Acoustic Emission - Standards and Technology Update

S T P 1353
Acoustic Emission: Standards and Technology Update
Sotirios J. Vahaviolos, editor
ASTM Stock #: STP1353
ASTM 100 Barr Harbor Drive West Conshohocken, PA 19428-2959 Printed in the U.S.A.
Library of Congress Cataloging-in-Publication Data

Acoustic emission : standards and technology update / Sotirios J. Vahaviolos, editor. p. c m . - (STP : 1353) Includes bibliographical references. "ASTM Stock #: STP1353." ISBN 0-8031-2498-8 1. Acoustic emission testing. I. Vahaviolos, Sotirios J. II. Series: ASTM special technical publication : 1353. TA418.84 .A263 1999 620.1 '27--dc21
99-38512 CIP
Copyright 9 1999 AMERICAN SOCIETY FOR TESTING AND MATERIALS, West Conshohocken, PA. All rights reserved. This material may not be reproduced or copied, in whole or in part, in any printed, mechanical, electronic, film, or other distribution and storage media, without the written consent of the publisher.
Photocopy Rights Authorization to photocopy Items for internal, personal, or educational classroom use, or the Internal, personal, or educational classroom use of specific clients, Is granted by the American Society for Testing and Materials (ASTM) provided that the appropriate fee is paid to the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923; Tel: 508-750-8400; online: http..//www.copyrig ht.com/.
Peer Review Policy

Each paper published in this volume was evaluated by two peer reviewers and at least one editor. The authors addressed all of the reviewers' comments to the satisfaction of both the technical editor(s) and the ASTM Committee on Publications. To make technical information available as quickly as possible, the peer-reviewed papers in this publication were prepared "camera-ready" as submitted by the authors. 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 the peer reviewers. In keeping with long standing publication practices, ASTM maintains the anonymity of the peer reviewers. The ASTM Committee on Publications acknowledges with appreciation their dedication and contribution of time and effort on behalf of ASTM
Printed in Philadelphia,PA October 1999
Foreword
This publication, Acoustic Emission: Standards and Technology Update, contains papers presented at the symposium of the same name held in Plantation, Florida, on 22-23 January 1998. The symposium was sponsored by ASTM Committee E7 on Nondestructive Testing. The symposium chairman was Sotirios J. Vahaviolos, Physical Acoustics Corporation.
Contents
Overview
vii
AE SOURCES: CHARACTERIZATION
Use of Acoustic Emission to Characterize Focal and Diffuse Microdamage in BonemR. M. RAJACHAR, D. L. CHOW, C. E. CURTIS, N. A. WEISSMAN, AND D. H. KOHN
CONCRETE APPLICATIONS
A Proposed Standard for Evaluating Structural Integrity of Reinforced Concrete Beams by Acoustic E m i s s l o n - - s , YUYAMA,T. OKAMOTO,M. SHIGEISH],M. OHTSU,
AND T. KISI-H
25
On the Necessity of a New Standard for the Acoustic Emission Characterization of Concrete and Reinforced Concrete Structures--E. o. NESWJSKI AE Evaluation of Fatigue Damage in Traffic Signal Poles--H. R. HAMILTON,HI,
T. J. FOWLER, AND J. A. PUCKETI"
41
50
INTmGRrrY AND LEAK
DETECTION/LOCATION METHODS
The Development of Acoustic Emission for Leak Detection and Location in LiqnidFilled, Buried Pipelines--R. ~ MILLER, A. A. POLLOCK, P. FINKEL, D. J. WATI'S,
J. M. CARLYLE, A. N. TAFURI, AND J. J. y1~7.71 JR.
67
Acoustic Emission and Ultrasonic Testing for Mechanical Integrity--s. J. TERNOWCHEK,

T. J. GANDY, M. V. CALVA, AND T. S. PATIT~SON
79
AE SENSORS, STANDAm)S, AND QUANTITATIVEAE
Calibration of Acoustic Emission Transducers by a Reciprocity Method--H. HATANO
93
DIVERSE INDUS1RIAL APPLICATIONS
Acoustic Emission Applied to Detect Workpiece Burn During G r i n d i n g - P. R. DE AGUIAR, P. WILLETI', AND J. WEBSTER
107
Analysis of Fracture Scale and Material Quality Monitoring with the Help of Acoustic Emission Measurementsms. A. NIKULIN,M. A. SHTREMEL,V. G. KnANZH~,
E. Y. KURIANOVA, AND A. P. MARKELOV
125
Characterization of Micro and Macro Cracks in Rocks by Acoustic Emission-G. M. NAGARAJA RAO, C. R. L. MURTHY, AND N. M. RAJU
141 156
Prediction of Slope Failure Based on AE Activity--T. SHIOTANIANDM. OHTSU

A E SOURCES: RESEARCH T o P I c S
Identification of AE Sources by Using SiGMA-2D Moment Tensor Analysism

M. SHIGEISHI AND M. OHTSU
175
TRANSPORTATION APPLICATIONS, STANDARDS, AND METHODOLOGy Practical AE
Methodology
for Use on AircraftmJ.
M. CARLYLE, H. L. BODINE, S. S. HFa'qLEY,
R. L. DAWES, R. DEMESKI, AND E. v. K. HILL
191
COMPRESSED GAS APPLICATIONS AND STANDARDS
Periodic AE Re-Tests of Seamless Steel Gas Cylindersmp. R. BLACKBURN Field Data on Testing of Natural Gas Vehicle (NGV) Containers Using Proposed ASTM Standard Test Method for Examination of Gas-Filled Filament-Wound Pressure Vessels Using Acoustic Emission (ASTM E070403-95/1)-R. D. FULTINEER, JR. AND J. R. MITCHELL
209
224
Acoustic Emission Testing of Steel-Lined FRP Hoop-Wrapped NGV Cylinders-A. AKHTAR AND D. KUNG
236 257 259
Author Index Subject Index
Overview
Acoustic Emission (AE) has been commercially available for more than thirty (30) years. Has any progress been made? The purpose of the Symposium held in January 1998 in Plantation, Florida was to discuss the evolution of the technology of AE over the years in instrumentation, applications, standards and codes and its overall worldwide acceptance. Authors have made comparisons between AE and other Nondestructive Testing (NDT) technologies as to their suitability in solving practical industrial problems worldwide. As the newcomer in the Nondestructive Evaluation (NDE) industry, AE was first tried on applications where other NDT technologies had previously failed or was used where wild financial cost savings were promised. The issue of suitability of AE for an application was never considered until the very late 70's and early 80's, when a new breed of industrial and university researchers entered the field in USA, Europe and Japan. AE "noise counting" was replaced with basic work on source characterization, wave propagation, mode conversion, the study of the inverse problem using a number of Green's functions, pattern recognition and, most importantly, they considered AE as a science, using all available tools at their disposal. While the university academics worked hard to identify certain AE waveform features with source and failure mechanisms, a number of industrial researchers explored a myriad of "Pseudo-sources" of AE and their statistical nature. Instead of absolute one-on-one correlations and exact location of defects, practitioners developed zonal location and data bases based on case studies that enabled them to relate AE to fracture mechanics, corrosion phenomena, and overall part integrity assessment, especially in composite structures first and then in pressurized systems and individual components. The introduction of artificial intelligence, coupled with existent data bases, led to the development of ready-to-use knowledge-based systems based on very complex structures that are found in power utilities, refineries, chemical plants, complex pipelines, wind tunnels, aircraft structures, etc. The hard work of the late 70's and early 80's by CARP (Committee on AE for Reinforced Plastics) and the wide application of AE in testing of Fiberglass (FRP/GRP) vessels and pipes rejuvenated the technology! Eventually they became ASTM Standards now widely in use. The well-publicized early failures of AE in several metal vessels tests, especially in Europe by INEXPERIENCED personnel, were now reconsidered. Unknown to most AE Researchers/ Practitioners a behind the scenes branch of CARP known as CAM (Committee for Acoustic Emission for Metal) start looking carefully utilizing vast experience in Fracture Mechanics, Civil Engineering, NDT and, most importantly, vessel construction maintenance and use, realized early on that the same inexperience that prevented the use of AE in FRP in the early 70's has prevented users to do Metal Vessel Testing by AE. With the help of t h e ' 'core members" of CARP, metal vessel testing was reconsidered, especially after the successes of MONPAC ~ (a commercially available knowledge-based expert system that formed the basis of acceptance of AE by American Society of Mechanical Engineers (ASME) and Department of Transportation (DOT) and, thus, gave credence to the newcomer NDE technology). In addition, the more than ten AE ASTM Standards and AE's acceptance by American Society for Nondestructive Testing (ASNT) as another major NDT technique and the establishment of Level III in AE were major steps forward for the technology worldwide.
vii
viii
ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY
In this Symposium basic important work is being presented that constitutes the basis for Natural Gas Vehicle (NGV) Cylinder Testing with AE, no matter how controversially some people might view their work. When properly applied, AE can save NGV assets for customers as the ASTM FRP vessel has done for the past 10-plus years. It is interesting to note that infrastructure and slope stability applications worldwide and especially in Japan are now to the point of standardization of existing working procedures. We were very much encouraged by the continuing success of the Reciprocity Method for Calibrating AE Sensors and hope that it eventually will become another ASTM Standard. As for the other applications, I can only comment on their existing uniqueness from micro damage in bones to burning of grinding tools in high speed manufacturing. We hope this publication will prove interesting to a wide spectrum of readers, especailly those who look for new AE Standards and are interested to explore the future directions for the application of the Acoustic Emission Technology.
Sotirios J. Vahavidos, Ph.D.
Physical Acoustics Corporation Princeton Junction,NJ 08550 SymposiumChairmanand Editor
AE Sources: Characterization
Rupak M. Rajachar, 1 Dann L. Chow, l Christopher E. Curtis, 2 Neil A. Weissman, 2 and David H. Kohn 3
USE O F A C O U S T I C E M I S S I O N T O C H A R A C T E R I Z E F O C A L AND D I F F U S E M I C R O D A M A G E IN BONE
REFERENCE: Rajachar, R. M., Chow, D. L., Curtis, C. E., Weissman, N. A., and Kolm, D. H., "Use of Acoustic Emission to Characterize Focal and Diffuse Microdamage in Bone," Acoustic Emission: Standards and Technology Update, ASTM STP 1353, S. J. Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999. A B S T R A C T : Fatigue of cortical bone results in the initiation, accumulation, and propagation of microdamage. AE techniques were adopted to monitor damage generated during ex-vivo tension-tension fatigue testing of cortical bone. The primary objectives were to determine the sensitivity of AE in detecting microdamage in cortical bone and to elucidate mechanisms guiding the onset of microdamage. Fatigue cycle data and histological data show that AE techniques are more sensitive than modulus reduction techniques in detecting incipient damage in cortical bone. Confocal microscopy revealed the ability of AE to detect crack lengths and damage zone dimensions as small as 25 gm. Furthermore, measured signal parameters such as AE events, event amplitude, duration, and energy suggest that AE techniques can detect and distinguish microdamage mechanisms spatially and temporally in bone. As fatigue processes continue, AE increases in terms of number of events, event intensities and spatial distribution. Diffuse damage appears to be a precursor to the development of linear microcracks. The spatial and temporal sequence of AE events enables differentiation between linear microcracks and more diffuse damage.
K E Y W O R D S : acoustic emission, bone, microdamage, confocal microscopy
l Graduate Student, Department of Biomedical Engineering, College of Engineering, University of Michigan, Ann Arbor, MI 48109-2125. 2 Undergraduate Student, Department of Biologic and Materials Sciences, School of Dentistry, University of Michigan, Ann Arbor, MI 48109-1078. 3 Associate Professor, Departments of Biologic and Materials Sciences, School of Dentistry and Biomedical Engineering, College of Engineering, University of Michigan, Ann Arbor, MI 48109-1078.
Copyright9
by ASTM International
3 www.astm.org
Age-related skeletal fragility is a significant medical and economic problem. In the United States, an estimated 250,000 hip fractures occur each year among those age 50 and over [t]. The cost involved in the treatment of these fractures is believed to be over 7 billion dollars [2,3]. In younger people, who have a greater physiologic capacity to withstand the functional demands placed on the musculosketetal system, stress fractures and impact/trauma-related fractures are also of significance. However, with age and pathologic conditions, the compensatory mechanisms needed to maintain the mechanical stability of skeletal tissue become significantly more impaired [3]. As a result, diffuse distributions of sub-threshold microcracks, which have been observed even in normal bone, increase exponentially in number with age. These cracks range in length up to 300 ~tm and may ultimately contribute to age-related property degradation and fatigue fracture in skeletally mature cortical bone (e.g. the dense bone of the mid-femur, or outer shell of the proximal femur and spine) and trabecular bone (e.g. the porous, spongy bone inside the cortical shell) [4]. Consequently, the processes guiding the location, orientation, size, and accumulation of microdamage are important in assessing the competence of skeletal tissues and in developing a basic understanding of structurefunction relationships. Cortical bone can be modeled as a fiber-reinforced composite [5]. Osteons or Haversian canals are cylindrical layers of bone around blood vessels. The inter-osteonal bone, which is arranged in lamellae, is called interstitial bone. Thus, osteons in bone, which are primarily oriented in a longitudinal direction, are analogous to unidirectional fibers in a composite. As in many engineering composites, a comparatively high strength/low toughness interface exists between each osteon or "fiber" and its surrounding interstitial bone or "matrix". This region is referred to as the cement line and is thought to be a site of relatively easy crack nucleation [6]. Also of importance to the overall structure of bone are embedded cellular compartments and their connecting network, lacunae and canaliculi, respectively [6]. These regions represent heterogeneous sites of porosity, akin to processing defects in man-made structural materials. Accordingly, these pores may serve as additional sites of crack nucleation and accumulation [7,8]. Bone is therefore a hierarchical composite structure subject to internal microdamage. Many of the difficulties which exist in characterizing the mechanics of structural composites also exist with bone. The long-term objectives of this research program are to model bone as a composite material and: 1) determine the effects of mechanical history on microdamage initiation and growth; 2) provide insight into mechanisms of damage initiation and accumulation; and 3) determine how damage phenomena are modulated by changes in tissue hierarchy. Previous investigations have attempted to use AE to monitor integrity of bone invitro [9-12], and as a non-invasive diagnostic in-vivo [13-15]. These studies made inferences that the AE detected was due to specific failure mechanism(s), but no direct correlations between AE and damage were established. Using more rigorous analyses, we have shown that AE techniques can be effectively used to distinguish mechanisms of crack nucleation, slow crack propagation, and rapid crack propagation, spatially and temporally, in biomaterials, microstructured materials and inhomogeneous materials [16,17]. The critical resolution of AE for detecting crack nucleation was shown to be on the order of 10 ~tm [16]. Because of the complex inhomogeneous nature of cortical bone,
RAJACHAR ET AL, ON MtCRODAMAGE IN BONE
application of these more detailed AE techniques may provide similar insight into the mechanical processes involved in the nucleation and growth of damage in bone. Moreover, the sensitive and non-destructive nature of AE testing may allow multiple crack sites to be distinguished spatially and temporally. In support of our long-term objectives, the specific aims of this project were to: 1) detect and characterize incipient microdamage in cortical bone via AE, 2) verify and quantify the microdamage histologically, 3) compare the sensitivity of AE and modulus reduction (AE) techniques, 4) compare microdmnage in bone of different initial stress intensities, and 5) associate AE signals with microstructural failure mechanisms.
Materials and Methods
Cortical Bone Specimens
The flow chart in Figure 1 provides an overview of the experimental design used in preparing and testing cortical bone ex-vivo. Cortical bone specimens were prepared from mature bovine femoral and tibial central diaphyseal sections. Each diaphysis was sectioned on a band saw into paratlelepipeds, such that the longitudinal axis of each parallelepiped was aligned with the long axis of the bone. These rough-cuts were then machined into smooth parallelepiped blanks (L = 120 mm, W = 12 mm, T = 4.5 mm) and gage sections were machined using a precision milling machine. Buffered saline irrigation was used during all machining steps to avoid heating the bone and to maintain tissue saturation. Two gage section geometries were created: V-notched specimens (p = 200 ktm, K t = 2.5), which provided a localized region of strain, and C-notched specimens (K t = 1), which provided a distributed strain region. Specimens not tested immediately after machining were wrapped in moist towels and stored at -65~
Fatigue Loading
Fatigue loading regimes were imposed using a servohydraulic mechanical testing machine and FLAPS mechanical testing computer interface program (Instron Corp., Canton, MA). During loading, specimens were kept moist using an ambient temperature buffered saline gravity drip, since drying results in an increase in the elastic modulus of bone. Initial elastic moduli (E0) were determined by subjecting the specimens to successive uniaxial tensile ramp loading and unloading cycles in the elastic range until a steady state value was reached. Peak stresses were approximately one-quarter of the yield stress for bovine cortical bone (~y - 140 MPa), and a steady state modulus value was typically reached in 5 cycles. The initial elastic modulus was then used to define a fatigue loading regime [18]. Fatigue loading was performed in uniaxial tension under load control. Specimens were loaded parallel to the longitudinal axis of the bone, following a sinusoidal waveform, over an effective strain range (AEeff = A(Y]E0) of 0 to 3000 ge, at a frequency of 1 Hz. The maximum strain was chosen such that it was in the upper range of physiologic strain experienced by cortical bone during normal loading [19].
I
I
Mature Bovine Bone(femur) DiaphysealLongitudinalCut End Milled Paralleleplped I
I
I
[
I Fatigue to
v-Notch I localizeddamagezone I I
C-Notch , diffusedamagezone I I
AEo.set I
I I
I
I 1%~.~
Fatigueto
FatigueLoadingConditions 3000petrain,1 Hzsinewave AE ParameteisMeasured events,Ioc.,ampL,(mergy,etc. 10 mmnotchedsection Histology bulkstainingandembedding 150 ~ sedalsections Confocal Microscopy grideachsectionandscan crackIoc.,density,INze,etc. 3-D reconstruction
I tf
I
I
Unloaded I control
RELATEAE TO FAILUREMECHANISMS
FIGURE
1 -
Flow chart of experimental design.
RAJACHAR ET AL. ON MICRODAMAGE IN BONE
Experimental Design
For each gage section geometry, a 30-specimen experimental matrix was used to study fatigue-related microdamage: 10 specimens were loaded in fatigue until the onset of AE, 10 were loaded in fatigue until there was a measured 1% modulus reduction (AE), and 10 served as unloaded histological controls. Each specimen was randomly assigned to one of these three groups. Modulus measures were made based upon real-time mean strain values, measured with an extensometer.
Acoustic Emission Analysis

AE was recorded using a planar array of four Physical Acoustics Corp. (PAC, Princeton, N J) nano-30 piezoelectric transducers having a broad-band frequency range of 125-750 kHz. Two sensors were placed on the shoulders of the 120 mm x 12 mm face, for longitudinal location, and the other two sensors were on opposite sides of the width of the specimen, above and below the notched region, for transverse location. The sensors were coupled to the specimens with an acoustic couplant and fixed in place using waterresistant surgical grade adhesive tape. AE data were collected, stored and analyzed with PAC LOCAN-320 data acquisition and analysis software. The pertinent operating parameters were: variable gain/total gain = 42 dB/80 dB; peak definition time = 500 gtsec; hit definition time = 2 msec; dead time = 1 msec; sample time = 100 msec; threshold = 1 V. A threshold value of 42 dB was used to eliminate background noise produced by specimen irrigation. For the tests stopped at the onset of AE, the first AE signals above 42 dB simultaneously detected at all 4 sensors were taken to signify the onset of microdamage. The following AE parameters were recorded and analyzed: AE source location, number of AE events, and intensities of AE events. Event intensities are a collective term for event amplitude, counts per event, event duration, event energy counts and event rise time. Subsets of events were also created, based on event location, fatigue cycle number, and stress range at which events were generated. Subsets of event intensities generated within different ranges of location, fatigue cycle number and stress level were also analyzed. Subsets of events were then analyzed by evaluating location distribution histograms (LDH) and intensity distribution histograms (IDH) of events. LDH and IDH are general terms for the distribution of events and event intensities as functions of location, fatigue cycle number or stress. Initial analysis of the spectral components of waves was also carried out. Digital transient capture of waveforms was performed using an F4000 Fracture Wave Detector (Digital Wave Corp., Englewood, CO). A maximum digitization sampling rate of 12.5 MHz was used and 1024 points of digitized data were collected from each waveform. Characteristic extensional and flexural waves were generated by breaking a lead pencil at multiple sites on selected bone specimens. Actual AE waves generated during testing of bone were recorded and digitized, and mode shapes and dominant frequency contents were determined and compared to the waveforms generated by the pencil breaks [20,21].
Histological Analysis
Following fatigue loading, 10 mm blocks from the region near the dominant AE sources were cut using a diamond wafering blade, and stored at -65~ in buffered saline solution until histological processing. Specimens were stained en-bloc using a graded series of 1% basic fuchsin solutions in ethanol [22]. Basic fuchsin is a fluorescent stain that preferentially marks exposed external and internal defects in biological structures. Bulk staining with basic fuchsin prior to histological sectioning enables visualization of microdamage and differentiation between cracking due to mechanical factors and artifactual cracks induced during histological preparation. Following staining each specimen was embedded in poly-(methyl methacrylate) and serially sectioned (-150 p.m thick) parallel to the 4.5 mm thickness. Each serial section was analyzed using laser scanning confocal microscopy (LSCM) to assess damage at the notch tip and throughout the bulk of the specimen. Three-dimensional data on crack morphology were obtained by using a z-axis reconstruction [23]. A l-ram square grid system was imposed on each histological section. Each grid space was analyzed for number of cracks, crack length, crack density, and crack angle.
Results
In support of Specific Aims 1 and 2, we were able to detect and characterize AE generated during fatigue of cortical bone and validate, via histological comparison to unloaded controls, that the sources of this AE were fatigue-induced microdamage. An LDH of AE events generated during fatigue of a V-notched specimen is shown in Figure 2. AE events are localized at the notch-tip (location 0). Representative LSCM photos of the notch-tips of fatigue and control specimens are shown in Figure 3. The highly luminous regions at the notch-tip of the fatigue specimen represent damage nucleation. More well defined linear (Mode I) microcracks are observed ahead of the diffusely damaged region. Quantitative histology revealed crack lengths as small as 25 p.m.
,~ 2O
<
10
E z
0
-40
..... . . .
-30
m, .
-10
~1,. .nn,. nn, , , . .

0 10 20
t 30
tm I 40
-20
Location
FIGURE 2 - Location distribution histogram (LDH) of AE events generated in V-
notched cortical bone specimen fatigued until the onset of AE. Events are localized at the notch-tip (location 0), indicating damage nucleation in this region.
FIGURE 3 - Confocal micrographs of V-notched specimens: a) control (unloaded) specimen, and b) specimenfatigued until the onset of AE. In the fatigue specimen, damage accumulation is seen at the notch-tip. Diffuse damage is observed at the notchelp, with well-defined linear microcracks observed ahead of the diffuse damage zone. (Both I OOX)
10
The average number of fatigue cycles to the onset of damage as determined by the two techniques, for V- and C-notched specimens, is presented in Table 1 (Specific Aim 3). For both specimen groups, incipient damage was detected at a significantly lower number of fatigue cycles with AE than AE (p < 0.01, p < 0.02, respectively, via Student's t-tests). TABLE 1 - Number of cycles to the onset of fatigue as determined by acoustic emission (AE) and modulus reduction (AE) techniques. Technique AE AE AE AE Specimen Type V-Notched V-Notched C-Notched C-Notched Averase NumberofCycles 8894 5317 32548 25697 8094 4241 40544 22828
Figure 4 shows LDHs of AE events generated in V-notched specimens fatigued until the onset of AE and until a 1% AE. Although both tests reveal AE accumulation at the notch-tips, the greater number of events generated in the AE specimen is commensurate with the longer fatigue regime of this specimen. Evaluating IDHs (Fig. 5), shows that peak and maximum values of signal intensities generated in bone that was fatigued until a 1% AE are greater than those generated in bone fatigued until just the onset of AE. Average intensities were also greater in the AE specimens. Comparison of confocal micrographs from AE and AE specimens (Fig. 6) suggests the presence of diffuse microdamage followed by development of linear microcracks as fatigue processes continue. Diffuse damage is characterized by relatively larger zones of staining opacity and minimally resolvable crack dimensions. While many histological sections exhibited Mode I microcracks, several sections revealed the presence of linear microcracks, adjacent to regions of diffuse damage, which were oriented parallel to the direction of loading (Fig. 6c). In support of Specific Aims 4 and 5, we were able to compare microdamage in bone of different initial stress intensities and associate AE signals with microstructural failure mechanisms. The mechanisms, observed histologically, were characterized based upon differences in specific AE signal parameters, as shown in the LDH and IDH in Figures 7 and 8. Specimens with a lower initial stress concentration generated a greater number of events, a greater spatial distribution of events and more high intensity events.
11
4 - Location distribution histograms (LDH) of AE events generated in Vnotched specimens fatigued until the onset of AE and until a 1% modulus loss. The greater number of events generated in the AlE specimen is commensurate with the longer fatigue process in this specimen. FIGURE
12
FIGURE 5 - hztensity distribution histogl"ams (IDI1) o f A E events generated in Vnotched specimens fatigued until tile onset o f A E and m~til a 1% modulus loss. A E signal hm, nsities in the A E specimens are lower than those o f the AE group, since the tests were stopped at an earlier stage o f fittigue.
13
FIGURE
6 - Confocal micrographs of a) diffuse microdamage (I OOX), b) Mode I linear microcracks (I OOX), and c) combination of diffuse damage and Mode H linear microcracks (60X).
14
7 - Location distribution histograms (LDH) of AE events generated in Vnotched and C-notched specimens fatigued until the onset of AE. The C-notched specimens exhibited a greater spatial distribution of AE events. FIGURE
15
FIGURE 8 - Intensity distribution histograms (IDH) of AE events generated in V-notched and C-notched specimens fatigued until the onset of AE. AE signal intensities are different in the two specimens, indicating different damage mechanisms.
16
Discussion
The objectives of this research were to determine the sensitivity of AE in detecting microdamage in cortical bone and to use the early warning capabilities of AE to help elucidate the underlying mechanisms guiding the nucleation and growth of microdamage in bone. LDHs (Fig. 2) and histology (Fig. 3) provided evidence that damage in cortical bone can be detected and characterized using AE. This is apparent by the prominent AE that was detected and microcracking that was observed at the notch-tips. Quantitative histology revealed that the AE was first generated at crack lengths and damage zone dimensions as small as 25 ~m. The fatigue data shows that AE techniques are more sensitive than modulus reduction techniques for detecting damage in cortical bone for each specimen geometry (Table 1). A greater number of AE events was also detected in specimens which were allowed to fatigue until a 1% AE was achieved (Fig. 3). The IDHs (Fig. 4) reflect the greater number of events, greater peak amplitude and greater number of high intensity events for the AE specimens. The damage inferred by the AE and AE techniques was subsequently verified histologically. All of these data support that conclusion that AE is more sensitive than modulus reduction techniques in discriminating incipient damage in bone. Currently, modulus reduction techniques represent the accepted real-time means of detecting damage in bone [24]. Since the sensitivity of AE is greater than that of AE, it is expected that there should also be a difference in the type and extent of microstmctural damage observed in the two groups. It is therefore hypothesized that different mechanisms of damage initiation and propagation in bone will express distinguishable AE signal profiles. The IDHs suggest these differences. Based on the relative amounts of diffuse and linear damage observed in control, AE and AE specimens, there, qualitatively, appears to be a progression of diffuse damage accumulation which ultimately coalesces into more distinct linear microcracks. There appears to be a greater damage density (i.e. # of cracks or # of diffusely stained regions/unit area) in AE samples compared to AE samples, and a threshold in damage density above which AE is generated. This threshold corresponds to an accumulation of plastic strain energy sufficient to generate detectable AE. Coupled with histological data, AE analysis suggests an initial diffuse matrix damage mechanism in the fatigue failure of cortical bone, analogous to microvoiding and craze formation in fiber reinforced composites. Further fatigue results in the formation of linear microcracks. LDH data allows the differentiation of diffuse matrix damage away from the notch-tip and damage at the primary fracture process zone. Linear microcracks were most often associated with cement lines, an observation made by others as well [7,25,26]. The specific nature of the interactions between microcracks and cement lines is unclear. Microcracks were observed to propagate both along and across cement lines. Combined with the fact that linear microcracks were observed to be oriented in both Mode I and Mode II directions, the overlapping of AE signal parameters in the different test groups is likely due to the combination of ultrastructural failure mechanisms observed.
17
The observation of longitudinal and shear oriented cracks motivated an analysis of AE waveforms. Based on the characteristic extensional and flexural waves generated during pencil breaks (Fig. 9), analysis of spectral components of AE signals generated during fatigue of bone revealed the following. At the notch-tip, longitudinal waves, with a dominant frequency range of 250-400 kHz, were primarily detected, whereas in the bone matrix, flexural waves, with a dominant frequency range of 100-180 kHz, were primarily detected. These characteristics are consistent with those of fiber reinforced plastics [27]. While it is acknowledged that modal analyses are limited to plate specimens and may not be applicable to in-vivo diagnosis of a whole femur, the modes of microfailure observed in this study are consistent with those observed in-vivo. Idealized testing on plate-specimens, which are amenable to modal analysis, may yield significant insight into damage mechanisms. Whether diffuse damage and modes I and II linear microcracking processes can be linked to specific AE signals is a continuing focus of this research. It may ultimately be possible to link damage location with particular initiation and propagation mechanisms, which may provide necessary information to the detection and prevention of catastrophic damage associated with many degenerative orthopaedic conditions. If a continuum can be established between diffuse damage and linear microcrack initiation and propagation mechanisms, it may also be possible to relate mechanical and biological components affecting the overall structure of cortical bone under different loading conditions. Thus, future work will not only attempt to quantify the relationship of the strain history of cortical bone to damage initiation and propagation mechanisms, but also to define local biologic conditions favoring damage mechanisms.
18
s) extensional w a v e
4.0
.~
E
0 > 0 n,
3.0t
20
1.0 0.0-
-1.0 -2.0 -3.0 -4.0 O.OE+O0 Time (seconds)
1.6E-04
b) flexural w a v e
6.0 4.0 2.0

a,.
E .<
0 > 0
0.0
I!
-2.04.0~.0 O.OE+O0
I,
II
II
I I,
j
1.6E-04
Time (seconds)
n,
FIGURE 9 - AE waveforms from lead pencil breaks." a) extensional wave, dominant frequencies- 300-500kHz, b) flexural wave, dominant frequencies ~ 100-200 kHz.
19
Summary and Conclusions

