropriate microprocessor for further processing and decision making. Quite apart from the unit cost of the sensor the removal of the need for a dedicated AE instrument and the simplification of system specification and installation increases the affordability and accessibility of the technology. CONCLUSIONS a) The concept of (SWS) Stress Wave Sensing has broad applicability to the monitoring and control of processes and the monitoring of machinery and structural condition. Even in the more complex application areas where the signal from the stress wave sensor needs to be further processed to identify the features of specific interest then there is considerable scope for reduction in hardware costs since the specification with regard to bandwidth and inherent noise level can be relaxed without any knock-on effect. for example. No further reproductions authorized. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. . DISCUSSION The abilities of AE technology and the requirements of industry in general have been reappraised and a new type of general purpose sensor has been created. Alternatively using generally available interface cards and software packages it is possible to input the stress wave sensor output into a PC in order to give the investigative user the maximum flexibility. Copyright by ASTM Int'l (all rights reserved). impact and machining type processes as well as for certain structural stability/integrity applications. The sensors are equally useful for monitoring frictional. In this way stress wave sensing can provide a very versatile approach to both condition monitoring and process monitoring/control. The adoption of an integrated sensor approach greatly simplifies the application of AE technology and therefore provides increased opportunities for its wider use. This approach has been used successfully to allow more sophisticated discrimination and control functions to be performed in an affordable manner and without the need for operator controls. The approach is particularly attractive in those applications where the source processes of interest are dominant since it will then usually be possible to act directly on the output of the sensor using. attention is drawn to the broader application of stress wave sensing than for the purely fluid related processes which have been described for brevity. 32 ACOUSTIC EMISSION noise as dry running occurs and this is illustrated by the plot of the sensor output shown in Fig 9. This finding is of particular relevance to the food and chemical industries where it is important that the fluid being pumped does not become contaminated by damaged stator debris. Finally.

SW Level . .05 0. Flow Rate 15 -.2 ..25 --. 0. 0 ' 0 10 20 30 NPSHa (m) FIG 7 RELATIONSHIP BETWEEN NPSHa & S W LEVEL Injection Pressure Normal m Low KXX~ '(XXX KXX~ . 10 -.5 20 0.5 -. ( X X)< KXX)~ ' ( X X)< 2 KXX)~ <XX)< KXX)~ ' ( X X)~ KXX)< KXX)< KXX)~ <XX)~ ~. HOLROYD ET AL ON STRESS WAVE SENSING 33 1.XX~ KXX~ KXX~ <xxx >(X X)< Kxx~< 0 1 5 3 6 2 4. . No further reproductions authorized. 0 .8 100 ZOO Iime (Seconds) FIG 9 EFFECT OF DRY RUNNING ON SW LEVEL Copyright by ASTM Int'l (all rights reserved). ~E 3 0. 0. Head Rise 1.25 r 1 . . Cylinder FIG 8 SW PULSE WIDTH ASSOCIATED WITH INJECTION PROCESS Onset of Dry Running .15 I / $ / 0. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 1 .75 L. I .

Copyright by ASTM Int'l (all rights reserved).11. . REFERENCES [i] Holroyd. 34 ACOUSTICEMISSION b) The stress wave sensor approach offers A/Z-like capability in an affordable and easy to use package which is appropriate to the industrial environment. c) Stress wave sensor outputs are suitable for further processing in either microprocessor based instrumentation or PC's. "AE from an Industrial Applications Viewpoint". T. Proceedings of IAES 9. London.1. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. T.. "AE as a Simple Sensor". Japan. J. 14-17th November 1988. 26th November 1985 pp. [2] Holroyd. Kobe. 5. No further reproductions authorized. 5. Proceedings of the Fifth National Conference on Condition Monitoring. J..

Storp. and K. The operation of such a sensor to detect the acoustic emissions accompanying an electron beam welding process is demonstrated and their application in a weld-monitoring system is described. Horst A. Germany 35 C o p y r i g h t 9 1991 b y ASTM I n t e r n a t i o n a l ~ee~. H. Bohm are research scientists.. No further reproductions authorized. "Monitoring Electron Beam Welding Process Using Electro- Magnetic Acoustic Transducers (EMAT's). Hermann J. Yamaguchi. Eds. Universitat Dortmund.. electromagnetic-acoustic transducer Dr. ASTM STP 1077." Acoustic Emission: Current Practice and Future Directions. 46 Dortmund 60.astm. A. electromagnetic acoustic transducers is described. process control. In this paper the application of non-contact. Storp. Storp and Mr. process monitoring. J. W. Roget. Weld monitoring applications place special requirements on both the electronic system as well as the sensor used to detect the AE signals.org Copyright by ASTM Int'l (all rights reserved). Professor Crostack is director at Fachgebiet Qualitatskontrolle. and Peter Bohm MONITORING ELECTRON BEAM WELDING PROCESS USING ELECTRO- MAGNETIC ACOUSTIC TRANSDUCERS (EMAT's) REFERENCE: Crostack. American Society for Testing and Materials. and Bohm. For this to be successful requires that the signals emitted during the welding process be detected and that a good correlation can be established between the welding parameters and the acoustic signals. . KEYWORDS: electron beam welding. H. 1991. Philadelphia.. P.. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Sachse. These requirements include a good directional selectivity and a capability for operation at high- temperatures which limits the use of piezoelectric sensors. J. ABSTRACT: This paper describes the use of acoustic emission analysis (AEA) to monitor and control an electron beam welding process. Crostack.

Improvements can therefore only be made afterwards. 36 ACOUSTICEMISSION INTRODUCTION The non-destructive investigation of welds imposes particularly high re- quirements on the test methods. Fig. according to size and position. but also very high interference noise may occur. i. Burst pulses can then occasionally be observed for a long time. continuous acoustic emission occurs together with high amplitudes which become a bundle of smaller ampli- tudes on turning off the welding current. During welding. the acoustic emission analysis offers the advantage. three different types of noise occur. .e. the acoustic emission of heating and cooling procedures are overlaid. 1. This is demonstrated by the following example of electron beam welding. Traditional non-destructive methods often detect flaws in welded seams only with difficulty. Furthermore.3/. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.BEAM (EB) WELDING PROCESS AND DATA OF ACOUSTIC-EMISSION ANALYSIS During EB welding. all methods have in common that welding flaws are generally checked only on the finish-welded component. Copyright by ASTM Int'l (all rights reserved). during welding. tests during welding are not yet common. Additional measurements are therefore necessary to suppress interference and separate different sound sources and these measures must be adapted according to the welding process/1. During welding. so that welding parameters can therefore be corrected at once. noise occurs both during welding and during the cooling phase. No further reproductions authorized. When the sound is registered parallel to the welding current. that flaws can be registered directly as they occur.2. In addition. in many welding methods. difficulties arise in the use of the acoustic emission analysis because not only various useful signals. FUNDAMENTAL CORRELATION BETWEEN ELECTRON . Here.

. ON ELECTRON BEAM W E L D I N G P R O C E S S 37 ~ J ] we(ding current time t Fig. 1 9 Acoustic emission from EB welding weldimG e(Isu) I~ per f o r m a n ~ ~. .. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. ii | .. . C R O S T A C K ET AL. No further reproductions authorized. 2 9 Schematic diagram of acoustic emission analysis during EB welding Copyright by ASTM Int'l (all rights reserved). == o == =_ == | | I pre a m p l i f i e r = n ~ 10"1Po E I ~ lO3mb~r) beam current lense current velocity penetration current Fig.

Thepenetrationcurrentrepresents a measure for the number of electrons which come out on the rear side of the component with opened seam channel and which can thus directly be correlated to the welding process. 3 : Influence of welding process on AEA Measurements on a 20 mm thick specimen (material 10 CrMo 9 10) show good means of identification of welding current. No further reproductions authorized. 38 ACOUSTIC EMISSION Whereas the correlations between the characteristic features of the acoustic signals and the sound sources during the cooling phase of the weld seams are satisfactorily clarified. Copyright by ASTM Int'l (all rights reserved). lens current. Fig. . 2 was set up. 3. the experiment shown in Fig. the evaluation of the signals emitted by the process is far more complex/4/. To investigate this testing problem. Fig. melting bath motion and acoustic signals. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The welding parameters beam current. drive speed and the pen- etrationcurrentcanbe monitored.

The large scattering range of the AE-signals is due to the integrally measuring acoustic transducer which also receives signals from the cooling seams in addition to the process noise. amplitude detector were used for the identification of the received signals. an increased location selectivity of the acoustic trans- ducer with regard to the interesting seam regions should result in a considerable increase in AE sensitivity for a separate monitoring of the welding process and the cooling weld seam. CROSTACK ET AL. The amplitudes of the RMSvalue. The welding current was temporarily increased in addition to the simulation of flaws. The open seam channel presses the sound-active slag out of the passage. An RaMSvoltmeter. . which can be seen from the corresponding curve course of the acoustic signals and/or characteristic break in the pulse-sum curve during the acoustic signal analysis. Copyright by ASTM Int'l (all rights reserved). a pulse-rate counter and a peak. This is a particularly interesting approach for possible process control. The sudden wide opening of the seam passage leads to the corresponding drop in sound intensity and an increase in amplitudes of the penetration current. After the pulse-like increase in intensity. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. ON ELECTRON BEAM WELDING PROCESS 39 The acoustic signals were acquired using various high-temperature resistant piezoeeramies at centre frequencies between 140 and 600 kHz were used. the peak amplitude and also the pulse-rate amplitude accordingly drop. the RMS meter shows the expected temporary increase in amplitudes. a pulse-sum counter. the acoustic emission hence allows process monitoring even with closed steam passage. No further reproductions authorized. thus enabling an early access to the welding process. the welding current was taken down too strongly. When the beam current is taken down at completion of the welding process. Thus all four analysis methods used show an increase in sound activity up to maximum at the beginning of the welding process with an increasing opening of the seam channel. In order to exclude the scattering influence on the welding process control. The signals are digitally filtered and then plotted on a multi-channel recorder and evaluated. Compared to the registration of the penetration current. and/or the gradient of the pulse- sum curve is reduced.

Depending on the geometry of the coil (in this case a meander-shaped coil) and the direction of the magnetic field lines (normal or parallel to the surface) the transducer may be made to favor the detection of shear. Fig. An alternating current voltage is thus induced in the coil whose am- plitudes can be correlated directly to the height of the particle deflection in the component. XS .. 40 ACOUSTIC EMISSION OPTIMIZATION OF THE AEA BY USE OF ELECTRODYNAMIC TRANSDUCERS This problem can be resolved by a face-to-face arrangement of acoustic emission transducer and electron beam.~~ ii/i _ . Further increase in the direction selec- tivity requires the application of the transducer in the near vicinity of the electron beam. No further reproductions authorized.0 C = f = wave v e l o c i t y frequency i @ rv ~ L ~ O = ARCSIN CL~-Q 9 . 4 : Geometry of a meander coil for sending and detecting different kinds of waves by an electromagnetic sensor Copyright by ASTM Int'l (all rights reserved). The electrodynamie transducer consists of a coil and a superimposed electric or permanent magnet. The experimental problem thus requires contact-free detection of the acoustic emission signals. ~ ~ . f Fig.4. ~-~x s L I=current Xs 0 = = transducer period Rayleigh wave L= l o n g i t u d i n a l wave | 0 | @ ~ = smear Nave = angle o f incidence ~z~~ . longitudinal or Rayleigh- waves. whereby the transducer is subjected to high temperatures during the welding process. . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. At present the only poss~le solution is the use of eleetrodynamic and/or permanent dynamic transducers. Its function as receiving transducer is that an ultrasonic wave within the region of the coil leads to a modulation of the static magnet field according to the wave frequency.

CROSTACK ET AL. After fixing the transducer period.5- lq I-1 (a) (b) Fig. In addition. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized. To increase the lateral resolution. the test frequency f. 5 9 Comparison of the directional characteristics of focus and conventional sensor increased directional selectivity of the focus transducer in comparison with a conventional transducer (5 (a)). an acoustic source was moved transverse to the focus point. The above mentioned Copyright by ASTM Int'l (all rights reserved). .sensor F piezoc eramic I01 ioni. the determination of the test frequency at a narrow- band range result in a particularly clear directional sensitivity of the transducer in the axial direction for each kind of wave mode. the wave velocity and the soundfield characteristic features is of particular importance for the electrodynamie transducer prepared with a meander-shaped coil.iiioiill nSor transducer 0. ON ELECTRON BEAM WELDING PROCESS 41 The correlation between the transducer period s. Fig. 5 (a). the sides of the meander can be addi- tionally arranged in a circular segment around a chosen focus point.x 1. the significantly higher absolut sensitivity of the focus sensor becomes evident. To show the influence of the coil geometry on its lateral resolution.0- ' ~ t ~ focus . The detection of the sound source either by a focus or the conventional sensor and plotting the coil voltages over the path of the sound source show a strong UI Um.

7. A piezoelectric transducer was additionally arranged on the front side of the specimen. Measurements up to now thus show that the direction selectivity of an electrodynamic transducer can be directly controlled in axial and radial direction. 42 ACOUSTICEMISSION method thus contributes to an increase in sensitivity of the electrodynamic transducers. Fig. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The permanent magnet dynamic transducer was installed on the specimen surface and its design was to receive ultrasound waves from the surface-near zone for its response frequency of 500 kHz. 6. For simplified illustration. eom j perm/anentmagnet/ / I plezocerarnlc V Fig. No further reproductions authorized. the signals received were digitally filtered and plotted on the recorder. After completion of the optimization of the transducer system. The schematic diagram of the experimental set-up is illustrated in Fig. initial measurements during welding are performed in practice. 6: Position of piezoceramic and electromagnetic sensor for monitoring welding process Copyright by ASTM Int'l (all rights reserved). .

. Therefore AEAis an interesting method for monitoring and regulating welding process. CROSTACK ET AL. this sensor has the following advantages: Copyright by ASTM Int'l (all rights reserved). More over there are clear advantages over conventional transducers owing to the transducer-specific location selectiv- ity. . . . . ~ ~ ^ ~ . . No further reproductions authorized. . . 7 : Acoustic emission amplitudes detected by EMAT and piezoceramic sensor CONCLUSIONS The measurements show that the acoustic emission signals can directly be correlated with weld beam position in material. ON ELECTRON BEAM WELDING PROCESS 43 The comparison of both curves shows that the received maximum signal height of the permanent dynamic transducer was reached when the welding beam was in the region of the focusing point of the transducer. . "~. The measured results hence show that the employment ofelectrodynamic transducers allow the detec- tion of acoustic signals from the weld seam. .. U(piezo)~ /~ 6rnV start end Fig. . One weak point in the pro- cess regulating line is the piezoelectric acoustic transducer which generally dispose of broad soundfield edges so that the direction of an acoustic event can not be determined accurately. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Compared to the piezoelectric transducer. A a)f . U(EMAT)~ 160pV IA A A . This problem can be solved by introducing permanent dynamic transducers into the process.

and Storp. Steffens. H. J~ptner. a Powerful NDT Tool. H. electromagnetic acoustic transducers. H. . DGM Informationsgesellschaft. Oberusel. H. the broad directivity exhibited by conventional. Intern. Our measurements with non-contacting electromagnetic acoustic transducers to detect AE signals during electron beam welding show that the signals can be directly correlated with the position of the weld beam in a test piece. Eisenbl~tter..-D. A suitable transducer geometry results in improvements in locating sources of emission both in the axial and radial directions. Anwendung der Schallemissionsanalyse zur 0berwachung des ElektronenstrahlschweiBprozesses. [4] Steffens." edited by J. 1980. H .-J.. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. [2] Crostack. and Crostack. W. W. D. . Hence the application of electromagnetic transducers as AE sensors is restricted to applications in which the generated acoustic signals are of sufficiently high amplitude. and Morlo. H. The non-contact detection of acoustic signals permits their use at high temperatures.. such transducers have the following advantages: .. Strahltechnik VII. 1988. permanent magnet.-D.: "Acoustic Emission Monitoring during Welding. piezoelectric acoustic transducers results in poor angular resolution which complicates the location procedures. H. Tokyo. It must however be pointed out that the absolute sensitivity of the electrodynamic transducer is approximately 30 dB lower than a piezoceramic transducer. 1975.-A. : Untersuchungen zur Anwendung der Schallemissionsanalyse fur die zerstorungsfreie Pr0fung yon AbbrennstumpfschweiBverbindungen.-A.. When compared to conventional piezoelectric transducers. 44 ACOUSTICEMISSION CONCLUSIONS AE measurements can be used to monitor and control a number of welding processes. Heft i0. When sources of emission are to be located during the weld process for control applications. . [3] Prine.: "Acoustic Emission Analysis-An Integral Method for Process Control during Welding" in "Acoustic Emission. Kern. This problem can be overcome by using appropriately designed. This sensitivity is further reduced if the gap between the coil and the test piece is increased.. " SchhweSBen und Schne~den 40 (1988).~ . R E F E R E N C E S [i] Crostack~ H." 5th. No further reproductions authorized.. Copyright by ASTM Int'l (all rights reserved). DVS- Berichte Band 34. Acoust$~ Emission SvmD.

. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized. AE Sources and Wave Propagation Copyright by ASTM Int'l (all rights reserved).

" Acoustic Emission~ Current Practice and Future Directions. We developed the advanced analysis system to evaluate AE signals quantitatively. The University of Tokyo. . our method of AE source characterization is described. Then the size. ceramics and composites. KEYWORDS: acoustic emission. Roget. 4-6-1 Komaba. ABSTRACT: The acoustic emission (AE) source characterization has been developed to understand the dynamic process of microfracture in metals. The transfer function of the measuring system is calibrated by the breaking pencil lead method. and Kishl. J. Finally. 6STM STP 1077. microfracture. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Moment tensor components are determined by the developed deconvolution algorithm.. AE signals can be represented as the c o n v o l u t i o n integral of the source f u n c t i o n due to microcracking. Source location of each AE is determined from the signals recorded using six transducers. 47 Copyright 9 1991 by ASTM International www. Secondly. the future aspects in AE source characterization will be discussed.astm. This analysis system was applied to the fracture toughness tests in v a r i o u s materials. orientation and nucleation time of microcracks are obtained from the moment tensor. Kishi is a professor of Research Center for Advanced Science and Technology. Enoki is a research associate and Dr. deconvolution. 1991. micromechanics. This paper is concerned with the theoretical background of AE source characterization.. M. "Development and Future Aspects in AE Source Characterization. Each dynamic Green's function of the compact tension specimen concerning each source location is calculated by a finite difference method. microcrack. Philadelphia. source characterization. Sachse. the dynamic Green's function of the media and the transfer function of the measuring system. An infinitesimal deformation in a material can be represented as a moment tensor by the 'eigenstraln method' of micromechanics. American Society for Testing and Materials. No further reproductions authorized. W. Meguro-ku. Eds. quantitative nondestructive evaluation. Dr. inverse problem. Tokyo 153. and K. T. dynamic Green's function. Manabu Enoki and Teruo Kishl DEVELOPMENT AND FUTURE ASPECTS IN AE SOURCE CHARACTERIZATION REFERENCE: Enoki. Yamaguchi. In this treatment a microcracklng can be modeled as a dislocation source with moment tensor components. fracture mode.org Copyright by ASTM Int'l (all rights reserved). Japan. The development of AE source characterization can be summarized firstly.

and then by carrying out the single deconvolution method in the time domain. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. also independently applied the simple deconvolution method and characterized intergranular microcracks in Ni-Cr-Mo steel by using the response function. and the Yobell specimen. Ohno & Kurlbayashi[4] 1 experimental o 1981 Wadley. which includes both the transfer function of the measuring system and the Green's function of specimen and can be experimentally calibrated by a breaking pencil lead. However.. as mentioned below.The history of AE source characterization. Some studies have attempted to characterize AE sources quantitatively on the analogy of seismology[2].mmarize the development of AE source characterization firstly. In this paper we s. We h a v e to e s t a b l i s h the relationship between microcracking and eigenstrain (or deformatlon moment tensor). this equation can be reduced to a simple linear c o n v o l u t i o n equation. Kishi et al. [4]. w h i c h can m e a s u r e the displacement of the surface. on the other hand. Table 1 -. Stacey & Baldwin[6] 6 half-space - 1986 Klm & Sachse[8] 6 plate A ~985 Ohtsu ~ h~-sDace o:perfectly carried out. [3] determined the volume and generation velocity of the cleavage and intergranular microcracks in a mild steel and electrolytic iron by u s i n g a c a p a c i t a n c e transducer. : not carried out Copyright by ASTM Int'l (all rights reserved). Year Author(s) Channel Green function Deconvolution 1981 Kishi. Sinzle ~ 9 o D v o l u t i o n Method In order to d e t e r m i n e the d e f o r m a t i o n moment tensor and characterize the AE sources. No further reproductions authorized. Wadley et al. and described as follows. A:partially carried out. Also we have to develop the method to obtain the moment tensor. if the mode of microcracking is the tensile type. Secondly we mention the theory of AE. the multiple deconvolution must be carried out in multiple convolution equation by using the recorded ~ E waveform with more than six channels. 48 ACOUSTICEMISSION METHODS OF AE SOURCE CHARACTERIZATION In the fields of micromechanics and seismology the deformations such as microcracks have been formulated analytically. Scruby & Shrimpton[3] 1 half-space o 1985 Ohira & Pao[5] 6 plate A 1986 Enokl & Kishi[9] 6 finite-body o 1986 Scruby. Finally some examples of analysis and the future aspects will be discussed. Their method could characterize the AE sources by general transducers and specimens. Those deformations in materials can be generally represented as nonelastic 'eigenstrain' in m i c r o m e c h a n i c s [ l ] . to which the theoretical Green's function can be applied. The development of AE source characterization can be summarized as shown in Table I. and the experimental and analysis system. . and then only the size of microcracklng is the unknown parameter and measuring with one channel can determine the size and generation velocity of microcracking. which was described in detail in [9]. An acoustic emission (AE) technique has been used as an almost unique method to detect dynamic deformation and fracture of materials with high sensitivity.

t). however the absolute value of the moment tensor and time function is not determined by his method. t) is small compared with distance between the position 9 and x'. t) denote an eigenstraln tensor in V. He determined the ratio of the moment tensor. and then both deconvolutlon algorithms in time domain and in frequency domain were developed and all the moment tensor components were determined. Finally. 9] Copyright by ASTM Int'l (all rights reserved). THEORY OF ACOUSTIC EMISSION D e f o r m a t i o n Moment Tenso~ Let V denote an elastic domain occupied by a given three dimensional body and e*mn(X . They determined the radiation pattern of the moment tensor from the peak amplitude of the first longitudinal wave under the assumption that longitudinal wave amplitude is proportional to the surface displacement. No further reproductions authorized. 12]. Firstly. . the source location of each acoustic emission was determined from the signals recorded by six multl-transducers. Kim et al. If the dimension of region occupied by eigenstrain e*mn(X. Secondly. They determined the moment tensor in 7010 A1 alloy by comparing the pulse strength and this far-field term. they used the far-field term of the Green' function in a semifinite medium. Ohlra et al. However. by comparing with the ratio which is calculated from the Green's function of an infinite plate. they used the far-field term of the longitudinal wave of the theoretical Green's function in an infinite medium under the consideration of the reflection at the surface. each dynamic Green function of the compact tension specimen concerning each source location was calculated by a finite difference method [i0]. [5] determined the moment tensor in A533B steel from the ratio of amplitudes between longitudinal and transverse waves. [8] determined the strength and time function of dipoles due to a thermal crack in glass by comparing the epicentral normal response and the Green's function in an infinite plate and by using the single deconvolution method. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. various simplified methods have been proposed in this problem to avoid solving the convolution equation directly. They obtained the time function of the moment tensor by using the single deconvolution method under the assumption that the components of the moment tensor have the same time function. As the Green's function of media. the transfer function of the measuring system was calibrated by a pencil breaking lead [Ii. Ohtsu [7] used only the first peak amplitude and simplified the Scruby's method in the result. that is. Scruby et al. ENOKI AND KISHI ON SOURCE CHARACTERIZATION 49 Simplified Amplitude Metho4 The deformation moment tensor has to be determined to obtain the mode and orientation of microcracking. and the shortest wavelength due to e*mn(X . then the point source approximation can be applied. Then we can state the displacement field due to e*mn(X. [6] carried out the analysis by using the strength of the first arrival pulse which is defined as multiplication of the first peak amplitude and time. t) as [i. Multiple Deconvolution MetHod The authors [9] determined the moment tensor in A470 steel by solving the convolution equation directly.

t) due to the discontinuity of displacement can be represented ~s ij(x. x. t).Dii . t)] vj(x) + [uj(x. t)] . 6i~ the volume of deformation domain. w h i c h is r e p r e s e n t e d by mode and intensity of an infinitesimal deformation. Let us consider the eigenvalue problem of Dik because Dik is s symmetry tensor of a second order. t)] vi(x)} 6(A . No further reproductions authorized. (A ~*mm6jk + 2p e*jk ) dr. and p . (2) where Cjkmn . . vi(x) . Djk(X. x.the normal to surface A. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. for an isotropic material.Jl D2 + J2 D .(3~ + 2~) ~*ii dV. and the comma indicates a differentiation and * means a convolution integral with respect to time. t) is a symmetry tensor of a second order. J3 . t) -Gij. We can calcurate three in~ariants Jl' J2 and J3 respectively Jl .2~2(~*ij~*ij)](dV) 2. Copyright by ASTM Int'l (all rights reserved). 50 ACOUSTICEMISSION ui(x'. t).Cjkmn e mn(X. is defined as. The deformation moment tensor Djk(X .x).the Lam~'s constants.k(X'.the displacement field in the direction x i at position x' at the time t due to an impulsive force in the direction xj at x at time 0. t) .(5) where [ui(x. Eigenstrain E i~(x. t) . D3 .(i/3)[(3A3 + 6A2~ + 6A~2 + 4~3)(~*ii )3 -(6A~ 2 + 12~3)~*kk(~*ijc ij) ** 3 +8~3~*ij~ jk ~ ki](dV) (3) Eigenvalues D are three solutions of the following equation. J2 . Equation 2 shows that independent six quantities of Djk are represented by seven quantities of ~ mn and dV. The deformation moment tensor D~k(X . t) . (4) M%r Model It is well known that the generation of microcrack can be modelec as the discontinuity of displacement on the faulting surface like dislocation[l].[(3A2 + 4A~ + 2#2)(~*ii )2 . (I) where Gij(x'. t) dV .J3 .O. which is the quantity to represent an infinitesimal deformation.(l/2){[ui(x . Kronecker's delta.the discontinuity of displacement on a surface A. t) * Djk(X.the elastic stiffness tensor.

xi(2) .[Us] v m / ([Uk] [Uk])I/2. and with direction x (I) .~ijk [uj] Vk. Substituting Eq 5 into Eq 3.[u] vi.two dimensional delta function on surface A. The deformation moment tensor is. which is called a vector dipole which has the same direction of arm and force. . x (2) and x ( 3 ) respectively.[ui] . the volume AV of microcrack and the angle 0 between [ul] and v i can be represented as respectively AV . we can calculate three invarlants Jr. x (2) is the vector in the direction of normal to the plane given by [u] and v.[Um] v m AA . Three elgenvectors x (I) . where [u]-([Us][Us])I/2. This Djk(X . x (2) and x (3) are respectively represented as xi(1) .[ui] + [u] vi. J2 and J3' and obtain three eigenvalues D (I) . t) is called a seismic moment tensor in seismology[2]. D (2) and D (3) of Djk from Eq 4 as D (I) . (7) D (3) . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. (ii) where Copyright by ASTM Int'l (all rights reserved).the area of surface A. and x (3) which is on the plane given [u] and v is normal to x (I) . (6) where ~A .v 2) / 3E. from Eq 2. This result shows that x (I) is the vector in the direction of isometric angle between [u] and v. ENOKI AND KISHI ON SOURCE CHARACTERIZATION 51 6(A . No further reproductions authorized. D (2) and D (). t) .((A + p) [Us] v m + p [u])) AA.x) .p [u])) AA.the permutation tensor. Using the trace of moment tensor. (9) cos8 . D (2) .((l + p) [US] v m . (i0) Assuming that a microcrack is a dlsc-llke crack subject to a normally applied stress a. Djk(X. Consequently the generation of displacement discontinuity [u] on the surface A with the normal vis equivalent to the generation of moment with magnitude D (I) . we can state the crack volume as[13] ~V .16 a 3 a (i .{A [Us] v m 6jk + # ([uj] v k + [Uk] vj)} AA. (8) xi(3) .Dil / (3A + 2~). where ~ i j k .I [Um] v m AA.

the nucleation time of microcracking. t) can be expressed as[4]. i.cted Signal V ~t Transd Transducer ~~Gij. (14) EXPERIMENTAL SYSTEM OF AE SOURCE CHARACTERIZATION Figure 2 shows the block diagram of the multichannel AE detection and recording system. v . The crack radius a is. from Eq i. (12) The nucleation velocity of microcracking v cane be defined as v . t) due to moment tensor Djk(X .{3 (i .the crack radius. No further reproductions authorized. t).v 2) o} 1/3 .a / At.~Detect / . 52 ACOUSTICEMISSION a .k(X'. t) * Gij.2v) Dii / 16 (I .the Young's modulus. from Eqs 9 and II. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.k' D(t) [u] ~ < ~ SourceFunction A Djk Microcrack . V(x'. Denoting the response function in the i-direction at the position x on the surface by Si(x. E . (13) where At . . t) = Si(x'. which is shown in Fig. t). Acoustic Emission Signals The detected signals of acoustic emission are generally different from the displacement field because of the response function of the measuring system.Poisson's ratio. where Gij. a .t z~t FIG. the detected signals V(x'. v(t) AE Signal Analysis 5i <: ~.k is a dynamic Green's Function and S i is a response function of measuring system. Copyright by ASTM Int'l (all rights reserved). x. i --The relation between source function D= k and detected signal V in acoustic emission signal processing. t) * Djk(X.

resolution of i0 bits and the record length of 2048 points. 12]. ENOKI AND KISHI ON SOURCE CHARACTERIZATION 53 Trans4uce~$ The transducer P50 (PAC) has a piezoelectric element of about i mm and a broad-band response up to 2 MHz. 6ch NF9913 AE9620 HP350 Computer) ~2MHz Low NoiseType 50ns.a[ 3Mwords FIG. 2 -. The transducers and measuring system are calibrated using a breaking pencil lead[ll. DC-20MHz. Wave Memory And then digitized and stored by the wave memory AE9620 (NF) with the maximum sampling rate of 20 MHz. Computer Data from the wave memory are transferred via a GP-IB interface and stored on magnetic disc in the model 350 computer (HP). The output from the transducer P50 (PAC) is fed into the AE analyzer AE9600 and the conventional AE parameters are stored on magnetic disc in a model 216 computer (HP) with the external parameters of load and crack opening displacement (COD). . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.6ch 80MDisc DC~20MHz Length 2kwords Plotter 20dBor 40dB Tot. preamplifiler The output from each transducer is amplified 20 or 40 dB by the low noise type preamplifier 9913 (NF) with a frequency response. Each dynamic Green's function of the ( AE Processor) HP216 I IDis t M682H :~ :IAE3201 IHGrd I (Transducer) (Preamplifier) (WaveMemory) (Computer) (Host Pico. Copyright by ASTM Int'l (all rights reserved). The AE a n a l y z e r A E 9 6 0 0 (NF) is also u s e d to m e a s u r e the conventional AE parameters of events and peak amplitude.The block diagram of the multichannel acoustic emission detection and recording system. No further reproductions authorized.

Fracture Toughness Tests Fracture toughness tests are carried out at a constant crosshead speed and AE signals are recorded during testing. No further reproductions authorized. where P is the total number of channels. . Copyright by ASTM Int'l (all rights reserved). Accurate source location is required in order tc understand fracture processes.The developed analysis system for AE source characterization. are transferred via a RS-232C interface and stored on magnetic disc in the model 350 computer (HP). 54 ACOUSTIC EMISSION compact tension specimen Iconcerning each source location which is calculated by a finite difference method. Source location of each acoustic emission is determined from the waveform recorded using six transducers.on Of Transducer ! > iDjk= WSi/Gij. k Moment Tensor dbeto Microcracking Ojk Location Size Orientation Fracture Mode Generation Time FIG. The location of each source event is determined by measuring the differences in the wave arrival time between two transducers[14]. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. k ~--IQf Media by FDM Si I // Gij. Suppose that Atil is the difference in the wave arrival time between i-th and j-th tran6ducers.ution ||Green Funct. 3 -. Let r ~ denote the transducer positions (i ~ i ~ P) and r denote the location of the source. We can represent the general equation for source location as Waveform by AE WaveCom I Location Pencil Lead due to i " I~ ' Breaking Microcracking qMicr ocracking aesponseFunctionI IOeconvo. ANALYSIS SYSTEM OF AE SOURCE CHARACTERIZATION Source Location Figure 3 shows the d e v e l o p e d analysis s y s t e m for AE source characterization.

No further reproductions authorized. Ir . for an isotropic materials. . r i) . experimental[4] or numerical researches on Green's function of media have been investigated. (15) where re(r. The e q u a t i o n s of wave m o t i o n in an elastic.Ir -rJl / re(r.Ir -ril / ve(r. Copyright by ASTM Int'l (all rights reserved). then a nonlinear least-square method can be used to solve the Eq 16 for source location r. ENOKI AND KISHI ON SOURCE CHARACTERIZATION 55 Atij . (17) Ch3 o ICh5 Ch9 4-o' x v.ril .Geometry of compact tension specimen and configuration of six P50 AE transducers.rPI. FIG. I S i ~ P-l. We simulate the dynamic Green's function of compact tension specimen by a finite difference method[lO].#v2u + (~+#)v(v'u). the Green's function which has the complex boundary conditions can be obtained only by the numerical simulation method.. Green's Funct%pD of Med~a The analytical[15]. isotropic and homogeneous medium in vector form are: pa2u/at 2 . However. (16) where . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Figure 4 shows the typical positions of six P50 transducers which are attached to the compact tension specimen.the longitudinal velocity of material. 4 -.x 'x . If P ~ 4. rJ).Ir . r i) . This equation can be reduced as. = Atip .the energy velocity of ri-r direction.

A model of ITCT specimen. z + A(Ux.x) . The boundary conditions are that all tractions on the free surfaces are zero. . . we get the following formulations as boundary conditions: azz . .the mass density of the material. 5. No further reproductions authorized. The initial conditions are U . 0 for tS0. (18) The boundary conditions for the other free surfaces and crack surfaces can be formulated in the same way. The free surfaces are shown by solid lines in Fig. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. . Copyright by ASTM Int'l (all rights reserved). for example. 6. .0. where Djk(X ) is the moment resulting fro~ the product of the force in the j-direction with the arm in the k-directlon. The length of a crack is represented by AA'. azy ~(Uz. . I I i s ! "~ I I I 0 Y FIG. 5. Here. . 5 -. Z LJJJJJJJJJJJ ! !' I st %% I l I ~I j l " ' T .y) R 0. (19) Let us consider the input function for moment tensors. On the upper surface shown by the shaded portion in Fig. . azx ~(Ux. 4.. In order to obtain the Green's function for a CT specimen. D~k(x)r(t). p . The pinholes are disregarded since the effect of a reflection wave from them on the AE source characterization is minimal for the sensor locations used in Fig. 8U/St . and r(t) is a nondimenslonal time function with some rise time. . . .z ) .0. at an inner point x. z + Uz.y + Uy. a force is applied at each of the two points x k+ and xk'-.(A+2~)Uz. x + Uy. the specimen is modeled by a rectangular parallelepiped with a macrocrack as shown in Fig. 56 ACOUSTICEMISSION where U R the displacements.

Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 6 -. The e x a m p l e of the response function Gi~ k at the position of transducers which are simulated by a finite d{~ference method. j. . j. 7 -. i= \ II {D :K In 12 ed 13 21 22 -& M 23 31 32 33 8 Time / ~s 5 FIG. Copyright by ASTM Int'l (all rights reserved).KE+I) 0 " 9 " 9 9 9 9 9 1 / I o (IE.KE) 9 0 o o ~ X "d 9 a macro _ crack Z= (KC+I k l o o o 9 9 9 Z=KCal o 9 9 X X 9 o 9 9 9 9 0 0 0 0 0 0 0 'X Finite difference mesh (y=j&l plane) FIG.j.~E+I) 0 0 0 0 0 ~ ([E+I.F i n i t e difference mesh around the free surfaces and crack surfaces.j.KE+I) (IE.KE) (IE+I. ENOKI AND KISHI ON SOURCE CHARACTERIZATION 57 (i. No further reproductions authorized.j.

JE. (20) Nine different forms of the moment tensor may exist. the following equations result: p@2U. J. At-O. PAl) at time pat.05 ~s and rise time tr-0. p02~j(X(x~. y and z-directions in Eq 17. k). where the sign of force is defined to be positive if it acts along the positive direction of the axis.)/0tz)/0t ~ V Z ~ ( x K-) + (A+~)UI'J(A+~)5. By considering the equafion of motion containing the body forces at two points. #-7. The Green's functions of a one inch compact tension (ITCT) type specimen shown in Fig. and p -7. and take an increment At in time. 4 are simulated using the p~e~ent FDM. JAI.p) represents the approximate components of displacement at a grid point (iAl.j. Thus. y and z- directions.67x10 Pa. In the application of the acoustic emission method. the interest of this paper is to obtain the Green's function of finite medium which includes a macrocrack. In order to obtain the Green's functions. It is well known that the discretized formulation generally converges to the exact solution of Eq 17. The stability condition of these equations is L . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The response functions at the six points of transducers are simulated. For simplicity we denote the point by (i. 7. Here i. . A-l.2 k+) + (xk+) + F'r(tl/Al~. The ~ t e r l a l is steel with3 the ~aterlal constants.#V^U4(x. respectively. ResDonse Function of Measur~n~ System The transducers and measuring s~y~tem are calibrated using a breaking pencil lead.j(xK'). where the input point is one of the experimentally localized points. x k+ and x k-. U(i. k and p are integers 0<i<IE.k.5 ~s are used. the nine couples of moment are applied to the three-dimensional localized points. as A1 is decreased while keeping the value of L constant. detection and evaluation of microcracks nucleated at the m a c r o c r a c k tip are essential. Before the fracture toughness testing. the values AI-0. No further reproductions authorized. 0~k_<KE. Copyright by ASTM Int'l (all rights reserved). and their directions are opposite.F'r(t)/AI 3. An example of the results at a transducer point on the surface is given in Fig. Then the response function of the measuring system is obtained from the result of the deconvolution with the known source function of the pencil breaking lead [12] and the above Green's function.At/AI ~ (~2 + 282)-1/2. The Green's function corresponding to the braking pencil lead is calculated by a finite difference method[10].07x10 i~ Pa. With consideration given to a reasonable amount of computer memory utilization. 58 ACOUSTICEMISSION where the size of each force F is Dik(X)/Al.87xi0 kg/m . (21) where pZ_~lp. Let us take the same increments AI in the x. the waveform data due to the breaking pencil lead at the surface of specimen are recorded.5 mm. J. k+ 2 . where IE. and KE represent the total numbers of the mesh points in x. 0<_J_<JE.

I. n-i . Eq 22 is written as J ui(t) .means Laplace transform. The c~ coeffiQients and T0(n ) are set to nonzero initial values. ENOKI AND KISHI ON SOURCE CHARACTERIZATION 59 Deconvolutlon in Time Domain Equation 14. .24.the waveform data due to microcracking. Using inverse Laplace transform. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The more improved c~. E E c J" giJ(n" . 2. 4.. ~ is calculated. given ui(t) and giJ(t) over a finite time interval[16]. we can find a discrete convolution integral equation in frequency domain as [17] J i-i . (24) j-1 k-i We develop the iterative algorithm to solve this nonlinear equations as follows: I. Solving the linear Eq 25. N. . (22) j-I where ~ ( t ) . i-I . 3. .. a moment tensor d~ is obtained Copyright by ASTM Int'l (all rights reserved). . Using these c6. (23) j-i The problem of sou!ce characterization is then to simultaneously determine T(t) and c J (I ~ j ~ J). Let N be the v@~ue of n which corresponds to the maximum time for which ui(t) and glJ(t) are specified. These procedur~ of alternately calculating cA and TM(n ) can be continued until the c~ and TM(n ) converge to fixed values. the Tl(n ) are calculated from Eq.k + I) T(k) At. J. . g~-J(t) the transfer function which is convoluted by the transfer function of measuring system and Green's function of media. 5. (25) j-I where . 3. can be written as J ui(t) . No further reproductions authorized. .E cJ giJ(t) * T(t). . i ~ n ~ N. These Tl(n ) are used to calculate the improved c~ from Eq 24. . Both ui(t) and giJ(t) are converted into ~ a n d ~ respectively. . . Deconvolution in Frequency Domain Applying Laplace transform to this equation. The algorithm to obtain moment tensor components is developed as follows: i. J-I .the unknown moment tensor. Assuming that the components of the microcracking moment tensor have the same time function T(t). I. 2.Z giJ(t) * dJ(t).. The LaplaGe transform is carried out. The convolution integral in Eq 23 reduces to a summation as J N ui(n) . dJ(t) . Tl(n ) are obtained by averaging. which is a convolution integral equation in time domain.

positions of transducers and dimensions of transducers. . Copyright by ASTM Int'l (all rights reserved). 14 later. As both transforms are performed by using the algorithm of the fast Fourier transform (FFT). No further reproductions authorized. the calculation time of deconvolution can be shorter than that of the algorithm in time domain. AE events can be classified two groups. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Figure 9 shows that source events under the low applied load are generated at the back of the fatigue precrack tip.85 mm and the notch of 0. In connection with the Laplace transform the time function has to be multiplied with e -St. while those under the high applied load are generated in front. so the discrete convolution error can be neglected if one chooses a suitable value for 6. 60 ACOUSTICEMISSION from H~.i mm width. Figure 9 shows the location data of AI-Li alloy in fracture toughness test with ITCT specimen of TL direction. The results were in good agreement with the observation and the location by AE.The results of location of AE sources detected during fracture toughness test in the SiC glass composite. It can be concluded that the former is due to cleavage at the precrack tip and the latter is due to separation crack in front by the observation of scanning electron microscope. The errors of source location are given by a sampling rate. 8 -. which is clearly shown in Fig. A sampling rate of 20 MHz and a FIG. RESULTS AND DISCUSSION S o u r ~ Figure 8 shows both the results of optical observation and the location results by AE of the SiC fiber reinforced glass composite using a half size of a standard i inch thickness compact tension specimen with 2.

Although the size of the piezoelectric element is about 1 mm. respectively. Eq 14 can be reduced to a simple linear convolution equation. a radius and nucleation velocity of microcracking can be calculated using Eqs 12 and 13. if the mode of mlcrocracklng is the tensile type. the error of positions in attachment of transducers is smaller than this. Then the experimental error on each coordinate is estimated to b e approximately I mm. Single Deconvolution ~ t h o d The AE waveform with more than six channels must be recorded and the multiple deconvolution must be carried out in Eq 14 to determine the deformation moment tensor and characterize the AE sources. ENOKI AND KISHI ON SOURCE CHARACTERIZATION 61 longitudinal velocity in AI-Li alloy give the maximum error of about 0.6 mm in source location. respectively. 9 9 RI-LI TL 1 2~ mm I FIG. which were 240 ~m and 200 m/s in the SiC fiber reinforced glass composite. Copyright by ASTM Int'l (all rights reserved).The results of location of A sources detected during fracture toughness test in AI-Li alloy. 9 -. Figures I0 and ii show the relations between radius a and velocity v in the SiC fiber reinforced glass composite and AI-Li alloy. . From the moment obtained by the single deconvolution method. and then only the slze of microcracking is the unknown parameter and measuring with one channel can determine the size and nucleation velocity of microcracking[4]. However. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized. The v a l u e s of m e a n radius and m e a n n u c l e a t i o n of microcrack were evaluated from the moment obtained by the single deconvolution method. and 45 #m and 50 m/s in AI-Li alloy. where o indicates AE under the low applied load and x indicates AE under the high applied load.

.go o > 0 t 0 sb 150 200 Radius. 200 Ai -Li TL 150 ul o D E o Do o 100 o o~ :) q ~176 ~ ~ ~ o ~= 45t4m >. 62 ACOUSTIC EMISSION 400 SiC / Pyrex 0 0 300 0 0 0 O E ~ 0 oo O 200 o ~o ~o O O > Oo 0 5 = 240 I. No further reproductions authorized.a / IJm FIG.The relations between radius and velocity of microcracking in AI-Li alloy. I0 . ii -. 50 o o D o o "o o 0 g = 50m/s . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.a / lain FIG. The relations between radius and velocity of microcracking in the SiC glass composite.tm 9~- u 100 0o 0 0 9 =200m/s 00 26o 460 6c)0 8oo Radius. Copyright by ASTM Int'l (all rights reserved). .

12. 0 x 10 . ENOKI AND KISHI ON SOURCE CHARACTERIZATION 63 Multiple Deconvolution Method The deformation moment tensor has to be determined to obtain the mode and orientation of microcracking. It can be c o n c l u d e d that the r e c o r d e d AE events due to m i c r o c r a c k i n g are identified as the separation cracks in front of the precrack. No further reproductions authorized. 1.0Z-]T(O 2.3Z -Z.Z.O [._._jK" Figure 13 i n d i c a t e s that the inclination of the microcrack plane to the main crack surface is 1060 and the inclination of the microcrack plane to the direction of the displacement discontinuity is 83 ~ This result has demonstrated that a microcracking occurs in mixed mode of tensile and shear. .ll Z. BS Z i i i i 0 Time / Ms FIG. Copyright by ASTM Int'l (all rights reserved). but the shear component is stronger.The obtained result of moment tensor due to a microcracking in AI-Li alloy. We also developed the experimental and analysis system by using AE waveforms in oil(t) .261 o 1 ~1.. Moment tensor Dik is determined by the frequency deconvolution method using some tim~ points from longitudinal wave arrival. This result demonstrates that the size of cracking with ~any mode can be obtained within about i0 % error by the single deconvolution method. The value of the crack radius from the multiple deconvolution method is about i0 % larger than that value from the single deconvolution method. 12 -. a microcfack is generated with a rise time of about 0.N At 0 E-1 L- 0 At = O. The crack radius a is estimated as 156 ~m from Eq 12 by assuming a . Figure 14 shows the microscopic fracture surface by scanning electron microscope. 17] .4 Nm . the displacement discontinuity [ui] and the normal v i are obtained from the d e t e r m i n e d m o m e n t tensor D. As shown in Fig. Applying the nonlinear least.4 #s. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.83] > "O . Figure 12 shows an example of moment tensor Dik due to microcracklng in AI-Li alloy. SUMMARY We have presented a theoretical consideration for an infinitesimal deformation in accordance with the method of micromechanics.1 . The estimated value of radius a agrees well with the size of the separation crack which is observed at the location of source event in front of the precrack tip.square method to Eq 6.3ay s. We have presented the multiple deconvolutlon method to determine the moment tensor[9. m 0 D .

FIG. 13 -.The microscopic fracture surface in AI-Li alloy.The results of the orientation. This method is remarkable for the dynamic Green's function of finite media by finite difference computer simulation and for the deconvolution with multiple Green's function in frequency domain. fracture mode and size of the microcrack in AI-Li alloy. the following conclusions could be drawn. was developed. x ~ [u1:o=55~ 0=15 ~ v :o =106 ~ 0=102 ~ Cos-l([ ~]. tq. 14 -. No further reproductions authorized. which can determine the moment tensor components in mlcroeracking. Z v z x. The experimental and analysis system for AE source characterization. Copyright by ASTM Int'l (all rights reserved). .v)=83 ~ a=156~m FIG. 64 ACOUSTICEMISSION order to evaluate this microcracking quantitatively. i. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. In this paper we applied this method to fracture toughness testing of the SiC fiber reinforced glass composite and AI-Li alloy.

theoretical Green's function cannot be applicable. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The AE source location could clear that AE source events under the low applied load due to cleavage were generated at the back of the fatigue precrack tip. used. It is obvious from the Eq 14 that the effect of transducer characteristic must be considered. 4. . The values of mean radius and mean nucleation of microcrack were evaluated from the moment obtained by the single deconvolution method. The computer aided system will become more important as well as another measuring system[19]. and means the AE source location possesses high potential. while t h o s e u n d e r the h i g h a p p l i e d l o a d due to s e p a r a t i o n c r a c k w e r e generated in front in AI-LI alloy. In the case that the geometry of the specimen is a plate and an AE source location is fixed. ENOKI AND KISHI ON SOURCE CHARACTERIZATION 65 2. and 45 #m and 50 m/s in AI-Li alloy. which were 240 ~m and 200 m/s in the SiC fiber reinforced glass composite. The results in SiC fiber reinforced glass composite were in good agreement with the observation and the location by AE. In this case. However. the moment tensor with the same components and the different time function may be failed to evaluate by the simplified amplitude method. FUTURE ASPECTS It has been well-known that the AE waveform is strongly influenced by the transducer response. However. such as for a compact tension specimen. the respective sensitivity between transducers and the frequency band of AE sources become more important. In the above respects. In the study of AE source characterization. the absolute sensitivity. the simplified amplitude method may be applicable under some restraint. the author's method may be the most general analysis method because transducer characteristics are calibrated and the Green's function is calculated numerically in the geometry of specimen with the effect of an existing crack. In the case that the AE source occur near the crack tip. size. which have a size of about 300 ~m. the e f f e c t of c r a c k in the G r e e n ' s f u n c t i o n m u s t be considered because the existence of a crack gives a strong influence on emitted waves. care is needed when applying this method because the rise time of time function can change the first peak amplitude as well as the first peak time[18]. Copyright by ASTM Int'l (all rights reserved). that is. respectively. The observation by scanning electron microscope verified that these AE sources were identified as the separation crack in front of the precrack in AI-Li alloy. in the case that the distances between transducers and source are not necessarily the same. as the future method of the AE source characterization. 3. The analysis neglecting the effect of the transducer cannot be applicable. The transducer must be calibrated in any way and the frequency band of the transducer must be checked also. the analysis method which is based on this multiple deconvolution method must be developed with the consideration of the data base technique or intelligent expert system. inclination of microcrack surface and f r a c t u r e m o d e of m i c r o c r a c k w e r e q u a n t i t a t i v e l y e v a l u a t e d . such as Kim et al. No further reproductions authorized. This method may be complicated and it may take more longer time than any other method to carry out the analysis. From the obtained result of moment tensor by the multiple deconvolution method.

T. C. G. [I0] Fukunaga. Dordrecht. M. 67-74. G.. Vol. pp. C. pp. C. 2005-2011. pp.19.. Ouantitative Seismologiy Vo1. and Sachse. K." Journal of AE. B. 35. 1988." in P~oceedings of the 1982 Joint CpnferenGe on Experimental Mechanics. 1982. 66 ACOUSTICEMISSION REFERENCES [I] Mura. "An Approach to Acoustic Emission Signal Analysis. Ohno. Simmons. and Baldwin. G. Martinus Nijhoff Publishers. and Kishi. pp." in Pros ~n AcoustiG Emission IV. (in Japanese) [5] Ohira. [9] Enoki. 1986. Vol. [18] Enoki~ M. pp." Journal of JSNDI. pp. C. and Kurlbayashi. "The Processing of AE Signals. 911-917. M. 1597-1612." IDternational Journal of Fracture. T. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Y. 100-106." Materials Evaluation. [14] Scruby." in Solid Mechanics Research for Quantltat~ve Non-Destructive Evaluation Martinus Nijhoff Publishers.. W. 1977. [3] Wadley.. H. W. 140-147. K. H. 1987. 35.. [8] Kim. (to be published). S. Y. Vol. E. 1987. 3. "The Inverse Source Problem for an Oblique Force on an Elastic Plate. Stress. H... JSNDI. 1981. G. Acoustics . J. M. W. A. StaGey. pp. 1981. K. "Crack Tip Shielding by Micro-cracking in Brittle Solids. 26-38. [19] Sachse. "Response Characterization of Piezo- electric Transducers and Wave Media for Acoustic Emission Source Wave Analysis.. and Kishl. pp. pp. [17] Enoki. pp. 31. N. and Kishi. K. [6] Scruby. 1981. 1982. . Y.. Vol." Journa~ AcousVical Society o~ America. M. 1988. Vol. "Acoustic Emission Source Wave Analysis for Dynamic Information on Micro Crack Formation. Vol." in Progress in Acoustic Emission ~V.295-310. pp. J.. 399-414. [16] Michaels. H. Enoki. Freeman and Company. and Richards. T.. 30. "Defect Characterisation in Three Dimensions by Acoustic Emission.. 29. 1985. San Francisco. and Pao.. [12] Ohisa. 1988. (to be published). T. 182-188. J.1605-1619. pp. 48. and Baldwin. JSNDI. 359-364. "Dynamic Green's Function of Finite Media by Finite Difference Method. pp. J. "Acoustic Emission Source Characterization of Microcracklng in A533B Steel. Kishl. A. and Kihara. A.. and Pao. P. [4] Kishi." Acta Metallur~ica. N. and Kishl. N. R.. and Shrimpton. 1988. 1984. No further reproductions authorized. "On the Determination of Moment Tensor from Acoustic Emission Waveform".. T.. N. and Reliability in D~i~n. Y. 411-423. 211-231. 1986." $ourn~l of Vibration. and Pao. pp. "Quantitative Acoustic Emission Source Characterization during Low Temperature Cleavage and Intergranular Fracture." Acta Metallur~ica. pp. Y... Copyright by ASTM Int'l (all rights reserved). 125-147. Micromechanics o~ Defects in Solids. JSNDI. K. "Theory and Analysis of Deformation Moment Tensor due to Microcracking. R. T. "Three-Dimensional Crack Location by AE. [7] Ohtsu. "Propagation of Elastic Pluses and Acoustic Emission in a Plate. Vol. [II] Hsu." ~gurnal APPlied MeGhanics. "A Study for Analysis System of Acoustic Emission Source Characterization. Society for ~per~menta~ Stress AnalYsis.." in PrQgress in Acoustic Emission ~V. 38. Scruby. [2] Akl. Vol. M.. 1980. Vol.. 77. B. T. B." ~nternational Journa~ of Fracture.1. W. Vol. G. "Characteristics of an Acoustic Emission Source from a Thermal Crack in Glass.. [13] Hutchinson. "Source Inversion Procedure for Acoustic Emission. T. [15] Ceranoglu. The Hague.. and Hardy. N." Journal of Physics D.

New York . American Society for Testing and Materials. Eds. point source. In most cases. Philadelphia. thin-film testing. By suitably processing these signals. Both line and point sources are generated by a transient. . and Sachse. K. By deconvolving the detected ultrasonic signals it is found t h a t a current pulse of short d u r a t i o n generates a step source with risetime a p p r o x i m a t e l y equal to treat of an i n p u t power pulse while a current pulse of long d u r a t i o n produces a linear r a m p source resembling the t e m p o r a l behavior of the t e m p e r a t u r e rise in the heated thin film. b u t other sources include high energy Dr. thermoelastic source. film adhesive strength.astm. STP 1077. No further reproductions authorized. electrical Joule heating of a long. = Acoustic Emission: Current Practice and Future Directions. Roger. =Joule Heating Line and Point AE Sources and the Adhesion of Thin Metal Films. thermoelastic gen- eration. It is also shown t h a t the failure of the film can be detected in the generated ultrasonic signals as well as in the m e a s u r e d electrical resistance of the film.and a short section of the thin film of narrow width. the transient heating is o b t a i n e d b y Q-pulsed laser b e a m s .. K E Y W O R D S : Q u a n t i t a t i v e AE.14853 U S A. Yamaguchi. Green's functions. 1991. It is shown t h a t the line and point sources are of dipolar type. and K.. W.org Copyright by ASTM Int'l (all rights reserved). line source. 67 Copyright9 1991 by ASTM International www. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Kwang Yul Kim is Senior Research Associate and Professor Wolfgang Sachse is a professor in the Department of Theoretical and Applied Mechanics at Cornell University in Ithaca. A B S T R A C T : This p a p e r describes a novel a p p r o a c h for generating line a n d point sources of u l t r a s o u n d from a thin m e t a l film a t t a c h e d to an insulating substrate. Sachse. information on the adhesion s t r e n g t h of the film to its s u b s t r a t e can be found. K w a n g Yul K i m a n d Wolfgang Sachse J O U L E H E A T I N G LINE AND P O I N T A E S O U R C E S AND T H E A D H E S I O N OF THIN METAL FILMS REFERENCE: Kim. W. m e t a l coatings INTRODUCTION T h e generation of acoustic or thermoelastic waves by the transient heating of a localized region of a specimen and its use for imaging and characterizing m a t e r i a l s are r a p i d l y coming into widespread use. respectively. Y. J.

The heating is obtained by sending a current pulse of various magnitudes and durations through the film.y. Gf.. heated by the resistance to the flow of an electrical cur- rent. will expand freely in the absence of a substrate.so) O(t). the thermal expansion of the film is constrained by the film/substrate interface. If the x-axis of a coordinate system is perpendicular to the length of the film and the y-axis is aligned along its length. y . If the linear coefficients of thermal expansion of the metal film and a glass substrate are denoted by a I and ag .. THEORY A thin metal film. respectively..t)Al = f wAl . The temperature rise of the film and the boundary of the substrate in contact with it.r ) dr = DzzCt) Al . uf and d are the Young's modulus.t) resulting Copyright by ASTM Int'l (all rights reserved).and line sources of ultrasound are generated by heating short and long sections of the film. then the force per unit length exerted by the film on the glass at the interface is given by Ed f . This outward force acts in a direction perpendicular to the film strip at both edges of the film. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. This can be realized with modern semiconductor fabrication techniques. It should also be possible to generate a variety of thermoelastic waves associated with different source types by a proper layout of the thin film on the substrate.. . respectively. t . can he written as EwdAl D==(y.t) Al GP . can be expressed by a Green's function according to [2] ufCr. . However. when the film is attached to a substrate. o (i) 1--Vf where E. thus constituting a horizontal dipole force. duration and magnitude of the current pulse and also its relatively low cost in comparison with other methods. located at y. ( a / . (2) l-u/ where w is the width of the thin film strip. resulting f r 0 m a p o i n t dipole D~z(t)Al.. t ) = /: Dz~(y.t) (3) where G P represents the Green's tensor associated with a point source and '*' denotes a convolution integral in time. is given by 8. respectively. No further reproductions authorized. Then the normal displacement u~(r.t) at location r and at time t. The advantage of using an electrical power pulse lies in its simplicity and versatility in generating elastic waves of various shapes and amplitudes through control of the risetime. . The displacement u~(r.~(r. Al. 68 ACOUSTICEMISSION electron beams or intense pulses of synchrotron-generated x-rays. It is demonstrated that both point. Poisson's ratio and the thickness of the film. ( r . In this paper another method of generating thermoelastic waves is described which is based on the electrical Joule heating of a segment of a thin metal film of narrow width which has been directly deposited on an insulating substrate [1]. then a horizontal dipole associated with a point source having a short section of the film length.

t) can be approximated by G=~ t). Cs source on a glass s u b s t r a t e with the re.(r.5.t ) = a=%Cr.% S x Ce=. (r. the -I . then... ..o .'~ -'~ 0 .0.O 9 ]o.5 IO..O I W h e n s is sufficientlylong.. -.O alized ray theory. . receiver at shown in Figs. .-~. Green's Function. O.~_J. Copyright by ASTM Int'l (all rights reserved). Epicenler I.(c).=(r. If the time of interest is limited to the ar. t) dy g E o. t) dy = D==(t) * Ce=. i... Off-Epicenter our experiments.! . s is given by u~(r. . .350cm//~s) for a horizontal dipole ceiver located on the opposite side are source. (a) Point source.O Time (Ms) that of an infinitelylong line source. f~. l ( a ) . Green's Function.= were available. . To c o m p u t e the val... KIM AND SACHSE ON JOULE HEATING LINE 69 from a line source of length. Calculated waveforrns Figure I: Green's functions of a plate for a short and a 1 0 c m long thin strip 0. Algorithms for c o m p u t i n g the point.0 = Le a~.t) = L l D==(t) * G=z. Time (Ms) ues of G=z. ..-_. (4) can be used. even t h o u g h using Eq. No further reproductions authorized..:. . ~5 I $ ~ To maximize Tp l in an experiment.#.. (5) would be more con- venient for times less t h a n Tel if G=~. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. .(b) Line Dip. In (e) Line Dip. Epicenler rival time.0.~. . Eq.O value of rpl of a p p r o x i m a t e l y 9/~sec for ~C) s the epicentral receiver. x g aL.. this corresponded to a I. 0 (5) for 0<t_<rp ~. (c) Line source.z l . .=(r. . corresponding to the first l. G~z.o " 215 ' co p e a r i n g in the above equations have been developed for the near field of a source I by Ceranoglu and P a o [3] using gener- -I. E t i i source Green's function of the p l a t e ap.=C.968cm thick (Ce -.582cm/#s.L 7.O where .t) (4) (a) Point Dip. (bI Line source. receiver at epicenter.Oi P-(longitudinal) ray emanating from the ro ends of the line source. y. which 1 P is the Green's function corresponding to -I. epicenter. receiver at 2 H from epicenter. Green's Function. .o Time (/zs) acoustic sensors should be placed at points equidistant from the ends of the film.=(r P .

T h e m e a s u r e m e n t s y s t e m is shown in Fig. also be possible to use films possesing lower conductivities. To o b t a i n Copyright by ASTM Int'l (all rights reserved).968 cm thick. 2. One was silver. which was deposited by sputtering. Each brass p l a t e h a d two electrodes on it.5 # m thick. polished brass plates were a t t a c h e d to the ends of the film strips with a silver epoxy cement. 15 cm wide and 0. one for the current excitation and the o t h e r for t h e p o t e n t i a l measurement.5 m m wide a n d 1 # m thick o b t a i n e d by evaporation. Both sensors were located m i d w a y between the ends of a thin film segment. It should. t i t a n i u m . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. T h e n o r m a l surface motions of the elastic waves were detected on the side opposite to the films by two. The other was a l u m i n u m .I Computer I DataDiskSt~ 1 F i g u r e 2: Electronic block d i a g r a m of line a n d point-source A E generation and the detection system. T h e results to be described were o b t a i n e d on two thin film specimens. T h e initial experiments were c o n d u c t e d on two film m a t e r i a l s which were chosen for their high electrical c o n d u c t i v i t y so t h a t the t o t a l electrical resistance was minimized to maximize the current and hence the power which could be applied to the thin film. .Sens6r~/ _ // / I~ Digitizer /I Glass i !// !1/ II ~ Charge~ mplJfiersV J171 X-Y Scopet Digitizer Micro. T h e spacing between the end pieces on the silver film was 2 m m giving an a p p r o x i m a t e point source while on the a l u m i n u m film they were s e p a r a t e d b y a b o u t 10 cm to give a line source. 1. I IACCurrent ] I TriggerOutI C+~ioJ-Probe/~+/1/lhin7 -V /// J|l IOCHHII I InTransi TriEgXetentr I H'l~g:eCGUr::::torC.I //' Filrn~AE~_. No further reproductions authorized. . 2 m m wide and 0. however. 70 ACOUSTIC EMISSION MEASUREMENTS T h e thin m e t a l films were deposited as strips on one side of a glass plate a p p r o x i m a t e l y 15 c m long. such as chromium. T h e excitation was o b t a i n e d using a current pulse g e n e r a t o r c a p a b l e of deliv- ering a m a x i m u m power of 20 k W in pulses of nearly constant a m p l i t u d e for a t i m e interval ranging from 2 # s to 500#s a n d o p e r a t i n g with a r e p e t i t i o n r a t e of 10 pulses/second. Two small. one m o u n t e d at epicenter a n d the other at 2 H away from t h e epicenter [4]. nickel. m i n i a t u r e high-fidelity capacitive transducers.

and 0. Both recorders operated under control of an interactive minicomputer-based data acquisition and processing system.3#m thick film of chromium. short section of silver film.5 I0.0 Time (Fs) Time (Fs) Figure 3: Short-duration excitation.0 2. 0. KIM AND SACHSE ON JOULE HEATING LINE 71 a ground plane for the capacitive transducers. whereas the generally flat portion of the source- . The dipole source time function D~=(t) obtained from Eq.0 2. (b) Point dipole source time function.0 1. The output of the charge amplifier was measured by a second waveform recorder operating similarly as the first.0 / i ~ I I I I I I 0.0 #sec which approximately equals that of the input power pulse.5. (a) Electrical Power. . 2 m m long.00 0.0 7. The source-time function is a step function with a risetime of about 1.50 Short Ag Film x x "~ L25 g. The magnitude of current flowing through the film was measured by an AC current probe whose bandwidth was 120 Hz to 60 MHz. This suggests that the temperature rise and hence the thermal dilatation of the film instantaneously follows the input power increment associated with the rising portion of the power curve.5 . 3(b).o time function after the ramp may be the result of the slow thermal diffusion after the input power begins to decrease abruptly. The measurements were carried out with current pulses of magnitude ranging from several to 100 Amperes with duration ranging from a few to 100 #s. The signal from each capacitive transducer was amplified by a charge amplifier whose bandwidth extended from 10 kHz to 10 MHz. RESULTS Short-duration Excitations In Fig. the entire bottom side of each specimen was coated with a 0. This relatively short length of film acts as a point source of ultrasound when excited by the current pulse. Ag FILM INPUT POWER POINT SOURCE: Recovered Source-time Function 2. No further reproductions authorized. The signals from the current probe and the voltage across the thin film were fed into a two-channel transient recorder operating at a sampling rate of 30 MHz and 10 bits.5 # m thick silver film.0 3.0 0. Copyright by ASTM Int'l (all rights reserved). l(a) is called the magnitude of the source and this is plotted in arbitrary units on the ordinate axis in Fig. 3(a) is shown the measured excitation current pulse which was applied to the 2 m m wide.0 (a) ((b) 2. (4) by deconvolving the displacement signal measured at epicenter with the Green's function shown in Fig.0 4. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.

Time (p.5 mm wide..00 0. '(~ ~. synthetic wave- forms were reconstructed and compared to those measured. 4(a)-(c). Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.5 cases was very good. l(b) 5. l(b). It was found that the source- time function determined from the epi- central response was again a ramp-step function whose risetime equalled that of 2.00 and (c). (3) with the source strength Time (/zs) shown in Fig. Measured / placement signal shown as a solid line ~ .. Copyright by ASTM Int'l (all rights reserved).(b) THERMAL LINE SOURCE: Epicenter duces essentially the same type of source.s) of source type. 4(b) with the Green's function shown in Fig. (c) Ob- computed waveforms is found to be ex.0 #m thick to which a short-duration pulse of maximum power of about 15kW was applied were analyzed similarly using the (a) LINE SOURCE'.00 23 Ampere current pulse 17#s in dura. 0. I0.0 F convolving the measured epicentral dis.0 termined by the temporal response of x the thermal dilatation of the film. . Figure 4: (a) Line dipole time func- son is made for the case of the off-epicen.5 5. 2H off-epicenter.00 '".. Shown in Fig. tion associated with the current pulse of tral responses in Fig.'~"--4. (b) Observed and recov- the agreement between the measured and ered waveforms at epicenter.0 75 using Eq. 4(a) is the time function of the line (e) THERMAL LINE SOURCE: 214 Off-Epicenter dipole source which was obtained by de.0 7.0 2. Measured C3 .. Theoretical A sample result obtained on an alu. served and recovered waveforms at cellent.50 the input power pulse. regardless Time (p. 4(b). Theoretical/ in Fig. the short pulse excitation pro. No further reproductions authorized. In every case long-duration. Hence. A small parabolic contribution is buried in this source-time function. i I ~ I L L I i ~ i ~ I 3 I time function since it is principally de.5 E 3 -2. be it a point or a line source. 4(a) for the aluminum film line source on a glass substrate is also shown in Fig. A comparison between the ex- perimental waveform measured at epi- ~ oo center and that computed theoretically 0... A similar compari. The agreement in all 0. 4(c).50 0 Long-duration Excitations -.I minum film which was heated with a 2 -5.. RecoveredSource-lime Funclion Green's functions shown in Figs. Once the source- time function was known..0 2.5 5. 10 cm long and 1. .s) tion is shown in Figs. 72 ACOUSTICEMISSION The line source responses of an aluminum film 1.

a temperature increase of 23. The resistance measurements showed that the temperature increase and hence the thermal dilatation of the metal films were almost linear with time with only a small parabolic term present.5 #m thick silver film from a glass substrate. 5(a). (a) Failed region.see) Figure 5: Failure of a 0. Knowing the coefficient of resistance. The polynomial is extrapolated to the time of the arrival of the excitation pulse so that the initial electrical resistance /to of the film can be determined. of the film material.oo ~ . the temperature rise of the film can be computed by solving for O(t) in Eq. The resistance of the film. KIM AND SACHSE ON JOULE HEATING LINE 73 The actual temperature rise in the thin film can be obtained from dynamically measured electrical resistance data.50 I Film width is 2 mm -1.u. . For example. R.00 Time (. In all cases. A Copyright by ASTM Int'l (all rights reserved). ~/. : : : : : : : : : : : : : : : : : : : : : : : : : : : 5. is expressed as _R(t) = R0 [1 + ff O(t)] (6) where R0 is the initial resistance of the film and -y is the temperature coefficient of unit resistance. Such measurements show that there is a sharp initial impedance transient which is followed by an slow increase in the resistance corresponding to the rise in film temperature until the current decreases abruptly at the end of the pulse. This is in agreement with the source-time functions found for the AE signals from the pulse-heated films.0 ~0. as a function of its temperature rise 0. for a short. (a) (b) THERMAL LINE SOURCE: Epicenter (Failure) o.0 20. thin film of silver on glass to which was applied a long-duration pulse of 22 Amperes. 10.0 ~ -0. Between the two impedance transients at both ends of the pulse lies the contribution of the pure resistance whose temporal behavior can be fit with a quadratic polynomial determined by a least-squares method.0 25. (6) for any instant during the excitation.7~ was determined at t = 17#sec just before the excitation pulse dropped to zero. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.0 15. . the determined value of /to was found to be in good agreement with the value determined via a static DC measurement. (b) Epicentral AE displacement signal exhibiting film failure. Film Debonding The debonding of the thin film from its substrate as a result of the accumu- lated thermal biaxial normal stresses at the interface is shown in Fig. No further reproductions authorized.

correspondence between the temperature rise of the film and the dipole source strength. oct*) (s) a=. The magnitude of the current measured when debonding oc- curred was about 30 Amperes. No further reproductions authorized.0 25. The off-epicentral signal detected by a receiver 2H from a silver film source is shown in Fig. wd (7) Because of the .00 (20.1MPa. the temperature at film failure can be determined from the mea- sured electrical resistance characteristics. 1.2~ Once this temperature is known. 74 ACOUSTICEMISSION Ag FILM IMPEDANCE (Failure) 5. 5(b). an alternative procedure is based on processing the emitted ultrasonic signals to determine the temperature of the film at the instant. : . clear indication of the film failure can also be observed in the emitted ultrasonic signals. For the silver film this was 64.0 20.O 5. (8).1 mfl/12/~ and Eq. + I ] I I I 1 l I I I I O.648 x 1O-2t .7#sec was 1. The case of the silver film is shown in Fig.116 x 1O-4t 2 ~. (4) to determine the dipole source strength at the instant of film failure.0 30. From this. the normal stress in the film at the instant of failure can be found according to a== = a~ .723 + 2. (6) the temperature rise of the silver film was calculated to be 65. It should be possible to deconvolve this signal which corresponds to Eq. the failure stress at the interface can be found E . t*. #(t*). Knowing this.0 15. T : i I I i L I [ I I I I .208 ~2 while the initial resistance obtained by extrapolation to the time of the pulse arrival at 0.0 ~ .0 10.50 R = 1. Copyright by ASTM Int'l (all rights reserved). uf Alternatively. . . Clearly evident in the data is the instant when the film begins to debond from the substrate.742 fl. of its failure. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.~ : . Using the value of ff equal to 4. the debonding stress acting at the interface can be obtained from Eq. = a~y -. The value of resistance measured at the instant of failure was 2. ~ 2. 1. 6.0 Time (~sec) Figure 6: Electrical resistance characteristics of the silver film showing film failure.

A. G. Japanese Society of Non- destructive Inspection. 1981. and Sachse. pp. (2) A short-duration current pulse generates a source whose time function is characterized as a ramp-step for which the risetime is approximately equal to that of an input power pulse. and Richards. 1989. exhibiting an almost linear behavior in time with a small non-linear contribution. L. 1988. Copyright by ASTM Int'l (all rights reserved). Y. Vol. Y. K. W. 86. Journal of Applied Mechanics... Quantitative Seismology: Theory and Methods. pp. Tokyo. Eds. Freeman. pp. the following conclusions can be drawn about thermoelastic waves generated by Joule heating sources: (1) A localized. in Progress in Acoustic Emission IV. 1989. and Pao. Vol. P... . K. General Responses". 1980. CONCLUSIONS Based on this work. Y. No further reproductions authorized. Y. "Thin-film Acoustics: Line and Point Source Generation and Testing of Thin Films". K. Higo. "Propagation of Elastic Pulses and Acous- tic Emission in a Plate: Part I. Review of Scientific Instruments. 3. Chapt. [5[ Kim. But this is far more difficult. Part III. 98-105. Journal of the Acoustical Society of America. 125-147. Vol.. ACKNOWLEDGEMENTS This work has been supported by the Office of Naval Research (Physical Acous- tics Program). Part II. transient Joule heating generates a thermoelastic pulse whose source-type is characterized by a dipole. H. REFERENCES [1] Kim. W. (4) The ultrasonic source characterization method can be used to study thermal properties of materials and to predict the temperature rise of a heated conductor.. 60. N. and Sachse. San Francisco. Castagnede. Vol.6]. 2785-2788. I. I. K. "Miniaturized Capac- itive Transducer for Detection of B r o a d b a n d Ultrasonic Displacement Sig- nals". Transactions ASME. K. Use of the facilities of the Materials Science Center at Cornell University is also acknowledged. Yamaguchi. hence the work to date has been based on dynamic resistance measurements as described in the previous paragraph [5. W.. B. Kimpara and Y. (3) A long-duration current pulse excites a source whose time function corresponds to the temperature increase of the thin film. Niu.. [3] Ceranoglu. 48. [4] Kim. and Sachse. 875-884. would be the extraction of that portion of the generated ultrasonic signal which corresponds to the failure of the film. (5) The line source generation from the metal thin film can be used to determine the adhesion strength of a thin conductive film on a substrate. Theory. KIM AND SACHSE ON JOULE HEATING LINE 75 More difficult than the above procedures. Epicentral Response. "Generation of Elastic Waves from Line and Point Sources by a High-current Pulse Method". pp. [2] Aki. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.

. W. D. No further reproductions authorized. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. "Determination of the Bond Strength of a Metal Coating on a Substrate'. A. B. in Interfaces Between Polymers. MRS Symposium Proceedings Series. Eds. pp. Rosenberg. Vol. 76 ACOUSTICEMISSION [6] Sachse. DeKoven. and Conway. J. Kim. Materials Research Society. PA. Pittsburgh. K. 1989.. Gellman and R. Metals and Ceramics. H. Copyright by ASTM Int'l (all rights reserved)... M. 249-259. 153. Y.

. 1991. Heiple. KEYWORDS: acoustic emission.. the energy release from fracture of the boron particles can be calculated from expressions which have been derived for the fracture of an Dr. Professor Carpenter (University of Denver) is a consultant. Carpenter. and K. This relation can be used to estimate the energy released by other comparable acoustic emission sources in samples of similar geometry from their acoustic emission signals. R. S. ASTM STP 1077. The energy released by fracture of the boron particles was estimated from elasticity theory using measured diameters of the individual fractures. S. 77 Copyright 9 1991 by ASTM International www. calibration. Eds. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. and Mr. inclusion fracture. "A Calibration Source for Acoustic Emission Analysis. H. The response of an rms voltmeter to individual acoustic emission bursts was found to be proportional to the energy of the burst (integral of the squared signal voltage).astm. The energy of the acoustic emission signals produced by boron particle fracture was found to be proportional to the energy released by the particle fractures. The boron particles were introduced into the aluminum using a powder metall~rgy approach. No further reproductions authorized. and Christiansen.. Yamaguchi. and Scott S. signal energy The fracture of boron particles in an aluminum matrix was proposed by Hamstad as a potential well-defined acoustic emission source [I]. C. Golden CO 80402-0464. Carpenter. Roget. Sachse. Boron particles were incorporated into 2219 aluminum tensile bars using a powder metallurgy approach. Box 464. J.. Since the mechanical properties of boron and aluminum are well known. Philadelphia. Clinton R.org Copyright by ASTM Int'l (all rights reserved).. Steve H.. S." Acoustic Emission: Current Practice and Future Directions. ABSTRACT: Acoustic emission produced by the fracture of boron particles in 2219 aluminum during tensile deformation was measured and characterized. Christiansen A CALIBRATION SOURCE FOR ACOUSTIC EMISSION ANALYSIS REFERENCE: Heiple. Christiansen is a Senior Development Engineer at EG&G Rocky Flats. Heiple is an associate scientist. American Society for Testing and Materials. Inc. W.

. Samples were fabricated using 250-350 #m boron particles. A similar. so it was expected that sufficient stress to produce fracture of the boron particles would be developed during plastic deformation of the aluminum matrix. no burst emission is produced during deformation. Acoustic emission was detected with a Dunegan-Endevco S140 transducer (resonant piezoelectric type. No further reproductions authorized. 140 kHz nominal resonant frequency). Copyright by ASTM Int'l (all rights reserved). 60-90 #m boron particles. and verified that the fractures occurred during tensile testing .000 psi).not during sample fabrication. recorded directly with a Soltec SDA2000 transient recorder. The metallographic examination determined the number of particles fractured during deformation. The correlation between the elastic energy released and the resulting acoustic emission signal is useful both for calibration purposes and for estimating the energy released by other sources from their acoustic emission signals. Following tensile testing.particularly in the peak-aged condition. First. and with no boron particles added. The pressings were hot-swaged (450-460 ~ into rod. this alloy is age hardenable to fairly high strengths for aluminum alloys. and tensile bars machined from the rod. and the level of continuous emission is very low . There was no detectable porosity in the resulting pressing. Particles in both the grip and gauge sections were examined. 78 ACOUSTICEMISSION elastic inclusion in an infinite matrix [2].5 hr) and aged to peak strength (190 ~ for 30 hr after a room temperature hold). one tensile bar containing large boron particles was examined in detail metallographically. Second. The amplified signals were characterized with an Acoustic Emission Associates Phoenix acoustic emission signal characterization system. estimated the size of the fractures. A block of 2219 aluminum was milled into small chips. The resulting compact was vacuum hot pressed (430 ~ 20. but less extensive. and measured with a Hewlett-Packard 3400A rms voltmeter. The tensile bars were solution heat treated (550 ~ 1. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Scanning electron micrographs of typical small and large boron particles are shown in Figure I. The aluminum matrix was gradually removed by electropolishing and the boron particles were photographed individually in a scanning electron microscope as they were uncovered. examination was carried out on a tensile bar with small boron particles. EXPERIMENTAL PROCEDURE Aluminum alloy 2219 was chosen as the matrix material for two reasons. Transducer output was amplified with a Panametrics 5050AE-160A preamplifier. Boron particles were mixed by hand with the chips and the mixture cold pressed.

(a) 60-90 ~m particles. ON A CALIBRATION SOURCE 79 Figure i. No further reproductions authorized. HEIPLE ET AL. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Copyright by ASTM Int'l (all rights reserved). (b) 250-350 #m particle. Scanning secondary electron micrograph of boron particles added to 2219 aluminum. .

As expected. never in the grips. with units of volt~sec. aluminum oxide and any other microstructural modifications introduced by the powder processing do not contribute detectable acoustic emission. Some particles were completely intact. The fracture surfaces are nearly parallel. The resultant quantity. and these small events are attributed to decohesion of the particles from the matrix. but is not actually an energy unit. Fractures with these characteristics were found only in the gauge section. and a few particles had multiple fractures. A sample without boron particles was fabricated and tensile tested to verify that the fabrication procedure did not affect the basic acoustic emission behavior of the material. The energy was calculated in this way for all of the larger signals from a large-particle and small-particle tensile bar. . and there was often aluminum between the broken pieces. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. An example of such a particle is shown in Figure 4. A comparison of the calculated energy in each of the acoustic emission signals with the height of the rms voltmeter spike produced is shown in Figure 5 Copyright by ASTM Int'l (all rights reserved). For the crushed particles. particularly for the large particle samples. No further reproductions authorized. the fracture faces were generally not parallel or perpendicular to the tensile axis. then the number of these events is comparable to the number of boron particle fractures observed in the sample gage length. the pieces were more widely separated. No burst emissions were detected and there was no continuous emission detectable at ii0 dB total amplification. an example is shown in Figure 3. The acoustic emission behavior of the powder-processed material was identical to that of wrought 2219 aluminum [3]. that is a point by point integration of the sign~l waveform squared. No continuous emission was observed at yield or during macroscopic plastic flow. i. Burst-type acoustic emission began in the boron-containing samples in the macroscopically elastic region near yield and continued through and well beyond yield. is proportional to energy. About half of the particles in the gauge section contained fractures of this type. fractures of this type were found in both the grip and gauge sections. the amplitude of the signals from the small-particle samples was less than from the large-particle samples. and approximately perpendicular to the tensile axis. Other particles were crushed during processing. Thus. so a direct comparison can be made of each acoustic emission signal and the response of the rms voltmeter to it. those large enough to individually produce a spike on the rms voltmeter. The number of acoustic emission bursts detected greatly exceeded the number of boron particles in the sample gage length. 80 ACOUSTICEMISSION RESULTS An example of a large boron particle fractured during deformation is shown in Figure 2. If only the largest events are considered. close together.e. One property of each acoustic emission burst which can be calculated using the transient recorder is its "energy". The time at which each signal occurred is also recorded by the transient recorder. Most of the events were relatively small.

Scanning secondary electron micrograph of a large boron particle fractured during tensile deformation. No further reproductions authorized. Figure 3. HEIPLE ET AL. The tensile axis is vertical. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Copyright by ASTM Int'l (all rights reserved). Scanning secondary electron micrograph of an intact large boron particle (following deformation) in the tensile bar gauge length. . ON A CALIBRATION SOURCE 81 Figure 2.

because the signal shape is quite variable. The tensile axis is vertical. by twin formation in uranium. Scanning secondary electron micrograph of a boron particle crushed during sample material fabrication. The energy released is proportional to the radius of the fracture cubed. Estimates of the energy released by martensitic transformations in Au-47. The resultant plot is given in Figure 7.5 at~ Cd and in plutonium. The calculated fracture energy releases were ordered from largest to smallest. the rms voltmeter response is proportional to signal energy. An expression for the strain energy released from cracking a spherical inclusion in an infinite matrix when the inclusion has elastic properties different from the matrix has been calculated by Kant [2] and reproduced in Reference 4. . and Figure 6. They were plotted against the measured acoustic emission signal energies (similarly ordered) for the 32 signals large enough to individually produce a spike on the rms voltmeter. No further reproductions authorized. 82 ACOUSTICEMISSION Figure 4. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. This plot serves as a calibration from which the elastic energy released by a real acoustic emission source in a sample of useful geometry can be estimated from the acoustic emission signal produced. Thus signals with quite different energies can have similar peak amplitudes. The radius of each boron-particle crack resulting from tensile deformation of the large-particle tensile bar was estimated from the scanning electron micrographs and the energy release then calculated. It should be noted that the correlation between signal peak amplitude and rms meter response isn't nearly as good as with signal energy. As expected. No noise correction is required because the actual signal during the time that the spike is produced exceeds background by at least two orders of magnitude. and Copyright by ASTM Int'l (all rights reserved). the applied stress squared. and is also a function of the elastic constants of the matrix and inclusion.

. Copyright by ASTM Int'l (all rights reserved). No further reproductions authorized.. Values referenced to transducer output. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Values referenced to transducer output.- o E o > y. RMS voltmeter response versus acoustic emission signal energy for small-particle sample. ON A CALIBRATIONSOURCE 83 30 LARGE P 20 o D. 0 ' I ' I 0 10X10 -12 20X10 -12 Signal Energy (volt z s e c ) Figure 6. w o n- 10 o E o > ' ' ' ' [ ' ' ' ' I ' ' ' ' -9 O lx10 -9 2x10 -g 3x10 SignalEnergy (vol~sec) Figure 5. m o SMALL P A R T ~ > P o f~ o (4 r~ . RMS voltmeter response versus acoustic emission signal energy for large-particle sample.j. HEIPLE ET AL.

Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Copyright by ASTM Int'l (all rights reserved). CONCLUSIONS Samples where the only source of acoustic emission is the decohesion and fracture of second phase particles with known mechanical properties have been successfully fabricated and tested.= u. with boron particle additions. over a wide range. a linear relation is found. The aalculated fracture energy for this particle was taken to be the sum of the energies for the two branches. 84 ACOUSTIC EMISSION o f- Ill S i m S P . . The acoustic emission from this material has been measured and characterized. The material used was 2219 aluminum. The estimates are in reasonable agreement with other available data. Details of these estimates are given elsewhere [5]. The response of an rms voltmeter to a single event is proportional to the signal energy. From this relation. provided the source velocity and sample geometry are similar to the boron particle fractures. referred to transducer output. by dislocation motion in beryllium were made from their acoustic emission signals. prepared using a powder metallurgy approach. Boron particle fracture energy versus acoustic emission signal energy. an estimate in absolute energy units can be made of the energy released in a sample by an acoustic emission source from the acoustic emission signal observed. No further reproductions authorized. i ~ i i i i i ! 0 lx1~ s 2 x 1 0 "~ Signal Energy (vol~sec) Figure 7. When the acoustic emission signal energies are correlated with the energy releases calculated for the observed boron particle fractures. where the signal energy is the integral of the voltage squared. The point plotted with a + is from a particle with a branching crack.

.. No further reproductions authorized. S. No. H. M. CA.. "Fracture of Boron Particles in 2219 Aluminum as a Known Acoustic Emission Source. S. S.-Dec. 1979o [3] Heiple.. C. Oct. 587-598. Carpenter. R.. "Acoustic Emission from Dislocation Motion in Precipication-Strengthened Alloys. 1981. Carpenter. H. S. 6. and Carpenter. HEIPLE ET AL. Albuquerque Operations Office. pp.. University of Denver. pp. [4] Heiple. C. 215-237. Nov.. 15." Doctoral Thesis. "The Elastostatic Axisymmetric Problem of Cracked Sphere Embedded in a Dissimilar Matrix.. R. ON A CALIBRATION SOURCE 85 ACKNOWLEDGMENT This work was supported by the Department of Energy. University of California. R. 4. Berkeley." Journal of Acoustic Emission. ]987. Their support is gratefully acknowledged. [5] Heiple. Copyright by ASTM Int'l (all rights reserved). and Carr. Unpublished Research.A Review: Part II. C. 1990." Metal Science Vol. S.." Accepted for publication in Acta Metallurgica. R. [2] Kant. and Christiansen. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.. J. . REFERENCES [I] Hamstad. A.. Vol. "Acoustic Emission Produced by Deformation of Metals and Alloys . H.-Dec. M.

ABSTRACT: A method is proposed to simultaneously determine source locations. source location INTRODUCTION Ultrasonic velocity imaging has been successfully used to study the effects of changing stress fields and map stress concentrations in Mr. and Hutchins.P. using both spatially uniform and restricted event coverage. Roget. KEYWORDS: ultrasonic imaging. and David A. 86 C o p y r i g h t 9 1991 b y ASTM I n t e r n a t i o n a l ~ee~.. Images of the velocity structure are sensitive to the state of stress in a sample. and can be used to map spatial variations in stress induced changes. Yamaguchi. acoustic emission. Young. W. Sequential passive imaging should prove to be a powerful tool for imaging the evolving in-situ stress field using only basic monitoring instrumentation. K... Hutchins SIMULTANEOUS VELOCITY TOMOGRAPHY AND SOURCE LOCATION OF SYNTHETIC ACOUSTIC EMISSION DATA REFERENCE: Maxwell. ." Acoustic Emission ~ Current Practice and Future Directions. Shawn C. S.C. Maxwell. "Simultaneous Velocity Tomography and Source Location of Synthetic Acoustic Emission Data. Synthetic arrival time data were used to attempt to passively image a velocity model. Maxwell and Professor Young are with the Rock Physics and Engineering Seismology Laboratory.org Copyright by ASTM Int'l (all rights reserved). Paul Young.astm. Eds. J. Sasche. passive tomography. This study documents the first phase of a passive imaging research project. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. however the restricted event coverage resulted in some loss of resolution. Canada K7L 3N6. Queen's University. the actual model was reproduced. R. R.. and passively image the velocity structure of a deformed sample. In both cases. 6STM STP I077. Kingston. Department of Geological Sciences. Professor Hutchins is in the Department of Physics at Queen's. D. The method utilizes arrival time data measured from recordings of induced acoustic emissions. American Society for Tes'ting and Materials. Philadelphia.A. No further reproductions authorized. 1991.

an examination of synthetic data sets. Further. source locations are calculated assuming a constant velocity. The travel times are then used to reconstruct or image the spatial variations of velocity. Similarly. in that the recorded waveforms of artificial ultrasonic sources are used to measure travel times through the sample. concurrent monitoring of induced activity and active imaging has proven to be a useful tool in examining failure mechanisms in both laboratory acoustic emission and in-situ induced microseismic activity studies [1. No further reproductions authorized. mining induced seismic events which result in damage to the mine workings.3]. more accurate source locations can be determined. which are sensitive to the state of stress within the sample. By accounting for velocity heterogeneities. active seismic velocity imaging using artificial seismic sources has been utilized in the delineation of stress concentrations. a research project has been undertaken to examine the application of passive imaging techniques in the investigation of rockbursts. which were used to calibrate and test a passive imaging algorithm. Generally. This is an active technique. and mapping of other geotechnical properties of in-situ rockmasses [2. by first applying the techniques to acoustic emissions recorded in the laboratory. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. since velocity images may be generated by exploitation of information contained in arrival time data sets of the recorded activity. As part of this research strategy. This study examines the application of passive imaging techniques. ON VELOCITY TOMOGRAPHY 87 deformed rock samples [i]. To develop and test methodologies to be used in the investigation of rockbursts in the mining environment. Passive imaging is a logical extension of concurrent monitoring of induced activity and active imaging. . Fermat's principle is used to linearize arrival times residuals of n events recorded at m stations as Copyright by ASTM Int'l (all rights reserved). The method was originally developed to invert local earthquake arrival times [4]. Late stages of mining in brittle rock types are occasionally plagued by the occurrence of rockbursts.3]. MAXWELL ET AL. This study documents the first phase in this research project. for ultrasonic velocity imaging using arrival time data from acoustic emission monitoring. velocity imaging can provide additional information for the study of the failure characteristics of rock samples in the laboratory. equipment. Furthermore. Passive ultrasonic imaging involves simultaneous determination of the source location and velocity structure by inversion of acoustic emissions arrival time data. Improved source locations are important since they are one of the most fundamental results of an acoustic emission experiment. and is some cases loss of life. controlled laboratory experiments of deforming rock samples were performed. PASSIVE IMAGING A linearized inversion method to simultaneously determine the three-dimensional velocity structure and source locations was used.

The center (y . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. and B is the corresponding mn x 4m matrix of travel time partial derivatives. The solution to Eq. and the calculation of its inverse can easily be computed by the individual inversion of the m-4 x 4 block submatrices. Computationally. (i) where T is a mn vector of arrival time residuals calculated from an initial parameterized velocity model and source locations. 3 requires the inversion of the p x p matrix 9 T (OA) and the 4~n x 4m matrix (B B) usually with some damping to ensure 9 o . No further reproductions authorized. SYNTHETIC EXAMPLES The synthetic model used is shown in Figure i. O .A T . (3 v . and h is a 4m vector of hypocenter perturbations. The size of the cube was normalized to be 5 units in each dimension.(OA)-IoT. creating 125 independent velocity parameters. the solution is the same as a least-squares tomography solution in addition to m least-squares source location solutions.ATB(BZB)-IB T. A tapered singular value decomposition method was used to invert the matrices [5]. Nonlinearity of the problem generally requires an iterative method. . This equation is simply a combination of the standard direct velocity tomography and source location equations. and A is the corresponding mn x p matrix of travel time partial derivatives. A cube was parameterized with 5xSx5 constant velocity blocks. which was normalized to Copyright by ASTM Int'l (all rights reserved). (B B) is a block diagonal matrix.BTAv). where.3) plane of blocks was modeled as a high velocity feature. h = (BTB)'I(BTT . 88 ACOUSTICEMISSION T = Av + Bh. v is a p vector of velocity model perturbations. ' T llnearlty and avold singularity in the matrices. The residuals are calculated assuming some initial parameters. with a 5% velocity contrast compared to the constant velocity background. ATT ~ ATAv + ATBh (2 BTT = BTAv + BTBh Solving these simultaneous equations for v and h gives. with ray tracing to account for ray bending effects A least squares solution of the problem results in two "normal equations".

The synthetic model is identical in each such layer corresponding to a horizontal slice through the model. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Corrections to these initial parameters were then computed by solving Eq. it is intended to isolate and analyze the resulting resolution in the final images caused by restricted spatial distributions of source and sensor locations. Although this is an oversimplification of real data. although surface mounting is more practical in real experiments. Calculation of the synthetic arrival times utilized straight ray paths. No errors were included in the arrival times. with 50 events confined to the sides of the high velocity layer. The first example assumes spatially uniform event coverage. each with distinct event spatial coverage. A plan view of a single layer of the velocity model in a plane perpendicular to the z-axis is shown in Figure 2. . unity. The second case considers more realistic event distributions. with spatially uniform event coverage. No further reproductions authorized. Events were simply modeled as point sources. Sensors were assumed to be located at each of the 8 vertices of the cube. MAXWELL ET AL. perpendicular to the z-axis. Two examples were considered. Initial parameters in the simultaneous inversion were a constant background velocity and associated least squares source locations. 3. In the case of the first example. ON VELOCITY TOMOGRAPHY 89 Z FIGURE I Physical model for the three-dimensional velocity structure. with equidirectional radiation patterns and zero rise times. two iterations were required to reproduce the velocity model Copyright by ASTM Int'l (all rights reserved). with 125 events located at the center of each cell.

The first case. Location errors were reduced from up to 109 of a unit cell. In this case. The imaged model appears identical to the actual velocity model (Fig. Figure 4 shows a plan view of the event relocations. Therefore. Conceptually the model could represent a stress concentration. 2). with spatially uniform event coverage is obviously not realistic. Further. The resulting image contained identical results for each horizontal layer in the model. events in the high velocity zone were not originally mislocated. Notice that due to the symmetry of the model. parameters to a precision of within 0. Figure 3 shows a plan view of a horizontal layer in the resulting velocity image. to less than 0. simple constant velocity source locations were biased for most events. However. No further reproductions authorized.019 after the inversion. Figure 5 shows a plan view of the velocity model produced by two Copyright by ASTM Int'l (all rights reserved). or horizontal slice perpendicular to the z-axis. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The 50 events were located at the center of each cell boundary. mislocations would be relatively larger for larger velocity variations. events were confined to the center of the cell boundaries on each side of the high velocity y13 layer.019. a second example is documented with restricted event coverage. of the velocity model used in the synthetic study. 90 ACOUSTICEMISSION FIGURE 2 Plan view. . inducing localized failure along forming or preexisting microfractures.

to locations determined by passive imaging. . Copyright by ASTM Int'l (all rights reserved). No further reproductions authorized. MAXWELL ET AL. t t 1 t t t f ~ t f FIGURE 4 Plan view of event relocations. corresponding to the velocity model in Figure 3. O N V E L O C I T Y T O M O G R A P H Y 91 FIGURE 3 Plan view of the passively imaged velocity structure with spatially uniform event coverage. Arrows point from constant velocity half-space locations. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.

92 ACOUSTICEMISSION FIGURE 5 Plan view of the passively imaged velocity structure with spatially restricted event coverage.5~ in the high velocity zone resulted. this study simply documents the development and testing of a proposed technique to passively image the velocity structure of a deformed sample. where a crystalline rock is injected with high Copyright by ASTM Int'l (all rights reserved). . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. so that errors of about 0. REAL DATA As previously mentioned. Compared to the first example. iterations of the simultaneous inversion method. the ray coverage is not uniform. which is representative of the average resolution. The image of the z-2 layer is shown in Figure 5. The data set will consist of arrival times measured from whole waveforms digitized on 16 channels. No further reproductions authorized. so that some cells were not transversed by a single ray. In this example there existed some subtle variation between x-y layers. an expanded version of the system described in reference [i]. The data will be collected on a hydraulic fracturing experiment in the laboratory. This caused the matrix OA to be singular. and damping had to be applied to quell the influence of near zero singular values. The damping caused some loss of resolution. The next phase of the project will be to invert a real data set.

In order to quantify the accuracy of the inverted parameters. This suggests that in stress monitoring using sequential passive imaging. for supplemental information of the physical state of the rock. and to map stress related changes in the rock. Further. These further analyses could include determining anisotropy effects and attenuation imaging. MAXWELL ET A L ON VELOCITY TOMOGRAPHY 93 pressure fluids. No further reproductions authorized. a more significant contribution to the errors in the arrival time data results from the non-zero rise times. Concurrent monitoring of the induced acoustic emissions and active imaging of the rock sample will be performed. straight ray theory has been applied. the covariance matrix should be examined. For ultrasonic frequency propagation in crystalline rocks. Iterative inversion steps with intermediate ray tracing are then required to converge to the correct velocity structure and ray trajectory. will now be discussed in relation to the real data set. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. These errors ultimately control the minimum velocity contrast that can be imaged. for more pronounced velocity contrasts ray tracing methods are required to account for ray bending effects. and more susceptible to errors. The finite rise times of real events increases along the raypath as energy is attenuated. CONCLUSIONS From the second example. the resolution matrix should be examined to access the significance of the inverted parameters [6]. the anomaly could still be reproduced. it is interesting to note that with the actual event locations spatially restricted to the velocity anomaly. the relative significance of which varies with the actual size of the fracture surface. However. However. The locations of the induced activity will be coupled to zones of anomalous stress states. The simplified assumptions of the source characteristics and the wave propagation made in the synthetic data set. Directional variations in the radiation pattern of the event will tend to increase the probability of arrival time picking errors as the signal-to-noise ratio decreases for raypaths close to nodal directions. Errors exist from the hypocenter point source location approximation to the finite sized event. In these synthetic models. Passive imaging of the recorded arrival time data of the activity should be able to resolve these spatial velocity variations. the evolving effects of the stress field can be imaged by monitoring the induced acoustic emissions. to validate the passively imaged velocity structure. The effect of arrival time errors on the velocity images will be to degrade the accuracy of the imaged parameters. the isotropic velocity image may then be used for more extensive data analysis. and hence velocity variations. This makes the actual arrival times difficult to measure. The results of a passive imaging calculation of the monitored acoustic emissions can then be compared to the results of a conventional active image. Further. More accurate ray directions also improve the accuracy of moment inversions and fault plane solutions [6]. Copyright by ASTM Int'l (all rights reserved). dispersion along the raypath results in long rise times. .

American Society for Testing and Materials... Seismic Tomography. "Acoustic Emission Analysis and Ultrasonic Velocity Imaging in the Study of Rock Failure. passive imaging provides information on the spatial variation of the stress sensitive velocity structure. P. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.. and Hutchins. A. G." Proceedings of the Second International Symposium of Rockbursts and Seismicity in Mines. [2] Young. R. [4] Spencer. S. 1987.. No further reproductions authorized. A. D.. [3] Young. pp 122-141." Acoustic Emission i Current Practice and Future Development~ ASTM STP 1076. Reidel Publishing Company. 1980." Geophysical Journal of the Royal Astronomical Society. Hutchins. Chow. Young. T. and Gubbins D. pp 122-141. [5] Nolet. REFERENCES [i] Falls. Vol. Sasche. R. 94 ACOUSTICEMISSION Thus. J. J.Ed. where monitors of arrival times of acoustical activity are already in place.. W.." Proceedings of the Second International Symposium of Rockbursts and Seismicity in Mines. . Minnesota. Minnesota. "Seismic Structure and Tectonics of Kilauea Volcano. H. "Travel Time Inversion for Simultan- eous Earthquake Location and Velocity Structure Determination in Laterally Varying Media. D. more accurate source locations. Philadelphia.. 1350. C... United States Geological Survey Professional Paper. Eds. R. D. 63. Financial support was provided by NSERC and Queen's University. "Seismic Imaging Ahead of Mining in Rockburst Prone Ground. No. 1988. C. "Geotomography in the Study of Rockbursts and Seismicity in Mines. ACKNOWLEDGEMENTS The authors gratefully acknowledge the assistance given by other members of the Queen's Rock Physics and Engineering Seismology Laboratory. and McGaughey. P. P. pp 95-116.. K. and ray directions for source mechanism studies. Roget. and W. 1988. Dordecht. D. Copyright by ASTM Int'l (all rights reserved). [6] Thurber. 1987. 1990. This method is particularly applicable to laboratory acoustic emis- sions and mining-induced seismicity. pp 919-934. Yamaguchi.

J. acousto-ultrasonics is not concerned with source location and characterization. 95 Copyright 9 1991 b y ASTM I n t e r n a t i o n a l ~. and variations of mechanical properties in fiber reinforced composites and composite-like materials [i]. adhesive bonds. These are factors that both singly and collectively influence acousto-ultrasonic measurements that. cable and rope. Sachse. W. . "Acousto-Ultrasonics . strength. The potentials and limitations of the technique are reviewed. Instead. adhesive strength The acousto-ultrasonic approach uses simulated stress waves and stochastic wave propagation to detect and assess diffuse defect states. acousto-ultrasonics deals primarily with the assessment of the integrated effects of diffuse defect states. American Society for Testing and Materials. paper and wood products. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. and alsc human bone. Basic methods and guidelines are discussed. therefore.An Update. Philadelphia. composites.." Acoustic Emission: Current Practice and Future Directions. thermo'mechanical degradation. 1991. OH 44135. stress waves. Roget.. and populations of subcritical flaws. correlate with mechanical Alex Vary heads the nondestructive evaluation group at the National Aeronautics and Space Administration's Lewis Research Center.like materials. Cleveland. Yamaguchi. The term acousto-ultrasonics denotes a combination of some aspects of acoustic emission signal analysis and ultrasonic materials characterization methodology. astm. Alex Vary ACOUSTO-ULTRASONICS: AN UPDATE REFERENCE: Vary. KEYWORDS: acousto-ultrasonics. damage conditions. Eds. org Copyright by ASTM Int'l (all rights reserved). 21000 Brookpark Road. The underlying hypothesis for acousto- ultrasonic materials evaluation is stated and needs for theory and signal analysis development are indicated. Acousto-ultrasonics has beer used to evaluate fiber reinforced composites. Unlike the usual acoustic emission practice. No further reproductions authorized. ASTM STP 1077. A. stress wave factor. and K. ABSTRACT: The acousto-ultrasonic technique was devised to assess diffuse flaw populations and any associated changes of the mechanical properties of composites and composite.

and structural damping measurements. Acousto-ultrasonics was seen as a means for assessing factors that influence strength properties in principal load directions in these structures. Moreover. These are in addition to more directly related techniques like acoustic emission. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.. The term "acousto- ultrasonics" was coined to describe the close relation with acoustic emission and the internal acoustics of solids [6]. dynamic response. . the objective should be the detection of global collections of subcritical flaws and material anomalies that diminish strength. etc. guidelines for making acousto-ultrasonic measurements are given along with limitations of the technique. Another incentive was the need to go beyond detection of overt flaws like cracks. pulse-echo. This chapter reviews recent experimental findings and progress in the application of acousto-ultrasonics. BACKGROUND The seminal idea for the acousto-ultrasonic approach was derived from the work of Egle who investigated stress wave simulation using various excitation methods [2]. toughness. The acousto- ultrasonic technique has been demonstrated to be highly sensitive to interlaminar and adhesive bond strength variations. It has been shown useful to assess micro- porosity and/or microcracking produced by fatigue cycling. The original incentive for introducing the acousto- ultrasonic approach was the recognition of nondestructive testing needs peculiar to composite structures such as fiber reinforced composite laminates and filament-wound composite vessels. "in- plane" parallel to fiber directions in composite laminates.g. 96 ACOUSTICEMISSION property variations and dynamic response. e. Acousto-ultrasonics belongs to a class of techniques that includes (coin) tap testing. Some current concepts of wave propagation modes that form the basis of the acousto-ultrasonic approach are reviewed. Sonic (coin) tap testing can be considered a primitive version of acousto- Copyright by ASTM Int'l (all rights reserved). dynamic resonance. Vary and his colleagues [4. delaminations or inclusions. and guided Lamb wave methods. Also.5] advanced the notion of using ultrasonically simulated stress waves and acoustic emission signal analysis methods to evaluate defect states and material properties. Egle and his colleagues were interested in simulating acoustic emission stress waves with the idea of improving methods for the characterization of acoustic emission signals and sources [3]. it was apparent that if a test piece has no dominant flaws. stiffness. No further reproductions authorized. The idea was to deal with large populations of minute flaws where it is both impractical and unnecessary to image each individual discontinuity.

An acoustic emission system in its elementary form constitutes half of an acousto-ultrasonic system: passive listening but no active interrogation. a sender and receiver. HYPOTHESIS The general objective of acousto-ultrasonics is to rate the relative efficiency of stress wave propagation in a material. The working hypothesis is that more efficient strain energy transfer and strain redistribution during loading corresponds to increased strength and fracture resistance in composites.. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. For many fiber reinforced composites lower attenuation means better stress wave energy transfer and hence. the transducers in the leaky Lamb wave approach interrogate through water in a immersion tank. like acousto-ultrasonics. This hypothesis is based on the "stress wave interaction" concept which holds that spontaneous stress waves at the onset of fracture will promote rapid Copyright by ASTM Int'l (all rights reserved). Using angle beams and immersion.g. utilizes two transducers. The leaky Lamb wave approach. VARY ON ACOUSTO-ULTRASONICS 97 ultrasonics. the leaky Lamb wave approach excites waves that are similar to "in-plane" waves generated in plates by the acousto-ultrasonic approach. this comprises the first step toward an acousto-ultrasonic configuration. the waves are affected by the same microstructural and morphological factors that determine material quality. It is worth considering the adoption of the idea that an acoustic emission system benefits from including pulsed transducers that help interrogate the condition of the material on which passive sensors are stationed. electric sparking. The combination of an acoustic emission sensor with an active pulser forms the basis for an acousto-ultrasonic system. The basic attribute measured is stress wave energy loss. and also with piezotransducers. This combination can in theory be used to continually check changes in the background accumulation of material damage and diffuse flaws against which acoustic emission events are monitored. Conceptually. No further reproductions authorized. e. by scatter attenuation. . Calibration of acoustic emission sensors is conventionally done with lead pencil or glass capillary breaks. In either case. except that with acousto-ultrasonics "tapping" is usually done with a piezotransducer while "listening" is done with a second piezotransducer. Instead of being attached to the test piece as in acousto-ultra- sonics. better transmission and distribution of dynamic strain energy. The true level of spontaneous activity in the material might be then taken as a function of the ratio of the acoustic emission rate versus the acousto-ultrasonic energy transmission level.

the acousto-ultrasonic approach requires that the received signal be the result of Copyright by ASTM Int'l (all rights reserved). But. prompt and efficient dissipation of stress wave energy away from crack nucleation sites is needed to assure that the energy is not focused or localized in a way that causes catastrophic fracture. as in flaw detection. changes wrought by thermo-mechanical degradation can be measured. For composites and adhesive bonds low attenuation will usually indicate high strength and impact resistance. and similar factors will affect stress waves and material response to stress wave interactions. Instead of well-defined wave reflection and propa- gation paths. further work is needed to establish the above-stated hypothesis in connection with crack nucleation and damage accumulation mechanics in composites. . No further reproductions authorized. for example. There are currently no entirely satisfactory micro- mechanical or constitutive models for describing inter- actions of ultrasonic stress waves with factors that govern mechanical properties and dynamic response. and matrix void content. Or. In lieu of these mechanisms. A pulsed sending probe is optimized for wave generation while the receiving probe is optimized for signal sensing. Better understanding is needed of the connection between the attenuation of the benign ultra- sonic waves used to interrogate composites and the attenua- tion of spontaneous stress waves during loading. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Experimental results have confirmed the ability of the acousto-ultrasonic approach to measure relative stress wave energy dissipation properties of composites. Acousto-ultrasonic measurements can also be applied to predict loci where failure might b~gin or h o ~ a material sample will react to loading. plastic deformation or microcrack deflections [7]. fiber-matrix bond quality. METHODOLOGY Acousto-ultrasonics is a generalized approach to ultra- sonic testing that uses a pair of ultrasonic piezoelectric probes in a send-receive configuration. Appropriate models should make it possible to specify the degree to which elastic properties. Acousto-ultrasonic interrogation can be used to assess morphological states just before or at the onset of frac- ture. fiber fraction. One very important implication of the hypothesis is that the wave attenuation properties of a material are pivotal. The usual and often most convenient arrangement is to have the probes on the same surface of a test piece. 98 ACOUSTICEMISSION macrocracking unless their energy is dissipated by other mechanisms. Acousto-ultrasonics differs from conventional ultra- sonic methods primarily in the nature of the received signal.

But. This produces stochastic wave propagation which is actually desirable because it ensures repeated wave interactions with key microstructural features and diffuse flaws. size. No further reproductions authorized. scattered and mode-converted (longitudinal and transverse) signals will arrive at the receiver.g. pure plate or Lamb waves are unlikely because (I) wave generation is not long duration. properties to be assessed. The receiver must be very sensitive and have a bandwidth covering all frequencies passed by the sample. This is to assure that there are multiple wave interactions with diffuse flaws and micro- structural features within the test volume. It has been observed that composite materials modulate and bandpass filter the acousto-ultrasonic signals. damage states. and types of flaws. In these cases it has been Copyright by ASTM Int'l (all rights reserved). signals running parallel to the fibers in a unidirectional composite panel may exhibit a bi-modal frequency spectrum: (i) a low frequency component because of guided plate waves and (2) a high frequency component because of transmission by fibers. single frequency but broadband pulsing. test piece (e. The signals arriving at the receiver probe often have spectra that indicate selective transmission of guided wave modes. shape). (2) lamina and fiber-matrix inter- faces will break up the wave propagation paths. ANALYTICAL In acousto-ultrasonics quantification of information contained in the simulated stress wave signals is done by calculating a stress wave factor (SWF). For example. VARY ON ACOUSTO-ULTRASONICS 99 multiple reflections and interactions with material micro- structure in a volume of material between sending and receiving probes. . thick- ness. Selection of the sending transducer center frequency and the receiving transducer bandwidth and sensitivity are pivotal in acousto-ultrasonics. This may also entail high ultrasonic frequencies that are quite strongly attenuated. For composite panels it is advantageous for the sender to produce wavelengths that are less than the thickness. In these cases the wavelength will be comparable to or less than major dimensions of the sample. for example. Bulk and guided waves and several types of plate waves can arise during acousto-ultrasonic testing [8]. There are many ways to define and calculate the SWF.. The definition of the SWF in any particular case depends on the nature of the signal. Acousto-ultrasonic signals often resemble acoustic emission "burst" waveforms. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. material. Flexural Lamb waves will tend to dominate in composite panels. In composite panels this means that overlapping.

Cepstrum analysis offers further means for signal analysis and SWF quanti- fication. 1O0 ACOUSTICEMISSION found practical to use signal analysis methods based on acoustic emission practice. and guided modes. second and third moments. the best correlation between the SWF and a property or flaw condition will be obtained by measuring the least dominant rather than the most dominant part of the signal. bulk. time domain quantities such as rise time. However. In the future. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized. ringdown counts and root- mean-square voltages. and root-mean-square voltage values tend to correlate with defect states and material properties. These involve power spectrum analysis and spectral distribution functions. plate. for example. they have usually been based on somewhat unsophisticated signal analysis methods. In many instances these simple definitions for the SWF suffice. One underexplored area is that of diffuse field theory for dealing with signals where phase information has been lost due to stochastic scattering and multiple reflections. advanced methods using inverse scattering theory. This is because guided wave modes at specific frequencies may correspond closely to particular elastic and mechanical properties. By selecting only certain portions of waveforms or spectra it is possible to greatly improve correlations between the SWF and specific material properties. The effects of interplay of these modes need to be evaluated by appropriate analytical methods. It is important to note that the SWF should not always be defined in terms of the entire waveform or spectrum. Experimental observations indicate that there are likely to be intermixed diffuse. homomorphic and diffuse field analysis will undoubtedly emerge as tools for dealing with acousto- ultrasonic signals [9-11]. Although a number of approaches for quantifying the acousto-ultrasonic SWF have been successful. voltage decay or ringdown count. For example. further work is needed to identify SWF quantification methods that relate directly to composite integrity. and thermo-mechanical degradation effects. Statistical analysis of spectral mean values. . Moreover. fiber-matrix bonding. peak voltage. Acousto-ultrasonics certainly has limitations common to all Copyright by ASTM Int'l (all rights reserved). and skewness can also form a basis for defining the SWF. Possibly. LIMITATIONS Acousto-ultrasonics has the capabilities and also the limitations associated with other ultrasonic techniques. further studies are needed to specify and measure the SWF in terms of specific wave propagation modes that arise under various material conditions and geometries.

. affect the material characteristics revealed [12. In addition to sample thickness and material variations all acousto-ultrasonic signals are affected by a number of factors associated with transducer-specimen attachment: applied pressure. The increased sensitivity gained imposes penalties because surface roughness effects. However. transducer alignment. This requires scanning by lifting and recoupling the transducers or by using probes that can scan while remaining in contact with the surface. Acousto- ultrasonics has also been successfully accomplished using probes encased in a squirter fixture where water jets couple the probes to a part and allow free scan movement. Using either laser beams or the water jet approach. the technique still requires a constant material thickness for comparative measurements of material property variations or defect states. that is. interferometric) detection of stress waves. The use of laser ultrasonics is being studied as an alternative to methods that rely on contact. the possible need for sacrificial layers. There is an essential difference with respect to ultrasonic techniques where the objective is only to detect the presence of overt flaws. Signal validity and reproducibility in acousto-ultrasonics require the utmost care in the test set-up and procedure. acousto-ultrasonics becomes vulnerable to material vagaries that affect probe sensitivity and signal reproducibility.. No further reproductions authorized. and the laser-optic (i.. resonance frequencies. i. bandwidth. the specimen surface roughness. type and amount of couplant. the spacing and location of sending and receiving transducers. There are further problems after the previous factors have been dealt with. probe alignment and coupling. By going beyond flaw detection. the Copyright by ASTM Int'l (all rights reserved). measurements must be repeated to map material variations. usually with dry coupling. noncontact acousto-ultrasonic methods should be able to accommodate the scanning of large areas and curved sur- faces. The latter can be accomplished by transducers stationed in the hubs of elastomer wheels that are rolled over the test surface. internal damping. Although quite capable of detecting overt flaws. Properties of the transducers. and so on. VARY ON ACOUSTO-ULTRASONICS 101 techniques that attempt to measure either absolute or rela- tive attenuation and velocity. potential laser damage.13].e. and inherent sensitivity. Non-contact laser methods have problems associated with thermal-to- acoustic energy conversion. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. In practice. by attempting to characterize defect states and subtle material anomalies. and associated factors impose variations on the signals transmitted by the bulk material. To assure stochastic wave propagation and to sense the collective effects of minute discontinuities requires using the highest possible ultrasonic frequency bands that can be transmitted and sensed.e.

acousto-ultrasonics is proving to be a valuable Copyright by ASTM Int'l (all rights reserved).24]. 102 ACOUSTICEMISSION acousto-ultrasonic approach was devised primarily to assess the integrated effects of diffuse flaw populations. With water jet fixtures and laser beams the spacing may be reduced. polymer composite cure state. Applications that demand resolution of small isolated discontinuities are not readily handled by the technique. The technique has been used to measure variations of the interlaminar shear strength of composites. As in the case of acoustic emission. impact damage. most of the uses reported to date have been with laboratory samples. This is determined in part by the size of the probes. Acousto-ultrasonics does not lend itself to high re- solution flaw imaging because of inherent constraints on transducer positioning and spacing. There are now numerous examples of successful and useful applications of acousto-ultrasonics. wire rope strength. Acousto- ultrasonics currently appears to be an excellent technique for ranking fiber composite laminates and adhesive joints according to thei~ strength variations [23. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. the spacing must not be so small that the volume of material interrogated does not allow a large number of multiple wave interactions. fatigue and impact damage can be closely measured with the acousto-ultrasonic stress wave factor [19-22]. Generally. Other techniques such as pulse-echo or through transmission C-scanning are likely to be more appropriate for high spatial resolution flaw detection and imaging. and thermo-mechanical degradation. APPLICATIONS The sensitivity of acousto-ultrasonics for detecting and measuring subtle but significant material property variations in composites has been amply demonstrated [i]. A typical minimum probe spacing may be approximately one centimeter. This latter attribute is primarily due to in-plane shear waves generated in bond zones. But. and adhesive bond strength [14-18]. filler content in wood and paper products. Micro-cracking. Among the most useful aspects of acousto-ultrasonics is the ability to locate weak areas and to assess degradation and diffuse flaw states in composites. CONCLUSION Although practical field applications of acousto-ultra- sonics exist. No further reproductions authorized. the tech- nique demands that a relatively sizable volume of material be examined for each position on the test piece. .

[i0] Karagulle. M. Materials Evaluation.. S. llth Sym- posium on Nondestructive Evaluation. E. [ii] Weaver. The Netherlands.. 1977. 8. 2. F. 134-152. Materials Evaluation. Rajapakse. M. 476-480. [14] Kautz. No. R.... and Bowles. VARY ON ACOUSTO-ULTRASONICS 103 tool in materials research. and Lee. 67-70. and McFarland. A. No. H. J.. pp. 40. and Phillips. Vol. and McFarland. M. H. M. British Journal of Non-Destructive Testinq. 3-10. J. Eds. M. No. 65. i. D. H. Solid Mechanics Research for Quantitative Non-Destructive Evaluation (J. Vol.. Achenbach and Y. Journal of Testing and Evaluation.. 185-191. 1198-1203. NDT Inter- national. May Copyright by ASTM Int'l (all rights reserved). Vol. pp. M.. 1988. [17] dos Reis. pp. The Netherlands. 47. pp. No. [6] Vary. Vol. 1988. No. Vol. 28. No. Ed. [16] dos Reis. [12] Bhatt. Journal of Acoustic Emission. American Society for Non-destructive Testing. [4] Vary. 6. and Lark. Plenum Press. K. G. E. 928-934. 1986.." U. 3. D. R.. Williams. No further reproductions authorized. 1979. No. 426-434. . 1985. Materials Evalua- tion. Dortrecht. J. D. pp. 12. L. Journal of Testinq and Evaluation. 1979. This can help gauge material changes that affect acoustic emission signal characteristics and also assess concomitant material damage and degradation. 1446-1454. J. B. pp. Rajapakse. [7] Vary. Dortrecht. Proceedings. 1987. Patent 4.. P. Duke. E. 21. New York. M. 247-257. pp. No... H.. "Control System for Processing Composite Materials. D. The under- standing gained during these test activities can aid in applying acousto-ultrasonics to the evaluation of the integrity and service life of structural composites. pp. L. 4. 650-654.) Martinus Nijhoff Publishers. Journal of the Acoustical society of America. M.. 7. 1987. 1981. i. Vol. 1984. [13] Russell-Floyd. pp. Vol. [3] Egle.268.. 242-258. 45. 1404-1412.. II. [9] Egle.. pp. No. Acousto-ultrasonics should be considered for in-situ monitoring during proof testing and materials tests that are usually monitored by acoustic emission.. Journal of the Acoustical Society of America. 43. [15] Hinricks R. 1989. Vol. 1987. 21. pp. 1976. G. S. pp.455. 1988. Solid Mechanics Research for Quanti- tative Non-Destructive Evaluation (J. Vol. 28. pp. No. Jr. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. and Brown. M. Materials Evaluation. 196-199. Achenbach and Y. [5] Vary. 1982. H. and Hogg. 4.. [8] Tang.. NDT International. L. 4. 70. Eds. D... D. R.Theory and Applications. M. Vol. 5. Jr. Vol. A. and Henneke.) Martinus Nijhoff Publishers. C. Vol. A. A. and Thuen J. J.S. D. [2] Egle. REFERENCES [i] Acousto-Ultrasonics . A.

and Prakash. T.. England. New York. 233-240... Y.8th Conference. pp. No. Composite Materials: Testinq and Des~qn . pp. pp. Mechanical Engineering. Yuceoglu and R. 1988. Henneke. Lee. Eds.Theory and Applications. 2. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. C. 311-321.. [19] Lorenzo. Canada. Composite $ ~ u c - tures. L. 1989.. 1988. Duke. 155-156. Vol. A. Advances in Aerospace Sciences and Enqineerinq. Tanary. Philadelphia. pp. G. Hesser. Mater- ials Evaluation. and Lampert. Copyright by ASTM Int'l (all rights reserved). STP 972. pp. S. Jr. 177-190. 55-60. S. N.. J. J. [18] Tanary. A. R.). 1502-1510. . K. 1980.. and Haddad. (U.. AGousto-Ultrasonics . Elsevier Applied Science Publishers. V. 380-397.. pp.. 1987. [20] Govada.. Vol. [22] Talreja. Vol. 1984. H. E. PA. S.. Ed. 1988. and Hahn. Amer- ican Society for Testing and Materials. and Talreja." Thesis. No further reproductions authorized. "Characterization of Adhesively Bonded Joints Using Acousto-Ultrasonics. pp. [23] Srivastava. 38. University of Ottawa.. R. No.. Materials Evaluation. [21] Williams. 8. [24] Fahr. R. American Society for Mechanical Engineering. 104 ACOUSTICEMISSION 1986. 47. R. 12. Plenum Press.

The original premise was that a section of material which transmitted the most energy from the sending transducer to the receiving transducer would transfer loading stress more e f f i c i e n t l y . ~991. . using through-thickness-transverse-resonance (TTTR)." Acoustic Emission: Current Practice and Future Directions. associated theory.. VPI&SU. Philadelphia. J. is an engineer at GE Nuclear Fuels and Component Manufacturing. Kiernan and John C. Jr. ASTMSTP 1077. T. Sachse. PO Box 780. W. Lamb wave theory. THEORETICAL BASIS OF THE ACOUSTO-ULTRASONIC METHOD REFERENCE: Kiernan. Additionally. Major emphasis is given to building a model of the AU method. American Society for Testing and Materials. J. and K. early research showed that AU energy measurement decreased as a function of fatigue cycles [2]. composites. physical interpretations. Kiernan. Duke is a professor of Engineering Mechanics. VA 24061-0219. Finally. Yamaguchi. No further reproductions authorized. Dr. C.org Copyright by ASTM Int'l (all rights reserved). KEYWORDS: acousto-ultrasonic. conclusions are presented identifying the modes of wave propagation influencing the AU measurements. Michael T. "Theoretical Basis of the Acousto-Ultrasonic Method. The AU method involves introducing a mechanical excitation at one position on a material surface and sensing how the disturbance is transmitted through the material by using a second receiving transducer at another position on the same material surface. Wilmington. a recent graduate of Virginia Tech. Duke. Lamb waves Composite materials develop complex damage states which demand practical nondestructive evaluation (NDE) techniques. The AU method is one of many NDE techniques under development for ascertaining the extent and mechanical effects of damage in composite materials. The paper begins with a brief introduction to the AU method and experimental results. Blacksburg. Jr. and elasticity. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. M. M/C K-51. Eds. 105 Copyright 9 1991 by ASTM International www. Roget. and hence be less prone to fracture.astm. This was successfully used to predict strength of composite specimens [ I ] . and comparisons between theory and AU results. NC 28402.. ABSTRACT: This paper provides a physical interpretation and theoretical basis for acousto-ultrasonic (AU) results. and Duke. Dr.

No further reproductions authorized. Following early AU experimental successes. including advance methods of signal analysis [3-5]. the same type of transducer was used for both the sending and the receiving transducer. iv. normal mode.25 cm diameter. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The method may be implemented inexpensively. as well as develop the means for material characterization. piezoelectric transducer was used for the sending transducer. Models of the AU method will help guide future AU research and development. A diagram of the AU set-up used in this work is shown in Fig. i i . Copyright by ASTM Int'l (all rights reserved). the AU method offers certain practical advantages for implementation: i. relations to material condition [8. and some work on developing a physical understanding of AU results [10.7]. 106 ACOUSTICEMISSION Beyond the proven experimental successes. iii. I. For the specimens utilized. 2. new applications [6. Wavemotion is directed in the plane of the plate. research on the AU method branched out in a number of different directions. A 1. in order to interact with the material in the direction of loading. AU RESULTSAND THEORETICAL RELATIONS Figure 2 shows the voltage/time plot obtained by performing the AU method on a 1. Requires access to only one surface of the material. this met Vary's original criterion of having the resonant frequency of the transducer relating to a wavelength on the order of the material thickness [ I ] .59 mm thick aluminum plate with the sending transducer approximately 25 mm from the receiving transducer. However. For the results described in this work.11]. an understand- ing of AU results for composite plates is s t i l l incomplete and no models exist for composite plate characterization. for the purpose of developing a refined AU technique and developing a basic understanding of wave propagation in composite materials. The purpose of this paper is to provide a physical understanding of the AU method and state the associated theory. The signals were acquired and analyzed using a personal computer incorporating an analog to digital converter board. which may be of use in other areas.25 MHz center frequency. Each component in the diagram has an effect on the results [12]. Rapid measurements are possible. wide band. since experimental results and the physical phenomena causing those results are different for different experimental arrangements. EXPERIMENTAL SET-UP Understanding AU results requires giving attention to experimental set-up and procedure.9]. . such as acoustic emission.

Schematicdiagramof the experimental set-up.-~ \-(CompositePlate) FIG. No further reproductions authorized./1 . 2 -. . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Preamplifier /r~ //. KIERNAN AND DUKE ON THE ACOUSTO-ULTRASONIC METHOD 107 v~t (Trigger Pulse) IBM-PC PanametricsPulser I I(t) Gate L[Signal to be~ vct) IR(t) I\ ~ (InputPulse) F(Received Signal) X~ I~-. FIG.Voltage versus time plot of an AU signal from a 1. .. I -.59 mm thick plate of aluminum..ndingTransducer~ " I [__L~-ReceivingTransducer I I ~ D~. Copyright by ASTM Int'l (all rights reserved).

The frequency and displacement equations for these modes are easy to calculate [13] and are given below: Symmetric Modes (a) us:B'sin(mxx3/2b) u1:0 m:I. i t was determined that the frequencies of resonant peaks were inversely dependent on plate thickness [12]. No further reproductions authorized. The long wavetrain would suggest that the AU signal is the result of some reflection pattern and not the result of a plane wave propagating between the sending and receiving transducer. . Fig. . . . 3.3 . (b) u1=C'cos(nxx3/2b) u3=O n=2. . 3 -. . Next. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.59 mm thick plate of aluminum. (b) u3=A'cos(m~x3/2b ) u~=O m=2. usually relate to resonant phenomena. .Amplitude versus frequency plot of a signal from a 1. 3x10 I ZxlO' IxtO" " FREQUENCY ( k ~ Z ) Fig.4 . Antisymmetric Modes (a) u~=D'sin(n~x3/2b) u3=O n=1. . . FFT data is provided in a plot of amplitude/frequency.3. The theory of through-thickness-transverse-resonance (TTTR) pre- dicts the frequencies associated with plane wave resonances between the top and bottom surfaces of a plate. Distinct peaks in the frequency response for physical systems. i t should be noted that definite harmonic components exist in the wavetrain and that these components tend to be modulated. . . Peak amplitudes occur at i and 2 MHz.7 MHz. with a smaller peak at . By comparing the amplitude versus frequency plots for plates of different thicknesses.4 . In order to obtain more information. TTTR is a l i m i t i n g case of the more general modes of Lamb wave propagation for plates. 108 ACOUSTICEMISSION The wavetrain is quite long (>150 microseconds) compared to the short pulse length (5-i0 microseconds). . Copyright by ASTM Int'l (all rights reserved). .

. AU results tend to indicate that for a given plate.0 0.0 2 2.5 Vs=(C44/p).0 7. Table 2 -.2. Similarly. In addition. respectively.8 3 5. These measurements were made for each of the harmonic components noted in the spectrum. Table I shows the predicted TTTR frequencies for the aluminum plate discussed previously.Cutoff Frequencies for 1. Basi- cally. The . as listed in Table I.Phase Velocity Measurements (I. For materials where the x~-axis is not in a plane of orthotropic symmetry. the value for the frequency at which peaks occur in the amplitude/frequency plot agrees with calculated TTTR values.0 >25. while the I and 2 MHz signal content showed group velocities at roughly 1500 meters per second. the 2 MHz peak agrees with the predicted value for both the f i r s t and second order symmetric resonances.6 5 9.5 5. Also. additional resonances would be expected.0 3. the shear wave v e l o c i t i e s would be affected by the coupling terms in the stiffness matrix.0 1 i . 3 agrees with the predicted value for the f i r s t order antisymmetric resonance. N: 1.0 15. No further reproductions authorized. estimates of group velocities were made by noting the speed of the modulating components.9 i .7 MHz signal content showed a group velo- city that was the same value as the phase velocity.33 1.59mm thick Al Plate) Frequency (MHz) Phase Velocity (km/sec) .0 Phase velocity measurements were made in order to gain more infor- mation on the modes of wave propagation.59mm Al Plate Symmetric Mode Ant i symmetric Mode Number Frequency (MHz) Frequency (MHz) 0 0. phase velocity versus frequency curves. this theory can be utilized to calculate Copyright by ASTM Int'l (all rights reserved). It can be used to derive dispersion curves. and E and u are the Young's modul~s and Poisson's r a t i o .13. The theory which governs wave propagation in plates is well developed [11. the resulting values are displayed in Table 2. . and to estimate group velocities.3 .0 4 4. TABLE I -.21]. C i i ' s are the stiffness matrix values. I f C44 does not equal Css. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.7 5.5 or (C5~/p)-5 Css=(1-u)[E/(1+u)(1-2u)] C44=C55=[E/2(I+u)] for isotropic materials where t is the plate thickness. Notice that the I MHz peak in Fig. . KIERNAN AND DUKE ON THE ACOUSTO-ULTRASONICMETHOD 109 Frequencies and Velocities fp=Nvp/2t fs=Nvs/2t Vp:(C3s/p) .4 It should be noted that the I and 2 MHz signal content showed phase velocities much higher than the values for plane waves in aluminum.7 3.75 2.14. . .

.............. Copyright by ASTM Int'l (all rights reserved)....8 ............. The Rayleigh/Lamb approach involves expressing the displacement vector as a scalar and vector potential and then solving for the potentials in the equations of motion by the separation of variables method...................... for the SV-wave.... Two basic approaches can be used to derive the equations for plate waves..... These can be expressed by the scalar potential......................................................... For anisotropic materials........ then ~:~oei(~x~• 2) H 2 = H o e i ( ~ x 1 • ....4 .........5 WAVE N U M B E R FIG................... the potentials are combined in an appropriate manner to satisfy the boundary conditions (stress free) at the top and bottom surfaces of the plate....... i f consideration is given only to the propagation in the x1-direction (see Fig..................... [14]........................ waves are assumed to r e f l e c t back and forth between the top and bottom surfaces of the plate.............................................................. 0 I I i i l i O 5 10 1.......2 0............... 0................................. 4 -.. 0........... and has been extended to calculate transient plate responses for an isotropic material [15]. In the partial wave method...... Next........................................................ the method of partial waves is more convenient method of solution than the classical approach...... Also.............. ~.... for the pressure wave and one vector potential.......2 .................................. 0.......... Dispersion curve for the Ist order antisymmetric mode for AI..............5 30 3...... Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.............................. The other more recently developed approach is the method of partial waves or transverse resonance [14]..... I)... PLATE WAVES IN ISOTROPIC MATERIALS Rayleigh-Lamb waves are produced by the introduction of a pressure wave (P-wave) and a v e r t i c a l l y polarized shear wave (SV-wave) to the plate.......... These waves are found by establishing the condition for transverse resonance through the thickness of the plate.. No further reproductions authorized........ ~2=v s 2 ( # 2 + ~ ) FREQUENCY (MHz) 1... The f i r s t is the more conventional Rayleigh/Lamb wave analysis................ as pointed out by Solie and Auld............................................... These combine to form traveling waves in the direction of wave propagation and stand- ing waves in the thickness direction of the plate............................. H2... ...5 20 2.............. 110 ACOUSTIC EMISSION displacement and stress fields.... Specifically.......... this method highlights certain physical features of plate wave propagation........................................6 ........