AE techniques were employed in association with histological analysis to determine the effects of strain history on the initiation and accumulation of fatigue microdamage in cortical bone. Microstructural damage mechanisms were identified and related to AE parameters spatially and temporally. Specifically, the following conclusions are drawn: (1) (2) AE techniques are effective in detecting incipient microdamage generated during fatigue of cortical bone. Histological comparison between loaded and unloaded specimens validated the AE technique and revealed that AE can detect microcracks in bone as small as 25 ~tm. Multiple microcracks and diffuse regions of damage may also be detected. Incipient fatigue damage is detected at a significantly lower of fatigue cycles with AE techniques, as compared to modulus loss techniques and, as a result, less AE is generated. As fatigue processes continue, AE increases in terms of number of events, event intensities and spatial distribution. Diffuse damage appears to be a precursor to the development of linear microcracks. The spatial and temporal sequence of AE events enables differentiation between linear microcracks and more diffuse damage.
(3)
(4) (5)
Acknowledgements
Supported by NSF BES-9410303, the Whitaker Foundation and the Natural Sciences and Engineering Research Council of Canada. We gratefully acknowledge Mitch Schaffler, Ph.D. of Henry Ford Hospital for his help in the histological processing and analysis, and Gordon Schneider of Digital Wave Corp. for his help performing the modal analyses.
References
[I]
[21
Melton, L.J., Eddy, D.M., and Johnston, C.C. Jr., 1990, "Screening for Osteoporosis," Ann. Int. Med. Vol. 112, pp. 516-528. Holbrook, T.L., Grazier, K., Kelsey, J.L., and Stauffer, R.N., 1984, American Academy of Orthopedic Surgery, Rosemont, IL. Birdwood, G., 1996, Understanding Osteoporosis and its Treatment. A Guide for Physicians and their Patients, Pearl River, New York. Schaffler, M.B., Choi, K., and Milgrom, C., 1995, "Aging and Matrix Microdamage Accumulation in Human Compact Bone," Bone, Vol. 17, pp. 521525.
[31
[4]
20
[51
Katz, J.L., 1981, "Composite Material Models for Cortical Bone," Mechanical Properties of Bone, S.C. Cowin, Ed., American Society of Mechanical Engineers Vol. 45, pp. 171-184. Martin, R.B., Burr, D.B., 1989, Structure, Function, and Adaptation of Compact Bone, Raven Press, New York. Frost, H.M., 1960, "Presence of Microscopic Cracks In Vivo in Bone," Henry Ford Hospital Bulletin, Vol. 8, pp. 25-35. Martin, R., 1982, "A Hypothetical Mechanism for the Stimulation of Osteonal Remodelling by Fatigue Damage," J. Biomech., Vol. 15, pp. 137-139. Knet-s, I.V., Krauya, U.E., and Vilks, Y.K., 1975, "Acoustic Emission in Human Bone Tissue Subjected to Longitudinal Extension," Mekh. Polim., Vol. 4, pp. 685690. Wright, T.M., Vosburgh, F., and Burstein, A.H., 1981, "Permanent Deformation of Compact Bone Monitored by Acoustic Emission," J. Biomech., Vol. 14, pp. 405-409. Fischer, R.A., Arms, S.W., Pope, M.H., and Seligson, D., 1986, "Analysis of the Effect of Using Two Different Strain Rates on the Acoustic Emission in Bone," J. Biomech., Vol. 19, pp. 119-127. Zioupos, P., Currey, J.D., and Sedman, A.J., 1994, "An Examination of the Micromechanics of Failure of Bone and Antler by Acoustic Emission Tests and Laser Scanning Confocal Microscopy," Med. Eng. Phys., Vol. 16, pp. 203-212. Wright, T.M., Arnoczky, S.P., and Burstein, A.H., 1978, "In-situ monitoring of ligament damage in the canine knee by acoustic emission," Mater. Eval., Vol. 37. Wright, T.M., Hood, R.W., and Flynn, W.J., 1981, "Acoustic emission monitoring in the diagnosis of loosening in total knee arthroplasty," 1981 Biomechanics Symposium, Van Buskirk, W.C. and Woo, S.L.-Y., Eds., American Society of Mechanical Engineers, pp. 203-212. Poliakoff, S.J., Miller, R.K., Jones, C.B., and Bright, R.W., 1989, "Acoustic emission monitoring of physeal separation: an experimental study," Trans. Orthop. Res. Soc., Vol. 14, p. 483. Kohn, D.H., Ducheyne, P. and Awerbuch, J., 1992, "Acoustic Emission During Fatigue of Ti-6A1-4V: Incipient Fatigue Crack Detection Limits and Generalized Data Analysis Methodology," J. Mater. Sci., Vol. 27, pp. 3131-3142.
[61 [7]
[8]
[9]
[lO]
[Ill
[12]
[13]
[14]
[15]
[161
21
[17]
Kohn, D.H., Ducheyne, P., and Awerbuch, J., 1992, "Acoustic Emission During Fatigue of Porous Coated Ti-6AI-4V Implant Alloy," J. Biomed. Mater. Res., Vol. 26, pp. 19-38. Caler, W., 1989, "Bone Creep-Fatigue Damage Accumulation," J. Biomech., Vol. 22, pp. 625-635. Lanyon, L.E., Goodship, A.E., Pye, C.J., and McFie, J.H., 1982, "Mechanically Adaptive Bone Remodeling," J. Biomech., Vol. 15, pp. 141-154. Kannatey-Asibu, E. and Emel, E., 1987, "Linear Discriminant Function Analysis of Acoustic Emission Signals for Tool Condition Monitoring," J. Mech. Sys. Signal Proc., Vol. 1, pp. 333-347. Kohn, D.H., 1995, "Acoustic Emission and Non-Destructive Evaluation of Biomaterials and Tissues," Crit. Rev. Biomed. Eng., Vol. 22, pp. 221-306. Burr, D.B., 1995, "Alterations to the En Bloc Fuchsin Staining Protocol for the Determination of Microdamage Produced In Vivo," Bone, Vol. 17, pp.431-433. Ross, M., 1995, Histology. A Text andAtlas, Williams and Wilkins, New York. Schaffier, M.B., Radin, E.L. and Burr, D.B., 1989, "Mechanical and Morphological Effects of Strain Rate on Fatigue of Compact Bone," Bone, Vol. 10, pp. 207-214. Carter, D.R., and Hayes, W.C., 1977, "Compact Bone Fatigue Damage - I Residual Strength and Stiffness," J. Biomech., Vol. 10, pp. 325-337. Schaffier, M.B., Pitchford, W.C., Choi, K., and Riddle, J.M., 1994, "Examination of Compact Bone Microdamage Using Back-Scattered Electron Microscopy," Bone, Vol. 15, pp. 483-488. Prosser, W.H., Jackson, K.E., Kellas, S., Smith, B.T., McKeon, J., and Friedman, A., 1995, "Advanced Waveform-Based Acoustic Emission Detection of Matrix Cracking in Composites," Mater. Eval., Sept., pp. 1052-1058.
[18]
[19]
[20]
[21]
[22]
[23] [24]
[25]
[26]
[27]
Concrete Applications
Shigenori Yuyama, ~ Takahisa Okamoto, ~ Mitsuhiro Shigeishi, 3 Masayasu Ohtsu, aand Teruo Kishi 4 A PROPOSED STANDARD FOR EVALUATING STRUCTURAL INTEGRITY OF REINFORCED CONCRETE BEAMS BY ACOUSTIC EMISSION
REFERENCE: Yuyama, S., Okamoto, T., Shigeishi, M., Ohtsu, M., and Kishi, T., " A Proposed Standard for Evaluating Structural Integrity of Reinforced Concrete Beams by Acoustic Emission," Acoustic Emission: Standards and Technology Update, ASTM STP 1353, S. J. Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999.
ABSTRACT: A series of studies has been performed to evaluate the structural integrity of reinforced concrete (RC) beams by acoustic emission (AE). Cyclic loadings were applied to RC beams with a single reinforcing bar, large repaired beams, beams deteriorated due to corrosion of reinforcement, and two beams with different damage levels in an aging dock. The test results demonstrated that the Kaiser effect starts to break down when shear cracking starts to play a primary role. It has been also shown that high AE activity is observed during unloadings after serious damage (slips between the concrete and the reinforcement or those between the original concrete and the repaired part) has occurred. A standard for evaluating structural integrity of RC beams by AE is proposed, based on these results.
KEYWORDS: acoustic emission, cyclic loading test, evaluation criteria, Kaiser effect, reinforced concrete, structural integrity
President, Nippon Physical Acoustics Ltd., 8F Okamoto LK Bldg., 2-17-10, Higashi, Shibuya-ku, Tokyo 150, Japan. 2 Chief researcher, Central Research Laboratory, Nihon Cement Co., Ltd., 1-2-23, Kiyosumi, Koto-ku, Tokyo 135, Japan. 3 Associate professor and professor, ,'espectively, Depamnent of Civil and Environmental Engineering, Faculty of Engineering, Kumamoto University, 2-39-1, Kurokami, Kumamoto 860, Japan. 4 Professor, Research Center for Advanced Science and Technology, The University of Tokyo, 4-6-1, Komaba, Meguro-ku Tokyo 153, Japan.
Copyright9
by ASTMInternational
25 www.astm.org
26
Introduction
In recent years the deterioration and cracking of concrete structures such as bridges and buildings has been a significant problem. Proper techniques for the inspection of damaged structures are important in making rational decisions regarding rehabilitation, repair or replacement. Thus, the development of techniques to evaluate degradation of concrete structures in long-term service has been one of the most important issues for an effective maintenance program. A series of studies has been performed to evaluate the structural integrity of rcinforced concrete (RC) beams by acoustic emission (AE). Cyclic loadings were applied to RC beams with a single reinforcing bar [/], large repaired beams [2], beams deteriorated due to corrosion of reinforcement [3l, and two beams with different damage levels in an aging dock [4]. The test results demonstrated that the Kaiser effect starts to break down when shear cracking starts to play a primary role. It has been also shown that high AE activity is observed during unloadings after serious damage (slips between the concrete and the reinforcement or those between the original concrete and the repaired part) has occurred. A concrete beam integrity (CBI) ratio, the ratio of the load at onset of AE and the maximum prior load, has been proposed as an effective criterion to measure the severity of the damage induced in the beams. The high AE activities during unloadings have been shown to be an effective index to estimate the level of deterioration. This paper proposes an AE test method for RC structures, demonstrating four case studies conducted for different types of RC beams. Test procedure and evaluation criteria are presented as guidelines for practical AE tests of RC beams.
Case Study 1: RC Beams with a Single Reinforcing Bar

Shown in Fig. 1 is a configuration of the specimen used for the cyclic bending test. A single reinforcing bar of 19 mm dia with lateral lugs is encased eccentrically in the rectangular concrete beam. Concrete cover (depth of reinforcing bar) is 30 mm. Compressive and tensile strengths of the concrete were 36.2 and 3.5 MPa, respectively. Six PAC R15 (150 kHz resonant) sensors were attached on the specimen to perform both a moment tensor analysis using the SIGMA code [.5] and parameter analysis. The specimens were subjected to repeated four-point bending loadings by a strain-control type machine. The maximum load of each loading cycle was increased gradually in order to investigate the relationship between the cracking process and AE behavior. Figure 2 presents the relationship between the number of AE hits and the applied load. AE signals are detected at a lower load than the maximum prior load (49kN) during the second loading. Accordingly, the Kaiser effect breaks down during the second loading. It was shown that the Kaiser effect starts to break down when the
YUYAMA ET AL. ON REINFORCED CONCRETE BEAMS
27
Z
9 '. e .', "..o*..,% : . : , . . ' , . . ,, . Q.. ," .,,"
:..:. @.::,.'
/:~.:i-:"..:-..t . ". ;.L::.~ DEFORM E D:BAR :

[ . . . . . . . . ' -...-...-.~~ _-.....;. ~.... ; .'....'.:-. ~':'.-: ...:..-;. :;
"J J
150
Ji
.l_
FIG.l-Configuration o f the specimen with a single reinforcing bar.
2500)--.T: 2 0 0 0 -
,, o
kl_l
1500-
f_..J
1V
rn 1000z 500-
49
LOAD
9'8
(kN)
188
'
196
245
FIG.2-Relationship between number o f AE hits and the applied load.

crack width exceeds 0.12 mm. The breakdown of the effect becomes clearer as the cracking progresses in the third, fourth and fifth loadings. High AE activities are observed during the third, fourth and fifth unloadings. The moment tensor analysis revealed that the contribution of shear cracks increases as the breakdown of the Kaiser effect becomes clearer with the progress of the fracture. It was also indicated that high AE activity is observed during the third, fourth and fifth unloadings after the maximum width of the surface cracks has exceeded about 0.25 mm. The moment tensor analysis found that the shear cracks
28
generated near the reinforcing bar is responsible fl)r this activity. "lhe origin ,,i the~ emissions was attributed to rubbings between the faces of the existing cracks or ttiction between the reinforcement and concrete.
Case Study 2: Repaired RC Beams

The configuration of the specimen and the locations of displacement transducers and AE sensors are shown in Fig. 3. Six PAC R15 sensors were placed lineally on the top plane of the specimen. The repaired part is in the tensile side of the specimen. The depth and length of the repaired part are 100 mm and 2200 mm, respectively. In addition to steel bars as reinforcement, stirrups were embedded in the specimen to prevent beams shear failure. The specimens were subjected to repeated four point bending loadings by a strain-control type machine. During each loading, measurements of AE, crack width, slip length between the repaired part and the original part, and strain of concrete and reinforcement were made by using AE sensors and two different types of displacement transducers. These measurements indicated that the initiation of the early tensile microcracks, main tensile cracks, local slips, and large-scale slips are clearly detected by AE hit measurement. It was shown that once large-scale slips have occurred at the interface between the original concrete and the repaired part, AE starts to emanate at
L CH.,6 CH.,5 (6)

1
480
-B -
_1 240_1 240 i
-
480 3
J CH.m2
9
WCH.4 (8)
3 IF.:"
~,-
7,
~ ,
CH.II 1
(7)
2
....
(9) (10)
4 5 " 5.~ '. ~ "
....... ~. ,-,---.,.----'-----'-----
.. - - - - c . . . . . ~
1_2ooJ2oo.I 2001_2oo.l.2oo 12oo_1

I C
[-
Two directional displacement transducer for sllp length Displacement transducer for crack width
mm
AE Sensor
FIG. 3-Configuration o f the repaired RC beams and the locations o f displacement lransduce~ and AE sensors.
29
100
'`''i''''l'``'i`'''`'''I'''`'''i''''''I''`''`n'''''''I'''''`''l''''','`i''''''t ....
MAIN
CRACK
INITIATION SLIP
L A R G E - S C A L E
I
,.
v
, . 9
..
5o
9 ..
.;. . .............. . . . . . . . . . . . . . . . .
1
MICRO-CRACK INITIATION i0
6
A
J l
2.5~ z
'"'
'"'
'"'
'"'
""
""
'"'
'"'
""
""
""
'"'
""
'"'
""
""1'"'1'"
'
3 0 0 0
6000
TIME
(SECOND)
FIG. 4-Amplitude and displacement histories. much lower load than the previous maximum load, that is, the Kaiser effect no longer holds for the next loading and high AE activity can be seen even during unloading. Thus, the breakdown of the Kaiser effect and the high AE activity during unloading can be a good indicator for the occurrence of large-scale slips in repaired RC beam. Amplitudes of all hits are plotted versus time together with the displacement in Fig. 4. It is obvious that the initiation of the early tensile microcracks or the local slips and the mechanical rubbings of the interlocked faces due to large-scale slips gave amplitude levels between 40 and 60 dB, while the initiation of the main tensile crack at 38.2 kN produced very high amplitudes that reached nearly 80 dB. Thus, the different AE sources could be clearly distinguished by comparing the amplitude data with the results of the visual observation and the measurement by displacement transducers. A concrete beam integrity (CBI) ratio, given below, was proposed as a criterion to measure the severity of damage induced in repaired concrete beams. CBI ratio = load at onset of AE / maximum prior load In the field of fiber-reinforced plastic (FRP) structures, AE tests have been widely used to evaluate structural integrity and testing has been standardized by ASME Code, Section V, Article 11. In this code, the Felicity ratio obtained from the ratio of the load at onset of AE and the maximum prior load gives the criterion to measure the severity of previously induced damage. It has been shown that the Felicity effect,
30
which is referred to as the breakdown of the Kaiser effect, is an indication of defects. The felicity ratio has been well accepted to examine structural integrity of chemical plant equipment such as pressure vessels, tanks and piping. However, beams, pillars, columns, and slabs are inspected in concrete structures. In the case of chemical plant equipment, structures are loaded by pressurization. In contrast, concrete structures are subjected to tensile, shear and bending loadings by jacking or running a heavy vehicle. As shown in the test results, failure mechanisms vary with the progress of damage in RC beams. It is obvious that the decrease of the CBI ratio is related to the generation and propagation of shear cracks. Thus, the ratio can be a practical index for evaluating structural integrity of RC beams. Taking these points into consideration, the CBI ratio was introduced [6]. Listed in Table 1 are CBI ratios for each loading cycle, based on AE hit rate activity. The ratios obtained by AE energy rate activity for all channels are also given in the last column. As shown in Table 1, the CBI ratio decreases from the fifth loading after large-scale slips have occurred due to the fourth loading along the interface between the original concrete and the repaired part. It continues to decrease as the damaged areas grow. It is known that the occurrence of large-scale slips is an essential feature for damage that may result in a serious disaster in repaired concrete structures. As shown above, the CBI ratio has very good correlation with the onset of large-scale slips and the growth of damaged areas. Thus, the CBI ratio can be a very useful and effective criterion to measure the severity of damage induced in repaired RC beams. It was also revealed that high AE activity is observed during unloadings once large-scale slips have initiated between the original concrete and the repaired part. The source of these emissions was ascribed to mechanical rubbings between the interlocked faces introduced by the large-scale slips. TABLE 1-Concrete Beam Integrity (CBI) ratios during the repeated loading tests of
repaired RC beams.
CH2 CH3 CH4 CH5 CH1 "~ 6 (hit) 1.25 1.16 1.20 0.69 0.53 0.40 CH1 ~" 6 (energy) 1.25 1.16 1.20 0.69 0.53 0.25
(hit)
2nd 3rd 4th 5th 1.25 1.16 1.20
(hit)
1.16 1.20
(hit)
1.25 1.16 1.20
(hit)
1.20
6th 7th
0.69 0.53 0.68
0.69 0.53 0.40
0.69 0.53 0.50
1.00 0.80 0.68
31
Case Study 3: RC Beams Deteriorated Due to Corrosion of Reinforcement Shown in Fig. 5 are dimensions (cm) of the specimen and sensor locations. Six PAC R6 (60 kHz resonant) sensors were attached on the specimen to perform the moment tensor analysis as well as AE parameter analysis. The lower quarter part of the specimen was immersed in a 3% sodium chloride solution and an anodic current was galvano-statically charged to the main steel bars until the maximum width of surface cracks due to corrosion of the bars reached 1 mm or 4 mm. Thus, three different types of specimens i.e. specimen with no corrosion damage and those with the surface cracks determined as above were subjected to repeated four-point bending loadings. Indicated in Fig. 6 are relationships between AE hits and the applied load for the specimens with the different deteriorated levels. It is observed that the Kaiser effect starts to break down during the third loading in the case of the specimen with no corrosion damage. However, it tends to break down during the second loading in the case of the deteriorated specimen (crack width 1 mm) and the breakdown is very clear during the second loading in the heavily deteriorated one (crack width 4 mm).
~<i (cm)
S
izo ..
Sensor
CH No.
CHI CH2 CH3 CH4 CH5 CH6
location
(cm) y
15 5 5 15 20 20
x
10 50 10 50 20 40
z
0 0 19 19
14 5
FIG. 5- The specimen and AE sensor locations for the repeated loading tests o f RC
beams deteriorated due to corrosion o f reinforcement.
32
10000 -
ll.-,* 8000 ml
(a)~
i , ~ , i , ,
Il l
E
6000-
,oooi
2000O.
98
196
LOAD ( k N )
10000-
I-,
8000-
(b)
6000
,oooi
2000 98
196
LOAD
I0000-
(kN)
(c)
I~ 8000
~=~ 6000
J
2000
-
O-
98
LOAD
(kN)
196
FIG. 6-Relationships between A E hits and the applied load for the specimens with
different deteriorated levels; (a)specimen with no corrosion damage, (b)deteriorated specimen (crack width 1 mm), (c)deteriorated specimen (crack width 4 ram).
33
TABLE 2-Concrete Beam Integrity (CBI) ratios for the second and the third loadings. Loading cycle No damage Crack width lmm
Crack width 4mm II III
1.00 0.71
0.28
0.75 0.49 0.10
CBI ratios for the second and the third loadings are summarized in Table 2. It is obvious that the CBI ratios exhibit smaller values than Case 1 because of the breakdown of the Kaiser effect during the second loading in the deteriorated specimens. It is also seen that the ratio becomes smaller as the deterioration due to corrosion of the reinforcement becomes greater. During the third loading, the CBI ratios are smaller than 1 for all the specimens. Again the ratios exhibit smaller values in the specimens with the greater deterioration induced by the corrosion. Thus, it has been confirmed that the CBI ratio can be an effective criterion to measure the severity of the damage due to corrosion of the reinforcement in RC beams. It is also observed in Fig. 6 that different levels of AE activity are detected during unloadings, depending on the different damage levels. In the specimen with no corrosion damage, relatively high AE activity is first observed during the 2nd unloading, as shown in Fig. 6(a). However, some AE activity is already detected during the 1st unloading in the case of the deteriorated specimen (crack width lmm). High activity is seen during the 2nd unloading (Fig. 6(b)). Quite high AE activity is observed during the 1st and the 2nd unloadings in the heavily deteriorated specimen (crack width 4 mm), as seen in Fig. 6(c). Thus the levels of AE activity during unloadings reflect the damage levels induced in the specimens. Since high AE activity corresponds to the occurrence of serious damage, it can be an effective index to estimate the level of deterioration.
Case Study 4: An Aging Dock

Structural integrity of RC beams was evaluated in an aging dock. Shown in Fig. 7 is a cross section of the tested pier. A repaired beam and a damaged one without repair were subjected to three loadings by running a dump truck with three different load levels i.e. empty (ll3kN), half load (142kN), and full load (171kN). Figure 8 schematically illustrates how the tests were performed. Two R15 and R6 sensors were attached on the beams. However, only the data collected by R6 sensors could be analyzed since there were no significant AE signals detected by R15 sensors due to high attenuation at higher frequencies. A strain gage was attached on the main reinforcing bar to measure strain changes during the loadings. Two cracks
34
tJ
4-~70 t7 HWL LW.L ~ 2.00 -+ 0.00
H.O0
5 0(3
f
i
q.t30
I. 3]23 ,
2=,C0 .
500
&O0
IO.f}0
:]30
9,
'~.x ~
-:'.:30
Sea
2,;:
13r
- 2 a Of)
[rn]
FIG. 7- Cross section o f the tested pier.
FIG. 8-Schematic illustration o f the loading test. the maximum opening width of which reached 0.8 mm were visually observed on the surface of the unrepaired beam and measurements of corrosion potential confirmed that serious damages due to corrosion of the reinforcement existed. Strain gage measurements showed that the strain change is much larger in the unrepaired beam than the repaired one for the same loadings. AE hit rate, strain and amplitude histories for the damaged beam are given in
35
40
30
.......................................
-
80 .~
Go
Lo
t
@ ................
Time
:,vlifiii, i,i il,,i

2o
90 80 70
0
(Empty) (Half loaded) (Fully loaded ..............................................
~--~ . . . . . . . . . . . . . . . . . . . . . . . . . . . .
$
...,.
t
............................
60 9. . . . . . . . . . . . . . . . . . .
.,-4
9 "
"-&l
<
! .~. t~ 50 :'* ............................. 9 9 :t"

",
.| ~"~ /
40
Time
FIG. 9-AE hit rate, strain and amplitude histories for the damaged (unrepaired) beam. Fig. 9. Although the repaired beam (no damage) was very quiet, high AE activity was observed since the first loading in the unrepaired one. The Kaiser effect breaks down during the third loading and high AE activities are seen during the second and third unloading. The AE source during the third unloading was thought to be frictions due to slips between the reinforcement and the concrete. The amplitudes from this source are smaller than 60 dB, as shown in Fig. 9. The CBI ratio is smaller than 0.6 during the third loading. Thus, it has been shown that the CBI ratio and the AE activities during unloadings are very good indicators for extensive deterioration in RC beams.
36
ACOUSTICEMISSION: STANDARDS AND TECHNOLOGY
Discussion
Cracking Processes and A E Behavior
AE behavior has been studied during cracking processes in different types of RC beams. Each fracture stage in reinforced concrete is schematically illustrated in Fig. 10. It has been reported that AE hits with medium level of amplitudes (40 to 60 dB) are emitted when early tensile microcracks initiate as the first stage of the fracture. If main tensile cracks are generated, AE signals with larger amplitudes (80 to 100 dB) are detected. After the width of surface cracks has exceeded a certain value as the damage progresses to the second stage of the fracture, internal cracks such as secondary tensile cracks or shear cracks start to initiate near the interface between the Fracture Stage
$
'
i
LoadS
9 . . /. 1 " .: .t 1 " / . / , Reinforcement Tensile microcracks
.1?.?
,l
$
Internal cracks [Secondary tensile c r a c k s ] or shear cracks
9 : "1. 7'.~'. q'.q~'.7 ~ )"
Slips between the reinforcement and the concrete III

9 9 . . . . . .
9
R e p a i r e d P a r t
"
Slips between the repaired part and the original concrete
FIG. lO-Schematic illustration o f fracture stages in reinforced concrete.
37
concrete and the reinforcement. In the cyclic loading test [ 7] of an L-shaped rigid frame, the Kaiser effect started to break down after the crack width reached 0.15 to 0.2 mm and high AE activity was seen during unloadings after shear cracks started to play a primary role. It has been thus shown that the Kaiser effect no longer exists and high AE activities are observed during unloadings when the fracture progresses to the second stage. This is because shear cracks start to play a primary role around the interface between the reinforcement and concrete, generating AE events due to mechanical rubbing of crack faces during loading and unloading. The amplitudes due to this mechanism are smaller than 80dB, as indicated in Figs. 4 and 9. When the fracture progresses to the third stage, slips (shear cracks) start to take place during loading and unloading between the reinforcement and concrete or between the repaired part and the original concrete. The amplitudes from this source are no larger than 60 dB as seen in Figs. 4 and 9. Thus, the damage level of RC beams is strongly related to the initiation of shear cracks. Therefore, it is possible to evaluate structural integrity of RC beams by analyzing emissions due to shear cracking.
Evaluation Criteria
In evaluating the structural integrity of concrete members, shear crack initiation at the interface between the reinforcement and concrete or that between the repaired part and the original concrete is vitally important. Therefore, the detection of shear cracks (slips) can be a practical criterion to measure the severity of damage induced in RC beams. As has been shown so far, the breakdown of the Kaiser effect and the TABLE 3-An example o f evaluation criteria tbr damages induced in RC beams.
Fracture Stage Early microcracks Main tensile cracks Secondary tensile cracks Internal shear cracks Slips between reinforcement and concrete
or
(Crack width:
Low
Damage Level W)
Amplitude (dB) 40 ~ 60
80~ 100
CBI Ratio Larger than
AE Activity during Unloading Low
(W<0.12 0.20mm) Medium (0.12 ~ 0.20mm <W)

1
40 ~
80
0.8 ~ 0.9
Medium
Smaller than High

40 ~ 60 0.8
High
Slips between repaired part and original concrete
(0.5mm<W)
38
high AE activity during untoadings correspond to the occurrence of serious damage such as the slip between the reinforcement and concrete or that between the repaired part and the original concrete. Therefore, they can be effective indices to estimate the level of deterioration. An example of evaluation criteria for damage induced in RC beams is given in Table 3. As demonstrated, the CBI ratio and the AE activity during unloadings, which can be obtained under repeated loadings with increasing maximum loads, are very useful for evaluating structural integrity of RC beams.
A Proposed Standard
In the present series of tests, AE signals were detected by PAC R15 (150 kHz resonant) and R6 (60 kHz resonant) sensors, setting the system examination threshold at 40 dB. An attenuation study made prior to the tests showed that R15 and R6 sensors can cover the areas of approximately 0.5 m and 2 m in the RC beams with enough sensitivity, respectively. Since the maximum sensor distance was smaller than 0.5 m for RI5 sensors and 1 m for R6 sensors and the amplitude range of the detected AE signals was quite large, the effect of either the sensitivity of the AE channel or the attenuation in the structure on the CBI ratio is considered to be small in the examinations. However, dependence of minimum detectable AE on the system examination threshold must be always taken into consideration. Therefore, an attenuation study and sensitivity calibration of the sensors by a pencil lead break or electronic waveform generator with a pulse should be performed prior to test. Based on the series of the tests we have conducted, the following brief test procedure and evaluation criteria are proposed for AE tests of RC beams. (1) AE Sensor Low frequency sensor (60 kHz resonant) : large area (whole structure) High frequency sensor (150 kHz resonant) : small area (2) Attenuation Study and Sensitivity Calibration An attenuation study and sensitivity calibration of sensors should be performed prior to test by a pencil lead break or electronic waveform generator with a pulse. (3) Loading Method Running a dump truck or trailer 1st loading : Empty 2nd loading : Half loaded (half filled with water) 3rd loading : Fully loaded (fully filled with water) A strain gauge should be preferably attached on the main reinforcing bar to
39
measure strain changes during loadings. The maximum load should be determined in accordance with the allowable load. (4) AE Data Analysis Hit rate Count (Energy) rate Amplitude history AE activity (hit, count, etc.) vs. load Linear source location by low frequency sensor if possible Zone location by low frequency sensor or high frequency sensor (5) Evaluation Criteria Serious damages (slips between the concrete and the reinforcing bars or those between the concrete and the repaired part) in RC beams are indicated by: CBI ratio < 0.8 High AE activity during unloading
References
[/] Yuyama, S., Okamoto, T., Shigeishi, M. and Ohtsu, M., "Quantitative Evaluation and Visualization of Cracking Process in Reinforced Concrete by a Moment Tensor Analysis of Acoustic Emission," Materials Evaluation, Vol. 53, No. 6, June 1995, pp. 751-756. [2] Yuyama, S., Okamoto, T. and Nagataki, S., "Acoustic Emission Evaluation of Structural Integrity in Repaired Reinforced Concrete Beams," Materials Evaluation, Vol. 52, No. 1, Jan. 1994, pp. 86-90. [3] Murakami, Y. and Yuyama, S., "Acoustic Emission Evaluation of Structural Integrity in Reinforced Concrete Beams Deteriorated Due to Corrosion of Reinforcement," Progress in AE 8 (JSNDI), Proc. 13th Inter. AE Symp., Nov. 27-30, 1996, Nara, Japan, pp. 217-224. [4] Kamada, T., Iwanami, M., Nagataki, S., Yuyama, S. and Ohtsuki, N., "Application of Acoustic Emission Evaluation of Structural Integrity in Marine Concrete Structures," Progress in AE 8 (JSNDI), Proc. 13th Inter. AE Symp., Nov. 27-30, 1996, Nara, Japan, pp. 355-360. [5] Ohtsu, M., "Acoustic Emission Theory for Moment Tensor Analysis," Res. Nondestr. Eval., Vol. 7, No. 6, 1995, pp. 169-184.
40
[6] Yuyama, S., Nagataki, S., Okamoto, T. and Soga, T., "Several AE Sources Observed during Fracture of Repaired Reinforced Concrete Beams," Progress in AE 5 (JSNDI~ Proc. 10th Intern. AE Symp., Oct. 22-25, 1990, Sendai, Japan, pp.345-353. [ 7] Yuyama, S., Okamoto, T., Shigeishi, M. and Ohtsu, M., "Acoustic Emission Generated in Corners of Reinforced Concrete Rigid Frame Under Cyclic Loading," Materials Evaluation, Vol. 53, No. 3, Mar. 1995, pp. 409-412.
Edouard G. Nesvijski 1 ON THE NECESSITY OF A NEW STANDARD FOR TIIE ACOUSTIC EMISSION CHARACTERIZATION OF CONCRETE AND REINFORCED CONCRETE STRUCTURES REFERENCE: Nesvijski, E. G., " O n the Necessity of a New Standard for the
Acoustic Emission Characterization of Concrete and Reinforced Concrete Structures," Acoustic Emission: Standards and Technology Update, ASTM STP 1353,
S. J. Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999. Abstract: The acoustic emission (AE) method, though rather difficult in application and interpretation of results, has a great potential for characterization of stress, bearing properties, fatigue, and fracture of materials. The existing NDT standards that employ AE cover only a limited number of materials and structures. Direct compilation of these standards for materials with distinctive properties is difficult and sometimes impossible. For instance, concrete is a "living" material and AE can be registered immediately after preparation of cement or concrete mix, then during setting, and later during curing. AE in hard concrete can be registered due to initiation and growth of cracks under different kinds of physical factors. Classification of the signatures for different stages of concrete life and service is given. Some new models of the quantitative AE analysis are presented in this work.
Keywords: acoustic emission, concrete, strength, stiffness, crack-resistance, signal
processing, standard
Introduction
In spite of the long period of development of construction technologies, concrete and reinforced concrete are still holding the leading position. Many concrete structures show symptoms of deterioration before their designed service life. Repair and replacement of concrete infrastructure is a very expensive and labor-consuming task. It has become a matter of great concern to preserve existing concrete structures and improve their safety and durability in the 21st century. A fast developing theoretical platform for understanding of concrete failure mechanics and accumulating experimental data give an opportunity for a closer look at new possibilities of concrete technology [1].
' Visiting Research Professor, Center of Technology, Federal University of Santa Mafia, Campus Universitafio, CEP: 97105-900, Santa Maria, RS, Brazil
41
Copyright9 by ASTM International www.astm.org
42
Concrete and Concrete Structures - "Orphaned Children" of the NDT Standards Family
Control and management are important components of the complex process of concrete technology development. NDT and NDE techniques are traditionally vital instruments of characterization for many materials. A well-developed system of standards was created to support high quality of materials and safety and durability of structures. There are a great number of studies, reports and other publications devoted to development of NDT for concrete and concrete structures [2-4], but only a few test methods are standardized [5]. This factor has a negative impact on usage of NDT for concrete technology now and will have a place in the future. Concrete and concrete structures still remain "orphaned children" of the NDT standards family. For example, acoustic emission (AE) standards well developed for other materials are out of sight for concrete and concrete structures testing: ANSI/ASTM Practice for Acoustic Emission Monitoring of Structures During Controlled Simulation (E0569), ANSI/ASTM Guide for Mounting Piezoelectric Acoustic Emission Sensors (E0650), ANSI/ASTM Practice for Acoustic Emission Monitoring During Continuous Welding (E0749), ANS1/ASTM Practice for Characterizing Acoustic Emission Instrumentation (E0750), ANSI/ASTM Practice for Acoustic Emission Examination of Fiberglass Reinforced Plastic Resin (FRP) Tanks/Vessels (El 067) and others.
Features of AE Applications for Cement-Aggregate Based Materials and Structures
The AE method, though rather difficult in application and interpretation of results, has a great potential for characterization of stress, bearing properties, fatigue, and fracture of materials. The existing NDT standards that employ AE techniques are of general character or have been worked out for specific materials, such as metals and alloys, and cannot be compilated directly for cement-aggregate based materials. Three main factors can explain the necessity for a new standard for the AE characterization of concrete and reinforced concrete structures. The first is connected with particulars of stress-strain relation in concrete under different loads (static, dynamic, chemical, thermal, etc.) which are responsible for different failure mechanisms. These mechanisms initiate a generation of signals with various AE signature characteristics for every one of them or their combination. The second factor represents specific conditions of propagation of acoustic signals in a multi component medium, such as different types of concrete. It is connected with the inhomogeneous structure of the material and contributes to particulars of AE measurement and analysis. These materials, from the acoustic point of view, can be described by models of polycrystalline elastic solid and by models of composite materials with stochastic structure. In this case propagation of AE signals are characterized by attenuation connected with high dispersion on inhomogeneity with a display of explicit frequency dispersion of velocity. The third factor is connected with specific problems of testing concrete and reinforced concrete structures and control their main characteristics: strength, stiffness, crackresistance. Additionally, there are a number of practical problems connected with AE analysis of pre-stressed reinforced concrete structures during their manufacture as well as during service and repair, and also of testing the steel-to-concrete bond [6]
NESVIJSKI ON NECESSITY OF A NEW STANDARD
43
There are also other features that are important for AE applications in concrete and reinforced concrete structures: design of AE sensors, signal processing, procedure of source coordinates location, etc., which have to be standardized.
Structural AE Sources in Concrete
The model of cement-aggregate material from the point of view of mechanics of failure may be presented as a matrix-filler composite. Here filler represents various kinds of aggregate with different shape, size, elastic and strength properties. Matrix may be represented by a model of polycrystalline solid consisting of various components, such as sand, hydrated cement, and their chemical compounds with a spreading net of pores and micro-cracks filled with water of air. Non-organic admixtures that are used now (fly and rice ash, silica fume and others) change the mechanical structure of concrete matrix. That is why it is possible to consider three main sources of AE due to different actions and loads. The first is AE signals emitted by structural changes in matrix. This physical process is rather difficult to describe because sizes and shapes of micro-cracks, their direction, and distribution in matrix are of random character. Two opposite processes may characterize stress-strain mechanism in matrix: opening and growth of micro-cracks and shut of pores. These processes initiate AE with different signatures. As they are of mass and continuous character it is very difficult to distinguish them from one another. In this case the signal reminds typically the "white" noise process. Analysis of this type of AE requires special instrumentation [7]. The second source is that AE can be initiated as a result of adhesion failure between matrix and filler surface. This AE source emits short pulses of different shape depending on filler shape and adhesion character. These pulses have sufficient amplitude for effective registration. The only problem is their different locations in the matrix. As a result they are registered as a package AE signal, which represent a combination of continuous AE from matrix and a sum of AE from matrix-filler sources. Sometimes there are considerable difficulties to distinguish a weak signal from a distant location in the shadow of a strong pulse from a close zone at the registration point. The third source is represented by a physical failure of filler. As a rule this type of AE can be observed only at a high level of stress before fracture or collapse of material or structure. AE pulses in this case have very high amplitude and are less frequent than other types of AE signals. It is necessary to add that at this stage micro-cracks develop to macrocracks in matrix and also emit high amplitude AE signals. They resemble AE filler pulses, but in a number of cases it is possible to distinguish them.
Features of AE Propagation in Concrete
Due to complexity of structure concrete has a strong filtrating effect n the propagating AE signals. The material can be represented as a chin of filters with changing from point to point characteristics. That is why the amplitude-phase characteristic of the signal changes greatly, and the registered signal may be completely different from the emitted one, or two similar signals may have absolutely different features at the point of registration. Typical AE pulses and their spectra are shown (Figure 1).
44
Figure
1 -
Typical AE Pulses in Different Kinds of Concrete: a - Hard Concrete," b Cement Stone Concrete; c - Lime Stone Concrete
NESVlJSKI ON NECESSITY OF A NEW STANDARD
45
Set of AE Applications Will Be Ready to Meet a New Standard Procedure
Estimation of the Relative Level of Stress