Additionally. are r e l a t i v e l y low. For the symmetric mode. these modes are characterized by rather high phase velocities and rather slow modulating (group) velocities. the curves turn upwards at larger slopes) with lower associated phase velocities. However as a general trend. including the displacements. indicating that the group velocity or rate of energy flow is r e l a t i v e l y low. These modes correspond to where the higher order modes cross the ordinate (~-axis) on the dispersion curve (see Fig. the f i r s t order symmetric curve has a negative group velocity. Thus. this can be described as a combination of motions which are symmetric and antisymmetric about the mid-plane of the plate. Moreover. these curves start with very high phase velocities and rather low group velocities and move to higher group velocities "(ie. and D are defined by @=(Asin~x~+Bcosex~)ei(~x~ -~t) and H~=i(Csin#xs+Dcos#x3)ei(~x~-~t) By examining the slowness surfaces for the SV and P waves in conjunction with the stress free boundary conditions. For a plate with traction free top ~nd bottom surfaces. where the stress components can easily be calculated using the constitutive relations. The manner in which the phase velocities drop from i n f i n i t y for the higher order modes.d i r e c t i o n . can be seen in Fig. u~=i (B~cos~x3+C~cos~x~ )e i (~xl -~t) us =(_B~sin~x3 +C~sin~x3 )e i (~x~ -~t) F~=(~2-#2)2cosebsin#b +4e#~2sinebcos#b=O. the higher the mode the f l a t t e r the curves tend to become (in general) and hence the less the group velocity.C. Since the phase velocity is defined as v=~/~. Also. In this special case. By combining four of the SV-waves with four of the P-waves. the corresponding stresses that must exist. The variables A. the Copyright by ASTM Int'l (all rights reserved). each with the appropriate amplitude and phase. the three types of plane waves each can propagate between the top and bottom surface. independent of the other waves. Generally. and the transcendental equations which describe the dispersion relations between the wave numbers. where the stresses can be calculated using the constitutive relations. when ~ becomes nonzero. near the TTTR condition. .B. 5. KIERNAN AND DUKE ON THE ACOUSTO-ULTRASONIC METHOD 111 The wave normal for the P-wave can be found to make an angle e1=• with the vertical axis and the SV-wave can be found to make an angle e3=• ) with the vertical axis. Fi i t is possible to determine the nature of the Rayleigh-Lamb modes that w i l l satisfy these conditions. No further reproductions authorized. the phase velocity for a thickness mode is i n f i n i t e . u~ =i (A~sin~x~ -D~sin~x3 )e i (~x~ -~t) u3= (A~cos~x3+D~cos~xs)e i (~x I -~t) F2=(~2_#2)2sinebcos#b+4e#~2cosebsin#b=O. for the antisymmetric mode. the amplitudes vary sinusoidally through the thickness. setting up separate plate waves which can be derived from the standard plate wave equations by setting equal to zero. Using symmetry arguments with the traction free boundaries. the motion can be shown to be harmonic in the x l . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. ul and u3. equations for both the symmetric and antisymmetric motions can be found. note the slope of the dispersion curves for the higher order modes. 4). the phase velocity is VD=~/~ and the wavelength is 2x/~. Simple thickness modes occur when displacement components are only a function of the x 3 coordinate and time.

3--. many ray angles of input (el) result in similar frequencies. the anisotropic constitutive relation greatly complicates the analysis of the plate wave problem. 5 -. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. A variety of extra plate wave modes are created due to the extra modes of plane waves which are possible.Lamb wave spectrum in Aluminum (after [17]). . Hence. I t is again very important to note that for higher order modes the frequency of the resulting wave changes very l i t t l e with ~. . Thus. hence a great deal of dispersion is associated with these higher order modes for a small frequency range. the higher the phase velocity and the slower the group velocity 9 Phase velocities and group velocities vary along each of these curves and vary from curve to curve. No further reproductions authorized.-. However. antisymmetric symmetric (Frequency)(Thickness) mil-MHz FIG. a lot of energy may be created around these frequencies.. PLATE WAVES IN ANISOTROPIC MATERIAL Several investigators have looked at the problem of plate waves in anisotropic materials. Further complications arise due to the fact that all modes are coupled into the same equations. ' I I ' ~ I (D G) cO c- bOO. dispersion curves) depend on azimuthal angle. Detailed analysis of the dispersion relations of anisotropic materials were not developed until Mindlin and Copyright by ASTM Int'l (all rights reserved). the full analysis of the anisotropic problem would be prohibitive without the use of the computer.VT r~ c- O- \ -. higher the mode. Ekstein did formulate the full analytical solution to the problem for the case of orthotropic symmetry in 1945 [16]. IO0 2~ 300 400 500 600 . Also. 112 ACOUSTICEMISSION V 5O0 ! I I l .. In fact. . 40 VL SO % -..<V. VT i00. 09 O c- >~ 3OO . As might be expected. . > 2OO (1) x /'5 . the properties of the Lamb waves (ie. hence vastly complicating numerical procedures. .

Calculated disturbances may then be used to determine exact magnitudes for the displacement fields calculated in the Lamb wave equations [11]. v. Setting the determinant of the system of equations to zero results in the dispersion relations. at k values corresponding to very small input angles e I .18] is used to provide the magnitude for the P-wave potential. In the method of partial waves. which yields information on the dependence of velocity on frequency. varies with frequency. Fig. the system can then be solved to find relative values of Cn indicating the relative displacements occurring in the plate. In particular. at I MHz. Also. corresponding to small input angles e l . and hk. Hence.59 mm thick aluminum plate. However. the . I t should be noted that this shows relatively similar behavior. others have used this type of approach [9]. the 2 MHz signal content may correspond to either the f i r s t or second order symmetric mode at small values of ~. of the 1. the small group velocities for AU signal contents agrees with plate wave theory. These can be used to derive the commonly used dispersion curves with the i n f i n i t e number of branches. this portion of the wavetrain was on the order of the pulse width. Furthermore. Notice.7 MHz signal content showed no disperison and separated into the early portion of the wavetrain for greater transducer separation (due to greater group velocity). uj=~jei(kixl ~t) becomes uj=~jeik(x~+12x~) k=~/v l~=k~/k The displacement field may be taken as a linear combination of the six allowed partial waves. that the only possible mode of propagation for the I MHz frequency content is the f i r s t order antisymmetric mode. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized. The . derived from plate wave theory. i f a point force e l a s t i c i t y solution [12. for various possible Lamb waves in aluminum plates. the major portion of the wavetrain seems to be made up of higher order Lamb waves with small values of ~.7 MHz signal content showed behavior which agrees with the fundamental symmetric mode. the particle displacement vectors of the partial waves are coupled to each other by reflections at the surface in accordance with Snell's Law. Cij. Mathematically this maybe expressed as 6 9 n u j ~ n ~ j (n)el k(x+Iz()z) Imposing the boundary conditions r. THEORETICAL RESULTS FOR THE ALUMINUMPLATE Figure 5 shows the plot of velocity versus frequency thickness. KIERNAN AND DUKE ON THE ACOUSTO-ULTRASONIC METHOD 1 13 co-workers used a variational approach [13]. 3. Elasticity solutions may be used to understand what disturbances may be created due to an applied force on a free surface. . Figure 6 shows how u3 at the top surface. Copyright by ASTM Int'l (all rights reserved). compared to the I MHz peak in the experimentally obtained amplitude/frequency plot.^=T3~=o2_=O leads to a system of six homogeneouslinear equations whe~ the coefficients of the Cn are now functions of p. More recently the method of transverse resonance or partial waves has been forwarded as a solution technique [14].

The major differences in the AU signatures for the composite plates are the shorter wavetrains and the existence of extra frequency components (see Table 3). 6 -.6 0. for an azimuthal angle of zero. this p~ot shows the same general behavior as the plot for the aluminum plate. the values of peak frequencies for the composite plate showed the same inverse dependence on plate thickness seen in the aluminum plate.2 0. in fact some of the peaks in the amplitude/frequency plot are undoubtedly the result of combined modes (this may explain the double peak in the low frequency range in Fig. In this work. Also. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The number of modes.4 0. Additionally. bandwidth of peaks. This approach involves comparing how the energy measurements in certain frequency ranges varies with azimuthal angle to how displacement values should vary with azimuthal angle.us displacement at the top surface as a function of frequency The AU examination of composite plates is complicated by the issues of inhomogeneity and anisotropy. as predicted by TTTR theory. However.2 FREQUENCY(MHz) FIG. The effects of inhomogeneity are minimized in this work by dealing with only unidirectional composites. An improved understanding of what modes are contributing most to certain frequency ranges may be obtained by noting how AU behavior varies with azimuthal angle and comparing this to theoretical predictions of how p a r t i c u l a r Lamb waves should vary wiZh azimuthal angle for a normal input pulse on the top of the plate.8 1 1. 8). 7 and 8. all peaks in the amplitude/frequency plot are at predicted TTTR frequencies. Material anisotropy makes i t necessary to perform AU measurements at d i f f e r e n t azimuthal angles r e l a t i v e to the fiber direction. Notice. No further reproductions authorized. and the nearness of t h e i r TTTR frequency values makes relating p a r t i c u l a r frequency peaks to p a r t i c u l a r Lamb modes more d i f f i c u l t . t h i s involves comparing phase Copyright by ASTM Int'l (all rights reserved). in order to f u l l y understand AU results. Figure 7 shows a voltage/time plot for a unidirectional gr/ep composite plate. 114 ACOUSTICEMISSION U-3 20 5 0 I [ I I 0 0. Table 3 displays theoretical values for TTTR frequencies for the plate used to obtain the voltage/time and amplitude/frequency plots in Figs. but ignore the effects of inhomogeneity. the experimental effects of other lay-ups have been discussed previously [19]. . This paper w i l l address the issue of anisotropy. an azimuthal angle of zero denotes the f i b e r direction and an azimuthal angle of 90 corresponds to the cross f i b e r direction.

32 MHz varies with azimuthal angle.92 1.2 1. for measurements made in the cross fiber direction.15 and . Voltage versus time plot for 24 ply graphite/epoxy unidirectional plate velocity and group velocity measurements. Fig.46 .i . The full application of this approach involves camparing vast numbers of complicated measurements and difficult numerical calculations. however results to date seem to validate the approach in a qualitative way [12]. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. unlike the general energy content. As an aside. 9 shows how the energy level for the frequency range between . KIERNAN AND DUKE ON THE ACOUSTO-ULTRASONIC METHOD 1 15 FIG. which is greatest in the fiber direction [19]. (MHz).23 . associated with the resonance of the shear wave polarized in the cross fiber direction. This seems to indicate that the mode of wave propagation for this frequency range is dominated by the f i r s t order antisymmetric mode.71 3 I. No further reproductions authorized.36 . (24-ply Gr/Ep) Mode Symmetr i c Anti symmetri c I .2 1. with interlaminar shear strength of unidirectional composites Copyright by ASTM Int'l (all rights reserved).31 2 .4 5 1. the tendency for this shear mode to dominate the signal in the 90 degree direction may explain the correlation of AU energy. TABLE 3 -. 7 -.TTTR Frequencies.69 . .62 .6 As an example. which would be expected to resonant in that frequency range. This frequency range shows the greatest energy level in the 90 degree direction.93 4 .8 1.

24-ply Gr/Ep.0E-08 ~ 6. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.0E-08 8.OE=OB- 1.0E-08 q. in order to substantiate the heuristic argument that the Lamb wave caused by the shear wave polarized to the cross fiber direction should be biggest in the cross fiber direction. No further reproductions authorized. Copyright by ASTM Int'l (all rights reserved).15.OE-07 9.15-.0E-O8- I.32 Mhz range. .2E-07 I.0E-08 ~7. A1(.32) 0.000E-03 8.000E-04 8. OE-O8 3.000E-04 2.32 MHz energy range.IE-07 I..000E-04 0 i i 0 0 2O 4O 6O 8O IO0 ANGLE FIG. Eventually.001 1.Variation of energy in the .Amplitude versus frequency for the signal shown in Fig. in the 90 degree direction [ I ] .0E-08 5.0E-08 2.000E-04 4.OOOE-04 6.3E-t3- L 2 3 q 5 6 7 B 9 I0 FREQUENCY FIG. 9 -. 116 ACOUSTIC EMISSION 1.000E-04 4.001 0.000E-04 2. 7.15-. The complications of the pseudo waves that actually occur in the various directions are i m p l i c i t l y accounted for in the Lamb wave equations.000E-04 6. displacement values for this mode at various azimuthal angles should be computed and compared to the variation of the energy in the . 8 -.000E-03 1.

and elasticity show promise in helping to understand AU results. No further reproductions authorized. hence causing a great deal of dispersion around the TTTR frequency and the associated modulating behavior. predict that the most energy would be directed in the fiber direction and that i t would show up early in the wavetrain. results indicate that the major energy content of the AU signal is higher order Lamb waves caused by plane wave interaction and reflection at various angles to the normal. these results highlight the great dependence material geometry has on AU results and provides a means for accounting for these effects. . CONCLUSIONS AU results indicate that higher order Lamb waves are dominating AU results for f l a t plates. hence blocking the rays emanating in that direction. due to the high slope causing a high group velocity. which agrees with experiment. Additonally. TTTR. A. A further complication to interpretting experimental AU results is the effects of energy flux deviation [12] on waves propagating in an anisotropic material. Lambwave theory. KIERNAN AND DUKE ON THE ACOUSTO-ULTRASONICMETHOD 1 17 Velocity measurements indicate that the bulk of the AU signal for composite plates is also made up of higher order Lamb waves. "Ultrasonic Evaluation of the Strength of Unidirectional Graphite/Polyimide Composites". usually requiring the use of coupled mode theory [14]. Calculations of dispersion curves and elasticity results for the 0 and 90 degree direction showed the large effect azimuthal angle can have on Lamb wave behavior. Specifically. The potential for u t i l i z i n g this theory to interpret slight variations in the signature are exciting. J. NASATM X-73646. and Bowles. Dispersion curve results [12]. REFERENCES [l] Vary. Some reasons for this behavior have been postulated [12]. which may help explain the vast differences in AU behavior with azimuthal angle. an indentation in a frequency peak may relate to a flaw blocking the input of energy for a certain angle e l .. however in most cases measured phase velocities were not nearly as high for composite plates as for the aluminum plates. using Leknitskii's solution [20]. but the computational details are extreme and other complications. K. Cleveland 1977. elasticity solutions show that most of the input is for low angles e resulting in the excitation of low wavenumber Lamb waves. In general. This problem may be crudely modeled with Lamb wave theory and elasticity. such as the existence of both normal and shear disturbances resulting in the elasticity results. Copyright by ASTM Int'l (all rights reserved). It is the nature of these modes to have quickly changing phase velocity and group velocity. In general. For instance. Elasticty results. the calculation of the dispersion curve for just one azimuthal angle is a sizeable computational task. also predict that the energy level early in the signal should be high for the fiber direction. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. ~he frequency range and low wavenumber associated with the input are both desireable for exciting higher order Lamb waves close to the TTTR frequencies. where the material thickness is on the order of a wavelength associated with the center frequency of the sending transducer. and hopefully eventually modelling them in composite materials.

"Quantitative Assessment of Damage Growth in Graphite/Epoxy Laminates by Acousto-Ultrasonic Measurements. R. 655-662. and Rose." Materials Evaluation. G. pp. pp. Copyright by ASTM Int'l (all rights reserved). Vol. H. J. New York. 1985." in Physics Review. pp. pp. "A Study of the Stress Wave Factor Technique for the Characterization of Composite Materials. Plenum Press. "Acousto-Ultrasonics Characterization of Physical Properties of Human Bones. Vol. 1988. E. Y. S. 40.. Vol. H. S. H. A.D. E." in Review of Proqress in Quantitative Nondestructive Evaluation. [4] Kiernan. "High Frequency Vibrations of Thin Crystal Plates. and Lee. "Waves and Vibrations in Isotropic and Anisotropic Plates. I I . . C. "PC Analysis of an Acousto-Ultrasonic Signal" Materials Evaluation. Williams. and Auld. 68. G. 10. and Kaul. "Application of Homomorphic Signal Processing to Stress Wave Factor Analysis" Materials Evaluation. K. No. R. 11-23." in Acousto-Ultrasonics: Theory and Application. pp 1446-1454. Y. D. S. H. Sept. O." in R.. ed . Plenum Press. H. [5] Karaguelle. No further reproductions authorized.. K. 1988. A. [15] Vasudevan. K. 1988. ed . J. C. [11] Williams.. C. No. Plenum Press.4A.. "Application of Acousto-Ulatrasonics for Predicting Hygrothermal Degradation." in Acousto-Ultrasonics: Theory and Application. 301-304. pp. Karaguelle. H. A. [7] Patton-Mallory M. I . . "A Physical Model for the AU Method " Ph. ed .. P. Vol. R. pp. 1985. Mindlin and Applied Mechanics. J... I I . Pergamon Press Inc. O.. P. [10] Kautz. New York. H. Cavano. "Trans-Ply Crack Density Detection By Acousto-Ultrasonics. 54." in Review of ProQress in Quantitative Nondestructive Evaluation. 1988. E. D. pp. 327-336. Plenum Press. Duke. "Elastic Waves in Free Anisotropic Plates" Journal of the Acoustical Society of America. A. 1985. Vol.. Roman. [14] Solie. eds. A. N. ed . Jr. Kautz. A.. Thompson and D.. Duke. 149-196.. Vol. 305-309. and Lee. [12] Kiernan.. 1099-1106. New York. Thompson and D. 319-326. Cleveland. 1988. K. and Mal. J. [13] Pao. 1988. . "Ultrasonic Input-Output for Transmitting and Receiving Longitudinal Transducers Coupled to the Same Face of an Isotropic Elastic Plate. 1973. 118 ACOUSTICEMISSION [2] Govada. Bivas... B.. "Ray Propagation Path Analysis of Acousto-Ultrasonic Signals in Composites. Margulies. Duke. K. I.. and Weinreb. Dissertation." Acousto-Ultrasonics: Theory and Application ed . E9] Phani. C. 1974. Govada. and Duke. 1989. Plenum Press. [3] Talreja.. C. Duke. 1945. New York. pp. K. 1344-1352.. Henneke." NASACR 17480. J.. E. and Henneke. Duke. J. New York. Chementi.. H. C. T. J.3B.." in Acousto-Ultrasonics: Theory and Application. New York. T. [6] Mittelman." in Acousto-Ultrasonics: Theory and Application.. Virginia Polytechnic Institute and State University. D. Plenum Press.. Leichter. 43. J. J. N.. pp.. 46. A. pp. Jr. C. K. J. D. 53-60 [16] Ekstein. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Plenum Press.. Duke. M. Chementi. Blacksburg. Oct.. S. Jr. and Bowles. and Anderson. No. [8] Hemann. "An Acousto-Ultrasonic Method for Evaluating Wood Products. E. eds. 1984. New York. 6. J. "Response of an Elastic Plate to Surface Loads and Buried Dislocation and Sources. May 1982. M. New York. I .

G. N. W. 1934. July 1988. New York. . May 1986. C. [18] Timoshenko. C.. Theory of E l a s t i c i t y of an Anisotropic Body. "Development of NondestructiVe Testing Methods For The Evaluation of Thin and U l t r a t h i n Sheet Materials" AFML-TR-65-320. McGraw-Hill. 1965. "Acousto-Ultrasonics as a Monitor of Material Anisotropy" Materials Evaluation. [19] Kiernan. M. J. Cleveland.. G. and Henneke. J. No. J r . Moscow. E. No further reproductions authorized. P. S. and Duke. Duke. and Schmitz. 1105-1113. Dec. Vol. Copyright by ASTM Int'l (all rights reserved). [20] Lekhnitskii. G. 8. "Low Frequency Plate Wave Modes" in Ultrasonic Stress Wave Characterization of Composite Materials. A. G. . 1981. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Henneke I I . E. Stinchcomb. Theory of E l a s t i c i t y . KIERNAN AND DUKE ON THE ACOUSTO-ULTRASONICMETHOD 119 [17] Wieczorek. . S... [21] S t i f f l e r . edited by J. pp. R. Mir Publishers. Dayton. NASA CR-3976. and W. and Goodier. 46. T. C. J r .. I I .

Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. . No further reproductions authorized. Signal Processing Approaches Copyright by ASTM Int'l (all rights reserved).

KEYWORDS: acoustic emission. J. Sachse. Roget. J. JAPAN. fracture behaviors. Hirotada Oyaizu. the needs of integrity monitoring of industrial structures are increasing rapidly. Johkaji and Mr. waveform parameters. 22-1. Correct characterization is desirable for material evaluation and improvement. Yamaguchi. Roppongi-7. signal processing. The performance of the advanced systems can be considered more powerful for material evaluation than that of usual systems depending on conventional AE parameters such as peak amplitude and event count. FRP INTRODUCTION For the analysis of fracture behaviors in materials. etc. research assistant. . Kusuo Yamaguchi. Also. Mr. and Yutaka Kobayashi ACOUSTIC EMISSION TECHNOLOGY USING MULTI-PARAMETER ANALYSIS OF WAVEFORM AND APPLICATION TO GFRP TENSILE TESTS REFERENCE: Yamaguchi. K. Mr. matrix-cracking and fiber-breakage.astm. and Kobayashi. Jun Johkaji. Yamaguchi. . the results indicated the high identification p o s s i b i l i t y on failure modes such as fiber-debundling. Oyaizu. on-line measurement is very important. H. American Society for Testing and Materials. Y. No further reproductions authorized. 1991. Minato-ku. New parameters such as energy moments and zero crossing number were used together with conventional parameters and showed the usefulness for waveform analysis and source classification.. The systems were applied to different experiments such as fracture-monitoring of metal-structures and testing of materials. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement... and K. Tokyo 106. Philadelphia. Johkaji. Acoustic Dr. "Acoustic Emission Technology using Multi- parameter Analysis of Waveform and Application to GFRP Tensile Tests. University of Tokyo. ABSTRACT: Advanced acoustic emission instrumentation systems which can use multi-parameter data obtained by wave-envelope processing of all detected AE's have been developed. Oyaizu. In tensile tests of GFRP. former graduate students of the Department of Electrical Engineering and Electronics." Acoustic Emission: Current Practice and Future Directions~ ASTM STP i077 r W. Kobayashi are Professor. Eds. 123 C o p y r i g h t 9 1991 b y ASTM I n t e r n a t i o n a l ~.org Copyright by ASTM Int'l (all rights reserved).. the Institute of Industrial Science.

13]. ordinary AE instrumentations.e. the results are not necessarily very successful. As for wave-features. use comparatively simple parameters such as peak amplitude or energy. .18. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized. fiber-debundling.7]. it was reported that the fiber breakage AE could be discriminated from the others very successfully by the amplitude distribution [14]. improvement of analysis method and better information acquisition system are necessary.16. Systems which extract waveform parameters including pattern parameters in real-time operation have been developed at the Institute of Industrial Science (IIS). Regarding the AE applications to composites of high strength fiber. in many cases. etc. for example. it still can be advanced by utilizing more information from detected waves [3. duration-time Td. wave-features and location. Some systems can use integrated wave-energy of every event [6]. Most analyses on AE waves by the conventional instrumentations depend on peak amplitude or peak energy distributions [7] or wave- energy distributions [6]. The AE technology has evolved to utilize multi-parameter processing recently. and they have been applied successfully to the fatigue tests of metal- structures [8]. Also friction AE's at Copyright by ASTM Int'l (all rights reserved).11]. the AE technology needs three kinds of information from detected waves as input. In many cases. pattern information of input waves should be employed. Some studies use wave frequency parameters [12. matrix-cracking and fiber-breakage in GFRP could be presumed by waveform-parameters [9]. except off-line waveform analysis system. [9]. As for information acquisition. the performance of those methods using only amplitude or energy data of AE waves is not satisfactory because of insufficient information for the detail analysis. The advanced AE system which can use wave- pattern parameters and other new useful parameters is developed to improve the performance of the technology and to contribute to the practical applications [3. the pattern parameters had not been available in usual instrumentations for a long time until 1988. To improve this situation. For the actual development of the AE technology. The multi-parameter signal processing method which employs pattern parameters seems powerful for analysis on fracture of composites which usually have complicated failure modes and behaviors. However.9]. the results of experimental analyses could not show very successful discrimination by the amplitude distribution [15.5]. total wave amplitude At and total wave energy Et besides sorts of count and maximum amplitude ap for analyses [6. In the case of popular FRP such as GFRP and CFRP. i. In general. but.19]. University of Tokyo [5. rise-time and duration-time for each event.2].4.10. 124 ACOUSTICEMISSION emission technology is being expected for such purposes as one of the most powerful tools [1. From the experimental results it seems that some failure modes. Basically.8. They are being applied to FRP material tests. accumulation of experience or database.17]. Recently the applications of multi-parameter analyses to material tests and structural monitorings are being increased [13. and new systems use such parameters as rise- time Tr. frequency. the AE technology has a limited performance because it is an indirect measurement method.

the system performance firstly depends on the availabSe input information. (1) All available waves or hits should be processed at the input stage as many as possible for the analysis of all possible AE sources at the later processing. Naturally the principles compete with each other in design engineering. CONCEPTS OF ADVANCED AE TECHNOLOGY In general.5. Additional application results are expected to be published. YAMAGUCHI ET AL. Among the popular and new parameters in the AE information. and the typical processing architecture is stated. The supply of the instrumentations which employed the pattern parameters was started by a manufacturer in 1989 and the instrumentations are now being applied to the experiments of subsurface and civil engineering in Japan and Canada. effective data-compression should be required for good compromisation. In this paper. envelope detection method. Then.2. For high analysis performance the following two principles should be considered in the design of input processing [3. from stochastic methods for macro-analyses to precise techniques for detailed analyses using micro-data of individual wave. Therefore. . Copyright by ASTM Int'l (all rights reserved). However. the fundamental principles in design concepts for the advanced AE instrumentations are explained. No further reproductions authorized. ON GFRP TENSILE TESTS 125 cyclic loading tests could be discriminated at high probability by the specific features of waveform-parameters and loading phase information [9].20].4]. Such new instrumentations will be available from some other manufacturers soon. Finally.4. Comparison of these features with those of the tests on metal is to be discussed in this paper. Therefore. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. There are variations of AE technology. Processing of envelope-detected waves can be the best solution in many cases. some GFRP tensile tests and the results are shown and discussed. In usual way the cost of the system has to be considered.3. as an application of the system. Also. (2) Every set of processed wave or hit data by the input stage should contain detailed information as much as possible to enable various and precise analysis by all possible ways at the later processing. kinds of wave moment as pattern parameters and some other parameters will be explained. in every AE instrumentation. the method and the capacity of the input processing stage essentially dominate the possible performance of the whole system. the kinds of information which are available in the present AE technology are shown. multi-parameter processing and the instrumentation are explained and discussed. and consequently increase of processor capacity and decrease of processing speed are brought on. The above two principles should be considered at the input stage in real-time operation. AE techniques analyzing the detected waves are used for the identification and evaluation of AE sources and the cause of the sources [1.

Table 2 shows the content examples of wave microdata currently used in the actual advanced system recently developed at IIS. The step (3) can be done by on-line operation or by off-line analysis according to the purpose and the cost. . energy. characterization.4. source- location. The consideration of the following characteristics are necessary for waveformparameters. very important problems are the selection of effective input informations and how to acquire and use them. except in the direct digitizing method.4. decision making The steps (i) and (2) should be done by on-line processing.10].5. grouping (3) higher-level grouping. severity assessment. Thus. The data of waveform parameters from every input AE or hit are called wave or waveform microdata in contrast to macrodata such as counts and distributions. Automatic processing of step (4) is necessary for on-line monitoring. No further reproductions authorized. (i) easy data-extraction of input waves in real-time operation (2) high data-compressibility from original waveform (3) good representation of the original waveform features (4) low sensitivity to noise and minor change of propagation conditions The parameters in the pattern and the time categories is related to the AE waveform. (I) wave (hit)-data extraction. INFORMATION AVAILABLE IN AE TECHNOLOGY AND WAVEFORM PARAMETERS In the AE technology. time. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Distributed processing system is a good and preferable structure for these three steps. pattern and frequency [3. Such AE waveform information is effective for fracture behavior analysis when the propagation distance is short [4.5] and for the propagation route identification when the monitoring area is large [3. noise rejection and data recording (2) combination of hit-data for event-data composition. 126 ACOUSTICEMISSION In most cases the data processing in advanced AE instrumentations consists of the following four steps [4. concentration and time-trend analysis.5]. Table 1 shows the sorts of information available or possibly available in the AE technology at present. estimation or evaluation.8]. environment analysis (4) presumption of object status or behavior. event-qualification. but the pattern parameters such as moments are the better representations of the approximate shape of the waveform- envelope than the time parameters like rise-time Tr and duration time Td. the pattern parameters can be general and it can fairly satisfy the above four requirements. but off- line processing or human decision on step (4) would probably be a good way for research and development. Copyright by ASTM Int'l (all rights reserved). Waveform parameters of input AE waves can be classified into four categories.

Channel number ch 6 Arrival time ts 32 Time Wave rise-time Tr 12 Wave duration time Td 16 Waveform Energy Peak amplitude ap 12 envelope Total xave energy Et 32 Energy moment i Teml 12 Pattern Energy moment 2 Tem2 12 2nd order energy moment Tem(2nd) 16 Wave Frequency Number of zero-crossing Nz 8 .Tem(2nd). ts.. Energy intensity) Waveform change Loading condition Load level and loading pattern Load sequence Load phase in cyclic loading (Ldph} Felicity ratio Environmental Atmosphere.Sorts of information available in AE technology. No further reproductions authorized. Td. At.) Frequency (Nz. Total (24bytes) ~ 190 Copyright by ASTM Int'l (all rights reserved)... etc) parameters Time (Tr. etc. Event count (Evc) Stochastic distributions (ap.. FFT...Tem. . Et..) Time trend Location concentration Event activity (Number. Tam(2nd).. ap 2. Length Parameter Categories Contents Symbol (bits) . Auxiliary data Aux 32 .. Macro data AE count (ring-down). Contents (vessel) condition Noise Process sequence or condition Plant operation TABLE 2 -. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. YAMAGUCHI ET AL.Contents-examples of wave (hit) m i c r o d a t a including waveformparameters..) Pattern (Tam. etc. etc. ON GFRP TENSILE TESTS 127 TABLE 1 -.. etc.) Location(Lc) Arrival time difference Nit sequence Zone Propagated xaveform (pattern analysis) Waveform Energy (ap.. ap 2.

Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.~ ~ ~ ~ i ~ _ _ _ Arriv'!"time ~ ~ ~ t'el te~ tiC'me t Tem].dt (2) n 1st order amplitude moment: Tam = ~ai-ti. .dt / Et .(Tem)2 (6) where d t is t h e s a m p l i n g period of A/D conversion.J Maxi!um ~Envelope detectedwaveformai(t) amplitudeap~.tt e. 1 -.ti-dt / Et i-e (4) n 2nd order amplitude moment: Tam(2nd) = Za~.Illustrations of w a v e f o r m parameters. 128 ACOUSTICEMISSION Wn i dow Tnz Zer~cr~ I ~Input AEwaveform(original) U time t RisetimeTr L~.ti2.dt / At (3) i-0 n 1st order energy moment: Tel = Zai2. Copyright by ASTM Int'l (all rights reserved).| I I ~--Tem2 ~ I I Tmaxl f~ DurationtimeTd. FIG. . n Total amplitude: At =Xa~-dt i=e (i) n Total energy: Et = Xai2.dt / At (Tam) 2 (5) n 2nd order energy moment: Tem(gnd) = Za~2. No further reproductions authorized.

(2). and the pattern parameters. Tam(2nd). Advantages and disadvantages of the envelope detection are as follows.22].2 (a). The Moment-parameters in Table 1 can be used for waveform analysis. the zero crossing count is a very simple application of higher order crossing (HOC) method.e. Wave rise-time Tr or duration time Td has been used for pattern classification successfully in some cases [21. This simple frequency parameter Nz is not available in usual commercial AE systems. multi-path effect and waves-overlapping. YAMAGUCHI ET AL. total wave amplitude At and total wave energy Et. Tem and Tem(2nd). zero crossings of the zeroth differentiation of the original time signal [23]. No further reproductions authorized. This might have been the reason why pattern parameters had not been used in commercial systems until 1988. Pattern parameters had been used only by the systems developed at IIS during the time from 1980 to 1988.(4). in the case of usual AE waveform as shown in Fig. Through the calculations of equations (I) to (6) in real-time operation. Usually the Ist order energy moment Tem is most stable and insensitive to such disturbances as noise. high data compression from the sampled data of original wave to the parameters can be achieved. the ist order moments calculated by eq. In general. it is the same parameter as ringdown count. Zero crossing number Nz indicates approximately the main frequency of the first part of a waveform. From another viewpoint. Without a fixed window time Tnz. Rise-time can be processed easily but it is less efficient than the moments in considering the point (3) waveform representation and in considering the point (4) insensitivity to noise. threshold-shift.l shows the illustrations of the waveform parameters in Table 2. Equations (1). the 2nd order energy moment Tem(2nd) or amplitude moment Tam(2nd) would be useful for wave discrimination between B1 and B2. In some cases as B1 and B2 in Fig. Therefore.(3) and (4) as well as the 2nd moments show the peakedness and the concentration of the envelope patterns. i.(5) and (6) are used for real-time calculations of the energy parameters. (i) the Ist(or odd) order moment shows the asymmetricity of the wave pattern (2) the 2nd(or even) order moment shows the peakedness or the concentration of the wave pattern However. the moments have the following two characteristics according to the orders. the Ist and the 2nd order amplitude moment and energy moment Tam. Copyright by ASTM Int'l (all rights reserved). . if the window time is proper and fixed. In the systems using the parameters illustrated in Fig. the ist order energy moment can be the most effective and convenient pattern parameter. envelope detection is adopted for the input A E w a v e processing prior to A/D conversion except the processing for zero crossing number Nz. Processing of pattern parameters needs more complicated and higher speed hardware in input stage than processing of the other parameters in Table 2 [5].2 (a)(b). Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. in usual AE applications.l.(3). ON GFRP TENSILE TESTS 129 Fig.

simple long. (i) loss of frequency information of the original time signal (2) loss of precise shape and phase of the original wavefront Copyright by ASTM Int'l (all rights reserved).reflections. No further reproductions authorized.Change of AE waveform information according to propagation path. (i) bandwidth reduction to enable 31ow sampling and easy processing (2) good representation of approximate wave contour pattern Disadvantages. 2 -.Relation between waveforms and the ist and the 2nd order moments (energy and amplitude). route. =ulti-pat~ ) Advantages. where f(t)'s are envelopes of amplitude or energy waves. 130 ACOUSTICEMISSION . TABLE 3 -. information fracture mechanism 'plate wave effects \ in AE wavef0rm source characters ] generation mechanis~ overlapping .~A / I s t moment : small\ f(t) j /~ ~2ndmoment : small) / \ B1 /lst moment : large I / ~ ~ . . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. propagation path short. ~ 2 n d moment large] > ts=O t (a) A usual case of AE waves of short distance A and long distance B1 f(t)J ~/~"A : the sam~iiilt(~iment : large ~ > ts=O t (b) A special case of long distance (multi-path or overlapped wave) B2 FIG. complicated Main source .

ON GFRP TENSILE TESTS 131 The advantages outweigh the disadvantages in many cases of actual application.5]. In general. the methods-selection for the purposes is very important and it is one of the essential part of the AE technology which needs much experience and knowledge. The level (4) processing for characterization. an advanced AE instrumentation should perform real-time and multi-parameter processing in the input stage and high-speed processing for event qualification and grouping after the input stage. The degree of the change increases by the propagation path characters from left to right in the Table. The applicable variation of grouping methods is one of the flexible characters of the advanced system. On-line processing can be extended to the higher processing level (3) which are now performed by off-line processing. Table 4 shows examples of selectable parameters. optional processing or analysis methods at the processing levels. evaluation or severity assessment would still needs human ability and knowledge. Also. YAMAGUCHI ET AL. Consequently. except special purpose systems. Suitable processing methods can Be selected in each level. by analyzing waveform parameters or wave microdata. The advanced AE instrumentation system which is based on multi- level distributed structures is usually applicable to the wide range of versatile and precise use [8. Copyright by ASTM Int'l (all rights reserved). while the methods of (2) would be useful for many structural integrity monitorings. Thus.9]. (i) direct source-event grouping by source mechanism identification or presumption (at a short distance) (2) indirect source-event grouping by propagation route identification or classification0(via long and complicated propagation route) The methods of (I) can be used mainly for the fracture analysis or the evaluation of materials. the advanced AE instrumentation should be based on a distributed high-speed processing architecture [4. MULTI-PARAMETER PROCESSING As stated in the previous section.(2) and (3). Table 5 shows examples of effective parameter combination for the experiment examples to which the advanced systems including the prototype have been applied. it should be flexible for the various applications and objectives. (1). Information in a detected AE waveform changes from that of the source to the propagation distance and the medium structure as shown in Table 3. Zero-crossing number Nz can be a good and simple compensation for frequency information of input AE waveform to some extent. For example. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. grouping of events or identification of event-sources can be performed in the following two ways. . No further reproductions authorized. the latest IIS system has two on-line processing levels (i) wave-data extraction and (2) combination of hit data and location which are explained in the section of CONCEPTS OF ADVANCED AE TECHNOLOGY.

. a r r i v a l time d i f f e r e n c e qualifi- cation location 4-. W e s ( 2 n d ) .Examples of multi-option processing or analysis method selection at each distributed processing level. Load: load level. Evr: event rate Lc: location. Ldph time t r e n d parameter d i s t r i b u t i o n s analysis number of groups. Td . p l a n t sequence Kaiser e f f e c t r a t i o F e l i c i t y r a t i o . . e t c . . . . Nz. Tem.2-dimensional) event. l i f e of groups l o a d i n g c o n d i t i o n s . Load. No further reproductions authorized. Tam. Ldph: load phase Copyright by ASTM Int'l (all rights reserved). . Tr.s e n s o r s l o c a t i o n and h i t sequense. ) parameter time (ts. 2 . etc. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 3-. Note: Evc: event count. ) Level 3 grouping c o n c e n t r a t i o n of parameters analysis analysis Lc. . zone grouping e v e n t data e v e n t data s t r u c t u r e filing macro data Evc. processing level processing p r o c e s s i n g o p t i o n or s e l e c t i o n waveform energy (ap. ) Level 1 extraction pattern (Tel. Lc. Evr processing parameter d i s t r i b u t i o n s (ap. ) frequency (Nz) i n p u t wave parameters noise parameter v a l u e l i m i t check extraction rejection abnormal waveform check a u x i l i a r y data indexed data processing data p r e . 132 ACOUSTICEMISSION TABLE 4 -. Et . master-slave & combination (1-dimensional) Level 2 waveform parameters (1. ap 2.q u a l i f i c a t i o n qualification coincidence. . . At. l o c a t i o n .

000-3. Lc tests in FRP (fiber breakage. Ldph. Copyright by ASTM Int'l (all rights reserved). The 3rd system has been developed recently [5] and is used for current study. Tam(2nd).} Tem:Evc.Examples of effective parameter-combination out of waveform parameters and other information. Lc. and it is still in use after some of the improvement. since 1972. Ldph components (Tem-ap) :Evc. . Tem(2nd).Ldph FRP cyclic ap:Evc. Besides the ability to perform multi-parameter processing. The Ist-generation was a multi-channel AE source location system equipped with arrival time difference detectors [24]. and applied to experimental applications. It was a fully digitized system specialized for arrival time-difference detection among input channels and it was used with an on-line mini-computer. University of Tokyo. Lc FRP tensile classification of presumed failure modes (Tem-Nz) :Evc. Et. Evc. ap. The preparation-time after every input is about 40 microseconds and the ultimate input speed is 25. debundling. No further reproductions authorized. Tem. Environment INSTRUMENTATION DEVELOPMENT AND APPLICATIONS Several AE instrumentation systems have been developed at IIS. maximum input speed is 2.Ldph of FRP piping grouping of friction AE sources Evc:Lc. The 2nd system was the prototype of advanced AE instrumentation. matrix cracking.Ldph loading tests grouping of friction AE sources Evc:Lc. At. one of the noticeable features of the advanced systems is its very high input processing speed. Load. YAMAGUCHI ET AL.000 hits~second~input channel.000 hits/second/channel according to the maximum signal duration of 300- 500 microseconds [5]. Ldph other effective combinations of. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. This system had introduced the dead-time technique and was adopted since by most usual AE instrumentations. debonding. Tam. ON GFRP TENSILE TESTS 133 TABLE 5 -.Ldph fatigue tests ap:Evc. etc. The system have been showing the advantages of multi-parameter analysis in the experiments on metal structures [8] and FRP tests [9]. parameter application combinations examples objectives Tem:Evc. In typical use. Nz.

whitening started at the tips of notches. Fig.4 (a) was applied to the tests.3 shows the experimental instrumentation setup for tensile and cyclic loading tests. AE detection started at No. No further reproductions authorized.9 loading when the final breakage occurred.l-7 (mostly at No.6 shows the stepwise loading and unloading process and the AE event rate at tensile test Ttl.3 loading. the dynamic range extension method in Fig. therefore. In these experiments. The dynamic range of 72dB is enough for usual tests but the range can be extended to ll2dB as shown in Fig. A DC- servomotor driven test-machine of a lead-screw type was used for the experiments. Pre-cracking was made in the (G-2) specimen for the tests Tcl and Tc2 by 288Kg pre-load which was 80% of the estimated maximum fracture load (360Kg).4 (a). Two different gain channels were connected to one sensor output and extended the dynamic range up to 64dB. Table 6 shows the types and the conditions of the loading for the tests.4 (b) by the same way as in Fig. and it spread out toward both of the longitudinal directions of the specimen. The envelope detected AE signals are processed at the AE waveform parameters extraction system which produces waveform microdata shown in Table 2. Test Results Fig. 134 ACOUSTICEMISSION APPLICATION TO GFRP TESTS Experimental Setup and Tests Fig.7 (a) and (b) show the distributions of energy moment(Tem) and peak amplitude(ap) during No. The 3rd system has 12 bits (72dB) converters. The waveform microdata are transferred to DEC 11/73 CPU which performs the processing of the level(2) and a part of the level(3) and records all the data sets in the disk for off-line analyses. Fig. A wider dynamic range than 48 dB was preferable for the GFRP tests. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. . At No. Copyright by ASTM Int'l (all rights reserved). peak amplitude and zero-crossing number besides location data were mainly used for the analyses. the waveform parameters of energy moment. The improved 2nd system has an 8 bits (48dB) A/D converter (0-10V analog input) at each input channel.3-7) and No~ loading. Loading ranges of 36-216Kg for Tcl and i08-288Kg for Tc2 were applied to the specimens at each loading cycle [9].4 loading. The purpose of the cyclic loading is to observe friction AE and to analyze its features which are possibly different from those of the other types of AE. The whitening had reached at the locations of about 15n~n from the center of the specimen at No. The stepwise tensile loading during the tensile tests (Ttl and Tt2) had the aims to slow the fracture progression and to make the AE measurement and the visual observation easy. Most of the AE sources were located near the center of the specimen.5 is the shapes and the specifications of specimens (G-I) and (G-2) used for tensile tests (Ttl and Tt2) and cyclic loading tests (Tcl and Tc2).

: [. ~. YAMAGUCHI ET AL. 3. No further reproductions authorized. FIG..Experimental instrumentation setup. .~ 39mY ........ ... .. I00 ~ V (38dB) (40dB) 72dB (ibit) 13~ V -* 39mY max. . 8 0 ~ V ..j . 3 -.. Copyright by ASTM Int'l (all rights reserved).$2: Sensor(Dunegan D9201) FIG. 2 .Dynamic range extension method by overlapped connection of two different-gain channels. ON GFRP TENSILE TESTS 135 ch input (o~1ov) sl 9 s0~ AE waveform 34dB parameters extraction = = 2 50dB system [The improved $2 ~ 34dB ~2nd system / amplifier Specimen 20dB Main amplifier Wave 1 I IT CPU (A/D converter range) memory lOV = 8bits Disk I Disk SI. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement...... gain (22dB) l~V (OdB) (a) The improved 2nd system (b) An example of the 3rd system- in the GFRP tests. mirL gain ch3.4 400mY sensor (112dB) output A/D c. (Sbits) 48dB . . ch3. . ... 2 { (12 i (12b~f (70dB) (TZdB) 64dB 48dB ll2dB "" C"'~'32dB ... "I..4 voltage input J~ 2OmV -~ lOV (8bits)' (86dB) 72dB chl.... . 4 -. . 2mY -~ l OV chl.

00 Tcl G-2 range: 3 6 . 5 -. . of loading tests specimen loading condition loading speed(mm/min.26 Specimen (G-l) t=3. 5 J I !! 260 [mm] !' Untreated woven cloth [0~ ~] / Unsaturated polyester VfR-0.) tensile 9 0.G F R P specimens./~.2 1 6 K g 100 (triangular) precracked preloaded: 288Eg period: 9 sec.288Kg 200 (triangular) precracked preloaded: 288Eg period: 9 sec. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. cyclic loading 10. 136 ACOUSTICEMISSION Aluminum plates sensor sensor 9 I ! ! K ' 130 42.5 ~ ! ! I ! 260[mm] Silane treated woven cloth [0~ ~] / Unsaturated polyester Vf~-0. 05 Tt2 G-2 stepwise loading / final / OO'at and unloading \breakage/i f i n a l stage/ cyclic loading 10.5~ 0 . No. No further reproductions authorized. Copyright by ASTM Int'l (all rights reserved).0mm sensor sensor 8. TABLE 6 -.T y p e s and c o n d i t i o n s of l o a d i n g for t e n s i l e tests (Ttl and Tt2) and c y c l i c t e s t s (Tcl a n d Tc2).00 Tc2 G-2 range: 108 . Omm Specimen (G-2) FIG. - ! | 160" " ' ! J I '! 130 42.26 t=3.05 Ttl G-I stepwlse loading I final ( O02at / and unloading breakage) final stage/ tensile 6 O.

lu ..o- o "~ ... ..0 2. .....9.. ..0 Energy Moment Tem [~sec] Peak A m p l i t u d e ap [mV] (b) L o a d i n g No..... ... .. .... 2.......i "t0.......... ~ i: i::i~:i::. ... .. .. ~ o .0 8.. . -TT-I .! ...J... ... 6 -. !-~ .... Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.0 8.. I-m'= .... No further reproductions authorized........ i oJ ...... . . .l-7 l o a d i n g and (b) d u r i n g No... 8 " 1 N< ... l ~ 7 o 8 o ~z ~ _I f I ~I I I 0 20 40 80 80 I00 120 O.. ... ........ I o .... ... .. Copyright by ASTM Int'l (all rights reserved)... .......0 8. 9 .. O N G F R P T E N S I L E TESTS 137 ..... .Stepwise loading and u n l o a d i n g process and AE g e n e r a t i o n rate during tensile test Ttl....0 6...........9 FIG....... co L~ O %0 i.D i s t r i b u t i o n s of energy m o m e n t T e m and p e a k a m p l i t u d e ap at t e n s i l e test Ttl........ YAMAGUCHI E T AL.. ! ..... (a) during No....0 4.0 4.~ .O 2. ..~ ) c) 0 2000 4000 6000 8' 00 10000 TIME (sec) FIG... ..... ~i O o v . c> ~ . ~o~Z~/ ~ ....__ .0 Energy Moment Tem [~sec] Peak A m p l i t u d e ap [mY] (a) L o a d i n g N o ........... 7 -...... i c) i~ Z -~... o ~ I i .... ... .:j21 .. Loadin( Number "N!-6" No . j ....... ... r-- ~2 t~ o. ..... ...... ! i O 20 40 60 80 I00 120 ~ 0..... . . . o ? v .........

Fig.6 (window time Tnz=50~sec. .32~ap~ 1. AE detection started from low load level of No. regularity of the AE generation at successive cycles such as Group A1 of Tcl and Group A2 of Tc2 could be observed at the load increasing and decreasing periods. and (b) shows the results at the No. o .Distributions of zero-crossing number Nz and energy moment Tem in tensile test Tt2 (specimen: G-2) by stepwise loading unloading (No. Fig. Three parameters. Fig.l-4 and (b) during No.. ap.9 (a) and (b) are the AE generations in the cyclic loading tests Tcl and Tc2 (G-2 specimens). . . Fig.4 0 0 ~ OOo o 0 9~ o o 9 9 o 9 .l loading and continued at the following loading. Unlike Ttl.).6b (0.9 (a).9 (b). . Copyright by ASTM Int'l (all rights reserved). Nz and Tem.2mV< ap) presumed as debundling AE presumed as fiber breakage AE FIG.6 loading of Tt2 when the final fracture occurred. were used for the AE analysis on Tt2. o . During the test Tc2. The final fracture occurred at No. In both tests. (a) during No.8 (a) and (b) show the distributions of zero-crossing number Nz and energy moment Tem during the tensile test Tt2 on the specimen G-2 by stepwise loading and unloading process similar to the test Tt2.12 shows the two dimensional distributions of Tem by different ap values of the AE's in group B1 during the test Tc2 (1-200 cycles). Fig. a) ~0 o 2O o N o o z 15 0o0 o o o tm .2mY) (l. No further reproductions authorized.l-6). 9 9 9 m g*0 0%900 o.l-4 loading of Tt2. The AE sources were located at considerably wider range from the center than those of Ttl. 138 ACOUSTIC EMISSION O:5 Events ~:I0 Events O:25 Events O:50 Events 25 9 9 . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.6 loading in the Tt2 tests. Fig. 8 -.8 (a) is the results during the No. Coordinate (0-i00) of the vertical axis is the load increasing phase and (100-200) is the decreasing phase. while almost no AE was detected at the peak load of the test Tcl. many AE's were detected near and at the peak load.ll is the Tem and ap distributions in group A2 of Fig. 9 o 9 9 9 9 o OO oo I O o o 0 o o o 0) 5 N O~i)O O 0 o o o o o 0 0 I I I I = t I I I I 30 50 70 90 110 0 50 70 90 110 Energy moment Tem ( !a sec ) Energy moment T e m ( lasec ) (a) Loading N0 I--4 (b) Loading No.10 shows the distributions of ap and Tem of the AE's in group A1 of Fig.

b~ " " ~ % "~::s'~ ~ '*" l ~ ' ~ ~ Group A2 U3 __._ l0 20 30 40 0 20 40 60 80 (a) Tcl (b) Tc2 FIG. .~ .." 9 - O 9 9 .. . No further reproductions authorized. . The first part of the results are shown.4 0... --::-. ii -. . . 09 Z 0 Z U 0 0 20 40 6~ 80 iO0 120 Ir 0. 9 o*~0.....2 0. . . Group A1 c 9 I ..6 o18 11o Energy moment Tern ( 1 J s e c ) Amplitude ap (mY) FIG.. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. .2 0.o. ~0 m o I r r I r~ n r f 0 20 40 ~0 80 100 [20 ~40 0.. -.~. . ..| * .6 0.0 0. YAMAGUCHI ET AL.o.S. 9 -.'v.8 Energy moment Tern ( ~ s e c ) Amplitude ap ( mV ) FIG.i 0 0 c y c l e s ) a n d (b) T c 2 (1-200 cycles)....0 . . C bC Group BI " " " 0 C3 b3 L~ " t --." 9 f% L_ "l.0 0..D i s t r i b u t i o n s of T e m a n d a p in g r o u p A1 (1-53 c y c l e s ) at t e s t Tcl....... i0 -. . ... . 9149149149149 9 9 9 1 7 6 1 4991 4 9 . -~ I 1 I f I f f f _.'.. .. . . .D i s t r i b u t i o n s of T e m a n d a p in g r o u p A2 a t t e s t Tc2. .tl.. Copyright by ASTM Int'l (all rights reserved).. .A E g e n e r a t i o n r e l a t e d t o l o a d p h a s e d u r i n g cyclic loading t e s t o f (a) T c l ( I . ON GFRP TENSILE TESTS 139 9 : EVENTS = 1 : EVENTS i0 0 0 O O = . .4 0. . - r ~ (-.... & . . . ~.

7 (b) but the change was not so clear. Similar features of Tem distributions of final breakage AE and Other AE's were observed in the tests of unidirectional GFRP and silane treated cloth GFRP in the past [25]. small amplitude AE's could be observed at low load levels from No.7]. is believed to be more reliable than analysis by using fewer parameters such as peak amplitude and wave-energy alone. .I Z .ol .9 loading as shown in Fig. In this case.7.-. These AE's could be presumed as debundling AE's generated by cracking inside the transverse fiber-bundles." . The similar feature is usually seen on the peak amplitude distributions in many test results by conventional AE systems [15. Tem. multi-dimensional analysis using parameters. Large amplitude AE's increased slightly at No. and the scatter of the source locations also supports the presumption. two types of AE could be clearly sorted by the energy moment values." / . the Tem threshold of about 57. This type of AE can easily be discriminated from fiber breakage by zero-crossing number Nz and pea k amplitude ap as shown in Fig. Copyright by ASTM Int'l (all rights reserved). ap or Et. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.7 (a) could be presumed as AE's from matrix cracking corresponding to the whitening by visual observation.16." . 12 -. The concentration of small Tem value in Fig./ / I h -7 / .5 microseconds can be applied for the discrimination of the two failure modes. Most of the AE's of large Tem during No.l loading. IF. In the case of tensile test Tt2 at which silane treated woven cloth was used in the specimen (G-2).--. 140 ACOUSTIC EMISSION Peak [mY] Amplitude / / / . etc. From these results." ! / /! v y v v V Y v" 20 40 60 80 100 120 !40 Energy Moment Tem [psec] FIG.l-7 loading shown in Fig. / B 0 U _ 1 .9 loading. No further reproductions authorized.Two dim4nsional distributions of energy moment Tem by different ap values of AE's in group B1 during cyclic loading test Tc2 (specimen G-2). m . Nz." . DISCUSSIONS Classification of Failure Modes From the results of the loading-holding-unloading tensile test Ttl on GFRP shown in Fig.8 (a)(b). Supposedly the breakage of most fibers occurred at the final fracture of No.7 (b) seems to show a feature of the AE's generated from fiber breakage.

3--3 whitening low l o a d debundling 35--70 5--14 0. Two AE groups detected at close loading phases can be d i s c r i m i n a t e d if they have different values of w a v e f o r m parameters from each other. .10 or Fig. the AE's which were observed at low or middle load level are p r e s u m e d to be friction AE's.2 level Table 7 shows examples of the t h r e s h o l d values for AE w a v e f o r m parameters applicable to the c l a s s i f i c a t i o n of the failure modes in the above cases. w h i c h c o r r e s p o n d to the same load phase in successive cycles. YAMAGUCHI ET AL. ON GFRP TENSILE TESTS 141 TABLE 7 -. were observed as shown in Fig.2--10 breakage deb0nding & matrix cracking 55--100 8--14 0.3--1. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.ll. Similar features have b e e n observed on the AE's in metal and they were presumed to be g e n e r a t e d from friction on crack surfaces [8]. p r o b a b l y the high peak load (about 80% of the estimated m a x i m u m fracture load of the specimen) caused some fractures near the peak load of m a n y cycles.9 (a) and (b). 1 0 and Fig. In the case of Tc2. Copyright by ASTM Int'l (all rights reserved). In the case of GFRP.An example of t h r e s h o l d list on w a v e f o r m parameters for fracture mode p r e s u m p t i o n in GFRP by AE grouping. The test Tc2 of w h i c h peak load was higher than that of Tcl p r o d u c e d the AE's of B1 near the peak load. The AE's of group B1 which were g e n e r a t e d near the peak load of each cycle in the Tc2 cyclic loading test showed d i f f e r e n t features from those of group A1 and A2. High u n i f o r m i t y of the AE g e n e r a t i o n d u r i n g specific loading phase in successive cycles supports this presumption. these friction AE's usually showed large T e m and small ap as shown in F i g . fracture modes Tem Nz ap generation (presumed) [ F sec ] [150 F s e c ] [=V] stage fiber final breakage 35--60 10--20 1. In the cyclic loading tests as shown in Fig. No further reproductions authorized. The comparison will be d i s c u s s e d in the following section.9. while almost no AE was g e n e r a t e d near the peak load of the test Tcl. Many rows of AE generation.ll. A l s o the AE w a v e f o r m parameters in a g r o u p such as A1 or A2 show c o n c e n t r a t e d d i s t r i b u t i o n s as Fig.

fracture surfaces which cause friction AE can be very complicated in composites. delamination and fiber breakage. 142 ACOUSTIC EMISSION The distributions of parameters (Tem. Further experiments and analysis using multi-parameter would probably make the characterization definitive. Now.e. (i) small(Tem)-small(ap). but the number is not so many. as shown in Fig. For example. For example.g. Comparison of Friction AE's Between Metal and FRP In the case of cyclic fatigue tests of FBR SUS 304 piping components.7 (b) and Fig. matrix cracking. The type of (3) small(Tem)-large(ap) showed approximately the same values of those parameters obtained at the final breakage of the tests Ttl and Tt2. It would be important to make clear which characters would be proper to presume the behaviors in the case of FRP. they are supposed as the most noticeable characters of friction AE from metal. These characters (i). The other items from (4) to (6) could not have been observed at these simple cyclic tests of GFRP. debonding. the difference of material characters between metal and GFRP is very large. (i) high regularity of repetitive AE generated at the same load phase in every successive cycle. (2) and (3) seem to support the presumption of the friction AE in Tcl and Tc2. The similar phenomena to (7) were observed during the Tcl and Tc2 tests. Copyright by ASTM Int'l (all rights reserved). However. Therefore. ap) in Fig. the following features on AE generation have been observed [8]. some fiber breakages are supposed to have been included in group BI.8 (b). e. No further reproductions authorized. Also the other items are considered useful for discrimination of friction AE in metal. These coincidences of (i). i. . The phenomenon of the item (3) has also been observed in the test Tcl. (2) and (3) are considered very important and useful. groups of AE's in rows on successive cycles (2) almost identical waveform of AE's in a group of the same load phase (3) almost no detectable AE at peak load (4) major AE generations during load increasing phase (5) high correlation of generation phase drift to the AE peak amplitude generated in the same group (6) shifting trend of generation phase of the AE's in a row to lower load level (7) clear life span of a group or a row of repetitive AE generations The AE's of group A1 and A2 in the cyclic loading tests of Tcl and Tc2 showed the same features of the items (i) and (2) which are the most powerful reasons to presume the AE's as to be generated by friction.12 seem to show a few types of AE in group BI. (2) large(Tem)-small(ap). (3) small(Tem)-large(ap) and large(Tem)- large(ap). Some generation mechanisms can be supposed for these types of AE. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.

for his helpful advice to complete this paper.. the following methods can be considered important and advantageous for the design of instrumentation. for his cooperation on the tests and paper preparation. (3) wave-pattern classification by pattern parameters including moments. Tr. matrix cracking and debundling can be identified by multi-parameter analyses. former member of their laboratory. the authors are much thankful to Dr. etc. Tem(2nd). Tem. (2) real-time processing of wave envelope. (5) distributed high speed processing. ON GFRP TENSILE TESTS 143 SUMMARY AND CONCLUSION On-line observation of material behaviors or structures integrity under load is a notable advantage of AE technology. and his staff for their kind help and advice on the experiments in their laboratory. Some of the failure modes such as fiber breakage. and his staff for their suggestions on the tests and the specimens. ACKNOWLEDGMENTS The authors are much thankful to Professor Teruo Kishi of Research Center for Advanced Science and Technology. Katsushi Kobayashi. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Nz. Multi-parameter analysis using waveform microdata could be a good way for relatively wide applications to material test and structure monitoring using AE technology. Finally. Et. Hsu of National Institute of Standards and Technology. No further reproductions authorized. ap. For the improvement. Interesting similarities between GFRP and metal under fatigue tests have been obtained. e. it has a limited performance because it is an indirect method of observation. are being undertaken. The technique would be a powerful tool especially for the fracture behavior observation and for the characterization of composites. (I) waveform multi-parameter analysis. Further experiments and more detailed analyses using more parameters. The authors are also much thankful to Professor Isao Kimpara of Faculty of Engineering. (6) multi-option processings or analyses. University of Tokyo. . In the advanced acoustic emission technology. United States Department of Commerce. Nelson N. Also. friction AE in cyclic loading can be discriminated from the other types of AE. (4) zero-crossing count to compensate frequency information. The authors thank Mr. YAMAGUCHI ET AL. the AE technology should utilize more information of the detected waves. University of Tokyo.g. Copyright by ASTM Int'l (all rights reserved). On the other hand.

K. J.. A~ and Sakakibara. Pergamon Press. pp. Published by The Japanese Society for Non-Destructive Inspection (JSNDI). . Kishi. Vol. 1988. "Instrumentation and Data Processing for Acoustic Emission Technology and Applications. Blessing. of AECM-3 (Paris). H. "The MONPAC System.. K. Smith. Johkaji. J. of AECM-3 (Paris). S. pp.. of the 16th EWGAE Conference r London~ 1987). Eds. pp. Oyaizu.695-701 [13] Ohtsu." Proc. M. Blessing.432-441 [7] Li..l-10 [5] Oyaizu. 1989. . Vol. E.. of the 9th International Acoustic Emission Symposium (IAES-9) r Kobe). Farley and R. K. I.367-370 [9] Yamaguchi. Vol. Iwamoto. 1988. S.T. T. pp. 1989. M... K. I.Kimpara and Y. Y. P. pp.16-27 [4] Yamaguchi. K. "Fatigue Fracture Mechanisms of Short Fiber Reinforced PET Composites by Acoustic Emission Method.. 1989. pp. No further reproductions authorized.J.V.Yamaguchi. A. "Acoustic Emission: Theory and Practice." Proc. B . 1985. Eds. 1988. 1989. JSNDI." Progress in AE IV (Proc.781-787 [12] Suzuki. "Features of Acoustic Emission from Fatique Crack in FBR Piping Component and its Generation Mechanism. I. Oyaizu.. J.. and Ishida.. and Zhao. J. "Recognition on Fracture Modes and Behavior in FRP by AE Waveform Microdata.L.. L.." Proc. "Pattern Recognition Analysis of Acoustic Emission from Unidirectional Carbon Fiber-Epoxy Composites by using Autoregressive Modeling.I989. of AECM-3 (Paris). of IAES-9~ Kobe). "Advanced AE Instrumentation System for Versatile and Precise Use by Waveform-Microdata Processing." Proc.. of the 2nd International Symposium on Acoustic Emission from Reinforced Composites (Montreal).Higo.4. 1989. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.3. pp. "AE Energy Signal Processing: An Overview. K. Yamashita. J. pp. No.. of IAES-9 F Kobe). and Ono. Published by American Society for Nondestructive Testing (ASNT). "The Monitoring Damage Growth Processes in Glass Fiber Reinforced Composite by Amplitude Analysis. Y. J.. P.Yamaguchi." Proc.6. 1986. of AECM-3 (Paris). and Swanson. H. T. M..W. ASNT.." Journal of Acoustic Emission. I.. pp.240-249 [Ii] Shiwa.Kimpara and Y." Proc. Eds.l.8. of the 7th International Conference on Non-Destructive Evaluation in Nuclear Industry (Grenoble).. "Acoustic Emission Signal Classification of Graphite/Polyphenylene Sulfide Composite Subjected to Mode II Fracture. J.Y. and Kobayashi.259-267 [6] Vahaviolos.268-277 [i0] Hoa. M. Imura." Progress in AE IV (Proc. M.. "New Direction in Testing. pp." Non- Destructive Testing (Proc. C . ASNT. ASNT.. 1988. K.. pp. of the 3rd International Symposium on Acoustic Emission from Composite Materials (AECM-3) (Paris). A. ASNT. Nichols Eds.Yamaguchi. Matsuo. pp.61-71 Copyright by ASTM Int'l (all rights reserved)." Journal of Acoustic Emission. COFREND and ASM. pp.Higo. T.90-95 [8] Yamaguchi. Conlisk.M." Progress in Acoustic Emission IV (Proc.. and Conlisk. J.Kimpara and Y. "Acoustic Emission Evaluation of Aramid Reinforced Aluminum Laminates.Higo.l-8 [3] Fowler. T.. The Society of the Plastics Industry (SPI). 144 ACOUSTICEMISSION REFERENCES [i] Scruby. H. 1987." Proc..C. JSNDI. K. and Jinen. J. No..2831-2838 [2] Fowler. and Yamaguchi.

of IAES-9 r Kobe). 1986.. Onoe. Dunegan and W." Composite Materials: Testing and Design (Seventh Con- ference) r ASTM STP 893." Progress in AE IV (Proc.. Lake Tohoe). 1985. "Monitoring Damage Progression in Center-Notched Boron/Aluminum Laminates Through Acoustic Emission. pp.Kimpara and Y. of WMAE. Ed.391-404 [21] McBride.M. ON GFRP TENSILE TESTS 145 [14] Madhukar. Hartman. 1989. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. I.s195-s198 [17] Maslouhi.H.J. (Proc. Vol. and Johkaji. "AE Role in the Diagnosis and Prognosis of Defects in Industrial Plant Steel Components." Progress in AE IV (Proc. JSNDI. pp. I..773-780 Copyright by ASTM Int'l (all rights reserved). Shimada. of IAES-4 (Tokyo). Hirai.N.D. pp. No further reproductions authorized. pp.Y.8. "Analysis of AE Signals in Time and Frequency Domains Coupled to Pattern Recognition to Identify Fracture Mechanisms in CFRP. Vol. and Awerbuch..Yamaguchi. Eds. and Roy.K. High Pressure Institute of Japan (HPI).. pp." Progress in AE III (Proc. J.59-66 [24] Yamaguchi. and Kishi. T. M.Yamaguchi.24-49 [19] Yamaguchi. C.L.. K. JSNDI. JSNDI.s4-s7 [22] Tonolini. and Kishi. (Proc.. and Eitzen.Higo. of the 2nd International Conference on Acoustic Emission. "Acoustic Emission During Load-Holding and Unload-Reload in Fiberglass-Epoxy Composites..Yamaguchi.. and Peters. K.Higo.l-2. "Multichannel Acoustic Emission Source Location System and its Application to Fatigue Test of Model Reactor Vessel.Kimpara and Y. of IAES-2 (Tokyo)... 1974. Vol. A. N. No.. HPI.. P.A New Acoustic Emission Signal Processing Method." Progress in AE IV (Proc. I... J. 1988. S. No. pp.. of IAES-9 r Kobe) . and Hamada. M. Dunhart Publishers. M." Journal of Acoustic Emission. Charlotte). and Nakai.s292-s296 [18] Yamaguchi.L. of IAES-8 w Tokyo). pp. (Proc. Eds. S. YAMAGUCHI ET AL.. of IAES-9~ Kobe). pp.554-563 [16] Shiwa. K.Yamaguchi.. "Higher-Order Crossings ." Journal of Acoustic Emission.J. . T.. ASTM. pp. K.4. Oyaizu. Enoki. Charlotte). 1988. Eds.Whitney. Yuyama.Higo." Journal of Acoustic Emission.8. 2/3. Eds.H. Eds. T. K. 1989." Advances in Acoustic Emission (Proc. of WMAE. Bowman. D. 1989.. K.8..337-367 [15] Shiwa. 1981. D. No. "Recent Developments in Acoustic Emission Testing of Chemical Process Equipment. Hamada. pp..Kishi. of International Conference on Acoustic Emission r California r 1979). pp.. Y. N. "Acoustic Emission Signal Analysis during Fatique Damage of GFRP. Ichikawa.Aoki and T.s62-s65 [23] Hsu." Proc." Proc.49-64 [20] Fowler. M. J.M. "Recognition of Fracture Modes in Fiber Reinforced Plastics by Acoustic Emission Waveform Parameters. No. "AE Source Location by Identification and Combination of Signals.T.Kimpara and Y.J. 1988. K. MacPhail. JSNDI. M.T. (Proc.F.. "Acoustic Emission Detection of Crack Presence and Crack Advance During Flight. F. and Shimozuma..T.82-I02 [25] Yamaguchi.K. 1986. of the World Meeting on Acoustic Emission (W~L~E). H. Pollard..l-2.S... pp.D." Journal of Acoustic Emission..G. Oyaizu.. 1978. "Acoustic Emission Source Location by Identification and Combination of Signals. Charlotte).l-2. Vol. H.