Application of AE for estimation o f stress in concrete and reinforced concrete structures was developed as a result of analysis o f experimental data o f testing concrete specimens by central axis compression and measurement o f the AE parameters during testing time [8]. Estimation of the relative level o f stress is based on the idea of building a dynamic model with identification of transfer operator F(t) according to the input action X(t) and the response Y(t). It can be described mathematically: input action X(t) like Dirac function 6 ( 0 , output action or response like convolution function Y(t). It is possible to give a physical interpretation with the help o f the small loads (SL) method:
Y(t) = F(t) | 8 ( 0
(1)
Application of additional SL to the specimens that are under various relative levels of stress (stress to ultimate load ratio) generates AE of different character. The AE signals are registered during a certain period o f time AT after these actions. This approach allows the development of a criterion similar to output action as a convolution function, where multiplication o f count velocity N of AE to period o f time AT after the SL action is the core of this function.
Analysis of Stress Using NN-Criterion

The AE response to small loads for different relative levels o f stress can be analyzed using NN -criterion (Nogin-Nesvijski criterion) [8]. Values o f NN -criterion are calculated for different relative levels o f stress~(t)/R, where R is the strength o f the concrete specimen. The formula for heuristic NN -criterion is presented below:
(2)
where: N, = the AE count velocity; t i = time interval between the moment o f application of additional small load and the moment o f the AE registration; B = weight coefficient, which depends on the material structure and particulars o f its response to an action i = 1,2...,k, where k is a number o f AE measurements.
Combined Method of Stress Evaluation

Combined method of stress evaluation is based on the joint measurements o f AE and infrared radiation (IR) as a result of application of SL [9]. AE is charaetefized by the NN - criterion; IR is characterized by the change o f the density o f thermal flow AW. The AE method with the help of the NN - criterion allows the evaluation of the relative level o f stress 8 R. The IR method give the possibility to evaluate the change o f stress ty. Thus, the
46
combined method uniting the advantages o f both methods could be used to forecast strength R or bearing properties &concrete structures R + ~T~
J f(NN)J
where: ~ = coefficient of the heat expansion of the material; T O= the initial temperature; Y0 = density of the material; ~ = specific heat of the material; f ( N N ) = experimental function; plus(+)= compression of the structural material; minus(-) = tension of the material.
(3)
Practical Applications of AE for Stress Analysis

The SL method has its limitations. For the problems of field-testing is can be substituted by the method of local small loads (LSL) There are two variants of creation of the LSL: mechanical loads can be created using a special device (indenter) on the surface, an AE transducer is included in the device for registration of the signals; thermal loads can be created by a special device with lens for concentration of point heating, by the pulse laser, or by the laser with continuous radiation
lhermal A E
The character of thermal AE is different for different types of concrete. It is possible to consider two thermal processes: beating and cooling. The NN - criterion may be applied for analysis of the relative level of stress in these cases.
Other Important AE Applications

The AE method can be used for estimation of anchor length in a pre-stressed reinforced structure [10]. It gives the possibility to analyze stress and / or strain in the anchoring zone. It also can be applied for quality testing of embedded articles in reinforced concrete structures and for measurement of stress in core structures of multistory monolithic concrete buildings during their construction [11]. Concrete, unlike other construction materials, has special acoustic effects when loaded. These effects could be used for forecasting of failure or durability prediction of the material [12,13]
Promising Directions of AE Testing of Concrete

Deeper investigation of the AE signal processing should give more information about structure of testing materials and dynamics of modification of their properties. One of the directions of the investigation is separation and analysis of AE signals at the point of their generation. Another direction is analysis of filtrating properties of concrete when the point of registration is at some distance from the AE source. These two approaches could allow solving the inverse problem of identification of concrete failure. It is possible to use
47
complete characteristics of the signal, for example, their complex spectra [14]. General 3-D images of complex spectrum of single AE pulses are demonstrated (Figure 2).
Figure 2 - General 3-19Images of Complex Spectrum of Single AE Pulses Registered for: a - Hard Concrete; b - Cement Stone Concrete; c - Lime Stone Concrete
48
Modeling of the AE testing may be built on the basis of dynamic stochastic models, when concrete will be represented by a "black box" schematic. In this case the AE process should be considered a material reply to the testing loads. Usage of such a model gives an opportunity to solve three main problems: forecasting of time series behavior using "lead time" data (data obtained during current experiments) for prediction of the material response to the similar actions and loads in the future or testing results for a twin material specimen during "lead time"; identification of the material properties as an estimation of transfer function for the "black box" model. In this case the problem of identification has been solved according to the results of the input and output actions. There is a transfer of notion of mathematical description of transfer function on characterization of the material. This logical substitution gives us an opportunity to formalize testing procedures and describe the material as a set of formulae, which can be used for quantitative and qualitative characterization of the material; this "black box" model is open for further development and could include other types of loads and actions and corresponding material responses to them.
Conclusion
The AE method Applications for NDT of concrete and reinforced concrete structures differ considerable from other materials (metals, alloys, plastics, ceramics, etc.). The practical application of AE for concrete structure characterizations and evaluation has specific tasks and that is why new approaches, equipment, and software are required. Standardization of all parts is the main streamline for successful application of the AE method in the laboratory and industry. But it is necessary to point out that this problem requires further investigation together with studies of mechanism of failure and fracture of concrete. New modem analysis for AE signal processing and imaging, and analysis of time series on the basis of stochastic dynamic models, should be provided for successful standardization of the AE testing of concrete and concrete structures.
References
[1] Ohtsu, M., "Quantitative Evaluation of Crack and Damage in Concrete by Acoustic Emission," Keynote Speech, Sixth International Symposium on Acoustic Emission from Composite Materials (AECM-6), San Antonio, TX, June 2-4, 1998. [2] Hardy, R.H, Jr., "Evaluating the stability of Geologic Structures Using Acoustic Emission," ASTM-STP 571, 1975, pp. 80-106. [3] Lord, AE., Jr. And Koener, R.M., "Acoustic Emission in Geologic Materials," Proceedings, Joint Meeting of Acoustic Societies of America and Japan, Nov.22Dec. 1, 1978, Publ. UCLA, Kanji Ono, Ed., pp. 261-307. [4] Nogin, S.I. and Nesvijski, E.G., "The Parametrical Points of the Cracking Process in Concrete under Compression," Journal of Concrete and Reinforced Concrete, No.3, 1980, Moscow, pp. 10-12.
49
[5] Nesvijski, E.G., "Perspective Integration of Ultrasonic Pulse Method for Testing of Concrete in the Frame of GOST, ASTM, and ISSO Standards, "Proceedings, ASNT's Spring Conference, Las Vegas, NV, 1995. [6] Nesvijski, E.G., Sagaydak, A.I., and Tukhtaev, B.H., "Quality Control of Embedded Articles," Journal of Civil Engineering Maretials and Structures, No.3, 1991, Kiev, pp. 23-25. [7] Nesvijski, E.G., "On Application of Acoustic Emission Method for Prognosis of Hard Concrete Properties," Proceedings, Fifth International Symposium on Acoustic Emission from Composite Materials (AECM-5), Sunsvall, Sweden, 1995, pp. 385 - 392. [8] Nesvijski, E.G. and Nogin, S.I., "Acoustic Emission Technics for Nondestructive Evaluation of Stress of Concrete and Reinforced Concrete Structures and Materials," Proceedings, Third Conference on Nondestructive Evaluation of Civil Structures and Materials, Boulder, CO, 1996, pp. 285-294. [9] Nesvijski, E.G. and Sagaydak, A.I., "'A Combined Method of Estimating Stress in Structures," Journal of Civil Engineering Materials and Structures, No.2, 1991, Kiev, p. 37. [10] Nesvijski, E.G. et al., "'Method of Testing Stress Conditions of Concrete and Reinforced Concrete Structures," Patent SU 1472820 A1, Publ. BI No. 14, 1989, Moscow. [11] Nesvijski, E.G., Sagaydak, A.I., and Vernidub, A.T., "Acoustic Emission Method Applied to Testing of Multi-Storied Concrete Monolithic Buildings," Journal of Civil Engineering Materials and Structures, No.2, 1992, Kiev, pp. 34-35. [12] Nesvijski, E.G. and Nesvijski, T.E., "Kaiser and Felicity Effects and Their Application for Evaluation of Concrete by Acoustic Emission," Proceedings, ASNT's Fall Conference, Seattle, WA, 1996. [13] Nesvijski, E.G. and Nesvijski, T.E., "Failure Forecast and the Acoustic Emission 'Silence Effect' in Concrete," Proceedings, ASNT's Spring Conference, Houston, TX, 1997. [14] Nesvijski, E. G., "New Possibility for Acoustic Emission Testing of Concrete Structures," Progress in Acoustic Emission IX, Transition in AE for the 21st Century, Proceedings, International Acoustic Emission Conference, 14th International Acoustic Emission Symposium, 5th Acoustic Emission Worid Meeting, Big Island, Hawaii, 1998, pp. II-119-125.
H. R. Hamilton, 11I,1 T. J. Fowler, 2 and J. A. Puckett3 AE E V A L U A T I O N OF F A T I G U E D A M A G E IN T R A F F I C S I G N A L P O L E S
R E F E R E N C E : Hamilton, H. R., 11I, Fowler, T. J., and Puckett, J. A., " A E Evaluation of Fatigue Damage in Traffic Signal Poles," Acoustic Emission: Standards and Technology Update, ASTM STP 1353, S. J. Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999. ABSTRACT: Two traffic signal structures recently collapsed in Wyoming. The collapse was a result of brittle fracture of the connection between the cantilever signal light support arm (mast arm) and the vertical pole (post) caused by wind-induced, largeamplitude vibrations of the mast arm. This paper presents the problem, the proposed test program, the results of initial development tests, and the effort to develop a viable AE testing technique along with evaluation criteria that can be used to screen in-service traffic signal poles. K E Y W O R D S : acoustic emission, traffic signal poles, wind-induced vibrations, fatigue, welds
Background
Two signal structures recently collapsed in Wyoming. The collapse was the result of the fracture of the connection between the cantilever signal light support pole (mast-arm) and the vertical pole (post) connected to the foundation, The Wyoming Department of Transportation (WYDOT) performed inspections of the state inventory of traffic signal poles and found that nearly one-third of the approximately 820 poles inspected had fatigue cracks visible at the surface of the post at the toe of the weld. The
i Assistant Professor of Civil Engineering, University of Wyoming, P. O. Box 3295, Laramie, WY 82071. 2 M. W. Kellogg, Adjunct Professor, Department of Civil Engineering, University of Texas at Austin, Austin, TX 78712-10",'6. 3 Professor of Civil Engineering, University of Wyoming, P. O. Box 3295, Laramie, WY 82071.
Copyright9
50 www.astm.org
HAMILTON ET AL. ON FATIGUE IN TRAFFIC SIGNAL POLES
51
crack lengths found in the inspection program varied in length from less than one inch (25 mm) to as much as twenty inches (500 mm). The findings prompted WYDOT to immediately initiate replacement of the poles with the most extensive cracking. The replacement poles were fabricated with a connection detail designed by WYDOT to have improved fatigue resistance over the existing detail. WYDOT also contacted other states to determine if the problem extended beyond Wyoming. Other states reported having had isolated failures of traffic signal poles. However, no states actually have inspection programs in place for routine inspection of traffic or sign structures. In addition, many traffic signal structures are owned by local authorities rather than the state DOT and may receive even less attention in that regard [1]. The University of Wyoming (UW) was asked to look at the problem and provide assistance. UW is currently involved in research on improving the very low damping (0.1-0.7 percent critical) in these structures, investigating the fatigue resistance of the old and new connection detail, and testing and implementing potential NDT methods to be used in regular inspection of traffic signal poles. This paper covers the concept of using AE to inspect traffic signal poles and presents the results of a pilot test conducted on a fill-scale signal structure.
Typical Pole Geometry

Wyoming uses a traffic signal pole that originated in California. The pole is manufactured from ASTM A53 Grade B welded steel pipe. The mast-arm and post are manufactured as a single piece and then hot-dipped galvanized for Mast-ar corrosion protection (FIG. 1). Both the mast arm and post are tapered sections. The baseplate on the post is bolted Post directly to the foundation anchorbolts. The mast-arm and post are field connected with two thick plates and high FIG. 1--Signalpole configuration. strength bolts (FIG. 2). One plate is welded to a 4-plate box that is, in turn, welded to the post. The other plate is cut and welded to the mast-arm as a socket type connection. All welds are fillet welds. There are generally two different types of connections used, the open and closed box connection. Both Wyoming failures occurred in the post material at the toe of the weld between the box and post, illustrated in the figure with bold lines.
Wind-Induced Fatigue
Prediction of the fatigue life in traffic signal poles under wind-induced vibrations is a complex problem. Because of the low stiffness and damping of these
52
structures, mild wind can cause significant vibrations of the cantilever element, which causes variable-amplitude stresses to develop in the connection. The vibrations occur at relatively low wind speeds causing many cycles at moderate stress ranges to develop in a short period of time. Wind speeds of 10-30 mph (16-48 kph) are almost a constant occurrence in Wyoming so it is possible that these poles are experiencing large numbers of cycles before similar poles in other parts of the country. This may mean that other parts of the country that use a similar detail could see similar problems surfacing in the next few years. continuous Weld
FIG. 2--Comparison of open and closed connections. McDonald et al. indicate that the primary contributor to vibrations in signal poles is galloping during wind speeds in the range of 10-30 mph (16-48 kph) [2]. Galloping is an aerodynamic phenomenon that causes the tip to displace vertically. The signal poles of the type used in Wyoming tend to vibrate predominately in the first mode as a single degree-of-freedom system with the maximum displacement occurring at the tip of the cantilever (FIG. 3). McDonald et al. also indicated that out-of-plane movement was caused by natural wind and truck-induced wind gusts and that the outof-plane vibration is In-plane vibrations: tip predominately first mode in displacement perpendicular to which the tip displacement is g r o u n ~ / maximum, causing torsion in the post. Amplitudes of 8 to 18 inches (203 to 457 mm) in a fullscale test specimen (with a 48 ft (13.7 m) cantilever) during galloping in actual wind displacement parallel to [.'1 conditions of 10-30 mph (16-48 ground surface ~ " kph) were found to occur. \ Fatigue research was Wind direction for ' x ~ conducted at Lehigh University galloping to occur for Caltrans (California Department of Transportation) in FIG. 3--Vibration of pole.
53
1980 on these types of Previous fatigue tests caused connections [3]. The cracking at toe of pipe/plate weld. results of the research indicated that there were Cracking has been found in no fatigue problems with the vertical post at the toe of the welds. The cracks have the connection for inplane loading of the arm. generally radiated from the comers and have been noted The fatigue tests were conducted using in-plane at both the top and the bottom of the box. This displacements with no pattern has been found in out-of-plane both the open and closed displacements imposed. connections. Most of the relevant research FIG. 4--Crackingfound during WYDOT inspections. conducted to date has focused on the in-plane displacements of the signal arm caused by galloping. However, the WYDOT field inspections following the failures indicated that the fatigue cracking is occurring at the top and bottom comers of the box connection indicating that there are significant out-of-plane stresses in combination with in-plane stresses (FIG. 4). It is expected that if the in-plane displacements were causing the damaging stresses, then the majority of the cracks would have been found in the top of the connection. This is an indication that the stresses caused by out-of-plane displacements may be a significant contributor to the damage. Another possible explanation for the fatigue cracking at the bottom of the connection is an "oil-canning" effect that may be induced by the flexibility of the thin vertical post to which the box is welded. As the mast arm vibrates, moment develops at the connection. This moment is transferred through the connection to the post. If the post is sufficiently thin then unfavorable flexural stresses could develop in the post that eventually initiate and grow fatigue cracks.
Post-Mortem Examination
A section of one of the Wyoming failed poles was sent to Lehigh University for examination. The results of the examination indicated that the cracking initiated at the toe of the fillet weld. There were several sites along the length of the weld where cracking initiated. These sites eventually coalesced into a single crack that led to failure. One interesting aspect of the examination was that the cracks grew from the outside and inside of the wall section and met at approximately mid-depth. This indicates that external visual inspection will not necessarily detect cracking if it has initiated on the interior. This also supports the theory that the post may have significant flexural stresses present that cause cracking on the interior as well as the exterior. Finally, the examination indicated that there were no significant weld defects that may have precipitated cracking.
54
Research Significance
There may be a significant problem with the fatigue strength of many traffic signal poles throughout the U.S. However, this problem may be going unnoticed because traffic signal structures are generally not inspected regularly. In fact, in many instances traffic signal poles are owned by local authorities and are not considered a structure at risk of fatigue or other damage that requires a regular inspection program. In addition, while the cost of a typical signal structure is not small, each signal structure does not represent the significant capital investment that a bridge or petroleum tank would. It is prudent, then, to develop a method of inspection (to supplement visual) for these structures that does not require an extensive amount of training and experience and that is somewhat automated. AE provides an excellent screening tool that can be used to quickly test a pole to determine if further more detail inspection is required. Acoustic emission testing has been used successfully for many years in the petrochemical industry for testing FRP tanks, steel tanks, gas pipelines, etc. and for testing insulated bucket trucks (ASTM E 1067, E 1419, and F914). Standard test methods have been developed for testing these structures using AE. The attractiveness of the method is that it is a global inspection method. AE provides a method to screen these structures in a relatively quick manner without significant loss of service time.
UW Research Program
Field Testing
Because of the question concerning how much the in-plane and out-of-plane vibrations contribute to the connection fatigue damage, the initial focus of the research will be to determine the actual dynamic movements that typical poles have in service. Two 50-ft cantilever poles, currently in service in Laramie, WY, have experienced large amplitude vibrations under windy conditions, and will initially be monitored for several months to determine the direction and magnitude of the vibrations. These poles have been in service for only three years and do not have detectable fatigue cracks, so the test equipment will then likely be moved to a signal pole in Cheyenne that has fatigue cracks typical of those found on many of the poles inspected. It is anticipated that the field-monitoring program will yield information regarding the relative in-plane and outof-plane movements. Later, these results will be used to formulate laboratory and numerical studies.
Computer Modeling and Static Testing

Computer models of the old and new connection detail will be developed to determine the stress distribution in connection due to dead load and wind events. Other issues that may be significant are the method of assembly and the resulting residual
55
stresses locked into the connection. Concurrent with the analytical modeling, static laboratory tests will be conducted on existing connections to confirm the results of the computer modeling. WYDOT, as a result of the pole replacement program, has a number of poles that have been taken out of service that will be used to conduct the laboratory testing. In addition, pole suppliers have offered to fabricate both new and existing connections for use in testing as a baseline for "new condition."
NDE and Fatigue Testing

The results of the field study will be combined with the computer modeling and static laboratory tests to develop an appropriate fatigue test scenario for the damaged and new tes" specimens. The objectives of the fatigue tests are to determine the constant-amplitude fatigue life and to calibrate the AE and ultrasonic (UT) test methods. It is unlikely that the specimens will be tested purely with in-plane or out-of-plane displacements, but rather a combination based on the results of the field study. Acoustic emission testing will be conducted during the fatigue testing. In addition, the fatigue testing will be paused periodically to perform static AE tests in order to calibrate the AE vs. damage level at different stages in the fatigue loading. Periodic ultrasonic testing will also be conducted during the fatigue tests.
AETesting
While project has not yet reached the point where significant AE work has been completed, it was felt that a test should be conducted on a full-scale specimen early in the project to determine if there were any adjustments necessary in the planned laboratory work to accommodate field testing issues. To the authors' knowledge, AE has never been used on traffic signal poles, so it was prudent to try it in the field on a full-scale structure prior to performing the laboratory work. This will avoid developing procedures in the laboratory that are not appropriate or applicable for field testing.
Results of Initial AE Tests
As a part of the initial phase of the research project for WYDOT, there was a full-scale pole installed at the Laramie district yard for use by the researchers. The pole was taken out of service because of cracking that had been found during the dye penetrant inspection program initiated after the pole failures. The pole is a typical 50 ft cantilever mast arm that holds three signals, two standard and one turn signal (FIG. 5). This particular pole was constructed with a closed-box connection.
Dye Penetrant and Ultrasonic Inspections

Prior to AE testing, the test pole was inspected using both dye penetrant and ultrasonic inspection. Both inspections found cracking at the comers of the box connection (in the post at the toe of the weld). The results of the dye penetrant testing
56
indicated that there were two cracks approximately 1.5 inches (38 mm) in length located at the comers. The inspection report does not specify which comers. Ultrasonic inspection indicated that there were cracks less than 1 in (25 mm) in length at two comers corresponding to AE sensors 2 and 3 (FIG. 6)
Instrumentation
The test was conducted using the MISTRAS 2001 system with two Physical Acoustics Corporation (PAC) AEDSP-32/16 Cards mounted in a portable computer. PAC R151 (125 kHz resonant frequency; 70-200 kHz operating frequency) sensors with integral amplifiers (40 dB) were used for acquisition of AE signals. Traditional AE
FIG. 5--Test pole configuration. feature data were recorded along with the applied load level. The load was fed directly to the AE cards with a 5 kip (22.2 kN) load cell placed in line with the cable used to pull on the end of the mast-arm. Wave form capture was also activated during the test. The following settings were used: 1. 2. 3. 4. 5. Threshold: 40 dB Sample Rate: 2 MHz Peak Definition Time: 200 ~ts Hit Definition Time: 400 ~ts Hit Lockout (rearm) Time: 2 ~s
Sensors were attached to the pole at each Of the four comers of the box connection (FIG. 6) and were attached with hot melt glue. Magnetic clamps were also used to hold the sensors in place. The ambient temperature was approximately 30 ~ F (-1 ~ C) and there was some difficulty in getting the glue gun to heat to the proper level so that the glue could be applied in a sufficiently viscous state to allow the couplant to be squeezed to a thin layer under the sensor. It is anticipated that in future testing the magnetic clamps will be used to keep the sensors in place and a more viscous couplant will be used.
57
Applied Moment
Load Application Load was applied to the end of the mast arm using a fabric cable and a winch. As noted previously, the load cell was installed in line with the cable to allow recording of the load history along with AE data. As an added check, the tip displacements were monitored to ensure
FIG. G-Location of AE sensors.
that the load remained in the elastic range. When considering the strength of the pole under static loads, the location that controls the capacity of the structure is the base of the mast arm. The calculated tip load required to yield the pole, based on the measured section properties and an assumed yield strength of 35 ksi (240 MPa), was 960 pounds (4.27 kN). The maximum load applied was 665 pounds (3.0 kN), which is 69% of the yield capacity. AE Test Procedures After the application of the sensors, lead breaks were made at approximately 3 in (75 mm) from each sensor. The average of three breaks each was: Sensor 1 - 83.7 dB, SenSor 2 - 82.7 dB, Sensor 3 - 82.0 dB, and Sensor 4 - 84.0 dB. After application and testing of the sensors, the pole was fitted at the tip of the cantilever with the cable, load cell and winch. The cable was attached to concrete blocks to provide the necessary dead weight. It is anticipated that in actual testing conditions the cable would be attached to a bucket truck. Acquisition was initiated and a background noise check was done for five minutes. The load was then applied incrementally in steps of approximately 100 pounds (450 kN) and held until the emission ceased. Unfortunately, the gain was not set sufficiently high to obtain good resolution in the load readings, as can be seen in the load history plot (FIG. 7). Load (Pounds (kN)) 700 (3.1), 600(2.7) .1
500(2.2) i
400(1'8)i 300(1.3) 200 (0.9),; ~,~
ii
;~
/
d 1 O0(0.4) r / 0 (0)
0 10 20 30 Time (minutes) 40
50
FIG. 7--Load History.
58
The load was taken to 665 pounds (69%) on the first loading and 657 pounds (68%) on the second loading. At full load the tip deflection was 22 in. (560 mm). This level of load causes an extreme distortion visually and it is hoped that the test load can be reduced (mostly for the comfort of the operator and general public).
Data Reduction
There was a large amount of AE data acquired during the first loading. Because this was the first time the pole had been loaded, it is expected that much of the acoustic emission was from the relieving of residual stresses in the welds and slip in the connection. The second loading produced much less emission than the first loading indicating that the material was exhibiting the Kaiser effect. The counts and load history plots for the first and second loading are divided into two graphs (FIG. 8 and FIG. 9). There are columns of high count bursts centered, in general, around the times when additional load is applied and the number of counts increases with increasing load level. This indicates that there are possible noise events occurring that are generating spurious AE data. The form of the data looks similar to that found in steel tank or rail car tests when there is mechanical rubbing present. To eliminate this spurious data, a post-processing filter (Swansong II) was used to eliminate the data near hits that are suspected of being spurious emissions [4]. The filter keys on low amplitude hits that have long duration and removes all of the data within plus and minus half a second of the telltale hit. Count and load history plots after filtering indicate that there is a large number of hits that have been removed from the acquired data set (FIG. 8 and FIG. 9). Load (Pounds (kN)) Log AE Counts 10 5 700(3.1) 104 103 102 10
1
600(2.7) :; ;1
;!
; Illilm..~
!'
"
500 (2.2) 400(1.8)
"
l:l'.~l,II I I: '
'" ii
30 35
300 (1.3) 200 (0.9)
I~..; 0 5 10 15 20 25 Time (Minutes)
100(0.4)
0 (0)
FIG. 8--Count and load history for first loading.
59
Log AE Counts 105 104 103 102 10 9 fVg J'

( '
Load (Pounds (kN)) / ~

,i
t700(3.1) 600(2.7)
Y"9 !:,..n -" .~" : l, ~" ; 9~~ i "

9
, ""~" t
!.
"
; 9
9
.''roll
= ...
9
.~
IV:
T
.~.
ma-I
~-
~"
+ 300(1.3) ,
400(1.8)
200(0.9)
"
! 9 -, 9
: 9
-_%!
"-
;~,i,',,
~'~-~
+
12
9 ~ m,,.a.~. ~
: :=':--=',
"~.~.".
::~:
~
i,,
+100(0.4)
, io (0)
4 6 8 Time (Minutes)
10
FIG. 9--Counts and load history for second loading. It is likely that there is slip occurring between the connection plates during the load increase as the frictional resistance on the surface of the plates (due to bolt tightening) is exceeded by the applied load (FIG. 10). This is Slip supported by the fact that spurious data are generated simultaneously with the increasing load. In addition, there is a significant amount of AE during low loads and unloading. While the wind was calm for most of the test there was an occasional 10-15 mph (16-24 kph) wind that would develop that may have caused spurious hits. The mast arm is not restrained out-of-plane and is free to vibrate. One option is to load the mast ann at an angle (rather than straight down), which will tend to FIG. lO-stabilize the out-of-plane movements caused by the wind. Connection slip. Source Location The MISTRAS rectangular location algorithm was used after the completion of the test to determine if there were any clusters of events that might indicate damage. The source location was run using unfiltered data and was plotted on a two-dimensional graph (FIG. 13 and FIG. 14). Note that there is a small number of events shown on the location graph compared to the total number of hits received during the test, indicating that there were a significant number ofnon-locatable "events." The graphs are plotted with the sensor location shown at each of the four comers of the graph (inside the circles). Refer to Fig. 6 for the location of the sensors relative to the box connection. Also, note that in these location graphs, the box connection is oriented 90 degrees from vertical.
60
Log AE Counts lOs 104

I
Load (Pounds (kN)) 700(3.1) 600(2.7) 500 (2.2)

t
103 102 10
'.
,1~
. 9
,,1+
g.
400(1.8)
'
figk
"i;s" '
9 .. +
"
+
; ~
9 9
.~;" .':..: . .
~ , o,g "o g , * ,~
300(1.3) 200 (0.9) 100(0.4) 35
5-
10"~15 20 25 Time (Minutes)
30'
(o)
FIG. 11 --Counts and Load History for First Loading after Swansong H Filter. Log AE Counts 105 104 103 10z 10
9
Load (Pounds (kN)) 700(3.1)
,:,.. .9 . .| . 9 I
.:!t
o" 9 ~176
600(2.7) 500(2.2) 400(1.8) 300(1.3) 200(0.9)

0 (0)
...
t,oo,o4,
f2
:~ .... ' 24' : = 4 == = f0 Time (Minutes)
FIG. 12--Counts and Load History for Second Loading after Swansong H Filter.
12005)
8(203)
(~)
og
4(102) 0
,'~
0
4(11)2)
8(2'03)
12(305) 16(406)
20(508)
FIG. 13--Event vs. Location for First Loading (in(mm)).
61
12(305) ! 8(203) 4(102).
|
~,-
o0|
4(1'02)
8(2'03)
12(505)
16(406)
20(508)
FIG. 14--Event vs. Location for Second Loading (in (mm)). The first load does not indicate any significant cluster of activity. However, the second loading indicates that there is a significant area of activity directly between sensors 2 and 4 indicating that there may be significant damage at this location (FIG. 15). An alternative explanation for the concentrated emissions is friction at the compression face of the connection. Emissions from this point would appear to be centered between sensors 2 and 4. The likely explanation is that there is mechanical rubbing occurring at the base of the post. The emissions from this rubbing is picked up simultaneously at sensors 2 and 4 and then simultaneously at sensors 1 and 3 at a time necessary for the emission damage to travel from the bottom to the top of the FIG. 15--Possible sources for location connection. cluster during second loading. A first hit analysis was also conducted on the filtered data to determine if zonal location would provide a better indication of the damage located using UT and dye penetrant. Table 1 shows the results of the analysis. The time of arrival was examined and the sensor with the lowest time of arrival was assigned one first arrival and any hits received for 3 ms following this first hit were ignored. TABLE 1--Number offirst arrivals.
Channel 1
2 3 4
First Loading 58
122 14 235
Second Loading 21
41 1 75
The data indicate that channel 4 has the most first arrivals followed by channel 2. This can again be explained by the possible rubbing of the base of the post. The spurious events generated by this rubbing are received first by either channel 2 or 4. As discussed previously, the cracks are at the comers where sensors 2 and 3 are located.
62
ACOUSTIC EMISSION: STANDARDSAND TECHNOLOGY
Summary
The AE testing did not indicate significant activity near the sensors where the cracking is located. However, there was activity noted in the area directly between sensors 2 and 4. This could indicate that AE is detecting damage that was not detectable with dye and UT inspection techniques or that the bottom two sensors are picking up spurious data from mechanical rubbing. As was indicated in the Lehigh inspection of the coupon, the cracks actually initiated and grew from both the exterior and interior of the post wall. Although unlikely, the cluster of AE events between sensors 2 and 4 may indicate a crack that has initiated on the inside of the pole but is too shallow to be detected by UT. Unfortunately, the pole is still in use for testing and it was not possible to determine the actual extent of damage in the pole at the time of AE testing. The test presented herein did not provide conclusive results. The AE test reported in this paper used a vertically applied load. It is suspected that the fatigue damage is caused by both inplane and out-of-plane movement of the mast-arm. This would suggest that the load should be applied at an angle rather than vertically (FIG. 16). This may activate more damaged areas than vertical loading. One additional advantage is that this tends to stabilize out-of-plane movements caused by wind during testing. FIG. 16--Testing configuration.
Implementation Strategies
There are several logistical problems related to the application of AE to traffic signal structures. The biggest question regarding implementation is how to test the pole with a minimum interruption to traffic. AE requires a load application and the most convenient location for this load application is under the cantilever, which necessarily will involve interruption of traffic flow. One option is to perform the AE inspections when the signal bulbs are being changed. The Wyoming DOT normally changes bulbs every two years to prevent burn out in service. I f A E tests were conducted during the light changes then the traffic control that is already in place could be used to complete the AE test. Sensors could be applied and checked while the crew was changing lightbulbs. After the lightbulbs are changed, the AE test could proceed.
63
Acknow~dgmen~ The authors wish to gratefully acknowledge the continued support of this research by the Wyoming Department of Transportation and Federal Highway Administration.
Disclaimer
This paper represents the views of the authors and not necessarily those of the Wyoming Department of Transportation and Federal Highway Administration
References
[1] B. Patrick Collins, personal correspondence, State Bridge Engineer, Wyoming Department of Transportation, Cheyenne, WY. [2] McDonald, J. R., Mehta, K. C., Oler, W. W., Pulipaka, N., "Wind Load Effects on Signs, Luminaires and Traffic Signal Structures," Texas Tech University Research Study No. 11-5-92-1303, July 1995.
[3] Mike, C., Fisher, J.W., Slutter, R. G., "Fatigue Behavior of Steel Light-Poles," Fritz Engineering Laboratory Report No. 200.81.714.1, 1981.
[4] Association of American Railroads Operation and Maintenance Department, Mechanical Division, Procedurefor Acoustic Emission Testing of Tank Cars and IMIOI Tanks, Issue 6.
Integrity and Leak Detection/Location Methods
Ronnie K. Miller, 1 Adrian A. Pollock, l Peter Finkel, l Daniel J. Watts, 2 John M. Carlyle, 2 Anthony N. Tafuri, 3 and James J. Yezzi, Jr. 3
T H E D E V E L O P M E N T OF A C O U S T I C E M I S S I O N F O R L E A K D E T E C T I O N AND L O C A T I O N IN L I Q U I D - F I L L E D , B U R I E D P I P E L I N E S
R E F E R E N C E : Miller, R. K., Pollock, A. A., Finkel, P., Watts, D. J., Carlyle, J. M., Tafuri, A. N., and Yezzi, J. J., Jr., " T h e Development of Acoustic Emission for L e a k Detection and Location in Liquid-Filled, Buried Pipelines," Acoustic Emission: Standards and Technology Update, ASTM STP 1353, S. J. Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999.
ABSTRACT: Acoustic Emission leak detection and location for liquid-filled, buried pipelines was studied using various leak sources and laboratory reference standards. These standards provided significant data that have contributed to the development of test procedures for generating two-phase flow and a better understanding of the leak mechanism. Results are discussed for leak signal enhancement due to two-phase flow, leak orientation and the presence of backfill surrounding the pipe. Field studies were performed on buried pipes to demonstrate that small leaks (less than 0.1 gallons/hour or 0.1 mL/second) can be detected under the right test conditions and that the location of these small leaks can be determined using a variety of location methods. Among the different location methods examined was a location technique referred to as "tuned" linear location. In addition, a modification to the conventional Signal Difference location technique is discussed.
1 Executive Director, Director of Education and Training, and Research Engineer/Scientist, respectively, Engineering Services & Inspection Group, Physical Acoustics Corporation (PAC), 195 Clarksville Road, Lawrenceville, NJ 08648. 2 Executive Director and Acoustic Specialist, respectively, Emission Reduction Research Center, New Jersey Institute of Technology, 138 Warren Street, Newark, NJ 07102. 3 Research Program Coordinator and Physical Scientist, respectively, U.S. Environmental Protection Agency (EPA) NRMRL/WSWRD/UWMB, MS-104, 2890 Woodbridge Avenue, Edison, NJ 08837.
Copyright9
67 www.astm.org
68
KEYWORDS: acoustic emission, leak detection, source location, buried pipelines, liquid-filled pipelines, laboratory reference standards
The objective of this development program is to determine if Acoustic Emission can be used to detect small leaks, on the order of 0.1 gallons/hour (0.1 mL/second), during annual testing on liquid-filled, buried pipelines. An additional objective is to develop the technology so that a maximum sensor spacing on the order of 500 feet (152.4 meters) may be used to detect the minimum leak rate. A basic understanding of leaks and Acoustic Emission was given by Pollock and Hsu [1]. In order to better understand the leak mechanism, to evaluate existing leak detection and location methods, and to develop new methods, a series of tests and experiments were carried out. From the onset of this work, it became clear that it would be impossible to always study functioning, full scale pipelines due to the logistics for testing, the cost of travel and the need to control all of the test variables. The solution to this problem was to create small specimens that could be studied in the laboratory as well as to fabricate a test facility that would allow the study of long sections of pipe. For the laboratory, we designed and fabricated small test specimens that became known as laboratory reference standards. They were designed [2] with the idea that whatever we did in the lab could be repeated in the field. This meant that we could develop new ideas and study most of the variables that affect leak detection and location before any field studies were attempted. In addition, we found that the series of tests that we subjected the standards to produce very repeatable results. The success with the lab standards was primarily due to the introduction of what we call PAC Plugs. These devices allowed us to reproduce, in both the laboratory and the field, very controlled and repeatable leak conditions. This made possible the transfer of information gathered in the laboratory to the field and reduced the total amount of field time and travel costs expended in the technology development. The SERDP Experimental Test Pipeline Facility [3] located at the EPA site in Edison, New Jersey, offered the opportunity to test and evaluate ideas developed in the laboratory on full scale piping systems. These systems are comprised of long pipe runs, elbows, tees, and in some cases, expansion loops. In its second generation, this facility now contains a 2 inch (5.08 cm) diameter pipe testing facility; a 12 inch (30.48 cm) diameter pipe testing facility; and a double-wall piping system with a 4 inch (10.16 cm) diameter carrier and 12 inch (30.48 cm) diameter conduit. There is insulation between the carrier and the conduit for the double wall system and a section with only the 4 inch (10.16 cm) diameter cartier.
MILLER ET AL. ON LIQUID-FILLED,BURIED PIPELINES
69
Experimental Method
The experimental work associated with this program can be divided into two categories. The first is laboratory work performed at Physical Acoustics Corporation, Princeton, New Jersey, on small lengths of pipe. The second is field work that was performed at the pipe testing facility located at the Environmental Protection Agency in Edison, New Jersey. A description of this work follows.
Laboratory Studies
Two different specimens were fabricated for this program. They are referred to as laboratory standards due to the fact that we have been able to standardize the testing performed on them and utilize the results in the field testing that followed. Two different standards were fabricated from steel piping, assembled and subjected to a series of tests using water and air to internally pressurize the different pipe sections. Threaded plugs with small machined holes were inserted in the standards so as to provide a well defined and controllable leak. Leak Plugs-Three classes of leak simulators/generators were investigated in this program. Details of these simulators can be found in other publications [2]. The majority of the lab studies were performed with a device referred to as PAC Plugs. By varying the pressure across the orifice in the plug, we could program various leak rates varying from 0 to about 2 gallons/hour (2 mL/second). Two Inch Laboratory Reference Standard-The 2 inch (5.08 cm) laboratory reference standard was fabricated from 2 inch (5.08 cm) diameter, Schedule 40 steel pipe. It was designed to handle internal pressure due to gas, liquid or combinations of the two mediums. For most studies, a PAC Plug was used as a leak generator and was located in the middle of the pipe section as shown in Figure I. This experimental setup allows the pipe to be rotated so that leak orientation may be evaluated while maintaining the same source-to-receiver relationship. The Plexiglass| housing allowed us to visually monitor each experiment and observe the development of the leak excavation path. In addition, it gave us a means of adding different types of soil for studying the effects of backfill. It was quickly realized that the study of leak orientation was meaningless unless backfill was present. We therefore combined these two experiments into one. In addition to these studies, we performed attenuation and damping experiments. The details of these experiments and results will be presented in future publications.
70
Figure 1
Two inch Laboratory ReJerence Stanchlrd.
Four Inch Laboratory Reference Standard--The 4 inch (10.16 cm) laboratory reference standard was fabricated from 4 inch (10.16 cm) diameter, Schedule 40 steel pipe that was left over from the installation of the double-wall piping at the EPA's test facility. The same provisions were made for accommodating all types of leak simulators and generators as shown in Figure 2. The design also allows for varying leak orientation and backfill as with the 2 inch standard.
Figure 2 - Four inch Laboratory Reference Standard.
Field Studies
Field studies were carried out in Edison, New Jersey at the EPA's Underground Storage Tank (UST) facility. A section of 2 inch (5.08 cm) diameter buried pipeline was modified by adding an additional 150 feet (45.72 meters) of straight piping terminated by an elbow, a tee with a 30 foot (9.14 meters) branch, and a 150 foot (45.72 meters) return run. The details of the entire piping system are shown in Figure 3.
MILLER ET AL. ON LIQUID-FILLED, BURIED PIPELINES
71
New Section
9
;, 777 ~ 1
Outlet Inlet ~ /
40'
40'
Top View ExistingSection Exposed Section
Surface
Side View
Figure 3 - Full scale 2 inch pipe testing system.
Shortly after the modification was completed, a tightness test was performed to check for leaks. Several were discovered in the inlet and outlet fittings and repaired. One was accidentally created at the location of a "blank" plug (no leaking orifice). This accidental leak provided us with the opportunity to study small leak rates, in this case estimated at 0.014 gallons/hour (0.014 mL/second) by filling a graduated beaker over a fixed period of time.. The new modification included removable spool sections that were designed to accommodate the PAC Plugs at various orientations as shown in Figure 4. Typically only one PAC Plug with a leak orifice is used while the other holes have "blank" plugs installed. All PAC Plugs were installed using an O-ring under the collar of the socket head cap screw.
Drill & Tap 5 holes 1/4 - 28 thd

\ A-~
ii,iiiiiiiiiiiiil)iiiiii.iiiii,iiiil)) ..................
I ---12"
--~1 A~
Section A - A
20' typical
Figure 4 - Spool section designed for PAC Plugs.
72
Results
Two-Phase Flow
Early studies on the 2 inch standard were performed without any backfill. Under these test conditions, leak orientation seemed to have no effect on the Acoustic Emission (AE) signal that was detected at the end of the standard (see the sensor mounted on the right end in Figure I). In fact the AE RMS (Root Mean Square) signal was considered to be extremely low for the given leak rate of approximately 0. I gallons/hour (0.1 mL/second). This was accomplished using a PAC Plug and an internal test pressure of 10 psi (68.95 kPa). A review of AE leak detection technology with several AE Service companies provided us with the information needed to overcome this low signal problem. We found that it is common practice to use an inert gas blanket to pressurize liquid product during normal field testing. When we tried this procedure in the laboratory the results were very dramatic. As seen in Figure 5, data collected between 500 th second and 520 th second of a laboratory experiment shows that the AE RMS signal level was below 0.12 volts. During this time, the flow of water through the PAC Plug was well behaved and considered laminar. As gas began to mix with the leaking water, we observed a transition from laminar flow to two-phase flow. This transition took place between the 530 'h second and the 560 th second. After the 560 th second, the flow can be described more as a fine mist while the AE RMS signal had increased to a maximum of near 1.2 volts. The order of magnitude change in the leak signal provided us with vital information to guide us in the development of field testing procedures. As will be discussed later, this method did indeed help to detect very small leaks during the field test
vo.,
1.2
Pule Liquid Leak
Traalsition
Two Phase Leak
9.6[-1
7.81s
BE
I.
2.4s
11
D Time (seconds)
Figure 5 - Leak RMS signal (Volts) versus time (seconds).
MILLER ET AL. ON LIQUID-FILLED,BURIED PIPELINES
73
at the EPA's test site in Edison, New Jersey.
Leak Orientation and Backfill
The effects of leak orientation and backfill were studied using dry sand for backfill. With a PAC Plug installed, the standard filled with water, and a 60 psi (413.7 kPa) test pressure, measurements were made with the plug oriented upward, sideways and downward. No gas blanket was used during pressurization. The results of the AE RMS signal level measurements (using a 15 kHz resonant transducer) are shown in Figure 6. The behavior of the 2 inch (5.08 cm) standard and the 4 inch (10.16 cm) standard are much the same. The strongest signals were detected (see Figures 7 and 8) when the plug was oriented upward or downward while the sideways orientation always generated a significantly lower signal. Verification of this behavior during field studies is expected in future studies.
(a) 2 Inch Standal
9 20 psi 9 40 psi [] 60 psi
(b) 4 Inch Standard
9 20 psi 9 40 psi I"160 psi
. 7000 ] ~ 600o 50o0 9 3ooo

r 2000 loo
=L 4000 !
r ,oo~]
upward sideways downward ' upv,ard Leak Orientation sideways do',mv,ard Leak Orientation
Figure 6-RMS versus leak orientation and pressure.
Leak Location During Field Studies
A series of experiments were performed on the pipe testing system, shown m Figure 3, using water and a nitrogen blanket for pressurization. During one test, a "blank" PAC Pug (no leaking orifice) had been installed in one of the spool sections to plug up one of the threaded holes (see Figure 4). A grain of sand became trapped between the O-ring (used to seal the "blank" plug) and the pipe. This resulted in a small leak estimated at 0.014 gallons/hour (0.014 mL/second).
74
Figure 7 -
Waveforms collected from the 2 inch standard with the PAC Plug oriented." (a) upward," (b) sideways; and (c) downward.
75
Figure 8 -
Waveforms collected from the 4 inch standard with the PAC Plug oriented: (a) upward; (b) sideways," and (c) downward.
76
Two sensors were mounted on the pipe with 25 foot (7.62 meters) spacing. One sensor was located 1 foot (0.30 meters) from the leak and the other 24 feet (7.32 meters). Tuned Linear Location-Location of this small leak was initially attempted using a conventional linear location approach in real time. The results did not accurately reveal the location of the leak source. By varying the detection threshold for the AE channel with the highest hit rate (this corresponds to the sensor closest to the leak), we were able to tune the system and produce the linear location plot shown in Figure 9.
NJIT
LEgit I
PROJECT
-" E D I S O N
Oct
82,96
17:29:86
Events
,
36~
iiiiii: iiii iiiiilili iiii _ i iiiilliilll,.illi_.iillillllilllliiiil,illi

:: "
iilliiiiiiillliiill illl
..-ill
iii:i:i::::i:i::i:i:ii:ii:ii:::i::i:iii::iiill
-it i ii!iiii!iiiiiii!iiii iii iiiiiiii!iii!i

Leak Location
H
I Sensor 2 Location [
I
Sensor 1 Location [
J
Figure 9 - Results of tuned linear location of leak source located l foot (0. 30 meters)from sensor 1. The location of the leak as shown in Figure 9 is within 1 foot (0.30 meters) of the actual leak location. The amount of threshold adjustment is determined by matching the hit rate on the two sensors that straddle the leak. Modified Signal Difference-Previous studies of leak location [4] describe how leaks can be located when the leak source produces a continuous signal. By measuring the signal level at each sensor location and taking the difference, the leak location can be estimated by comparing the differential measurement with signal difference versus position curve generated from an attenuation plot. In this program, we extended this concept by applying the same principle to the peak Amplitude measurements taken from transient leak signals. The method cannot be utilized without an attenuation curve to reference the differential peak Amplitude measurements. Even more important, the attenuation curve must be developed for a distance equal to or greater than the sensor spacing used. In this study, we developed an attenuation curve over a distance of 150 feet (45.72 meters). This was done using: a 0.3 ram, 2H lead break source; a 0.5 mm, 2H lead break source; and a spring loaded center punch. The data was fitted together using a spline technique to produce the attenuation
77
curve shown in Figure 10. Applying the differential peak amplitude measurement technique gave us a leak location within 1 foot (0.30 meters) of the actual leak source.
9O 80 70 84
~so ~4o
2O 10
O' 50 100 D i s t a n c e (ft.) 150
Attenuation on 2" Pipe at 60 kHz. Figure 10 - Attenuation curve generated using three different AE simulators.
Conclusions
As a result of our studies, we have concluded that small leaks can be detected under the right test conditions. This became apparent after we had investigated the variables that affect the leak signal (orientation and backfill) as well as the use of a gas blanket for generating two-phase flow. The laboratory reference standards proved to be useful for gaining a better understanding of leak mechanics without having to study full scale systems. That understanding can be put to good use in the field to improve the chances of detecting and locating small leaks. We have found that several methods are available for leak location. The accuracy of threshold-dependent techniques ("tuned" linear location) and signal level techniques (modified signal difference) appeared the same for the cases that we studied.
Acknowledgments
The development of Acoustic Emission for leak detection and location in liquidfilled, buried pipelines was undertaken by Physical Acoustics Corporation and their
78
sponsor, the Emission Reduction Research Center, a National Science Foundation funded Engineering Research Center, located at the New Jersey Institute of Technology, Newark, New Jersey. Support and funding for this work has been provided by the Strategic Environmental Research and Development Program (SERDP). The participating SERDP agencies are: (1) The Environmental Protection Agency, National Risk Management Research Laboratory [EPA is the lead Agency]; (2) The U.S. Navy, Facilities Engineering Service Center; (3) The U.S. Army, Construction Engineering Research Laboratory; and (4) The Department Of Energy, Oak Ridge National Laboratory.
References
[ 1] [2]
[3] [4]
Pollock, A. A., Hsu, S-Y S., "Leak Detection Using Acoustic Emission," Journal of Acoustic Emission, Vol. 1, No. 4, 1982. Miller, R. K., Pollock, A. A., Watts, D. J., Carlyle, J. M., Tafuri, A. N., and Yezzi, J. J., Jr., "A Reference Standard for the Development of Acoustic Emission Leak Detection Techniques," accepted for publication in NDT&E International on May 17, 1998. SERDP Experimental Test Pipeline construction contract, awarded to Turner Construction Company, 31 March 1997, NJIT Purchase Order P801468. Miller, R. K. and Mclntire, P., Nondestructive Testing Handbook, Vol. 5, Acoustic Emission Testing, American Society for Nondestructive Testing, Columbus, Ohio, 1987.
Samuel J. Ternowchek,1 Thomas J. Gandy,2 Mauricio V. Calva,3 and Tom S. Patterson4
ACOUSTIC EMISSION AND ULTRASONIC TESTING FOR MECHANICAL INTEGRITY
REFERENCE: Ternowchek, S. J., Gandy, T. J., Calva, M. V., and Patterson, T. S., "Acoustic Emission and Ultrasonic Testing for Mechanical Integrity," Acoustic Emission: Standards and Technology Update, ASTM STP 1353, S. J. Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999. ABSTRACT: Today's owner/operator of process equipment is required to implement "good engineering tests" to ensure the mechanical reliabilityof his equipment. There are several ways this can be accomplished. One or more nondestructive testing (NDT) techniques, such as Acoustic Emission, Ultrasonics, Radiography, Magnetic Particle, Visual and others can be used. Most of these require the equipment be taken off line before the inspections can be performed. In some cases, when internal inspections are used, the cost of the inspection increases significantly. This paper describes an inspection program which uses Acoustic Emission (AE) as the primary technique for performing a global evaluation of the equipment's mechanical integrity. When the AE results indicate potential problems, a second NDT method, Ultrasonics (UT) shear wave analysis or thickness gauge, is utilized to determine the source mechanism, its characteristics and its boundaries. The two techniques complement each other quite well. This approach to equipment reliabilityoffers several advantages: 1) the test, in many cases, can be performed on-line with product; 2) the equipment is evaluated under its normal operating conditions; 3) different damage mechanisms such as cracking corrosion and embrittlement may be detected with a single test; 4) the cost savings can be substantial over other methods. In addition to describing the AE technique utilized, its advantages and disadvantages and an improved analysis method, there are examples of results produced on several recent tests, a 60 ft. diameter sphere, a 50 ft. diameter storage tank and a jacketed process vessel.
1 Director of PAQS, Physical Acoustics Corp., P.O. Box 3135, Princeton, NJ 08543. 2 Southeastern Regional Field Test Manager, PAQS, Physical Acoustics Corp., 600 Kenrick, Suite D2, Houston, TX 77060. 3 Northeastern Regional Field Test Manager, PAQS, Physical Acoustics Corp., P.O. Box 3135, Princeton, NJ 08543. 4 Senior Test Engineer, PAQS, Physical Acoustics Corp., P.O. Box 3135, Princeton, NJ 08543.
79
80
KEYWORDS: acoustic emission, pressure vessels, inspection, ultrasonics, NDT, process safety management, fitness for service
Introduction
In today's environment of company "right sizing" and reduced operating budgets, the need for improved, accurate and large-scale inspection techniques has never been greater. Engineering and plant inspection departments are in greater need for methods which can evaluate large and complex equipment quickly and provide indications of potential problem areas. These areas can then be focused on during shutdown or turnaround inspections. Since its commercial inception in 1968, one technology, Acoustic Emission (AE), has emerged from the laboratory as having the capability to meet this requirement. AE offers a number of advantages over conventional inspection techniques for inspecting plant equipment. These will be discussed in detail in this paper, however, a few include: in-service as well as new equipment inspection; global monitoring of large complex equipment; non-intrusive; and a measure of structurally significant anomalies. These features provide a valuable tool in the evaluation of an equipment's structural integrity and its fitness for service. Based on the results of an AE test, engineering and inspection personnel are better able to determine what, if any, further inspections should be performed and where. This can lead to the savings of inspection dollars by focusing the follow-up efforts in the areas in greatest need and not re-inspecting areas which are not in need of it. By definition, Acoustic Emission (AE) is the transient elastic wave generated by the rapid release of energy from a localized source within a material. Since its development in the early 1960's, AE has been growing in its use and acceptance by industry. From the laboratory tests of the 60's on through today's use in the field and laboratories, AE has been shown to be a valuable tool for detecting and understanding the response of a material or structure to applied stress. The basic concept of AE is shown in Figure I. A stimulus is applied to a material or structure so as to cause localized yielding. This yielding will release a stress wave which propagates elastically through the structure. At some point it reaches the surface and stimulates the piezoelectric sensor. This sensor converts the mechanical energy to an electrical signal which can then be amplified and processed for analysis. AE testing differs from other NDE in two ways: 1) the signal that is detected is generated by the material itself and 2) the method is evaluating the response of the material or structure to applied stresses, hence it is a dynamic technique. These two factors provide the basic concept for applying AE to vessel testing and other structures.
Advantages of AE
There are a number of advantages that AE offers for the inspection of pressure vessels.
TERNOWCHEK ET AL. ON ULTRASONICTESTING
81
Acoustic Emission (AE) Process
Stimulus
Stimulus
Measurement
4tm
ousfic sion
ve
9R e c o r d i n g 9Interpretation 9E v a l u a t i o n
FIG. 1 -
Definition: Acoustic Emissions are transient elastic waves generated by tln~ rapid release of e1~gy from localized sources within a material.
Some of these include: a) In-service test - Acoustic emission is applied in three ways to pressure vessels: 1) During initial hydrostatic testing after fabrication; 2) when equipment is being recertified for alternative use and 3) in-service inspection. The most often used is the in-service. AE provides the ability to evaluate the structural integrity of a vessel on-line, under the operating conditions it experiences while in use. This valuable information, as to how the vessel is performing in-service and whether degradation is occurring while in use, can help optimize when and where other inspections are performed. Its use can either extend the time between other inspections or provide advance warning of an impending failure. b) Test entire structure - by carefully determining a given sensor's area of coverage, complete equipment can be tested with a relatively small number of sensors. This capability of AE makes it most attractive for large columns, spheres and heat exchangers where access and inspectability are very difficult and time consuming. c) Detect structurally significant defects - because AE is utilized under the actual loading conditions the structure experiences in-service, it has the capability to differentiate anomalies that are growing versus those that are dormant or structurally insignificant. With other techniques, it is possible to establish the presence of an anomaly but it is difficult to determine its effects when the structure is in-service.
82
d) Measure of structural severity - utilizing today's testing procedures and analysis, it is possible to provide a measure of the structural severity of an active AE source. The most ot~en used programs are MONPAC| ~ and MONPAC PLUS| These programs provide a means of determining the structural significance of the AE data when testing to prescribed procedures. A grading of the AE data is provided with recommendations for follow-up inspections. e) Non-invasive - the AE test uses sensors that are mounted on the outside of the vessel wall. There is no requirement to have access to the inside of the vessel when performing the test. On insulated vessels, only small access holes are required for sensor attachment, the remaining insulation remains undisturbed. f) Permanent record - the AE test instrument records data in to files which are stored on the computer. These files are available for replay and further analysis. They can be archived and stored for future reference. A library of files can be compiled for comparing similar type structures or comparing a number of tests over several years on the same structure. Types of Problems Detected with AE AE is a stress-related phenomena. In order to detect discontinuities and damage in equipment, the technique must be applied when the equipment is being stressed. It cannot detect problems in areas where there is no loading of the structure. In fact, careful attention must be paid to the loading schedule ifAE testing is to be successful. The stress concentration created by defects will produce the localized yielding, which is detected by the sensors, while the remaining structure is silent. Since different defect types have different stress concentration factors associated with them, some will emit AE at a lower load level than others, hence, they are more easily detected. Some examples of detectable defects in new equipment include, but are not limited to, the following: a) Welding defects - there are a number of weld problems that produce AE during hydrostatic testing. These include cracks developed due to hot cracking, cold cracking, intergranular and transgranular microcracks, base metal cracking in the heat affected zone (HAZ), incomplete fusion, undercut, lack of penetration, porosity and inclusions. As noted above, the ability to detect these defects is dependent on the stress concentration factor of the defect and the ability to produce localized yielding at the stress levels applied. It is very possible to have a defect type be very emissive in one case and not as emissive in another case, simply because of this. b) Casting defects - again, as in welding, casting defects that produce stress concentrations may be detected by AE. These include: gas and blow holes which produce porosity and cavities; inclusions such as scale oxides or dross entrapped within the casting; shrinkage cavities; hot tears and cracks due to unfused chaplets; dissimilar metal chills and weld repairs. c) Forgings - A number of forging defects may be detectable with AE should it be loaded sufficiently to produce localized yielding. Some of these are: laminations caused by inclusions or blow holes in original ingots; laps caused Monsanto Corp., St. Louis, MO, and Physical Acoustics Corp., Princeton, NJ.
83
by metal protrusions left from previous passes; cracks caused by rolling or forming and certain types of notches. d) Heat treat - when heat treating fails to reduce residual stresses or produce the desired changes in metallurgical structure, a significant amount of AE can be generated during the hydrotest. Failures in the heat treating process which produce localized stresses, excessive variations in metallurgical structures, or failure to stabilize specific phases, all have the potential to generate AE. In addition to new fabrication, AE is used extensively as an in-service test technique. While it would be of historical value to have AE test information on equipment from the time of fabrication throughout its life, it is not necessary. Much equipment is tested after being placed in use with substantial results. Typically an in-service test is intended to detect and locate defects that are growing under the service loads. Some oftbese are: a) Mechanical fatigue - variation in process may cause mechanical fatigue damage to a structure. The variation can be in internal pressure or pressure cycling. It can be a result of variation of process flow. It can also be external as in the case of vibrations. The result is to initiate and grow cracking under the inservice conditions. The purpose of this AE test is to detect and locate these cracks prior to catastrophic failure. b) Thermal fatigue - variation in process temperature can cause crack initiation and growth. The crack initiation can be the result of mechanical load concentration in areas such as stiffness, nozzles, hangers and knuckle areas. These areas experience high local flexural stresses as a result of expansion and contraction. Or, it can be the result of creep damage. It is the case that thermal damage is one of the main service related damages detected with AE testing in utilities and refineries. c) Hydrogen damage - as opposed to thermal fatigue, hydrogen damage and hydrogen embrittlement results from the formation ofmicrofissures caused by the absorption of hydrogen by the steel vessel. Hydrogen damage causes a loss of ductility and strength in the steel. It has been shown that AE can detect hydrogen damage during in-service testing. d) Hydrogen blistering - this is the result of atomic hydrogen contained in a void or lamination of a steel and changing to molecular hydrogen. As the amount of molecular hydrogen increases, pressures in the void increase causing internal splitting and surface blisters. Left undetected, rupturing of the vessel could result. AE has been able to detect hydrogen blisters at a relatively early stage. e) Stress Corrosion Cracking (SCC) - this phenomenon which results from the effects of corrosion and tensile stress has been detectable by AE in a number of different materials and processes. SCC typically produces a stable cracking process that increases the probability of detection with AE. It has been successful in detecting SCC in austenitic stainless steel as well as SCC in carbon steel used for ammonia vessels. f) General corrosion - during the in-service test of many vessels, tanks and railroad tank cars, acoustic emission is generated as a result of spalling of corrosion product. Unlike the above applications, the signal source in this case may not be the localized yielding stress of a defect, but a result of the frictional rubbing of the corroded product. This does produce a detectable signal that
84
can be used to indicate the presence of corrosion.
Testing Method
An outline of a typical testing procedure is shown in Table 1. It shows the sequence of activities that would normally be performed. While different structures (e.g. piping, vessel, tanks, etc.) may have procedures specific to themselves, the outline shown can be utilized in many cases. TABLE 1--Test procedure.
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.
Determine area of coverage for a sensor based on signal attenuation on the tank. Mount sensors based on above area of coverage. Perform equipment verification check. Determine test load level and load schedule Perform background noise check. Begin filling the tank according to the specified load schedule. Verify quality of AE data. Complete loading. Evaluate test data per evaluation criteria. Determine location of active sources. Determine whether any follow up NDT is required.
A key step in the test procedure is number 9, "Evaluation of test data per evaluation criteria." This is an important step since it is here that the tester must decide whether a problem exists in the test structure. The most often used criteria today is MONPAC| or MONPAC PLUS| This is a technology program developed by Dr. T. Fowler 1~1at Monsanto Chemical Company and marketed to industry by Physical Acoustics Corporation (hence MON-
85
PAC). A typical evaluation would comist of five criteria, including: 1. Activity during hold periods - this can indicate continuing yielding or damage. 2. On going activity during loading - this can indicate widespread damage such as corrosion. 3. Amount of activity - can indicate presence of defects. 4. Large amplitude hits - can indicate presence of a growing defect. 5. Cumulative energy- c a n indicate that a defect area is responding to increases in stress. Within this technology package is a method that allows the significance of the data to be determined should it not pass all of the above criteria. Based on the energy parameter, a "Severity" and a '~l-Iistorical Index" value are calculated and plotted on an empirical scale. The scale is graded from A to E with a corresponding significance level. These are: A) Nfinor indication. No follow up required. B) Note for future tests. Check for surface defects such as corrosion, pitting, etc. C) Indication requiting follow-up evaluation which can include further data analysis. D) Significant indication. Follow up with NDT. E) Very significant indication. Discontinue operations until NDT inspection performed. Using the above criteria, the AE tester has a means of evaluating his test results and determining whether any follow up action is required. This has worked well in many situations. With the performance of a large number of tests, PAQS has developed a large database of test results. Periodic reviews of the database allowed the data techniques to be refined and, in fact, a complimentary analysis developed. This analysis is based on the characteristic AE response to stress levels. Designated the Emission Source Rating (ESR), it is a method of analyzing the source response to stress. It has been observed that the emissions from an anomaly will increase as stress levels increase. Using this as a basis, a numerical scale was developed which complements the better grades of the MONPAC| System. The ESR ratings are: 0. InsutScient data. 1. Not responsive to stress; no follow up required. 2. Response to stress but not indicative of significant defect growth. No follow up required. 3. Response to stress. Shows moderate increase with stress. Follow up NDT recommended. 4. Response indicative of significant yielding. Follow up NDT required. By using the combined analysis, a more complete test result can be produced. The following are three examples of the use of both the MONPAC| and ESR analysis for determining what areas of the structures should be subjected to follow up UT inspection.
A. Above Ground Storage Tank