To accomplish this. It is shown. is battery-powered. Eds. several studies have investigated the feasibility of using acoustic emission (AE) to monitor the integrity of aircraft structural components during flight. AIE has the advantage of being capable of monitoring large components with a single sensor. in-flight monitoring. data acquisition INTRODUCTION Since the mid-1970s. Philadelphia. Current Practice and Future Dirwtiqns. ABSTRACT: Results are presented which show that it is possible to detect crack growth and crack presence in airframe components during flight. 146 Copyright9 1991 by ASTM International www. and can be conveniently used in hard-to-reach locations. P. T. This m u l t i p l e . L. MacPhafl. MacPhafl. Pollard. Paul S. D. J a m e s D. W. using a multiparameter criterion to identify signals originating at a crack in the presence of airframe noises. S. Peters) are faculty members in the Department of Physics at The Royal Military College of Canada. McBride and Major Pollard (assisted by Mr. ASTM STP 1077. Yamaguchi. J. Sachse. crack growth.astm. and K.org Copyright by ASTM Int'l (all rights reserved). These studies are completely catalogued in an a n n o t a t e d bibliography of acoustic emission [1. and MCpl. and Peters. Michael D. The principal problem of acoustic emission monitoring is the unambiguous identification of signal sources [eg.. MacPhafl. a data acquisition system has been developed specifically for in-flight applications. crack growth.2].. S. Roget. To accomplish this. structural noises). Ontario. Peters ACOUSTIC EMISSION DETECTION OF CRACK PRESENCE AND CRACK ADVANCE DURING FLIGHT REFERENCE: McBride. Dr. No further reproductions authorized. Kingston. that crack advance of less t h a n 1 m m 2 was readily detected during flight and the fracture-related acoustic emission signals unambiguously identified. Of particular relevance is the work of Scott [3]. Bowman. we have developed a data acquisition system specifically for in- flight AE monitoring. "Acoustic Emission Detection of Crack Presence and Crack Advance During Flight". AcoustiC Emission.c r i t e r i o n s y s t e m greatly e n h a n c e s the confidence level for the unambiguous separation of crack-related data from airframe noise. Canada K7K 5L0. Stuart L. Successful development of this technique will offer tremendous savings by reducing the need for major disassembly in order to inspect critical load-bearing components. David T. Hutton and Lemon [4[ and Scala et al [5]. crack face rubbing. D. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.. KEYWORDS: acoustic emission. and allows for dual-channel m u l t i p a r a m e t e r processing of the data during flight. 1991. This problem is addressed here. Pollard. This system is stand-alone. D. . is truly a passive technique. J.. Mr.. M. McBride. Bowman.. American Society for Testing and Materials. Bowman.

The lower of the two thresholds was selected to be -52 dB relative to 10 mV as measured at the preamplifier input. No further reproductions authorized. model R501. Final analysis and interpretation are accomplished using spread sheet software. digital data acquisition system used here was designed and constructed by us specifically for the recording and interpreting of acoustic emission data during flight. containing a well-documented crack. event risetimes for 6 dB change in amplitude. signal risetime. The type of integrated sensor used is Physical Acoustics Corporation. bandwidth 300 kHz to 1 MHz) located inside the sensor casing.General specifications for the RMC digital data acquisition system for in-flight acoustic emission monitoring applications. event decay times and event peak amplitudes. 1). These studies established the importance of the difference in arrival time of an event at different locations. Extensive screening of data. All of the above data are compressed into an event record which includes the time of occurrence of the event at each sensor.8]. along with the prototype data acquisition system. To provide m ax i m u m flexibility. the data acquisition system can be powered either by the aircraft electrical system or by batteries. The resulting data set is then extracted from the data acquisition system via an RS-232 interface and stored on disk on a portable personal computer. event durations. envelope followed and peak detected.5 crux 25 cm Weight 2kg Mass Data Storage transfer to laptop PC via RS232 interface Copyright by ASTM Int'l (all rights reserved). The output of the peak detectors and accelerometer are digitized by an A/D convertor and stored in memory (Fig. MCBRIDE ET AL. which is attached to the support frame in the i nst r u m en t at i o n bay of a Tornado aircraft. logarithmically amplified. TABLE 1 -. EXPERIMENTAL The Data Acqui@ition System The dual-channel. All of these parameters are necessary to isolate crack related events from other noise sources during flight and are recorded by the data acquisition system used here. ON CRACK PRESENCE DURING FLIGHT 147 This s t u d y also i n c l u d e s the use of an inertially-loaded specimen. and the magnitude and variation of the applied stress at the time of occurrence of the event. The design is based on criteria derived from the RMC work of almost a decade in the area of acoustic emission monitoring during flight [6. Aircraft manoeuvres produce crack advance in the specimen under known g-loadlng conditions and with superimposed airframe noise. This test apparatus. the difference in arrival times at two sensors (At). Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The efficacy of these parameters was also considered by Hutton and Lemon [4] and Scala et al [5|. field analysis and interpretation can be carried out immediately. . The outputs of the two 500 kHz piezoelectric sensor elements are each amplified by an integrated preamplifier (nominal gain 40 d B . 2 Channels AE 60 dB dynamic range 2 Analog Channels 0-10 V full-scale deflection Power 10 Watts maximum Memory internally battery-powered RAM Dimensions 2 3 c r u x 13. The resulting signal is buffered.7. has been flight tested in both the Canadian CF-5 and the British Tornado aircraft. The output of each envelope follower is separately fed into the digital data acquisition system where the times of preselected amplitude threshold crossings 6 dB apart are recorded. Table 1 lists the general specifications of the apparatus.

1 threshold s t a t u s > i ' ~ ~ Ch. __ Pulse length Ithreshold Decay time -n i..S c h e m a t i c d i a g r a m o f t h e a c o u s t i c e m i s s i o n s i g n a l c o n d i t i o n i n g a n d d a t a acquisition computer. ._ Integrated amplitude r Transducers (to NO) Acoustic Emission Signal Conditioning Threshold Bank-Switched comparators Memory from Ch 1follower envelope 9 ~ / C h . 148 ACOUSTICEMISSION Logarithmic Envelope AE Amplifier Follower Inputs To threshold comparators J Buffer ~ Amplifier Peak Detector Peak . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.L At between channels Threshold setting ~ "1cur 4 channE A/D Ib. 1 -.sK . Channel number from Ch 2 9 Peak amplitude envelope follower Load Accelerometer Load Accelerometer / I1 ~1 I time Time of AE event Peak amp Ch. 1 / Ires' ~176 Risetime Peak amp Ch. 2 iselec t -. serial link to monitoring computer Data Acquisition Computer FIG.ovo. Copyright by ASTM Int'l (all rights reserved).

9 kg weight being subjected to the aircraft acceleration directed through the base of the inertial loading apparatus. model EGD-240) mounted in the loading frame support block. Silicone fluid provides proper acoustic coupling of the fatigue specimen to the loading frame to ensure that airframe noises are transmitted to the fatigue specimen for detection by the acoustic emission sensors. Inc. The inertially-loaded specimen was acoustically coupled to the aircraft support frame in the instrumentation bay. The stress required to propagate the crack is applied to the specimen via the coupling pin. FIG. This is achieved as a result of the lever action of the 0.5 g perpendicular to the plane of the aircraft The aircraft acceleration is sensed by an accelerometer (Entran Devices.. 2 -. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.Schematic diagram of the inertial loading apparatus and precracked 7075-T651 aluminum test specimen. Copyright by ASTM Int'l (all rights reserved). .9 kg inertial load was subjected to an acceleration in excess of 3. ON CRACK PRESENCE DURING FLIGHT 149 The Inertiallv-Loaded Specimen Figure 2 shows a schematic diagram of the inertially-loaded 7075-T651 a l u m i n u m fatigue specimen clamped in the inertial loading frame support block. No further reproductions authorized. the test specimen was precracked to a crack length which would cause crack propagation when the 0. MCBRIDE ET AL. Prior to mounting in the loading frame.

5 m m d i a m e t e r pencil f r a c t u r e a s a s i m u l a t i o n source. RESPONSE MAPPING OF INERTIAL LOADING APPARATUS -k 40 Base of + inertal-loading + ~'30 + f r a m e ~ + + -- On sample +~+ + " outside sensor E 20 + + ++ + :. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.5 • 2 B a s e of i n e r t i a l . .+ ++ T + + ++ :t+ . T h e wide r a n g e of v a l u e s of At a n d r i s e t i m e for s i g n a l s o r i g i n a t i n g i n t h e b a s e of t h e l o a d i n g f r a m e is d u e to m u l t i p l e reflections within the structure. P o s i t i o n of S o u r c e At 0Jsec) r i s e t i m e (~tsec) Crack location 24 + 2 3 + 2 O n specimen.= Crack ~array + ~ ~ . a s i n d i c a t e d i n T a b l e 2 a n d s h o w n i n Figure 3. 150 ACOUSTICEMISSION System Calibration A d e t a i l e d c a l i b r a t i o n of t h e a c o u s t i c e m i s s i o n s y s t e m ( i n e r t i a l l o a d i n g f r a m e . The m e a s u r e d arrival t i m e differences (At) a n d r i s e t i m e s are listed i n T a b l e 2 for v a r i o u s locations. outside s e n s o r a r r a y 41 + 4 2. ~+[ : -H-. fatigue s p e c i m e n .M e a s u r e d v a l u e s of At a n d signal r i s e t i m e for v a r i o u s locations. 3 -. p u l s e d YAG l a s e r a n d f r a c t u r e . T h e s o u r c e s i g n a l s u s e d are t h e h e l i u m g a s jet. D a t a r e c o r d e d d u r i n g fatigue p r e c r a c k i n g of t h e s p e c i m e n r e s u l t e d in At=24 • 6 llsec a n d r i s e t i m e of 3 • 2 llsec for c r a c k a d v a n c e e v e n t s o c c u r r i n g a t m a x i m u m load. TABLE 2 -.r e l a t e d e v e n t s f r o m i n c o m i n g a i r f r a m e noise. No further reproductions authorized. Copyright by ASTM Int'l (all rights reserved).r e l a t e d e v e n t s generated during fatigue crack growth and overload in the laboratory. I ' I ' I ' I 0 20 40 60 80 100 At (psec) FIG. T h e s e two p a r a m e t e r s were extremely efficient a t isolating c r a c k . s e n s o r s a n d d a t a a c q u i s i t i o n a p p a r a t u s ) w a s c a r r i e d o u t for s o u r c e s i g n a l s i n j e c t e d a t v a r i o u s locations.M e a s u r e d m e a n v a l u e s of At a n d signal risetime for v a r i o u s locations.~~Jp 9 location + n" 10 k + + +++ + 9 :[: 0 ' I It .l o a d i n g f r a m e 65 _+ 15 16 + 15 50 . p e n c i l lead f r a c t u r e . F i g u r e 3 s h o w s t h e m e a s u r e d r e s u l t s o b t a i n e d u s i n g 0.

. C o m p a r i s o n of Figures 3 a n d 5 show t h a t the majority of detected signals arrive at the sensors via the base of the inertial-loading frame.63 m m 2 resulted from the first overload c r a c k advance (4. only those events with At = 24 + 6 llsec and risetimes of 3 -+ 2 psec were accepted as crack-related events.3 g and 5.~ 2. These include all sources (crack advance. Copyright by ASTM Int'l (all rights reserved). These events result from crack advance. respectively.5 g.2 g.0 I i I I I I I I 0 10 20 3O 4O 50 60 70 8O T i m e ( m i n rel to t a k e o f f ) FIG.3 gmax manoeuvre).5 g. 67 m i n 38 sec and 68 m i n 30 sec. Figure 6 s h o w s the effect of applying these conditions to the data of Figure 4. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. MCBRIDE ET AL. ON CRACK PRESENCE DURING FLIGHT 151 The Test Flight Results Figure 4 shows the test flight profile as m e a s u r e d by the d a t a acquisition system.. No further reproductions authorized.7 g.3 gma~manoeuvre. crack face rubbing and airframe noises. 5.o | @ o r ~ lI r 0. 4 -. Included in the flight profile are three aircraft m a n o e u v r e s at 15 m i n 12 sec.Aircraft acceleration as f u n c t i o n of time for all of the events detected during flight. Fracture of the specimen occurred during the 5. E a c h recorded data point corresponds to the detection of a n event.o 2 3. Hence.p a r a m e t e r filter. c r a c k face rubbing and superimposed airframe structural noises). relative to takeoff. These m a n o e u v r e s resulted in successively increasing m a x i m u m g values of 2. Note t h a t all b u t the final two high-g m a n o e u v r e s are removed from the d a t a set by t h e d u a l .0 ~ TESTFLIGHTPROFILE O1 "f-" 4. respectively. The m a x i m u m load u s e d to p r e c r a c k the s p e c i m e n c o r r e s p o n d s to t h a t applied b y the inertially loaded s p e c i m e n for a n aircraft a c c e l e r a t i o n of 3. Figure 5 s h o w s the s c a t t e r plot of signal risetime and difference in arrival time of each acoustic emission event and noise signal detected during the test flight (Fig. 4) for c o m p a r i s o n with the calibration d a t a (Fig.0 i o . Optical and electron microscopic e x a m i n a t i o n of t h e fracture surface revealed a n increase in c r a c k face area of 0. This is the 4. 3). and applied a s e q u e n c e of increasingly large s t r e s s e s to the fatigue c r a c k via inertial loading. t h e first overload occurring in the s p e c i m e n during flight is identified with the first m a n o e u v r e which exceeded 3.2 ~ manoeuvre. 7 P.0 . 4. Based on this comparison.

o .0- l.0 ' I = I ' I ' I ' I I I I 0 10 20 30 40 50 60 70 80 Time (rain rel to takeoff) 5. Note t h a t t h e s e l a t t e r e v e n t s o c c u r only d u r i n g p r o g r e s s i v e l y h i g h . .0.Signal risetime a s a f u n c t i o n of difference in arrival time.. r oo 0 o . 0. _ o ~ ? o 0 o I ~ ! =o 40 ~ 1o l& At (psec) FIG. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.1:: 2.o.0.< 1. 6 -.C o m p a r i s o n of t h e o c c u r r e n c e of all e v e n t s d e t e c t e d d u r i n g t h e t e s t flight ( u p p e r graph) with t h e o c c u r r e n c e of e v e n t s u n a m b i g u o u s l y identified a s c r a c k .%.E o oo 0 o o 0 0 0-0~176 o o 20 ~o o o o oo o o .g m a n o e u v r e s w h i c h provide t h e s t r e s s e s r e q u i r e d for c r a c k advance. Total of 30 Events e- o At = 24 • 6 l~.0 i .0.0- p .0 _ .ram ~F>oo $ o <>o ~.~ 4. ro : 0. ' I ' I ~ ~ ~ 40 ~o ~ . No further reproductions authorized. l 5. o o ~ ~. ALL SIGNALS RECORDED I' DURING TEST FLIGHT 600 Events Detected r 4.~ 1.r e l a t e d (At = 24 + 6 psec a n d risetime of 3 + 2 ~Isec).o Time (rain tel to takeoff) FIG. Copyright by ASTM Int'l (all rights reserved).0- o 3.c O risetime = 3 + 2 psec 3. 5 -. 152 ACOUSTIC EMISSION 60 O0 O0 ~ 0 0 0 0r TEST FLIGHT SIGNALS 50 0 o o ~" 40 ~'~ 30 % 0 0 ~) o-.0 . CRACK-RELATED EVENTS AT HIGH-G MANOEUVRES @o ~.0- 'r< 2.

Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.r e l a t e d events.0 2.r e l a t e d e v e n t s (At=24 + 6 psec. ~ a A .4 ' I ' I ' I ' I ' I ' I 0.0.5 3. i . T h e u p p e r g r a p h i n c l u d e s all d e t e c t e d e v e n t s . No further reproductions authorized. X~.2-'10verloadManoeuvre | 67 rain 38 sec rel.. m 0. T h e d i s t i n c t i o n of c r a c k a d v a n c e events from other crack-related events can be accomplished using the Kaiser Effect criterion a s d e s c r i b e d in t h e text.0 Time (sec) . I n t h e lower g r a p h .4 m i = I m .'~ r '"7 CRACK-RELATED EVENTS 4. o n l y t h e c r a c k . w h i l e t h e lower g r a p h i n c l u d e s only t h o s e e v e n t s w h i c h a r e c r a c k .0 Time (sec) FIG. A 8 o A A~.3 gmax m a n o e u v r e w h i c h r e s u l t e d i n c r a c k a d v a n c e . Copyright by ASTM Int'l (all rights reserved).2.5 1. e a c h of t h e e v e n t s o c c u r r i n g i n t h i s m a n o e u v r e is identified s e p a r a t e l y a s c r a c k a d v a n c e or o t h e r c r a c k . 7 -. 67 min 38 sec rel.63 mm 2 z~ A a .6 l:-craAoancevon ] Other Crack-related Events < t v 3.0 1. t o t a k e o f f A.0 1~ 2.sec my 9~ .5 1.r e l a t e d s i g n a l s detected d u r i n g t h e 4.~ 4.0 0.r e l a t e d .5 3. m . MCBRIDE ET AL.8 3.3 gmax m a n o e u v r e s .0 2. A Z~ <{ 3. m .4 - TEST FLIGHT SIGNALS Overload Manoeuvre ~x~ ~ . r i s e t i m e = 3 + 2 psec)are s h o w n .8 A ~ 3.5 2.0 0.0 "1 risetime = 3 + 2 p. [ T o t a l of 17 Events 9 v o <o 3.0. .Lsec 9 V V (a 4. By r e q u i r i n g c r a c k a d v a n c e e v e n t s to b e r e s t r i c t e d to t h o s e w h i c h o c c u r a t l o a d s w h i c h h a d n o t b e e n a c h i e v e d p r e v i o u s l y (Kaiser effect). 4.C r a c k . to t a k e o f f O '= "1 At=24+61..~ crack increment ..6 < 3. ON CRACK PRESENCE DURING FLIGHT 153 Figure 7 s h o w s a n e x p a n d e d view of t h e d a t a w h i c h o c c u r r e d d u r i n g t h e 4.

CRAD 144688RMC01 and DAS Eng 84778ACCA01). 154 ACOUSTICEMISSION SUMMARY AND CONCLUSIONS A precracked 7075-T651 a l u m i n u m specimen m o u n t e d in a n inertial loading apparatus was subjected to stresses large enough to propagate the crack during flight. Defence Research Establishment . Stone of the Royal Aerospace Establishment.Pacific. Warton. Canada (ARP 3610-208. D. Miller. England. Farnborough. More detailed analysis was later carried out to confirm the results and to separate crack advance events from other crack-related events. The specimen and loading apparatus were part of a secondary structure bolted directly to the support frame in the i n s t r u m e n t a t i o n bay of the aircraft.J. under contract to Dr. Acoustic emission data and aircraft acceleration were recorded using a data acquisition system designed at RMC specifically for in-flight monitoring applications. and the in-service monitoring of failure-prone airframe structural components during flight. The selection of events which are unambiguously caused by crack-related sources was carried out using a very restrictive dual criterion (At = 24 + 6 ~tsec and risetime of 3 + 2 ~sec) derived from specimen calibration and comparison of the test flight data with laboratory crack growth results obtained for 7075-T651 specimens with geometry and sensor configurations similar to that of the specimen used for the test flight. The 30 crack-related events selected in this m a n n e r from 600 events detected during the test flight occurred as the result of high-g manoeuvres during which crack advance would be expected. The crack-related events were isolated immediately following the test flight. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. These crack propagating stresses occurred during specific test flight manoeuvres of a British Tornado aircraft. National Defence Headquarters. ACKNOWLEDGEMENTS Supporting funds were provided by the Department of National Defence. and Major W. The project was monitored by Mr. Continuing work in this area includes application of the RMC data acquisition and analysis system to the monitoring of airframe fatigue tests. Ottawa. .I~ Sturrock of the Materials Section. Copyright by ASTM Int'l (all rights reserved). W. Ontario. No further reproductions authorized.E. we confirm the feasibility of the unambiguous detection of crack growth and presence in 7075-T651 during flight. The Tornado test flight reported here was provided by British Aerospace. Thus.W. the monitoring of a slowly growing fatigue crack in a secondary structure during flight. Eleven of the 30 crack-related events are attributed to crack advance by application of a temporary Kaiser Effect criterion. Victoria. provided that detailed calibration of the structure is carried out. British Columbia.

H. A Bibliography with Abstracts. [8] McBride. [6] McBride. No further reproductions authorized. pp. ed. L. Review of Progress i n Quantitative Nondestructive Evaluation. "Effect o f Crack Presence on In-Flight Airframe Noises in a Wing Attachment Component". W. and Maclachlan. Copyright by ASTM Int'l (all rights reserved). 1979. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. P. "In-Flight Acoustic Emission MonitorinG of a Wing Attachment Component".. 229-235. Vol. October 1982. 3. and Maclachlan. "An Analysis of Acoustic Emission Detected During Fatigue Testing of an Aircraft". pp. J.. J o u r n a l of Acoustic Emission. and Bowles. I. I983. F. 2A. S. D. J. S. "Acoustic Emission Due to Crack Growth. October 1982. San Antonio. Proc. B. [4] Hutton. 223-228. and Maclachlan. R.A.. pp.J. Leonard. 709-718. T. I. N. S... Review of Progress in Quantitative Nondestructive Evaluation. F. New York. pp.~i0n. "Bibliography Update".K. J.. i . 210-219. J o u r n a l of Acoustic Emission. T. [7] McBride. 1985.i 0 . pp. Crack Face Rubbing and Structural Noise in the CC-130 Hercules Aircraft".. Plenum Data Company. J o u r n a l of Acoustic Emission. Journ~l 0fAcoustic Emi.E.G. Vol. W. [2] Droufllard.. J a n u a r y 1984. C. W. pp. .. Vol.M. 1. Vol. ON CRACK PRESENCE DURING FLIGHT 155 REFERENCES [i] Droufllard. Coyle. "Development of Acoustic Emission Methods For In-Flight Monitoring of Aircraft S t r u c t u r e s " . L. MCBRIDE ET AL.1982- 87. Acoustic Emission. L. and Lemon. [5] Scala.. Y. 459-470. 4B. 13th Symposium on Nondestructive Evaluation. Vol. [3] Scott. NTIAC. S.

Philadelphia. N." Acoustic Emission: Current Practice and Future Directions.W. 1991. make it very difficult to achieve reliable integrity evaluation. Wood and R. 2234. Yamaguchi. A.org Copyright by ASTM Int'l (all rights reserved). plant monitoring. A. R. and event capture capacity. Sachse. when combined with the operator's decision making ability in real time.. Harris are both scientists in the Commonwealth Scientific and Industrial Research Organization. American Society for Testing and Materials. Harris STRUCTURAL INTEGRITY EVALUATION USING AE TECHNIQUES RgFERKNCE: Wood. threshold settings. PMB 7. remnant life prediction INTRODUCTION. Brian R. ABSTRACT: Acoustic Emission structural integrity evaluation tests rely on the validity of the detected acoustic emission data and the ability of the practitioner to both analyze and interpret the data to make a correct decision on the future use of the structure tested. R. it is not possible to know that all the available data has been assessed properly. integrity evaluation. expert systems. Eds. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. and Harris. Roger. J.astm. W. B. The determination of the structural integrity of a plant or B. KEYWORDS: acoustic emission. W. Henai. Australia... ASTM STP 1077.S. Wood and Robert W. "Structural Integrity Evaluation Using AE Techniques. The equipment sensitivity. The main problem is to interpret the data and so any monitoring programs should involve Expert Systems in conjunction with both acoustic emission and other operational monitoring data to maximise the utilization of the large amounts of information now available from operating plants and structures. Unfortunately with acoustic emission surveillance testing. 156 C o p y r i g h t 9 1991 b y ASTM I n t e r n a t i o n a l ~. and K. . No further reproductions authorized. Division of Geomechanics.

Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The advent of new alloys and materials used in the construction of plants enables designers to use these materials closer to the upper limits of their physical properties. Thus there is a need to utilize acoustic emission (AE) technology to monitor many structures as part of a package of monitoring techniques. the details of the elastic wave propagation between source and transducer. No further reproductions authorized. electronic instrumentation. and the characteristics of the transducer and monitoring equipment. The result of all the previously mentioned approaches to plant operation is that without correct and accurate surveillance the community can expect an increase in failures of critical plant components. A factor in many situations is that while some maintenance programs may be restricted. The detected pulse will be different from that generated by a source as a consequence of the microstructure of the material. Temperature and stress changes are usually considered Copyright by ASTM Int'l (all rights reserved). so the interpretative procedures must take into account as many factors as possible. some will be reflected. BASIC THEORY. Some of the important factors that can affect any propagating elastic wave are the source attributes. The elastic wave will initially radiate as a bulk wave with the wavefront determined by the nature of the source and the elastic properties of the material (especially the extent of any anisotropy). and newer techniques are being introduced into plants and structures not originally designed for changing loads or process applications. structure geometry. higher loads are being applied to some older structures. WOOD AND HARRIS ON STRUCTURAL INTEGRITY EVALUATION 157 structure and the location of any defects are of prime concern for both statutory authorities and plant operators particularly as the current trend is towards more cost effective operations and quick returns on investments. One obvious example is the effect of increases in axle loadings of both road and rail freight vehicles on bridges which can be in excess of fifty years old. . The surface wave then can be detected by an appropriate transducer placed on or very close to the material surface. The transducer produces an electrical pulse which can be analyzed to provide information about the original acoustic emission source and hence facts about the structure. but any monitoring scheme will not provide sufficient information unless there is satisfactory interpretation of any acquired data. Simple analysis of the detected pulses may not always give results with a high degree of credibility for both source location and identification. transducer characteristics. wave propagation along boundaries. When ~he wave reaches an interface then some of the incident energy will be transmitted into the adjoining medium. and some will be converted into a surface wave which will propagate along the boundary. When a material or object is placed under stress then any defect which is activated by that stress will release energy and so act as a source of elastic waves. event location and pulse parameter measurements [i]. higher production rates are being expected from some older plants. wave propagation in the material. The main analytical techniques employed are event counting.

These effects should be measured prior to each monitoring program as the geometric and structural conditions will vary. Simple pattern recognition applied to the experimentally measured PSD's can provide further information about source attributes. The fact that the transfer functions relating the Copyright by ASTM Int'l (all rights reserved). propagation. Welds in the propagation path will result in interesting forms of waveform modification. Typical waveforms obtained from an FRP structure are a simple pulse (figure i) and one where reflections and other iruhomogeneities in the propagation path have considerably modified the pulse (figure 2). Then if these waves reinforce with other waves arriving at the transducer site by different paths. and detection of the signals. No further reproductions authorized. otherwise incorrect deductions may be made. . since they do not give a true energy value. is to consider the situation where an operator depends solely on the results obtained with equipment that measures only peak amplitude and rise and arrival times. The effects of dispersion (velocity of propagation is a function of frequency) together with the material and surface attenuation which are also functions of frequency will alter the propagating pulse. a situation can arise where components of the detected waveform will arrive after several reflections. A simple demonstration of the need for intelligent interpretation of acoustic emission data. A simple ray tracing approach will show that for a given geometry there will be many paths that allow wave propagation from source to transducer to occur so that as a result of the many reflections that can occur at interfaces. however this does require a wideband response and also critical spatial positioning of the transducers. The disciplines of electronic engineering and electronics cannot be simply used without regard to materials science and the nature of elastic wave propagation. The operator must have some understanding about the processes involved in the generation. some of the peak amplitudes in the resultant wave can have larger values than the peak associated with the first arrival. but only an indication of the energy released by the source. An electrical signal may be viewed in either the time or more familiar frequency domain to provide a framework to determine much of the information contained in a signal. A desirable end-point for all of the processing is to present the data in ways which will readily provide a real time indication of structural integrity. The power spectral density (PSD) of a source can be estimated if some knowledge about the transfer functions operating in the chain of events between source and transducer is available. 158 ACOUSTICEMISSION insignificant factors compared with the above under normal operating conditions. The results of the various analysis techniques applied to the detected acoustic emission signals form a knowledge base which is used to extract information about the nature of these signals. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. so that the approach is impractical for large transducer separations. Some parameters of use in the analysis procedures can be derived from the peak amplitude distributions of the detected pulses and for want of a better name these have been designated as indicative energies [2].

Figure 2. However. means that a true energy cannot be obtained. then valid comparisons both within a test and between tests are possible on the same structure. No further reproductions authorized. . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Alternatively the data can be plotted and continuously updated during a monitoring period so that areas of probable significant activity can be readily identified. provided all the equipment and the transducer locations remain the same. Simple AE waveform O --~50us4- O Time in Microsecs. These indicative energy values can be plotted as a function of time to predict maintenance requirements and estimate the remnant life of the structure. Modified AE waveform Copyright by ASTM Int'l (all rights reserved). are not all known or capable of measurement. Figure l. A graphical indication of structural integrity can be gained b y v i e w i n g these indicative energy values plotted as a function of an appropriate activlty/stimulus factor producing what has been termed a localised energy graph. O O t Time in Microsecs. WOOD AND HARRIS ON STRUCTURAL INTEGRITY EVALUATION 159 observed waveform to that generated by the source.

. and use this information to provide some estimate of the integrity of the structure.. I !. Hoisy mteNal IF~. The simple decision tree of figure 3 has only "high" and "low" for the amplitude of the AE.. A "clean" steel may generate less than 50 events before failure.. the largest peak amplitude and/or weighted average (or possibly some of the indicative energy values). The knowledge base for such an expert system in a controlled stimulus situation..... . will take as input pertinent parameters such as count rate...+'s quiet mteNal [ IFI : 60 ] Figu~ 3.. The values of this index are to a certain extent arbitrary and consensus needs to be reached on the choice of the values and their interpretation.. .... . ial I~ny events Nois] Mte~. ial High amplitude /.+... Ibis9 mteeial .. An example of the type of decision tree that may be followed is given in figure 3 where the situation is considered for constant AE during a stimulus hold period.P! : I Fev events ... "many" and "few" for the count rate. which will allow the externalization of the interpretive procedures into a computer based system to carry out the %ogical steps needed to make a valid assessment... . ie9 tNe (oe integeit9 assessment... .... High as~plitucle . Copyright by ASTM Int'l (all rights reserved)... Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized. 160 ACOUSTICEMISSION Eventually some of the extensive knowledge required by an operator will be encapsulated within the framework of an expert system. Quiet ~tes... . 9:-2" Quiet Mte~iat jm+ ] -::'.IIITj i '(:'. Quantification of these arbitrary values is very dependant on the realistic measurement of the AE source activity characteristics of the material concerned. An extended knowledge base would have many more options for the values and relate them to actual experimental data...ial t !P! :6e I I . ial 9 is9 mtel. together with information obtained during laboratory evaluation of the material of which the structure is made and plant measurements (temperature... while a fibre reinforced plastic would generate many thousands of events under the same load conditions.:2.. .........Lov a~lit~... and "quiet" or "noisy" for the material characteristics...mtel. pressure etc)......... . . The estimate of mechanical integrity has been expressed as an integrity failure index (IFI) which has a value of i00 for completely suspect and 0 for not suspect... Quiet .. Lov ~litude ..:...

This traumatic experience often results in significant damage being done to the plant or structure. and it is possible that up to 60% of the operational degradation may be initiated by transient operating conditions specifically in the first 48 to 72 hours of operation. No further reproductions authorized. Prior to start-up. The use of AE techniques to monitor structures often employs a i0 hour monitoring program which is about 0. The use of continuous or periodic AE monitoring using the techniques described previously and taking into account the plant material characteristics together with test procedures which recognise the Kaiser Effect. although other mechanisms such as fatigue and creep will both initiate and produce defect growth during plant operation. The exact nature of the regions of inflection and the approximately linear sections of the graph will vary for different situations. A typical form of the relative damage versus log time graph is given in figure 4. Under "normal" plant operation the source/defect activity which may be connected with mechanisms such as fatigue. is usually associated with sustained long term deformation. The results of such a program have been used to provide a current evaluation of structural integrity and so gain certification for future operation. makes it now feasible to use AE assessment as the basis for continuous plant operation (no statutory shut downs) when coupled with other NDT techniques. . On start-up the plant or structure is placed under considerable stress both localised and overall (particularly if it is a high temperature high pressure vessel). WOOD AND HARRIS ON STRUCTURAL INTEGRITY EVALUATION 161 PHILOSOPHY REGARDING AE MONITORING.02% of the life of a vessel in operation for 5 years. however for the testing to have a high level of confidence it is necessary for the structure to be subjected to a new peak stress level in the surveillance to overcome the AE load hysteresis phenomenon k n o w n a s the "Kaiser Effect". The growth of any defect or AE source will be dependant on the material properties and the plant operating conditions. Short term tests may be conducted with little prior notice and minor disruption to plant operation. creep. A number of plant monitoring programs and separate tests where vessels and samples have been stressed to failure indicate that there is a relationship between indicative energy 1 and indicative energy 2 which produces a local and/or overall indication of relative damage [3]. The most common use of AE is for structural integrity evaluation since the technique can locate active sources and provides some "information about them. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.1% of the operating period. Since any defect/source activity will vary depending on the surrounding stress it is not possible to get a definite measure of the damage/degradation of the structure. If the monitoring becomes an annual occurrence then the testing time rises to around 0. stress corrosion etc. if there are realistic plant operating conditions. any plant or structure should be in a new condition. Copyright by ASTM Int'l (all rights reserved).

a laboratory investigation of a structural steel. This analysis can be applied to data obtained d u r i n g a test p r o g r a m and may be related to both the whole structure or specific areas of interest for that test program p r o v i d i n g a short term evaluation of integrity.. integrity and remnant life for both specific areas of interest or the entire structure. 162 ACOUSTIC EMISSION After a p r e s c r i b e d period (typically 3 to I0 years d e p e n d i n g on the structure and its operating conditions) there will be a statutory shut-down and maintenance program to restore the plant or structure to a near new or defect free condition.. stability and failure. No further reproductions authorized.. Relative damage curve. . effects of operational excursions on the integrity. The values of the indicative energies will show no d i s c e r n a b l e trends for a defect free structure. however when significant defect activity occurs. AE m o n i t o r i n g during start-up and preferably also d u r i n g operation. Log(Ti~e) Figure 4.... then there will be trends similar to those depicted in figure 5 where parameters related to two of the indicative energies (the scaling will be material dependant) are plotted as a function of time where the stimulus is precisely controlled (e. As this a p p r o a c h is further investigated and validated it should be possible to estimate - the present structural integrity. and a vessel constructed from the same steel [3]. Copyright by ASTM Int'l (all rights reserved).. The next start-up will again cause degradation and so the cycle continues. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.i . abnormal changes in structural integrity.. p~dicted maintenance shutdown I da~ge ' Failu~ I .g. This interperative approach has been supported in a number of extended investigations on different FRP booms. hold period) and other parameters are still monitored such as the strain in FRP structures. will highlight and assess the severity of any source a c t i v i t y s o that operational conditions detrimental to structural integrity will be determined.. It may also use data from a number of tests on the same structure u s i n g the same transducer locations over an extended p e r i o d of time to provide a long term estimate of the operation. or some function of the applied stress [3]. Historical AE data will a l l o w the d e t e r m i n a t i o n of a cumulative indication of remnant life w h i c h can be graphed.. . An indication of the structural integrity and the remnant life appears to be related to the separation of the two curves with the cross-over point h a v i n g a special significance. timing of routine and preventative maintenance scheduling..f~". remnant life of the structure. . effects of changes in operating conditions on the integrity. ("7:) / ( a) ..m...

critical areas in an industrial plant. and in some cases the source may be identified. potential problem areas could be located. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The equipment would tend to be less expensive and dedicated to a specific use but capable of some diversification in its application. storage tanks. The extraction of meaningful conclusions from all the gathered data subsequent to the application of the various analysis procedures requires a distillation of the methods used by a skilled practitioner in the field of acoustic emission. Otherwise important data can be missed since it is not always possible to intentionally over-stress the structure to stimulate defect activity and hence important data will arise at unpredictable times. The analysis and interpretation of the indicative energies and remanent life evaluation will only be a viable tool after the correct relationship between material properties and AE activity is better defined and wave propagation factors are included in any AE analysis techniques. Data from permanent monitoring systems would give advance warning of areas requiring inspection. Since many of the steps used in the interpretation of the data involve Copyright by ASTM Int'l (all rights reserved). FUTURE DEVELOPMENTS. it is preferable that continuous monitoring be employed. WOOD AND HARRIS ON STRUCTURAL INTEGRITY EVALUATION 163 laai~ative Ener~7~ ~ IIlIIIIIlltlIIIJlllIIllllIIIIllIllllllIlllllIlilllIIIIlll Lo~Time Figure 5. process control. The results of permanent monitoring should be used to plan both the requirements and needs of maintenance and shut down programs. civil structure monitoring. Structural Integrity Evaluation Graph. The future development of AE techniques is most likely to be in long term monitoring or permanent monitoring of structures such as pipelines. and geotechnical applications. pressure vessels. Although it is possible to monitor entire plants and/or structures at regular intervals over long periods of time. . No further reproductions authorized.

A. R. there will be an increasing use of expert systems to provide. Acoustic Emission team who participated in both the laboratory and field test work associated with this report. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The authors wish to express their appreciation to the other members of the C. 1988. 164 ACOUSTICEMISSION some form of decision tree. If a sufficiently large amount of data about actual waveforms and the associated conclusions becomes available. ~ C E S [I] Harris. 1982. and computer based pattern recognition will play an important part in this work. and Interpretation of Acoustic Emission Signals. "The Detection. 4." Metals Forum (Australia). A. 373-381. B. [2] Wood. W. in a computer literate form. R.. pp. the thought processes used by the skilled operator. L. The Japanese Society of NDI.O. R. R. 125-135. . Copyright by ASTM Int'l (all rights reserved). and Wood. R.. The Japanese Society for NDI.R." Progress in Acoustic Emission IV. B. B.. ACKNOWLEDGEMENTS. W.I. then it may be possible to use a learning algorithm such as that associated with neural networks to provide a workable computer hased interpretative procedure. M. pp. [3] Wood. 5. Noyes. "Acoustic Emission Applied to Railways. "Comparison of Laboratory and Field Tests During Acoustic Emission Monitoring of Pressure Vessels.. and Harris. 210-215. R. The extraction of parameters that can be used to provide useful descriptors of the complex waveforms that are observed is still a fruitful field for research. and Harris. 1986.." Progress in Acoustic Emission III. No further reproductions authorized.S. Transmission. No. Vol. W. A. pp.

inverse problems. KEYWORDS: Neural signal processing. learning. 14853 . neural networks. It is shown that the characteristics of the source can be estimated from the AE signals or vice versa by the auto-associative recall from the correlation memory. c. Sachse. Yu- goslavia and Wolfgang Sachse is a professor in the Department of Theoretical and Ap- plied Mechanics at Cornell University in Ithaca. University of Ljubljaaa. E. No further reproductions authorized. non-destructive testing tool is becoming well established. W. professor on the Faculty and grad- uate student of Mechanical Engineering. Eds. Acoustic Emission: Current Practice and Futur e Directions. A S T M S T P !07_7. I. process monitoring INTRODUCTION The use of acoustic emission (AE) measurements as a powerful. Philadelphia. w .f. simulated source of discrete acoustic emission events in a block of material and to process the AE signals generated during a metal drilling operation. ]65 Copyright9 1991by ASTMInternational www. Data is input into the system as a vector composed of either AE signals or their spectra and encoded information about the source. and Govekar. New York. E. Igor Grabec. Experiments are described which utilize a system running on a minicomputer to process signals from a localized. This is performed by presenting experimental signals to the system and adap- tively forming a memory whose output is an autoregression projection of the input.. Ljubljana. American Society for Testing and Materials.org Copyright by ASTM Int'l (all rights reserved). source characteriza- tion. quantitative AE. Discrepancies between the input and output are applied in a delta learning rule. The mapping of AE signals from the sensors to the descriptors of the source and vice versa is accomplished by learning in the system.U S A. The processing system resembles a neural network including an associa- tive memory. "Solving AE Problems by a Neural Network". [1]. Roget and K. Ya- maguchi. Wolfgang Sachse and Edvard Govekar SOLVING AE PROBLEMS BY A NEURAL N E T W O R K REFERENCE: Grabec. . J.. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.. Essential to its success in many Igor Grabec and Edvard Govekar are respectively. ABSTRACT: This paper summarizes recent work utilizing an adaptive learning system to the characterization of acoustic emission phenomena.astm. K. adaptive systems. 1991. Sachse.

yet at the same time avoid the difficulties related with an exact Copyright by ASTM Int'l (all rights reserved). called a deeonvolution. Various studies of the cutting process have shown that the spectra of the emitted acoustic signals have very complex structures [6]. the chaotic dynamics associated with the process is too complicated to be easily analyzed to obtain a theoretical basis of the emitted acoustic signals and their relation to details of the processing operation [7]. the determination of the response characteristics of the specimen using elastodynamic theory and a well-characterized. This is of practical importance for an automatic identification and control of the process.3]. No further reproductions authorized. In this approach. the sound generating process is continuous and it is therefore convenient to use a spectral distribution of the emitted signals for the description of the process. The ultimate goal of such an approach however is to make a reasonable estimation of the parameters related to the manufacturing operation from the AE signals. In most cases. the multi-component nature of many important AE sources and elastodynamic phenomena present serious obstacles for the practi- cal implementation of quantitative AE procedures in most non-destructive testing applications. as for example the cutting of metal. This generally represents a so-called ill-posed problem requiring some sort of regularization in order to permit a stable processing of the experimental data [4]. The last thirty years have seen the increasingly rapid development of various systems for sensing and processing of AE signals by which various AE source phenomena can be detected and characterized. of the detected signals [3]. One is thus usually forced to rely on an empirical description of the process. The success of this approach depends critically on the micro-mechanical modeling of AE sources. a major emphasis of recent research has been towards the development of quantitative signal analysis tech- niques by which the detected ultrasonic signals can be processed to recover the detailed characteristics of the source or the properties of the medium [2. . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. In order to preserve the preciseness that is inherent in a quantitative description of AE phenomena. the source characteristics are recoverd from the detected AE signals using an inversion procedure. an analysis of the generated AE can be expected to provide important information related to the process [6]. But even when a suitable mathematical method of regularization is found. 166 ACOUSTICEMISSION applications is a signal processing procedure by which information about the source of emission can be extracted from the signals detected at one or more sensors on the surface of a structure. broadband signal detection system. Many manufacturing processes. In order to establish a direct connection between the parameters describing a source and the AE signals detected by a sensor. Therefore. The source information includes its location and charac- teristics where the latter are directly related either to failure processes occurring in the material comprising the monitored structure or to changes in a process which is being monitored. It would therefore be of advantage to find a quantitative method of analysis by which a rigorous elastodynamic treatment of the inverse problem could be avoided. Even for the simplest case of orthogonal cutting. are efficient generators of sound. One example is the minimization of the mean noise energy which leads to a so-called optimal deconvolution [5].

a series of various acoustic. ON SOLVING PROBLEMS BY A NEURAL NETWORK 167 elastodynamic modeling. represents a brain. the brain then forms a n internal representation of images of the environment [15. Since this corresponds to the main task of AE analysis. an adaptive system. it is useful to review some of the characteristic properties of biological neural networks. caused by the interaction between neurons. visual. For example. If in certain situations the sensory signals provide only partial information about the environment. an adaptive AE analyzer must be capable of storing in an internal memory. the brain is capable of supplying the missing information from the image in its memory by an associative recall from the previously formed representation. it can generate an appropriate output from only a partial input. This information is presented to the system during learning and it is subsequently able to utilize this stored information for completion of partial data so as to obtain a good estimate of the features by which an AE event is characterized [8. Here "appropriate" has to be understood in terms of some measure of statistical error which is determined during an analysis of the system's performance. an intelligent being is usually able to imagine where the sound is coming from and what its source is. The ability of an animal to recognize events in its environment from the sounds it has sensed was developed during evolution. The purpose of this article is to review the structure of such a processing system and. For this task an animal needs no theory of sound generation and propagation or the corresponding physical models.11]. The same adaptive system as that used for the analysis of discrete AE events has recently been applied to the characterization of drilling and laser-based processes [13. In technical terms. in its most complex form. to demonstrate its applicability to the analysis of AE signals. when hearing a sound. By a self-organizing process. but rather. Although it is still not known exactly how the brain operates. only experience gained by learning. provided that the environment has not changed. resembling the structure of'a neural network has recently been proposed for obtaining approximate solutions of direct and inverse AE problems [8]-[12]. Before proceeding to describe the formulation of the system. using examples. while the internal image can be interpreted as the output from the network [15. touch and other signals from the corresponding sensors is presented to the neural network which. Systems of this type have been already extensively studied in relation to re- search of pattern recognition. one can therefore argue that in many NDT applications.14]. .16]. the sensor signals can be treated as a multi-component input to a neural network. The main goal of the previous work has been to find an appropriate learning procedure and a structure of the information processing system which facilitates this. GRABEC ET AL. One of the fundamental tasks of learning is then to adapt the response characteristics of the neural network so that when it is used. adaptive systems and artificial neural networks [15]- [19].16]. The system operating in this way is said to be capable of performing an auto-associative recall [15]. Much research has also been completed related to the various representations Copyright by ASTM Int'l (all rights reserved). No further reproductions authorized. information about various possible AE events. During this learning. one of its most characteristic features can be illustrated by an example. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.

f . position or force or from quantities which have been derived from these.20. kurtosis. f = G -1 . The source characteristics and the corresponding detected signals are related by the Green's tensor and a convolution in space and time. including its temporal as well as its spatial characteristics. FORMULATION OF THE SYSTEM We consider an AE system to be a linear. the signals input into an information processing system or obtained from it should be interpretable in terms of directly measurable physical quanti- ties. the time-dependent displacement signal u ( r . such as displacement. u (2) This last equation corresponds to the elastodynamic equation together with the applied boundary conditions. However. In order to achieve the latter. auto-associative processing system which resembles a neural network having multi-component inputs and outputs whose application is the processing of AE signals [8. energy. In view of all the aforementioned requirements. t ) detected at the sensor position r (-. u = G * f (1) The inverse of this expression permits a determination of the source function. the pre~processing is most frequently applied in order to present empirical data in a form most appropriate for the clas- sification of clusters [20]. That is. t) located at the source position r ~. 168 ACOUSTICEMISSION of signals needed to obtain an amplitude. No further reproductions authorized.21]. In order to facilitate a simple physical expla- nation of the results of such processing. the representation of empirical data by a set of quite arbitrarily selected statistical parameters is often used. it follows that all the information about the phenomenon can be best represented by a multi-dimensional. we wish to describe now in more detail an adaptive. A typical example is the representation of an AE event by its risetime. the results obtained by such processing of AE data often have no physical basis nor can they be compared with the detailed results obtained from a quantitative analysis of the AE phenomenon.x. time-dependent variable. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. the spectral density of the sensor signal. Be- cause there may be many transducers simultaneously used in an experiment. time and spatially invariant operation of the system [15. In this system. we first describe how an exact physical description of the AE phenomena can be approximated by a discrete system.11]. u] = 0 (3) Copyright by ASTM Int'l (all rights reserved). decay time. In the field of pattern recognition. c -1 . pulse duration etc. This is written in simplified form as. f] . which all describe a point in a multi-dimensional feature space [22]. The required data transformations can be performed in the pre-processing portion of the system or included with the neural network proces- sor [21]. [c. elastodynamic system. as for example. Both of the previous equations can be written at once in the form [u. For this purpose. y z) results from the time-dependent force density field f ( r ' .. .

W. on average. No further reproductions authorized. W . Y = X . the dis- crepancy. The corresponding memory tensor then appropriately models the unknown response characteristics of the material. complete information about an ultrasonic phe- nomenon is supplied by the measured quantities. In an empirical treatment. . However. The inpu t to the system is the vector X and the corresponding output of the memory is the vector Y = W * X . The discrepancy V is the difference between the input and the estimator and it is given by V = X . GRABEC ET AL. sometimes called the pattern vector X(r. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. t) . It is therefore sometimes called the memory tensor. That estimator which is obtained after adaptation to all training input vectors is called the ultimate estimator. The problem is then how to determine the unknown tensor W .X (6) For an arbitrarily chosen memory tensor. the discrepancy generally differs from zero. According to the given definition. Eq. (5). The system consists of two essential parts: the memory and the subtractor. reduces to zero. f]. w . The output Y is an estimator of the input vector. If instead one uses the concatenated vector. at least for those input vectors which have previ- ously been presented to the system during learning. 1 is utilized. x = o (4) where the tensor W is equivalent to the generalized matrix expression W = G -1 0 (5) It is seen from the above that the concatenated vector quantity X is mapped back onto itself by the tensor W . From past research of adaptive Copyright by ASTM Int'l (all rights reserved). (3) becomes x .[u. The memory is represented by the tensor W which has the structure described by Eq. For this purpose the adaptive system shown in Fig. ON SOLVING PROBLEMS BY A NEURAL NETWORK 169 V=X-Y 0 0 Y=WX Recall: ~ = Figure 1: Schematic of an adaptive system. the vector X represents the measurable source quantities and the detected AE signals while the tensor W represents the response or dynamical characteristics of the material or structure. by an adaptive modification of the memory tensor.

. w w (9) With these finite-dimensioned quantities. ) (8) Furthermore. 170 ACOUSTIC EMISSION systems. In addition.uk. the convolution operation appearing in the previous equations becomes a matrix multiplication. X (N)) to which the memory can be simultaneously adapted. the input pattern vector is mapped into a reduced input vector. At the start of learning. . C is a constant and X r is the transposed input vector while the operator | denotes the outer product. the rate of adaptation can be controlled. However. . where IlZll denotes the average Z2-norm of all the input samples. f 2 . Instead.fl. where X(r. we choose W -. W Rxx = 0 (10) Copyright by ASTM Int'l (all rights reserved).. in contrast to that of Eq. IIXII2). (5). No further reproductions authorized. the funda- mental scheme of the adaptive system's operation remains essentially unchanged. We therefore must assume that the mea- sured data is mapped onto a finite number of digitized experimental data from the displacement signals and the corresponding source data.. Generally. W. We therefore assume that there is available a series of N experimen- tal samples (X (1). ( 7 ) b y C < 1 / ( N . f . That is. various rules of adaption have already been discovered [15]-[21]. X(2). X.f] ~-+ X = (Ul.t) = [u. there exist no computers possessing a continuous. . has non-zero elements also on its diagonal elements. with the simplest one given by the delta learning rule which is expressed as •w = c (v | x T) (7) Here. . In actual experimental situations the input data is never detected simultane- ously everywhere in space nor at all frequencies for all time. U2. During learning...0.. A statistically equivalent treatment of all samples is achieved if the com- plete series of experimental sample data is presented many times to the system. distributed memory by which the memory tensor W could be calculated. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. the matrix W converges to the solution of the equation Rxx . (5) cannot be strictly realized in practice. Such iterative formation of the memory matrix is generally called learning. For this purpose it is reasonable to choose the adaptation constant appearing in Eq. By adjusting the value of the constant C . If there are k signal components and l source components.. with this assumption the memory tensor is also mapped into the reduced memory matrix. only one sample of the vector X does not contain sufficient informa- tion about the total response characteristics of the material and the measurement system. the system memory is described by a more general matrix which. Because of such an arbitrarily truncated description of the signals and because of possible non-linearities in the ultrasonic phenomenon itself. the structure of mem- ory matrix as described by Eq.

they were continuous. The example s t u d i e d was the ability of the p r o p o s e d s y s t e m to learn and to analyze AE signals excited at several fixed locations by forces acting in various directions. Then. In one corner of the specimen were m o u n t e d two miniature. amplifiers. in which Rxx is the correlation m a t r i x of the i n p u t vector X. GRABEC ET AL. waveform recorders or a s p e c t r u m an- alyzer and a c o m p u t e r with a p p r o p r i a t e d a t a acquisition software and processing software implementing the processing operations described in the previous section.3 mm. In this mode. No further reproductions authorized. EXPERIMENTS In order to d e m o n s t r a t e the applicability of adaptive s y s t e m we present here the results of analysis performed during e x p e r i m e n t a t i o n with two acoustic emis- sion phenomena. generates the e s t i m a t o r Y which. C h a r a c t e r i z a t i o n of DisCrete A E Events In the first t y p e of experiment. an input vector which is only p a r t i a l l y filled with experimental data. Such an o p e r a t i o n corresponds to the auto-associative recall of missing information [15. The d i a m e t e r of the sensors' active area was 1.W 9X therefore represents a linear autoregression in the multi-dimensional space of the i n p u t vector variable X. ON SOLVING PROBLEMS BY A NEURAL NETWORK 171 Sensor #1 "~ Sensor #2 Figure 2: Sources and transducers on the corner of an a l u m i n u m block. the feedback to the m e m o r y is t u r n e d off. T h e e s t i m a t o r Y -. 2. . The test specimen was an alu- m i n u m block of dimensions 450 x 180 x 350 m m 3. The sources were activated by breaking a pencil lead of d i a m e t e r 0. After learning is complete.16]. in t u r n . the A E signals were of discrete type. sensors. T h e experiments were performed using a setup consisting of a specimen. the system can be used as a t r a i n e d analyzer. T h e solution of this equation is a projector to the eigenvector of the correlation m a t r i x corresponding to the largest eigenvalue. These are i n d i c a t e d b y Copyright by ASTM Int'l (all rights reserved). while in the other. b r o a d b a n d piezoelectric transducers as shown in Fig. will include also an e s t i m a t i o n of the missing data.5 m m act- ing in various directions at each of the source locations. the processing s y s t e m was applied to solve a forward and an inverse AE problem. In one of these. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The s t r u c t u r e of the a d a p t i v e system described here is analogous to a simple neural network in which the synaptic joints between neurons represent m e m o r y elements.3 m m and of length 2.

Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. evenly extracted from 1024 point-long time records. fs. 3(a) which practically coincide with the lines representing the input signals. . The complete experimental data for each AE event was thus represented by a pattern vector comprised of 150 components. No further reproductions authorized. 3(b). . f 2 . 172 ACOUSTICEMISSION Figure 3: (a) The set of concatenated training signals and the corresponding output signals (superimposed). while 22 components were appended for encoding the source data by two descriptors. the norm of the discrepancy V fell below 10 -4. fs0 in the figure. The training set of pattern vector samples is shown in Fig. The signals from the transducers were amplified by 40 dB and recorded by a two-channel waveform digitizing system. Coincidence of both data sets indicates a good adaptation of the system. (b) The memory matrix corresponding to the synaptic weights of the neural network. The digitiza- tion rate was 10MHz. The sample pattern vectors were composed of digitized data from both sensors and data about the orientation of the released force and the location of the source point which was entered via the keyboard. These sample vectors were then presented to the processing system in twenty iterations. This is represented by the dotted lines in Fig. The source position and the direction of the force were encoded by assigning to each a constant corresponding to an appropriate component out of eleven available while the others were set equal to zero.and twelve cross-correlation fields of the segments comprising a sample vec- Copyright by ASTM Int'l (all rights reserved). The AE signal from each sensor was represented by 64 components. The initial portions (left side) of each record correspond to the signals from both transducers while in the latter portions (on the right side) are the encoded source descriptors. . 3(a). The components of the memory matrix W can be treated as a data record so that the entire matrix can then be graphically represented as shown in Fig. . During the adaptation. This indicates that a very accurate reproduction of the input signals was attained. . making it nearly impossible to delineate between the training and learned signals. The displayed data consists of sixteen characteristic regions corresponding to four auto.

(b) The set of output signals including the recovered source descriptors of position and force orientation obtained from the data set (a). 4(a) and 5(a) while the corresponding output signals are presented in Figs. The results of Fig. ON SOLVING PROBLEMS BY A NEURAL NETWORK 173 AE SIGNALS SDUACE DAFA RECOVERED T = 1. the missing AE signal portions in the input vectors were recalled by the source descriptors as shown in Fig. These input signals are shown in Figs. The recovery of the source descriptors is a consequence of the cross-correlation portions present in the memory of the system. For the solution of the forward or inverse problems. This recovery corresponds to an approximate solution of the forward elastodynamic problem in which AE signals are determined for specified characteristics of the source and the dynamic response of the sample. the output represents an optimal mixture of those vectors of the training data whose source descriptors are similar to those in the presented input vector of the test data.5 Figure 4: (a) The set of concatenated input signals. 5(b). However. representing the partial infor- mation supplied to the adapted system. we present here the results of two types of outputs excited by artificially truncated input pattern vectors. corresponding to those formed from the prod- ucts of the signals and the source coordinates. 4(b) and 5(b). No further reproductions authorized. the system can be used as an adapted analyzer. We note that the recovered AE signals do not correspond exactly to the learned signals.5 = 1. The apparent corruption by noise is a consequence of the similarity between the training data which are not fully orthogonal to each other. respectively. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. To show this. The first type of input vectors included only the signals from the transducers while the second type was formed using only the source descriptors. tor. 4(b) demonstrate that the source descriptors of source position and orientation are correctly recovered in the processor's output signals. In the second type of processing. the off-diagonal or cross- correlation elements of the matrix. It is also characteristic that in the elements of the vector containing the encoded source data. disturbances appear in Copyright by ASTM Int'l (all rights reserved). This example demonstrates the use of the processing to obtain a solution of the AE inverse problem. . GRABEC ET AL. After the learning has been completed. are essential.

A compromise between an exact description and the heuristic empirical method presented here leads to an optimal multi-component deconvolution. The drilling of a mild steel plate was selected as a representative processing operation. 174 ACOUSTICEMISSION SDURCE DATA BE SIGNALS RECOVERED I = l. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. A time-invariant operation of the system will be demonstrated in the following example. it is only these parameters of the source that can be recalled from the memory.4 Figure 5: (a) The set of input signals. A further limitation of the example above is that the processing was not time-invariant and so the described procedure is only applicable to process those signals in which a proper triggering of the waveform recorder is possible when recording the AE signals. Unfortunately.S T= 0. similar to the one-component version [5]. . we are trying to avoid. The right peak represents the force direction and the left peak. the output signals indicating the similarity between patterns and their cooperative formation of the output data. (b) The set of output signals generated by the source descriptors in (a). one can optimally translate the signals in time [12] or assume that the memory has a multi- dimensional structure [21]. it appears that an exact treatment of this problem leads back to a rigorous description of the AE phenomena by elastodynamic models. Alternatively. The acoustic emission signals from a cutting process can be treated as a sta- Copyright by ASTM Int'l (all rights reserved). Characterization of Continuous AE A second type of experiment was performed to demonstrate the applicability of the proposed learning system to the analysis of acoustic signals emitted during a manufacturing process. comprised of the source descriptors. Both of these approaches are still under investigation. In the above example the time-dependance of the AE sources was neglected because the encoded source descriptors were specified only in terms of the source's orientation and its location. Its development for the analysis of AE signals remains a topic of future research. Hence. No further reproductions authorized. The AE signals are approximately recovered. as stated earlier. because of their complexity. which. its location.

The experiments were carried out on a steel workpiece of dimensions 145 x 60 x 15 mm 3 using a 3 mm di- ameter drill. one of the vector components was set equal to 1. The cutting speed was 55 m/min at the feed rate of 4 #m/revolution. 7. 8. The signals were normalized and then transformed into thirty components corresponding to the signal portion of the input vector. Therefore three repeat samples were taken to represent each operation. The corresponding typical spectral density records are shown in Fig. To these were added an additional 10 components to encode the logarithmic scale factor of the signal am- plitudes. while the remaining were left equal to 0. After learning. the system could be used as an adapted analyzer to which the AE signals were input and the missing descriptors of the process were estimated by the auto-associative recall. etc. GRABEC ET AL. at A and B. were statistically changing.. which were approximately logarithmically distributed over the spectral interval between 100Hz and 5kHz. For the learning. This encoding required an additional ten components of t h e pattern vector. This was applied as the input for the development of the auto-associative memory of the processor. In this case the amplitude distribution in that segment of the pattern vector representing the descriptor of the cutting process does not Copyright by ASTM Int'l (all rights reserved). The input vector was then completed from the keyboard by appending a descriptor related to the drilling process. because of the chaotic nature of the drilling process. Typical examples of pattern vectors from the training set are shown in Fig. then the recall is comprised of a mixture of the corresponding descriptors. ON SOLVING PROBLEMS BY A NEURAL NETWORK 175 Spectrum ~ Personal Analyzer Computer Ori. The features which were encoded in- cluded the drilling position. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. For each feature. I Workpiece I W 'B' Figure 6: Experimental setup used to investigate the drilling process. The representative signals from such a process are conveniently described in terms of their spectral density which is easily determined using a standard spectrum analyzer. The experimental setup that was used for this set of experiments is shown in Fig. the spectral density records measured over short time intervals of a few seconds. No further reproductions authorized. The drilling at site A was repeated with a worn drill. Training samples were recorded at free run and while drilling at two different sites on the plate. tionary random process over short time intervals. when the input vector is similar to more than one training vector. The normalized records were repre- sented by 20 components. However. The AE signals from the transducer were transformed in the spectrum analyzer into their corresponding spectral densities. 6. Not surprisingly. the same delta rule as used in the previous example was applied. The complete pattern vector thus consisted of 40 components. the drill condition. .

~6 0.002 0 . 176 ACOUSTICEMISSION < 1 0 -3 FREE RUN 5'x 3 2 1 0 0 2 3 4 .000 0~ 0. The vertical.004 0. i . 01 ORILIINC ON P O S I T I O N " B " i .002 0 2 3 4 . 0.~ B.~08 B. rkHz] . No further reproductions authorized.z] s Figure 7: The set of typical AE spectral densities recorded during various phases of the drilling operation. Copyright by ASTM Int'l (all rights reserved). linear amplitude scale is the spectral density in volts. . 2 3 " f [k. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.i D R I L L I N G ON POSITION "A" 9 t i 3 0 1 0 0.004 0. 01 YORN OUT DRILL ON "A" Q. [k.( I ~ .

which can be performed sequentially by standard maximum searching algorithms but this is not elaborated here. A signal from the previous layer. Fig. Because of laterial inhibition. 9 shows the records of the input signals together with the corresponding. a second layer was intro- duced into the network. the input vectors consisting of only pre-processed AE signals were supplied as input to the system. which interact by mutual inhibition [19]. the correct value of the descriptor cannot be eas- ily determined. while the others were inhibited to remain in a low excitation state. An estimate of the system's performance with this example shows that it produced an approximately 90% correct estimation of the process descriptors. ON SOLVING PROBLEMS BY A NEURAL NETWORK 177 LEN~NINING INPUT SISNAL$ a} F r e e nun b} Onl]llng on position "A" c} Onilllng on position 'B' d) Worn out drill on 'A ~ ~ b Figure 8: The set of completed training signals corresponding to the drilling op- eration. With this. corresponding to the descriptor amplitude. A typical example is the strength Copyright by ASTM Int'l (all rights reserved). The second layer was composed of laterally interconnected neurons. No further reproductions authorized. was input to each neuron. recovered outputs in which the process descriptors were esti- mated by the auto-associative recall. GRABEC ET AL. In the majority of cases the proper descriptor of the process is already revealed from a visual examination of the vectors. Errors cannot be completely avoided because of the inherently chaotic nature of the machining process and the dependence of the spectral distribution on the properties of the environment which are not very reproducible. The effect of the second layer is equivalent to searching for the maximum of a set of dig- itized data. The excitation signals from this layer could be applied for the automatic control of the manufacturing process or for displaying the optimal descriptor. In order to test the ability of the system to identify features of the drilling pro- cess. only that neuron which was most strongly excited by the input signal attained a high excitation state. . In order to circumvent this decision problem and to permit an automated identification of the manufacturing process. the maximum of the amplitude distribution in the descriptor portion of the output pattern vector was determined and the cor- responding desciptor index was supplied as a representative one for the process. exhibit a sharp peak and hence. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.

. Copyright by ASTM Int'l (all rights reserved). The descriptors are correctly recovered by the auto-associative recall. 178 ACOUSTICEMISSION FREE BUN INPUT SIGNALS FREE P~I OUTPUT SIGNALS DRILLING ON PO~III.O~ 'A' INPUT SIGNALS DRILLIN~ ON POSITION 'A' OUTPUT SIGNALS DRILLING ON POSITION 'B' INPUT SIGNALS DRILLING ON POSITION "B' OUTPUT SIGNALS WORN DRILL ON 'A" INPUT SIGNALS WORN DRILL ON 'A" OUTPUT SIGNALS Figure 9: The set of input and corresponding output signals from the auto-associative memory. No further reproductions authorized. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.

However. In a sense. For this purpose no elastodynamic theory was needed since the method is completely empirical. it was conve- nient to include in the processing network a second layer by which the optimal descriptor of the process was determined. . which. then the characterization of the process from these signals becomes less reliable. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. in those instances when the characteristics of the signals emitted during different phases of a process are very similar. GRABEC ET AL. is also only possible in special cases. demonstrated that a trained auto-associative system is capable of solving simple inverse or forward problems related to the determination of AE sources from the detected signals and vice versa. the neural network cannot be properly trained and for this reason it cannot yet be applied to many structural integrity monitoring applications. but whose operation during the processing phase is linear. When a manufacturing process in its various phases generates acoustic emissions of significantly different spectral distributions then the proposed adaptive analyzer is capable of correctly recalling the descriptors of the process from the detected AE signals. By employing the spectral density data in the formation of the pattern vectors. However. The prob- Copyright by ASTM Int'l (all rights reserved). The processing system is comprised of an associative memory which resembles a neural network. A further step would be the connection of both or even more layers and the application of back-propagation algorithms for a complete adaptation of the network [161-[191. The example in which discrete AE signals were analyzed. The applicability of the proposed auto-associative system to the analysis of continuous AE signals was demonstrated. The structure of the processing procedure was obtained from consideration of an exact description of the elastodynamic phenomena and it resembles a simple neural network. Simulated signals obtained bybreaking a pencil lead are somewhat artificial and possibly not possessing equivalent source characteristics as those to be detected in the actual application. In the example presented here. a linear system is inherently limited in its ability to identify features of machining processes which are characteristically non-linear chaotic. As a consequence. up to now. such an AE system resembles the auditory system of an intelligent being. No further reproductions authorized. This problem corresponds to the reliable calibration of an AE system. provided that a spectrum analyzer is in- cluded as a pre-processor in the measurement system. CONCLUSIONS We have summarized in this paper recent work utilizing an adaptive learning system to the characterization of acoustic emission phenomena. Two examples were described which illustrated the operation of such a system. a time-invariant presentation of the AE signal characteristics was obtained. One important concluding remark can be made. It is recognized that one of the obstacles for the further application of the method presented here is the lack of typical AE sources needed for the train- ing of the system in real applications. The derivation presented here corresponds to a quasi-linear system in which the learning was achieved by a non-linear delta rule. ON SOLVING PROBLEMS BY A NEURAL NETWORK 179 of the clamping of the specimen and the cutting tool.

in Acoustic Emission Hand- book. 14.. Sect. Chapt. The first problem appears to be solvable by a self-organization of infor- mation processing units [23]. Eds. VIII. Eds. Surrey. Dornfeld. C. J. REFERENCES [1] Miller. parallel computers. Acoustic Emission Handbook.. D. an optimal statistical estimation. pp. Columbus. Inverse Problem Theory. 1988. pp. ACKNOWLEDGEMENTS This work has been supported by a contract to the Faculty of Mechanical Engineering of EK University by the Research Community of the Republic of Slovenia. Rajapakse. and McIntire. 28.. R. R. London. may be applicable [24]. non-linear. A. OH. D. No further reproductions authorized. I. pp. Partial support has also been derived from grant MSM-8904384 from the National Science Foundation. "Applications of Quantitative AE Methods: Dynamic Fracture. 1987. Butterworth Scientific. 219 .. Columbus. . J.. UK. 141-210. Dordrecht 1987. S. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. capable of simulating the operation of neural networks in real time. K. Amsterdam. Vol. A. 180 ACOUSTICEMISSION lem remains how best to develop a general. "Process Monitoring with Acoustic Emission". R... I. Achenbach and Y. There are two main problems. "Quantitative Acoustic Emission Techniques". Academic Press. [5] Grabec. B.. One is related to the non-linear formation of the memory during learning and the second involves the non-linear recall during analysis. Elsevier. Hamilton. in Ultrasonics In- ternational '85: Conference Proceedings. Copyright by ASTM Int'l (all rights reserved). [3] Sachse.. 1988. while for the second problem. W. ASNT. Materials and Transducer Characterization".. Martinus Nijhoff Publ. "Chaotic Dynamics of the Cutting Process". OH. based on a conditional average. 41-64.. "Optimal Filtering of Transient AE Signals". Vol. in Research Techniques in Nondestructive Testing. pp. Ltd. [4] Tarantola. 5 of Non-destructive Testing Handbook. in Solid Mechanics Research for Quantitative Non-destructive Evaluation. Ed. and Yee. P.. become available.. Vol. adaptive system by which manufacturing processes could be identified on the basis of acoustic emission signal analysis. It should be possible to implement both solutions in adaptive systems comprising a new gen- eration of adaptive analyzers which can be realized as soon as low cost. [6] Miller. K. 5 of Non-destructive Testing Handbook. 1985. [2] Scruby.. 1985. [7] Grabec. Vol. 4. International Journal Machine Tools Manufacturers. K. pp.224. 1988.. 467-511. Use of the facilities of the Materials Science Center at Cornell University is also acknowledged. Sharpe.. Mitchell. 19-32. ASNT.

UK. North-Holland. and Winter. 1984. in: Progress in Acoustic Emission IV.. Ltd. John Wiley and Sons. I. B. "Experimental Characterization of Ultrasonic Phenomena by a Learning System". "Application of an Intelligent Signal Processing System to Acoustic Emission Analysis". Cambridge. Self-organization and Associative Memory. Journal of the Acoustical Society of America. Journal of Applied Physics. 1989. I. I. 1987. The Adaptive Brain. "Experimental Characterization of Ultrasonic Phenomena by a Neural-like Learning System". and Hart.. 1. 66. Surrey. 1226-1235. Stockholm (1989). "An Introduction to Neural Computing". I.. No further reproductions authorized. pp.. W. [10] Grabec. I. McClelland. 211-218. 85. Zgonc. Butterworth Scientific.. pp. K.. and Sachse. 8A. W. [18] Arbib. pp. T. UK. 25-39. 3-16. pp. [12] Grabec. ON SOLVING PROBLEMS BY A NEURAL NETWORK 1 81 [8] Grabec. Pattern Classification and Scene Analysis. 21. pp. Ed. Chimenti. and Sachse. Tokyo. In Press. Springer Verlag. [9] Grabee. and Sachse. A. L.. Brains. R. S. I and II. 3993-4000. J. I. "Application of a Neural Network to Analysis of Ultrasonic Signals". [16] Rumelhart.. [15] Kohonen. and Sachse. I. Vol. [17] Grossberg. pp. Machines and Mathematics. 1988. O. Ed. [14] ~ribar. [20] Duda. New York. W..... MA. Vols. 75-80. M. Computer. 1988. D. "The Application of an Intelligent Signal Process- ing System to AE Analysis". 649-656. W. . Kimpara. 1984.. 1988. GRABEC ET AL. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.. Butter- worth Scientific.. and Sachse.. E. R. and Peklenik. pp. D. New York. R. in UI'89: Conference Proceedings. J. Mo~ina.. W. 1989. Plenum Press. CIRP Symposium on Manufacturing. Vol. Neural Networks. [19] Kohonen. Copyright by ASTM Int'l (all rights reserved). P.. in Review of Progress in Quantitative Nondestructive Evaluation.. pp. Springer Verlag. T. "Optoacoustic Monitoring of Laser Manufacturing Process by a Neural Network".. New York. Amsterdam. Japanese Society of NDI. 1973. 1988. Surrey. Vol. E. Thompson and D. "Monitoring of a Drilling Process by a Neural Network". 1989. 1987. E.. O. 1989. Vol. "Neural Nets for Adaptive Filtering and Adaptive Pattern Recognition". [13] Govekar. Parallel Distributed Processing. [21] Widrow. Ltd.. Eds. Vol. New York. [11] Grabec. I.. and Grabec. The MIT Press. J. in UI'89: Conference Proceed- ings. Grabec. 796-801.

"Self-organization Based on the Second Maximum-entropy Prin- ciple". Miller and P. Section 1 in Acoustic Emis- sion Handbook. Cornell University. I... D. . and Sachse.. C. R. Artificial Neural Networks. pp. "Automatic Modeling of Physical Phenomena: Application to Ultrasonic Data". Columbus. 12-16... No further reproductions authorized. Institute for Electrical Engineers. NY. W. 182 ACOUSTICEMISSION [22] Spanner. "Fundamentals of Acoustic Emission Testing".. pp. ASNT. I. Vol. Ithaca.. Copyright by ASTM Int'l (all rights reserved). 27-33. K. V. [241 Grahec. A. OH. 5 of Non-destructive Testing Handbook. Submitted for publication. R. 1989. and Pollock. Eds. J.. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. A. Conference Publication No. Notvest. 313. K. 1989. 1988. Mustafa. Brown. Materials Science Center Report #6771. McIntire. London. [23] Grabec. Hay.

Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Structural Monitoring Applications Copyright by ASTM Int'l (all rights reserved). No further reproductions authorized. .

. KEYWORDS : periodic inspection. flaws detection. ABSTRACT : Gas cylinders and large transport vessels used for industrial compressed gases are subjected to hydrosta- tic retests. France 185 C o p y r i g h t 9 1991 b y ASTM I n t e r n a t i o n a l ~. Yamaguchi. This method allows safe extension of the interval between re-tests. 1991. In North America. "Periodic Inspection of Compressed Gas cylinders and Transport Vessels by Using Acoustic Emission Testing".5 times (in Europe) or 5/3 times (in North America) the service pressure. This test consists of an external and internal visual examination of the vessel. Herv4 M. Acoustic Emission : Current Practice and Future Directions~ ASTM STP 1077.org Copyright by ASTM Int'l (all rights reserved). acoustic emission. pressure vessels. or also if the permanent expansion exceeds i0 percent of the measured total expansion. H. 75.M. Containers for compressed gases used by industrial gas companies. .astm. Application of acoustic emission testing to periodic inspec- tion of tubes in compressed gas service [i] has allowed some industrial gas companies (since early 1983). The vessel fails the hydrostatic test by leaking or bursting. gas cylinders. No further reproductions authorized. more than 1 m in water capacities (Fig.. and K. i) are subjected every 5 to i0 yr to an official test. 75321. Roget.Sachse. to use acoustic Dr Barth41~my is Materials Manager at L'Air Liquide. J. W. this hydrostatic testing is performed in a water jacket. whether single gas cylinders or big transport vessels. The cylinder or transport vessel should not have any leak at the test pressure. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Philadelphia. followed by a hydrostatic test at 1. American Society for Testing and Materials. In this paper are presented results of an investigation concerning how to apply acoustic emission tests during such periodic inspections to detect flaws.e. Barth414my PERIODIC INSPECTION OF COMPRESSED GAS CYLINDERS AND TRANSPORT VESSELS BY USING ACOUSTIC EMISSION TESTING REFERENCE : Barth414my. i. In conclusion a method for the periodic inspection (hydrostatic test with an acoustic emission testing) of such vessels is proposed. quai d'Orsay. Eds. internal corrosion and cracks. Paris.

We have. The maximum spread (ms) of the notch during the pressure cycles is recorded 9 When the crack is initiated.e. Additional tests [5]. carried out investigations to check by ourselves the actual possibility of detecting flaws affecting compressed gas vessel [4]. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. These artificially produced flaws were mainly fatigue cracks 9 The method consists of creating. 1 . 186 ACOUSTICEMISSION emission testing in lieu of hydrostatic testing [2]. tests were carried out : 9 on internally corroded cylinders 9 on cylinders with artificially produced flaws. this spread Copyright by ASTM Int'l (all rights reserved). [3]. The crack initiation and propagation are monitored using an extenso- meter (Fig. flat and notch with a rectangular bottom (see Fig9 2) on which a crack is initiated by hydraulic cycling from atmospheric pressure to the maximum service pressure of the cylinder. No further reproductions authorized. FIG. 3). [6] more recently allowed us to propose a re-testing method of compressed gas vessels making use of Acoustic Emission.Tube trailer for transport of gaseous hydrogen VARIOUS TYPE OF FLAWS Previous investigations [6] showed us that in the case of compressed gas vessels two types of flaws are likely to occur : 9 corrosion or oxidation located inside or outside the cylinder 9 cracking located on the cylinder or tube wall. at this time. a geometrical flaw i. . To check the reliability of the AE method proposed. on the external surface of a cylinder.

were also made. No further reproductions authorized. Other types of artifi- cial flaws. In addition. residual thickness 5 to 6 mm) NOTCH : L e n g t h : I00 mm f Depth : 5 mm FIG.5 mm FLAT (dimension : Length = 200 mm.3 . NOTCH MADE WITH A M I L L CUTTER Length : 100 mm Width : 5 mm Depth : 2 to 2. to make cracks of different depths. . or arc burns. 2 . B A R T H I % # M Y ON GAS CYLINDER INSPECTION 187 increases and continues increasing during propagation.Shape of the notches machined on the gas cylinders ec ~ EXTENSOMETER / z ED e: F A T I G U E ~ Q_ CD r~ r% O- :z uJ m~ o_ tq c3 7 I ~ CRACK INITIATION X LATENT PERIOD I NUMBER OF CYCLES FIG. these cracks being initiated on the external surface. It is therefore possible by monitoring this parameter (ms). flaws involving a notch. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. it is possible to confirm their presence using magnetoscopy before acoustic emission testing.Principle of the method for fatigue cracks production and detection using an extensometer Copyright by ASTM Int'l (all rights reserved).

558 m in diameter and 10. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 5) . They were performed during the hydrostatic pressurization. . the total count per unit of time (Fig.70/1. AE equipment used was previously described [6]. and when that emission corresponded to the flaws artificially produced (Fig. 4).20/0.30/0. in order to evaluate the difference ( ~ P ) between detection pressure and burst pressure. Copyright by ASTM Int'l (all rights reserved). No further reproductions authorized. The flaw was considered to be effectively detected as soon as a pronounced increase in emission occured.Chemistry. one located close to each end.40 Heat treatment Quenched in oil or other suitable medium and tempered at 640~ Mechanical properties Yield Ultimate Elongation strength tensile strength (percent) (MPa) (SPa) 750/850 880/1050 16/18 METHOD AND TEST CRITERIA Tests were carried out on vessels equipped with two acoustic emission sensors.e.15/0. heat treatment and mechanical properties of cylinders tested Chemistry (percent by weight) C Mn P (max) S (max) Si Cr Mo 0.20/1.5 m in length). from atmospheric pressure up to a pressure which could at least detect the flaw.05 m 3 (50 i) but more than 15 tests were carried out on flawed big transport vessels (sometimes more than 1 m 3 in water capacity.035 0. The emission were recorded also as a count rate i. the largest ones being 0. and in most cases even up to the failure of the vessels..38 0.035 0. this makes it is possible to easily define the threshold rate above which the defect being detected can actually be considered dangerous.00 0. TABLE 1 -.40 0. 188 ACOUSTICEMISSION Such tests were mainly carried out on Cr-Mo quenched and tempered steels (see Table i) gas cylinders with a water capacity of 0.20 0. In our case this threshold count rate was I0 ~/I0 sec).

ZI. . 4 .I]. ~"7 ~ 2z. ~ . 25600 TOTAL COUMT V5 TIME 480 20480 32B 15360 248 10240 15B 51~0 88 0 8 0 10 20 30 40 70 80 90 100 110 TIME (x i0 sec) FIG....8 10E03 10E02 168 10E01 88 10E00 8 0 '10 ~0 30 40 70 80 90 100 10 TIME (x i0 sec) FIG.. .~.E x a m p l e of crack d e t e c t i o n - AE r e c o r d e d as total count versus time + p r e s s u r e NUMBER PRESSURE (BARS) 10E05 COUNT RATE V5 TIME z.il.A. BARTHELC:MY ON GAS CYLINDER INSPECTION 189 NUMBER PRESSURE (BARS).. n o t c h .~o~o . It was found that DP is equal to the m i n i m u m p r e s s u r e w h i c h c o r r e s p o n d s to a t h r e s h o l d count rate of 103/10 s.I i .II!i I.. 5 .~. Copyright by ASTM Int'l (all rights reserved). i. Peak a m p l i t u d e s of acoustic e m i s s i o n r e c o r d e d w h e n testing c y l i n d e r s with fatigue cracks were in the range of 50 to 70 dB for p r e s s u r e g r e a t e r than or equal to DP.i 328 ~ . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.AE r e c o r d e d as total count rate versus time + p r e s s u r e AE TEST R E S U L T S Some results c o n c e r n i n g fatigue cracks..Ii I.. Iiii.~. .E x a m p l e of crack d e t e c t i o n .~o~. .i~.t y p e flaws and arc burns on gas c y l i n d e r s are c o m p i l e d in Table 2..II'.OO 10E04 IIIIIIIIIIZI. No further reproductions authorized.

Copyright by ASTM Int'l (all rights reserved).DP versus DP.38 21 Crack 200 331 1.34 250 1.27 435 (30 bar) SP = Service pressure and/or cycling pressure to obtain fatigue crack.49 4585 2. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.5 SP and 2.52 4677 2.32 400 28 bar) Notch 2015 5207 2.68 2400 1.29 22 Type of SP BP BP DP DP A P = BP-DP flaw (psi) (psi) SP (psi) SP (psi) Notch 2015 2437 1. notch and arc burn types of flaws - summary of AE results on gas cylinders Type SP (i) BP (2) BP DP (3) DP A P = BP-DP flaw (bars) (bars) SP (bars) SP (bars) Crack 200 297 1.5 SP A P is almost nil when the flaw is very serious (BP ~ 1.58 4007 1. Recent tests carried out on flawed tubes with notches (large transport vessels) c o n f i r m e d results initially found on cylinders.49 5021 2. .2 SP) and when the flaw is extremely small (BP ~ 2.21 2437 1. 190 ACOUSTIC EMISSION TABLE 2 .655 304 1. On the graph DP is expres- sed in terms of SP. No further reproductions authorized.99 1200 82 bar) Notch 2015 5021 2.Crack.83 293 (20 bar) Burns 2015 5020 2.52 27 Crack 200 362 1.97 5700 2.49 0 Notch 1800 3020 1. BP = Burst pressure DP = Acoustic emission detection p r e s s u r e I000 psi = 69 bars AE results are also reported on the graph (Fig.21 0 Notch 2015 5077 2. It can be seen from this graph that : A P is at an o p t i m u m when burst p r e s s u r e occurs b e t w e e n 1.4 258 1.81 304 1.5 SP).09 219 1.33 620 (43 bar) Arc 2015 5993 2.6) giving P = BP .25 19 Crack 200 280 1.52 58 Crack 200 219 1.09 0 Crack 200 269 1.485 276 1. with BP : burst p r e s s u r e and DP : flaw d e t e c t i o n p r e s s u r e using acoustic emission.

5 /3 Z 2. . lla and llb). 70 o_ <~ 60 f ~_ 50 => c. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 9 and i0).9 25 END 150 310 2.. No further reproductions authorized. 6 . &O u~ 30 A 20 c.2 1.5 3 DP/SP PRESSURE ( DP and BP) VERSUS SERVICE PRESSURE 0nd BP/ S~ FIG. p r e s s u r e was increased up to the rupture p r e s s u r e (Fig.48 70 Copyright by ASTM Int'l (all rights reserved). 10 LU 0 I i i i 1.88 268 1.96 15 MIDDLE 180 338 1. For such tests.. BARTHI~LC:MY ON GAS CYLINDER INSPECTION 191 l I I X NOTCH ARC 8URNS X O CRACK DP BP O CYLINDER WITHOUT .07 285 1.Sensitivity of AE detection versus flaws depth Typical examples of notch locations on large transport vessels are given (Fig. -- APPRECIABLE FLAWS SP SP Q_ c~ ' B0 X c~ .- DP DP I P = BP-DP the notch SP (bar) SP MIDDLE 150 310 2. TABLE 3 -.07 295 1. Some results also are compiled in Table 3 where it is shown that it was p o s s i b l e when using only two sensors per v e s s e l (one at each end) to detect flaw located either in the m i d d l e of the tube or at one end.Typical examples of notch detection on large transport vessels Location of SP BP BP . 7 and 8) as well as detection graphs (Fig.

Large transport vessel N ~ 1 . 4 . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.IM L " V ~ 100.. - LOG EVIga'S 1.'B8 12:5~:5Z ECLAT. 192 ACOUSTICEMISSION Events II Ill] i Lll lJll II I11t 11. O.00 '.Example of notch location by AE ! LO(. 8 . . "0 T 01234567 FILliP: i. 0. Of" 1B) .L II I I 2..5 I0 Location ( meters ) FIG. FIG.00 (R4RS..|iii 1. II I ' X II FOR SET(S) ' 0 ~ FILE: RIRLIg05 10/20. 7 .5 7.Example of notch location by AE Copyright by ASTM Int'l (all rights reserved).00 I] LOCATION .Large transport vessel N ~ 2 .46 5 "|* + | " =" . No further reproductions authorized.

08 T SEC ~. No further reproductions authorized. 08E4"01 ' ~ ' ' ' 1508.AE versus time O. BARTHI=LI~MY ON GAS CYLINDER INSPECTION 193 Events 100 _ Test pressure ( 4~330 PSI ) 50_ bsrs 0 I I 1 I I 100 200 300 400 FIG.TUBE FIG. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Large transport vessel N~ 1 . 9 . . 1 0 s ' I I I l O.AE versus time Copyright by ASTM Int'l (all rights reserved).Large transport vessel N~ 2 . i0 .S8 FILIlER: HISTORY F O R SET(S) 0 FILE: AIRLIQ05 10/20/B8 12:55:52 ECLAT.

194 ACOUSTIC EMISSION FIG.T y p e of rupture Copyright by ASTM Int'l (all rights reserved). lla .Large transport vessel[ N ~ 2 . . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized. llb .Large transport vessel N ~ 1 .Type of rupture FIG.

........... . 12 .... It was c o n f i r m e d that in these cases high emissions m a i n l y occur when pressure is less than service p r e s s u r e (Fig.. No further reproductions authorized. .AE recorded as count rate versus time + p r e s s u r e Copyright by ASTM Int'l (all rights reserved). .. as c u r r e n t l y required in North A m e r i c a [6]....:... .. . i. ..Example of corrosion detection . . Z40 158 08 0 10 zo 3o 40 ~'g 80 9o I00 ~10 TIME (x i0 sec) FIG. are more accurate than hydrostatic tests p e r f o r m e d in a water jacket for d e t e r m i n a t i o n of expansion of the cylinders.. .. .. . NUMBER PRESSURE (BARS) I~2480 TOTAL COUHT VS TIME t... ..........~ :.. . .. ... .. . 328 o. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. ....AE recorded as total count versus time + pressure NUMBER PRESSURE (BARS) 10E05 COUNT RATE VS TIHE 408 IBEO4 320 I L]2t f 10E03 240 10E~2 150 lOEB~ 8U 10E00 0 ~0 28 30 40 TO 80 913 1OO 11{} TIME (x i0 sec) FIG. . 13 .. 12 and 13). ." ... 4 4 0 ....~8 81920 ~ . .... ................ ." ' ...-.Example of corrosion detection ... . . .. The AE events recorded during the b e g i n n i n g of the p r e s s u r i z a t i o n are due to the fracture of oxide layers. " "'% i 64.. It was also shown that hydrostatic tests (without p e r m a n e n t expan- sion test) p e r f o r m e d at the same time as acoustic emission tests on vessels.. BARTHELI:MY ON GAS CYLINDER INSPECTION 195 A d d i t i o n a l tests were also p e r f o r m e d on internally corroded cylin- ders [6]........ .. ..

called crack retardation. If extensive emissions are noted at pressures below service pres- sure. 14. INTERVAL BETWEEN TWO PERIODIC INSPECTIONS If acoustic emission testing is performed during hydrostatic testing.000 cycles. 2 and 3). using two sensors located at each end of the vessel. if there is no crack propagation. It was shown [6] that the "retardation effect" caused by the overload does block the crack (the propagation is stopped) for 1. The critical level of emission above which signals are considered as "extensive emissions" will depend on the AE equipment used. we have shown that it will also detect crack propagation. In addition. it was already mentioned that the c~itical count rate is equal to iO /I0 s. Therefore. the result is given for one of these tests.500 cycles). Vessels should be examined through visual inspection and verification of thickness of the identified areas. and with appropriate measuring equipment. caused by the various fillings. we recor~nend to carry out hydrostatic pressurization from atmospheric pressure to tp. For one of this equipment used. 196 ACOUSTrCEMISSION APPLICATION TO PERIODIC INSPECTION OF CYLINDERS AND TUBES From our results. In Fig. using the system which has been used for testing the performance of artifical flaws of the fatigue crack type. we decided to check this.500 to 3. plastic deformation caused by the overload will occur at the tip of the crack. Copyright by ASTM Int'l (all rights reserved). Vessels should be examined through visual inspection and ultrasonic examination of the located emitting areas. No further reproductions authorized. Such a plastic deformation results in blocking crack propagation. it can be noted that in extreme cases (50 times a year) and when using the minimum delaying effect (1. it is desirable to perform acoustic emission up to test pressure (tp). This well known phenomenon has been extensively investigated for aluminium alloys [7]. These will be performed without permanent expansion testing making it possible to inspect several vessels in a single operation while using multi-channel testing equipment. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. When considering that compressed gas cylinders and tubes are filled between 5 to 50 times a year. a system which allows the detection and monitoring of crack growth (Fig. the vessel probably has a crack. Nevertheless. it would take 30 yr to trigger a new crack propagation process. . it follows that in order to best detect cracking. following the blockage caused by the hydrostatic overloading. during the subsequent pressure cycles. At the same time acoustic emission will be monitored. If extensive emissions are noted for pressures greater than service pressure. the vessel probably has internal oxidation.

000 7. it normally take another 1. and 1.000 10.000 6. This gives an additional safety margin which shows that it is possible to safely extend the retesting period when AE is performed during the hydro-test. fatigue cracks with greater difficulty than oxidations. 8. Another safety factor that one could considered as well is the number of cycles for crack propagation. CONCLUSIONS Acoustic emission of compressed gas cylinders and tubes will detect : internal oxidation and corrosion very easily .500 cycles for crack propagation up to the leak. since there will be then a safety factor greater than 3 on the life of the container. it is completely reasonable and safe to recommend a periodic inspection every i0 yr. // .000 5. Our investigations showed that.67 times service pressure in North America). . for a gas cylinder. it is best to perform this test during pressurization up to the test pressure (1. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 14 . the pressurization is stopped as soon as the crack is detected by AE.000 NUMBER OF CYCLES FIG. No further reproductions authorized.000 9.Delaying effect in crack propagation by overloading Consequently.000 i ! 2. It was checked that if. . to be sure of detecting cracks.5 times service pressure in Europe. BARTHI~LI~MY ON GAS CYLINDER INSPECTION 197 MAXIMUM SPREAD (micrometer) i LOADING (270 BARS) 250_ 200 LOADING(250 BARS) AE NEGATIVE 1 150 100 LOADING(250 BARS) AE NEGATIVE 50 . Copyright by ASTM Int'l (all rights reserved).

D. 198 ACOUSTIC EMISSION Consequently.. 161 . "The Effect of Overloads Upon Fatigue Crack Tip Opening Displacement and Crack Tip Opening/Closing Lads in Aluminium Alloys. Vol. [6] Barth414my H. 1983. [3] US Department of Transportation . "Acoustic Emission Test : An Alternative to Hydro-Retest for Compressed Gas Cylinders" Symposium AFIAP. we have recommended a method for the periodic inspection (hydrostatic test with an acoustic emission testing) of compressed gas containers. P. March-April 1988. ASME Journal of Pressure Vessel Technology. M. D. while still maintaining a very satisfactory safety margin. ii0. pp 234-240.L. Cannes. Mar. on Fracture. Steel Tubes in Compressed Gas Service". 1986. to apply this method of periodic testing to trailer tube (which are normally re-tested every 5 yr). "5th Int. pp 10-15. J. Copyright by ASTM Int'l (all rights reserved).Research and Special Programs Administration. REFERENCES [i] Blackburn. Oct. May 1986. pp. Apart from the numerous technical advantages in comparison to the periodic inspection as presently performed. and Davidson. "Acoustic Emission Testing and Structural Evaluation of Seamless. France. [4] Barthelemy H. DOT-E8944. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement... Vol..Flaw Detection Using Acoustic Emission Testing" ASME Journal of Pressure Vessel Technology. [2] US Department of Transportation . May 1986. March 29-April 3. [5] Barth41~my H.167. 1981. and we have shown that these tests can be performed every i0 yr. Paris.. "How to Detect Flaws in Compressed Gas Cylinders and Tubes by Acoustic Emission" IOMA 1987 Annual Meeting Speach and IOMA Broadcaster. No further reproductions authorized.R. it appears to be of a high economical interest. 108. "Periodic Inspection of Compressed Gas cylinders and Tubes . [7] Lankford. 1985. and Rana. DOT -E 8944 (First revision). Conf. Dec. .Research And Special Programs Administration.

KEYWORDS: acoustic emission. Sachse. With regard to AE the goal of these experiments was the va- lidation and further development of AE during hydrotest. 1991. i). . All of these had artificial defects representing all types of defects which can occur.. W. J. localization. "Detectability of De- fects in Reactor Pressure Components by Location and Interpreta- tion of AE-Sources". ABSTRACT: In the last years many hydrotests. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Phila- delphia. and K. Acoustic Emission: Current Practice and Fu- ture Directions. The AE generated during these experiments has been recorded and analysed. Techniques for evalu- ation of AE-data have been developed which allowed the separation of AE-events from crack growth and crack surface friction by using the distributions of AE signal parameters.. Christoph Sklarczyk and Eckhardt Waschkies DETECTABILITY OF DEFECTS IN REACTOR PRESSURE COMPONENTS BY LOCATION AND INTERPRETATION OF AE-SOURCES REFERENCE: Sklarczyk C. residual stress In the frame of the German reactor safety research programme more than 20 experimental pressure hydrotests have been performed in the last seven years. Crack growth can be detected with high sensitivity at thermal shock and fatigue tests whereas detectability of crack surface friction increases with the amount of compressive stres- ses between the crack surfaces. ASTM STP 1077. No further reproductions authorized. Waschkies are research scientists at IzfP (Institut f~r zerst6rungsfreie Pr~fverfahren). C. E. cyclic fatigue tests and thermal shock tests have been carried out on reactor pressure vessels. American Society for Testing and Materials. signal ana- lysis. Eds. At hydrotest only a low level AE is produced by nongrowing cracks due to the lack of compressive stresses. cracks. Drs. and Waschkies. D-6600 SaarbrOcken. All of them were surveyed by acoustic emission (AE). with the aim to de- termine the conditions of detectability of growing and nongrowing cracks by acoustic emission (AE). pressure vessel.astm. Roget. Three pressure vessels were available for these trials (fig. especially the rise time. Sklarczyk and E. University.org Copyright by ASTM Int'l (all rights reserved). FRG 199 C o p y r i g h t 9 1991 b y ASTM I n t e r n a t i o n a l ~.. Several com- mercial multichannel AE-systems were used and moreover more capable sy- stems were developed. Yama- guchi. Building 36. which exhibited crack-like defects.

The main reason for this was that crack growth of defects during hydrotest did not oc- cur. No further reproductions authorized. AE could not be measured and evaluated reliably. As far as AE caused by friction did not occur or only on a very low level. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. . 200 ACOUSTICEMISSION FIG. 1 Fressure vessels tested by AE-method a) large vessel at MPA Stuttgart b) HDR vessel c) intermediate scale vessel ZB2 The experience leading to this very expensive series of experiments was that application of AE as available on the market on hydrotests of commercial vessels did not lead to conclusive results. In this paper we give a brief summary of the improvements which we- re developed and approved during these experiments. Copyright by ASTM Int'l (all rights reserved). This paper deals mainly with localization and signal processing for interpretation of different AE-source mechanisms like crack growth and crack surface friction.

Beside the signals from the nozzle also AE-signals from the flaws F8 and FI0 are distinctly lo- cated. I). The details of this algorithm and the appro- val of it will be published elsewhere. The efficiency of this algorithm is demonstrated in fig. The obtained results are shown in fig. Commonly three probes are used for planar localization. The inserted velocity was c = 3. especially when these signals propagate by disper- sive guided waves. These are in our case only the signals around the nozzle. which is propagating with this velocity. the real signal propagation velocity depends on amplitude of the signal. 0 < c < cI (cl: longitudinal wave velocity) In this case. We have developed a four probe localization algorithm which considers the propagation velocity as unknown. 2 on an experimental vessel with flaws (indicated in fig. No further reproductions authorized. In this case the sound veloci- ty has to be inserted as a known constant value to calculate the source location. vessel c in fig. A hydrotest was performed and surveyed by AE. In the area of the flaws F8 and FII and the middle weld seam much more signals were extracted by the four-probe-localization algorithm (right side of fig.FII at the inner side of the vessel wall six AE-probes were installed. much more signals are located. Copyright by ASTM Int'l (all rights reserved). 2). If triggering is done by constant amplitude thres- hold level. In the four probe case the velocity is assumed to be the same for all passes from the source to the four probes. Along the corner of a patch which was inserted in the vessel and contained the three flaws F8 . The achieved improvement is obvious. The localization method with 3 sensors handles the propagation velocities ci from the AE-source to the sensors i as known values which are equal for all sensors i and fixed: C i = C = fixed r i x rj path of flight difference ci(r i . . 2. SKLARCZYK AND WASCHKIES ON PRESSURE COMPONENTS 201 LOCALIZATION AND CHARACTERIZATION OF AE-SIGNALS Localization of AE-Sources Localization of AE-sources is not only important for determination of defect location but also to eliminate spurious noise.5 m/ms. on the left side the three-probe-results are displayed.rj) = Atij ~tij time of flight difference (ri: distance between sensor and source) In this case only that part of AE-signals is located correctly. c i = c. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 2) and defected weld- ments (ZB2. In the case of localization with 4 sensors the sound velocities c i are also regarded as equal but not as a known fixed value but as unknown. In many practical cases the real sound velocity is not con- stant and not known. Localization was performed both with the planar three-probes-method with pre-given constant sound velocity and by the above mentioned four-probe-method where no value for the signal propagation velocity was inserted.

. 202 ACOUSTIC EMISSION F I O ~ GO O O o Q o O 0 00 0 o O 0 SE-probes J ~.oo2t2l Ill 2~n ~o Oo o / -. This can be demonstrated by following examples: Fig. 2 Locations and numbers of AE-events of a hydrotest at the ZB2-vessel (comparision between two localizations algorithms) Interpretation and Classification of AE-Events Apart from the already mentioned hydrotests many trials have been carried out with different loading conditions in order to simulate the different possible types of real crack growth during the lifetime of a reactor pressure vessel. see also [1-3]) which resulted in a macroscopic crack growth by fatigue and corrosion. . ~ 0 2 I o e ell OO 0 s q ~ Oil . No further reproductions authorized.tensile tests in laboratory on fracture mechanical specimens made from reactor steels In the frame of these investigations it has been found that the AE- signal parameters rise time (= time from the start to the absolute ma- ximum of the signal). o. . . 41K3 ]- Iocalisation with 3 Iocalisation with 4 / weld seam weld seam transducers transducers FIG. Cooling down was perfor- med by injecting cold water at the location to be shocked. o . The cracks opened at cooling down and closed at reheating [1-3]. vessel b in fig. . Following loading conditions have been app- lied: .many thousand fatigue cycles by water pressurisation on a pressure vessel to produce real cracks and in service-crack growth (ZB2. . 3 0 I I 11724 3 O I 0 l O I'" O~O000000000 O00oO0000000 h~ o~ J . . 111 I Q ~ | O O O e # 0 0 | 0" o 0 ~ 0 0~000~ t O 0 e I O 0 0 t I I e 0 00000 000000 "~6 "66~6 o oo 6 C . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. . "" O'eo I ~2o12 Oll O0 O0 0 O0 O0 O0 O0 0 E Ooo=oCO0 oo000 nozzle ~ I ) o 0 r 00000 OOOOO000oO000 000000000 9 3 o I ~-T T f-o-r-olt o t o F8 O00OOOO00000 9. Z . . energy and duration exhibit characteristical di- stributions which can be used for classification and interpretation of AE-events from different sources like crack growth and crack surface friction whereby the signal rise time plays the most important role. I o n . : R l e e OlO e ! o I o l i t o o Oll ---. . . . . The interior of the vessel was filled with hot pressurized water (593 K. o . . i.oon . o 7T-i-~"iz o n o. OOOOOGO00000 00000~000000 0~000o000000 Io . Each thermal shock at the nozzle consisted in a cooling phase of 120 s and a reheating phase of 240 s. . i) . 3 shows that there exist two distinct maxima in the rise time distribution of located AE Copyright by ASTM Int'l (all rights reserved). . ~0 0 / ~ 11 e.some thousand thermal shock tests at the edge of a nozzle and at the cylindrical part of a reactor pressure vessel (HDR. o . . ii MPa). . vessel c in fig. .

The- refore it can be concluded that friction events exhibit long rise times under the given experimental conditions. where not separated but we- re concentrated to one single signal. No further reproductions authorized.us] FIG. 4 Rise time distribution of AE-events during a hydrotest at the HDR-vessel Copyright by ASTM Int'l (all rights reserved). The only pos- sible AE-source during this hydrotest was crack surface friction. c r a c k s u r f a c e friction 50O0 c ~P > 4000 3000 .Q crack growth E 2000 c 1000 0 i i 1 ! I 10 ~ 101 102 10 3 10 4 risetime [US] FIG. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. During this hydrotest no growth of the existing cracks occurred. signals. c r a c k s u r f a c e friction o) r 10.) In fig. (The signal dead time used was 819 ~s. . 3 Rise time distribution of AE-events during thermal shock test at the HDR-vessel (sum over 400 cycles) 12. r > "a . i. 10 3 10 4 risetime [. The AE-events with short rise time in fig. 3. 3 can thus be assigned to crack growth.e.IQ E 4 0 10 -1 A 10 0 10 ~ . 10 2 . . who- se time difference between end of the first signal and beginning of the second signal amounted to less then 819 ~s. 4 it can be seen that most AE-events detected during a hydrotest on the same vessel possess long rise times similar to the right distribution in fig. SKLARCZYK AND WASCHKIES ON PRESSURE COMPONENTS 203 recorded during some hundred thermal shocks on the nozzle of a pressure vessel (HDR).

5 Rise time distribution of crack propagation (solid) and crack surface friction (dashed) for CT-specimens The limiting value t o between these two groups depends on the expe- rimental conditions. Correspondingly the distinctly longer rise times of AE-events from crack surface friction are obviously due to longer source lifetimes or to groups of events from different touching points which interfere and cannot be separated. . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. t o is determined by the measured rise time resul- ting from broadening of an infinitely short ultrasonic test pulse (si- mulated crack growth signal) by sound propagation from the source to the transducer. It is given in our examples by t o = 50 . Copyright by ASTM Int'l (all rights reserved). This coincides with the above given interpretation that short rise times are due to crack growth events. Fig.g. The right distribu- tion (dashed curve) has been recorded during cyclic loading of pre- cracked specimens at low loads where no crack growth occurred. ~ f l 0 cD A i ! I 10 A: '. / . 204 ACOUSTICEMISSION These results areverified by fracture mechanical tests on CT-spe- cimens (thickness 50 mm). 5 displays two distinct rise time di- stributions of these tests: AE-events of the left distribution (solid curve) have been recorded during crack propagation. to = 30 ps O) II crack surface friction "E cD 20 > crack growth . This is not valid for e. Thus rise time reflects the sour- ce lifetime if the signals are recorded in the nearfield. This agrees to the results gathered during the hydrotests and thermal shock experiments at the HDR-vessel. The preference of the rise time compared to the other signal para- meters is due to the fact that rise time is approximately independent from signal amplitude. As a result it can be conclu- ded that small rise times are related to crack growth and longer rise times are related to crack surface friction. The va- lue of t o has to be estimated theoretically or experimentally for each test-setup. Then this AE is due to crack surface friction.. No further reproductions authorized. (The near- field of a sound source extends to about five plate thicknesses [6].100 ~s for the HDR-nozzle (thermal shock) and t o = 30 Us for the CT-specimens. -1 0 1 2 3 4 ~" Iog(risetime [us]) FIG. E 0 I I I .) Very short life times in the range of 1 Us or less for microscopic crack growth events have been found by Wadley and Scruby and other re- searchers [5].q t. the signal duration which obviously depends on amplitude.

Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 6 all signals with short rise times have been elimi- nated. . 6). In fig. 7). These calculations show that the conditions which are nessessary for a copious AE due to crack surface friction are fulfilled during thermal shock: a) relative move- ment of the rough crack surfaces (fig. It can be seen that these maxima approximately coincide with maxima of those AE-signals (dashed curve) which exhibit long rise times [2]. 6 AE-event rate (crack surface friction) and CMOD vs. see fig.S O E . SKLARCZYK AND WASCHKIES ON PRESSURE COMPONENTS 205 crack opening crack closing 1000 -1. reheating I FIG. 6) and b) compressive normal stresses between the crack surfaces (fig. No further reproductions authorized. 7 in the lower right corner) and at the beginning of reheating phase (t = 125 s) there exist compressive stresses near the inner sur- face of the wall. 7 shows the stress distribution in the vessel-wall of the HDR for several points of time during one thermal shock cycle. cycle- time during thermal shock test at the HDR-vessel This interpretation is also supported by fracture mechanical calcu- lations: The solid curve in fig. 6 shows that the maxima of velocity of crack mouth opening displacement (CMOD) are found at the beginning of the cooling (0 s) and reheating (120 s) phase during thermal shock at the HDR-vessel. .o 250 \ u C 0 . It can be seen that at the very beginning of the cooling phase at 0 s (this is equivalent to 360 s. Thus these time intervals where compressive stresses arise coincide with the maximum AE (fig. Copyright by ASTM Int'l (all rights reserved). | O 0 100 200 360 cycle time [s] I cooling .5 co 9 -~ 1\ 750 \ 500 / \ \ CMOD I I I~ I ~AE -1 E -O. Other calculations have shown that test phases with high rates of frictional AE are connected with large compressive stresses (up to 300 N / ~ 2) in the austenitic cladding material [4]: Fig.

8 gives an example of AE detected during fatigue crack growth (500 cyc- Copyright by ASTM Int'l (all rights reserved). No further reproductions authorized. negative signs at the y-axis mean compressive stresses. 125s t = 360 s -600 -600 0 do 60 9o 20 o do 6o 9o 12o wall thickness [mm c) d) wall thickness [mm] t: time after cooling FIG. Transfer Functions of AE-Events If the distance between transducer and source is greater than about five plate thicknesses it is nessessary for interpretation of AE-events to perform a transfer correction of the signal parameters because sound waves are changed strongly by propagation in the component or specimen. For dispersive lamb wa- ves the rise time is proportional to the distance.600 O 30 6'0 90 120 0 30 60 90 120 a) wall t h i c k n e s s [ m m ] b) wall thickness [ram] 600 600 -j E 300. The inner surface of the vessel lies at 0 mm. . This has been inve- stigated for various types of crack growth: fatigue crack growth. 206 ACOUSTIC EMISSION 600 600 E E 300 \ ~E E 300 ~. The transfer functions are given elsewhere [6]. crack propagation under monotonic increasing load and corrosion cracks. 7 Stress in the wall of the HDR-vessel during thermal shock test. At the above descri- bed experiments a transfer correction was not nessessary because the transducers layed in the nearfield of the sources. DETECTABILITY OF CRACKS In the frame of the above mentioned tests it has been shown that the detection limit of spontaneously occurring stable crack growth by using AE-method lies in the range of a few mm 2 [7]. o 0 = 9 -3oo cooling "~ . E 300 E E Z 0 0 f == begin of reheating end of reheating ~ -aoo "~ -3oo t. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Fig.300 end of cooling t=5s t -120 s -600 .

ZB2) exhibited only low ma- ximum amplitudes. 9 gives an example: fig. In fig. The AE-events recorded during hydrotests on pressure vessels (e.e. One further aim of these tests was to examine whether AE-monitoring is a proper method to find non-growing defects during hydrotests. The AE-peak at pressure maximum can be assigned to crack growth be- cause of the maximum stress intensity factor at the cracks whereas AE at lower pressure is likely to be due to crack surface friction (fig. 8 b) and are averdriven.ack velocity. 1 and 2).01 mm2/s was detectable [8]. .g. 9b it can be seen Copyright by ASTM Int'l (all rights reserved). i. even if macroscopic cracks exist in the vessel wall. SKLARCZYKANDWASCHKIESONPRESSURECOMPONENTS 207 les.1 "~ ~) i~" 1 friction crackgriwth . 9ooo. Fig. . T = 513 K) of defects situated on the inner side of the wall of the pressure vessel ZB2 (fig. steel type and structure. 9a shows that AE-events located during a hydrotest (ZB2) by using the four-transducer-algorithm are distributed over the whole pressure range and are low in quantity. close to the threshold. these amplitudes are beyond the measuring range of the AE-equipment. Most amplitudes of crack growth are higher than those of friction noise (fig. An important criterion is the c. 8 Distributions of AE-events from 500 fatigue cycles at the ZB2- vessel The detection limit of crack growth depends on the experimental conditions like stress state. but these cracks did not grow. 8 a). No further reproductions authorized. The results show that the detectability of nongrowing cracks during a hy- drotest cannot be quantified up to now and depends on the special con- ditions of stress state and crack morphology. == <1 0 8 20 pressure [aPa] a) AE-distribution versus pressure crack growth (overdriven amplitudes) ~4500- > friction _ o ~ E 0 250 amplitude [mV] 5O0 b) amplitude distribution FIG. A fatigue crack growth in the range of 0. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.

t:time) from 0. During hydrotest this curve is shifted towards higher tensile stresses (curve b).i MPa/min. The increasing of the pressure rate ~p/At (p:pressure. No further reproductions authorized. 18o t n E g o. 9 Distributions of AE-events from a hydrotest at the ZB2-vessel The absence of an high amplitude-AE from crack surface friction can be explained by the lack of compressive stresses in the vessel wall [4]. I0 which shows that there are only small compressive stresses in the vessel wall at the beginning of the hydro- test (curve a). . In normal cases the conditions to generate frictional AE are not present during hydrotests of pressure vessels. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Copyright by ASTM Int'l (all rights reserved). to about 20 MPa/min. 208 ACOUSTIC EMISSION that these signals exhibit low amplitudes which are in contrast to the high amplitudes recorded during crack growth (fig. i 1 ~ 0 10 2O 3O E p r e s s u r e [MPa] a) AE-distribution versus pressure 200 E CD > CD E 0 . yielded no better detectability. 0 250 500 amp. This is displayed in fig. A l r.tude [mY] b) amplitude distribution FIG. 8 b).

106. Proceedings of the 8th International Conference on NDE in the Nuclear Industry. in preparation Copyright by ASTM Int'l (all rights reserved). 1989.. Acta Metallurgica. B. Maier. G. and Waschkies.. 151 . F. E. G. G. Vol.. 17-20 Nov. H.. and Knoch.J.15.200 t- "~(noI. Nuclear Engineering and Design. Kissimmee. 600 . Waschkies. 1988. E.. 19. Waschkies. 1981.. and Neubrech. Cologne. N.154 [4] Gries. SKLARCZYK AND WASCHKIES ON PRESSURE COMPONENTS 209 . Vol.. Waschkies. 112. G. and Shrimpton. and Klein. Vol. 29. and H611er. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 1989.. Scruby. 399 .. Conference Proceedings of the 13th M FA-Seminar. P. . Nuclear Engineering and Design. T= 50~ E .414 [6] Waschkies. E. Vol. 8 . E0. Z. 399 . G.9 Oct. F. Walte. 351 .6 MPa. P. C. 1986.. 197 . P.. E.403 [5] Wadley. 600 9 i 0 o 0 10 20 30 40 50 60 70 80 90 100 110 wall thickness [mm] FIG.. Florida. FRG). 1986. H.. 139 . E.. -400 o a: after stress relieving (residual stress) b: during hydrotest (p= 15.. Vol. International Journal for Pressure Vessels & Piping.202 [2] Deuster.157 [7] Deuster.-.. H. 1984. M. E.. K.. No further reproductions authorized. FRG. Stuttgart.2 r c 400 200 i b E == E 0 a . and H611er. Walte.. 371 [3] Neubrech. NDT International.. i0 Stress in the wall of the HDR-vessel REFERENCES [i] Waschkies. Hepp. . E.. G. 1987 [8] Final report of the research project RS 150675 of the GRS (Gesellschaft for Reaktorsicherheit.

Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Deformation Studies Copyright by ASTM Int'l (all rights reserved). No further reproductions authorized. .

and Roman I . Yamaguchi. Israel. Chairman of the d i v i s i o n of Applied Physics and Materials Science. 1991. Philadelphia. 213 Copyright9 by ASTM Intemational wvwastmorg Copyright by ASTM Int'l (all rights reserved). Eds. Two types of AE were observed. " E f f e c t of Pre-Exposure to Water on the Acoustic Emission Behavior of 2091-T3 AI-Li Alloy".. The Hebrew Univ. especially in the yield region. No further reproductions authorized. i t is believed to stem from hydrogen charging that took place during pre-exposure. Jerusalem 91904. American Society for Testing and Materials. AE characteristics were evaluated during t e n s i l e deformation utilizing a micro-computer based system and an rms voltmeter. and which also caused reduction of the t e n s i l e elongation of the material. . Although the precise mechanism of the observed e f f e c t is not known. hydrogen embrittlement. As a result these alloys are good candidates Dr. ultrasonic field. The pre- exposed specimens gave rise to more intense AE a c t i v i t y manifested in higher peak amplitude AE events. followed by a change to burst type AE coinciding with the onset of PLC serrations. researcher. The growing interest in AI-Li alloys for advanced structural application is due to the decrease in density and increase in the e l a s t i c modulus with the addition of lithium (up to 3. . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. School of Applied Science and Technology. aluminium-lithium alloys. i n i t i a l l y . hydrogen charging. KEYWORDS: acoustic emission. Zeides. Sachse. and Professor Roman. Roget and K. ABSTRACT: The e f f e c t of pre-exposure to water with imposed weak ultrasonic field on the acoustic emission (AE) behavior of 2091-T3 AI-Li alloy was studied. A s i g n i f i c a n t difference in AE behavior between as-received and pre-exposed material has been observed. AE was of a continuous nature with a maximum at the yield. Acoustic Emission: Current Practice and Future Directions~ ASTM S-~:P 1077. W. J.5 w/o) [ I ] . Felix Zeides and Itzhak Roman EFFECT OF PRE-EXPOSURE TO WATER ON THE ACOUSTIC EMISSION BEHAVIOR OF 2091-T3 AI-Li ALLOY REFERENCE: Z~ides F.

AET). The rms voltage level increased with load and exhibited a prominent peak close to the yield load. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized. AET). AET) with 125 to 250 kHz band pass f i l t e r and a rms voltmeter (2Ol.displacement curves for both the as-received and the pre-exposed samples are shown in Fig. The samples which were of the sheet thickness (2. and in both cases. samples) are discussed.) comprized of 175 kHz resonant transducer (MAC 175. whereas the behavior of dynamic strain aging stems from the localized nature of slip brought about by the interaction of dislocations with mobile solute atoms in these alloys. isotropy and corrosion behavior of these alloys is desired [2].5 w/o Cu. i t was suggested that the continuous peak associated with yielding is due to repeated shearing of the coherent 6' precipitates. non exposed.5 mm) and had a 32mm gage. 1 w/o Mg AI-Li alloy with different heat treatment conditions [3]. Loyd machine with a strain rate of lO. 214 ACOUSTICEMISSION for replacing the existing high strength aluminium alloys. Copyright by ASTM Int'l (all rights reserved)./s in the as-received condition or after immersion in d i s t i l l e d water in an ultrasonic bath for three months. 2 w/o Li. 4and were tested in a J. AET corp. All samples were polished prior to testing. primarily in aerospace applications. better understanding of fracture.J. In the present study. were machined from a 2. preamplifier (160. the e f f e c t of pre-exposure to water on the acoustic emission (AE) behavior of 2091-T3 AI-Li alloy was examined. The decrease in rms voltage level during yielding was followed by numerous large AE bursts during work hardening which parallel stress level serrations (indicative of dynamic strain aging). before this materializes. Acoustic emission characteristics were evaluated during tensile deformation u t i l i z i n g a micro-computer based system (Model 5000. Results and Discussion Typical load-displacement and the accompanying acoustic emission intensity . . Similar AE behavior was observed in another study on a 2. Continuous type AE was observed almost from the beginning of elastic deformation. However. Possible origins of the observed changes in AE behavior (as compared with that of reference. EXPERIMENTAL Longitudinal f l a t tensile samples employed in this study. l .5mm thick sheet of 2091 AI-Li alloy in the T3 conditions. the AE behavior is q u a l i t a t i v e l y similar. There.

9 z~ z8 ~z 34 . . The dissimilarity is two fold.' ~ ~ ~U~BER OF I R T E R V ~ L $ = 6@ ~NTERV~L S I Z E . of both the as-received material and the material exposed to water. The more intense AE activity after pre-exposure is also manifested in higher peak amplitudes of the AE signals as shown in Fig. f i r s t i t can be noted that the pre-exposure markedly reduced the tensile elongation of the material. 100000 100~0 180e 108. 19. for which the peak of continuous emission close to the yield load was of higher intensity. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. which shows the peak amplitude distribution of AE date in Fig. I. No further reproductions authorized. D a t a were accumulated for the entire tensile test. | Figure 2: Cumulative distribution of events by peak amplitude for the unexposed (lower chart) and pre-exposed specimens. g ' ' ' -. a significant difference is noticeable. the pre-yield acoustic emission activity was more pronounced for the material pre-exposed to water. . . Copyright by ASTM Int'l (all rights reserved). . . . Second. ZEIDES AND ROMAN ON BEHAVIOR OF 2091-T3 AL-Li ALLOY 215 Figure l: Load and AE intensity vs. Inspire of the similar general behavior as demonstrated in Fig l. . . . . I . 2. displacement curves for unexposed (a) and pre-exposed (b) tensile samples. .

'~(ZE = [ Figure 4: Cumulative d i s t r i b u t i o n of events by peak amplitude f o r the d i f f e r e n t stages of deformation of the pre- exposed sample.~ s'z s'5 LOG OF CUM. t~ zs' 31' ~4-' '~7 4e' 43' 46' 49' ~2' ~5 ~'8 6'1 6'4 6'7 7'~ 7'~ 7'~ 7'. and i t can be seen that most of the higher amplitude AE a c t i v i t y o r i g i n a t e s in the p r e . IN NaOH water s o l u t i o n not subjected to the u l t r a s o n i c f i e l d . tO. 10080 1808 100 . Copyright by ASTM Int'l (all rights reserved). tO.~'. i n d i c a t i n g AE a c t i v i t i e s of higher amplitude as a r e s u l t of pre-exposure. The evolution of the d i s t r i b u t i o n s shown in Fig. D I S T . t h a t of the pre-exposed material extends to 78 dB. The o r i g i n of the observed higher AE a c t i v i t y in the pre-exposed samples is not c l e a r . The upper curve represents AE p r i o r to AE continuous peak. 2 which are accumulated over the e n t i r e t e s t are shown in Figs.y i e l d regime. However. . O I S T . I OF EUENT$ BY PERK RHPLITIJOE LOB) FOR SE~ISOR ! tlUHSER OF INTERVAL8 = 60 INTERVRL . 216 ACOUSTICEMISSION While the d i s t r i b u t i o n f o r the as-received material extends to about 55 dB. The upper curve corresponds to the burst type AE and the lower one to the continuous type AE. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. I~ S I Z E : I Figure 3: Cumulative d i s t r i b u t i o n of events by peak amplitude for the unexposed specimen. In addition s i m i l a r behavior was found in the specimens pre-exposed to O. Careful examination of the samples p r i o r to and during t e n s i l e t e s t i n g revealed the existence of surface p i t s and formation of corner micro-cracks (during t e s t i n g ) only in the pre-exposed samples. No further reproductions authorized. LO000 i080 log .46' ' 4~' 52' 55' ~8' bl' 64' 67' 70 73 7r14979' 8'2 8'5 LOG OF CUR. These cracks did not appear when the corners were rounded before the t e s t . I OF EIJEtlTS BY PEAK AHPLITUOE ( 0 8 ) FOR SENSOR L NUMBER OF I N T E R V A L 8 ~ 68 INTERVP. the pre-exposure e f f e c t both on t o t a l elongation and on AE behavior s t i l l persisted. t i S 28' 41" -~4' ":~7 4e' 4. the middle curve r e l a t e s to the AE continuous peak and the lower curve is f o r the burst type AE. 3 and 4.

and Johnson C. J. 22. S. In both cases. October 1986. and Karmarkar. I t is not yet clear which mechanism is responsible for the higher AE a c t i v i t y of the pre-exposed t e n s i l e samples. that AE can be used as a very sensitive tool for evaluating environmental effects in AI-Li alloys. [5] Zeides. ACKNOWLEDGMENTS Discussions with Prof. REFERENCES [ I ] Divecha. We are grateful for the support of portions of this research by grant No. . J. Vol. of Acoustic Emission Vol. pp 111-115. Advanced Materials & Process Vol. D. No. pp 74-79. Jerusalem. F. 4. i t should be realized. i t c e r t a i n l y relates to the hydrogen charging that took place during the pre-exposure.- Copyright by ASTM Int'l (all rights reserved). I. pp 1521-1529. of Material Science. P. Vol.. of Acoustic Emission. Kanji Ono in the course of this study are g r a t e f u l l y acknowledged. J. I . 85/00103 from the U. .S. Israel. J. ZEIDES AND ROMAN ON BEHAVIOR OF 2091-T3 AL-Li ALLOY 21 7 Thus. pp 4 7 . Roman. the observed effects can not be related to a specific environment. 8. September 1989. cracking of hydrides that possibly formed or hydrogen induced internal microcracking.. The possible sources for the hydrogen induced modification of AE in t e n s i l e test are hydrogen induced modification of p l a s t i c deformation in tension. J. to be published. A. [3] Roman. 2/3. 130. In addition. I. i t has been found that large amount of hydrogen was introduced into the pre-exposed samples [4]. No further reproductions authorized. 1987. Ono K. ... and Roman. No. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement... H. however. [4] Zeides. 4. [2] Lavernia.-Israel Binational Science Foundation (BSF). No. comparison between AE during uniform tension and uniform compression has been employed and the conclusion.. 1987.. F. In another study.4 9 . 3. No pre-exposure effects were found in the compression test. E. N. and Grant. that the major source of AE in the studied AI-Li alloys was plastic deformation has been reached [5].

e. AE technique has shown an a b i l i t y to explain the micromechanisms of dynamic changes in materials and to solve problems related to structural monitoring. Ni3(AI. For 316 SS. alloy PE-16 Acoustic Emission (AE) from metals has been studied for about f o r t y years starting with the pioneering work of Kaiser in 1950 to evaluate the dynamic behaviour of materials [ I ] . Raj and T.Yamaguchi. American S-oETety for Testing and Materials. Eds. Division for PIE and NDT Development. Kalpakkam. Indira Gandhi Centre for Atomic Research." Acoustic Emission: Current Practice and Future Directions. Jayakumar ACOUSTIC EMISSION DURING TENSILE DEFORMATIONAND FRACTURE IN AUSTENITIC ALLOYS REFERENCE: Raj. and T. deformation. The studies in PE-16 have shown that AE can be correlated with the type of interactions i . 1991.Ti). serious interest in the potential of t h i s technique for materials research began only in the 1970s. . Time and frequency domain analyses have been used to characterise the dislocation behaviour. stainless steel. J. .Jayakumar is research s c i e n t i s t in the same Division. Effects of second phases on the AE are considered. S a ~ .astm. India. . Examples of dynamic changes where AE technique promises to generate Baldev Raj is Head. "Acoustic Emission during Tensile Deformation and Fracture in Austenitic Alloys. A method to amplify the weak AE signals in 316 SS is highlighted. B. e .. 603 I02.Roget and K.. Philadelphia. W . Jayakumar T. fracture. However.org Copyright by ASTM Int'l (all rights reserved).. KEYWORDS: acoustic emission. Correlations of dynamic strain ageing in 316 SS with AE signal are presented. B. e . For PE-16.. ABSTRACT: The nature of the Acoustic Emission (AE) generated during tensile deformation and fracture of two austenitic alloys i . the second phases are carbides and inclusions. No further reproductions authorized. whether Orowan looping and/or p a r t i c l e shearing is operating. 218 Copyright9 1991 by ASTM International www. AISI type 316 Stainless Steel (SS) and Nimonic alloy PE-16 are presented and discussed. the second phases are carbides and coherent precipitates i. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. ASTM STP ]077.

Steels of AISI type 316 and similar to this are major structural materials for chemical and nuclear plants. As regards PE-16. . Both the parts are based on work carried out at the authors' laboratory. Four types of arrangements are designated as Copyright by ASTM Int'l (all rights reserved). PE-16 can have several combinations of microstructural constituents that may help obtain useful correlations with AE signals. There were only a few inclusions in the nuclear grade 316 SS. Influence of temperature (298-873 K) during deformation on AE signals for solution annealed nuclear grade austenitic stainless steel material is also reported. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The sizes of these inclusions were in the range 0.0 um. behaviour of microeracks and d u c t i l i t y of coatings/hard layers.5 to 1. No further reproductions authorized. The present studies have been specifically undertaken in order to establish correlations of mechanisms (deformation and fracture) with AE signals. Three types of tensile specimens and four types of mounting arrangements were used in the investigations. The paper discusses the characteristics of AE generated during tensile plastic deformation and fracture of two austenitic alloys i . PART A STUDIES ON AUSTENITIC STAINLESS STEELS Present studies on austenitic stainless steels describe the nature of AE signals and t h e i r correlations with sources operational during tensile plastic deformation and fracture in the following microstructures. RAJ AND JAYAKUMAR ON AUSTENITICALLOYS 219 new information are kinetics of dislocations. Table 2 gives the inclusion ratings for the commercial grade 316 SS used in t h i s study. The paper is divided into two main parts. there has not been any systematic study on the characterisation of AE generated during plastic deformation and fracture of this alloy. . The studies of AE in these alloys are s i g n i f i c a n t for the following reasons: AE investigations for materials research or structural i n t e g r i t y monitoring have been very limited for steels similar to AISI type 316 SS. (b) austenitic matrix with inter and intragranular carbide precipitates. (c) carburised case and (d) austenite matrix with inclusions. AISI type 316 stainless steel and Nimonic alloy PE-16. (a) single phase solution annealed. The f i r s t part is related to studies on 316 SS and the second to PE-16. I t is known that this steel emits weak AE making i t d i f f i c u l t to obtain good correlations with microstructures and dynamic changes. e . Materials and Test Arrangements The chemical compositions of the stainless steels used in the present investigations are given in Table I. PE-16 is a nickel base alloy suitable for high temperature operations and finds applications in aerospace and nuclear industry.

C and D types. T-Thin series. Solution annealing of tensile specimens was carried out in a vacuum better than 10-5 Torr at 1323 + 2 K for I/2 h in a three zone furnace.O013.5 0. B. Cu-0. S-0. Mo-2. type A arrangement has been used. C-0.33 1. Inter and intragrannular precipitates M23C6 were introduced by thermal ageing in vacuum better than 10-5 Torr in the temperature range 773-1073 K for a maximum duration of 2000 hrs. For transfer function and acoustic amplification studies.132 Commercial 316 Ni-12. No further reproductions authorized. carried out as per ASTM: E45-85) AT AH BT BH CT CH DT DH 1. C-0.0 2. Cr-16. Mo-2. AET 5000 system and lwatsu 2100 B signal analyzer were used for recording and analysing the AE signals. type 316 Mn-l. B-Alumina type. For finding the influence of inclusions.Chemical compositions of stainless steels used in present investigations Material Chemical Composition ( Wt % ) Nuclear grade Ni-12. For studying the influence of progressive plastic deformation on frequency spectrum of AE signal generated. Time-Temperature conditions employed are indicated in Fig. For studying the influence of single phase matrix and inter and intragranular carbide precipitates on time domain AE signal.3. B type arrangement has been used for 32 x 6. B type arrangement (B2) was used. Nb. high temperature testing and transfer function and acoustic amplification studies. Si-0. 1 to II in this figure indicate time-temperature thermal ageing treatments with respect to sensitisation curve of this material. Tensile tests were performed in an Instron 1195 machine.06 1.28. type C arrangement was used.46.5. H-Heavy series A.Ol. C-Silicate type. B type arrangement (BI) was used.01. Figures l(a-d) shows the arrangements.12 ( A-Sulphide type. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.02. Ti-O. (VlRGO 14 SB) B-O. For high temperature deformation.69.Inclusion ratings in AISI type 316 commercial grade austenitic stainless steel (average of 60 f i e l d s .2 [2].0 0. P-O.054. 220 ACOUSTICEMISSION TABLE 1 -.06 TABLE 2 -.5.43. D-Globular type.0 4. type D arrangement was used.006. Table 3 gives details of experimental conditions for microstructural variables. Specimens Nos.64.0.0 2. Carburisation of tensile specimens made from nuclear grade steel Copyright by ASTM Int'l (all rights reserved).35 x 3 mm gauge length specimens designated as B1 and 50 x 20 x 3 mm gauge length specimens designated as B2. . Cr-16.