A carbon steel storage tank of approximately fitty feet in diameter and thirty feet tall was AE tested. The tank was used to store caustic soda solution. The tank had a rubber liner that deteriorated. The caustic solution leaked and attacked the tank walls near the base. The tank was removed from service and the liner replaced. During replacement, repairs were made in the area where damage was noted. Atter repairs were completed, the tank was relined and a
86
hydro test performed. AE monitoring was performed during the hydro test. Upon completion of the test, the data was analyzed using the two methods outlined above. The results are listed in Table 2. In these areas, where both methods recommended follow up be performed, UT was conducted. The areas are shown on Figure 2. This is an unwrapped view of the tank. The four areas for follow up UT inspection are noted at the bottom of the tank. The UT results showed indication that varied in length from 0.75 inches to six feet. Depths varied from 20~ of wall thickness to 50~ These results were very important to the tank owner. Further repaks were made to the tank before putting it back in service.
N W S Ir
I
27
X
I
~
X
I
30
X
I
32 X ~ X ,'44 X 35 X 36 X
29
31
~ S L ~ N S O R SPACING 1~7" 18
X X
17 X ' ~
19
X
20
X
21
X
~
X
23
X
24 X
25 X
26
SI~/SOR S P A C I N G '!.5.7"
(%
n 'it'x fl'x~n
K)I,LOW.T.,'P.'It~EA#1
x n
'
x '~ "Im
nnmx
,3
,,
n x I1 x 11
,,
,,
rOLLOW,.IJ~'~ t4l
I~OLLOW-UPA,.REA t-'OI.J.OI~'.UPAIIh,A It~l it2
I ~--SF..NSOR SPACLNG 9,Jl'
FIG. 2 - Caustic storage tank shell; A E sensor locations; NDE follow-up locations.
B. Sphere
A carbon steel sphere was tested while AE monitoring was in place. The sphere was fitSy-six feet in diameter and was supported on the ten tubular legs with "X" braces. The sphere was used to store LPG It was tested with LPG while in service. Based on previous experience with this type of structure, a/l ESR ratings of three or gTeater were inspected. The ESR rating took precedence over the MONPAC| for this test. The results produced three areas of activity that were UT inspected. In two areas, cracks ranging in length from one half inch to two and a half inches were detected. These were deemed unacceptable and required immediate repaks. The third area had wall thinning up to 30% of the wall thickness. The results are shown in Figure 3.
TERNOWCHEK ET AL. ON ULTRASONIC TESTING
87
TABLE 2--Combined evaluation summary- initial load data. Sensor M-P Evai. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 Fail Fail Fail Fail Fail Fail Fail Fail Fail Fail Fail Fail Fail Fail Fail Fail Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass M-P Int. C C C D A C C C C C D D D C C B N/A INSIG INSIG INSIG N/A INSIG INSIG INSIG INSIG N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A M-P Sev. 119.70 142.80 196.10 186.90 33.70 112.60 140.20 208.00 208.00 252.20 344.70 339.60 307.30 261.10 204.40 69.20 ... 18.50 11.80 16.90 ... 11.50 06.80 07.60 ... ... 03.00 09.00 07.00 ... ... 26.00 ... 07.00 11.00 ... DTI ESR 2 2 3 3 2 2 3 3 3 2 3 3 3 3 2 2 0 1 1 1 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 Recommendations
Note on future AE test Note on future AE test Follow up-inspection Follow up-inspection Note on future AE test Note on future AE test Follow up-inspection Follow up-inspection Follow up-inspection Note on future AE test Follow up-inspection Follow up-inspection Follow up-inspection Follow up-inspection Note on future AE test Note on future AE test No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up No follow-up
M-P = MONPAC-PLUS| ESR = Emission Source Rating C/D = Corrupted data
DTI = Dunegan Testing & Inspection INSIG = Low levels o f AE activity N/A = Not available due to low emission levels
88
Wall Thinning~
up to 30%
Cracking
V2"-1"
TOP HEMISPHERE
Cracking
!i
BO'ITOM HEMISPHERE F I G . 3 - Defect location.
C. Horizontal Vessel
A horizontal vessel was tested using Acoustic Emission monitoring. The vessel was thirty seven feet long and ten feet in diameter. It is a phosphate processing vessel. The vessel is insulated. The test was performed in service using product and increasing pressure. The AE results showed primarily low level activity. The MONPAC| results were "A" ratings which would indicate minor sources. The ESR ratings were similar except for the sensor on the top of the vessel. This was sensor number eight. This area was followed up with UT thickness. The results indicated wall thinning was occurring in the vapor space of the vessel. Additional readings were taken to determine the extent of the degradation. These results are shown in Figure 4.
TERNOWCHEK ET AL. ON ULTRASONIC TESTING
89
North End
Bottom 9 Top 3
13 10 4 6
Bottom 12
FIG. 4 - Defect location.
Summary
AE has been shown to be a very useful tool in evaluating the overall structural integrity of plant equipment. The technique is fast, cost effective and relatively easy. AE allows the UT inspector to focus his efforts in the areas where anomalies which affect structural integrity are located. He does not perform inspection in areas where there aren't any problems. It should be noted that in the examples provided, the UT inspector was aided by the AE tester when performing his follow-up. This proved to be very useful since the AE inspector provided insight for the UT indications that might not have been obvious otherwise. The two working together made UT indication analysis more complete.
Acknowledgment
We are deeply grateful to our fellow employees at Physical Acoustics Quality Services for their assistance in preparing this paper. We also want to thank Ms. Janet Beyrouty for her patience in preparing this paper.
REFERENCE: [1] Fowler, T. J., et al., "The MONPAC | System", Journal of Acoustic
Emission, Vol. 8, No. 3, 1989. Acoustic Emission and Ultrasonic Testing for Mechanical Integrity, ASTM STP 1353, Sam Ternowchek, Physical Acoustics Corporation, Princeton, NJ, 08543.
AE Sensors, Standards, and Quantitative AE
Hajime Hatano l CALIBRATION OF ACOUSTIC EMISSION TRANSDUCERS BY A R E C I P R O C I T Y METHOD
REFERENCE: Hatano, H., "Calibration of Acoustic Emission Transducers by a Reciprocity Method," Acoustic Emission: Standards and Technology Update, ASTM STP 1353, S. J. Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999. ABSTRACT: The paper reviews and discusses the background, methodology and application fields of absolute calibration of acoustic emission transducers by means of the reciprocity method. For characterizing and calibrating acoustic emission transducers, a number of methods utilizing various mechanical sound sources appear in the literature. Reciprocity calibration of acoustic emission transducers was proposed some time ago, and has since been employed by transducer manufacturers and laboratories An outstanding advantage of the reciprocity method is that absolute sensitivity, including frequency characteristics and impulse responses, both to Rayleigh wave and longitudinal wave, can be determined by means of purely electrical measurements without the use of mechanical sound sources or reference transducers. The procedure for calibration was standardized as NDIS 2109 by the Japanese Society for Non-Destructive Inspection. KEYWORDS: acoustic emission, transducer calibration, reciprocity method, Rayleigh wave, longitudinal wave Calibration of acoustic emission transducers is of significant importance in acoustic emission measurements, not just for quantitative evaluation, but also for mutual comparison of data obtained by different laboratories [1-3]. For characterizing and calibrating acoustic emission transducers, a number of methods utilizing various mechanical sound sources, for instance, a falling steel ball, a breaking pencil lead, a sand blast, and an electric spark, appear in the literature [4]. However, by means of these methods it was difficult to obtain the absolute sensitivity because the characteristics of the sound sources were not clear and the reproducibility was inadequate [5]. A seismic surface pulse method was developed by Breckenridge and his colleagues, wherein the i Professor, Department of Applied Electronics, Science University of Tokyo, Yamazaki, Noda, 278-8510 Japan.
g3
94
t-
"["
r ~
"
2/'r
(a)Moment tensor. (b)Displacement. (c)Frequency spectrum. FIG. 1 -- Schematic illustration of acoustic emission generation. breaking of a glass capillary is employed for the sound source and a capacitive transducer is used for the reference [6, 7]. Reciprocity calibration of acoustic emission transducers was proposed more than two decades ago, and has since been employed by transducer manufacturers and laboratories [8, 9]. An outstanding advantage of reciprocity calibration is that absolute sensitivity, including frequency characteristics and impulse responses, both to the Rayleigh wave and longitudinal wave, can be determined by means of purely electrical measurements without the use of mechanical sound sources or reference transducers. Once reciprocity calibration has been carried out. sensitivity of an optional transducer, which is not necessarily reversible, can be determined by a relatively simple procedure by using a calibrated transducer as the reference for sound transmission or reception. The procedure for reciprocity calibration was standardized as the ND1S (Standard of the Japanese Society for Non-Destructive Inspection) 2109: Methods for Absolute Calibration of Acoustic Emission Transducers by Reciprocity Technique [10]. The present paper reviews and discusses the background, methodology and application fields of absolute calibration of acoustic emission transducers by means of the reciprocity method. Transducer Calibration in Acoustic Emission Measurement In an analogy of seismology, generation of acoustic emission is attributed to the change of moment tensor due to the release of inelastic strain such as crack propagation and plastic deformation in solids [11]. Figure 1 shows a schematic illustration of the relation between the change of moment tensor, resultant displacement waveform of acoustic emission, and frequency spectrum of the waveform. In figure, r r is the transition time of the moment tensor, namely, the duration time of each crack propagation or dislocation movement. Since the pulse width of the acoustic emission corresponds to r ,, the distribution range of the frequency spectrum is inversely proportional to r r [1]. If the duration time is supposed to be O. i/L s, the frequency spectrum will attain a range as high as I0 MHz.
HATANOON A RECIPROCITYMETHOD
95
FIG. 2 -- Propagation of acoustic emission waves. The waveforms and frequency spectra of electrical signals detected by acoustic emission transducers are, however, quite different from those shown in Fig. 1. Figure 2 shows the propagation paths of the acoustic emission from the source in the object to the transducer attached to the surface. The acoustic emission undergoes reflections and mode conversions along the propagation paths. This means that the transfer function from the source to the transducer is complicated in both the time and frequency domains. Figure 3 shows how the frequency spectrum of the original acoustic emission is deformed during propagation and detection. The detected spectrum is expressed as the product of the original acoustic emission spectrum, and transfer functions of the propagation paths and transducer sensitivity. The detected signals are strongly affected by the characteristics of the transducer, since the frequency bandwidth of the transducer,
C~JACOUSTIC EMISSION
TRANSFER FUNCTION
FREQUENCY
FREQU~ENCY
@ TRANSDUCER SENSITIVITY
0 ELECTRI CAL S I GNAL

@x|174
IL
-~-NOISE
FREQUENCY
FREQUENCY
FIG. 3 -- Schematic illustration of acoustic emission spectrum.
96
which usually employs a piezoelectric-ceramic element, is much narrower than that of the original acoustic emission spectrum.
Definition of Transducer Sensitivity

A definition of transducer sensitivity is important for interpreting the results obtained by the transducer. For conventional microphones and loudspeakers, which are usually placed in gaseous or liquid medium, sound pressure is taken as the measure of sensitivity. Since acoustic emission transducers, in contrast, are mounted on the free surface of a solid medium, it is difficult to determine their pressure sensitivity. Either displacement amplitude, velocity, or acceleration should be taken as the measure of sensitivity for acoustic emission transducers [8]+ For instance, free-field velocity sensitivity of the acoustic emission transducer M,, is defined as M,, E,, w,,
( 1)
where Eo is the output open-circuit voltage, and w,, is the vertical component of displacement velocity of incident elastic wave at the point where the transducer is to be placed. Vertical components are most important for calibration, since acoustic emission transducers are generally attached to the object by means of liquid couplant, and very small portions of horizontal components are transmitted to the transducers.
Wave Mode for Calibration

In a bounded elastic medium, as shown in Fig. 2, various wave modes are possible, while acoustic emission transducers generally assume different sensitivities to different wave modes. For estimating actual characteristics of transducers, it is necessary in the calibration to use the wave mode identical to that of the acoustic emission waves detected in the actual object. In plate objects such as vessel walls, it was shown that Rayleigh waves or Lamb waves are dominant [8]. As for these wave modes transducer sensitivity is subject to the aperture effect. Figure 4 shows the mechanism of the effect, where the crests and
TRANSDUCER A
TRANSDUCER B EIGH/LAMB ~AVE
FIG. 4 -- Aperture effect on Rayleigh / Lamb wave.
HATANO ON A RECIPROCITYMETHOD
troughs of the incident Rayleigh or lamb waves cancel out each other within the transducer aperture. Recently, application fields of acoustic emission measurements have expanded to include various bulky objects, for instance, thick-walled vessels and huge concrete structures. For such applications, sensitivity to longitudinal waves is of primary importance. In addition, longitudinal wave sensitivity represents the fundamental characteristics of acoustic emission transducers, since these transducers usually employ thickness-mode piezoelectric elements. In the actual acoustic emission measurements, acoustic emission waves of various modes, in correspondence to the shape and size of the object and to the positional correlation between the source and transducer, are introduced into the transducer. To carry out calibration of sensitivities to all the incident waves is too complicated. Characteristics of acoustic emission transducers are represented by both Rayleigh wave and longitudinal wave sensitivities [9,10]. Method of Reciprocity Calibration Figure 5 shows the fundamental aspects of the method of reciprocity calibration Three reversible transducers 1, 2, and 3 are prepared, and three independent transmission/reception pairs are configurated through a transfer medium. The magnitudes of the transmission signal current and reception signal voltage, lij and E~, respectively, are measured on each pair, where the subscript ij corresponds to transducer i for transmission and j for reception. If the reciprocity parameter H, which is dependent not on the transducer design but on the mode of elastic waves, constants of medium, and definition of sensitivity, is given, absolute sensitivity is determined by purely electrical measurements.
97
[12-14].
TRANSDUCER
1
TRANSDUCER / / ~
2
TRANSFER
HEDIUM
TRANSMSSION I
RECEPT ON I
FIG. 5 -- Three transmission/reception pairs for reciprocity calibration.
98
For instance, magnitude of sensitivity of transducer 1, M m, is given as

\ I/2 Mol = El2 123
Ii2 E2~ Reciprocity parameters essential for Rayleigh wave and longitudinal wave
(2)
calibrations, H Rand Ht,, respectively, were derived as a function of frequencyfas follows [8,9]: 1 + O" ( 2 ~ i/2 (3)
HR = 2~__.E___kRX~ ~RD~ )
H,. = 2 f (1 + O')(I - 20")
E(I-o)D,
(4)
where Here, Dk and Dr, are the distances between the transmission and reception transducers in Rayleigh wave and longitudinal wave calibrations, and E, ,~, and p are Young's modulus, Poisson's ratio, and density of the transfer medium, respectively. X and Y are constants which have been obtained from the numerical solutions to Lamb's equation as a function of Poisson's ratio [9,15]. For both the Rayleigh wave and longitudinal wave calibrations, a cylindrical solid block is commonly used as the transfer medium [9]. Figure 6 shows schematically the oRo, 9 L' Ro'. []
7
assumed propagation paths.
"'--.s?. Lo
[]
(b)Longitudinal wave. (a)Rayleigh wave. FIG. 6 -- Transducer arrangements on a cylindrical medium and
HATANO ON A RECIPROCITYMETHOD
99
transducer arrangements on the medium, and assumed propagation paths of various elastic waves. Figure 6 a shows the setup for the Rayleigh wave calibration. Both the transmission and reception transducers are placed on the top plane of the cylindrical medium with a distance DR apart from each other, and the direct Rayleigh wave R 0 is employed for the calibration. The longitudinal wave L' reflected at the bottom of the cylinder, and the Rayleigh wave R0' reflected at the edge of the top plane, are assumed to be possible spurious waves, which reach the reception transducer subsequent to the direct Rayleigh wave Ro. Figure 6 b shows the setup for the longitudinal wave calibration. The transmission transducer is placed on the top plane of the cylindrical medium, and the reception transducer on the bottom plane where their axes coincide with each other. The direct longitudinal wave Lo is employed for the calibration here. The longitudinal wave L~' reflected at the side of the cylinder, and the shear wave Sz' converted from the longitudinal wave ~ are supposed to be possible spurious waves.
Transmission Signal and Transfer Medium

Tone burst signals with the squared-sine envelope, in place of continuous waves, were employed for the transmission signal in order to discriminate between the direct wave and the subsequent spurious waves on the basis of their arrival times [9]. The bandwidth of the tone burst, which determines the frequency resolution of calibration, decreases with increasing duration. However, maximum duration of the tone burst is limited by the minimum difference of the arrival times. Namely, half of the duration of the squared-sine envelope should not exceed the time difference in order to measure the amplitude of the direct wave signal from the peak value without interference from the subsequent spurious waves. As a larger dimension of the transfer medium causes a greater difference in the arrival times, a large cylinder of forged steel ( E = 2.1x10 lj N/m 2, o" = 0.28, p = 7.7x103 kg/m 3) with a diameter of 1.1 m, height of 0.76m and weight of about 6 tons was prepared [9]. Ultrasonic testing was conducted at 2 MHz throughout the block and no detectable flaws were recorded. In addition, ultrasonic attenuation was measured in order to confirm that mechanical losses of the medium was small enough for the calibration. The transfer medium was designed so that the difference of arrival times of the direct and spurious waves was longer than 0.1 ms for any case in either the Rayleigh-wave or longitudinal-wave calibration. Duration of the tone-burst signal was set to 0.2 ms. Finite difference simulation was conducted using an axisymmetric model to study the wave propagation in the transfer medium. Figure 7 shows results for the longitudinal wave calibration, wherein the 10-mm-diameter sound source was placed on the top plane. In Fig.7 a, the direct longitudinal wave L0 reached the bottom of the medium with a propagation time of about 133/z s. Figure 7 b shows the result at 234/~ s, where the reflected longitudinal wave Lj' reached the center of the bottom.
100
FIG. 7 -- Finite difference simulation of wave propagation in a cylindrical
transfer medium.
Calibration Results Figure 8 shows the calibration results for one of the three identical transducers (0.14ZIO), which employed a 10-mm-diameter cylindrical piezoelectric element with a nominal thickness resonance of 140 kHz [9]. In the figure, absolute sensitivities were shown as a function of frequency; the bold line represents the Rayleigh wave s e n s i t i v i t y and the thin line the longitudinal wave sensitivity. Below about 150 kHz the two sensitivities were consistent. On the other hand, beyond that frequency, the Rayleigh wave sensitivity rapidly decreased in comparison with the longitudinal wave sensitivity due to the aperture effect described in the preceding section.
40
==
8 30
v
20
z r~ ',./3 F-
IO
RAYLE GH I ~AVE
I I
{./'}
I00
200
FREQUENCY ( k N z )
FIG. 8 -- Absolute sensitivity of O.14Z10 transducer.
LONGITU
!
300
400
HATANO ON A RECIPROCITY METHOD
101
Comparison with Surface Pulse Method

For the seismic surface pulse method, theoretical dynamic displacements of the surface were calculated on the basis of Lamb's theory. The reciprocity parameter for the Rayleigh wave calibration was derived from Lamb's theory as well [15]. There is a common theoretical basis in the two calibration methods. For the Rayleigh wave calibration, round robin experiments were carried out in a collaborative effort between the United States and Japan [16]. Six transducers, of two different types, were each calibrated three times by the surface pulse method, and three times by the reciprocity method. They were then recalibrated at NBS to assure that no changes had occurred. Although the procedures are very different, absolute sensitivities of the transducers as obtained by either method agreed remarkably well. Both the surface pulse method and reciprocity method have merits and demerits. The procedure for each method was specified as Standard, namely, ASTM Designation E 1106: Standard Method for Primary Calibration of Acoustic Emission Sensors, and NDIS 2109, respectively [5,10]. There are other calibration methods, such as the laserinterferometer standard transducer, which could also be documented as standard. In general, the standard might leave the choice of method open with some provision for adding other methods to the list if they become feasible.
Measurement of Impulse Response by Reciprocity Method

By means of reciprocity method, impulse responses of acoustic emission transducers, which are essential for the quantitative analysis of acoustic emission waveforms [2,3,17], can be determined [18]. On the basis of the newly derived complex reciprocity parameters, frequency characteristics of phase, in addition to magnitude, of absolute sensitivity were obtained, and the impulse responses were determined through inverse Fourier transform. Three acoustic emission transducers (0.67Z3), which employed a 3-mm-diameter
2.0
RAYLEIGH W V AE
4.0
LONGI TUDI NAL WAVE
N"
~
1.0
~v
0.0
o.o
-2.0
~,. -1.0 (./3 ~, -2.0 0.0
"I
-4.0
t i
si0
10.0
15.0
20.0
25.0
TIME (~)
TIME (~,s)
FIG. 9 -- Impulse responses of O.67Z3 transducer.
102
disk-type piezoelectric element with a nominal thickness resonance of 670 kHz, were used as the objects for the reciprocity calibration of impulse response. Figure 9 shows the impulse responses of one of the transducers to Rayleigh and longitudinal waves [18]. The duration and period of oscillations of the impulse responses were somewhat different between the two wave modes. These results suggest that waveforms of detected acoustic emission are liable to change with the wave mode even if the waveforms of the incident elastic waves are identical. Based on the impulse responses determined by the reciprocity calibration, matched filters were constructed [19]. Consequently, the signal to noise ratio was enhanced in the actual acoustic emission measurements, and acoustic emission signals and spurious noises were discriminated between.
Conclusion
The background, methodology and application fields of reciprocity calibration of acoustic emission transducers were reviewed and discussed. A significant advantage of the reciprocity method is that frequency characteristics and impulse responses of absolute sensitivity to Rayleigh and longitudinal waves can be determined by means of purely electrical measurements without the use of mechanical sound sources or reference transducers. In addition, the transfer medium is not limited to electrically conducting material, since no reference such as the capacitive transducer is required to measure the displacement of elastic waves. Mechanical losses of the transfer medium were not considered in the calibration, since forged steel with a comparatively low mechanical loss was employed. For more precise measurements or for transfer media with higher mechanical losses, for instance, concrete and composite materials, the effects of the mechanical losses and, if possible, methods for compensating for the losses should be important subjects in the next stage of research.
Acknowledgments
The author wishes to thank T. Watanabe and S. Hashirizaki of Nippon Steel Corporation for fabricating the forged steel transfer medium, and T. Chaya, S. Watanabe, K. Jinbo, and M. Koshimura of Science University of Tokyo for their efforts in making experiments. He further acknowledges D.A. Watson for reading the manuscript.
References
[/]
Hatano, H., "Quantitative Measurements of Acoustic Emission Related to its Microscopic Mechanisms," J.Acoust.Soc.Am., Vol. 57, 1975, pp. 639-645. Hsu, N.N., Simmons, J.A., and Hardy, S.C., "An Approach to Acoustic Emission
[2]
HATANO ON A RECIPROCITY METHOD
103
Signal Analysis - Theory and Experiment," Mater.Eval., Vol. 35, 1977, pp. 100106.
[31
Scruby, C.B., Collingwood, J.C., and Wadley, H.N.G., "A New Technique for the Measurement of Acoustic Emission Transients and Their Relationship to Crack Propagation," J.Phys.D: Appl.Phys., Vol. 11, 1978, pp. 2359-2369. Hsu, N.N. and Breckenridge, F.R., "Characterization and Calibration of Acoustic Emission Sensors," Mater.Eval., Vol. 39, 1981, pp. 60-68. Hatano, H., "Calibration Standardization of Acoustic Emission Transducers," Proc.of14th Inter. Cong.on Acoust., Beijing, Vol. 4, 1992, Invited Paper, No. LI-3. Breckenildge, F.R. and Greenspan, M., "Surface-Wave Displacement: Absolute Measurements Using a Capacitive Transducer," J.Acoust.Soc.Am., Vol. 69, 1985, pp. 1177-1185. Breckenridge, F.R., "Acoustic Emission Transducer Calibration by Means of the Seismic Surface Pulse," J.Acoust.Emiss., Vol. 1, 1982, pp. 87-94. Hatano, H. and Moil, E., "Acoustic-Emission Transducer and its Absolute Calibration," J.Acoust.Soc.Am., Vol. 59, 1976, pp. 344-349. Hatano, H. and Watanabe, T., "Reciprocity Calibration of Acoustic Emission Transducers in Rayleigh-Wave and Longitudinal-Wave Sound Fields," J.Acoust.Soc.Am., Vol. 101, 1997, pp. 1450-1455.
[4]
[5]
[6]
[7]
[8]
[9]
[lO] Hatano, H.., Moil, Y., Kishi, T., and Yamaguchi, K., "On NDIS 2109: Methods for
Absolute Calibration of Acoustic Emission Transducers by Reciprocity Technique," 4th World Meet. on Acoust.Emiss.and 1st Int. Confon Acoust.Emiss.in Manufct., Boston, Amer.Soc.for Nondest.Test., Columbus, 1991, pp. 147-154.
[11]
Aki, K. and Richards,P.G., Quantitative Seisnu)logy, Vol. 1, W.H. Freeman and Company, New York, 1980.
[12] MacLean, W.R., "Absolute Measurement of Sound Without a Primary Standard," J.Acoust.Soc.Am., Vol. 12, 1940, pp. 140-146. [13] Cook, R.K., "Absolute Pressure Calibration of Microphones," J.Acoust.Soc.Am., Vol. 12, 1941, pp. 415-420.
104
[14] Foldy, L.L. and Primakoff, H., "A General Theory of Passive Linear Electroacoustic Transducers and the Electroacoustic Reciprocity Theorem I, II," J.Acoust.Soc.Am., Vol. 17, 1945, pp. 109-120; Vol. 19, 1947, pp. 50-58.
[/5]
Lamb, L, "On the Propagation of Tremors Over the Surface of an Elastic Solid," Philos.Trans., R.Soc.London, set. A, Vol. 203, 1904, pp. 1-42.
[16] Breckenridge, F.R., Watanabe, T., and Hatano, H., "Calibration of Acoustic Emission Transducers: Comparison of Two Methods," Prog.Acoust.Emiss., Vol. 1, 1982, pp. 448-458.
I/7]
Michaels, J.E., Michaels, T.E., and Sachse, W., "Applications of Deconvolution to Acoustic Emission Signal Analysis," Mater. Eval., Vol. 39, 1981, pp. 1032-1036. Hatano, H., Chaya, T., Watanabe, S., and Jinbo, K., "Reciprocity Calibration of Impulse Responses of Acoustic Emission Transducers," IEEE Trans. UFFC., Vol. 45, No. 5, 1998, (in press).
l/S]
[191 Kawano, K., Kawauchi, T, Koguchi, H, and Hatano, H., "Measurement and Analysis of Acoustic Emission Utilizing Multi-Task Function of UNIX," Proc.of Spring ConfofAcoust.Soc.Jpn, Vol. 2, Mar. 1996, pp. 909-910, (in Japanese).
Diverse Industrial Applications
Paulo R. de Aguiar, t Peter Willett,2 and John Webster a ACOUSTIC EMISSION APPLIED TO DETECT WORKPIECE BURN DURING GRINDING
REFERENCE: de Aguiar, P. R., Willett, P., and Webster, J., "Acoustic Emission Applied to Detect Workpiece Burn During Grinding," Acoustic Emission: Standard and Technology
Update, ASTM STP 1353, S. J. Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999. A B S T R A C T : Overly-aggressive or otherwise inappropriate grinding of metals can produce an undesirable change in metallurgical properties of the material being processed; usually this is referred to as workpiece "burn". In this experimental paper the acoustic signature of grinding is collected, and compared to the processed workpiece condition, for thirteen data sets including both relatively hard (Inconel) and soft (52100 bearing steel) metals. This work is distinguished by its use of a high sampling rate (2.56 MHz) in data acquisition and in its processing of the raw, rather than RMS/filtered, data samples. Signs of burn are seen in the frequency domain, and in the correlation between wheel rotations. K E Y W O R D S : grinding, acoustic emission, burn, signal processing
Introduction
Interest in the application of intelligent machines to industry is growing. However, as yet, numerically-controlled (CNC) machining is insufficiently reliable for stand-alone operation - it is common to observe a machine's operator using a CNC system simply to correct the parameters or identify tool wear or breakage. As such, a recent research thrust has evolved around the monitoring aspect of intelligent control, and naturally acoustic emission (AE) has a crucial role in this. The problem of interest to us is the detection of superficial workpiece burn, which can occur during the cut of a workpiece by the grinding wheel, when the energy in the contact area generates a temperature increase sufficient to produce a local phase change. This is sometimes visually observable by a bluish temper color on the workpiece, but more generally requires time-consuming chemical means for its after-the-fact determination.
Researcher, Center for Grinding Research and Development, U-119, University of Connecticut, Storrs, CT 06269. 2 Associate Professor, Department of Electrical and Systems Engineering, U-157, University of Connecticut, Storrs, CT 06269. a Technical Director, Center for Grinding Research and Development, U-119, University of Connecticut, Storrs, CT 06269.
107
Copyright9
WWW.SStIII.org
108
The application of AE to monitoring of grinding and machining is not new, and a number of excellent papers have been written on the subject; we list several of these [ 112, 14]. For the most part these researchers have relied on AE power (sometimes called root-mean square, or RMS, AE) and statistics derived therefrom to determine grinding condition and performance. What distinguishes our work from the previous is the use of a high-frequency (2.56 MHz) analog to digital converter (ADC) coupled with massive data storage, and the consequent availability of essentially raw AE signals. This paper deals with the analysis of data records relating to thirteen grinding runs, involving two different materials. In each case, AE data was sampled for the entire grinding pass (30 million data samples, approximately 12 seconds each) and analyzed off-line, with results compared to inspection of each workpiece for burn and other metallurgical anomaly. A number of statistical signal processing tools have been evaluated, and an emphasis has been placed on those which are amplitude-independent; that is, features which are unaffected by overall AE power. This is due to the fact that AE power can rise and fall during a grinding process for reasons having less to do with the workpiece condition than with geometry. Unfortunately, most of these amplitudeindependent indicators have failed to recognize burn. A notable exception is a novel statistic which, in the absence of an accepted term, we shall refer to as wheel-period correlation. It turns out that AE data arising from one revolution of the grinding wheel correlates surprisingly closely with succeeding revolutions. Interestingly, this phenomenon can be employed to determine highly accurately the grinding wheel rotational speed; and further, it appears that variation in wheel speed can indicate burn. More important, we have found evidence that the strength of the correlation is also predictive of burn.
The Experiments
The experimental set-up is shown in Figure 1. Specifically, data from thirteen different experiments involving two different workpiece types were collected, one of these being a relatively difficult-to-grind metal (Inconet 718), and the other an easy-togrind bearing steel (52100). Most parameters were kept constant among all tests; however, the depth of cut, ranging from very light to aggressive, was varied. All workpieces were examined post-mortem, and signs of burn or other stress noted. Grinding parameters include: 9 Nominal wheel peripheral speed: 75.8 m/s (approximately 7500 rpm) 9 Nominal workpiece feed rate: 12.7 mrn/s 9 Coolant type: Master Chemical VHP 200 9 Coolant flow rate: 29 gallon per minute 9 Grinding wheel type: WOLFCO, CBN 1012, 100/120-CBN, M.O.S. 6115 9 Grinding wheel diameter: 7.5 inches 9 Grinding apparatus: Edgetek Superabrasive Machine Data was retrieved from a single Physical Acoustics PAC U80D-87 sensor, mounted directly on the workpiece via adhesive. The data acquisition system employed was a Hewlett-Packard E1430A run in continuous-sampling mode at 2.56 x 1 0 6 samples per
DE AGUIAR ET AL. ON WORKPIECE BURN DURING GRINDING
109
second. The A/D accuracy was 16 bits per sample, and appropriate anti-'aliasing filtering was performed internally to the HP system. The wheel spindle power, and both normal and tangential wheel/workpiece forces were also monitored on separate low-rate ADC channels. A typical data trace is shown in Figure 2. It is apparent that this signal is strongly autocorrelated. In fact, much of this autocorrelation is due to the sensor itself, whose function is aided by internal resonance. A "pencil-break" test provides an example of the overall system (workpiece propagation, AE sensor, and electronic filtering)
FIG. 1 - The grinding apparatus
impulse response and transfer function, as shown in Figure 3. A pencil-break test is a reasonably impulsive event over the frequency range of interest here, and might thus be a means of extracting the exact system response to be used in a pre-processing step of deconvolution. However, the response to such a pencil-break is dependent on a number of factors, not least of which is geometry (the pencil-break was not in the exact same location as the grinding action, and in any case this grinding interface itself moves), and hence the result is in fact useful only as a guide for the type of system response to expect. It is worth mentioning that both deconvolution based on this exemplar impulse response, and also adaptive (Wiener) whitening [15, 16] were performed on these signals, but did not aid in detection of burn and were not pursued further. Details of the tests are shown in Tables 1 and 2. In the case of the Inconel workpieces, burn can be estimated only visually. In the case of the 52100 workpieces burn was assayed visually and through laboratory testing (nital etch and surface hardness). Burn, when it occurred, was found at or near th e outfeed edge; this is due to coolant starvation from spray off this edge and away from the grinding interface. Generally, the more aggressive the cutting depth, the more likely burn was to occur. Depth of cut is an important parameter, since the length of time between first contact and the wheel axis being directly over the workpiece's infeed edge (full contact) increases with depth of cut a according to
t full_contac t - t first_contac t = l
V
a3f~- a 2
110
in which D is the wheel diameter, and v is the workpiece feed rate. This is naturally accounted for in all subsequent analysis.
6000
4000
2000
3= -2000
-40130
-8000 2760.01
I 2760.02
= 2760.03 t i m e (ms)
I 2760.04
r 2760.05
2760.06
FIG. 2 - A typical "raw" data trace from a benign grinding regime. The data is shown in 16-bit notation," that is, the maximum positive and negative levels are __+32768. TABLE 1- The parameters f o r the Inconel workpieces. All workpieces were 76.5 mm in length. Burn location refers to the distance from the infeed edge and the first visible sigm o f burn. In test number six the workpiece was ramped in such a way that contact was not established until 21 mm from the infeed edge. Test 1 2 3 4 5 6 Depth of Cut 0.025 in 0.010 in 0.005 in 0.035 in 0.045 in Burn Location 65 mm 68 mm 62 m m 50 mm 60 mm Comments Slight burn Very slight burn No visible burn burn Heavy burn Heavy burn, ramp cut
The Signal Processing