~ AE SENSOR 225 (c) 5O 35 1 II0 180 ' 3T.K . No further reproductions authorized..DPEAR~sTEUMRByEL a) (b) '.I -.. ~ C=l I TE..PUT AE SE.oAo.SO. 'c(' AE SENSOR '. (d) AE SENSOR FIG. . RAJ AND JAYAKUMAR ON AUSTENITIC ALLOYS 221 ~LOADING FIXEDTO INSTRON GRIPS CLAMP- SENSORS (~ THERMOCOUPLE CABLES 9 ".ff t.S. .Types of specimen and mounting arrangements (a) Type A (b) Type C (c) Type B (d) Type D Copyright by ASTM Int'l (all rights reserved). SPECIMEN 9 TENSILESPECIMEN FIXEOTO INSTRON GRIPS . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement..~.'~ ' I IooTPOT . ~ I (~'-FURN ACE i = pIuGH L TE..

r- ~D kOkO X X c~ X X 0 0 X X X (D C~JCM c~ O O.e" I LDC~J C~J C~J LD L~ CO O 0 c- O . 222 ACOUSTICEMISSION 4-.~-. 'T C~J ~c) CO CO C~J E o~ .1 S E ~'-..) c- % ~ E q.mO o Or) ~ % C0 E 0-4-) EC~ r0 ~ c- ILl b') I--4."0 . ~ E U r---. .(-~ .S .~ -"r..S s O 0 ~ O E O E O E ~ O ~ O ~ O ~ ~1.-~ X ~-~ S.r. O LrDL~ CD CD ~'~ O tO (i) .. COCO C~ cO LO 4-) E ~ "".~ 0 o~ ~ 4-'~ ~.J r~ Z S k-- Copyright by ASTM Int'l (all rights reserved).~ . .r."0 I-. ~CO C~ ~ oh E . I 4-)0") 0 O . r0 r'.E ~00 1 Uk- 4J ~-~ ~ O ~J I 9r . r-- E or. O~ o) -k~ E (D E c3 ~D_ ~ ~_ ~ O ~- (D ~--~-~ 0000 CO CO CO CO ~ X X X X X X X c. LO 0 r~ lJ CO O 0 0 0 O 0 C'O CO ~ 0 % E ~J COr-. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. ~.. c. ':~ U I".r.E ~ O.r- L. c- c- O E S.~ CD 0 .r.~ rw O . No further reproductions authorized.".

l ( d ) and the experimental set-up in Fig.7 V peak-to-peak was applied. which is the frequency corresponding to the f i r s t transmittance peak [3]. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The specimen mounting arrangement is shown in F i g .2 to 0.. experimental set-up for transfer mens of AISI type 316 Nuclear function and acoustic amplifica- Grade Material. before processing and analysis. The signals received were d i g i t i s e d at a sampling rate of lO 7 samples per second or a data length of 2K bytes.l and the AE signals generated during deformation were picked up by the receiving sensor. under off-normal conditions of o i l leakage.REAHPLIFIER . No further reproductions authorized. .) -' FIG. The location of AE sensors and methods of i n j e c t i n g acoustic wave were similar to that used f o r transfer function studies. I n j e c t i o n of waves has been done only for certain s t r a i n ranges enabling comparision of data obtained during this s t r a i n range with that obtained at the other s t r a i n levels where i n j e c t i o n was not done.Time -temperature thermal FIG. 173 23 4 OSC0LLOSCOPEl '/73 [ WAVEFORfl ] t73 I I I ~ ~ENERATOR lo-' ~t 101 10 ~ 10 ~ AGEING TIME (hrs. The frequency of the sighal was varied from lO0 kHz to 550 kHz.8 x lO-4s . kHz. The frequency selected for the injected signal f o r acoustic amplification studies was 156 kHz. acoustic wave with suitable amplitude and frequency was injected using a broadband AE sensor placed at one end of the specimen. For the transfer function studies. simultaneous i n j e c t i o n of acoustic signal and recording of AE data was used. tion studies. Copyright by ASTM Int'l (all rights reserved).3. The acoustic signal transmitted through the specimen was picked up by another broadband AE sensor.3 -. For acoustic a m p l i f i c a t i o n .73 1073 " l 7 -[~"OUT PUTSIONAL n. A fixed frequency acoustic signal (156 kHz) with varying amplitude in the range 0. RAJ AND JAYAKUMARON AUSTENITICALLOYS 223 was carried out in o i l contaminated sodium at 873+2 K for a period of lO00 hours to represent the carburisation phenomenon that takes place in t h i s steel when used as construction material for sodium loops of fast breeder reactors. The specimen was p l a s t i c a l l y deformed at a strain rate of 2.2 -.Block diagram of the ageing conditions for t e s t speci. The studies on transfer function and amplification required i n j e c t i o n of ultrasonic waves in the frequency range I00 kHz to 600 RE(EIVMG S(NSOR lOdlS T1213 J~l ~.

5(a)) is attributed mainly to dislocation m u l t i p l i c a t i o n by Frank-Read (FR) and grain boundary (GB) source operations and motion of dislocations. In order to ensure that no external noise was associated with extended grips and other components. a few occasional bursts at intermediate strains and a f i n a l strongest burst at f i n a l fracture. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. coupled with consistent observations in various specimens with different microstructures give credence to the use of extended grip assemblies for acquiring weak AE signals. Natural frequencies for 32 x 6. I t was found that an ideal Kaiser effect operated in the austenitic materials tested in the study. No further reproductions authorized. strain plot (Fig. Solution Annealed AISI 316 Nuclear Grade Austenitic Stainless Steel Solution annealed nuclear grade AISI type 316 stainless steel is a weak emitter as evidenced by low peak amplitude (9 to 15 dB) emissions during uniform plastic deformation (Fig.5(a)) shows a number of bursts. The response of the system for certain simulated frequencies was also found to be consistent. 224 ACOUSTIC EMISSION Ensurin 9 Worthiness of Captured Signa ! The threshold and gain for various series of experiments were so selected that no external noise was picked up. preloading using a dummy specimen to more than 1. were calculated. Ringdown counts (RDC) vs. with no material damping. was observed to be repeatable. The burst in RDC during i n i t i a l 2% plastic strain range (Fig. Displacement response for 180 x 50 x 3 mm specimen at 150 kHz. This has been carried out using a computer code. Theoretical analysis. This was repeated for a number of cycles. there was no necessity for removal and r e f i x i n g of transducer when the specimen was changed. the a c t i v i t y is mostly below the threshold Copyright by ASTM Int'l (all rights reserved).35 x 3 mm gauge length specimens for the t h i r d mode in the i n i t i a l and f i n a l (prior to necking) stages of plastic deformation were found to be 7.737 kHz respectively. This eliminated the possible variation in the acoustic coupling between the transducer and the specimen. natural frequency response of various types of assemblies was determined . discussed below. Programme for Automatic Finite Element Computation (PAFEC) [3]. Though these sources are operating throughout the plastic strain region. In order to make sure that the natural frequency response for different types of specimen geometries. . The response from a pencil lead breakage with a fixed gain and threshold conditions. The values are much less (I0 -fs m) than the s e n s i t i v i t y of the piezoelectric transducer (I0 -T4 m) at 1 microbar. When the transducer was fixed onto an extended grip assembly. There is a burst at very low strains (during the i n i t i a l plastic strain range upto 2%). The material does not show s i g n i f i c a n t variations in AE 'rms' signal (averaged for 4 seconds) during the complete tensile test. The background noise spectrum generated from a broad band sensor was used to compare the system response for various experiments.5 times the maximum expected load for any of the specimens tested was carried out. This ensured that acoustic amplification values are not influenced by displacement response of the specimen due to injection of acoustic energy. and gripping assemblies would not interfere with the AE signals.122 and 5.4 (a)).

Reasons f o r occasional burst emissions are explained l a t e r in the t e x t . The predominant frequency present in the frequency spectrum should be inversely proportional to the duration of the dynamic event. Similar feature was also observed in this study for RDC vs. 5 ] . However. These 200 SOLUTION ANNEALED I 3o 150 r ~ T H E R M A L L Y. --. the burst is a smooth curve in e i t h e r count rate or Vrms. The occasional bursts in RDC beyond 2% p l a s t i c s t r a i n and before necking are possibly due to the following reasons: ( i ) Decohesion and f r a c t u r e of inclusions .4 -. s t r a i n plot ( F i g . one or a combination of the three p o s s i b i l i t i e s discussed above appears l o g i c a l . Phase transformation during t e n s i l e deformation is ruled out as the source of AE because i t is known that formation of martensite does not occur in type 316 stainless steel during deformation at 300 K [ 4 .1 50 59 611 PEAK AMPLITUDE (riB) PEAK AMPLITUDE (dB) FIG. The burst is usually narrow and falls off rapidly with increasing s t r a i n . materials are known to have a single burst in AE near the p l a s t i c strain of 2% of the material. The f r a c t u r e surface was indicative of d u c t i l e f a i l u r e by microvoid formation and coalescence. In many materials. eddy current t e s t i n g and X-ray d i f f r a c t i o n analysis. RAJ AND JAYAKUMARON AUSTENITICALLOYS 225 level beyond y i e l d i n g e~cept a few occasional burst emissions. No further reproductions authorized.Peak amplitude d i s t r i b u t i o n plots for the AE events obtained from (a) solution annealed and (b) thermally aged specimens. ( i i ) Microcracking of matrix and (iii) Statistical p r o b a b i l i t y of strong acoustic a c t i v i t y from dislocations during deformation [ 3 ] . It is d i f f i c u l t to establish s p e c i f i c c o r r e l a t i o n of mechanisms with bursts.7hrsADED zo m 100 10 ~ so 0 n 0 t e . This is a t t r i b u t e d to the measurement mode of 'rms' value using AET 5000 system.1 50 59 68 6 14 23 32 ~.-- 0 6 14 23 32 ~. The observed burst in RDC during i n i t i a l p l a s t i c s t r a i n range upto 2% in this material is in agreement with the reported observation in metals and alloys having a single phase FCC structure [6]. . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. . Frequency spectrum analysis of AE signal: In order to f i n d out the influence of progressive p l a s t i c deformation ( s t r a i n ) on AE generated. The absence of martensite in AISI type 316 stainless steel a f t e r deformation was confirmed by examining the t e n s i l e tested specimens by f e r r i t e s c o p e . 5 ( a ) ) . . no burst in Vrms was observed. frequency analysis of AE signal has been carried out. We a t t r i b u t e the burst in RDC at f r a c t u r e to f i n a l linkage of microvoids and microcracks formed during early stages of deformation including necking. . . However. Thus the measurements can provide information on the time scale of the AE process and can thus be used to evaluate dynamics of source events [7] Copyright by ASTM Int'l (all rights reserved).

5 10 15 20 2S 30 3$ ~. the predominant frequency increases from 0.. aged specimen.5 SO 55 60 30 59 % STRAfN (TOTAL) % TOTAL STRAIN- FIG. The value of predominant frequency in the necking region is arrived at by averaging ten AE signals in each of the three specimens.50 ~o u-o ~.14 MHz which is about I/3 to I/4 of those found in the uniform deformation region of the stress . whose l i f e times are reported [8] to be higher than d i s l o c a t i o n related AE phenomena. No further reproductions authorized. The lower value of predominant frequency in the necking region could be associated with microvoid coalescence and / or crack growth. AE source related to waiting time of d i s l o c a t i o n at obstacles resulted in reasonable values of mobile d i s l o c a t i o n density (103 .L 200 150 3000 100 50 l z I I f z f f i 9 i 0 . 4 ( a ) ) . frequency of AE signal with ution annealed and (b) thermally strain..5 -. Copyright by ASTM Int'l (all rights reserved).0 4.104 /cm 2) [ 7 ] .66 MHz varying l i n e a r l y with t o t a l s t r a i n (Fig..475 MHz to 0. The s h i f t in predominant frequency during 1300 ~ 7O0 650 .. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.6 -. the predominant frequency has a low value of 0. b) In the necking region.. Acoustic a m p l i f i c a t i o n : AE generated during t e n s i l e deformation in AISI type 316 nuclear grade a u s t e n i t i c stainless steel are weak ( F i g .Variation in predominant counts with s t r a i n for (a) sol. 6).~I~'A ~ ~ z.Variation of ringdown FIG. Attempts to correlate t i m e . 250 o. . .f l i g h t with predominant frequencies resulted in unreasonable calculated values of mean free path (a few m i l l i m e t r e s ) . An innovative technique has been developed to amplify such signals [ 9 ] . ..00 z 3S0 6000 300 l THERHALLY AGEO c~ 1073K / 17Hr~..s t r a i n curve... uniform p l a s t i c deformation is explained on the basis of dynamic events related to waiting time of dislocations at obstacles [ 7 ] .oo 6O0 o o :o ~ 1 7 6 550 10 20 30 40 SO 60 663 ~0 0 A OA6u & X TOTAL STRAIN 500 ~.o f . 226 ACOUSTICEMISSION Two main observations are as follows : a) Beyond 2% p l a s t i c s t r a i n and upto the onset of necking.

although no such suggestion has been made in the l i t e r a t u r e . 7. and tend to promote fracturing of these sites [ I 0 ] .e 12S| O.Time signal and its auto-power spectrum for a typical continuous type AE signal. Figure 8 depicts net acoustic energy (total energy minus the energy due to the injected component) of various continuous AE signals recorded as a function of the strain levels for both injected and non- injected conditions. I t is thus plausible that externally injected acoustic signals of a certain frequency can interact with subcritical (energetically) AE sources and give rise to enhanced acoustic emission signals which otherwise would not h a v e been released at those stress levels. RAJ AND JAYAKUMAR ON AUSTENITICALLOYS 227 I t has been reported that spontaneous stress (elastic) waves emitted by various nucleation (microfracture) events interact with other potential nucleation s i t e s .$p n -7. The possible explanation of amplification is that the injected p SeC* p SSC. 7) are in a r b i t r a r y units. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Further. No further reproductions authorized. The data for the injected case at a strain level where the new frequencies are dominantly seen is given in Fig. Copyright by ASTM Int'l (all rights reserved). I t was not necessary to eliminate any other energy components as no harmonics corresponding to the injected signal frequency were observed.. -12Sm IIYY[IIVIIIVYf#fYIYIflYIIYVl Z2. The values of Y-axis (Fig. 20~J !o4. The AE signals emitted under either injection or otherwise were acquired at various strain levels. I t should also be emphasised that the emission of frequencies in the neighbourhood of the injected frequency is typical for a l l values of the injected voltages.6p -122. I t must be emphasised that the amplitude level of the frequencies without injected acoustic wave are more than one order of magnitude less dominant than that with the frequency components with the injected acoustic wave as may be noted from Fig.5 PIHz MHz Z~ NQN-INJEETEO INJEETEO FIG.1p I J I I I I I I I I I I I ~ I I I I 2.7 -. similar p o s s i b i l i t y of stimulating subcritical sources in the case of continuous AE appears to be more favourable considering the energetics involved. . I t has also been reported from damping studies that plastic deformation can be produced by injecting waves of very large strain amplitude [ I I ] . The spectrum of the signals recorded with the injected condition contains a strong peak at the injected frequency in addition to new frequencies. Figure 7 shows typical time domain signals and the corresponding auto-power spectra of continuous type AE signals for the non-injected and the injected conditions. r r r f J p i r i ilitLitlA tAittiimih ' ' ' ' ' ' i m. 7.

. These would then result in the enhancement of integrated power. there is no reported l i t e r a t u r e on the AE behaviour from the serrated p l a s t i c flow occurring in AISI type 316 stainless steel.. there would be a d i s t r i b u t i o n of relaxation time scales implying that.. .xx 9 x . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.. 05u o A_ 0. in general....~.5 % STRAIN (ENGG) FIG....13]..8 -... there is systematic changeover from i r r e g u l a r to A to B to C type serrations with a change in the t e s t i n g conditions [14].. there w i l l be a d i s t r i b u t i o n of frequencies around the injected frequencies with the maximum amplitude corresponding to the injected frequency (due to the resonant type of i n t e r a c t i o n ) . Further.... ... Associated with the d i s t r i b u t i o n of b a r r i e r s . Influence of temperature: Various investigators have studied AE generated during serrated flow from various materials l i k e A1 a l l o y s .. ~.. brass etc.. ! . A A A~. ...... Since...... Further.... environment of the activated energy releases. ... . INJECTEO) x = 9 x 9 x . there is a p o s s i b l i t y of the release of acoustic energy at other frequency bands when new subcritical sources are stimulated under the influence of injected acoustic waves. ~ .t~V 0. The acoustic a c t i v i t y increased with increase in temperature of deformation upto 773 K.. this material is ideal f o r c o r r e l a t i n g the AE behaviour with various types of serrations.. .. No further reproductions authorized. An w 10 84 A Aoo A < o 9 o ~ ~176 o~ 9 o oo o ~b o Q cr A " ^ UNJECT[m t 9 x x xx " ~ x x (WITHOUT (~ 9 100 xx ... and correlated them with the types of serrations [12.. There is.. since the relaxation time scales are expected to depend on the local 10 84 9 .~.. 228 ACOUSTICEMISSION acoustic wave would t r i g g e r the groups of dislocations arrested at obstacles to overcome energy b a r r i e r stimulating the s u b c r i t i c a l sources to contribute additional sources of AE. There was reduction in acoustic a c t i v i t y when Copyright by ASTM Int'l (all rights reserved)...7V a. ~ . The proposed mechanism explains both the amplification of the t o t a l AE power and the frequency response under the influence of injected signals. 5 10 15 20 25 30 35 40 1... i t is expected that there would be a d i s t r i b u t i o n of energy barriers which can only be overcome by increasing the amplitude of the injected acoustic wave. however no compelling reason that new frequencies should be found in every s i t u a t i o n ....Acoustic amplification for various injected voltages.. However... .......

No further reproductions authorized.1 2.7 2.5 . burst type AE occurred during serrations. AE occurred during load drops.3 ?. AE recording system and s t r i p chart recorder [3.15].0 1. are not known and ( i i ) the time scale involved for dislocation phenomenon associated with serrations is orders of magnitude lower than that of load measuring system. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.0 II < 0. F in Fig. 9(b) or during small load fluctuations (points E. elongation for a few representative serrations.9 ELONGATION (mi~ - FIG. At 548 and at 773 K. At 588 and at 873 K. ~verage velocity or both. I | i I 1 J l _ l I I i 10+7 10. when the serrations occurred over a longer period (as shown in Fig. 9 (c)) taking place immediately after C type serration. Continuous type AE was also observed I?]K (c) < (a) 583K (b) 773K F E I ! I A t 3. AE occurred during load drops for a l l serrations and during load rise for a few serrations. Rate of load drop rather than the magnitude of load drop or load rise decides the energetics of AE signals. Copyright by ASTM Int'l (all rights reserved).AE rms voltage peaks in correlation with change in load during serrated flow in austenitic stainless steel at various temperatures (a) 583 K (b) 773 K (c) 873 K. Eventhough the amount of load drop at point A was smaller compared to that at point B in Fig. . 9(a). Caceres et al.1 11.9 1.2 7. which cause d i f f e r e n t types of serrations and which also strongly affect AE.3 1~3 ?.4 2. the former generated higher rms voltage since the load drop occurred at a faster rate.9 11. [13] reported generation of AE during load rise or drop in different materials and correlated them/with the processes related to dislocation phenomena.9 -. They occurred either during load drop or during rise or during both f o r different serrations. In general. The serration changed over from irregular (548 K) to type A (588 K) to type A+B (773 K) to type B (lower strains at 873 K) and type C (higher strains at 873 K). I ~ i J I I .1 2.5 2. RAJ AND JAYAKUMAR ON AUSTENITICALLOYS 229 temperature of deformation was further increased to 873 K. The results of our work indicate that a d e f i n i t e conclusion is not always possible because ( i ) the r e l a t i v e contribution of change and rate of change in mobile dislocation density. There is one to one correspondence between serrations and AE generated. Figure 9 shows the portions of load and AE rms voltage vs.

twin boundaries and dislocations [16]. 4 (b)) 773K 10a . 15 .40 dB. .( io i . IO -. In a solution annealed and thermally aged material. I t can be seen that there is higher AE a c t i v i t y in thermally aged specimen (RDCs). No further reproductions authorized. FflAC]UR( x TOTAL TiLL FRACTURE FIG. Copyright by ASTM Int'l (all rights reserved). This broad peak was observed in a l l the thermally aged specimens where grain boundary cracking occurred. 17. The maximum amplitude values corresponding to second broad peak (for example 40 dB in Fig. the expected sequence of precipitation as a function of time at a given temperature would be grain boundaries. i i__ AGEING TIME (hrs] 106 973K I l~' I07)K t i i i _ |'~ I 101 10 ~ I0 1 101 ':0 ~t AGEING '(IME (hrs. Thermally aged specimens show an additional broad peak in the amplitude range. Figure 5(b) gives typical RDC vs. Figure I0 shows representative plots indicating variations of AE data obtained during t e n s i l e testing of thermally aged specimens. Figure 4 (b) shows typical amplitude plots for AE events generated in thermally aged (1073 K. Figure I I shows variation of the maximum values of various AE parameters with thermal ageing. AGF'ING 11ME ~rsJ 9 UPTO2 . 230 ACOUSTICEMISSION Effect of Second Phase : Inter and Intragranular Carbides in Nuclear Grade AISI 31'6 Stainless Steel Predominant precipitates formed after the thermal ageing treatments are inter and intragranular M23C6 precipitates. ~ STRAIN 0 UPTOUNIFORHELONGATION V DURINGNECKINGELONGATION8.)-. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.Typical plots showing variation of cumulative ringdown counts / events from specimens aged at various temperatures and time durations. total s t r a i n for thermally aged specimen.7 hours) specimen cumulated t i l l uniform elongation.

In Fig. I0. There is however a change in the points on the time scale at which the particular variations have taken place.3 % strain level. there was a gradual change even within the mixed mode type. No further reproductions authorized. Increased AE a c t i v i t y at a l l temperatures at higher thermal ageing times is attributed to increased grain boundary cracking due to favourable size and morphology of precipitates. I t was observed that there is a progressive change in fracture mode from f u l l y ductile type by microvoid formations and coalescence to mixed mode type containing ductile microvoids (dimples) on grain facets and intergranular cracking with the increase in duration of thermal ageing at each of three temperatures. 973K 1073K 1500 150 1300 130 ~0 Y 1100 110 30 <c 900 90 20 :_ 700 70 10 soo so o I I I lO' lO~ lO' AGEING TIME (rain) AGEING TIME (rain) 9 PEAK AMPLITUDE/EVENT x RINDOOWN COUNTS/EVENT o EVENT DURATION/EVENT FIG. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Figure 2) and decrease in grain boundary cracking.II -. (ii) Reduced dislocation mean free path resulting from the f i n e l y dispersed intragranular carbide precipitation thus reducing the statistical p r o b a b i l i t y of detection of weak AE phenomena as reported by Agarwal et al. As the ageing time increased. in the sense that the dimples on grain facets were reduced and the intergranular cracking was increased [17]. The decrease in AE a c t i v i t y at 973 K and 300 hours is attributed to desensitisation (point 8. The arguments though a t t r a c t i v e are not amenable to be isolated or confirmed in this present analysis.Maximum values of various parameters of AE events below which most of the events generated are present. RAJ AND JAYAKUMAR ON AUSTENITICALLOYS 231 for various thermally aged specimens were used for making t h i s plot. . [18] and ( i i i ) Size and morphology of grain boundary M23C6 precipitation at lower times of ageing treatments is such that i t strengthens the boundary and reduces the dislocation a c t i v i t y and d e t e c t a b i l i t y . The decrease in AE a c t i v i t y upto 2. The cracking propensity was observed by SEM studies to increase as the ageing time increased. Copyright by ASTM Int'l (all rights reserved). i t can be seen that similar trends in the variations of events/ringdown counts with duration of ageing are observed for deformation upto 2.3 % strain during i n i t i a l time periods of thermal ageing can possibly be due to: (i) Diffusion of chromium or substitutional elements to dislocations thereby reducing t h e i r mobility and d e t e c t a b i l i t y .

The second reduction in AE a c t i v i t y during uniform elongation for specimens aged at 973 K for 300 hours is attributed to the occurrence of less i~tergranular cracking due to desensitisation (point 8.(ii) and ( i i i ) as suggested for dynamic phenomena upto 2. 232 ACOUSTIC EMISSION The i n i t i a l drop in AE a c t i v i t y between 2. Reduction in ringdown counts per event and SOLUTION ANIIEALEO CARDURIZED i .12 -. I 23s %6 t. 17. The increase in ringdown counts for a specimen aged for 300 hours at 973 K is due to the a v a i l a b i l i t y of higher cross-sectional area for fracture.5 hours at I073 K is attributed to the decrease in the net available cross- sectional area for fracture by progressive intergranular cracking during uniform elongation.5 hours specimen as compared to I073 K.3 % strain and s t a r t of necking at 773 K and 973 K can be explained by the same arguments as (i). I13. Fig. This is due to reduced intergranular carbide precipitation and hence reduced grain boundary cracking (confirmed by SEM) during uniform elongation. . l l ) . The i n i t i a l decrease is not observed for I073 K. The d u c t i l i t y effects are less pronounced for I073 K. 300 hours as compared to thermally aged specimen at 973 K. The reduction or increase in AE events or ringdown counts generated during necking elongation corresponding to increase or reduction in propensity for intergranular cracking during uniform elongation observed in specimens thermally aged at 973 K and I073 K could not be seen in a specimen thermally aged at 773 K.7 hours specimen.) This increase in d u c t i l i t y value does not decrease AE a c t i v i t y . No further reproductions authorized. Effect of carburisation on acoustic emission during tensile testing.3 % strain. Copyright by ASTM Int'l (all rights reserved). The reduced intergranular cracking is also reflected in the increase in uniform elongation value from 43 to 46 % and total elongation from 50 to 57 % for thermally aged specimen at 973 K. The reduction in ringdown counts during necking elongation and fracture in a specimen aged for I08 hours at 973 K and for I13. time duration per event. peak amplitude per event (Fig. (Uniform elongation shows an increase from 50 to 52 % and total elongation increases from 59 to 62 % in the former specimen. This is due to the p o s s i b i l i t y that the size and morphology of grain boundary precipitates at this temperature are favourable for cracking at lower thermal ageing times too. This is also confirmed from the observation that higher peak amplitude events corresponding to grain boundary cracking have been generated after lower thermal ageing times at I073 K (Fig. The increase in values of such parameters as ringdown counts per event. I06 hours. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 2). l l ) are due to contributions from more and more intergranular cracking.~ 6&5 % STRAIN FIG.

SEM revealed mixed mode type fracture in the core and intergranular cracks in the carburised case. shape and d i s t r i b u t i o n s of inclusions in different specimens. 40 % of the total events occurred during yielding and upto 2.3 % strain to the end of uniform plastic deformation for 973 K and 1073 K thermally aged specimens. e . The material is s i l e n t almost during the entire plastic deformation stage. trends in AE a c t i v i t y for 2. The present study compares AE release from commercial and nuclear grade AISI type 316 stainless steels. Also. In addition to normal peaks at yielding and fracture. In case of commercial grade steel. size.3 % plastic deformation.5(a)). The r a t i o of ringdown counts upto 2% strain to the total ringdown Copyright by ASTM Int'l (all rights reserved). . Effect of Inclusions in Nuclear and Commercial Grades of AISI type 316 Austenitic Stainless Steels Comparison of nuclear and commercial grades of a material is logical for identifying AE generated by fracture and debonding of inclusions. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. .3 % strain and from 2. The s i g n i f i c a n t difference between the two grades is t h e i r inclusion content. strain plots show considerable variations in AE obtained from different specimens of the same batch of materials. Optical metallography showed extensive inter and intragranular carbide precipitation in carburised specimen upto a case depth of 6 % resulting due to carbon pick-up from o i l contaminated sodium and beyond this depth precipitation was observed to a less extent (due mainly to thermal ageing). the RDC vs. an additional broad peak was seen extending over a strain range 5 to 15 % with a maximum at 9 % strain. in carburised specimens. Figure 13 shows typical variations of AE ringdown counts with t o t a l strain in commercial grade stainless steel. No further reproductions authorized. I t agrees with the observation. This variation in acoustic a c t i v i t y (but similar burst emission in the range 2 to 25 %) is attributed to variations in size. In the solution annealed material. There is four to f i v e times increase in the total AE events in the carburised specimen compared to solution annealed specimen. The type. Thus. I t has been inferred from AE studies that inclusions that fracture at high stresses and with higher crack velocity produced stronger AE events. RAJ AND JAYAKUMAR ON AUSTENITICALLOYS 233 time duration per event at 973 K beyond 108 hours of ageing is attributed to the reduction in cracking tendency. a large number of events in the carburised specimen were of higher peak amplitudes. i . On the other hand. time for both solution annealed specimen and specimen carburised in o i l contaminated sodium. The study brings out the potential of AE technique for characterising mechanical properties of case hardened layers. there exists a p o s s i b i l i t y ( i n d i r e c t l y ) to characterise inclusions and t h e i r behaviour in a particular material by observing AE a c t i v i t y . AE ringdown counts with total strain plot for nuclear grade material is representative of d i f f e r e n t specimens as the variations from specimen to specimen are minimal (Fig. Influence of carburised case: Figure 12 shows plots of AE events vs. AE was observed throughout the tensile t e s t . composition and other features of inclusions have been characterised in numerous AE investigations [19- 21].

Copyright by ASTM Int'l (all rights reserved). . 1000 [] NU[LEARBRAOE oAlx OX~ } COMMERCIA~RAOE L 0 &x 100 Ae o 8 8 16 24 32 ~. 234 ACOUSTICEMISSION counts generated upto necking reduced d r a s t i c a l l y in commercial grade material ( t y p i c a l l y 1. In one typical commercial steel specimen.total strain for commercial grade AISI type 316 stainless steel.Typical plot of RDC vs. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 96% of t o t a l ringdown counts generated before fracture were generated during 2 to 1211~ t ~6400 r i 20 30 41 SO TOTALSTRAIN 9 FIG. No further reproductions authorized.0 48 56 64 PEAK AHPLITUDE (dB) FIG.2% in one p a r t i c u l a r case) compared to nuclear grade material ( t y p i c a l l y 57%). This is attributed to the presence of inclusions in the former causing a reduction in the dislocation mean free path which in turn reduced the AE a c t i v i t y generated before 2% t o t a l strain. 13 -.Logarithmic cumulative peak amplitude d i s t r i b u t i o n of events generated in nuclear and commercial grade AISI type 316 stainless steel specimens. 14 -.

the commercial grade specimens show similar slope ( b .4. the net cross . This agrees well with higher AE a c t i v i t y before 25 % total strain level in commercial grade steel. T i / l . Si/0. The figure shows generation of more number of events with higher peak amplitudes in commercial grade specimens compared t ~ those in nuclear grade specimens. No further reproductions authorized. Figure 14 shows logarithmic cumulative peak amplitude d i s t r i b u t i o n plot for events generated in nuclear grade and commercial grade steel specimens. As most of the separations of cross-section of the fractured region in commercial grade steel occurred during e a r l i e r stages of plastic deformation (by inclusion decohesion and cracking).5. The ringdown counts at fracture in commercial grade material are reduced considerably compared to nuclear grade material. Mn/O. Fe/33. More number of AE events with lower PA are detected possibly due to larger mean free path in nuclear material as compared to commercial grade material. This agrees well with higher AE levels in a l l the specimens tested. I t has been found by SEM observations using stop tests at various strain levels that fracture/decohesion took place before 25 % strain levels in a l l specimens tested. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 14. A I / I . This is attributed to the large number of ringdown counts generated during decohesion and cracking of matrix and inclusions. .parameter from 24 dB to 48 dB) indicating that the events generated in commercial grade specimens r e s u l t in a characteristic b .sectional area separated at fracture is reduced r e s u l t i n g in lower amount of ringdown counts at fracture. Also.26. PART B STUDIES ON NIMONIC ALLOY PE-16 Studies have been carried out on the influence of the following three types of second phases in a ~-austenite matrix :(a) MC type carbides. Mo/3. the figure indicates that there are large number of events with lower peak amplitude (PA) in nuclear grade steel. 2 .07. Materials and Test Arrangement The nominal chemical composition (wt %) of Nimonic alloy PE-16 used for investigations is as follows: C/0. RAJ AND JAYAKUMARON AUSTENITICALLOYS 235 25% of the s t r a i n in comparision to 33% in the nuclear grade material. The RDC at fracture are found to be consistently higher (for d i f f e r e n t specimens tested) in nuclear grade material compared to those generated in commercial grade material. Copyright by ASTM Int'l (all rights reserved). Stop tests had revealed that AE a c t i v i t y due to inclusion decohesion and fracture is in the peak amplitude range of 24 dB to 48 dB.3. Cr/16. The total ringdown counts observed before fracture are almost doubled in commercial grade steels as compared to nuclear grade material. (b) Xland (c) conjoint presence of ~ a n d MC.8. Even though the number oF events generated in d i f f e r e n t specimens have varied at higher amplitude levels. The representative data for one nuclear grade and data from four different lots of commercial grade specimens are included in Fig.parameter. 2 4 . Lower PA events are attributed to dislocation generation and motion at the early stage of deformation.

Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Total system gain was 99 dB. Threshold voltage was 0. to obtain Y' with 9 nm average size (type N-3).03 and Ni/balance. PE-16 t e n s i l e specimens. Deforma-i AE parameters during p r e .2 x 10-4 s . B/O. S/0. operating deformation processes and various time domain AE parameters f o r d i f f e r e n t specimens are given in Table 4.27. Zr/O.Microstructural d e t a i l s .35 x 3 mm were tested at 298 K at a nominal s t r a i n rate of 5.y i e l d type struc. AE was recorded with 375 kHz resonant sensor (250-500 kHz f i l t e r ) . No further reproductions authorized. Thermal ageing was carried out on type N-I specimens at 973 K f o r 24 hours.tion Total Total Max. operating deformation processes and various time domain AE parameters f o r d i f f e r e n t specimens Specimen ~icro. p a r t i c l e shearing process * * * Size of ~ = 20 nm~ Orowan looping processes * * * * i) Dislocation a c t i v i t y at Y / c a r b i d e interface and i i ) Decohesion and Fracture of MC * * * * * All the above deformation processes.O015. Double thermal ageing was carried out on type N-I specimens at 1173 K f o r 2 hours and 973 for 24 hours to obtain u 2 4 7 MC microstructure (type N-5).06.Mo) (type N-2).75 V. Influence of MC type carbides: Figures 15(a-b) show 'rms' voltage of AE signal vs. values ture process events RDC PA(dB) ED(us) RT N-I x * 63 780 27 650 225 N-2 r. having gauge dimensions 32 x 6.004. low volume f r a c t i o n of 7~(I0 nm) Deformation Mechanisms~ AE Results and t h e i r Correlations Microstructural d e t a i l s .I . Solution annealing was carried out at 1313 K f o r 4 hours f o r a l l specimens to obtain single phase austenite (type N-I).Mc **** 400 19500 45 2500 175 N-3 Y+ r ' ** 3550 32000 40 2000 500 N-4 r + r~ 875 32750~ 32 I000 325 N-5 Y+~'+Mcl 3825 44000i 40 2000 500 * Dislocation m u l t i p l i c a t i o n with Frank-Read and grain boundary sources and dislocation motion ** Size of YJ= 9 nm. In a l l the specimens tested. This was done in order to h i g h l i g h t the AE data during micro and macro y i e l d i n g . plots showing 'rms' voltage v a r i a t i o n with s t r a i n were made with the X-axis scale being selected to be ten times more (expanded) ~pto around 2 % t o t a l s t r a i n as compared to that used beyond 2 ~ t o t a l s t r a i n . 1 (a)) arrangement was used for a l l the investigations. . 236 ACOUSTICEMISSION Co/0. Approach for ensuring worthiness of captured signal was the same as mentioned f o r a u s t e n i t i c stainless steels. Cu/O. TABLE 4 -. s t r a i n plots for solution annealed condition Copyright by ASTM Int'l (all rights reserved). Type A (Fig. 1023 K f o r 24 hours to obtain r' with 20 nm average size (type N-4 ) and 1173 K for I/2 hour to obtain MC (M = Ti.

.. % TOTAL STRAIN ~ (b) zi..2 IO. . .25 i. . .. . .11 .c o.25 ..7! Ill i 131 l 2..5 0 0.5 1..11 1. . ~ I J I .l'l ' ' It..a 11 i13 3.. A T O T A L STRAIN (a) 1. xl li.~ "1o~8 2ito ~1. .z .l 16. ..7 TOTAL STOAlil (e) FIG..-- > 1.s .0 j ' 20 15 ' 25' ' 32 % TOTAL STRAIN . .'.1. L_. ~ 0.li ll &J " !q. 3..6 2. L . RAJ AND JAYAKUMAR ON AUSTENITIC ALLOYS 237 1.i li~l 27. .15 . . 7 .75 ' 1. i .1 22.2 3'9 9 "~$ 10..5 1..V a r i a t i o n in AE rms v o l t a g e w i t h % t o t a l strain (a) Type N-1 (b) Type N-2 (c) Type N-3 (d) Type N-4 and (e) Type N-5. -. . .5 2i 4 3. Copyright by ASTM Int'l (all rights reserved)..'..I 303 % TOTAL STRAIN (d) u ! z e. J .. cr 0 ' 0.1 I . No further reproductions authorized.. - (c) 1...ff I 1 i l I.Xo ~.4 i 3.5 1. 7 i 15 z~. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. J.0 L 0 . 2.0 0.1 lll.5 1.

Only occasional bursts were generated in solution annealed specimen~ Burst emissions represented by high ringdown counts and 'rms' voltage peaks observed in specimens with MC type carbides during the early part of sample laading can be attributed to decohesion of bigger size and favourably oriented carbides (with respect to the t e n s i l e axis). No further reproductions authorized. number of burst emission peaks seen in 'rms' voltage vs.24]. The yielding starts on a microscopic scale i n i t i a l l y at favourable regions and progresses throughout the gauge volu~e as the load increases. This is attributed to reduced ligament length for incremental joining of the microcracks (formed by decohesion and fracture of MC carbides during plastic deformation) during f i n a l fracture.15(d)). Influence of coherent 7': The large scale increase in acoustic a c t i v i t y indicated by large number oF AE events (Table 4) and very high 'rms' voltage peaks i~ the y i e l d region in specimen of type N-3 (Fig.24]. However when the process of deformation changes from p a r t i c l e shearing to predominantly Orowan looping (confirmed by TEM studies) as in the case of type N-4 specimens there is considerable reduction in the number of AE events (Table 4) and complete absence of AE 'rms' peak in the y i e l d region (Fig. 238 ACOUSTICEMISSION and microstructure with MC type carbides. This agrees with the results of Gladman et al [22]. AE events with r e l a t i v e l y higher PA are generated.-38 % (Fig. As 7' size approaches the optimum size of I0 nm. have indicated that depending on the type. However. This also results in increase in 'rms' value oF the signal. The broadness of the 'rt~s' voltage peak during y i e l d i n g stage in type N-3 specimens is attributed to progressive y i e l d i n g taking place in the complete volume of the specimen. s t r a i n plot and also high energetic acoustic events indicated by peak amplitude (PA) in thermally aged specimens compared to solution annealed specimen are attributed to the events generated by decohesion and fracture of MC type carbides [23. 15(b)). decohesion and/or fracture can s t a r t at small strain levels well below y i e l d point. Very high acoustic a c t i v i t y generated during plastic deformation s t r a i n range 8. Generation of higher ringdown counts during post-yield and pre-necking region of s t r a i n in specimens of N-4 type was observed. size and d i s t r i b u t i o n of the second phases. This was Copyright by ASTM Int'l (all rights reserved). Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The plots show occurrence of burst type AE in thermally aged specimens throughout the t e n s i l e test. .26]. Generation of AE events with higher event duration and r i s e time during yielding stage in types N-3 and N-4 specimen as compared to those generated in solution annealed specimens (type N-I) indicates that events associated with p a r t i c l e shearing and Orowan looping processes are characterised by slow build-up and longer duration compared to the dislocation processes in solution annealed specimens. 15(c)) is attributed ]o deformation taking place by p a r t i c l e shearing process [23. AE during fracture in thermally aged specimens are of lower energy. The above observations are in agreement with those found in p r e c i p i t a t e hardenable stainless steel and some aluminium alloys wherein the influence of shearable precipitates under d i f f e r e n t ageing treatments have been studied [25.24]. The favourable regions for early yielding are grains favourably oriented with respect to tensile axis and microscopic regions with f i n e r size of 4'. Gladman et al. ringdown counts generated in specimens of type N-4 are comparable to these in type N-3 specimen [23.

Bhattacharya.Kalyana- sundaram. Influence of conjoint presence of Y~ and MC: Generation of higher events and ringdown counts in N-5 type specimens compared to those in N-3 type specimens is due to contribution from decohesion and fracture of MC present in specimens of N-5 type. event duration and rise time for the events generated are d i s t i n c t l y d i f f e r e n t . for their encouragement an~ support. InFluence of carbide precipitates and inclusions can be characterised by conventional time domain AE parameters in AISI type 316 stainless steels. Mr.B. Prof.Rao.P.Placid Rodriguez.C. Bangalore and Prof.Sundaram. Even conjoint presence of phases like and carbide precipitates do not a l t e r the unique signatures due to and MC type carbides when they operate independently. Indira Gandhi Centre for Atomic Research and Dr.A. RAJ AND JAYAKUMAR ON AUSTENITICALLOYS 239 attributed to higher work hardening rate that is associated with Orowan looping process.K.O. ACKNOWLEDGEMENTS Authors are thankful to Mr. CON]LUSIONS Frequency specLrum analysis and acoustic amplification provides means to characterise weak AE sources due to dislocation movements in solution annealed austenitic stainless steel. Metallurgy and Materials Programme. Division for Post- Copyright by ASTM Int'l (all rights reserved).C.Jha and Mr.D. Indira Gandhi Centre for Atomic Research. 15 (e)). The characteristic signatures during plastic deformation are unique. . The energy of events generated with this microstructural condition is lower as compared to N-2 type specimens.Prabhakar of In~ian I n s t i t u t e of Technology Madras.B. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Sincere thanks are due to Prof. I t was observed that characteristic AE signatures observed in specimens containing Z'(type N-3) and MC (type N-2) i n d i v i d u a l l y have been maintained even when the two phases are present conjointly in the microstructure (Fig.K. e . .S.Ranganathan of Indian I n s t i t u t e of Science. the maximum value of peak amplitude. energy. No further reproductions authorized. Head. Mr. for many useful discussions during the course oF the studies~ Authors are also thankful to Mr. The AE events associated with the above mentioned deformati)n processes can be distinguished since the AE parameters i . Director. These is attributed to possible reasons such as: ( i ) the reduction in i n t e r f a c i a l energy between matrix and MC and ( i i ) higher attenuation of AE signal in t h e presence of ~' in the matrix.V.Rajagopalan. The strength of AE signals in PE-16 material decreases in the foll)wing order ( i ) Decohesion and fracture of MC type carbides ( i i ) Particle shearing process ( i i i ) Orowan looping process (iv) Dislocation generation an~ m u l t i p l i c a t i o n by operation of Frank- Read and grain boundary sources and mo~ion of dislocations.

Kalyanasundaram. "Influence of Deformation on Sensitisation kinetics and its Microstructure Correlation in a Nuclear Grade 316 Stainless Steel". March 1982.H. Non destructive Testing..C. [2] Mannan. Ph. February 1981. [II] Swift. "Frequency Spectrum Analysis of Acoustic Emission Signal during Tensile Deformation in an AISI 316 Type Austenitic Stainless Steel".12.H. Oxford. [5] Seetharaman. "Stimulated Amplification of Acoustic Emission Signals during Deformation". Vol.82-91. "Untersuchungen uber das auftreten Geraushen beim Zugversuch". pp.2. Samuel.. Transactions of Indian Copyright by ASTM Int'l (all rights reserved). ..l.H. and Carpenter. 1976. and Rodriguez. [4] Mannan.E. Spectral Analysis of Acoustic Emission". September 1984. pp. pp. Pergamon Press.R..l-8.K. No. for many useful discussions d~ring the preparation of the manuscript.S. ~nd Rodriguez.C.S.35. Journal of Applied Physics. presented at the Third Acoustic Emission Symposium.Kimura.P. "Acoustic Emission and Deformation Bands in AI-2. Kuganagi. Kalpakkam. Vol.18.2211-2215. and Krishnan. Ph.B.P. pp.L.R.S. pp.P. "~nfluence of the Martensitic Transformation on the Deformation Behaviour of an AISI 316 Stainless Steel at Low Temperatures"~ Journal of Material Science.. and Parvathavardhini. Thesis.5%Mg and Cu-30%Zn". Ananthakrishna.A.Baldev. U.V. Technische Hochshule.2851-2864. pp. Barat.255-257. [14] Mannan. [8] Curtis.H. RFP-4046. I. No. in Advances in Fracture Research Vol.H.T.S. 240 ACOUSTIC EMISSION Irradiation Examination and NDT Development.L. Vol...D.M.. and Jayakumar.3-10. Colorado. and Nakasa. Scripta Metallurgica. Indira Gandhi Centre for Atomic Research.H. pp.G. Samuel. No. Tokyo. Bangalore.. Vijayalaxmi.G. [3] Raj. "Observations on Predic~i)n of Non-self-similar Sub-critical Crack Growth and Stress Intensity Distributions".K.2.3.4.. Voi. Vol.. Joseph. April 1947.R.4i7:~25.Ao.K.G. ~p. "Internal Friction of Zinc Crystals".J. No. N.G. and Rodriguez.D. Jha. and Rodriguez. Dayal. "Dynamic S~rain Ageing in Type 316 Stainless Steel". Acta Metallur~ica.Baldev. Journal of Nuclear Material. "Acoustic Emission produced by Deformation of Metals and Alloys".7. N) 8. [9] Raj.. No further reproductions authorized. December 1987. Report No. 1987. "Acoustic Emission for Characterising Defon~ation and Fracture during Tensile Testing in Austenitic Stainless Steels".Baldev. Vol. 126.W. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement..C. [12] Heiple.K.H. Indian I n s t i t u t e of Science. Voi. and Richardson. 1982.15.H.. "Stress-Strain Relation for 316 Stainless Steel".. April 1974. 1989.P...523-530.. Scripta Metallur~ica. No. Acta Metallur~ica. [7] Raj.P.. 1950. [13] Caceres.S. (In Press). "Acoustic Emission Characteristics during Tensile Tests of various Metals with several Types of Crystal Structures". August 1989. REFERENCES [I] Kaiser. [I0] Smith. I. No.16.. Rockwell International. Munich. Thesis. [6] Imeda..37.B.

RAJ AND JAYAKUMAR ON AUSTENITIC ALLOYS 241

I n s t i t u t e of Metals~ Voi.36, No.4 & 5, August/October 1983,
pp.313-320.
[15] Raj,Baldev, Jayakumar,T., Kalyanasundaram,P., Barat,P.,
Jha,B.B., Bhattacharya,D.K. and Rodriguez,P., "Comparative
Studies on AE Generated during Luder's Deformation in Mild
Steel and Portevin-Le Chatelier Effect in Austenitic Stainless
Steel", Journal of Acoustic Emission, Vol.8, No.l-2, January-
June 1989, pp.S149-S152.
[16] Lai,J.K.L. and Wickens,A., "Microstructural changes and
variations in Creep D u c t i l i t y of Three Casts of Type 316
Stainless Steel", Acta Metallurgica, Voi.27, No.l, January
1979, pp.217-230.
[17] Raj,Baldev and Jayakumar,T., "Acoustic Emission during Tensile
Plastic Deformation and Fracture in Austenitic Stainless
Steel,"Journal of Acoustic Emission, Vol.7, No.3, July 1988,
pp.SI-Sl2.
[18] Agarwal,A.B.L., Fredrick,J.R. and Felback,D.K., "Detection of
Plastic Microstrain in Aluminium by Acoustic Emission",
Metallurgical Transactions, VoI.IA, No.4, April 1970, pp. I069-
~071.
[19] Heiple,C.R. and Carpenter,S.H., "Acoustic Emission produced by
Deformation of Metals and Alloys", Journal of Acoustic
Emission, Vol.6, No.3, July-September 1987, pp.177-204.
[20] Shibata,M. and Ono,K., "Stress Concentration due to an Oblate
Spheroidal Inclusion", Materials Science and Engineerin 9,
Voi.34, No.I, January 1978, pp.131-137.
[21] Shibata,M. and Ono,K., "Internal Stresses due to an Oblate
Spheroidal Inclusion: Misfit, Inhomogenity and Plastic
Deformation Effects", Acta Metallurgica~ Voi.26, No.6, June
1978, pp.921-932.
[22] Gladman,T., Holmes,B. and Mcirror,l.D., "Effects of Second
Phase Particles on Toughness and D u c t i l i t y " , in Proceedings
Conference on Effects of Second Phase Particles on the
Mechanical Properties of Steel, The Iron and Steel I n s t i t u t e ,
Scarborough, Cox and Wymon Limited, London, 1971, pp.68-73.
[23] Jayakumar,T., Raj,Baldev, Bhattacharya,D.K., Rodriguez,P. and
Prabhakar,O., "Effect of Heat Treatment on Acoustic Emission
in Nimonic Alloy PE-16 during Tensile Deformation",
Transactions of Indian I n s t i t u t e of Metals, Vol.40, No.2,
April 1987, pp.147-150.
[24] Jayakumar,T., Raj,Baldev, Bhattacharya,D.K., Rodriguez,P. and
Prabhakar,O., "Studies on Acoustic Emission in Nimonic Alloy
PE-16 during Tensile Deformation and Fracture", presented at
Twelfth W o r l d Conference on Non-destructive Testing,
Amsterdam, The Netherlands, 1989.
[25] Scruby,C.B., Wadley,H.N.G., Rushbridge,K.L. and Stockholm-
Jones,D., "Influence of Microstructure on Acoustic Emission
during Deformation of Aluminium Alloys", Metal Science,
Vol.15, No.ll & 12, November-December 1981, pp.599-608.
[26] Rushbridge,K.L., Scruby,C.B. and Wadley,H.N.G., "Origin of
Acoustic Emission in Aged AI-Zn-Mg alloys I: The Base Ternary
Alloy", Materials Science and Engineering, Voi.59, No.2, June
1983, pp.151-168.

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

Yasuhiko Mori and Teruo K i s h i

RELATIONSHIP BETWEEN ACOUSTIC EMISSION AND FLAW SIZE IN Si3N4 CERAMICS

REFERENCE: Mori, Y. and K i s h i , T . , " R e l a t i o n s h i p b e t w e e n
A c o u s t i c Emission and Flaw S i z e in Si3N4 C e r a m i c s , " A c o u s t i c
Em i s s i o n : C u r r e n t P r a c t i c e and F u t u r e D i r e c t i o n s , ASTM STP
1077, W. Sachse, J . Roger and K. Yamaguchi, E d s . , American
S o c i e t y f o r T e s t i n g and M a t e r i a l s , P h i l a d e l p h i a , 1991.

ABSTRACT: A c o u s t i c e m i s s i o n b e h a v i o r was i n v e s t i g a t e d in t h e
Si3N4ceramics i n o r d e r to s t u d y t h e e f f e c t of f l a w s i z e on
fracture strength. A c l o s e r e l a t i o n between t h e g e n e r a t i o n
of a c o u s t i c e m i s s i o n and f l a w s i z e was found. I t was found
t h a t a c o u s t i c e m i s s i o n s i g n a l s can be d e t e c t e d o n l y j u s t
b e f o r e specimen f a i l u r e w i t h f l a w s i z e of 15 z m , ' a c r i t i c a l
f l a w s i z e e s t i m a t e d f o r t h e m a t e r i a l s , or l a r g e r . Whereas
f l a w s s m a l l e r than 15 zm began t o g e n e r a t e e m i s s i o n s a t t h e
s t r e s s l e v e l , a p p r o x i m a t e l y 50 g o f t h e s t r e n g t h o f t h e
specimens f a i l e d . In t h e l a t t e r c a s e , i t was concIuded t h a t
e m i s s i o n s were due t o m i c r o s c o p i c f r a c t u r e e v e n t s a r o u n d
inherent flaws.

KEYWORDS: a c o u s t i c e m i s s i o n , s i l i c o n n i t r i d e , artificial
flaw, flaw s i z e , n o n d e s t r u c t i v e t e s t , f r a c t u r e

Ceramics have been used f o r s t r u c t u r a l m a t e r i a l s b e c a u s e of t h e i r
i n t r i n s i c p r o p e r t i e s , such as high s t r e n g t h or high h e a t r e s i s t i v i t y .
ttowever, t h e b r i t t l e f r a c t u r e occurs e a s i l y in c e r a m i c s , t h e r e f o r e
more work i s b e i n g d e v o t e d t o e n h a n c i n g t h e f r a c t u r e t o u g h n e s s o f
ceramics [1-3]. In t h e f r a c t u r e of c e r a m i c s t h e c r i t i c a l crack size
Is v e r y s m a l l , and i t i s g e n e r a l l y a c c e p t e d t h a t c r a c k s can o f t e n
o r i g i n a t e from m i c r o s c o p i c i n h e r e n t f l a w s and c o a i e s c e [ 4 , 5 ] . it is,
t h e r e f o r e , i m p o r t a n t t o u n d e r s t a n d t h e dynamic m i c r o s c o p i c f r a c t u r e
process in d e t a i l s .

On the o t h e r hand, n o n d e s t r u c t i v e t e s t i n g (NDT) t e c h n i q u e for

Dr. Y. Mori i s a s s o c i a t e p r o f e s s o r a t C o l l e g e o f I n d u s t r i a l Tech-
n o l o g y , Nihon U n i v e r s i t y , 2 - 1 I z u m i - c h o - 1 , N a r a s h i n o , C h i b a 275,
Japan; P r o f e s s o r Dr. T. K i s h l i s a t R e s e a r c h C e n t e r f o r Advanced
S c i e n c e and Technology, The U n i v e r s i t y of Tokyo, 6-1 gomaba-4, Neguro-
ku, Tokyo 153, Japan.

242
Copyright9 by ASTM International www.astm.org
Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

MORI AND KISHI ON FLAW SIZE IN Si3N4 CERAMICS 243

quantitative assessment of ceramics r e l i a b i l i t y is also important.
Some i n n o v a t i v e t e c h n i q u e s have been s t u d i e d and i n v e s t i g a t e d for
t h e i r a p p l i c a b i l i t y t o - c e r a m i c s , such as m i c r o - f o c u s e d x - r a y , s c a n n i n g
a c o u s t i c m i c r o s c o p e (SAM), s c a n n i n g l a s e r a c o u s t i c m i c r o s c o p e (SLA~)
and so on [ 6 , 7 ] . A c o u s t i c e m i s s i o n (AE) i s one of t h e most e x p e d i e n t
techniques both to improve fundamental understanding of fracture
dynamics and t o e v a l u a t e s t r u c t u r a l i n t e g r i t y . M a n y r e s e a r c h e r s have
made r e m a r k a b l e advances [ 5 , 8 - 1 1 ] .

Japan Fine Ceramics Center (JFCC, Nagoya Japan) has promoted R & D
p r o j e c t on t h e d e v e l o p m e n t o f NDT t e c h n i q u e s f o r t h e f i n e c e r a m i c s ,
contracted w i t h Government, s i n c e 1987. The p r e s e n t st u d y was c a r r i e d
out as one of t h e p a r t of th e JFCC's p r o j e c t .

In t h e p r e s e n t p a p e r , a c o u s t i c e m i s s i o n d u r i n g t h e f o u r - p o i n t bend
t e s t both in h o t - p r e s s e d and in p r e s s u r e - l e s s s i n t e r e d Si3N4 specimens
with a r t l f l c l a i d e f e c t s has been a n a l y z e d . The purpose of t h i s st u d y
is t o i n v e s t i g a t e th e i n f l u e n c e of f l a w s i z e both on a c o u s t i c e m i s s i o n
and on s t r e n g t h of t h e m a t e r i a l . This paper a l s o d e s c r i b e s an a t t e m p t
t o s e e k an e x p l a n a t i o n for the generation of acoustic emission
o b s e r v e d in the p r e s e n t m a t e r i a l , by c o m p a r i n g w i t h t h e r e s u l t s
o b t a i n e d by t h e f r a c t u r e surface topography analysis (FRASTA)
technique.

EXPERIMENTAL PROCEDURE

Specimens and F o u r - p o i n t Bend T e s t

H o t - p r e s s e d Si3N4 (tlPSN) and p r e s s u r e - l e s s s l n t e r e d Si3N4 (PLSSN)
were used as t h e t e s t m a t e r i a l s f o r t h i s e x p e r i m e n t b e c a u s e of t h e i r
p o t e n t i a l i m p o r t a n c e as a s t r u c t u r a l ceramics. Material properties
a r e l i s t e d in Table 1.

The specimens f o r t h e f o u r - p o i n t b e n d i n g t e s t were machined from
the d i s k s f a b r i c a t e d by a hot p r e s s or by a s i n t e r l n g , and m a c h i n i n g
damage was removed m e c h a n i c a l l y by p o l i s h i n g t o p r o d u c e an o p t i c a l
finish. The specimen was t h e r e c t a n g u l a r bar w i t h dimensions of 3 by

TABLE 1 - - P r o p e r t i e s of m a t e r i a l s used.

HPSN PLSSN
Mean g r a i n s i z e , am r r
Density 3.35 3.27
S t r e n g t h ( 4 - p o i n t bend t e s t ) , MPa 863 1050
Young's modulus, GPa 306 260
Poisson's ratio 0.26 0.2%
Fracture toughness, ~Pa/~ 5.3 a 6.3 o
Sound v e l o c i t y ; P-wave, km/s 10.3 10.0
S-wave, km/s 6.0 5.5
a by i n d e n t a t i o n m i c r o f r a c t u r e (IM) method [12].
b by IM method and i n d e n t a t i o n s t r e n g t h - i n - b e n d i n g
(ISB] method [13].

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

244 ACOUSTICEMISSION

Straight notch j ~20 size,pro
a) ( ~ 2 o - so

Knoop indentation
b) ( i 2o-4oo

FIG. 1 - - Schematic illustration of artificial flaws introduced
into specimens.

4 b y 40 mm ( F i g . 1).

I n o r d e r t o e x a m i n e t h e e f f e c t o f d e f e c t s b o t h on t h e s t r e n g t h o f
the specimen a n d on a c o u s t i c emission, artificial defects were
introduced into the specimens. Straight-through notches which had a
d e p t h v a r y i n g f r o m 20 t o 50 z m w e r e made a t t h e c e n t e r o f t h e t e s t
p i e c e b y a 20 zm t h i c k d i a m o n d c u t t e r (Fig. la). gnoop i n d e n t a t i o n s
were placed on the bend surface of the specimen using a routine
microhardness tester at load levels o f 0 . 1 t o 150 N, t a k i n g c a r e t o
accurately align the indentation diagonal perpendicular to the
longitudinal axis of the specimens (Fig. lb). The resulting
dimensions of the indentations w e r e 20 t o 400 z m. In genera] for
indentation flaws, residual stresses are induced by the indentation
,required to create the flaw [14,15]. Such r e s i d u a l s t r e s s e s would aid
the applied tensile stress in producing fracture, thus leading to a
l o w e r m e a s u r e d KIc v a l u e . In this experiment, neither annealing nor
removal of material from the surface was treated on the Knoop
specimens to eliminate the residual stresses near the flaw.

Four-point bend tests were performed using a universal testing
m a c h i n e a t a c o n s t a n t c r o s s h e a d s p e e d o f O.1 mm/min a n d / o r 0 . 5 mm/min.
A b e n d t e s t j i g w i t h i n n e r s p a n o f 10 mm a n d o u t e r s p a n o f 30 mm was
used. A l l t e s t i n g was p e r f o r m e d i n a l r a t room t e m p e r a t u r e .

AE M e a s u r e m e n t

Two 400 kHz r e s o n a n t t y p e hE s e n s o r s were e m p l o y e d t o d e t e c t and
locate emission events. Detected signals were amplified 60 dB b y a
preamplifier and analyzed by a commercial AE a n a l y z e r . Total
a m p l i f i e r g a i n o f 90 dB a n d t h r e s h o l d l e v e l o f 100 gY a t s e n s o r o u t p u t
were c h o s e n .

NDT E x a m i n a t i o n

Flaw d e t e c t a b i l i t y of nondestructive testing (NDT) m e t h o d s w h i c h
are expected to be techniques for the inspection of ceramics were also
examined.

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

MORI AND KISHI ON FLAW SIZE IN Si3N4CERAMICS 245

TABLE 2 -- Results of NDT examinations.

Flaws and sizes, ~ m
NDT Methods Notch Knoop
20 50 20 50 i00 400
Fluorescent penetrant 0 0 X X X ...
M i c r o - f o c u s e d X-ray (10 zm) 0 0 X X X ...
U l t r a s o n i c (15 MHz) 0 0 X X X ...
SAM (50 MHz) 0 0 0 0 0 ...
SLAM (100 MHz) 0 0 ... X X X
O: d e t e c t e d , X: not d e t e c t e d

P r i o r t o t h e bend t e s t , a l l s p e c i m e n s were examined by m i c r o -
focused (10 Zm) X-ray technique, ultrasonic inspection
(15 M H z ) , s c a n n i n g a c o u s t i c m i c r o s c o p e (SAM, 50 Mtlz), s c a n n i n g l a s e r
acoustic m i c r o s c o p e (SLAM, 100 MHz), and f l u o r e s c e n t penetrant
inspection. The t e s t r e s u l t s a r e shown in Table 2. Notch f l a w s were
d e t e c t e d by a l l of th e methods employed in t h i s s t u d y , however, gnoop
f l a w s , which were t h e r e s u l t i n g i n d e n t a t i o n c r a c k s , were d e t e c t e d o n l y
by SAM. D e t a i l s of th e e x a m i n a t i o n have been r e p o r t e d e l s e w h e r e [16].

RESULTS AND DISCUSSION

Test R e s u l t s

A l l specimens were l o a d e d in th e f o u r - p o i n t b e n d i n g t e s t u n t i l t h e
s p e c i m e n s were f r a c t u r e d , and t h e a c o u s t i c e m i s s i o n s were m e a s u r e d
throughout the t e s t . F i g u r e 2 shows two examples of b e n d i n g l o a d , P,
a p p l i e d v s . e l a p s e d t i m e , t , c u r v e f o r Knoop s p e c i m e n s . For a l l
specimens t e s t e d , u n s t a b l e f r a c t u r e s o c c u r r e d at a c r i t i c a l applied
s t r e s s l e v e l w i t h o u t an obvious p r e c u r s o r compliance change.

Acoustic emission total e v e n t c o u n t s , N, a r e also plotted in

Knoop / Knoop 20pro -1.0
-O.5
20- Z
20-
Z -~ Z
5,,.~
n 10- Q.
IO-~N
0
b)0 0
0 f, , ~ , I ' ,I, , ' ' ' L I ~ '" ' I . . . . I
0 50 100 50 I O0 150
a) t/s tls

FIG. 2 - - Two examples of t h e f o u r - p o i n t bend t e s t results.
a) Knoop 100 ~m and b) Knoop 20 ~m specimens (PLSSN).

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

246 ACOUSTIC EMISSION

F i g . 2. In t h i s s t u d y , e m i s s i o n e v e n t s l o c a t e d on t h e p a r t o f t h e
specimen w i t h i n i n n e r l o a d i n g p o i n t s of bend t e s t j i g were a n a l y z e d as
e f f e c t i v e a c o u s t i c e m i s s i o n s i g n a l s g e n e r a t e d by f r a c t u r e e v e n t s in
t he specimen. Here th e sound v e l o c i t y of 6000 m/s was used.

AE e v e n t c o u n t s measured were few in a l l specimens, and t h e r e was
no d i f f e r e n c e i n a c o u s t i c e m i s s i o n b e h a v i o r b e t w e e n HPSN and PLSSN
specimens. The AE t e s t r e s u l t s showed t h a t t h e r e were two t y p e s of
e m i s s i o n b e h a v i o r s d u r i n g t h e bend t e s t , i.e., (A) e m i s s i o n s were
o b s e r v e d o n l y when th e specimen f a i l u r e o c c u r r e d , as seen in F i g . 2a,
and (B) t h e o c c u r r e n c e o f e m i s s i o n s began a t l o a d l e v e l s a b o u t or
b e l o w 50 % o f t h e f r a c t u r e strength of t h e s p e c i m e n , F i g . 2b.
Therefore, the stress level at which high emission activity was
firstly o b s e r v e d d u r i n g t h e bend t e s t was e v a l u a t e d . This stress
level is referred hereinafter t o as aAE and a a f o r the bending
s t r e n g t h of each specimen.

R e l a t i o n Between A c o u s t i c Emission And Flaw S i z e

F i g u r e 3 shows t h e r e l a t i o n s h i p s b e t w e e n f l a w s i z e , s t r e s s aAE
(open s y m b o l s ) and a B ( s o l i d symbols) f o r HPSN and PLSSN s p e c i m e n s .
In F i g . 3, t h e f l a w s i z e of a n o t c h or a Knoop i s e x p r e s s e d by t h e
e q u i v a l e n t c r a c k l e n g t h , aeq, in terms of t h r o u g h - t h e - t h i c k n e s s c r a c k
in an i n f i n i t e p l a t e . For a Knoop f l a w , t h e r e s u l t i n g indentation
c r a c k c o u l d not be o b s e r v e d f o r a l l of t h e specimens, though s c a n n i n g
e l e c t r o n m i c r o s c o p e (SEM) o b s e r v a t i o n on t h e f r a c t u r e s u r f a c e was
made, t h e r e f o r e , t h e e q u i v a l e n t c r a c k l e n g t h was deduced by assuming
a s e m i c i r c u l a r f l a w shape w i t h a r a d i u s o f gnoop d i a g o n a l m e a s u r e d .
The d a t a f o r t h e specimens whose f a i l u r e d i d not occur from t h e f l a w
s i t e a r e p l o t t e d i n t h e h a t c h e d r e g i o n a t ] e f t of p l o t . H er e, t h e
s t r e n g t h d a t a f o r gnoop s p e c i m e n s i n v o l v e s the e f f e c t of r e s i d u a l

oo0 I
I o-B " m I
1000 ~ 1 "---.o;, . o,~,,, , o I

100

so ';,IT'

1 5 10 50 100 200
aeqlpm
FIG. 3 - - R e l a t i o n s h i p s between e q u i v a l e n t c r a c k l e n g t h aeq,
s t r e s s a ^ [ a n d a B.

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

MORI AND KISHI ON FLAW SIZE IN Si3N4 CERAMICS 247

indentation stresses.

R e g a r d i n g the f r a c t u r e s t r e n g t h , i t i s concluded from F i g . 3 t h a t
the v a l u e of about 10 zm i s e s t i m a t e d as the minimum f l a w s i z e which
a f f e c t e d the f r a c t u r e s t r e n g t h of the p r e s e n t m a t e r i a l s . In a d d i t i o n ,
t h e s t r e n g t h of s p e c i m e n s h a v i n g a f l a w l a r g e r t h a n 10 zm can be
e x p r e s s e d by a s t r a i g h t l i n e of s l o p e - 1 / 2 , t h a t i s the l i n e a r f i t in
a c c o r d a n c e w i t h the r e l a t i o n f o r s t r e s s i n t e n s i t y f a c t o r of K=a'/~aeq.
The r e s u l t a n t v a l u e of f r a c t u r e toughness KIC = 5.5 ~Pa~/m from t h e f i t
l i n e c o r r e s p o n d s v e r y c l o s e l y with the Kit v a l u e s of 5.3 MPa~ f o r HPSN
and 6.3 ~ P a ~ f o r PLSSN o b t a i n e d u s i n g f r a c t u r e t e s t p r o c e d u r e ( s e e
Table 1).

On the o t h e r hand, the r e l a t i o n s h i p between a c o u s t i c e m i s s i o n and
f l a w s i z e , t h a t i s a p p l i c a b l e both to IIPSN and to PLSSN, i s found as
shown i n F i g . 3. In t h e r e g i o n of aeq s m a l l e r t h a n about 10 zm, t h e
l e v e l of aAE i n c r e a s e s w i t h the i n c r e a s e of aeq, though the c o r r e s p o n d -
i n g f r a c t u r e s t r e n g t h f o r each specimen remains c o n s t a n t . The p I o t s
of a ~ o f the specimens with a flaw l a r g e r than 15 ~m l i e more or l e s s
on a l i n e p a r a l l e l to the g l c c o r r e s p o n d i n g l i n e s drawn on p l o t s of a B.
By u s i n g n o t a t i o n , K~, the r e s u l t i n g v a l u e , gAE = 5 ~ P a ~ , g i v e s a good
c o i n c i d e n c e t o g x c v a l u e s of t h e m a t e r i a l s . In a d d i t i o n , a c r i t i c a l
c r a c k l e n g t h i s e s t i m a t e d by a c o u s t i c e m i s s i o n to be about 15 zm. It
is of i n t e r e s t t h a t t h i s f l a w s i z e o f 15 zm i s c o m p a r a b l e t o t h e
critical f l a w s i z e which a f f e c t e d the fracture strength of the
materials.

M i c r o f r a c t u r e P r o c e s s Leading To A c o u s t i c Emission

F i g u r e 4 shows the examples of the m a c r o s c o p i c f r a c t u r e s u r f a c e s
of PLSSN n o t c h specimens, illustrating three different fracture

FIG. 4 - - Macroscopic f r a c t u r e s u r f a c e s of PLSSN notch specimens,
illustrating three different fracture surface profiles.

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

248 ACOUSTICEMISSION

surface profiles. The c r a c k p r o p a g a t i o n d i r e c t i o n i s from l e f t (notch
r o o t ) t o r i g h t in each p h o t o g r a p h . With i n c r e a s i n g a B, t h a t i s , w i t h
d e c r e a s i n g notch d e p t h , b, a f r a c t u r e s u r f a c e p r o f i l e r e v e a l s a much
roug h er s u r f a c e from a f l a t s u r f a c e . I t i s i n t e r e s t i n g t o compare t h e
f r a c t u r e s u r f a c e p r o f i l e w i t h t h e a c o u s t i c e m i s s i o n b e h a v i o r and t h e
f r a c t u r e s t r e n g t h of t h e specimen. The rough f r a c t u r e s u r f a c e p r o f i l e
such as a) i n F i g . 4 was o b s e r v e d o n l y i n t h e s p e c i m e n s w i t h a f l a w
s m a l l e r than about 10 ym ( F i g . 3 ) , and i t s a c o u s t i c e m i s s i o n b e h a v i o r
f e l l u n d e r t h e C a t e g o r y B, m e n t i o n e d a b o v e . On t h e o t h e r hand, t h e
specimens showed C a t e g o r y h ' s e m i s s i o n b e h a v i o r had a p r o f i l e of c) or
b) on r a r e o c c a s i o n s . I t was e s t i m a t e d t h a t t h e s u r f a c e p r o f i l e of b)
which l a y between p r o f i l e s a) and c) c o r r e s p o n d e d t o t h e change i n t h e
l e v e l of a ~ n e a r an aeqof 15 ~m in F i g . 3.

The f r a c t u r e s u r f a c e s shown i n F i g . 4 and t h e f a c t t h a t the
s p e c i m e n s showed t h e f r a c t u r e s u r f a c e p r o f i l e o f F i g . 4a) b r o k e i n
small pieces at the fracture site suggests that the crack branching
t o o k p l a c e in t h e s p e c i m e n s w i t h h i g h e r s t r e n g t h . In g e n e r a l , t h e
crack branching occurs a f t e r c r a c k grew to a c e r t a i n l e n g t h . There-
f o r e , a c o u s t i c e m i s s i o n d a t a shown in F i g . 3 can be i n t e r p r e t e d as
follows. In the high s t r e n g t h specimens that not did fail from
artificial f l a w s i t e , a c o u s t i c e m i s s i o n began a t s t r e s s e s as low as
100 MPa. This e m i s s i o n is a s s o c i a t e d with the microscopic f r a c t u r e
e v e n t s i n i t i a t e d a t t h e n a t u r a l f l a w s t h a t e x i s t e d on or b e n e a t h t h e
t e n s i l e s u r f a c e of a bend specimen where t h e high t e n s i l e s t r e s s i s
achieved during stressing. On t h e o t h e r hand, i n s p e c i m e n s w i t h
larger artificial f l a w s , b e c a u s e of t h e s t r e s s c o n c e n t r a t i o n a t the
f l a w t i p , t h e c o n t r i b u t i o n s of n a t u r a l f l a w s to t h e e m i s s i o n a r e few
here. This e x p l a n a t i o n w i l l be s u p p o r t e d by t h e f o l l o w i n g d i s c u s s i o n .

In o r d e r t o seek an e x p l a n a t i o n f o r th e a c o u s t i c e m i s s i o n o b s e r v e d
in t h e p r e s e n t m a t e r i a l , f r a c t u r e s u r f a c e t o p o g r a p h y a n a l y s i s (FRASTA)
technique d e v e l o p e d by T. Kobayashi was employed. FRASTA t e c h n i q u e
u n d e r d e v e l o p m e n t a t SRI i s a p r o c e d u r e f o r r e c o n s t r u c t i n g the
f r a c t u r e p r o c e s s by c o m p u t e r m a t c h i n g o f c o n j u g a t e f r a c t u r e s u r f a c e
t o p o g r a p h s [17]. The f a i l u r e p r o c e s s produces i r r e g u l a r i t i e s on t h e
fracture surface, and h e n c e t h e s e q u e n c e o f m a i c r o f r a c t u r i n g is
r e c o r d e d in t h e topography of t h e f r a c t u r e s u r f a c e . The t o p o g r a p h y of
the s u r f a c e s i s e x t r a c t e d by p h o t o g r a p h i n g t h e s u r f a c e s in s t e r e o w i t h
a SEM, and by u s i n g a e r i a l c a r t o g r a p h i c t e c h n i q u e . In p r e p a r a t i o n f o r
r e c o n s t r u c t i n g t h e f r a c t u r e p r o c e s s , u s i n g a computer, t h e t o p o g r a p h i c
map of one f r a c t u r e s u r f a c e i s i n v e r t e d and superimposed on t h e map of
the c o n j u g a t e s u r f a c e . The two maps then a re a d j u s t e d so t h a t i d e n t i -
f i a b l e c o r r e s p o n d i n g p o i n t s on t h e two s u r f a c e s a r e i n a l i g n m e n t and
overlap everywhere. The f r a c t u r e p r o c e s s i s r e c o n s t r u c t e d by i n c r e a s -
i n g t h e d i s t a n c e between t h e c o n j u g a t e maps in s m a l l i n c r e m e n t s . The
r e s u l t s can be d i s p l a y e d as f r a c t u r e d - a r e a p r o j e c t i o n p l o t s .

FRASTA was c u r r i e d out by gobayashi on a specimen of PLSSN used in
t h i s s t u d y [18]. The specimen was a smooth rod specimen w i t h gauge
l e n g t h of r mm. The s p e c i m e n was f r a c t u r e d in t e n s i l e t e s t a t
room t e m p e r a t u r e , and FRASTA was t h e n c a r r i e d o u t on t h e f r a c t u r e
surfaces. In t h e t e n s i l e t e s t , acoustic emission located at the
f r a c t u r e s i t e of t h e specimen began at a s t r e s s l e v e l of 340 MPa, and
the specimen f a i l e d , with i n i t i a t i o n s i t e a t the specimen s u r f a c e , at
a s t r e s s l e v e l of 486 MPa. The mean t e n s i l e s t r e n g t h of the m a t e r i a l

Copyright by ASTM Int'l (all rights reserved); Mon Dec 26 18:40:22 EST 2011
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.

Okada and S i n e s r e p o r t e d Copyright by ASTM Int'l (all rights reserved). No further reproductions authorized.i n f r a c t u r e in the specimen such as t h a t shown i n F i g . expected that the final failure of the specimen occurred from the crack which is approximated with a s e m i c i r c l e w i t h a r a d i u s o f a b o u t 135 zm (aeq = 59 Zm) i n F i g . r e s u l t e d in the s o u r c e s of e m i s s i o n s o b s e r v e d a t an e a r l y s t a g e o f t h e s t r e s s i n g . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. such as t h e g r o w t h and c o a l e s c e n c e of microscopic inherent f l a w s and t h e n u c l e a t i o n of new m i c r o c r a c k s [ F i g . 5 . and w h i t e a r e a s are the f r a c t u r e d a r e a s . . In t h e s e p l o t s t h e dark a r e a s i n d i c a t e c o n t i n u o u s m a t e r i a l .a r e a p r o j e c t i o n p ] o t s o b t a i n e d by s e p a r a t i n g t h e c o n j u g a t e t o p o g r a p h i c maps (FRASTA t e c h n i q u e ) . On f o l l o w i n g sequence in which t h e maps were s e p a r a t e d up to 5 zm. and a l s o shows m i c r o c r a c k s or c a v i t i e s . and the a c c e l e r a t e d specimen f a i l u r e took place. was 516 MPa with shape p a r a m e t e r of Weibull d i s t r i b u t i o n of 12. 6 1 7 Zm s e p a r a t i o n and b) 2. showing t h e f r a c t t ~ r e d .861 zm s e p a r a t i o n { a f t e r Kobayashi [ 1 8 ] ) . 5. which a r e e n c l o s e d w i t h a s e m i c i r c l e w i t h r a d i u s about 50 zm (aeq= 22 Zm) below t h e specimen s u r f a c e . though the c r a c k e x t e n s i o n toward t h e r i g h t i s a l s o found. It is. 5 a ) ] . For t h e m i c r o s c o p i c f r a c t u r e p r o c e s s i n c e r a m i c s . 5a) and 5b) r e v e a l s t h a t d e s p i t e a s e p a r a t i o n i n c r e m e n t o f 1.. The c r a c k p r o p a g a - t i o n d i r e c t i o n i s from top (specimen s u r f a c e ) t o bottom.3 M e a / i (see Table 1). From t h e above d i s c u s s i o n s .a r e a p r o j e c t i o n p l o t s o b t a i n e d by m a t c h i n g and s e p a r a t i n g t h e c o n j u g a t e t o p o g r a p h i c maps. An e x a m i n a t i o n o f F i g s . It i s a l s o e x p e c t e d t h a t t h e m i c r o s c o p i c f r a c t u r e e v e n t s . i t i s c o n c l u d e d t h a t the o c c u r r e n c e of a c o u s t i c e m i s s i o n s g e n e r a t e d b e f o r e the specimen f a i l u r e were due to the p o p . 5 b ) . MORI AND KISHI ON FLAW SIZE IN Si3N4 CERAMICS 249 FIG. 5 b ) ] . F i g u r e 5 shows the r e s u l t of FRASTA. p o s s i b l y formed d u r i n g the s i n t e r i n g p r o c e s s of t h e m a t e r i a l . The r e s u l t i n g v a l u e o f 462 Mea a g r e e s w e l l w i t h t h e m e a s u r e d v a l u e o f 486 MPa. F i g u r e 5a) illustrates t h e specimen s t a t e a t the o n s e t of l o a d i n g . a) 1 . therefore. This v a l u e was used t o compute the f r a c t u r e s t r e n g t h of t h e specimen u s i n g t h e s t r e s s i n t e n s i t y f a c t o r g I c o f 6.F r a c t u r e d .244 ~m t h e e x i s t i n g c r a c k i n c r e a s e d r a p i d l y i n s i z e up t o a b o u t 135 ~m [ F i g . i t was r e v e a l e d t h a t a f t e r t h e c r a c k was a r r e s t e d f o r a w h i l e t h e c r a c k s t a r t e d t o grow.

which was almost t h e same as a c r i t i c a l f l a w s i z e e s t i m a t e d by a c o n v e n t i o n a l m a t e r i a l s testing. The r e l a t i o n s h i p between a c o u s t i c e m i s s i o n and c r a c k l e n g t h o b t a i n e d i n F i g . ACKNOWLEDGMENTS This work i s r e l a t e d to t h a t c a r r i e d out in Japan Fine Ceramics C e n t e r (JFCC) f o r R e s e a r c h on t h e D e v e l o p m en t o f NDT T e c h n i q u e s f o r the Fine Ceramics. c o u l d p r o p a g a t e more r e a d i l y than m a c r o s c o p i c c r a c k s and t h a t c r a c k s can c o a l e s c e n c e in a v e r y s h o r t time [4]. whereas f l aws s m a l l e r than 15 #m began to g e n e r a t e e m i s s i o n s at t h e s t r e s s l e v e l . I t was known t h a t e m i s s i o n s were due t o m i c r o s c o p i c f r a c t u r e e v e n t s around i n h e r e n t f ] a w s . f o r t h e use of h i s FRASTA data. 250 ACOUSTICEMISSION t h a t m i c r o s c o p i c c r a c k s . gobayashi of SRI I n t e r n a t i o n a l . However. The r e s u l t of FRASTA p r e s e n t e d h e r e was o b t a i n e d from SEM f r a c t u r e s u r f a c e p h o to g r a ph s t ak en a t low m a g n i f i c a - tion. which may not be d e t e c t a b l e by n o n d e s t r u c t i v e t e s t i n g . FRASTAw i t h h i g h e r . Copyright by ASTM Int'l (all rights reserved). generated emissions only just before specimen failure. w h i c h m u s t be s u c c e s s f u l l y detected by NDT f o r t h e p r a c t i c a l use o f c e r a m i c s . No further reproductions authorized. 3). 3 shows t h a t a c o u s t i c e m i s s i o n can d e t e c t the n u c l e a t i o n o f c r a c k . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The s u b j e c t of u s i n g t h e a c o u s t i c e m i s s i o n t e c h n i q u e as an NDT f o r c e r a m i c s r e m a i n s q u e s t i o n a b l e as t h e a c o u s t i c emission could not predict the failure o f a s p e c i m e n w i t h a f l a w l a r g e r t h a n 15 z m (Fig. a c r i t i c a l s i z e of t h e f l a w from which a c r a c k i n i t i a t e s and e x t e n d s under l o a d i n g may be e s t i m a t e d . T. by u s i n g a c o u s t i c e m i s s i o n measurement. Thus. SUMMARY A r e I a t l o n s h i p between a c o u s t i c e m i s s i o n and f l a w s i z e . n u c l e a t i o n of m i c r o c r a c k s w i t h a d i a m e t e r of 15 t o 30 zm c o u l d be d e t e c t e d at an e a r l y s t a g e of s t r e s s i n g [5]. Flaws of which s i z e s were l a r g e r than 15 #m. The a u t h o r s would l i k e t o t h a n k Mr. The a u t h o r s e x p r e s s t h e i r i n d e b t e d n e s s to Dr. f o r m e r l y p r o f e s s o r of Nagoya I n s t i t u t e of Technology. was found. . FRASTA t e c h n i q u e gave an objectivity t o t h e e x p l a n a t i o n of a c o u s t i c e m i s s i o n s o u r c e i n t h e materials.l e s s sintered Si3N4. Wakayama e t a l . K e n i c h l r o K i t a d a t e and Makoto Ueda o f JFCC f o r p r o v i d i n g t h e s p e c i m e n s and p e r f o r m i n g NDT e x a m i n a t i o n s on t h e s p e c i m e n s . i t was not p o s s i b l e t o e s t i m a t e such m i c r o s c o p i c f r a c t u r e process. i t i s of i n t e r e s t t h a t a c o u s t i c e m i s s i o n i s s e n s i t i v e to such f r a c t u r e e v e n t which can not be d e t e c t e d by NDT. b e l o w 50 ~ o f t h e f r a c t u r e s t r e n g t h of each s p e c i m e n . s u g g e s t i n g t h a t .m a g n i f i c a t i o n SEM p h o t o g r a p h s may be a b l e t o g i v e i n f o r m a t i o n s about t h e s e problems. from which as f u r t h e r s t r e s s i n g i s a p p l i e d t h e o n s e t o f f i n a l c r a c k growth in th e specimen o c c u r s . a ] s o r e p o r t e d t h a t by q u a n t i t a t i v e a c o u s t i c e m i s s i o n a n a l y s i s . a p p l i c a b l e both to the hot-pressed and t h e p r e s s u r e .

Mori. pp. June 1987. 302-309. J. F. Plenum P r e s s . Vol. 1975. P. [4] T. The Jap. J. P r o c e e d i n g s of the 4th I n t e r - n a t i o n a l Symposium on F r a c t u r e Mechanics of Ceramics. Goebbels and ti. 98. 1986. Tokyo.6. Vo]. Faber. T a l t y . p. Ohta. g. R. No. C l a u s s e n . T. pp.A. G. 18A. [11] M.S. N o n . [5] S. 160. L. J o u r n a l of the American Ceramic S o c i e t y . A. [2] R.61-R-3. 194. P r o g r e s s i n A c o u s t i c Emission I I I . No further reproductions authorized. 1975. e. 1983. B. 1. Vol. 559. N o n . M a r s h a l l . [14] J. M. 255-260. Vol. [16] JFCC Report No. Soc. T. 722. Okada and G. Soc. A n t i s . 895-924. Ando. Soc. 539. 58. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Japan Fine Ceramics C e n t e r . J o u r n a l of the American Ceramic S o c i e t y . p. J o u r n a l of M a t e r i a l S c i e n c e . Vol. 1984. l . g i t a d a t e .d e s t r u c t i v e I n s p e c t i o n . Vo. pp. 719-725. [6] D. e e t r o v i c .. M e t a l l u r g i c a l T r a n s a c t i o n s A. 1985. ~asudevan. R. [13] P. R. 1987. p. [12] G. J o u r n a l of JSNDI.d e s t r u c t i v e I n s p e c t i o n .4. C h a n t l k u l . Soc. R. and A. g i t a d a t e . No. No. 37. 113. 1977. T. h. C h a n t i k u l . B l a c k s b u r g . Soc. U. Vol. Japan Fine Ceramics C e n t e r . Hashimoto and H. Copyright by ASTM Int'l (all rights reserved). . 19. Evans and K. [10] T. F r a c t u r e Mechanics of Ceramics. Vol. V o l . 64. P r o g r e s s i n A c o u s t i c Emission I I . K i s h i . J o u r n a l of the American Ceramic S o c i e t y . S t e l n b r e c h . J o u r n a l of M a t e r i a l s S c i e n c e L e t t e r . Tetelman and A. 1984. J o u r n a l of the American Ceramic S o c i e t y . Munz.61-R-3.d e s t r u c t i v e I n s p e c t i o n . 672. No. Non- d e s t r u c t i v e I n s p e c t i o n . 2A. Lawn and D. B. S i n e s . MORI AND KISHI ON FLAW SIZE IN Si3N4 CERAMICS 251 REFERENCES [1] N. [3] A. K. 327. The Jap. No. [8] A.d e s t r u c t i v e I n s p e c t i o n . 533. 1988. T. R e i t e r . p. J o u r n a l of JSNDI. M a r s h a l l . 10. A n t i s . p. The Jap. P r o c e e d i n g s of N a t i o n a l Conference on AE. Kishl and S. N o n . Matsuda and K. Lawn and D. [7] M. Vol. P a b s t . pp. G. The Jap. 3-4. Kohara. Stceb and R. N o n . I w a s a k l . B. No. Vol. [18] JFCC Report No. N i s h i n o . B. p. H. 1984. T. Ando. Y. No. F e t t . Evans. 66. Shockey. p. Kobayashi and D. 9A. S. p. June 1987. 67. pp. 36. New York. G. f f i s h i . J o u r n a l of the American Ceramic S o c i e t y . p. 1983. The Jap. 1981. [15] T. p. 1941-1948. J o u r n a l of the American Ceramic S o c i e t y . 56. pp. 1981. 92. p. iakayama. November 1987. Matsuda and K. 1982. [17] T. K. 64. [9] K. Knehans and R. 653-660. Jacobson.

1991. Steve H. Expressions have been derived for the energy Dr. CHINA 252 C o p y r i g h t 9 1991 b y ASTM I n t e r n a t i o n a l ~.org Copyright by ASTM Int'l (all rights reserved). ABSTRACT: It has been known for many years that the fracture and/or debonding of Inclusions or second phase particles during deformation can act as a source of acoustic emission.H. Zumlng ~. Roget and K. No further reproductions authorized. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Eds. A number of experimental studies of this phenomena have been reported. Excel lent theoretical investigations of the fracture of an elastic spherical particle in a plastic matrix have been carried out and expressions have been derived for the energy released durlng fracture. Carpenter and Zuming Zhu A COMPARISON OF THE ACOUSTIC EMISSION GENERATED FROM THE FRACTURE AI~3 DECOHESION OF GRAPHITE NODULES WITH THEORETICAL PREDICTIONS REFERENCE: Carpenter. KEYWORDS: acoustic emission. ls a distinguished s c l e n t l s t a t I n s t l t u t e of Metals Research. S. inclusion fracture. Current Practice and Future~Irectlons. The purpose of this paper Is to report on the acoustic emission generated from the fracture and/or decohesion of spherical graphite ncxJules during the compressive deformation of nodular cast Iron. To date there has been a great deal of experimental work reported and some excellent theoretical work as well.. f o r m e r l y a v l s l t l n g s c h o l a r at the U n i v e r s i t y of Denver. ASTM $TP 1077. Carpenter ls a Professor of Physlcs at the U n i v e r s l t y of Denver. Zhu.astm. Denver. American Society for Testing and Materials. "A Comparison of the Acoustic Emission Generated From the Fracture and Decohesion of G r a p h i t e Nodules with Theoretical Predictions. The theoretical work has concentrated on the fracture of an elastic spherical partical in a plastic matrix. The correlations of the experimental acoustic emission data with theoretical predictions will be reported. A recent review [2] has provided a discussion of the current understanding of acoustic emission from the fracture and decoheslon of inclusions and second phase particles." Acoustic Emission. CO 80208. second phase particles. however there Is little work where a direct comparison of theory and experiment can be carried out. and Zhu. However. Dr. Shenyang. W. nc~lular cast iron The fracture and/or debonding of inclusions and second phase particles during deformation have been recognized as possible sources of acoustic emission for many years [I]. Sachse. Yamaguchl. . Philadelphia. J. there is little work where a direct comparison between experimental data and theoretical predictions can be carried out.

Typical a c o u s t i c emission reported in terms of the RMS voltage and the applied stress for the compressive loading of a nodular cast Iron sample are shown as a function of the crosshead displacement in Figure 2. No further reproductions authorized. The RMS data presented in this paper have been corrected for noise and are referenced to the transducer. Data will be presented in this paper on the acoustic emlsslon generated from the fracture and d e c o h e s i o n of g r a p h i t e nodules d u r i n g the compressive deformatlon of ductile cast iron..136 mm to 0. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Care was exercised to eliminate any and all sources of extraneous noise during testing. N. Measurements were made at a gain of 105 dB with a frequency bandpass of 100 . The average nodule diameter varied from 0. Carefully prepared nodular cast iron is a material in which it is possible to obtain uniform spherical particles. Richard Salzbrenner. Samples with five different graphite nodules diameters were supplied and tested. i. The change in average nodule diameter varies over a factor of four. Samples have been carefully prepared with a constant carbon concentration and a uniform dlstributlon of spherical graphite nodules of near the same diameter.5]. with a resonant frequency of approximately 140 kHz. The acoustic emission measurements were carried out using standard commerlcal equipment. The data shown in Figure 2 are for a group 4 sample. The RMS voltage generated by the acoustic emission was m e a s u r e d by a Hewlett-Packard 3400A voltmeter. Compression samples 7mm x 10 mm x 20 mm were deformed at a constant crosshead speed of 0. CARPENTER AND ZHU ON DECOHESION OF GRAPHITE NODULES 253 released during the fracture of the spherical particle. the spherical graphite nodules.M. RESULTS The a c o u s t i c e m i s s i o n g e n e r a t e d during the c o m p r e s s i v e deformation of nodular cast iron has been investigated and reported previously [4. The primary difficulty In evaluating the val Idlty of these e x p r e s s i o n s by experimental data Is obtaining suitable test samples from which to o b t a i n the data. for this investigation. The RMS data are characterized by a peak near yield and a much broader peak well past yield. A series of samples with different diameter nodules have been tested. M e a s u r e m e n t s of the average nodule diameter were performed by Salzbrenner [3].0064 cm/min. F i g u r e I is a m l c r o g r a p h showing typical mlcrostructure for group 5 samples (see Table I below).031mm. EXPERIMENTAL PROCEDURE A series of well characterized ductile cast iron samples having uniform chemistry but different graphite nodule diameters were supplied by Dr. Albuquerque. Correlations between the experimental acoustic emission data and theoretical predictions as a function of average nodule diameter will be presented and discussed. .300 kHz. The transducer used was a resonant piezoelectric (PZT-5) type. The peak at yield is believed due to avalanche motion of Copyright by ASTM Int'l (all rights reserved). The r e q u i r e m e n t for s a m p l e s w i t h well characterlzed spherical second phase particles with small variations in diameter is difficult to satisfy. The measurements reported on in this paper are in general agreement with the earlier results. Sandla National Laboratories.e.

16 0.80 0.~6 2.123 0.088 0.20 0.31 0.. average nodule diameter = 0. The broad RMS peak past yield has been attributed to the fracture and/or decoheslon of the graphite nodules [4.40 3.031 mm.48 1. TABLE I San~le Nodule F~IS I~IS RMS Strain at Yield Stress Group Diameter Peek(1) Pe~k(2) [Pe~(2) Onset Stress Peek(2) N~.03 1. as expected.00 295 525 5 0. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.78 214 450 2 0.20 300 780 Copyright by ASTM Int'l (all rights reserved).031 0. ~m uV uV .136 0.83 0.88 177 460 3 0.04 3.47 1. 254 ACOUSTICEMISSION Figure I. . No further reproductions authorized. The strain at the onset or beginning of the second RMS peak.77 0.078 0.5].76 1. (x 183) dislocations and will not be discussed further in this paper. were found to be dependent on the graphite nodule diameter.46 1.88 199 500 4 0.Typical microstructure of a group 5 sample.Bg]uV % m~a m~ 1 0.62 0. The column RMS [Peak(2) " Bg] is the maximum value of the RMS bursts at the second peak less the value of the background or continuous emission at that point. T a b l e I p r o v i d e s a listing of the a v e r a g e s of experimental data from at least three compressive tests of samples from each sample group. the maximum value of the second RMS peak and the applied stress at the maximum of the second RMS peak.

.Acoustic emission In terms of the RMS voltage and applied stress for the compressive loading of a group 4 nodular cast iron sample. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.0 . . . Scruby [7] and Kant [8] have estimated the energy released from the fracture of an elastic spherical Inclusion in a infinite plastic matrix where the matrix and the Inclusion have different elastic properties. .. CARPENTER AND ZHU ON DECOHESION OF GRAPHITE NODULES 255 0. . . DISCUSSION An examination of the experimental data In Table I indicates that there are definite correlations between a number of the experimental measurements and the average nodule diameter. .4 9 9 | 9 9 | 9 9 | " 9 | 9 9 ! 800 M E o 0. 0 0 20 40 60 80 100 12u T e s t t i m e (rains) Figure 2. u is the shear modulus. D is the distance the crack Is located below the surface and A is the area s w e p t out by the crack. . . Only the the RMS Peak(1) and the yield stress fail to show a distinct correlation with the average nodule diameter. .. < 0. cr is the applied stress. . . .2" 400 0 "0 > Q) 0. Clearly if the particle cracks Copyright by ASTM Int'l (all rights reserved). Earlier work has shown that the yield properties are not dependent upon the nodule size In ductile cast iron [6]. . .1" 200 D. Both found that the energy released should be given approximately by E = (4/3)(I .3" 600 v 0. . No further reproductions authorized. The other experimental data show an ordering in magnitude in accordance with the average nodule diameter.v)(o-2r3/E) (I) n where v is Polsson's ratio. Gerberich and Jatavallabhula [9] have d e r i v e d the f o l l o w i n g expression for the RMS voltage generated during the fracture of a spherical particle of radius r: VRM S = [2K"(f)/uD][~Ar] (2) where K"(f) is a function of frequency. . n. r is the radius of the crack and E is the Young's modulus of the inclusion.

It should then be possible to also use equation (I) to predict the value of RMS voltage generated.6 corr = 0.e. 9 . .. 1.- ~ 1.2 (Applied stress)(Average radius cubed) Figure 4. No further reproductions authorized.o ~ 0. Check on equation (2). Check on equation (I). .2 20 40 60 80 (App|iedstresssquared) x (average radius cubed) Figure 3. . Figure 3 shows a plot of the maximum value of the second RMS peak versus the applied stress squared times the average radius cubed. . 9 . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Copyright by ASTM Int'l (all rights reserved).4 ~ c o r r = 0 ~ O 9 0. In both cases good straight lines are obtained. . Equation (I) l~e~icts that the RMS voltage generated will be ~roporr to [o.1 0. Recently published work [10] has shown that the RMS voltage generated by an acoustic emission event is proportional to the energy in the acoustic emission waveform.2 U. Hence.Maximum value of the second RMS peak versus the applled stress times the average radius cubed.e. i.r ]. .0 0. equation (2). .0 1. both treatments predict an [r ] dependence but disagree on the stress dependence. 256 ACOUSTICEMISSION completely across Its diameter then equation (2) predicts that the RMS voltage generated during fracture of the spherical particle will be proportional to [o-r ]. equation (I). Assuming that the values of RMS voltage and applied stress at the second peak maximum are the best choices to correlate with the average nodule diameters it is possible to evaluate the validity of equations (I) and (2).943 iz 0. i.4 ~ 1. .943 i i i 0. Figure 4 shows a similar plot of the maximum value of the second RMS peak versus the applied stress times the average radius cubed.Maximum value of the second RMS peak versus the applied stress squared times the average radius cubed..6 .

0 x 0. .r ] and [o-r ] should glve a straight line fit for equations (I) and (2) respectively. . then the height of an RMS burst less the background.-. Check on equation (I). similar reasoning can be used to analyze the nodular cast Iron data. 0. . CARPENTER AND ZHU ON DECOHESION OF GRAPHITE NODULES 257 A paper [10] contained in this STP reports on the a c o u s t l c emission from the fracture of boron particles in a 2219 aluminum matrix.4 ~ 1. 1. It is possible to estimate the validity of equations (I) and (2) by using the values of RMS(2)[Peak .6 t-.RMS(2)[Peak . l. No further reproductions authorized. > 9 .. The authors of that investigation report that there were many more small bursts of e m i s s i o n than could be accounted for by the number of particle fractures.Bg]. i i i n. Copyright by ASTM Int'l (all rights reserved). . As can be seen. Clearly the [r ] dependence dominates. With this background information. 9 . Plotting RMS(2)[Peak . The results are shown in Figures 5 and 6. In that Investigation a small countable number of boron particles were used. They postulate that the additional smaller bursts of acoustic emission were caused by the partial decoheslon of the boron particles from the aluminum matrix.e. 9 .Bg] versus applied stress squared times the average radius cubed. the value of stress at the second maximum and ~h~ average nodule size.2 20 40 60 80 (Applied stress squared) x (average radius cubed) Figure 5. the individual RMS voltage bursts rising from the background in Figure 2 are due to the fracture of the graphite nodules and that the background emission Is most likely due to the decohesion of the graphite nodules from the ferrite matrix. a3good straight line fit is obtained in both cases. 9 .Bg] against [o. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. If this explanation of the data is correct.. should be the appropriate measure of the RMS voltage (or energy since the two are proportional) generated from fracture of a graphite nodule.

Their result is the fol lowing: o.. However. If one assumes.Bg] versus applied stress times the average radius cubed. Copyright by ASTM Int'l (all rights reserved). 7~at there is little change in q(e) then o. Gerberich and Jatavallabhula [7] have also derived an expression for the tensile stress applled to the matrix which is necessary to fracture a spherical Inclusion of radius r. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Check on equation (2).17 which Is between the two predicted values.8 o 9 9 x 0. ~ c is the specific work of fracture and E is Young's modulus for the particle. Using this result in equations (I) and (2) leads to a prediction that the RMS voltage generated during the ~racture of a spherical p a r t i c l e should b~51~_roportlonal to [r ] for equation (I) and proportional to [r " ] for equation (2).1 0. Log-log plots of the maximum value of the second RMS peak and RMS(2)[Peak . .ses slowly with strain. No further reproductions authorized.0 0.2 "O == 0. Both plots show good straight I Ines. 258 ACOUSTIC EMISSION 1.4 "O t-. = q(e)-l[6E*~'c/2r]I/2 (3) where q(e) is a compllcated strain concentration factor which increa. 2 L rt" 0.Bg] against the average nodule radius are shown in Figures 7 and 8.RMS(2)[Peak .is proportional to [r -I/ ]. the data for the RMS second maxima give a slope of I while the data for RMS(2)[Peak .Bg] gives a slope of 2.2 (Applied stress) x (average radius cubed) Figure 6.0 u.

. .. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.Log-log plot of the maximum value of the second RMS peak versus the average nodule radius. No further reproductions authorized. CARPENTER AND ZHU ON DECOHESION OF GRAPHITE NODULES 259 10 oq (z .01 Average nodule radius (mm) Figure 7. Copyright by ASTM Int'l (all rights reserved).Bg] versus the average nodule radius. CONCLUS IONS The acoustic emission generated from the fracture and/or decoheslon of spherical graphite nodules during c o m p r e s s i v e deformation has been analyzed In terms of theoretical predictions. Using the predicted dependence of the applled stress on the average nodule dlameter. o m ~ . it was shown that the value of the RMS voltage bursts less the background voltage agrees very well with the theoretlcal predictions for the fracture of an elastic particle In a plastlc matrix.01 Average nodule radius (mm) Figure 8.01 er . the average nodule diameter and the RMS voltage produced by the fracture and/or decohesion of spherical graphite nodules has been investigated.Log-log plot of RMS(2)[Peak . The dependence of the applied stress. The nodule diameter was found to be the predominant factor in determining the resulting acoustic emission.

Berkeley.W.Wolf. Proceedings of the Second Acoustlr Emission Symposium. Albuquerque. 1976.J. ed. Warrendale. S..K.C. New Mexico 87185. 1987..M. A. E93 Gerberlch. C. D. [5] Egle. [2] Helple.Buck and S. pp $184-$187) also in this volume ASTM STP 1077. July.S. Vol. R. REFERENCES Eli Ono.J. No further reproductions authorized.6. Portions of this work were supported by the U. K. pp 1-63. Sandla Report. and Jatavallabhula. C. Sandia National Laboratories. .. AERE-R 11262. Vol. Charlotte. pp 1037-1044. pp 177-204 and 215-237. Ouantltative Acou~tlr Eml~lon Techniques. AEWG. K. Rocky flats Plant. C. 260 ACOUSTICEMISSION ACK NOWL EDGEME NTS The authors appreciate helpful dlscusslons wlth Dr. Sandia National Labs. Japan.. Sec. N.. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.R... [3] Salzbrenner. 1987. Tatro.Heiple of Rockwell International Corp.. University of California. Thanks are also extended to Dr. Richard Salzbrenner. 1986.M.E. R.R. and Brown. DE-FGO2-85ER45182. W. Nondestructive Evaluation: Mlcrostructqral Characterization and Reliabilitv Strategies. R. Materlals Evaluatlon. C. Oxfordshire. 1981...H. S. Tokyo.B. SAK086 . 1979... 1974. This support Is gratefully acknowledged. S. PA.. Tokyo.A. pp 288-304. Japan Industrial Planning Association. S. 22. (extended abstract in Journal of Acoustic Emission. Proceedlnos of the Third Acoustic Emls$ion__SMmposl_um . for the sample materials used in this Investlgation. Vol. Copyright by ASTM Int'l (all rights reserved).H. [6] Salbrenner. World Meeting on Acoustic Emission. 4. [7] Scruby. Department of Energy through Grant No.0470.S. TMS-AIME.. Journal of Ar Emisslon. 1984 E8] Kant. U.P. Vol. AERE Harwell. pp 2135-2147. Japan.. O. and Carpenter. Doctoral Thesis.. 8. 1989. Carpenter. and Chrlstiansen.H. The Elastostatic Axisymmertic Problem of a Cracked Sphere Embedded in a Dissimilar Matrix. F. Japan Industrial Planning Association. CA.R. and Higglns. pp 319-348. E1o] Heiple. E4] Carpenter. C. 39. Journal of Materlal Science.

t t i r o k a z u Kobayashi. c r a c k detection. Ltd. Yamaguchi. H. The crack growth rate of the gears can be estimated using the relation between the acoustic emission event counts and the intergranular fracture area on the crack surface. C e n t r a l E n g i n e e r i n g L a b o r a t o r i e s . ABSTRACT: Acoustic emission measurements of carburized steel were carried out in room-temperature fatigue tests at the constant load-amplitude condition. T a k u r o Yamaguchi a n d Kimihiro S h i b a t a EVALUATION OF FATIGUE CRACK GROWTH RATE OF CARBURIZED GEAR BY ACOUSTIC EMISSION TECHNIQUE REFERENCE: 0 b a t a . S h l b a t a a r e r e s e a r c h e r a n d s e n i o r r e s e a r c h e r . r e s p e c t i v e l y . Kobayashl and Dr. Yamaguchi. It can be concluded that the acoustic emission technique was power- ful and quantitative method to evaluate the fatigue crack behavior of carburlzed mechanical parts. J a p a n . fatigue c r a c k g r o w t h .org Copyright by ASTM Int'l (all rights reserved). Aoki. and S h i b a t a . C o l l e g e of I n d u s t r i a l T e c h n o l o g y . Kanagawa 237. acoustic emission. Roger and K.. Kobayashi. i n t e r g r a n u l a r fracture A carburized treatment is one of the surface hardening process of the steel. Aoki are research assistant. Y. Yoshlhiko 0 b a t a . However.. g r a d u a t e s t u d e n t and p r o f e s s o r . K e n . Eds. Chlba 275.i c h i r o Aoki. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 1991. K. Yama- g u e h i a n d Dr..astm. " E v a l u a t i o n o f F a t i g u e C r a c k Growth Rate of C a r b u r i z e d Gear by A c o u s t i c Emission Technique". K.. ASTM STP 1077. N a r a s h i n o . Mr. . KEYWORDS: earburized steel spur gear. W.. Yokohama. r e s p e c - t i v e l y . Hr. 261 C o p y r i g h t 9 1991 b y ASTM I n t e r n a t i o n a l ~. Sachse. and has been applied to power driving mechanical parts such as shafts and gears. Nihon U n i v e r s i t y . P h i l a d e l p h i a . N i s s a n Motor Co. A c o u s t i c Emission: C u r r e n t P r a c t i c e and F u t u r e D i r e c t i o n s . T. where high fatigue strength and wear properties are required. Many emissions were also recorded in the spur gear tests. Japan. J. 0 b a t a . American S o c i e t y f o r T e s t i n g and M a t e r i a l s . in order to evaluate the crack growth behavior.. the fatigue crack growth behavior Dr. No further reproductions authorized. An extremely large number of acoustic emission events were recorded near the maximum load in a flexural fatigue cycle.. The increasing behavior of acoustic emission event counts was associated with the distribution of fractured intergranular facets in the fracture surface..

No further reproductions authorized. A few r e p o r t s on t h e f a t i g u e c r a c k i n i t i a - t i o n l i f e and p r o p a g a t i o n r a t e in c a r b u r i z e d l a y e r s h a v e b e e n p r e - s e n t e d n o t o n l y f o r t h e g e a r specimen b u t a l s o f o r s m a l l .05 I.f a t i g u e s p e c i m e n s o f c a r b u r i z e d s t e e l and t h e s p u r g e a r s in o r d e r to d i s c u s s a p p l i c a b i l i t y of the acoustic emission technique for the evaluation of the fatigue crack growth rate. On t h e o t h e r hand. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. .s i z e d s p e c i - men [1-2]..7 ram. Copyright by ASTM Int'l (all rights reserved). F o u r t y p e s o f f l e x u r a l s p e c i m e n s w e r e p r e p a r e d by t h e h e a t t r e a t m e n t and t h e s h o t . 1 .p e e n i n g . b) s p u r g e a r FIG. 262 ACOUSTICEMISSION in c a r b u r l z e d s t e e l is n o t w e l l u n d e r s t o o d p a r t l y due t o l a c k o f c r a c k d e t e c t i o n methods. EXPERIMENTAL METHOD M a t e r i a l s and Specimen Materials used are Ni-Cr-Mo s t e e l ( J I S SNCM420tt) f o r f l e x u r a l s p e c i m e n and Cr s t e e l ( J I S SCr420H) f o r s p u r g e a r . I t s e f f e c t i v e c a s e d e p t h was a b o u t 0. 1. But few r e p o r t s f o r a c o u s t i c emission(AE) in c a r b u r i z e d s t e e l h a v e b e e n p r e s e n t e d [3].p e e n i n g p r o c e s s as shown in T a b l e 1. The p r e s e n t p a p e r d e s c r i b e s t h e a c o u s t i c e m i s s i o n b e h a v i o r in n o t c h e d f l e x u r a l .Specimen s h a p e and s i z e . t h e s p u r g e a r was p r e p a r e d using o n l y one c a r b u r l z e d c o n d i t i o n without s h o t . 7O R1 14] 56 a) n o t c h e d f l e x u r a l specimen 13 1. The s h a p e and t h e s i z e f o r b o t h s p e c i m e n s a r e shown in Fig.

Therefore. recorded near the maximum l o a d r a n g e i n a f a t i g u e c y c l e ( p e a k .7mm) a n d s h o t .V a r i e t y a n d i t s s y m b o l f o r f l e x u r a l specimen.p e e n i n g HOLDER~00DLDINGI LOADCELL " ~ T O O T H . a s s h o w n i n Fig. Two AE s e n s o r s w e r e m o u n t e d t o t h e b o t h e n d s o f t h e b e n d i n g s p e c i m e n a s s h o w n i n F i g . 2 -. BN : non-carburized (normalized heat treatment only) BS : carburized ( e f f e c t i v e c a s e d e p t h i s 0. a n d t h e c r a c k growth length from the notch root measured by mobile optical micro- scope as a function of the number of fatigue cycles.Schematic diagram of spur gear test machine. Schematic diagram for the gear test machine i s s h o w n i n F i g . t h e i n t e r g r a n u l a r fracture was not observed in the non-carburlzed specimens. 4. . T h i s i s a s m a l l (3 mm d i a m e t e r . B o t h n o t c h e d f l e x u r a l and gear specimens were loaded in air at room temperature with a constant amplitude as a f u n c t i o n o f s i n e wave. 2. OBATAET AL. and is used non-reproducible such as the strain gage. 2.7mm) BP : carburized ( e f f e c t i v e c a s e d e p t h i s 0. It can be suggested that non-carburized steel gives us few emissions even for high-strength m a t e r i a l s s u c h a s BN s p e c i m e n o f which tensile strength i s a b o u t 1000 MPa. Fatigue and Acoustic Emission Test A closed-loop electro-hydraulic test machine was used in the fatigue test.l o a d AE). I n t h e g e a r t e s t . T h e y a r e 8 mm d i a m e t e r a n d 15 mm h e i g h t s e n s o r w i t h a r e s o n a n t f r e q u e n c y o f 200 kHz. Copyright by ASTM Int'l (all rights reserved). R e c o r d e d s i g n a l p a r a m e - ters each event are the number of fatigue cycle. Good correlation b e t w e e n e v e n t c o u n t s a n d c r a c k l e n g t h s was o b t a i n e d f o r e a c h s p e c i - men. the difference of arrival time (delta-t) and the event ampli- t u d e . a large number of emissions for carburized specimens caused by the intergran- ular fracture. No further reproductions authorized. the load level in a cycle. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. RESULT FOR FLEXURAL SPECIMENS Effect of Carburizing Treatment on Emission Activity Figure 3 shows the cumulative event counts.5mm) BD : carburized ( e f f e c t i v e c a s e d e p t h i s 0. On t h e o t h e r h a n d . T h e s t r e s s r a t i o was 0.1~ T e s t f r e q u e n c y was 4 Hz i n m o s t t e s t s . ~ ~ P U S H E R I ACTUATOR FIG. T h e e m i s s i o n a c t i v i t y of carburized specimens i s 100 t i m e s larger than that of non-earburlzed specimens.. Intergranular fracture mode was o b s e r v e d i n t h e r e g i o n o f e f f e c t i v e case depth at the notch root and at the specimen side in each carburized specimen. The delta-t was n o t r e c o r d e d in the gear test. ON CARBURIZEDGEAR 263 TABLE 1 . a 150 kHz r e s o n a n t s e n s o r was m o u n t e d a t t h e place a s s h o w n i n F i g . 3 mm height) caseless sensor. 1.

5 f o r b o t h s i d e s o f t h e specimen. In t h i s specimen. 6. 264 ACOUSTICEMISSION a) n o n .F r a c t o g r a p h o f intergranular fracture at notch root of BD c a r b u r i z e d specimen E v a l u a t i o n o f F a t i g u e C r a c k Growth R a t e The A E e v e n t r a t e and c r a c k g r o w t h l e n g t h o b t a i n e d b y t h e m i c r o - s c o p e a r e shown in Fig. 1 H z .c a r b u r i z e d ( B N ) specimen b) c a r b u r i z e d ( B D ) s p e c i m e n FIG. However emission d a t a r e p r e s e n t more h o m o g e n e o u s c r a c k g r o w t h b e h a v - ior than t h a t of the microscope. 3 . and t h e n g r a d u a l l y i n c r e a s e d w i t h increasing crack growth rate.D i f f e r e n c e o f AE c u m u l a t i v e e v e n t c o u n t s b e t w e e n n o n . . The h i g h e m i s s i o n r a t e was o b s e r v e d up t o 15000 c y c l e s e v e n t h o u g h c r a c k g r o w t h r a t e was n o t h i g h w h i l e a c r a c k p r o p a g a t e d in a c a r b u r i z e d l a y e r . This I n t e r e s t i n g AE c h a r a c t e r i s t i c s a r e a s s o c i a t e d with t h e d i s t r i b u t i o n o f t h e i n t e r g r a n u l a r f r a c t u r e on t h e c r a c k e d s u r f a c e . f i r s t t h e e m i s s i o n r a t e b e c a m e low. FIG. T h e r e f o r e two r e l a t i o n s must be u s e d in t h e c r a c k g r o w t h l e n g t h e s t i m a t i o n b y t h e e m i s s i o n e v e n t c o u n t . It can be c o n c l u d e d t h a t emission data faithfully represent the crack growth behavior through the thickness.. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Copyright by ASTM Int'l (all rights reserved).c a r b u r i z e d specimen and c a r b u r i z e d specimen. When a c r a c k p a s s e d t h r o u g h t h e c a r b u r i z e d l a y e r in f r o n t o f t h e n o t c h r o o t .. This exam- p l e o f t h e BS specimen was r e c o r d e d a t 0 . No further reproductions authorized. 4 . The c r a c k g r o w t h r a t e o b t a i n e d b y t h e o p t i c a l m i c r o s c o p e and AE method is shown in Fig. Both d a t a r e p r e s e n t t h e same b e h a v i o r . u n h o m o g e n i z e d c r a c k g r o w t h b e h a v i o r t h r o u g h t h e t h i c k n e s s was o b s e r v e d as shown in t h e m i c r o s c o p e d a t a . The b o u n d a r y b e t w e e n two r e l a t i o n s was a s s o c i a t e d w i t h t h e e f f e c t i v e c a r b u r i z e d d e p t h .

.b . . Copyright by ASTM Int'l (all rights reserved). AE e v e n t r a t e a n d c r a c k growth length behavior as a function of number of fatigue cycle. -6 BS3 tO . ON CARBURIZED GEAR 265 BS3(O. and increases with increasing fatigue cycles. RESULTS FOR SPUR GEARS F a t i g u e a n d AE B e h a v i o r Figure 7 shows a relation b e t w e e n t h e maximum s t r e s s a n d t h e number of failure f a t i g u e c y c l e s . . . OBATA ET AL. the opening a n d c l o s u r e AE s h o w s q u i t e low a c t i v i t y except a specimen tested a t 10 ttz. T h a t is. T h i s AE p a t t e r n is ob- served in all four gears tested. An e s t i m a t e d f a t i g u e strength at 107 c y c l e s i s 650 MPa. . 6 . A v e r y i n t e r e s t i n g result was obtained in the case of fractured i n b o t h t e e t h a t 2 Hz a s s h o w n i n F i g . .C o m p a r i s o n of crack growth rate with AE m e t h o d and optical microscope observation. The peak- l o a d AE i n i t i a t e s at about a half stage of the fatigue life. 5 . Two t e e t h w e r e s i m u l t a n e o u s l y fractured in a g e a r w h i c h was t e s t e d a t 2 Hz.. 8 . i I i i i i I i i i i ~ I i t i i I i //~ [ ca ~ 2E E o \ \ 4 z 14 \ 2 z 0 o 5 10 15 20 25 N / 103 c y c l e s FIG. o n e t o o t h was fractured in the other three gears. . lHz) I0 o 8 J 09 sisiddeeAB J . No further reproductions authorized. Figure 8 shows load level distributions o f AE a s a f u n c t i o n of t h e n u m b e r o f f a t i g u e c y c l e s . ~ ~bo ~10 -? -**~0 \ \ 0 optical microscope 9 acoustic emission 10"8 = I = I I W I I I i I I I = 0 l 2 3 Aa / mm FIG. i . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. . On t h e o t h e r h a n d . . On t h e o t h e r hand. . AE d u e to crack propagation is observed at two load levels as at the peak-load and in the unloading range.. We c a n c l e a r l y discriminate the peak- l o a d AE d u e t o c r a c k p r o p a g a t i o n a n d AE a t t h e l o w e r l o a d l e v e l due to friction of crack surfaces and mechanical noises. I .

................-....................... Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement........ 8 ... 7 .. ... FIG.... X 600 ....... 266 A C O U S T I C EMISSION 1200 I ~1100 ~ 1000 m m 9OC 800 7O0 :iiiiii:-iii!ii! iiiiLLiii:i:::::: ...................R e l a t i o n between maximum s t r e s s and number of failure fatigue cycle in spur gear..... i . No further reproductions authorized. i ... 500 Illlll I I I IIitll I I II IIill I 1 111111 104 105 108 107 Nf / cyc I es FIG. Copyright by ASTM Int'l (all rights reserved)...AE l o a d l e v e l distribution at different test frequencies in spur gear tests.

l .l o a d AE e v e n t rate behavior in spur gear tests.. I o 105 15 2O 0 5 10 15 20 25 N / 103cycles N / ]03 c y c l e s c) 4 Hz d) 10 Hz FIG. * ' i ' * i L I i i i i I ~ i ~ h 2. . . . . 10. A t t h e S t a g e I. T h a t is.0 \ z 9 z2 \ \0. . t h e crack propagates the region where the carburized layer exists at only side surfaces. . OBATA ET AL. No further reproductions authorized.. . o !i o \ \ 41 5\ z 2 0 0 . __ . . t h e c r a c k p r o p a g a t e s in the carburized layer through the tooth width.. . ~ l ~ . . . t h e c r a c k l e n g t h c a n b e e s t i m a t e d u s i n g two l i n e a r r e l a t i o n s . . .P e a k .. [ . Therefore it is considered t h a t AE s i g n a l s o f t h e s e g e a r t e s t s w e r e e m i t t e d d u e t o the intergranular fracture.. l ~ . T h e AE b e h a v i o r o b t a i n e d f r o m c a r b u r i z e d gears is almost the same as that of the flexural specimen test. ON CARBURIZED GEAR 267 G7 (1Hz) G6 ( 2 H z ) 6 i i L . 15 x \2! x 0 o ~'b l .0 . .. . . i [ . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. . Copyright by ASTM Int'l (all rights reserved). 9 . ~ .5 z z o ~ J 0 ' ' ' ' 7 ' ' ' ' ' ' ' ' 1 . j. o <1. The increasing p a t t e r n of AE did not depend on t e s t frequency and also the absolute value of the r a t e did not change a t d i f f e r e n t frequencies from 1 H z to 4 Hz. 10 . Figure 9 shows the r a t e of the peak-load AE as a function of the number of fatigue cycles. 5 i b ~ i L 113 I i h ~ i N / 10 3 cycles N / 103 c y c l e s a) 1 Hz b) 2 Hz G5 (4Hz) G4 ( 1 0Hz) 6 . ~ h . . ~ l .5 ~4 i g \1.. A t t h e s t a g e II. o n t h e o t h e r h a n d . The boundary be- tween two stages is estimated as the effective case depth at the tooth root. . g i v i n g u s f e w e r AE e v e n t s . Estimation of Fatigue Crack Growth Rate There is no doubt that it is easy to evaluate the fatigue crack initiation life from the onset of the peak-load AE a s s h o w n i n F i g . The schematic diagram of the relation between the distribution of the intergranular fractures on crack surface and the cumulative event counts of the peak-load AE i s s h o w n i n Fig. . j i 5 ~ . giving us an extremely l a r g e n u m b e r o f AE e v e n t s . 9.

1~ 71 67 16 7 6 1 7 6 1 7 6 1 7 6 1 7 6 1 7 6 1 7 6 1 7 6 .:. The final crack length at the unstable fracture of the tooth has not yet been measured. .~176176176176176176176176176176176176176176176176176 .:.6.:.:.:. a = K/W x ZNhE (2) where W is width of the tooth and K is a constant. ~ 1 7 6 1 7 6 1 7 6 1 7 6 1. It can be considered that the hE technique is very useful to evaluate fatigue crack growth behavior for carburized gears.-4 o o i!iiiill. Therefore we t r i e d t o e s t i m a t e t h e c r a c k l e n g t h u s i n g t h e f o l l o w i n g a s s u m p - tions and conditions. 7.l o a d h E c u m u l a t i v e e v e n t c o u n t ( A N ~ ) i s p r o p o r t i o n a l to the intergranular fractured area(A~G) r e p r e s e n t e d b y Eq.:.:.:.:. That is. Copyright by ASTM Int'l (all rights reserved). Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 1) c r a c k i n i t i a t i o n is onset of the peak-load hE 2) u n s t a b l e f r a c t u r e o c c u r s a t 0. 268 ACOUSTIC EMISSION ~176176176176176176176176176176176176176176176 .:. e f f e c t i v e .:.:. F i g u r e 11 s h o w s t h e f a t i g u e c r a c k growth rate of the carburized gears obtained by the peak-load AE event counts.:. ..:.. 10 . However the final crack length was e s t i m a t e d a s a b o u t t h e e f f e c t i v e case depth at the tooth root.~ ~176176176176176 I H Z / . C o n s t a n t K/W values u s e d a r e 1 . because o f AE r a t e behavior. No further reproductions authorized.:. ' ~ 1 7 6 1 7.v.S c h e m a t i c d i a g r a m b e t w e e n d i s t r i b u t i o n of intergranular o n c r a c k s u r f a c e a n d AE c u m u l a t i v e e v e n t c o u n t s ..:. .~/depth case ~a FIG.:. Transgranular .(2).:.7 mm c r a c k l e n g t h a t s t a g e I 3) t h e p e a k . crack growth length(Aa) is obtained b y Eq. 0 x l 0 -4 r a m / c o u n t t e s t e d a t 1 Hz t o 4 Hz a n d 2.>:.6x10 4 m m / c o u n t t e s t e d a t 10 Hz. there is no evidence of decreasing b e h a v i o r o f AE r a t e u p t o t h e t o o t h f a i l u r e .(1) Am = K x EN~ (1) Thus.6~1 7 ~ 6 .:...

z - Z E ~ . I t is suggested that a simple and t r a d i t i o n a l m e a s u r e m e n t with o n e s e n s o r c o u l d be ap- plied to evaluate the crack propagation behavior for carburized g e a r f a t i g u e t e s t s . However r e l a t i v e l y h i g h t h r e s h o l d must be s e l e c t e d in the noisy test at high test frequency. F o r example. T h e r e is n o t a b i g d i f f e r - e n c e b e t w e e n o v e r a l l AE and t h e p e a k .Test frequency dependence on AE cumulative event counts in spur gear tests. . 0 overall ~E I05 9 peak-load AE o o 0 \ 104 2: 10 3 I t I I0 f / Hz FIG.2 0.. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.4 0. ON CARBURIZED GEAR 269 G e a r (G= 1 0 7 8 M P a ) 10"6 O ~i0-7 . although relatively f ew emission are expected. No further reproductions authorized. 9-e. 13 e x h i b i t s t h e same as t h e p e a k .8 Aa / mm FIG.l o a d AE up t o 4 Hz. Copyright by ASTM Int'l (all rights reserved).6 0. D 4Hz - 1 0 -8 (]-" 0.F a t i g u e crack growth rate of spur gear teeth estimated by AE method P o s s i b i l i t y o f 1 . l l .c h a n n e l AE Measurement t o E v a l u a t e F a t i g u e C r a c k An a u t o m a t i c a n d s i m p l e m e a s u r e m e n t i s r e q u i r e d i n t h e AE a p p l i c a t i o n s u c h as c r a c k m o n i t o r i n g o f p r a c t i c a l m e c h a n i c a l e l e - ments. OBATA ET AL. 12 -. F i g u r e 12 shows t e s t f r e q u e n c y d e p e n d e n c e on c u m u l a t i v e AE e v e n t c o u n t s a t d i f f e r e n t l o a d l e v e l s .l o a d AE b e h a v i o r as shown in Fig. t h e o v e r a l l AE b e h a v i o r as shown in Fig. Rela- t i v e l y many e v e n t s o f t h e o v e r a l l AE a t 2 Hz r e s u l t e d f r o m t h e fact that cracks propagated at both teeth.

T.. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.l o a d r a n g e in a f a t i g u e c y c l e . Kawabe. K. Our main r e s u l t s a r e as f o l l o w s : 1. S h i b a t a . J o u r n a l o f The I r o n and S t e e l I n s t i t u t e o f J a p a n . Konuma. v o l . Y. No further reproductions authorized. . A c a r b u r i z e d h e a t t r e a t m e n t f o r s t e e l g i v e s us an e x t r e m e l y l a r g e number o f AE e v e n t s d u r i n g t h e f a t i g u e c r a c k g r o w t h . P r o g r e s s in A c o u s t i c Emission IV. " E f f e c t o f C h e m i c a l C o m p o s i t i o n and Heat T r e a t m e n t on M e c h a n i c a l P r o p e r t i e s o f C a r b u r - i z e d S t e e l s " . 1982. CONCLUSION In o r d e r t o e v a l u a t e t h e f a t i g u e c r a c k g r o w t h r a t e o f c a r - b u r i z e d s p u r g e a r s in p u l s a t o r g e a r t e s t s . K. REFERENCES [1] F u r u k a w a . Nishi. K. F a t i g u e c r a c k g r o w t h r a t e c a n be e s t i m a t e d u s i n g t h e r e l a t i o n b e t w e e n AE e v e n t c o u n t s a n d d i s t r i b u t i o n of intergranular frac- t u r e s on t h e c r a c k s u r f a c e .. pp 596-605 [2] Asami. (1988). No. 1979. 346. A simple and t r a d i t i o n a l AE m e a s u r e m e n t c a n be a p p l i e d as a c r a c k e v a l u a t i o n method o f t h e p u l s a t o r s p u r g e a r t e s t . K. "Many A c o u s t i c Emission E v e n t s from P r o p a g a t i n g F a t i g u e C r a c k on C a r b u r l z e d L a y e r " . Matsumoto. " E f f e c t o f C o m p r e s s i v e R e s i d u a l S t r e s s on F a t i g u e Strength of Tufftrided Steel".. pp 499-505 Copyright by ASTM Int'l (all rights reserved).. AE t e s t w e r e c a r r i e d o u t u s i n g f l e x u r a l s p e c im e n s and s p u r g e a r s . Behavior of overall-AE event rate in spur gear tested a t 4 Hz. Th ese e v e n t s a r e r e c o r d e d in t h e p e a k . pp 638-643 [3] Obata. 3. v o l .. 65. a n d S a k a n i wa.. No.. which h a v e a high signal to noise ratio. and a s s o c i a t e d with i n t e r g r a n u l a r f r a c t u r e in t h e r e g i o n o f t h e c a r b u r - ized layer. Aokl.. 2. H. 8.. J a p a n . Kobe. 31. Journal of the Society of Materials S c i e n c e .. 13 . K. S. 270 ACOUSTIC EMISSION overall AE(4Hz Vth=O. 5mV) o "-6 \4 Z z 0 I 0 5 10 15 20 N / 103 c y c l e s FIG. T.

. No further reproductions authorized. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Novel Applications Copyright by ASTM Int'l (all rights reserved).

velocities 2. KEYWORDS: Erosion. U. INTRODUCTION The wear of important components in industrial plant caused by dust entrained in gas streams can often cause problems.7. this usually being in the range 2 to 200 m/s. The velocity and time dependence and the absolute values of the peak impact force and impact time thus obtained compared well with predicted theoretical values. J.5 .astm. ASTM STP 1077.J. . Scruby CHARACTERISATION OF DUST IMPACT PROCESS AT LOW VELOCITY BY ACOUSTIC EMISSION REFERENCE: ~uttle. The erosion rate depends strongly upon the particle velocity. The largest particles inevitably cause the greatest damage. Sachse..org Copyright by ASTM Int'l (all rights reserved). deconvolution. and K. American Society for Testing and Materials. ABSTRACT: An acoustic emission (AE) technique is developed for monitoring hard particle impact for eventual use as an on-line erosion rate sensor. Buttle and Christopher B. 1991. W. C. elastic and plastic impact. Erosion rate is a maximum at normal incidence for 'brittle' materials whilst for 'ductile' materials the erosion process is more rapid for larger angles Drs.1 m/s) on a aluminium target was investigated. "Characterisation of Dust Impact Process at Low Velocity by Acoustic Emission". The quantitative technique developed was also used to size the particles. The incident angle of the eroding particle is also important. Roget. but where erosion is known to be a problem these are usually removed by filtration of the gas stream. Eds.B. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Yamaguchi. Acoustic Emission: Current Practice and Future Directions. Oxon OXII 0RA. particle. The particles responsible can be of a wide range of sizes from just a few to several hundred micrometres in diameter. Philadelphia.i00 ~m. In a preliminary study the elastic impact of bronze and glass particles (size range 50 . For example heat exchangers. ducts and especially gas-turbines associated with coal-fuelled combusters can become irrevocably damaged if unchecked [i]. Didcot.K. This fundamental approach provides a framework for later extension to plastic particle impact. Harwell Laboratory. D. No further reproductions authorized. giving the correct mean particle size to within 4%. Green's functions. and Scruby. The plate Green's function and system response were deconvolved from AE signals due to impact at epicentre. David J. 273 C o p y r i g h t 9 1991 b y ASTM I n t e r n a t i o n a l ~. Buttle and Scruby are research scientists at AEA Technology.

composition and state of both the target and erodant [1-3]. Waveform analysis of the observed AE signals could therefore yield information relating to such parameters as the particle size. V(t). will therefore be dependent. There is a need for a robust probe for the in-situ monitoring of erosion. Parameters relating to the particle and impact dynamics are then deconvolved using a Green's function formalism and the results are compared with theoretical predictions. The Laser Doppler Anemometry technique has been used to measure gas flow in rigs [4. making it unlikely to be used for routine plant monitoring. 5-10% in the form of compression waves. but also upon the propagation of the sound in the target medium and the functional response of the acoustic sensor. No further reproductions authorized. These acoustic waves will propagate through the target according to the elastic properties of that medium before being detected by a suitable acoustic sensor. and the roughness. THEORY Acoustic Emission Signal Analysis When hard projectiles make an impact with a massive plate their kinetic energy is redistributed into a combination of plastic deformation in the target surface. it would be most useful for characterising the erosion process provided that plasticity of impact could be correlated with removal of material from the target. the theoretical aspects of elastic wave generation from simple transient force impact upon relatively thick sections has already been solved numerically [8]. angle of impact and degree of elasticity or plasticity of the impact itself. This Copyright by ASTM Int'l (all rights reserved). Thus an acoustic emission (AE) technique. . If this last parameter could be deduced. which offers the potential for on-line monitoring and should be inexpensive. of which approximately 2/3 is in the form of Rayleigh waves. either directly or indirectly. The resulting signal. Thin Layer Activation is an established technique for monitoring erosion and wear in internal combustion and gas turbine engines [6. which should be capable of operating in the often harsh environments of high temperature and pressure. The work presented here is a first stage for development of an AE technique for erosion rate monitoring. an AE technique has not yet been employed industrially in erosion monitoring. velocity. Although. not only upon the projectile and dynamics of the impact itself. is proposed as a method which fulfils the requirements of a robust probe for in-situ monitoring. measurements from individual elastic particle impacts upon plates are made. After laying a theoretical framework for determination of the characteristics of AE waveforms from elastic and plastic impact. The fraction of energy dissipated as elastic waves is typically only a few percent [9]. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. but the cost is relatively high and the component to be monitored must be portable. but does not address the erosion process itself. to the best of the authors' knowledge. 274 ACOUSTIC EMISSION of incidence.5].7]. elastic waves propagating away from the impact site and kinetic energy in the rebounding projectile. Other factors which influence erosion are the shape of the particles. and the rest in shear waves.

it is necessary to determine and deconvolve the propagation and detector functions from the observed AE signal.c) summarise these forces for normal impact.e. In our work the size of the acoustic source was small compared to the propagation distance and so the source could conveniently be modelled as a point. We shall consider each of these functions in turn. Gi~.e. Propagation Function for a Plate The elastic waves generated by the impact will propagate through the target medium according to the transfer function of the plate before being detected. Figure l(a. j=3). Thus the determination of the impact source function should yield the impact time and the magnitude of the force. the source can be represented by a cluster of orthogonal forces. Thus the displacements are given by the convolution integral: Ui(t) = !=i Gij(t) * fj(t) (2) where fj(t) = point force component acting in jth direction. in order to extract the information concerning the erosion process. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. force dipoles. The determination of the spatial dependence of the function should indicate the degree of plasticity of the impact and should also therefore relate to the rate of erosion. D(t) the detector function and '*' represents convolution in time. describes . it is assumed that all parts of the source radiate with the same time dependence. This can be calculated from the appropriate wave equation which describes the dynamic displacements in time and space according to the elastic constants of the material. The spatial form of the function will depend upon the nature of the impact. If the impact is at some oblique angle there will be an additional horizontal component to the monopolar force for both elastic and plastic collisions (figure l(b. The solution for a delta function source in time and space is known as the Green's 9 function.d)). If the plate normal is defined as the z direction (i. If then. which is contained in S(t). No further reproductions authorized. Particle Impact Source Function Particle impact will create an impulsive force at the surface of the target with a duration equal to the impact time. G(t) the propagation function. J This . A highly plastic impact would however result in forces parallel to the surface in addition to a normal force. a perfectly elastic normal impact will generate a force normal to the surface. both of which depend upon the particle size and velocity. Hence. . For normal impact plastic deformation can be assumed to produce pairs of equal and opposite forces. BUTTLE AND SCRUBY ON DUST IMPACT PROCESS 275 can be represented by: V(t) = S(t) * G(t) * D(t) (i) where. S(t) is the acoustic source function. . ith component of dlsplacement at the observatlon posmtzon due to a polnt force acting at the source location in the jth direction as a function of time. the. i. then Copyright by ASTM Int'l (all rights reserved). For example.