A number of statistical signal processing tools were applied to the data collected. Most of these, such as moments (e.g. kurtosis) and predictability (the output of an adaptive Wiener whitener), did not correlate well with grinding quality, and we do not report them here. For the most part an attempt was made to normalize with respect to AE power, since this can rise or fall for reasons having nothing to do with burn.
111
TABLE2 - The parameters for the 52100 workpieces. All workpieces were 78mm in length. Burn location refers to the distance from the infeed edge and the first visible signs of burn. Metallurgical softening is denoted by TB (temper burn); a hard Martensite layer is denoted by RL (rehardened layer).
Test 1 2 3 4 5 6 7 Depth of Cut 0.002 in 0.010 in 0.020 in 0.030 in 0.040 in 0.050 in 0.070 in Burn Location 65 mm 68 mm 62 50 60 60 mm mm mm mm Comments slight pervasive TB and RL slight pervasive TB and RL no burn slight TB at end moderate TB at end; some at start heavy TB at end; moderate at start heavy TB, RL at end; heavy TB at start
The Spectrum
It is natural to investigate the behavior of the power spectrum. To calculate this, we have used fast Fourier transforms (FFTs) of length 1024, and averaged the magnitudesquares of these according to the Bartlett procedure [ 16]. The power spectral density of received data before contact and during benign grinding are shown in Fig. 4. It is clear that active grinding is characterized by enhanced high-frequency power. In Figures 6, 7, and 8 we show the short-time power-normalized spectrum (that is, the power spectrum estimated from a block of data, divided by the average AE power during that block). In these plots lighter shading corresponds to higher power at the given frequency and time;
x 10 ~ 2 1.5 i i 1
0.5
o
-1 -1.5
0.02 0.04 0.06 0.08 0.1 0,12 0.14 0,16 0.18 0.2
~ -0.5
time ( m s )
1 i
._~o.e
0.6
~0.4 c m 0.2 i
200 400 600 800 1000
I
1200 1400
frequency (kHz)
FIG. 3 - Above: the system response, as determined from "pencil-break" data, amplitude is in A/D units. Below: the system transfer function, units normalized to give maximum unity value.
112
logarithmic pre-processing was performed to avoid dynamic range problems in these plots. The Nuttall Statistic A data-processing step attracting a great deal of interest of late is Nuttall's "powerlaw" statistic [13]. The form of the statistic is
Tr
zklxkl5
in which X(k) denotes the k th F F F output, and summation is over any specified range of frequencies. The denominator enforces that this statistic be power-independent; the exponent of five was found to be a robust choice for most applications. Essentially, the power-law statistic measures a frequency-domain moment - in fact, if the exponents above were 4 and 2 instead of 5 and 2.5 this would be easily seen to be a kurtosis. Nuttall has derived this statistic based upon considerable analysis showing
t0'
E4
2
o
o
J
200 400
600 800 frequency(kHz)
1000
1200
1400
x 10r 10
E 2
'D C "C
,-~
2
00
200 400 600 800 frequency (kHz)
1000
12100
1400
FIG. 4 - Above." a typical power spectrum o f data observed before wheel/workpiece contact is made. Below: a typical power spectrum during grinding.
1 13
that its level is considerably enhanced when transient events (of almost any nature) are present in the block of data being processed. In grinding, physical phenomena such as coolant boiling, fracture (cracking), and simple grain-passage (the normal grinding wear mechanism) can all be thought of as transient burst-energy events, hence our motivation for exploring this statistic here.
Correlation 1: Wheel-period Correlation

In Figure 5 is shown a typical AE autocorrelation signal. It is somewhat remarkable that the AE signature from one revolution of the wheel over the workpiece is highly similar to the AE signature from the next revolution - this is most likely due to the wheel/cutting profile remaining more or less constant. W e use the value of this correlation "peak" both as a statistic in itself, and also as a measure of wheel rotation period. Again, remarkably, AE is therefore able to measure rotational wheel speed very accurately. W e note that to measure the peak's height and position we have used threepoint parabolic interpolation; from Figure 5 it is seen that the actual peak may lie in between two samples.
1.2 1
0.8 g ~ o.6
~
o
0.4
0.2
I
-0.
10
I 15
I 20
25
30
40
45
50
time (ms)
0.6 0.4 .~_ 0.2
o
-0.2
-0.4
I I I I I
7.92
7.94
7.96
7.98 time (ms)
8.02
1.04
FIG. 5: Above: the measured autocorrelation coefficient from a typical grinding AE signal. Note the peaks at times corresponding to the reciprocal of the wheel rotational frequency. Below: a zoom view of the first autocorrelation peak.
114
The autocorrelation can be measured directly; but since this is computationally burdensome a fast alternative means was used [ 16]. Specifically, we estimate 1 2N-li, . 12 j2ztkm/2N P(m) =-~--G~, Z, lakl e ~.lV k=0 For any m ~ {0 ..... N-1 }, where
(2)
Xk = ~ Xne-j2~n/2N n=O
N-!
(3)
meaning, in essence, that we take the 2N-point zero-padded FFT of a block of data of length N, take the magnitude-square, and then take the 2N point inverse F I T . Since 2Npoint F F F can be calculated using O[2Nlog2(2N)] operations, all autocorrelations can be determined very quickly. It should be noted that the means described above yields a biased autocorrelation estimate; division of ~ (m) by l-lml/N removes the bias. In Figures 6, 7 and 8 the autocorrelation has been scaled by the AE power, or ~ (0).
Correlation 2: Wheel-period Power Correlation

It is arguable that the autocorrelation measure described above is overly sensitive to random effects such as phase noise (it is unlikely that a wheel revolution corresponds to an integral number of AE sampling points) and grain slippage (a small wheel deformation can result in a large loss of correlation). Thus, we propose to perform the same operations, but instead on filtered power data. Specifically, in equation (3) the raw signal x, is replaced by L-I 2 (4)
Yn =(l=_~L+lO--[ll/L~n+ll--Y
in which the mean value y is subtracted. It is anticipated that this statistic retains a more accurate and robust estimate of the wheel profile than the previous correlation. In our tests, we used the value L=10.
The Results
Statistics from Inconel tests 3 (no burn) and 6 (angled cut; heavy burn), and from 52100 steel test 7 (heavy burn) are shown in Figures 6, 7, and 8, respectively. AE and wheel motor power data from these same tests are shown in Figures 9, 10 and 11. It is clear from the normalized spectrum that normal non-destructive grinding has a somewhat disordered or "dappled" spectrum; but at the onset of burn the spectrum remains relatively consistent. We note this feature, but as of yet have been unable to summarize it into a usable scalar statistic. The normalized Nuttall power-law statistic does not appear to offer any compelling evidence of burn. It appears that burn is characterized effectively by increased wheel-period correlation; the effect is amplified, as expected, when the second (filtered power) correlation is
1 15
examined. It is speculated that this is due to metallic softening during burn: the wheel "rubs" and "plows" rather than "grinds," and its change from revolution to revolution is concomitantly lower. Further evidence of this is available from the wheel speed data, in which it is evident that the wheel increases its speed - the time between correlation peaks is lower - when burn is occurring. (It is likely possible that this effect could be measured by a substantially less involved technique than high-rate AE; but we have noted its existence for the first time during our AE study,) Comparing Inconel and 52100 plots it can be observed that correlation is greater in the latter than the former; this is as expected, since 52100 bearing steel is relatively soft and plastic. If wheel-period correlation is to be used as an indicator for burn, a relative increase rather than an absolute level should consequently be sought.
FIG. 6 - Data from the third Inconel test. (a) normalized frequency spectrum. (b) Nuttall statistic. (c) wheel-period correlation. (d) wheel-period power correlation. (e) wheel speed measured from AE. (f) AE power. The vertical dashed lines represent: time of first contact; time of full contact, time of end contact.
116
FIG. 7 - Data from the sixth Inconel test. (a) normalized frequency spectrum. (b) Nuttall statistic. (c) wheel-period correlation. (d) wheel-period correlation. (e) wheel speed measured from AE. (f) AE power. The vertical dashed lines represent: time of lull contact; time of observed burn, time of end contact. As this was an angled cut, first and full contact times are coincident.
117
FIG. 8 - Data From the seventh 52100 test. (a) normalized frequency spectrum. (b) Nuttall statistic. (c) wheel-period correlation. (d) wheel-period power correlation. (e) wheel speed measured from AE. (f) AE power. The vertical dashed lines represent: time of first contact; time of full contact; time of observed burn, time of end contact.
118
ACOUSTIC
EMISSION:
STANDARDS
AND TECHNOLOGY
14001200
looo
Q.
| 600400-
2~
I i 1
J
..
2
i
4 i f
6
7~
8
J
10
i
12
t
lOI
0 r t-
0___J
I I
l a
0 150 o100 50
0 I II
6
ill
10
~
~.,.-12
i
II
6 time(s)
10
12
FIG. 9 - AE power, wheel power draw (expressed as a percentage of full 35 horsepower), and ratio of these, for lnconel test 3. The vertical dashed lines represent: time of first contact; time of full contact, time of end contact.
~~
ae wheel
0
i i i
ratio power
Q Q
i i i i i
power
(jrl 0 I 0 Q
~ C:~ I
(jrl 0 I
r~
~-~
EJ m G3 c
4~ J~
33
m
.I-0 z
cr)
~
CO
~:~
m c z c."11
~2
z Q
:3:) z
z 0
I
I I I I I
.--k
to
120
~i
2000 1500i
O
~1000
t~
500
I 1
10
,.15 ~
0 'Ji 0
i 1
i 3
7
'1
8
I
g
I
10
100
50
I I
i II I J I r I II I
5 time (s)
10
FIG. 11 - AE power, wheel power draw (expressed as a percentage of full 35 horsepower), and ratio of these, for 52100 test 7. The vertical dashed lines represent." time of first contact; time of full contact; time of observed burn, time of end contact.
121
The wheel-period power correlation statistic is shown for all grinds (except the first 52100 test, left off for reasons of space) in Figures 12 and 13. Comparing this with the data in Tables 1 and 2 it is seen that burn correlates well with an increase in this statistic. The table data was mostly gathered through visual inspection; the statistical data appears to predict bum somewhat earlier (52100 test 5) and more often (52100 test 1) than the inspection data. The AE power data shows the familiar "horns" of high amplitude on entry and on exit. However, AE power either alone or with wheel motor power does not appear to be a reliable indicator of workpiece burn.
(a)
1 1
(b)
tO
tO
-~ 0.5
0
\
2 4 6 (c)
I I I I
~_ 0.5
0
~ v , ~ . ,,, 10
0 .
. 2
. 4
(d)
'~A'-'~^ 8 10
tO
S
0.5
_~ 0.5
0 0
O "O
\
i i ~.,,,~.^ ,~d
. ,v4^,,/v.~ , . M
6 (e)
10
8
I
10
(f)
1
/
co
_~ 0.5
0.5
0
8
i i
6 time (s)
10
6 time (s)
10
FIG. 12 - Wheel-period power correlation from Inconel tests 1 through 6. The vertical dashed lines represent: time of first contact; time of full contact; time of observed burn, time of end contact. The first and full contact lines are coincident in test 6, the angled cut.
122
(a)
t~ tO
(b)
C 0
_~ 0.5
0
I t I
"~ 0.5
0
I t
o ''~
1
6 (c)
10
o
1
iI
10
(d)
I
_~o.s
0 U
I I I
iI i i
-~ 0.5
0
I I I
~ ii i i
I
I I
0 "~
6 (e) I
10
2
9
4
, ~
10
(f)
. ' I
tO
t
t-
_~o.5
0 U
I r I
~ r , ~
0.5
0
6 time (s)
10
I I I I 2
'
"r
6 time (s)
10
FIG. 13 - Wheel-period power correlation from 52100 tests 2 through 7. The vertical dashed lines represent." time o f first contact; time o f full contact; time o f observed burn, t,:me o f end contact. Summary In this paper we have described analysis of high sampling rate AE from a series of ghnding tests with the idea o f finding a statistic indicative of workpiece burn. While our remits are preliminary, it appears that the metallic softening accompanying burn causes th~ AE signal to change less from wheel revolution to revolution than would be observed if he metal were undamaged. This is observable as an increased correlation statistic, and als) in a degree of self-similarity between short-time spectra. It has also been observed,
123
somewhat surprisingly, that wheel speed (rotational velocity) can be determined quite accurately from AE data.
Acknowledgment
This research was supported by the National Science Foundation under contract DMI9634859. The authors also wish to express their appreciation to Alan Chasse for his help in performing the grinding tests.
References
[ 1] Bennett, R., "Acoustic Emission in Grinding," University of Connecticut Master's Thesis, 1994. [2] Berkovits, A. and Fang, D., "Study of Fatigue Crack Characteristics by Acoustic Emission," Engineering Fracture Mechanics, Vol. 51, No. 3, 1995, pp. 401-416. [3] Bifano, T. and Yi, Y., "Acoustic Emission as an Indicator of Material-Removal Regime in Glass Micro-machining," Precision Engineering, Vol. 14, No. 4, 1990, pp.219-228. [4] Chang, Y. and Dornfeld, D., "Chatter and Surface Pattern Detection for Cylindrical Grinding Using a Fluid Coupled Acoustic Emission Sensor," International Conference on Machining of Advanced Materials, NIST, 1993, pp. 159-167. [5] Dornfeld, D. and Cai, H., "An Investigation of Grinding and Wheel Loading Using Acoustic Emission," ASME Journal of Engineering and Industry, Vol. 106, No. 1, 1984, pp. 28-33. [6] Dornfeld, D., "In Process Recognition of Cutting States," JSME International Journal, Vol. 37, No. 4, 1994, pp. 638-650. [7] Eda, H., Kishi, K., Ueno, H., Kakino, K., and Fujiwara, A., "In-Process Detection of Grinding Burns by Means of Utilizing Acoustic Emission," Bulletin of JSPE, Vol. 18, No. 4, 1984, pp. 299-304. [8] Kluft, W., "Monitoring the Grinding and Dressing Operations Increases Output and Quality and Reduces Cost and Waste," Proceedings of Fifth International Grinding Conference, 1990, p. 526. [9] Konig, W. and Meyen, H., "Acoustic Emission in Grinding and Dressing: Accuracy and Process Reliability," Proceedings of Fourth International Grinding Conference, 1990, p. 525. [10] Akbari, J., Saito, Y., Hanaoka, T., and Enomoto, S., "Using Acoustic Emission for Monitoring of Grinding Process of Fine Ceramics," JSME International Journal, Vol. 38, No. 1, 1995, pp. 175-180. [11] Liang, S. and Dornfeld, D., "Detection of Cutting Tool Wear Using Adaptive Time Series Modeling of Acoustic Emission Signal," ASME Conference on Sensors for Manufacturing, 1987, pp. 27-38. [12] Emel, E. and Kannatey-Asibu, E., "Tool Failure Monitoring in Turning by Pattern Recognition Analysis of Acoustic Emission Signals," ASME Journal of Engineering and Industry, Vol. 110, 1988, pp. 137-145.
124
[ 13] Nuttall, A., "Detection Performance of Power-Law Processors for Random Signals of Unknown Location, Structure, Extent, and Strength," Naval Undersea Warfare Center Technical Report 10751, 1994. [14] Waschkies, E., Sklarczyk, C., and Hepp, K., "Tool Wear Monitoring at Turning," ASME Journal of Engineering and Industry, Vol. 116, 1994, pp. 521-524. [ 15] Haykin, S., Adaptive Filter Theory, 3rd edition, Prentice-Hall, 1996. [ 16] Proakis, J. and Manolakis, D., Digital Signal Processing, 3~dedition, Prentice-Hall, 1996.
Sergey A. Nikulin, 1 Mstislav A. Shtremel,~ Vladislav G. Khanzhin, l Elena Y. Kurianova I and Anton P. Markelov I ANALYSIS OF FRACTURE SCALE AND MATERIAL QUALITY M O N I T O R I N G W I T H THE HELP OF ACOUSTIC EMISSION MEASUREMENTS
REFERENCE: Nikulin, S. A., Shtremel, M. A., Khanzin, V. G., Kurianova, E. Y., and Markelov, A. P., "Analysis of Fracture Scale and Material Quality Monitoring with the Help of Acoustic Emission Measurements," Acoustic Emission: Standards and Technology Update, ASTM STP 1353, S. J. Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999. ABSTRACT: The results of a study of the mechanism and kinetics of metallic material fracture (dual phase steels, binary Zr-2,5Nb and multicomponental Zr-l,2Sn1Nb-0,3Fe alloys, reinforced surface layers and protective coatings, multifilamentary Nb3Sn and Nb-Ti based superconductors and HTSC-based composition wires) are given. The quantity fracture analysis is based on measurements of acoustic impulse peak (maximum) amplitudes by non-resonance sensors for linear measurement of acoustic shifts and crack parameters measurement. The developed methods of absolute calibration of AE equipment were checked by testing various types of materials and crack parameter measurements in laps and fractures. The calibration dependencies for quantity measurements are shown. The possibilities of AE for quality analysis and for characterizing of materials in the process of various mechanical tests and in the process of pressure processing with the help of a developed experimental computerized AE system are demonstrated. KEYWORDS: quantitative AE-measurements, AE materials quality monitoring, computerized AE system, deformation and fracture, materials testing, pressure processing
At the Moscow State Steel and Alloys Institute (MSAI) the basis of the program for developing AE quantitative methods is formed by procedures and instruments that are based on the notion of the linear relation between the maximum peak amplitude of the acoustic field and the elastic energy of the AE source at the rate of the AE source evolution close to the sound velocity. The previous AE methods are 1 Professor, professor, leading scientific officer, senior scientific officer, and senior scientific officer, respectively, Moscow State Institute of Steel and Alloys, Lcninsky av. 4, 117936, Moscow, Russia.
125
Copyright9
www.astm.org
126
ACOUSTIC EMISSION:STANDARDSAND TECHNOLOGY
oriented on obtaining the maximum sensitivity at the expense of using resonance detectors which provided just the qualitative analysis of processes due to an intensive sound reverberation in a metal. Therefore, the instruments and techniques of measurements developed and employed in this work are based on non-resonance damped wide-band sensors. This allowed linear measurements of acoustic displacements and, using the latter, one can assess the dimensions of radiation sources, in other words, cracks. There are two ways of practical implementation of AE-method in MSAI. The first way is monitoring of quality of materials with the help of AE measurements during laboratory research, which is called "Quality monitoring of materials," that allows the evaluation of ductility margin, fracture resistance, resistance to SCC, etc. The second is AE-monitoring of processing of materials and AE-monitoring of work conditions of technological equipment, which is called "Technological monitoring." In the framework of "Quality monitoring of materials" and "Technological monitoring" various methods of quantitative analysis of processes and materials with the help of AE were developed. The present work deals with the results of a 10-year research work, carried out in MSAI [3,10-24]. The physical principles of quantitative AE measurements were presented earlier in [3] and were delivered by authors ft the Russian - American Conference "The Ways of Development of Nondestructive Control" (Moscow, 1989). In the framework of this research specific methods and equipment were developed for quantitative AE analysis. In combination with other methods of investigation, such as metallography, electron microscopy, fractographical analysis, AE measurements will allow analysis in real scale of time various processes of deformation and fracture of tested materials. The developed equipment and methods for AE-measurements are used for carrying out a wide range of practical problems: development of cryogenic and coldresistant dual phase steels [10, 11], increased crack resistance of reinforced surface layers [ 12] and protective coats [13], development of technology for processing Zrbased tubes for nuclear power reactors [14-16, 23], development of technology for manufacturing for multifilamentary superconductors for large scale magnetic systems [17,181.
Experimental Procedure
Phenomenologyand Principlesof EquipmentDevelopment

An acoustic impulse in material is formed when elastic relief of an area takes place AV - d 3 (d = diameter of fracture zone). The time of impulse formation is tp = d/c, where c is fracture rate. Further continuous reverberation is accompanied by consecutive transmission of oscillation energy to the lower frequencies by excitation o longitudinal oscillations, and then flexural oscillations of the sample and the system "sample - machine." Therefore, the maximum frequency of AE source is fo - 1/tp.
NIKULIN ET AL. ON FRACTURE SCALE AND MATERIAL QUALITY
127
In the laboratory test a frequency of source (crack) oscillation is several orders higher than frequency of longitudinal and flexural oscillations of the sample. So, frequency characteristics of the AE device influence the physical sense of a signal: amplitude of an AE signal of high frequency depends on the dimensions of the sample and decreases when the hardness of the loading unit is raised. The inside crack, which is formed at a rate o f V - c (c is speed of sound in metal), can be characterized by different dependence of fracture period on linear dimension of a crack d. That is why a depends on bandwidth of a device fracture dimension, measured with the help of the AE methods. Minimum resolved distance between the sources is Almin ~ c/fo. If the upper bound o f a frequency band of a device is fo -1 MHz, then Almin - 1 mm. So, the registered acoustic impulse is insignificant depending on the specific characteristics of microoscillators - "elementary" microcracks in structure with specific size of d _= 10100 ~m and it is determined by averaged change of stress field. The equipment for signal detection and registration should match the subject of inquiry by pulse period - to pulse duration ratio. If for the period of time tp the sample with cross-section F o is distracted, the registration device with time constant "c cannot separately detect the crack steps on surface area less than F =_(X/tp)Fo. It is impossible to detect the steps of a crack of a sample with cross-section Fo ~ 1 cm 2 for the period oftp ~ 10.3 s (linear crack rate M 0 m/s) on the surface area less than 0.1 mm 2, if the bandwidth is fo - 1 MHz. As usual in microstructure there are no structure elements with the surface area of 0.1 mm 2 (liner size d - 0.3 mm) and more. That is why, when the main crack is formed, high frequency detection reflects not elementary, but complex events in microstructure. In that case the registered amplitude depends on the time interval between the events, united in a group. So, in mechanisms of high rate fracture the registered AE allows the detection of a single step of crack. Pulse height distribution and frequency allocation provides information about the stages of crack development when heterogeneous crack is formed.
AE-Devices for "Monitoring of Materials" and "Technological Monitoring"

The principles stated above are fixed in the basis for development of the following AE equipment: 1. Small size laboratory equipment for AE-measurements of tested materials. 2. Microprocessor detectors of AE signals - the devices with built-in calculating machines for automatic detection, storage and processing of information about AE signals during mechanical tests of materials. 3. Information-measuring system on the base of IBM PC. Small Size Systems for AE-Measurements -The objectives of laboratory AEmeasurements of fracture are comparisons of events in structure with generated AE signals. The peculiarity of an AE signal, generated by a microcrack, is a low level of a signal - which can be compared with the noise level of receiving amplifying devices.
128
High sensitivity of AE-method is achieved by implementation of resonance sensors [1-9]. A multireflected acoustic wave passing through a piezoelectric element, increases the ratio "signal/noise" of the output electric signal to 102-103. But a continuous reverberation makes it difficult the identify an output signal and does not allow to detection of a single stage of the fracture process. In the program for developing AE quantitative measurements it was intended to use another method, based on implementation of nonresonance damped piezoelectric elements [3]. In that case the sensor from the opposite side to the tested sample is damped by acoustic trap, which absorbs an acoustic wave, passing through the piezoelectric element. This leads to increase of time resolution of devices and allows the detection of the formation of inside cracks with the help of AE, to calibrate equipment for crack size measurements and to make quantitative analysis of kinetics of damage accumulation in deformed material. In a specially developed wide band piezoelectric converter for broadening the band of linear transformation the AE piezoelectric element is damped by a conic acoustic trap. Amplitude-frequency characteristic of a sensor is linear on the level of 3 dB in frequency band not less than 20 MHz. The design of the sensor allows the registration of a signal directly from the working part of the tested sample. The acoustic sensor contacts the sample through the oil layer. The developed small-size device for AE-signals registration in dynamic range (Fig. 1) consists of a set of wide band piezosensors; a set of small-size amplifiers (Af = 0.01+15 MHz, 32 dB); preamplifier, connected directly to the AE sensor; a unit for signal processing, which consists of a power unit of preamplifier, passive RC-filter additional amplifier with step change of amplification factor (0.8, 14, 20, 40 dB) and a high speed (50 V/~ts) peak detector. The reservation of dynamic range of amplitudes during the laboratory registration of AE signals, is insured by recording of a detected signal with the help of a multichannel high speed recorder, or directly in the memory of the PC. The dynamic range of peak amplitude registration is not less than 72 dB. The AE impulses of 0.1 las is linearly processed. Time resolution of impulses is determined by the frequency limit of a detector, which for a high speed recorder is fmax = 150 Hz. Peak (maximum) AE impulse amplitude Up is expressed in dB (Vp = 20.1g(Up/Un)) in relation to average noise impulse amplitude (U,), which is determined for each sample after the second loading, the first stage of test, in a zone of Kaizer effect [ 19]. Microprocessor Detectors of AE-Signal -Microprocessor detectors of AEsignals (MDS) [21 ] were developed for registration, transformation, preliminary processing and recording of results of quantitative analysis of deformation fracture and process kinetics. The built-in calculating machine allows one to make a direct superposition of deformation diagrams of mechanical tests and AE-diagrams when less than 107 are cracks formed in one test, and processing according to one of the programs in the software packet. The device transmits digital information by communication channels for a distance of up to 100 m; indicates mechanical and AE parameters; registration of input information arrays; and monitoring of the work condition of the device.
129
$
IO n'
(D
j J
1 - samples; 2 - AE-transducer
FIG. 1- Schematic of acoustic emission facility and test schemes.

Multichannel Computerized System "Test AE" Type - The "Test AE" type system, developed by scientists at MSAI, allows measuring, processing and archiving of AE signals [ 16]. Such an information measuring system is designed by a modules principle. AE-signals are detected by a processor. Preamplifyling and the filtration of signals is carried out by analogue, connected to a multichannel module of AE signals numbering. This module is on an interface board, which is mounted in an IBM PC. Data input is fulfilled by a 16-channel digital module. Numbering of signals is fulfilled by a 12-digit analogue-to-digital converter. Maximum beat frequency is up to 20 MHz; for preliminary digital processing of signals a RISK-processor is mounted in the interface board. The interface of a system for processing signals is based on the technology of virtual (image) devices, such as Labview software, which is used for the realization of algorithms of information processing and data archiving.
130
AE Devices for Monitoring of Pressure Processing of Materials -High sensitivity of AE radiation allows the implementation AE method in conditions of high background noise for monitoring pressure processing of metals, for example, drawing of a wire made of composite materials. The faults (breakdowns) of the technological process can lead to nonuniformity of the plastic deformation of a wire, which can lead to formation of cracks, separation of components and fiber break in composite wire. AE signals from such defects can be detected by a "Test-AE" computerized system in the drawing process. The following problems were solved when the set of devices was developed: 1) Detection of acoustic signals directly in a zone of minimum distance from defbnnation zone, in conditions of high deformation rate, high temperature and vapor of technological liquids. 2) Extraction of legitimate acoustic signals from the deformation zone of the background of technological noise of the loading unit and industrial noise. Block-schema of AE device is shown in Figure 2,
composite wire; drawing tools; AE-transducer; preamplifier; unit of information pre-processing; 6 - additional amplifier; 7 - signal processingunit; 8 - IBM PC type computer; 9 - signalization unit; 10 - defect
1-
'0
2345i. . . . . . . . . .
o'i
I i
Ii
'
FIG. 2 - Information and measurement unit for drawing AE-control of composite
superconductors.
The information measuring device has 16 channels. The AE-detecting block consists of a measuring cell with 2-12 piezoelectric sensors (3), connected to them are small size preamplifiers (4) and electronic devices of preliminary processing of AE signals (5). The amplification is 60 dB in a frequency band of 0.1 to 20 MHz with time resolution 10.4 s. The complex and measurements are controlled by 12 channels with the help of a "Test-AE" computerized system.
Results and Discussion
AE Monitoring of Materials
The development of the AE method to quantitative monitor materials is implemented by us in two main directions.
131
13.
r (r
II1 69
450 300 150 0
60 -o 40 E
0.
CL
20 o~ Q)
0.1 Strain 0.2 a)
e
O3
-r
450
60 ~E
3 [ [0 1~ 5 [ 0
0
0 ~I'll
J~ ,i.~, 0.1 b)
4 O
-~
20 o.
0.2 Strain 0
FIG. 3 - Diagrams of strain and AE upon tension Zr-l.3Sn-l Nb-. 4Fe samples
containing fine particle (a); and aggregates of coarse particles (b).

The first direction is the use of the AE method as an indicator of the initiation end or intensity of deformation and rupture processes that proceed during material testing. In this case AE measurements were carried out directly in the process of tensile testing materials using various schemes (tension, bend, torsion, etc.). AE and loading diagrams were analyzed together with the results of the metallographic and fractographic studies into changes that take place in the structure or fracture. This makes it possible to set up unambiguous correspondence of the acoustic signal to the event that gave birth to it. This complex approach allowed the analysis of the fracture processes in various materials and the acquisition of the information that could not be obtained if the indicated methods were used individually. The second direction in the evolution of the quantitative AE analysis is a direct measurement of the dimensions of the fractured areas using AE signals. Some results of material monitoring are given below.
Assessment of Ductility Margin of Materials-The AE method was widely used by the authors to assess the ductility margin in mechanical tests for tension and bend of
132
high manganese steels [10, 1 t] and Zr alloys (Zr-l.3Sn-lNb-0.4Fe; Zr-2.5Nb) in different structure conditions [ 14-16, 23]. The joint analysis of strain and AE diagrams demonstrates the interrelationship between the ductility determining route of plastic flow stability loss and the mode of AE. Figures 3 and 4 illustrate the AE upon the plastic flow stability loss due to a "geometric loss of strength" by the steels and Zr-alloy where they are of uniform
oo O.
.= 600 4O0 200

0.05 Strain 0.10
6o 40
}
m
20 -~ 0
a)
m n
oo
~2
nn
600
400 200
I
so .~,
40
~ID
20 -~ 0 a.
I~
I tllllllllhh,J~ll ~
0.05 b) Strain
,i,,,~l]r
0.I0
FIG. 4 - Diagrams of strain and AE upon tension of hydrogenated Zr-2. 5Nb samples
containing fine (a) and coarse (b) hydrides.

plastic strain and the level of AE does not exceed the level of a noise. The figures illustrate the earlier loss of flow stability due to the formation of an "internal" neck induced by microcracks imitating structure defects (aggregates of coarse particles, brittle secondary phase, etc.). Here, before a load drop into the test several strong AE impulses are recorded due to microcrack openings which are corroborated by the metallographic and fractographic analyses of samples. Thus, the structure determined differences in the way the flow stability and the deformability of materials are lost are unambiguously revealed in the joint analysis of
133
the deformation and AE diagrams. This allows the application of AE measurements for monitoring the alloy quality in the standard mechanical tests. Analysis of Crack Resistance of Coats -The AE method was employed to study the processes of surface crack initiation and evolution in hardened layers having variable toughness (e.g., carburized layers in steels [12]) and in thin layers of protective coats (e.g., Cr coats on superconductors [13]). In the tests for static and dynamic crack resistances of carburized layers the measurements of AE together with metallographic and microfractographic analyses made it possible to determine differences in the structure mechanisms and the kinetics of a fracture over the layer depth (Fig. 5). The feasibility is shown using AE to measure individual crack ramps and predict the depth of a crack penetration into a layer.
Crack length, mm
a
b
3,oI
Peak amplitude, dB
15 10 5
G ~ _
120 100 80 60 40 20
zs i 2,o!
1,~ 1,~
o,~
I I
0 0,4 0,8 1,2 1,62,0 2,4 2,8 Crackopening,mm
200
time, s
250
300
FIG. 5 - Diagrams of strain, AE (a) and forecast of a crack length on AE (b) in carburized steel. Mechanism and kinetics of fracture of electrolytic chrome coatings on multifilamentary superconducting wire have been investigated with the AE method when developing superconductors for the ITER magnetic system [ 13]. The objective in this work is to assess the structure and the strength of 3 types of chrome coatings, different in view of their production technologies, for choice of the best coating quality. Resistance of layers to cracking characteristics of chrome electrolytic precipitates were determined using the AE measurements at the wire specimen tension. Figure 6 shows typical AE-diagrams and synchronized with them by the time works the diagrams of wire specimens deformation. There the kinetic functions F~N(8) and types of histograms of AE pulses amplitude distributions are also presented. AE measurements allows the extraction of the deformation at the beginning of active chrome destruction at tension el. Kinetics of damage accumulation was studied by integral functions of a number of AE pulses specimen deformation EN(e).
134
Technological Monitoring AE Analysis of Material Damageability in Operations Composite Superconductor Twisting-The technological ductility of composite superconductors is limited by a crack formation at the drawhole-die interface. The studies into the kinetics of a damage accumulation upon twisting complex composites, i.e., multifilamentary superconductors containing up to 50 000 filaments became possible only after special instruments and techniques of AE measurements were created [18].
500 Stress (MPa)
,,#
600 Peak amplitude (dB) 450
EN
"x
250 <~(~)
~N(c)
60 ..... 40 151) 20 3~
~h
~,l~J
25
4 S~ain (%)
Peak amplitude (Vp)

i
750 Stress , (MPa) 500 i

i
EN(e) l
Peak amplitude (dB) 100 > 1000
EN
50
500
o i,
gl
ill JJh,,,I
2 4 g2 Strain (%)
ill
25
0 b
0
I 2
Peak amplitude (Vp)
FIG. 6 - Typical AE diagrams, stress-strain curves, and types of histograms of AE peak amplitude distributions: (a) - milk Cr; (b) - hard Cr coating. The AE measurement implemented together with electron microscope and fractographic analyses revealed the main stage of deformation and crack formation as well as the degree of damageability of superconductors at different stages of twisting. This allowed the assessment of the ductility margin of superconductors having various designs and the control of their damageability in the process of twisting.
135
The AE diagrams of twisting superconductors of different designs reveal three distinctive stages that differ in the acoustic radiation power (Fig. 7). The first stage is distinguished by an increase in the AE power upon transition to the plastic deformation of a superconductor. The second stage corresponds to a uniform deformation and is described by AE with a small amplitude of signals. The duration of this stage depends on the ductility margin of a superconductor. Upon going to the third stage one observes a raise in the power and a broadening of the amplitude spectrum of acoustic signals that accompany the formation and evolution of defects, i.e., cracks. AE Flow Detection upon Pressure Processing-The usability of AE for flow detection upon working was analyzed using as an example the process of drawing composite superconductors based on the intermetallic A15 compound type and Nb-Ti alloys containing up to 50 000 filaments in a bronze (copper) matrix.
Peak amplitude, dB AE total counts, imp. -1000 800 600 400 200
40
30
5~ t
'~
0 , 0
2O Revolutions 40
20
FIG. 7 - Acoustic emission of twisted composite conductor. Upon drawing the friction couple "wire-instruments" are effectively controlled by the AE method in a wide range of strain rates. A disturbance of the drawing process can result in a non-uniform plastic strain of a wire that induces tensile stresses in the center that might lead to a crack formation, component lamination and breakage of filaments in a composite wire. The AE signals formed by those defects are steadily revealed by the designed instruments directly in the process of wire drawing. Therefore, another possibility of the instrument and AE method of a drawing control is an isolation of the moment crack are formed and accumulated in a strained material. It is obvious that the acoustic signal pattern is capable of expressing and most informatively reflecting the greater number of sophisticated technological processes of metal working. Crack Measurement by Acoustic Emission To analyze the fracture processes by the AE signals are to be quantitatively compared to the events that generated them. For this purpose one is to provide and calibrate the single-valued linear relation "crack increment-acoustic signal amplitudeamplitude of recorded electric signal." The single-hyphen valued relation between the
136
amplitudes of acoustic field and electric signal is given by linearization of sensor by mechanical damping ofa piezoelement. In this case a sensitivity drop is unavailable, but, e.g., in our instrument [3] the signal is at the noise level at a microcrack diameter o f d = 10~a, But the simple summation of the impulse numbers ("counting rate," "total count") may give the process kinetics even if all the recorded acts of fracture are of the same scale. It is not acceptable when impulse amplitudes vary by an order and more when single acts are replaced by "cooperative" ones, The numerical investigation of the problem on a single source immersed in a semispace [3] shows that with the unchanged released energy W and the depth of a source location r0 the peak (maximum) value of linear surface displacement Up above the source depends on the time of its action t: Up x t/W ~ const (1)
A brittle crack is opened up at the rate of the order of the sound velocity. At the stress c and elastic modules E an increment of elastic energy is U = o'/2E, a crack with diameter D releases the energy W = Uxd 3 during the time t = d/s and at little variable stress one obtains the proportionality between the peak displacement in an AE impulse and a crack area F =_dZ: Up ~ d 2~ F (2)
The linear detection with recording the extreme values of the impulse amplitudes in a wide dynamic range is promising because the scope of a fracture may be seen. Dependence (2) was checked by measuring the displacement Up with a linear detector (from the electric signal Vp for intemal cracks upon tension round samples of high manganese dual phase steels [3] and ofhydrided Zr-2.5Nb alloy [I 5]. In the hydrogenated Zr-alloy there were available platelet hydrides of the axial orientation; the fracture shows internal cracks along hydrides crossing the cup. For 138 cracks in high manganese steels a linear dependence was found lgVv(tgF) with correlation coefficient o f r = 0.92 (Fig, 8a) [3]. For 15 cracks in the Zr alloy the amplitude-area relationship is also linear lgVp(lg F) with r = 0,91 (Fig. 8b) [15]. The linear dependence Up - F (with r = 0.91) has been corroborated later for brittle cracks in 38XH3MqbA steel (Fig. 8d) [3]. Thus for the internal cracks with the cross section o f d = 0.04 + 2 mm (recording piece by piece) under the constant conditions of measurements and with the invariable geometry of an experiment the proportionality is demonstrated for the peak amplitude and the crack area (which corresponds to the single value relation between the amplitude and power released by a crack). The event of a crack propagation from the outer surface is distinguished by the fact that the time of a ramp is determined by the largest one of the crack sizes - the sample B width and the acoustic impulse - from a volume discharge V N A12 x B for
137
tgvp
2-
4
. o
j
5,0
i
OgOo~oo
1 3
t 4
I 5
a)
1 6
Ig F, ~ira2
5,5
6,0 lgF,/.an 2
b)
8000
/
6400
0
4800
fJ
3206
/
i,
/g
o ~ ~o
1606
5 -0,7
I,
-0,3 c)
0,1 IgAL, mm
o,1
o,2
d)
0,3 F, ram2
-3
-2
-1
lg F, mm 2
e)
FIG. 8 - Structural calibration of AE-equipment at measurement o f crack sizes. Internal cracks: manganese dual phase steels (4); Zr - 2, 5Nb alloy containing hydrides (b); and 38XH3M@A steel [24] (el). Surface cracks." converted l & ~ steel (c); and corrosion cracks in 30XFCH3A steel [24] (e). the time t ~ B / c so that the amplitude o f the displacements Umax ~ A L 2 is weakly dependent on the sample B width. The A E peak amplitude reflects its least size - the depth o f the ramp Up ~ A L 2.
138
The experimental check has been implemented for surface cracks of a brittle fracture of a carburized layer upon static bending. For 13 surface cracks with a straight front and an invariable sample width the peak amplitude of signals is linearly related to a crack ramp depth lgAL (Fig. 8c) [12]. The summation of the arising corrosion crack area in terms of the AE amplitudes with the calibration on their final area in SCC tests of 50 and 30XFCH3A type steels having the structure of low tempered martensite [24] gives the dependence lg Up = lg F (with r = 0 , 9 5 ) (Fig. 8e). The proper reproducibility of the peak amplitudes of single sources makes it possible by going from one sample to another to calibrate the instrument against an acoustic impulse and to retain once directly constructed graduation for crack dimensions. The existence of the same graduation is shown for changes in the linear dimensions of cracks in the range of two orders [7]. The linear "amplitude-crack size" correspondence is to be expected at the invariable rate of propagation of all the cracks. The possibility is not excluded of an incomparable graduation for a brittle and "slow" tough crack. But the event of the coincidence of "the amplitude - crack size" graduation for tough and brittle cracks (Fig. 8) points to a higher rate of a band breakage in a group of pores near a tough crack. Thus, the AE measurements of a crack increment is feasible under the following conditions: 1. Discrete mode of process. The time between crack ramps is more than the constant of the instrument time. 2. Relatively high elastic energy of release in a single fracture event and high rate of opening which allows the use linear detection for cracks with sizes of 10 tam and higher. 3. Wide dynamic range of instruments that allows measurement of large cracks without losing small ones. 4+ The feasibility of direct graduation by fractographic measurements of cracks under identical conditions.
Conclusion
The quality level of materials can be determined only by complex analysis of AE measurements data and comparison of these data with standard mechanical tests. At the MSAI the basis of the program for developing AE quantitative methods is formed by procedures and instruments that are based on the notion of the linear relation between the maximum peak amplitude of the acoustic field and the elastic energy of the AE source at the rate of the AE source evolution close to the sound velocity. The developed system of AE quality monitoring of materials united all the experience which was obtained for the period of 10 years in testing of materials in various technological conditions with the help of the AE method. The system allows the conductance of a complex analysis of mechanical characteristics and the results of
139
AE measurements. This allows a unique scientific and technological information to be obtained. The quantity fracture analysis is based on measurements of acoustic impulse peak (maximum) amplitudes by non-resonance sensors for linear measurement of acoustic shifts and crack parameter measurements. The developed methods of absolute calibration of AE equipment were checked by testing various types of materials and crack parameter measurements in laps and fractures. The calibration dependencies for quantity measurements are shown. The possibilities of AE for quality analysis and for characterizing of materials in the process of various mechanical tests and in the process of pressure processing with the help of developed experimental computerized Test-AE system are demonstrated. The high sensitivity of the test-AE system provides a new level of quality monitoring of materials.
References
[1]
Sagat S., Ambler J.F.R., Coleman C.E. "Application of Acoustic Emission to Hydride Cracking," AECL-9258 Chalk River Nuclear Laboratories, Chalk River, Ontario, July 1986, 8 p. Coleman C.E. "AE from Zirconium Alloys During Mechanical and Fracture Testing," AECL-91tl, Chalk River Nuclear Laboratories, Chalk River, Ontario, October 1986, 21 p. Khanghin V.G., Shtremel M.A., Nikulin S.A. "Assessment of Internal Crack Sizes from Peak AE Amplitudes," Defectoskopiya, 1990, No.4, pp. 35-40. Dunegan H.L., Harris D., Tatro C., Eng. Fract. Mech., No. 1, 1968, pp. 105-110. Hartbower C.E., Gerberich W., Zeisbonits H. "Investigation of Crack Growth Stress-Wave Relationship," Journal Eng. Fracture Mech., Vol. 1, No. 12, 1968, pp. 13-28. Mirabile M. "Non-destructive Testing," Vol.8, No.14, 1975, pp. 77-85. Scruby C.B., Wodley H.N.J., Hill J.J. Journ. Appl. Phys., No.16, 1983, pp.. 1069-1083. Liptai R.G., Harris D.O., Engle R.B., Tatro C.A. "Acoustic Emission Techniques in Materials Research," International Journal Non-destructive Testing, Vol.3, 1971, pp. 215-275. Williams R.S. "Modeling of Elastoplastic Fracture Behavior using Acoustic Emission Methods," J. Metals, Vol.31, No. 10, 1979, pp. 21-25. Nikulin S.A. "Two Variants of the Loss of the Plastic Fflow Stability and Alloys Ductility," Phisika Metallov and Metallovedenie, Vol. 81,. No.3, 1996, pp. 36-49. Nikulin S.A., Khanghin V.G. "The Fractures Classification from AE Measurements," Zavodskya Laboratoriya, No.2, 1991, pp. 61-63.
[2]
[3] [4] [5]
[6] [7] [8]
[9] [10]
[11]
140
[12]
[13] [ 14] [15]
Khanghin V.G., Nikulin S.A., Stremel M.A. et al. "The Studies of Carburized Layers Fracture," Fizika Himicheskaya Mechanica Materialov, No.I, 1990, pp. 91-95. Nikulin S.A., Khanghin V.G., Shikov A.K. et al., IEEE Trans. on Appl. Superconductivity, Vol. 5, No.2, June 1995, pp. 325-328. Nikulin S.A., Markelov V.A. et al. "Influence of Structure on Strain Giagrams of Zr-2,5%Nb Alloy," Izvestiya AN SSSR, Metals, 1990, No.3, pp. 134-139. Nikulin S.A., Shtremel M.A., Khanghin V.G., Fateev B.M., Markelov V.A. "Influence and Hydrides on Ductile Fracture in the Zr-2,5%Nb Alloy," Nuclear Science and Engineering, Vol. 115, 1993, pp. 193-204. Nikulin S.A., Goncharov V.I., Shishov V.N. "Effect of Microstructure on Ductility and Fracture Resistance of Zr-l,3Sn-lNb-0,4Fe Alloy," Eleventh International Symposium on Zr in the Nuclear Industry, ASTM STP 1295, Garmish, German),, 1996, pp. 99-112. Shikov A.K., Nikulin S.A. et al. "AE-Monitoring Composite HTS Conductors," Sverhprovodimost, Vol. 6, No.2, t993, pp. 429-431. Nikulin S.A., Khanghin V.G., Kurianova E.Y., Markelov A.P. "Influence of Desigh Oparameters on Mechanism and Kinetics of Twisting Effected Cracking in Composite Superconductors," IEEE Trans. on Appl. Superconductivity, 1997, (to be published). Karser J., PhD Thesis, Hochshull, Muncher, Germany (1950), see also Arch. Eisenuttenwes, 24, 1953, p. 43. Khanghin V.G., Tumanov A.V., Nikulin S.A. Prib. Texh. Ehks., No.l, 1991, p. 241. Nikulin S.A., Khanghin V.G., Kurianova E.Y., Markelov A.P. "Acoustic Emission Technology for Quantitative Monitoring SCC," 5th European Conference on Advanced Materials, Processes and Applications, 21-23 Apri, 1997, Maastricht, NL, pp. 238-241. Tumanov A.V., Khanghin V.G., Nikulin S.A. "Microprocessor Defector of AE Signals," Prib. Texh. Ehks., No.5,1987, p. 244. Nikulin S.A., Shtremel M.A., Markelov V.A. "Influence of Secondary Phase Particles on Zr-alloy Plastic Flow Stability and Fracture", Colloque C6, Journal de Physique III, Vol. 6, October 1996, pp. 133-143. Krupin Yu. A., Kiselev I.K. "Effect of Structural Factor on Corrosion Crack Resistance Parameters," Mater. Sci. and Engin., A130, 1990, pp. 29-35.
[16]
[ 17] [18]
[19] [20] [21]
[22] [23]
[24]
G. M. Nagaraja Rao,t C. R. L. Murthy,2 and N. M. Rajul CHARACTERIZATION OF MICRO AND M A C ~ O CRACKS IN ROCKS BY ACOUSTIC EMISSION
REFERENCE: Nagaraja Rao, G. M., Murthy, C. R. L., and Raju, N. M., "Characterization of Micro and Macro Cracks in Rocks by Acoustic Emission," Acoustic Emission: Standards and Technology Update, ASTM STP 1353, S. J.
Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999. ABSTRACT: Rock is a natural brittle material, and under compressive stresses fracture and failure occur by initiation, growth and interaction of microcracks forming macroscopic cracks and finally leading to a fault. As stress vs. volumetric strain curve can give only qualitative results, both for understanding the phenomenon of fault formation in terms of microcracking stages quantitatively and also for accurate prediction of impeding failure it becomes essential to identify the stages of the stress vs. volumetric strain curve by a suitable technique. Thus, in this respect the advantages of acoustic emission, a real-time on-line monitoring technique, is investigated. Conventionally, plots based on cumulative events, event rate and amplitude distribution are used to understand crack growth. While these parametric plots give the trends in terms of their AE activity, the exact characteristics of events in relation to the phenomena at different stress levels of micro and macro crack progression to failure are not easily discernible. So, in the present work based on the recorded wave form parameters, events are classified into four groups as ix, ~, "/and ~ and are characterized as micro and macro crack phases which correlate with ultrasonically imaged data. Classification of events into micro and macro crack phases gives a better understanding of fault formation in rock materials and also the effect of stress, temperature, macrostructure, mineralogy, etc. KEYWORDS: rock, micro and macro crack, volumetric strain, acoustic emission, parametric plot, ultrasonic imaging, cluster
1 Scientist and Director, respectively, National Institute of Rock Mechanics, Kolar Gold Fields - 563 117, Karnataka, India. 2 Associate Professor, Department of Aerospace Engineering, Indian Institute of Science, Bangalore - 560 012, Karnataka, India.
141
Copyright9
www.astm.org
142
Rocks are highly anisotropic in nature and their mineralogy and composition vary widely. They contain cracks, pores, joints, faults, folds and other defects. Fracture or failure of rock occurs by nucleation, extension, interaction and coalescence of microcracks forming a macroscopic crack. And, microcracking occurs whenever the local tensile stress exceeds the tensile strength. Though it is well established that with the increase of stress new cracks form, they interact and coalesce forming macroscopic cracks. A survey of the existing literature does not indicate any significant study relating to the generation of new cracks and their growth in phases [1- 6 ]. Understanding the interaction and coalescence of cracks is essential to get an insight into the fracture phenomenon in rocks.
200 0.14 f 0.12 150 r~ 100
o,Stago3_ r
Stage 4
0.08
Stage 6
50
Axial Strain Lateral Strain Volumetric Strain . (0.1) 0 ! 0.1 ~ I 0.2 0.3
0,04 0.02
50
% Strain Figure 1 - Stress vs. Strain f o r Rock.
100 lso 200 250 300 Stress (MPa)
Figure 2 - Volumetric Strain vs. Stress.
A qualitative picture of microcrack growth can be obtained from a stress vs. strain curve and typical expected curve is shown in Fig 1. While the axial strain curve shows a more or less linear portion except at lower stress levels and near the peak stress, the lateral strain curve is always nonlinear. And as both these curves do not provide detailed and quantitative information to understand the various stages of microcracking development, Brace et al. proposed volumetric strain as a sensitive indicator of microcracking development and suggested the following formula for its calculation [7] AV
e v =--=t;
Vo
x +2e l
where e = Volumetric strain

V
e x = Axial strain e t = Lateral strain.
NAGARAJA RAO ET AL. ON MICRO/MACROCRACKS IN ROCKS
143
It can be observed from the volumetric strain vs. stress curve (Fig. 2) that with an increase in stress volumetric strain increases, reaches a maximum value and then decreases. And, in the present work the stages in the microcrack development under uniaxial compressive stress are identified as six compared to the five stages usually reported in literature [7-9]. They are: Initial nonlinear region depicting closure of preexisting cracks and propagation of suitably oriented cracks is stage 1. In the second stage, the linear elastic region is obtained by drawing a linear regression line. Stage 3 marks the beginning of a nonlinear region or in-elastic deformation which arises due to the formation of new cracks. In Stage 3 volumetric strain increases with the increase of stress, reaches a maximum value and then decreases. The transition region where volumetric strain remains constant is stage 4. Probably this is the stage where crack interaction and coalescence may be beginning at localized regions within the sample. In stage 5 the volumetric strain decreases with increase of stress. In this region crack interaction and coalescence predominate. And, strain localization leads to the formation of macrocracks that occurs beyond 95% of the failure stress. Stage 6 corresponds to the peak stress marking the beginning of the post-failure region. And, with load control tests brittle rocks fail violently after reaching the peak stress which is referred as the uniaxial compressive strength. Out of these six stages Stage 3, 4 and 5 are important as new cracks form, grow and interact. From the foregoing it is clear that the stress - strain curve provides information only on a macroscopic scale and is qualitative regarding the above mentioned stages making a detailed understanding of each stage difficult. Thus, to assess the effects of various stages leading to failure, visual and nonvisual techniques are used. The conventional method is to observe the microcracks through an optical microscope. But this method of observation has certain drawbacks. The changes that occur during unloading and the preparation of the sample are not known. Further, there is some uncertainty as to whether what is being observed is actually the same microstructure that was present when the sample was under load. And, during the preparation of thin section, new cracks and other surface damages are inevitable with the resulting determination of crack density, crack aspect ratio and crack type becoming uncertain. The greatest disadvantage of microscopic observation is that only a small section of the sample is examined, which may not represent statistically a true picture of crack population and the technique of crack observation is not suitable for studies on the rate of crack growth. Several indirect methods of monitoring crack growth and their interaction have been developed. Of these, acoustic emission, a real time on line monitoring technique, was used to study the initiation and growth of cracks. The study of acoustic emissions associated with the microcracking offers an excellent means of observing indirectly the microscopic process that occurs during deformation. These processes can then be correlated with the observed macroscopic stress-strain properties. In the present work since the information available from acoustic emission monitoring is indirect, to correlate these results ultrasonic imaging is also used as a complementary technique. Specific stages of microcrack development, growth and coalescence to form macrocrack is best presented by ultrasonic imaging and the complementary information of the evolution of these stages is best understood by acoustic emission monitoring.
144
Acoustic Emission Technique for Studying the Cracking Phenomena

Acoustic emission may be defined as transient elastic waves generated by the rapid release of strain energy in a material. In geological materials, which are basically polycrystaUine in nature, acoustic emission which are of burst type may originate due to friction between interlocking grain boundaries, initiation and propagation of microcracks, crushing of pores, etc. Each material deforms in its own characteristic manner. A number of micro and macro processes contribute to the deformation and deterioration of a material under strain and to the resulting series of emission events. Thus, the events emitted by the material contain information regarding the general deformation processes as well as what happens at the flaws that exist. Acoustic emission signals are analyzed to understand the nature of source and deformation mechanism. By an analysis of these AE events it is possible to understand crack nucleation, propagation, coalescence and formation of a macroscopic crack. A large amount of literature is available on acoustic emission studies of rocks under a variety of loading conditions. Several approaches have been adopted to understand the various aspects of microcracking and fracture processes, for example: 1) Recording the events expressing them in the form of event rate or cumulative events [6,10,11]. 2) Determination of source locations and the study of mechanism of individual events [12-15]. 3) Study of amplitude distributions | 16-20]. 4) Investigation on frequency characteristics of emission events [10,11,21].
Experimental Set Up
In the present work, granite has been chosen for investigation as this rock is fairly homogeneous and a large amount of literature is available on its deformation and strength behavior. The granite rock selected for the studies was from a quarry near Kolar Gold Fields. The modal composition of the essential minerals of the granite is given in Table 1. TABLE l--Modal Composition of Minerals in Granite. Mineral Plagioclase Quartz Microcline Biotite Hornblende Chlorite Percentage 37 31 18 7 6 1
Granite blocks of the size 45 cm x 30 cm x 15 cm were collected from the quarry. NX size (~54 mm diameter) right cylindrical core samples were drilled from these blocks, cut to the desired length, ground and polished using a surface grinder. The straightness and flatness were in accordance to the tolerances prescribed by the ISRM standard. The length to diameter ratio was maintained around 2.5. Acoustic emissions were recorded during the
145
uniaxial compression testing of granites using a wide band sensor and the equipment details are given elsewhere [22]. Data Analysis Depending upon the application and type of phenomenon being studied, different approaches can be adopted for analyzing AE signals. Broadly, they can be categorized as time domain analysis and frequency domain analysis [23]. Due to convenience in handling high data rate, time domain analysis has been the most common approach for burst type of signals. Time domain analysis usually consists of studying a number of parametric plots that are generated based on individual or cumulative activity or activity rate. The simplest of them is the estimation of cumulative events, cumulative ring down count, event rate and rate of ring down count. This data indicates the condition of the component under test in terms of cumulative or rate of damage or defect growth, thereby providing an early warning of the impending failure. In the work reported here, analysis of acoustic emission events was done based on parametric plots and classification of events to understand the microcrack development.
Parametric Plots
Plots based on cumulative event, event rate (no. of events per unit stress) and amplitude distribution used to draw inferences about microcrack development yield the following information and is presented separately under cumulative plots, event rate plots and amplitude distribution.
80,000. o 60,000"
9 40,000; ~
20,000
~ o o
,/
50 100 150
,.., 2,500 ~ 2 ,ooo
~'~
~.~ 200
,., 1,500 5OO" 0 0 I 40
; . = 1,000"
iiItiT t
, _ rUl
80 120 160 Stress (MPa) 200
Stress (MPa) Figure 3 --Two Regions of Microcrack
Figure 4 --Four Regions of Microcrack
Development.
Development.
Cumulative Events--Figure 3 shows cumulative events vs. stress where two regions
of microcracking are identified. Region I: Increase of cumulative events with the increase of stress is small. This region corresponds to stage 1 and 2 in the stress vs. volumetric strain
curve.
146
Region II:
Here cumulative events increase rapidly with the increase of stress. This region represents a large amount of microcracking activity which corresponds to stage 3, 4, and 5 in the stress vs. volumetric strain curve.
Event Rate--Figure 4 shows the variation of event rate vs. stress. Here event rate is expressed as events per unit stress, i.e. dn/dg. Event rate vs. stress curve can be divided into four regions as shown in Fig. 4 and gives better information compared to cumulative plot.
Region I: The event rate which is initially high, decreases immediately to a low level with the increase of stress due to presence of pre-existing cracks. The source of AE may be due to the closure of pre-existing cracks, propagation of suitably oriented cracks, or friction between the crack surfaces. This region corresponds to stage I in the stress vs. volumetric strain curve. Here the event rate is more or less constant with the increase of stress, this region corresponds to elastic region (stage 2) in the stress vs. volumetric strain curve. In this region, the event rate slowly increases with the increase of stress due to stress induced microcracks. It corresponds to stage 3 in the stress vs. volumetric strain curve, which marks the beginning of dilatancy. Here the event rate rapidly increases with the increase of stress. This region corresponds to stages 4 and 5 in the stress vs. volumetric strain curve. In this region a large variation in event rate is observed with the increase of stress. The event rate does not increase continuously, but it always increases and decreases with the increase of stress, which implies that there are obstacles for the growth of cracks.
Region II:
Region III:
Region IV:
Amplitude Distribution--Figure 5 shows amplitude distribution at different stress levels and it is clear from this figure that as the failure stress is approached events shift towards higher amplitude. Classification of Events
While parametric plots presented give the trends, the exact characteristics of the events in relation to the phenomena at different stress levels of micro and macro crack progression to failure is not easily discernible. In fact from any AE event a number of parameters (rise time, ring down count, event duration, peak amplitude and energy) can be determined which help in obtaining this information. Thus, an attempt is made in the following to study the qualitative and quantitative changes that are discernible from the values of these parameters. So, rise time, ring down count, event duration, peak amplitude and energy were recorded for each event. Events are classified into various groups based on the recorded wave form parameters. Except the peak amplitude, for all the other
NAGARAJA RAO ET AL. ON MICRO/MACRO CRACKS IN ROCKS
147
4oo]
5oo]
300
0-50 %
2,000
50-70 %
1,6001
~ 1,200 t
400
. , . , . ,~'~ ,,,n ,, .
200t 100
O/ 0 5,0001 20 40 60 80 100 Amplitude (dB) 70-95 ~ I
] 100
m4,oool ~=3,000t
1,OOOl
d"-h J ~. /
/ \
I I t~ I
01 : ~___:____J. . . . ~.'7--4 0 20 40 60 80 100 Amplitude (dB)
3O0 L~ 200 100 ~ 0 ....:----7-------;--- . . . . . 0 20 40 60 80 100 Amplitude (dB)
4~176
500 t
0 0 . 20 . . 40 . 60 . . 80 . . . Amplitude (dB)
, [q~95-97 %1
1000,
800 I 200 ]
0/ . . . .
200t
1
/.
0 20
9~~%
40 . . . . . 20 40 60 80 Amplitude (dB) 100
8~1 ..... 0
, .~-~
100 99.7-100 %1
40 60 80 Amplitude (dB)
701
0 20
9949" 1
40 60 80 Amplitude (dB) 100
50 40
~ 30 ~ 20
10 0
__:____.____:_.___:____./,
0 20 40 60 80 Amplitude (dB) 100
Figure 5--Amplitude Distribution at Different Stress Levels. parameters individual values vary from very low to very high, making it difficult to group the events based on them. The peak amplitude varies to a maximum of 100 dB which permits grouping of these events. The steps involved in classifying the events are: Step I: Classification of events into six groups : 44-50, 51-60, 61-70, 71-80, 81-90, 91-100 dB.
148
Step lI:
Determination of ring down count, energy and event duration for these groups of events based on a point plot which is also known as correlation plot or scatter plot or cross plot. Here each AE hit will produce a single point at the appropriate place on the screen or print out. The place on the screen shows the value of the x and y parameters for that hit. This plot is useful for discriminating between different kinds of sources. Table 2 gives the ring down count, energy and event duration for the sample tested (values given in Table 2 correspond to the maximum value). From Table 2 it can be inferred that events in the range 44-50 and 51-60 dB have low ring down counts, energy and event duration, where the events belonging to groups 81-90 and 91-100 dB have high values. Determination of initiation stress for these six group of events based on the point plot of amplitude vs. stress. Table 2 also gives stress values for six groups of events.
Step III:
From Table 2 the following inferences can be drawn. 1) Events belonging to the group 44-50 dB and 51-60 dB are initiated at the beginning of stress application. They have low values of ring down count, energy and event duration, which are classified as "c(' type events. 2) 81-90 and 91-100 dB events are high amplitude events, they initiate more or less at the same stress level, their ring down count and event duration are comparable. They are grouped together and named as "8" type events. 3) The remaining two groups of events 61-70 and 71-80 dB, do not have identical features and are treated as separate groups, named as "13" and " T "In brief, the events have been classified into four groups as: Event type ot l~ T 8 Amplitude (dB) 44-60 61-70 71-80 81-100
149
TABLE 2-- Wave Form Parameters and Initiation Stressfor Six Groups of Events Amplitude (dB) 40-50 51-60 61-70 Ring down count 19 100 250 Energy 14 I00 450 Event Duration (micro second) 480 1800 4000 Event initiation Stress (MPa) Observed from the beginning of stress till before failure Observed from the beginning of stress till before failure Observed from the beginning of stress but number events are very low. 82 112 115
71-80 81-90 91-100
800 2000 2000
1800 5000 10000
10000 19000 20000
.-.1,000, ~ 600] =o'~ 400 0 40 p~
~ ,000.
~'~ 400] 80 120 160 200 Stress (MPa) 0 40
t ,ltl][ 80 120 160 200 Stress (MPa)
~ 1,0001
~ I 600] ='= ~" = 400t
s 800t
7
~~
~1,000.
800]
S
.11
ms6001~t o='~ 4001 200"
40
80 120 160 200 0 40 80 120 160 200 Stress (MPa) Stress (MPa) Figure 6--Event Rate vs. Stress for ~,~,?,and S Events.
150
Characteristic Features of the Events'