. for each of the examples of figure 1 are given by: Elastic Plcstic {a) (c) Normal 0 ~ i / / / / / / ! ~. For plastic impact the component G33 ~ is included for the case where the plasticity may extend to depths ' greater than the acoustic wavelength thus incurring an additional vertical dipole moment.i * Dii for oblique plastic impact (6) where Dii = the force dipole components. Particte / / / / / / / / " ~ ~:--TQrge t surface 1 (b) (d) Oblique \ 0 \ 0 //////7/" / / / / / / / / T" ) Velocity direction Force monopole -> Force dipole Figure 1 Schematic of forces during (a) normal and (b) oblique elastic particle impact and (c) normal and (d) oblique plastic particle impact.. One of the force monopole terms can Copyright by ASTM Int'l (all rights reserved). The third index of the Green's function represents differentiation with respect to that index. No further reproductions authorized.. 276 ACOUSTICEMISSION the vertical displacements at the detector. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. U = G * f for normal elastic impact (3) 3 33 3 U = ! G3i * fi for oblique elastic impact (4) 3 =l * for normal plastic impact (5) U 3 = G33 * f3 + ~ =i G3i. U3.i Dii =_nd U 3 = !=i G3i * fi + G3i.

9 mm 2) with an omni-directional sensitivity and a frequency response which falls only slowly at high frequencies (-12 dB at i0 MHz). there being some loss of low frequency after 6 ~s in both signals and a strong reflection of the compression waves in the first example delayed approximately 0. Figure 2 shows the observed epicentral response of the transducer together with its charge amplifier for (a) a sharply focussed i0 ns laser pulse (ablation mode) and (b) a partially focussed pulse (thermoelastic mode). Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized. Detai}s of this transducer design and sensitivity can be found in [ii]. For comparison. BUTTLE AND SCRUBY ON DUST IMPACT PROCESS 277 always be set to zero by choice of axes. The agreement is good. The overall signal fidelity was good. The agreement between the Green's functions and the transducer response was sufficient to indicate that.0~s after the peak corresponds to the shear wave arrival and was thought to be due to a small error in the shear wave velocity used for the Green's function calculation. This was a point-contact transducer (contact area 0. (a) the ablative laser source can be represented by the single function G33. Details of the deconvolution method used are given elsewhere (12). Copyright by ASTM Int'l (all rights reserved). and G31 i = G32 2 = -=G~3. The detector function was obtained explicitly by deconvolving the Green's function (unsmoothed) from the experimental signals of figure 2. The ideal response of a transducer to a short impulse would be a Gaussian with a small risetime. and (b) the thermoelastic laser source can likewise be represented simply by the waveform G31 i (as the three spatial derivatives of the green's function dip61es are proportional).3 (u constant) when measurements are made at e~icentre ~i0]. is given in figure 3. The smaller bipolar signal occurring 1.5 ~s. The transducer and charge amplifier response. a high fidelity piezo-electric transducer was used.5~s following the initial peak. The decay time is longer than the risetime and the reflection from within the transducer element is clearly seen to occur almost 0. Note that the Green's function of figure 2(a) corresponds to an elastic impact of a particle at normal incidence (equation 3). The two Green's functions G32 and G~I are equal. convolved with a Gaussian of width (at half pSak) 120 ns are also shown. the Green's functions G33 and G31 i. but there was some loss of low frequency content and small reverberation following the early wave arrivals [11. A Nd-YAG pulsed laser was used as a standard acoustic source. The sensitivity was calibrated absolutely in terms of output signal level per unit surface displacement by comparison with a capacitance transducer on a standard test block. This reflection was believed to occur from the back face of the piezoelectric element. thus obtained.12]. . at epicentre. Detector Function In order to minimise the distortion of the AE signal voltage due to an imperfect detector.

. Goussion ! 0!.0 r (b) Time/ Hs o-'s 1~ o ~ 1-5 ' 2-0 ' 2.sss V v V v V v 300 E Loser Colibration - c / 200 ~ ~ j ~ .5 to 7.O. The particle drop length was adjustable enabling the impact velocity to be selectable in the range 2. .J 0-5 1.0 2-5 3. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.0 3-5 z. Wave arrivals also ~hown. This was further ensured by positioning a mask plate above the target which contained a hole of diameter 1. Z. EXPERIMENTAL The Free-fall Impact System A long vacuum tube arrangement was used [12] whereby particles were allowed to fall freely under gravity before impact onto the target. No further reproductions authorized.G33 ii. The vacuum Copyright by ASTM Int'l (all rights reserved). The target consisted of a 5..8 mm thick aluminium plate with a transducer located on the underside such that the particles strike the target close to the epicentre.5 2. 4-5 54) 5-5 ' 60' ~ -10t3 ='~-150 V 631.12 ~s width Gaussian.5' 3<) . The number of particles falling was controllable. .1 m/s.1 * Goussion -200 Figure 2 Comparison of transducer/charge amplifier response to (a) G33 and (b) G31 i both convolved with 0.5 mm.0 4.5 5-0 5-5 6. 278 ACOUSTICEMISSION (a) ~00 P s ppp pps pss pp~op.0 1.5. A vibrator 'shakes' particles out of a dispenser positioned at the top of the evacuated tube.. 3.

i.u-~ i . where G(t) is the appropriate propagation function. the particle impact force function must be longer than the laser pulse. indicating that. 40 P k k / ~ I . was maintained between i0 -4 and i0 -s torr which was considered low enough to prevent significant 'drag'. The glass particle size distribution was measured under an optical microscope.d o= =r 8o o '~. BUTTLE AND SCRUBY ON DUST IMPACT PROCESS 279 200 16o . Copyright by ASTM Int'l (all rights reserved). Two experiments were carried out: (a) Bronze particles impinging obliquely at 0. RESULTS AND DISCUSSION AE waveforms from particle impact (figure 4(a)) were similar to the AE system response to an ablative laser pulse. filtered with a bandwidth of 30 kHz to 7 MHz (3 dB points) and stored digitally (i0 ns sampling time) onto the hard disk of a personal computer. Spherical bronze particles with a diameter between 53 and 75 um or glass particles with a diameter between 80 and i00 v~nwere used. The sensitivity of the AE signal to impact angle between 0 and 20 degrees for bronze particles was small (data not shown) and this was confirmed by comparison with theoretical waveforms calculated from D(t) * G(t).t t g -4o Figure 3 Deconvolved transducer/charge amplifier response to impulsive force generated by pulsed laser. The experimental signals differed primarily in having longer risetimes and lower bandwidth than in figure 2(a). i0 and 20 degrees to the target normal at a single velocity and (b) glass particles impinging normally at a selected range of velocities. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. containing both normal and horizontal components to the impact force. Acoustic Emission System and Measurements The AE signals detected at the transducer were amplified. . A.e. No further reproductions authorized. The glass particles were almost perfectly spherical and data from these impacts was used for an absolute comparison with theoretical predictions. although still of short duration.

No further reproductions authorized. since a high sensitivity would tend to hide the effects of plastic deformation in later studies.98 and 0. The impact force function. Note different time scales.e. whilst theoretically it should be 0. figure 2(a). 280 ACOUSTIC EMISSION The clearest effect of small angle oblique impact upon the AE signals was to reduce the amplitude of the first compression wave arrival and increase that of the first shear wave arrival. particularly in the ZO 10 C 0 -10 -- -20 I I I I 2 4 6 8 10 T i m e / ~Js 001A o:rll 0. For example at i0 and 20 degrees impact the first compression wave arrival was reduced in amplitude by a factor 0.985 and 0. The relatively high noise level following the initial impact force is due to errors in the deconvolution process. Copyright by ASTM Int'l (all rights reserved).00 ~ I 1 " 2.94 respectively. The AE signal of figure 4(a) is shown deconvolved in figure 4(b). was derived by deconvolution of the propagation and system response functions. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. i. This insensitivity to impact angle for small angles from the target plane normal can be considered to be beneficial to the technique.940 in excellent agreement. .0 2 3'0 3~5 -(>02 I v " Tirne l ~ s v -004 Figure 4 (a) Typical AE signal from bronze particle impact at 7.0 I m/s and (b) deconvolved force impact signal. S(t).

No further reproductions authorized. Two useful parameters which can be extracted from this signal are the impact time and the peak impact force. . The impact force function was deconvolved (assuming no plasticity during impact) for a total of 8 1 A E signals. 0-I 0.46m/s i .05 o 4 ' /o'3% '. = i . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. t i E 0-I 0"05 0 %4 v I I ~ I I t (d) V =/4-27raYs (h) V = 7-09rn/s = ._ = ='~A . 0'I 0.09 m/s. Figure 5 shows an example for (o) V: 2-/~9m/s (e) V= 5. Z 0 ~u 9 ~ " " - I I I I I | .='~'~..'o '#o ' .05 0 . Copyright by ASTM Int'l (all rights reserved). .49 to 7.L_ _ =. = .. . . . This implies that the impact velocity was never high enough to result in significant plastic deformation. BUTTLE AND SCRUBY O N DUST IMPACT PROCESS 281 determination of the inverse function..13rn/s 0. . = i .P "6 (c) V = 4-05 m/s (g) V = 6.05 I. =. I I (b) V=3-3irn/s (f) V : 5-93m/s .-d%S. Glass particle impacts on the aluminium target plate were observed for the whole available range of impact velocities from 2. and to noise in the original AE signal itself. = . i i . .o Time / ps Time / ps Figure 5 Deconvolved force impact signals from AE signals for several impact velocities.1 0. The shape of the AE signal did not change appreciably with increasing velocity (data not shown) although the amplitude increased as would be expected.

r. fell between the minimum and maximum bounds of the theory.94 x ( )2. the mean experimental peak impact force data were generally less than the mean theoretical values. These results show a remarkable agreement considering that the theory is based on semi-empirical calculation. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. . 282 ACOUSTIC EMISSION each impact velocity. is given by Hutchings [13] as: 2"5 r (i-~12) (i-~22) t e l = 2. However. 89. Observation (a) indicates that the acoustic technique was capable of giving realistic peak impact force values whilst observation (b) implies the coefficient of restitution was less than unity. We observe: (i) The peak impact force increases consistently with increasing impact velocity and (ii) the impact time is not strongly dependent upon the impact velocity. No further reproductions authorized.2 and 10O ~m impinging on an aluminium target.5 for 100Nm particies [14]. Thus the impact time.2 (equation 8) whilst the acoustic data suggested a power of 1. v and E are Poisson's ratio and Young's modulus respectively for the target and particle materials. The results are shown togenher with the experimental values measured from the deconvolved signals in figures 6 and 7 for impact force and time respectively. Using the formula above the theoretical impact time and peak force were calculated for glass particles of diameter 80. The results indicate that: (a) Nearly all the peak impact force and impact time data measured from the deconvolved acoustic waveforms.7. In the extreme limit where the particles stick to the surface of the target the maximum force will only be half as great. Note that figure 5(d) shows a double impact (observed also in the original AE waveform). The data of figure 6 suggests a value of 0.15. the peak impact force developed is given a (12. These sizes were chosen because they were the minimum. and v the impact velocity. An alternative approach is to use the acoustic data together with the theory (equation 8) to calculate the particle size. This was done and the resulting particle size distribution for the 8 1 A E signals is Copyright by ASTM Int'l (all rights reserved). tel .s [(i-~i ~ 2) + (i-~22) E ~2.13): Fel = [8 9 23 ~ 3 p 3 v 6] 1. (b) At any given impact velocity the scatter in the peak force or impact time was similar to the scatter expected due to the particle size range.5 r2 (8) 1 2 where the coefficient of restitution is assumed to be unity. This was subsequently measured using the Laser Doppler Anemometry technique to be 0. The deduced peak impact force and impact time data were directly compared with semi-empirical calculations for normal elastic particle impact on an elastically deforming plane. The impact force depends upon the impact velocity raised to the power 1. mean and maximum particle sizes used. which is in reasonable agreement.Sp v I"S [ E + E ]2.5 (7) 1 2 where p is the density of particle material of radius. Similarly.

2~ \ / ~T".09 8OHm " 0. . 0.2 a. E 0.2 pm ~0.3 / / 80~m 0.06 "~ 0. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.04 0.20 Experimental data w i t h s c a t t e r .02 I I I I . Also shown are theoretical curves for minimum. mean and maximum particle diameters.05 0. Also shown are theoretical curves for minimum.I 2 3 4 5 6 7 Impact velocitylms -1 Figure 6 Experimental measurements of impact force together with scatter as a function of impact velocity for glass particles.4 --Theoretical curves lOOpm / 89.0:3 0.5 Experimental data with scatter 0." z 0.08 0. 0. No further reproductions authorized. mean and maximum particle diameters. Copyright by ASTM Int'l (all rights reserved).~Theoretical curves 100pro 0.. BUTTLE A N D S C R U B Y ON DUST IMPACT P R O C E S S 283 0.07 E 0 ".L.10 89.1 I ! I I I 2 3 4 5 6 ? Impact velocitylms -1 Figure 7 Experimental measurements of impact time togethen with scatter as a function of impact velocity for glass particles.

The two distributions compare well. . 50 60 70 80 90 100 110 120 130 Size rcnge / jJm 36 F 28 24- c 2O L 16 t. (ii) errors in determination of the peak impact force. For example. No further reproductions authorized. The distribution determined from the acoustic data is broader due to a combination of factors such as (i) variations in the coefficient of restitution from one impact to the next caused by variations in surface conditions. A different coefficient of restitution would give a different result but the particle radius depends only on the square root of the impact force (equation 8).n n. For comparison figure 8(b) gives the size distribution determined from optical microscopy measurements taken of particles from the same stock. the mean particle size differing by just 4%. . assuming a coefficient of unity only results in a 10% error in the mean particle size.l_ 12 8 5(? 60 70 80 90 100 110 120 130 Size range / jJm Figure 8 Size distribution of glass particles (a) calculated from deconvolved AE impact signals and (b) measured using an optical microscope. 284 ACOUSTIC EMISSION given in figure 8(a) (coefficient of restitution of 0. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 16- 12 10 8 u_ 6 4 2 . Copyright by ASTM Int'l (all rights reserved).5 assumed). and (iii) lack of sphericity in the particles.

Academic Press. Wear of Materials. "A Single Particle Optical Counting Instrument for On-Line Simultaneous Measurement of Drop Size. and Scheibel. L. C. Bellanca.Atomisation and Spray Systems.. C. and both the peak impact force and impact time deduced.) Erosion: Treatise on Materials Science and Technology.. C.K. and Saunderson. Gaithersburg is gratefully acknowledged for supplying the computer code for Green's function calculations.i00 ~m over the velocity range 2. JPGC-Pwr-57. Joint ASME-IEEE Power Generation Conference Milwaukee. U. (ed. The Laser Doppler Technique. J..5 nm) for an impact at i0 m/s was determined as %1. E.E. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.7. The impact force function was deconvolved from these signals.M.1 m/s. Hemsley. D. H. J. London.. J. P.. . if the current noise level were maintained." Harwell Report AERE-RI2663. 16. Martin. D. This work has included: (a) Derivation from first principles of the elastic waveforms from normal and oblique. P. [5] Yeoman. (b) Measurement and analysis of acoustic signals from bronze and glass particle impact in the size range 50 . M. D.S. 1979. The results compared well with semi-empirical theoretical models. J. BUTTLE AND SCRUBY ON DUST IMPACT PROCESS 285 As a final consideration. New York. National Bureau of Standards. Vol. Blatchley. American Society of Mechanical Engineers. Hadded. where it is hoped to relate acoustic parameters to the effects of plastic deformation.. The authors also would like to thank Dr D H Saunderson for many helpful discussions. 1980. Wiley Publications. [3] A.. International Conference on Liquid . and Bates. Particle diameters were also determined from the AE particle impact signals and the resulting size distribution was consistent with that measured optically. [6] Duncan. ACKNOWLEDGMENTS This work has been carried out as part of the UKAEA Underlying Research (Non-nuclear Energy) Programme. R. 1985.5 .... S. 1981. Imperial College. [4] Drain. Velocity and Concentration in Sprays and Spray Systems". L. elastic and plastic impacts. A higher impact velocity would reduce this still further. D.. N Hsu of the US Departmen of Commerce. Wisconsin. Harwell Laboratory. [2] Preece.. the minimum particle size detectable was estimated. 1988.. No further reproductions authorized. D. London.4 ~m. Feb. The diameter of a bronze particle necessary in order to give a peak displacement larger than the sensitivity limit of the transducer (1. J. Sioshansi. S Martin and T Pinfold for LDA measurements and C Duffil for construction of the particle free-fall vacuum system. "Erosion in the Power-Generation and Utilisation Industries. CONCLUSIONS A broadband high-fidelity acoustic technique has been developed for characterising particle impact on plates. C. O. New York.. Copyright by ASTM Int'l (all rights reserved). Hemsley. REFERENCES [i] Buttle. This has provided a framework for future experiments on plastic particle impact. 1985. M. Oct.

4(i).. 1989. P. S. C. T. G... [8] Ceranoglu. "Energy Absorbed by Elastic Waves During Plastic Impact". B. 1977. Copyright by ASTM Int'l (all rights reserved). 1985. "Acoustic Emission Measurements Using Point-Contact Transducers". Harwell Laboratory. Didcot. 1983. [12] Buttle. 12. Journal Physics D: Applied Physics Vol. M. Y. pp. Harwell Report AERE-RI3439. [13] Hutchings. 48. I. pp.K. Didcot. K. A. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. and Pao. New York. AEA Technology. "Impact and Rebound Velocity Measurements on Small Particles". [9] Hutchings. B.K. D. AEA Technology. C. Simmons. Part 3: General Responses". Journal Physics D: Applied Physics Vol. "Strain Rate Effects in Microparticle Impact". and Stockton. U. pp. No further reproductions authorized. American Society of Mechanical Engineers. Accepted for publication in Wear (1989) and Harwell Report AERE-RI3028. J. R... Harwell Laboratory. Vol. Apr... Pinfold. C. 1979. Journal of Applied Mechanics. [ii] Scruby. LI79-LI84. I. U. 9-18. Journal of Acoustic Emission. Harwell Report AERE-RI0401. "Propagation of Elastic Pulses and Acoustic Emission in a Plate. pp. 1981. N. Sept. 1982. A.. Part i: Theory. AEA Technology. N. 1988. i0. and Scruby. Mar. and Wallace-Sims. [14] Martin. 1819-1824. . Didcot. "Dynamic Elastic Displacements due to Pulsed Laser Absorption at the Surface of an Infinite Plate". M. G. Trans ASME. 286 ACOUSTIC EMISSION [7] Sioshani. U. M.K. 125-147. [10] Wadley. "Characterisation of Particle Impact by Quantitative Acoustic Emission". ASME 83-JPGC-24. Harwell Laboratory.. H. J. R. Part 2: Epicentral Responses..

. Hitachi Ltd. machinery is used under severe conditions. better reliability and preventive maintenance are required. Minoru Yanagibashi. Acoustic emission (AE) is a useful means of machine condition diagnosis. and chemical plants. W. rotating machinery.. M. Kouichi Sato. I. Hitachi-shi. Eds.. The AE technique can detect incipient failure and locate the source. For instance.. Japan. and Kazuo Takikawa APPLICATIONS OF ACOUSTIC EMISSION TECHNIQUES FOR DIAGNOSIS OF LARGE ROTATING MACHINERY AND MASS PRODUCTION PRODUCTS REFERENCE: Sato. and Takikawa. ABSTRACT: In order to maintain efficient plant operability. Ohira-cho. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Tochigi-ken. mass production products INTRODUCTION In large plants such as power plants. J. .. I.. then take measures to prevent serious damage. T. and applications for the diagnosis of rotary compressors in air conditioners. Sato and Tanaka are engineers at Tochigi Works~ Hitachi Ltd." A~oustic Emission: Current practice and Future Directions. American Society for Testing and Materials. Roger. Sato and Yoneyama are researchers at Energy Research Laboratory. 1168. 317. 800. 287 C o p y r i g h t 9 1991 b y ASTM I n t e r n a t i o n a l ~. Japan. Ibaraki- ken. 1991. Yanagibashi. 316~ Japan. Ichiya Sato. Ohse-cho. KEYWORDS: acoustic emission.. Sato. Hitachi-shi. Ibaraki-ken. and K. ASTM STP 1077. Toshiyuki Tanaka. "Applications of Acoustic Emission Techniques for Diagnosis of Large Rotating Machinery and Mass Production Products. K. operators must find the cause of trouble. development of diagnosis systems using algorithms obtained from AE characteristics analyses. steel plants. Shimotsuga-gun. machine condition diagnosis. Philadelphia. T. Yoneyama.. Tanaka. K. No further reproductions authorized. 329-44. Yamaguchi..org Copyright by ASTM Int'l (all rights reserved). Moriyama-cho.astm.. Takao Yoneyama. Yanagibashi and Takikawa are engineers at Hitachi Engineering Service Company. Sachse. K. 2-9-i. This paper describes applications of AE techniques for large rotating machinery diagnosis. When a machine experiences some problem during use.

In the inspection of mass products~ automatic inspection and high accuracy diagnosis are required. The acoustic emission technique can satisfy all these conditions. In order to perform high accuracy diagnosis in large plants and mass production factories~ the following conditions are necessary. Next~ we developed a total diagnosis system with digital processing techniques using the algorithm obtained from the analyses ~3]. In these applications~ the relationship between various abnormal conditions and the AE characteristics were investigated systematically. (c) Location of abnormal sources. 288 ACOUSTIC EMISSION the machinery in thermal power plants are being operated by daily start stop~ or weekly start stop schedules. (b) Early detection of abnormal condition. (a) Detection of primary abnormal phenomenon. Therefore~ in order to maintain efficient plant operability~ better reliability and preventive maintenance are required. The bearing diagnosis system was developed in order to detect bearing damage at a early stage in journal bearings for steam turbines. First~ we have developed individual diagnosis systems with analog processing techniques. On the other hand~ automation of mass production factories which manufacture consumer products or automobile appliances is rapidly progressing. Generators -For Rolling Mills # Mass Production I AE Wireless Monitor Products .For Rotor Diagnosis Diagnosis System FIG. This paper describes applications of acoustic emission techniques for diagnosis of large rotating machinery and mass production products. AE wireless monitor was developed in order to detect and transmit directly AE signals from an operating rotor Eli. The rubbing diagnosis system can detect a slight rubbing phenomenon and locate the source in a rotating machine which is giving rise to abnormal vibration ~I~. Figure 1 shows the machine condition diagnosis systems. 1 -.The machine condition diagnosis system. No further reproductions authorized. So~ development of high accuracy diagnostic techniques is very necessary. . (d) On line inspection of these above conditions. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Individual Diagnosis System Total Diagnosis System with Analog Processing with Digital Processing Rubbing Diagnosis System f Machine Diagnosis Bearing Diagnosis System System -For Turbines. Copyright by ASTM Int'l (all rights reserved). generators ~i] and rolling mills ~2].

Envelope detection waveforms and frequency spectra in the rubbing test. As can be seen from upper waveforms~ it is difficult to distinguish rubbing phenomena from background noise.2 -.3 -. Figure 3 shows envelope detection wafeforms and frequency spectra in the rubbing test. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. FIG. ON TECHNIQUES FOR LARGE ROTATING MACHINERY 289 For application to mass production products~ we developed a diagnosis technique with the digital system for rotary compressors of air conditioners ~4].A view of the rubbing diagnosis test. FIG. Each rubbing test piece is composed of a steel rod with an aluminum oil splatter guard fixed to one end. RUBBING DIAGNOSIS Large scale rotating machines have a potential for metal to metal contact between rotors and stators. The test rotor is a turbine rotor for a 350 MW high-intermediate pressure steam turbine. Copyright by ASTM Int'l (all rights reserved). Figure 2 shows a view of a rubbing diagnostic test. SATO ET AL. . No further reproductions authorized. AE sensors were mounted on the bearings.

Copyright by ASTM Int'l (all rights reserved). Figure 4 shows a schematic diagram of the rubbing diagnostic technique. The upper waveforms are amplifier output signals from two sensors. No further reproductions authorized. It is difficult to distinguish the difference between normal condition and rubbing. A rubbing source can be located by correlating the rubbing signals detected by the two sensors. FIG. 290 ACOUSTICEMISSION In order to extract rubbing signals from the noise~ a frequency analysis of the signal was performed after envelope detection. .5 -.Rubbing signal in rubbing diagnosis test. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.4 -.Schematic diagram of rubbing diagnostic technique. Then. This technique separates rubbing signals from background noise using a variable bandpass filter which has a center frequency identical to that of the rotation signal. source location is obtained from the time difference AT. Such analyses have shown that the main frequency component of signals caused by rubbing is equal to the frequency of the rotation signal. Figure 5 shows examples of rubbing signals observed in a actual steam turbine. FIG.

This causes the oil to thin. In this case. FIG.7 -. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. an excessive load on a bearing causes an increase in the temperature of the lubricating oil. The main factors contributing to journal bearing damage are radial plane damage and thrust batting as shown in Figure 6o In radial plane damage. some bearings have been damaged due to an increase in oil temperature or excessive load on the bearing. No further reproductions authorized. which results in the bushing face of the bearing being worn off. Copyright by ASTM Int'l (all rights reserved). This method lacks accuracy. SATO ET AL. Rubbing phenomena can easily be detected. Therefore.Classification of slide bearing damage. it was found that a rubbing source was at a vicinity of the sensor $1. Up to now~ monitoring of the journal bearings were performed by temperature measurement of lubricating oil.AE waveforms produced by thrust butting. The thrust batting is a metallic contact between the sleeve side and the bushing side.6 -. . BEARING DIAGNOSIS We applied the acoustic emission technique to diagnose backup roll journal bearings for a rolling mill. ON TECHNIQUES FOR LARGE ROTATING MACHINERY 291 The lower waveforms are corresponding signals that have been processed with the rubbing diagnostic technique. FIG.

Figure 7 shows AE waveforms produced by thrust butting. This system can be monitored by scanning several bearings during the milling process. J I I FIG.. (Object: Journal bearing fdr backup roll of rolling mill) AE Sensor BUR: Backup Roll Pre Amp~ier Journal Bearing / " /~l//~--iD ~ l . The system is being used successfully in an actual mill.8 -. As seen in this figure~ the waveform characteristics are burst type~ and the frequency spectrum has a wide band. The AE waveforms have a long duration time. .A diagram of a monitored rolling mill using the bearing diagnosis system. ~ .. Figure 8 shows an example AE waveform and the frequency spectrum of an envelope detection signal caused by radial plane damage.~ . No further reproductions authorized. From these results~ it was found that the frequency spectra of signals after envelope detection are useful for diagnosis of journal bearing damages. Et " '~ I I 01 20 4O 6O 8O 100 (fr~ 1Hz~ Frequency (Hz) (2V/div) (0.9 -. Figure 9 shows a rolling mill with a bearing diagnosis system whichwas developed through the artificial damage test and signal analysis. ~. Copyright by ASTM Int'l (all rights reserved).AE waveform and its frequency spectrum of envelope detection caused by the radial plane damage.5s/div) FIG. 292 ACOUSTIC EMISSION In order to observe acoustic emission behavior during damage of journal bearings~ artificial damage tests were conducted with an actual mill. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The AE signals were analyzed for waveform and frequency characteristics. .

A block diagram of the diagnosis system. Using waveform characteristics~ AE signals are classified as either a continuous type or a burst type signal. As shown in Table I~ the basic algorithm of the system discriminates between various abnormal conditions in machinery using AE waveform characteristics and frequency characteristics of envelope detection signals.10 -. ON T E C H N I Q U E S FOR LARGE ROTATING M A C H I N E R Y 293 DIAGNOSIS SYSTEM Basic Diagnosis Algorithm A machine condition diagnosis system was developed using an algorithm obtained from the AE characteristics analyses. The narrow band type is further identified as either a rotation tuned type or an untuned type. TABLE i -. The AE and external signals are converted in to digital signals~ and are memorized in a digital memory. The AE signal is categorized by six types of abnormal conditions. Frequency Characteristics Wavefor m Characteristics of Detection Signal Continuous Type Burst Type Rotation Tuned A 1 A 2 Narrow Band Type ( Rubbing ) (Rotor Crack) Type B 1 B 2 Untuned Type (Metal Wipe) (Fatigue Crack) Wide Band C 1 C 2 Untuned Type (Radial Plane Type ( Bea ring Ti I t ) Damage ) Note : ( ) is An Example of Abnormal Condition I IRotation signal FIG. SATO ET AL. Typical phenomena are shown in parentheses in this table. Copyright by ASTM Int'l (all rights reserved). Waveform and frequency analyses are done by a personal computer~ as well as evaluation of the analyses' results. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Figure I0 shows a block diagram of the diagnosis system.Basic algorithm of the diagnosis system. . Then~ using the frequency characteristics~ the AE signal is identified as either a narrow band type or a wide band type. No further reproductions authorized.

In order to analyze the waveform and frequency characteristics~ such parameters are calculated by a personal computer. Figure 13 shows the algorithm of the basic diagnosis software. This includes mean value~ AE event~ duration time~ peak value r rise time~ AE energy and FFT. The frequency band judgment is decided by employing a frequency band judgment constant Kfw as shown in the following equation: Ap / A m ~ Kfw (i) where Ap = highest peak value of power spectrum (PSD) in the frequency spectrum~ A m = mean value of PSD in the frequency spectrum~ Kfw = frequency band judgment constant. Table 2 describes the calculation method of the waveform parameters. : Duration Time T : Memorizing Time Tri : Rise Time FIG. and continuity judgment. It consists of three judgment functions: frequency band judgment~ tune/untune judgment. ~' < 9 Rubbing Graphic Display Duration Time < 9 Rotor Crack Data Display Peak Value I 9 Bearing Damage < Trend Display Rise Time I Finish AE Energy I FFT I[ FIG. 12 -. The frequency analysis is performed by the FFT calculation. Condition Set.Illustration of waveform parameters.II -. Mean Value Basic Diagnosis <~ < Diagnosis AE Event Application Diag. 294 ACOUSTICEMISSION Figure ii shows a flow diagram of the diagnosis system software. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. Copyright by ASTM Int'l (all rights reserved).Flow diagram of the diagnosis system software. . Vt~ : Threshold Level TG. If equation (i) is satisfied~ the AE signal is determined as a narrow band type~ otherwise it is a wide band type. No further reproductions authorized. Figure 12 shows one a~proach for waveform analysis.

If the equation is satisfied~ the AE signal is judged as the rotation tuned type~ otherwise as untuned.i-1 v0 Rise time T . = ~ ~ V'2 (Verier) "" i~1 'Continuous Continuity judgment Burst Tune/untune(/Tune judgment \ Untune Continuous Continuity Frequency / Narrow band judgment Burst band judgment Normality / Normal Wide band Continuity judgment / judgment Continuous Burst Abnormal FIG. Duration time 1 N (Tdurat) To~rat:~ ~ T~.13 -..~ V n i-I AE event N is total counts over (N) the thereshold level Vth.. ON TECHNIQUES FOR LARGE ROTATING MACHINERY 295 The algorithm for tune/untune judgment classifies as a rotation tuned type or untuned type using the following equation.. SATO ET AL..Calculating method of waveform parameters.. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. (mrise) i=1 AE energy V . = L....Algorithm of the basic diagnosis software. I I-Fp/Frl ~ Se (2) where Fp = frequency at the highest peak in the frequency spectrum~ Fr main frequency of the rotation signal~ S e = permitted error for the tune/untune judgment. No further reproductions authorized. Waveform parameters Calculating method Mean value 1 rl (V~) V~ao =-.. TABLE 2 -. Copyright by ASTM Int'l (all rights reserved). Peak valUe ( Vpeak) v . =N ! ~ tr. .

(2) If Tdmax is larger than Tcont (continuous judgment time)~ the signal is the continuous type. the signal is the continuous type. (i) If there is no AE event. 296 ACOUSTIC EMISSION Figure 14 indicates the algorithm for the continuity judgment. No further reproductions authorized.rnent T i m e FIG. The system has basic diagnosis and application diagnosis software. (2) or (3) are not met. N=O (/Yes \ No Continuous / Yes Type T~a• [40 /NoIT~at. The judgments of waveform analysis are performed with the following waveform parameters: AE event (N). normal conditions have to be determined by a normality judgment algorithm. .T=. Using the algorithms for the three judgment functions as described above. Otherwise it is the continuous type. BI. and Tdura t is larger than Tburs t (burst type judgment time). the AE signal is categorized into six types of abnormal conditions which are denoted as AI. maximum duration time (Tdmax) . A2. (4) If conditions (i). mean value (Vmean) ~ mean duration time (Tdurat).An algorithm for continuity judgment. mean value of peak amplitude (Vpeak). (3) If conditions (i) or (2) are not met. Suitable application software is provided for rubbing diagnosis. The continuity judgment of the AE signal is processed as follows.t/No V~ean " Burst IN : AE Event Tcont : C o n t i n u i t y Judgment T i m e Vpeak: Peak Value Kpw : Waveform Judgment Constant Vmean: Mean Value mdurat : D u r a t i o n T i m e mdmax: M a x i u m D u r a t i o n T i m e Tburst : B u r s t Type Jud~. rotor crack diagnosis and bearing diagnosis. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. the signal is the burst type. which classifies the signal into continuous type or burst type. Copyright by ASTM Int'l (all rights reserved). Thus. Generally.14 -. and VDeak/Vmean is larger than Kpw (waveform judgment constant)~ the signal is the burst type. B2~ CI and C2 type. abnormal conditions and source locations can be monitored in detail.

20 (3. (Volt • 0. Volt.16 0. T~i~ (msec) 0. Volt.26 8.29 0.15 0.03 9.84 0.51 CH2 (Volt) 0. LOCATION (m) : -3.03 (0. DATA 1986/06/26 TIME 13:31:07 DATA# 17 SENSOR # 1 SENSOR # 2 1. No further reproductions authorized.%) 50. SATO ET AL.36 0.01 (0. V~e.An example of "graphic display" obtained with the rubbing diagnosis software. Figure 15 shows an example of "graphic display" obtained with the rubbing diagnosis software.62 6.00 5. 15 -.6%) 27. SIGNAL T Y P E : At AI 10.17 4.An example of "data display" obtained with the basic diagnosis software (object:steam turbine).11 CH1. From the experimental results~ it is seen that the source location of the continuous signals is determined accurately.22 (4. EXT CHt (Volt) 1.10 7. TABLE 3 -.65 0. . From the experimetal results.41 0.39 2. Table 3 shows an example of "data display" obtained with the basic diagnosis software. Spect rum 7.Fr (Hz) 50.52 0.1 F1 (Hz. Copyright by ASTM Int'l (all rights reserved).3 F3 (Hz. md~rat (n3sec) 0. it is found that the abnormal condition belongs to the A1 type. %) 28. %) 274.03 (0. ON TECHNIQUES FOR LARGE ROTATING M A C H I N E R Y 297 Evaluation Test on Rubbing Diagnosis The system evaluation test was done with the data of rubbing diagnosis tests of steam turbine.2%) 2. N (Count/sec) 126.8%) 72 F2 (Hz.00 96. Vme. Vpea~ (Volt) 0.8%) 7.2%) 7.4 T O T A L (Volt) 5.~ (Volt) 0. Volt.17 5.2%) 50.01 (0. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.30 FIG.93 0.65 3.70 0.29 0. This means that the rubbing signal is judged as a continuous type signal and its frequency characteristics are rotation tuned type belonging to the narrow band type.

. and it is driven by a motor mounted on top.. 298 ACOUSTIC EMISSION DIAGNOSIS OF MASS PRODUCTION PRODUCTS Mass production product diagnosis techniques were developed so that the rotating slides of compressors can be monitored nondestruetively. / / . and a vane. When any of these sliding surfaces of a compressor is damaged as described above~ the operating efficiency of the compressor decreases and an abnormal stoppage during operation may result.. / ~ . If techniques become available and allow evaluation without disassembly.~ ' . _ Suction Suction space ~ V / / \ \ way ~ V a n e FIG.. ~ way //~ ~ ( 0 ) . The crank shaft is supported by an upper bearing~ and a lower bearing. (O) :Dia nostic Object (O)~J Crank shaft A =.. ~ ~ . the inspection efficiency improves and the development of new products would be promoted." ~ ':'~ane I Roller + t ~ . Upper bearing~ ~~ Hence early detection of the damage in sliding surfaces is important for improved operaing efficiency and reliability.~ .. The key part of the rotary compressor consists of a crank shaft. No further reproductions authorized.Structure of rotary compressor.. (3) Slide surfaces between the outside of the roller and vane. as illustrated in the lower drawing of Fig. a roller.:... The objectives of diagnosis are to detect damages in the following portions of the sliding surfaces.. ....~ ~ AE sensor]__~ I \~x ~ . The reliability of rotary compressors in production has been evaluated through visual inspection of disassembled parts after long-term testing. This will also enable 100% inspection of compressors in production. (i) Slide surfaces between the crank shaft and bearings. 16 -.. 16. This means the premature failure of the air conditioner that uses such a compressor. The vane slides on the rotating roller~ and moves in and out of a slit in a cylinder that forms the compression and suction spaces. /~ Crank shaft ~ ~ ~ I : Roller /~(O)/Discha rge Compression space. (2) Slide surfaces between the roller and the faces of the beaings. (4) Slide surfaces between the inside of the roller and crank shaft. Copyright by ASTM Int'l (all rights reserved).. ~ ...-"~A~))/[ : 1 _ _ Bearing Lower bearing . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The crank shaft and the roller move in a rolling piston motion. Structure of a Compressors and Diagnostic Objectives Figure 16 shows the structure of a rotary compressor used in room air conditioners.. motor~_ Crank shaft .

ON T E C H N I Q U E S FOR LARGE ROTATING M A C H I N E R Y 299 AE Behavior under Normal Conditions and Compressors with Defects Acoustic emission behavior under normal conditions was observed with an AE sensor which was mounted on the compressor chamber as shown in Fig. 18.~ (a) Normal Nothing Rotation (b) Abrasion Upper journal Upper bearing 10/. 16.N-.~ _ ~ _ n _ R _ n _ FIG. J r--TOZ Rotation -. Suction way Suction way ~ Oiso wayrge ~ // Roll~f-~'. This indicates burst type AE signals generated in synchronization with a rotation signal which is obtained from a motor control signal. Figure 17(a) shows a typical AE waveform under normal operation ("normal AE"). we conclude that the generating mechanism for the normal AE signal is the butting phe- nomenon of the vane at the top dead center of the roller~ as shown in Fig. Roller (a) Vane is pushed to a wall (b) Vane butts a wall plane plane of suction side of discharge side FIG~ -. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. From these. . In order to evaluate the AE phenomenon under normal operating conditions further~ experiments were performed using test compressors on which several sensors were mounted.. with a maximun~ sensitivity of -70dB (ref. AE signal of excepting the burst type signal is low level.Examples of compressors with artificial defects and their AE waveforms. No further reproductions authorized. Copyright by ASTM Int'l (all rights reserved).. SATO ET AL. This AE sensor was a broad band type..hi Rotation - (c) Wear Pin T r-T l Roller outside. In Fig. 18(a)~ the vane is pushed to the suction-side wall Obj~t Defect places Surface roughness AE waveforms ( A f t e r ol3eration) Lower bearing 10/.17 -.The generating mechanism of normal AE. iV/pbar) at 160kHz.

A block diagram of the diagnosis system. (' CompositeTypeDetector ) ~" RotaryCornpressor ) -. It also includes measurements of surface profiles after operation. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.. Copyright by ASTM Int'l (all rights reserved). These may mask normal burst type AE signals as in the case of wear. The artificial defects to simulate abrasion damages were made by a punch on the upper and lower journals of crank shafts.. 300 ACOUSTICEMISSION by the roller.. (1) AE characteristics of compressors with artificial abrasion damages include a number of burst type signals.. 18(b). 19 -. Figure 17(b) and (c) show examples of compressors with artificial defects and their AE waveforms. These are generated between normal burst type AE signals from vane butting. (2) AE characteristics of compressors with artificial wear damages are exemplified as the continuous type signal. In order to investigate AE patterns of rotary compressors with defects~ experiments were conducted using compressors with artificial defects. Wear damages were simulated using a sandpaper on the pin surface of crank shafts and on the nose of a vane.20 -. ration s~vL or!s AE Sensor 4~ AE Signal ~q .An AE detection and signal processing diagram for diagnosis of rotary compressors. In Fig. No further reproductions authorized. AE patterns of the compressors with defects have the following characteristics. the normal AE is generated when the vane butts the discharge-side wall as the roller reaches the top dead center position... Mainamplifier C~176 E sensor Digital memory ~tPers~ computer FIG.. . I Signal [~lWaveform ~_~ I /Averaging t IAnalysis Abnormal RotationSignal I / Judgment l FrequencyI J Analysis I -I FIG.

O N T E C H N I Q U E S F O R L A R G E R O T A T I N G M A C H I N E R Y 301 Signal Analysis Method Signal analysis of the acoustic emission signals was conducted with the machine condition diagnosis system. as shown in Fig. The AE and external sensor signals are converted into digital forms. . 20.21 -. is obtained by the following equation. The results of AE events are useful to discriminate abrasion damaged slides from normal products. Threshold Level FIG.~ I-1 i-L_. the following waveform parameters are calculated with averaging of the processed signals: AE events. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. the results of frequency analysis are useful to discriminate slides with wear damage from normal products. Threshold Level Wesr Jill ~ I ~ 59Hz Rotat ion ~ .Typical signal analyses data of AE detection signals. In order to develop the diagnosis algorithm for rotary compressors~ the following parameters were extracted to evaluate the results of the signal analyses. Krf = Afr/Aft (3) where Krf = rotating tune factor~ Afr = PSD of the fundamental rotation frequency in the frequency spectrum. Rotation-. Aft = total PSD of the whole frequency band in the frequency spectrum. Figure 21 shows typical data of frequency and waveform analyses. described previously. No further reproductions authorized.. and minimum values. a rotating tune factor. I-I I-] ~{. In order to analyze the waveform characteristics.. . S A T O ET AL. and are stored in a digital memory. Figure 19 shows its block diagram.h Threshold Level (5050rpm) N : : ~J" : : . From the frequency analysis. is obtained with the following equation: Kind of AE Detection Frequency AE Event Count Defects Waveforms Spectra Normal (4970 rpm) ~ i 3Hz N i. the mean-to-peak ratio~ Kmp . mean~ peak. Waveform and frequency analyses are performed by a personal computer~ which also evaluate the results of these analyses. As can be seen from the figure. Copyright by ASTM Int'l (all rights reserved)... Krf .-~.. J-'l n lllllll. Rotation ~ .LL. From the results of waveform analysis. Signal analyses are performed by waveform analysis and frequency analysis using envelope detection signals.

124 0.213 0.~ N Kmp diagnosis Disassemble Normal Without Nal 2.394 0.376 0. diagnosis algorithm for rotating slides of rotary compressors was developed using the resultant correlation.3 0. The results obtained were correlated either to the AE events counts and the rotating tune factor.22 -.Basic diagnosis algorithm for rotary compressors.37 2. as shown in Fig. . Table 4 shows examples of results which were obtained in the diagnostic tests. Copyright by ASTM Int'l (all rights reserved).3 0. Diagnosis System and its Application to Factory Inspection Diagnostic tests were conducted with numerous compressors of various types. 22. 302 ACOUSTIC EMISSION Kmp = Vmean/Vpeak (4) where Kmp = mean-to-peak ratio. Revolution AE parameters Results of Results of Samples (rpm) K. Vmean = mean value~ Vpeak = highest peak value. It consists mainly of three judgment functions: TABLE 4 ..79 (Wear) VaneNose Abnormal Abrasion No.46 (Abrasion) at Bearing Peak ~ High Abnormal judgment \ L o w (Wear) Tune ver Event / O Tune judgment \Normal judgment Normal \ Untune Event ~ Normal judgment \ Over Abnormal (Abrasion) FIG. After the tests.29 1.Examples of results which were obtained in the diagnostic tests. No further reproductions authorized.22 Damage Normal Without No.7 0. or to the mean-to-peak ratio and the rotating tune factor.2 3.4 5. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.12 2.18 Damage Abnormal Wear of N~3 5. these were disassembled and inspected visually.0 0.12 1.

Using the algorithm. SATO ET AL.24 -.The on-line test in an air conditioner factory9 FIG. . Copyright by ASTM Int'l (all rights reserved). No further reproductions authorized. This diagnostic algorithm can be applied to all types of rotary compressors for air conditioners 9 The detector used in the system is a composite type consisting of an AE sensor and a rotation signal sensor. The structure of the composite type detector developed for on-line inspection~ as shown in Fig. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. the AE signal is categorized as normal or as one of two types of abnormality. FIG. The rotation signal sensor is an accelerometer which detects vibration signal tuning to determine the rotational frequency 9 The detector with a magnetic holder can be easily mounted on the compressor under test. wear or abrasion. 20.23 -.An example of data which were obtained in factory tests. ON TECHNIQUES FOR LARGE ROTATING MACHINERY 303 9 Tune judgment using the rotating tune factor 9 9 Peak judgment using the mean-to-peak ratio~ and 9 Event judgment using the AE event count.

and Watanabe.. T. No. T. pp. and Hata. T. "Diagnosis of Backup Roll Journals for Rolling Mill by Acoustic Emission". H. T. 7.~ Yoneyama. ~4] Sato.. Tanaka. Vol. pp. Figure 23 shows a photograph of the on-line inspection in an air conditioner factory. 1988.. Yoneyama. S. The developed diagnostic techniques should also be applicable to other mass produced goods. .. K. (i) Each individual diagnosis system with analog processing techniques. No. S. Sato.. T. 1-10. we have developed machine condition diagnosis system with acoustic emission techniques. Inoue.. Koga. K.. "Rotating Machinery Diagnosis with Acoustic Emission Techniqes".. Journal of Acoustic Emission. Figure 24 shows an example of data which were obtained in factory tests. and Suzuki. Tokyo. Suzuki. ~2~ Inoue... No further reproductions authorized. T. Yoneyama. T. CONCLUSIONS In order to perform high accuracy diagnosis in large plants and mass production facilities.. Copyright by ASTM Int'l (all rights reserved). Journal of Acoustic Emission. Vol.. 304 ACOUSTIC EMISSION Results of AE tests correlated well with the visual inspection of disassembled rotary compressors following life-time tests. REFERENCES ~I] Sato. 1983. Kakehuda. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. pp. 17. Some products with such defects were discovered by the tests.. T. the digital system for rotary compressors of air conditioners was developed. 4.. T. 4. I. I. Proceedings of the 8th International Acoustic Emission Symposium. pp. 173-178.. Ozaki. Kawasaki Steel Giho. 1. Many products more than several hundred thousand have already tested until July 1989.. 1986. No. S. April 1985. T. Sato. I. N.. M.9 Yoneyama. (3) For application to mass production products. 2. Sasaki. Sasaki. Vol. and Hiruoka. "Diagnosis of Rotating Slides in Rotary Compressors using Acoustic Emission Technique". 65-72. (2) The machine condition diagnosis system which uses a digital processing technique was developed for the diagnosis of various abnormal conditions of rotating machinery.. rubbing diagnosis... 52-58. The developed AE system was applied to on-line inspection of air conditioners on factory floor with satisfactory results. were developed and are in use in actual plants. ~3] Sato. bearing diagnosis and rotor diagnosis system. and is applicable in on-line inspection of these products. 1/2.

No further reproductions authorized. March are engineers at Engineering Laboratory. of the AE signal integrated over 1 second that is proportional to the average cavitation energy. R. An inexpensive AE transducer. RMS. Ourang Derakhshan. and RMS circuit is described with a gain of 22 dB and a frequency range of 10 kHz to 1. Data from two laboratory controlled cavitation experiments is presented. The bubble collapse is accompanied by a high amplitude pressure pulse. Cavitation in hydromachinery has been known to the scientists and engineers since the early part of this century [1]. Tennessee Valley Authority. of Mechanical Engineering. Jones and Mr. R. Keith Jones. K. damage to the turbine blades and also to the surrounding surfaces) costs the hydropower industry enormous amounts of revenue for repairs and replacement of hydroturbine systems annually. Acoustic Emission (AE).. 1991. J. Kaplan-type and Francis-type turbines had a similar shaped increased RMS AE curve with wicket gate opening but the numerical maximum values are different in each hydroturbine installation. R." Acoustic Emission: Current Practice and Future Directions. Root Mean Square (RMS). and K. J. which at times reaches damaging shock wave proportions.e. TN 38505. This high amplitude pressure pulse is the cause of cavitation damage (pitting of the boundary surface). . and March. O. Houghton are affiliated with Dept. ABSTRACT: Acoustic Emission measurements are shown to be a reliable indicator of cavitation pressure pulses. Jones. Thus it often happens that the hydroturbines are operated in cavitating range. Energy. W. March CAVITATION MONITORING OF HYDROTURBINES WITH RMS ACOUSTIC EMISSION MEASUREMENT REFERENCE: Derakhshan. Philadelphia. low cost. and Patrick A. Sachse. ASTM STP 1077. Cookeville. The cavitation damage to the hydroturbines (i. and then the bubble collapses once it has moved to a region of higher pressure. Inception occurs when the pressure of the liquid is reduced rapidly to or below that of its vapor pressure. A. The system is tested on two TVA hydroturbines. Mr. Tennessee Technological University. Cavitation is defined as the process of generation (inception) and collapse of vapor and/or gas bubbles in liquids. monitoring system to inform the operators of the cavitation status of their systems. Richard Houghton.. Houghton. It is considered as one of the important damage mechanisms that must be dealt ~vith. filter. American Society for Testing and Materials. P. A real-time continuous RMS AE monitor is recommended on each hydroturbine for the evaluation and control of cavitation erosion.. J. TN 37828. Roger. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.. Mr. Yamaguchi. KEYWORDS: Cavitation.org Copyright by ASTM Int'l (all rights reserved).25 MHz. Derakhshan and Prof. "Cavitation Monitoring of Hydroturbines with RMS Acoustic Emission Measurement. Consequently the hydropower industry for years has been interested in some type of real-time. Although the designers try to rectify cavitation inception in their designs. amplifier. it is not possible to do so efficiently for all operating and environmental conditions. Eds.astm.. The method of display is the root mean square. 305 Copyright9 1991 by ASTM lntcmational www. Norris.

306 ACOUSTICEMISSION This paper reports on a low cost solution to cavitation monitoring of hydroturbines. The energy of a voltage signal. Also for the purpose of matching the impedances between the transducer and the amplifier. and (2) placed on the outside wall of the tank. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. The cost of this system is about $180. as measured using a thermocouple power meter. 1) at the Tennessee Valley Authority (TVA) Engineering Laboratory in Norris. Copyright by ASTM Int'l (all rights reserved). A 10 kHz highpass.(30 plus 20 hours of labor for assembly and testing.~$ where R represents the balancing resistance used in the power meter. is a Valpey-Fisher VP1093 Pinducer. is E = ~ l f/ v2(t)dt. This RMS value can then be displayed in variety of forms. and T is the time duration of the whole digitized waveform over which the voltage signal is averaged. To reduce cost and amount of electronic components. a number of single spark generated bubble collapse tests were performed. it was decided to use the RMS voltage of the AE event as the indication of the AE intensity level. was used to determine the effect of cavitation bubble cloud on the propagation of the pressure pulses. then comparing the Vr. The transducer in turn converts this displacement into a voltage signal. If the exact value of the energy is of no interest. The signal from the transducer is amplified by an amplifier built around the AD5539JN ultra-high frequency Op-Amp. For this tests the AE signal is from two transducers: (1) located on the test specimen. The monitoring system employs a transducer which is positioned on the exposed metal liner of the draft tube.~. double pole filter was found to be necessary to reduce the extraneous background signals. used for detecting the pressure pulses transmitted through the metal liner.e. steel pipes) are addressed. Vr. . but rather its relative value. 1 yields E = TV~ (3) R r. the system only needed to distinguish between no cavitation. It has a constant gain of about 22 dB from 100 Hz to 3 MHz and a maximum signal to noise ratio of 62 dB. (1) and since root mean squared voltage V~. In Acoustic Emission (AE) an energy release is transmitted through the medium in a displacement form and is sensed by a transducer. 2 in Eq.~a is sufficient. a 1 Mfl input impedance buffer was added between the transducer and the amplifier. INSTRUMENTATION The preliminary plan for the monitoring system was to verify that the cavitation bubble collapse has the characteristics of an AE signal source and then to design the monitoring system. To investigate the cavitation bubble collapse as a viable source for Acoustic Emission (AE) purposes. a broadband transducer with a good frequency response up to 1. No further reproductions authorized. and intense cavitation. of the signal is 2 = ~1 f0 T V2(t)dt. The energy level of the cavitation was the parameter that was selected to be monitored. (2) then substitution of Eq. The Cavitating Jet Erosion Test Facility [2] (see Fig. whose signal is then amplified and converted to its root mean square (RMS) value. In these tests such issues as the attenuation of the pressure pulse in water with distance and its transmissibility through a second medium (i. Tennessee. That is to say.25 MHz. The transducer.~8. cavitation.

7K SlF~NALDUTPUT 500p OtJT_~_~ L T~. No further reproductions authorized. Presently the averaging time is switchable for either 64 • 10 .Schematic of the monitoring system circuitry. TAv~ = Is POSITI[]N ~. 1 . and presently a D C panel voltmeter is being used. Beeswax is used as couplant between the draft tube liner and the transducer. DERAKHSHAN ET AL. 2). The output signal of the amplifier is then fed to a RMS circuitry built around the AD637JD TRMS converter chip whose averaging time duration T (as shown in Eq.TVA Engineering Laboratory Cavitation Jet Erosion Test Setup. can be changed by changing the averaging capacitor.7~K ~--I2. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. T4VE = 64n~ FIG. =_T_ Fft~. . Copyright by ASTM Int'l (all rights reserved). 1. 2 . The circuitis placed within a metal box which shields the sensitive circuitry from electromagnetic and high frequency radio disturbances. ON MONITORING OF HYDROTURBINES 307 FIG. The D C output of the system can be displayed in many ways.ITION I. .s or 1 second. Figure 2 shows a schematic of the monitoring system circuitry.

ooooo|Er O.Response of the NBS Standard Reference Material displace- ment transducer and the pinducer to a seismic surface pulse.2t VOLT~. 3 and 4. RLOGXO O~ -20" ' * y. -60. lmH VOLTSl "5"S!000e --0. -80.. 4 . Pinducer The pinducer was compared using the Seismic Surface Pulse method [3] against a NBS Standard Reference Material(SRM) AE displacement transducer.0000 SECONDS 0. and Fig. Figure 3 shows a typical response of the SRM and the pinducer to the same capillary glass break input. using single bubbles generated in water by single sparks. 308 ACOUSTICEMISSION EXPERIMENTAL STUDIES Experiments were performed to evaluate the nature of the pressure pulse amplitude of a bubble collapse.) PZN O. .o -0-2t -o.0O004 (. Such issues as the change in amplitude and frequency content with distance and with passing through steel pipes are addressed.00004 0.O0008 O. Copyright by ASTM Int'l (all rights reserved).000.Frequency spectrum of the pinducer.oo~a o. However the usefulness of the transducer as an Acoustic Emission sensor had to be verified first. 3 . 4 shows the frequency spectrum of the pinducer. Suitability of the pinducer as an AE transducer is self evident from Figs. Also investigated is the transmission of the pressure pulse in a cloud of cavitation bubbles. No further reproductions authorized.2 FIG. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. zk6 2k6 HERTZ FIG.

Through Steel Pipe Transmission A number of different thickness.) steel pipes with 101. From the above mentioned experiment it was found that the pressure pulse of a bubble collapse did not follow the acoustic laws for a spherical point source.23 mm (1/4~ 3/8. At each location the pressure pulse signature was captured using a Data Precision Data 6000 digital data acquisition and analyzing system at a sampling rate of 300 ns. ON MONITORING OF HYDROTURBINES 309 Pressure Pulse In Water A spark generating system was used. AI.0 0.75 x 1 meter to a depth of about 152.5" bubblecollapse spark VOLTS 0. The electrodes were lowered into a plastic tank approximately 1 x 0. respectively. Copyright by ASTM Int'l (all rights reserved). The relative energy of each captured signature was then found by first squaring and then integrating it (Eq. and 7/8 in. In Fig.53. 12. was placed at different distances from the electrode gap. 004 ~ECONOS FIG. which consequently produces a vapor bubble due to point vaporization of the water between the electrodes. Figure 6 shows a sample pressure pulse and its energy waveform as calculated using the above procedure.5 to 1 mm) between two tungsten electrodes using a solenoid relay switch. Figure 7 is the log-log plot of the relative energy of the pressure pulses against the distance from the electrode gap. The pinducer was placed against the outside wall while the electrodes were positioned.A typical pressure waveform from a spark generated bubble collapse. 5 . No further reproductions authorized. 4 /~F capacitor to about 2500 volts using a DC power supply. The discharge causes a spark across the gap. 5 the rectangular pulse and the pulse immediately following it are the signature of the spark jump across the gap. see Fig. However more experimental data are needed (larger range of distances) to he able to show a governing law for cavitation energy attenuation.4 mm and the pinducer. 1/2.0 -0. 0. and/or the pressure front may not yet be fully developed. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.60 mm (4 in. The reason for such a displayed form was not investigated since the spark signature was not of any immediate concern. After a radius of 19 mm the energy is seen to decrease at a rate of 3. The spark is produced by first charging a 4000-V. high speed filming (3000 frames/sec) was used. 9. opposite to the pinducer inside the pipe.5. 5.) inside diameter were placed in the tank filled with water. To verify the presence of the bubble and that the second pulse is due to the bubble collapse. The next two pulses are initial and rebound bubble collapse.70. The acoustic law states that energy attenuates inversely with square of distance from the source. covered with petroleum jelly. DERAKHSHAN ET AL. . The capacitor is then discharged through the gap (approximately 0.5 0. The lower energy level for radii less than 19.35. and 22. 1). 002 0. This technique has been previously used by references [4-6]. There is also the indication that the rate of energy decrease to be even higher for radii larger than 60 ram.0 mm is believed to be due t o the disturbance of the bubble growth by the transducer. 6.

95 kHz peak is very close to the longitudinal natural frequency across the wall thickness..A sample pressure pulse and its calculated energy.0036 [..eSE-6 I -0. ... 0. Figure 8 shows a sample waveform. The 279.80 mm (2 in. 7 .. ..2S I S~Y - U..18 mm (1/8 in. 310 ACOUSTICEMISSION LO. . .it~ting Jet Erosion Laboratory tests were conducted at Tennessee Valley Authority (TVA) Cavitating Jet Erosion Test Facility to provide a comparison of the AE signals from a test specimen Copyright by ASTM Int'l (all rights reserved).0038 SECONDS 0...0001~ . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. 0.. "o~ > "3'51 " ~ .53 mm thick pipe. Distance on log axes (through water transmission). . For each pipe two sets of data were captured. No further reproductions authorized.. 6 .7~ O. .0040 FIG.75 ~f -s 0.=ONOS .. . .) from the inside wall..O001V.0036 0. one set with the electrode gap at about 3. . Distonce (ram) FIG. for the 9. 0..) from the inside wall and a second set with the gap at about 50.0640 EO.Plot of energy vs. Cav... as sensed by the pinducer with the electrode gap at center of the pipe and its frequency content.0o38 S. .

2 VOLTS -09 -0.P46 -20 VV~A~L. In the case of the jet erosion tests. Theoretically this frequency is calculated to be approximately 270 kHz as compared to the 279. The setup included two 300-500 kHz A E T A E transducers whose signals were amplified. 1.4E6 (b) FIG. no relation with distance has yet been established [7]. The pressure pulse is short in duration and strong enough to pass through such materials as steel. The signals were then recorded using a personal computer equipped with an 8 bit A / D board. with increasing pressure and cavitation. Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement.6 -0. At each pressure. There is the indication that the wall thickness can be measured from the frequency content of the pulse as sensed by the pinducer on the outside wall of the pipes.'~AA. ON MONITORINGOF HYDROTURBINES 311 PI46 0. J~A~AA 9 I'" -v 1 .0002 (.. Copyright by ASTM Int'l (all rights reserved).0E6 1.14% to 6.5 M P a (500 psi) down to 3. the RMS output from the tank wall mounted transducer decreased (almost linearly). 9.0002 0. Fig.95 kHz found experimentally. DISCUSSION OF RESULTS It was verified that the cavitation bubble collapse can be used as an A E source. the signal from each transducer was-recorded at a sampling frequency of 10 MHz and their RMS value was noted. see Fig. Also a RMS circuitry with a 1 second averaging time was used to measure the related energy of the cavitation. 8b. DERAKHSHAN ET AL. This is a striking agreement. The tests were conducted with an initial jet pressure of 51. There are frequency peaks coinciding with the longitudinal natural frequency of the walls as calculated using theory 9 This frequency for the 9 .0000SECONDs 0.) thick pipe is noted on Fig. No further reproductions authorized. 8 . .Through steel pipe sample waveform and its frequency content 9 with a transducer located directly on the test specimen and another transducer located on the outside wall of the test chamber. It has not been possible to draw any reliable conclusion on the change in the frequency content of the pressure pulse as it moves through water.53 m m (3/ 8 in.5 M P a (500 psi).6 I 0. 0.4E6 HERTZ 1. Although there are some differences in the frequency contents.8 M P a (7500 psi) in decrements of 3.0E6 0.63% for the four different thickness pipes [7].) ~L. The percentage difference in the calculated and measured frequencies ranged from 0. filtered and displayed on a digital oscilloscope. after allowing for stabilization.

..@ . I . ... I . Fig. I .0 0 1.D C Converter Chip ?i ~ ...B 2. there is only a need for a l l 0 . The system -3 dB points fall around 9 kHz and 7 MHz.D C ConvePter Chip Time Conetint . it was tested for longevity... 10. . a s h a r p 1. . I . . .8 RDG37 R M S . I . On the other h a n d the RMS o u t p u t from the specimen m o u n t e d transducer.0 v 3. Furthermore.... .. These tests m e t w i t h success and since a DC power supply is included in the circuitry.. 5. This result again shows t h a t cavitation signals can be used in measuring wall thickness. ::~ .t i m e mode.. I . pressure. T E S T I N G AND A P P L I C A T I O N OF T H E M O N I T O R I N G S Y S T E M Since the m o n i t o r i n g system is to operate in r e a l ... . . 9 . 10 . | e e c 1888 ~see 3@@@ 4888 ~e88 see8 7@8@ esee Pressure (psi) FIG. ... I ..t o .t o . . I . 312 ACOUSTICEMISSION There was a visible b u b b l e cloud t h a t increased with pressure. . . w i t h no Copyright by ASTM Int'l (all rights reserved).14 MHz [8]. 93 8 ~ . I .. Tank-Mounted Trtnsducer ~h~-''~'~''* ~ RDG37 R M S .2 MHz frequency peak was present in the frequency s p e c t r u m of t h e t a n k wall m o u n t e d t r a n s d u c e r at all pressure levels. . This result supports the feasibility of using the T R M S chip to provide an indication of cavitation level when the transducer is m o u n t e d on the cavitating piece..V AC power source. T h e frequency response of the amplifier-filter c o m b i n a t i o n was verified as can be seen in Figs. ? 2s[ i I 18 L. . This bubble cloud appears to a t t e n u a t e the signal before it reaches the transducer. No further reproductions authorized. 8 18~8 2888 38@8 4888 58@8 GSB@ 7 8 8 8 B@88 Pressure (psi) FIG.. .. ! ... I . 4. . and reliability of the circuitry.RMS o u t p u t of the specimen m o u n t e d t r a n s d u c e r vs. ! . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. .. .. .RMS o u t p u t of the wall m o u n t e d t r a n s d u c e r vs. . I .. 11 and 12. . This frequency corresponds to the calculated wall resonance frequency of 1. .. .... . pres- sure.. . . . increased w i t h pressure and cavitation. I . .

. .i i 9 I ~ ii i i |it! i i iiilii ii It i! i i i ~ I i i i! i -" i ...~ ilil I l i i tlii '~ i/ . operated by Tennessee Valley Authority. one is for decreasing gate opening from 100% to 10% and the other for increasing gate opening from 30~ to 100%. ~ i l i " ' :t: ' ! i" i ! t i i i l 9 i . "4-' "+'~-':i :. the system was then applied to the hydrosys- tem.i..Phase response of the amplifier-filter combination.l'. Figure 14 shows the RMS output of the cavitation monitoring system against the percentage wicket gate opening for unit ~ 3 of Douglas and that of unit # 1 of Chickamauga hydroplants.. and the rate of the deviation to the exact value was within plus or minus 0. !.~ ..I ! i i lid ! 1 i iliii ill! i i!iil o .i'J. triangular. DERAKHSHAN ET AL. . i ' it: ! '9 i i iit. i i liil I 9 I :. Tests have been conducted at Douglas and Chickamauga hydroplants.. i. From the two different RMS voltage sets (notice the maximum RMS values)..!i!l i ~i "iii : 'fi i i 71i i t I'tl~ lli i ii:i.! il " / i..li.i 9 ~i' . ii i i iii . . Mon Dec 26 18:40:22 EST 2011 Downloaded/printed by (PDVSA Los Teques) pursuant to License Agreement. i-o i i i . ... 11 .i i t i'~-'i't !! ill . 9 ! l.. No further reproductions authorized.. It would require long term statistical study of the cavitation of each hydroturbine in order to be able to draw a cavitation performance function with Copyright by ASTM Int'l (all rights reserved). After the above tests were performed. i I !!!!!!'" . iI. it can be seen that there can be differences from hour to hour at any particular hydroturbine. with Francis and Kaplan turbines respectively. . ii ' ..0 m and a period of two hours had elapsed when the increas- ing gate opening data were taken. 12 . 'lq z i i i t i t.. i ii iii II! i . ON MONITORING OF HYDROTURBINES 313 significant phase shift from 100 kHz to 3 MHz.u. . i iii~ . lii"i l i lilqi....'ii i lI i i i t ~:!' ! i! lii!i i ! -~i..ii--"r i .. Since each unit had a dif- ferent cavitation intensity level the ordinate has been normalized as a percentage of the maximum RMS value for each curve. . i i i ill.. In case of Douglas Hydroplant two sets of data are presented.. 9i iiii . "i ~'" ]: i i 1: i li 4fl o ~ i!'! i ill i I ' o . IIIIT " ill . ..i. t't. .. : .i i " . : i!l.:::''i.~ { :ifil .i .t .i'ITl.~s output of the system was checked against known value sinusoidal.. .m. i tlI iii~ ~ .~ t.i lmii i li i ' i I !i :Isili !i{i i i 1i iqli ! :ill --4-4-4.. and square waveforms. i i ill i ild. i ' i i ilil ' [ i " il i i i 'iiilll~ i "iiii! . In this case the head and tail water levels had increased by about 1.: .0029 . Also the Vr. ii 'Nil . "~ ~.I : i "i -"i i.Frequency response of the amplifier-filter combination. . . . : i.iii"