Figure 6 shows the event rate (dn/dG) vs. stress for four types of events, From these figures it can be inferred that the cz, 13, ~/and ~5type events occur at different stress levels. ot - type events: These are low amplitude events, observed immediately with the application of stress. At low stress levels they show a high event rate which decrease to a very low value within a few MPa of stress rise. At about 50 60% of the failure strength initially the event rate increases slowly, further at about 80-90% of its failure strength the event rate shows a sudden increase, reaches a maximum and then decreases till failure. Event rate vs, stress can be divided into four regions. The first two regions are not important as the events are due to pre-existing cracks and localized microcracking. Region III and IV are important as the stress induced microcracks are preduced in this region. - type events: These are observed immediately after the application of stress, but their event rate is very small which decreases immediately to a negligible value with the increase of stress. They are once again observed at about 50-60% of the failure strength and increases with the increase of stress till failure. ~/- type events: These events are observed only in the dilatant region (between 60% - 80% of the failure strength) of the stress vs. volumetric strain curve. Their event rate continues to increase with the increase of stress till failure. ~5- type events: These events are observed around 80-90% of the failure strength. Their event rate continues to increase with the increase of stress till failure. The total number of events observed and their rate decreases in the order of or, ~, ~/ and & All these four types of events appear at different stress levels. They initiate at a particular stress level, increase slowly in rate with the increase of stress and beyond a stress level the event rate increases rapidly. From this observation it is possible to generalize the variation of event rate as a function of stress for four groups of events as shown in Fig. 7. o~ type events which occur from the beginning of stress application are due to microcracks. 13 type events which are observed mostly in stage 3 are also due to microcracks. Although both cz and 13events are due to microcracks, with the available experimental evidence it is not possible to distinguish them but they may be inferred as crack initiation and extension respectively. The amplitude distribution near failure stress is shown in Fig. 5 which indicates a shift in amplitude towards higher amplitude events as the failure stress is approached. Before failure of the sample two things should occur. 1. Macrocrack initiation. 2. Macrocrack extension. ~/& ~ type events occur before failure. It can be inferred that ~ type events are due to macrocrack initiation that arises due to coalescence of microcracks and 8 type events
151
which appear after 7 type events are due to macrocrack extension. This result can be summarized as: tx type events represent microcrack initiation. type events represent microcrack extension. type events represent macrocrack initiation. 8 type events represent macrocrack extension.
Stress
Stress
eO
t~
Stress
Stress
Figure 7 -- Characteristic Feature of Four Types of Events.
Ultrasonic C-Scan Imaging

So far we have presented results of acoustic emission which shows the development of micro and macro cracks at different stress levels. And, AE is an indirect technique, to supplement the observations of microcrack development ultrasonic C-scan imaging was carried out at specific stages. The details of ultrasonic imaging are explained elsewhere [22]. Figure 8 shows the ultrasonic images obtained at stage 0, 3, 4 and 5 of the volumetric strain vs. stress curve. Stage 0 represents a sample in the unstressed condition. Based on the amplitude of the reflected ultrasonic waves, damage has been classified into micro and macro crack phases. Microcracks initially appear as random isolated points, but with the increase of stress coalesce forming clusters (group of microcracks), cluster density and coalescence increase with the increase of stress. And, most of clusters coalesce as the failure stress approaches. Further, coalescence of clusters leads to macroscopic cracks. Ultrasonic images clearly show the formation of microcracks and their interaction to form a macroscopic crack.
152
Based on the acoustic emission monitoring and observation of ultrasonic images, it can be inferred that with increase of stress the microcrack damage development will have the following phases.
Figure 8 -Ultrasonic Images at Different Stages of Deformation.
Conclusions
AE results are analyzed in two stages, based on simple parametric plots and grouping of events. While parametric plots give a qualitative development of microcrack in terms of identifying regions, grouping of events leads to the classification of AE events into micro and macro crack phases. Based on cumulative event plots, two regions of microcracking activity were identified. The event rate plot gives a better picture where microcracking activity is divided into four regions. But still the available information on microcrack development is qualitative. Classification of events based on wave form parametrics has
153
revealed four types of events which are identified as micro and macro crack phase events. Those with amplitude less than 70 dB are identified as microcrack events and greater than 70 dB as macrocrack events (Table 3). TABLE 3 - - Micro and Macro Crack Phases Phase Microcrack initiation Microcrack extension Macrocrack initiation Macrocrack extension Event type ~ I] ~/ 8 Peak amplitude (dB) 44-60 61-70 71-80 81-100
The total number of events observed and their rate decrease in the order ~t, I~, ~/and 8. All these four types of events appear at different stress levels. They initiate at a particular stress level, increase slowly in rate with the increase of stress, and beyond a stress level the event rate increases rapidly. Ultrasonic images clearly show the initiation and growth of micro and macro cracks. Classification of events into micro and macro crack phases gives a better understanding of fault formation in rock materials and also the effect of stress, temperature, macrostructure, mineralogy, etc.[22].
References
[I]
Patterson, M.S., Experimental Rock Deformation - The Brittle Field, SpringerVerlag, New York, 1978. Tapponier, P. and Brace, W.F., "Development of Stress-induced Microcracks in Westerly Granite," International Journal of Rock Mechanics and Mining Science & GeomechanicsAbstracts, 13, 1976, pp. 103-112. Wawersik, W.R. and Fairhurst, C., "A Study of Brittle Rock Fracture in Laboratory Compression Experiments," International Journal of Rock Mechanics and Mining Science & Geomechanics Abstracts, Vol. 7, 1970, pp. 561-575. Peng, S. and Johnson, A.M., "Crack Growth and Faulting in Cylindrical Specimens of Chelmsford Granite," International Journal of Rock Mechanics and Mining Science & Geomechanics Abstracts, 9, 1972, pp. 37-86. Wong, T.F., "Effects of Temperature and Pressure on Failure and Post Failure Behavior of Westerly Granite," Mechanics of Materials, 1, 1982, pp. 3-17. Scholz, C.H., "Microfracturing and the Inelastic Deformation of Rocks in Compression," Journal of Geophysical Research, 73, 1968, pp. 1417-1432.
[2[
[3]
[4]
[5] [6]
154
[7] 181 [91
Brace, W.F., Paulding, Jr. B.W., and Scholz, C.H., "Dilatancy in the Fracture of Crystalline Rocks," Journal of Geophysical Research, 71, 1966, pp. 3939-3953. Scholz, C.H., The Mechanics of Earthquakes and Faulting, Cambridge University Press, Cambridge, 1990. Martin, C.D. and Chandler, N.A., "The Progressive Fracture of Lac du bonnet Granite," International Journal of Rock Mechanics and Mining Science & Geomechanics Abstracts, Vol. 31, No. 6, 1994, pp. 643-659. Boyce, G.M., Mc Cabe, W.M., and Koerner, R.M., "Acoustic Emission Signatures of Various Rock Types in Unconfined Compression," Acoustic Emissions in Geotechnical Engineering Practice, ASTM STP 750, V.P. Dnevick and R.E. Gray, Eds., ASTM, 1981, pp. 142-154. Ohnaka, M. and Mogi, K., "Frequency Characterization of Acoustic Emission in Rocks Under Uniaxial Compression and Its Relation to the Fracturing Process to Failure," Journal of Geophysical Research, Vol. 87, No. B.5, 1982, pp. 38733884. Lockner, D. and Byerlee, J.D., "Acoustic Emissions and Fault Formations in Rocks," Proceedings of the First Conference on Acoustic Emission/Microseismic Activity in Geologic Structures and Materials', H.R. Hardy Jr. and F.W. Leighton, Eds., 1977, pp. 99-107. Scholz, C.H., "Experimental Study of the Fracturing Process in Brittle Rock," Journal of Geophysical Research, 73, 1968, pp. 1447-1454. Byerlee, J.D and Lockner, D., "Acoustic Emission During Fluid Injection in Rock," Proceet.fings of the First Conference on Acoustic Emission/Microseismic Activity in Geologic Structures and Materials, H.R. Hardy Jr. and F.W. Leighton, Eds., Trans. Tech Publications, 1977, pp. 87-98. Nishizawa, O., Onai, K., and Kusunose, K., "Hypocenter Distribution and Focal Mechanism of AE Events During Two Stress Stage Creep in Yugawara Andesite," Pure Applied Geophysics. 122, 1984, pp. 36-52. Mogi, K., "Study of Elastic Shocks Caused by the Fracture of Heterogeneous Materials and Its Relations to Earthquake Phenomena," Bulletin of Earthquake Research Institute, Tokyo University, 40, 1962, pp. 125-173. Mogi, K., "Magnitude Frequency Relation for Elastic Shocks Accompanying Fractures of Various Materials and Some Related Problems in Earthquake," Bulletin of Earthquake Research Institute, Japan, 40, 1962, pp. 831-853.
11Ol
I111
il2l
1131
1141
I151
1161
[17]
155
[181
Scholz, C.H., "The Frequency - Magnitude Relation of Microfracturing in Rock and Its Relation to Earthquake," Bulletin of Seismological Society of America, 58, 1968, pp. 399-415.
[191 Atkinson, B.K. and Rawlings, R.D., "Acoustic Emission D u ~ g Stress Corrosion Cracking in Rock," Earthquake Prediction-An International Review, Maurice Ewing series, Vol.4, D.W. Simpson and P.G. Richards, Eds., American Geophysical Union, Washington, DC, 1981, pp. 605-616. [20]
Mogi, K.,"Earthquake Prediction Program in Japan," Earthquake Prediction- An International Review, Maurice Ewing series, Vol. 4, D.W~ Simpson and P.G. Richards, Eds., American Geophysical Union, Washington, DC, 198l, pp. 635666. Ohanaka, M., "Acoustic Emission During Creep of Brittle Rock," International Journal of Rock Mechanics and Mining Science & Geomechanics Abstracts, Vol. 20, 1983, pp. 121-133. Nagaraja Rao, G.M., "Studies on Strain Rate and Thermal Exposure Effects on Initiation and Growth of Cracks in Granite by Ultrasonic and Acoustic Emission Techniques," Ph.D thesis, Indian Institute of Science, 1997. Murthy, C.R.L, Dattaguru, B., Ramamurthy, T.S., and Rao A.K, "Studies in Acoustic Emission Signal Analysis," Project Report, ARDB-STR-5015, 1982.
[211
[221
[23]
Tomoki Shiotani 1and Masayasu Ohtsu ~ PREDICTION OF SLOPE FAILURE BASED ON AE ACTIVITY
REFERENCE: Shiotani, T. and Ohtsu, M., "Prediction of Slope Failure Based on
AE Activity," Acoustic Emission: Standards and Technology Update, ASTM STP 1353, S. J. Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999. ABSTRACT: Slope failure occurs transiently due to the brittle nature of failure, of which mechanisms are dependent greatly on the ground properties. Because the failure often produces large-scale damage, techniques for predicting slope failure are in urgent demand. In the present paper, the applicability of AE to the prediction of slope failure is discussed. Firstly, characteristics of AE wave attenuation in the ground are examined. Efficient wave-guide materials, which enable us to detect AE waves generated in the ground with good sensitivity, are studied. Secondly, in order to estimate the failure process inside the ground, a new procedure is proposed to determine the b-value in realtime from the peak-amplitude distribution of acquired AE waveforms. Finally, curvefitting techniques such as graphical analysis and the rate process analysis are applied to AE activity to predict the slope-failure time. KEYWORDS: b-value, graphical analysis, rate process analysis, slope failure, waveguide.
It is well known that AE is useful in giving a variety of information on the fracture state ranging from micro-fracture of crack formation to macro-fracture of crack coalescence. In the case that AE is applied to the monitoring of slope failure phenomena, it shows a great promise in predicting the eventual final failure. This study demonstrates that fracture prediction is possible in laboratory models, by utilizing AE signals radiated i Researcher, Technological Research Institute, Tobishima Corporation, 5472 Kimagase, Sekiyado, Higashi-katsushika, Chiba 270-0222, Japan. 2 Professor, Department of Civil Engineering and Architecture, Kumamoto University, 2-39-1 Kurokami, Kumamoto 860-8555, Japan.
156
Copyright9
www.astm.org
SHIOTANI AND OHTSU ON SLOPE FAILURE
157
from the slope failure. Firstly, characteristics of AE wave attenuation in the ground are examined. Efficient wave-guide materials, which enable us to detect AE waves generated in the ground with enough sensitivity, are studied. Secondly, in order to estimate the failure process inside the ground, a new procedure for calculating the b-value is proposed. Finally, curve-fitting techniques such as graphical analysis and rate process analysis are applied to AE activity to predict the slope-failure time.
Characteristics of AE Propagation in the Ground
A body wave radiated from a simple AE source is known as a spherical wave. In this radiation process, the AE wave loses its energy due to propagation, and the attenuation is normally called "geometrical damping," represented by
A = A0r-"
(1)
where A0 is the amplitude of source, A is the amplitude at the distance o f r from the source, and n: 2 (semi-infinite media); 1 (infinite media); 1/2 (surface wave). The AE wave also attenuates due to absorption. The absorption results from internal friction. Taking into account the absorption, Eq 1 is modified,
A = Aoe-~r -"
(2)
where 2 is the ratio of equivalent attenuation to the absorption, and is given by 2 = 2nhf
V
(3)
where h is the ratio of absorption,f is frequency and v is the velocity of waves.

Characteristics of Attenuation in Toyoura Sand
Due to the homogeneous distribution of the grain size, Toyoura sand is recognized as standard sand for various tests. Dry, wet and saturated conditions of the sand are adopted in the attenuation tests. Twelve AE sensors of 60 kHz-resonance type are set in the sand as shown in Fig. 1. An artificial AE is generated by hi~ing nails of stainless steel, and is detected by the sensors. Peak amplitudes of acquired AE signals are fitted to Eq 2, and 2 is determined by the method of least squares, where v of 220 m/s is measured.
158
FIG. l--Experimental set up of AE attenuation in sand. According to the experimental results, 2, ranged from 0.01 to 0.05. Using derived 2, h was also obtained by Eq 3, Hence, AE attenuation characteristics in frequencies were determined by Eqs 2 and 3.
Results of Attenuation Characteristics in Sand
Figure 2 shows results of AE attenuation at frequencies of 1, 5, 10, 30 and 50 kHz, when 2 is 0.3. Attenuation characteristics are 1.91 dB/cm at 1 kHz, 5.3 dB/cm at 5 kHz and 6.6 dB/cm at 10 kHz. At 30 kHz, AE extremely attenuates when it propagates longer than 8 cm. It is also found that the distance of 5cm at 50 kHz is utmost to detect AE signal efficiently. Accordingly, the effective frequency range for AE monitoring of slope failure is under 10 kHz.
159
FIG. 2--AE attenuation in each frequencies, where ,~ is 0.3.
Effective Wave-Guides [I]

Because the attenuation of AE through the ground is higher in general, wave-guides leading weak AE signals to AE sensors have been used. In this case, the purposes of the waveguides are to drive AE signals to the AE sensors as to reduce the attenuation to the minimum, and allow the wider range monitoring. For the materials of wave-guides, because of their low attenuation characteristics, metallic guides are often used [2,3]. It is known, however, that the wave-guides could not detect AE successfully due to mis-matching of the acoustic impedance. As for other materials, surrounding the wave-guides with gaanular soil [4] and filling rosin into steel pipes [5] and so forth have been performed. For the examination of characteristics on wave-guides, a few papers have so far been reported [6, 7]. These papers mainly described about the characteristics of wave-guides after AE has already attained to the wave-guides. Accordingly, their results are only applicable to the case that the ground would deform in the vicinity of the wave-guides and these AE are detected as deformation-related emissions of the ground. Therefore, when AE generated due to fracatre of the ground is monitored with the wave-guides, further studies are needed. A most important characteristic to be examined on these wave-guides is their adaptability to the geotechnical materials. When AE traveling through the materials is incident to the wave-guides, wave transmission and reflection are made at the boundary between the materials and wave-guides. The wave-guides for this purpose should feature excellent characteristics in the wave transmission.
160
Wave Transmission Loss at Interfaces

An interface is defined as the boundary between two media with different acoustic properties. The transmission loss of an incident wave is dependent on the difference between acoustic impedances Z~oaand Z ~ g ~ . , which are defined as density times wave velocity. The amplitudes of the reflected and the transmitted waves, Ar and A,, are given by,
Ar = A, (Z2 - Z~) Z, + Z 2
(4)
A, = A,
2Z2 (Z 2 + Z~)
(5)
where Z~is the acoustic impedance of the medium where the wave propagates and is incident to the interface, Z2 is the acoustic impedance of the medium where the wave is transmitted from the interface, and A, is the amplitude of the incident wave. The ratio of At~A, is the coefficient of reflection, R,
R = Z2 - ZI
(6)
Z2 + ZI
The traditional and still widely used wave-guides are made of metal [1,2]. This implies that an incident wave is almost reflected at the soil/metal interface, because Z~ is even smaller than Z2 in Eq 6. Therefore, in the case where the wave-guides are applied to the detection of AE underground, the materials of wave-guides should be examined for their acoustic properties.
Effective Wave-Guides
Taking into account characteristics of AE propagation, new wave-guides are proposed. The wave-guide is combined "Unplasticized Polyvinyl Chloride (PVC) Pipes" with "water". The PVC pipe is devised for the decrease of the transmission loss between soil and the outer portion of wave-guides, and water filled in the pipes is for the decrease of attenuation under propagation. AE sensors are set at proper places in the water, and they detect AE waves propagated in the water without notable attenuation.
161
FIG. 3--Configurationsof wave-guides.
Experimentof Wave-Guides
Tested materials for wave-guides are an aluminum pipe and PVC pipe (Fig. 3). The length of wave-guides is 500 ram; the diameters are 70 mrn (PVC) and 78 mm (aluminum), and the thicknesses are 11 mm (PVC) and 20 mm (aluminum), respectively. Two AE sensors, denoted as A and B, are placed on both sides of the wave-guides. Figure 4 shows the setup of AE sensors in the wave-guides: (a) AE sensor attached to the internal wall of the pipes; (b) AE sensor in water; (c) AE sensor directed vertically; (d) AE sensor suspended in the water; and (e) AE sensors conventionally mounted. The wave-guide is set in Toyoura sand of 50% relative density. Five 60 kHz-resonance type AE sensors are arranged at 3 mm separated from the wave-guide as shown in Fig. 3, from #1 through #5. AE measurements are performed by Mistras DSP (PAC), in which each AE wave is recorded at a sampling frequency of 1MHz and with 2k words.
162
FIG. 4--Setup of AE semors in the wave-guides.
Experimental Results
Figure 5 exhibits the AE amplitude detected by the AE sensors of A and B, and shows in decibel terms in which artificial AE output referred to 140dB. "Considered transmission loss" in the figure shows the average of the amplitude when the artificial AE is generated at pulsers #1 and #5 in Fig. 3. The amplitude also includes the effect of the attenuation of AE propagation in the water. "Considered transmission loss and propagation attenuation of 25 cm" shows the results of the average of the amplitude when the artificial AE is generated in the middle of the wave-guide at pulser #3. In the horizontal axis, methods of AE sensor setups and materials of wave-guide are indicated. "Alum, non-fill, hor," for example, represents the case where the wave-guide of aluminum is not filled with water, and AE sensors are mounted horizontally. From Fig. 5, the proposed wave-guides denoted as "Vinyl-chl, w-fill, ver" shows better characteristics than the monolithic aluminum, and the amplitude is twice as high as those of the aluminum pipes. Therefore, it is concluded that the proposed wave-guide combining PVC pipes with water reveals excellent characteristic for detecting AE under ground.
Improvement for Calculating b-Value [8]

AE Amplitude Distribution
Because AE peak amplitude is associated with the magnitude of fracture, the b-value that is defined as a slope of the amplitude dislribution is known as an effective index related to the states of the fracture (see [9,10]). The b-value is originally defined in seismology. In the case of AE applications, however, there are some problems to be solved. Here, one calculation method of the b-value suitable for an AE technique is proposed.
163
FIG. 5--AE amplitude detected by AE sensor of A and B on various wave-guides in decibels.
0.15 ]
14~---~-L~'~
1st
2nd cycle
o."1
,A
=i~ i
3rd
cycle
0.09 0 200
: 400 600 Number of AE hits
; 800 1000
FIG. 6--A result of cumulative calculation of b-value when one set of AE data is executed three times in an overlapped way.
Quantitative Calculation Procedure of b-Value
Determining the b-value, first, the number of the peak amplitudes should be set to be calculated. Roughly classified, two methods for determining the calculation number have been
164
adopted: a) accumulated number from the beginning data; and b) number per unit time. in the case of a), the number determining the b-value is increasing with elapsed time, while in b), because the AE activity increases exponentially with approaching final failure, it is apparent that the number determining the b-value is growing with passing time, and therefore it leads to inconclusive evaluation when the b-value is calculated. Figure 6 shows a result of cumulative calculation of b-value when one set of AE data is executed three times overlapped. If the bvalue is not dependent on the number of data, the result of each cycle should be of the same tendency. It is found, however, that the derived b-value converges to 0.12 with reiteration. Consequentl), these results imply that, for one fracture phenomenon, when the same magnitude of plural fractures occur, there is a great possibility that the earlier fractures are overestimated and later fractures are underestimated. Accordingly, it is important to determine the b-value that the constant number of data is applied to the calculation in real-time. To improve the calculation of the b-value, the number of AE data is formulated by,
fn(a)da=fl
(7)
where n(a) is a number of AE at da and flis a number of AE data. 50 to 100 offlvalues are thought to be an appropriate number by the results of the correlation coefficient when data is fitted to a Gutenburg- Richter's equation.
Qualitative Calculation Procedure of b-Value

The value of AE peak amplitude is varied with such monitoring conditions as: methods of sensor setup, AE traveling media, AE occurrence location and so forth. Accordingly, the AE amplitude distribution is also dependent on these conditions. As a result, it is required for calculating the b-value that a method for determining the amplitude range, which is not to be dependent on the above-mentioned conditions, be established. Because amplitude distributions are unchangeable under the monitoring conditions, statistical values of the distribution are also invariables. Then, applying statistical values such as mean and standard deviation for determining the amplitude range, the effective calculation of b-value is possible. We let the mean of amplitude distributions as/1 and the standard deviation as or, the upper amplitude w 2 and lower wt be formulated as l~,~a~crand/~-~ respectively. Setting accumulated amplitude over w~ and w2, as N (wl) and N (w2), which is obtained by,
N(w 1) = N(kt - ct2cr) = f_~,,o n(a)da
(8)
165
N(w~_) = N(p + a,cr) = ~
n(a)da
(9)
where, the range of amplitude would be (c6+tz2)cr, then Ib-value (Improved b-value) is given by
Ib = l~176 N(w~ ) - logl0 N ( w 2)
(10)
(a, + a2)cr
where, a~ and or, are constants.
E
E
22" 20181614-
[ ] Horizontaldisplacement A Vertical displacement
~, 128
~. ._ a
lo8-
6420 0 --
2000
4000
6000
8000
10000
Elapsed time (sec)
FIG. 7--Results of surface displacements.
Applications of lb- Value A Failure Testof a Full-Scale Embankment Slope--A full-scale embankment slope of 3 m height was fractured by seepage water. AE due to friction between steel wave-guides and soil is monitored. AE signals are detected by 15 kHz-resonance AE sensors, and processed by LOCAN-AT (PAC). Results ofDisplacemem--Figure 7 shows the results of surface displacements. Conceming vertical displacement, remarkable behavior could not be observed. For horizontal displacement, a similarly notable tendency as a precursor of final fracture could not be confirmed.
166
0.6
.................................................................................................. i....................................................................................................... 16oo ]
[ Cumulative I
0.5
1400
Ib-value.. 9 0.4 _= ,>

e~
t
zeroed
1200
I000
_=
=
03
800
"= 0.2 400 b-value per 500 sec
0.1
200
al
0
start of interval
I
2000
4000
6000
8000
10000
Elapsed time (sec)
FIG. 8--Results of AE behavior in a failure test offull-scale embankment slope.
Results ofAE--Figure 8 shows AE behavior in a failure test of full-scale embankment slope. The Ib-value in the figure is calculated by Eq 10 setting flis 100, a, is 1 and ~ is zero. Conventional b-values per unit time of 500 sec and zeroed at start of interval are also shown with cumulative AE hit. For the conventional b-values per unit time and zeroed at start point, steep changes as to give useful information could not be observed. Around 5000 sec, increasing of AE hits is observed, then after 6000 sec, AE hits are activated remarkably. From these results, the period after 6000 sec is predicted as an unstable fracture state. Although, it is difficult to estimate how large scale of fracture is progressing within the slope and whether there is in hazardous conditions or not. Paying attemion to the Ib-value, it is obvious that Ib-vaiue tends to increase until 7000 sec and drop at 7200 sec. Because the increasing trend of the bvalue exhibits the predominance of micro-fracture occurrence. On the other hand, decreasing of the b-value demonstrates the predominance of macro-fracture, and thus it is estimated that the main failure of the slope occurred at 7200 sec. Therefore, the proposed Ib-vaiue could provide useful information related closely to the states of the fracture.
Prediction of Slope Failure Time [11]
In order to estimate the slope failure time, a graphical analysis method based on slope surface strain is defined by Saito [12] and is widely adopted. The analysis, however, is not only

Final exp.re~sio.n ~ '~'
.
. . . . . . . . .
167
!
. . . . . . .
Creep rupture
~'
:A'I
N /'~
L A I . ~ , ~ " / I Midpointbetween ~ndA~ A',

t, b. Time t~ ~" (Rupture-life)
FIG. 9--Schematicprocedure of the graphical analysis method. limited to the strain in geomaterials, but also is applicable to cumulative AE number for forecasting failure time. Here, AE techniques are applied to the tilting box tests [13] of model slope, and two methods are examined for their effectiveness: AE graphical analysis combined with the Ibvalue, and AE rate process analysis [14].
GraphicalAnalysis Method
In the slope failure test of soil materials, application of stress leads to the stage of transient creep, where strain rate increases suddenly at the beginning and then decreases continuously with time. Then the stage of creep follows with steady-state strain rate, and it tums to an accelerating stage leading to final failure. Generally, these three stages are termed primary, secondary, and tertiary. Figure 9 shows the procedure of the graphical analysis method. To trace the rapture-life curve, firstly, the standard time oft~ and At of strain at tl are determined, and A 3 is a strain at t3 which is the time when the forecasting is performed. A'~ is obtained as a middle point between A~ and a projected point to vertical axis by A3. A2 is given as an intersection point between a parallel line to horizontal axis through A't and the strain c u r v e . A ' 3 is given as a projected point o f A 3to this horizontal line. M and N are middle points of A'~A2and A'IA'3; M'A2 and N'A2 are obtained as equals MA2 and NA2 along a vertical line through A2 and t2. Finally, top,are forecasting time t, at the time of t3 is determined as an intersection point between a parallel line to the horizontal axis through M' and a line connected A'~ and N'. Because the method does not require computation, it has been used in practical applications. However, it is noted in this method that the rapture forecasting time is strongly dependent on t~ initially determined as the standard time.
168
FIG. I O--Estimatedrupture curves by graphical analysis on AE parameters.
Application of Graphical Analysis to AE 70meHistory

Because an experimental AE time history is similar to that of displacement, particularly in the tertiary creep range, an applicability of the graphical analysis to AE time history is examined. Figure 10 shows the results of estimated rupture curves from three AE parameters, where 800, 850, 890, 900 sec are adopted as the standard time. 890 sec in the figure is identical to the time when Ib-value is first dropped. It is found that the graphical analysis is applicable to AE, and the failure time could be predicted as the same as that of the displacement. In AE parameters, a cumulative AE event curve is the most effective for earlier forecasting of final failure. Concerning the optimum standard time, the result of 890 sec by Ib-value leads to successful prediction.
Application of Rate Process Theory to AE Activity Rate Process Theory--In concrete materials under uniaxial tests, Ohtsu [14] leads Eq 11
under the condition that AE activity is closely associated with the crack occurrence process. Where a and b are coefficients to define a hyperbolic approximation of the probability function of the rate process model, and C is a constant of integration. These are determined by the
SHtOTANI AND OHTSU ON SLOPE FAILURE
169
1201 10~ 9 600
AE Event(50%) r=0.981 Failure tir~e:1190s~

I
':t
i
< ~1
AE Event(70%) r-=0,978
"
..: :~s ~ i~
Estimated failure time:950s': j

Evaluated AE event .x..: , , '
400
.' 200 Empirical AE event i 0 200 400 600 800 1000 1200 Elapsed time (sec) AE Event(60%) r=0.990
qo,
0
Evaluated AE event E m p i r i c a l ' "
(~OO/o!_._L-- .....
200
i!
480 600 800 1000 1200 Elapsed time (sec) AE event(80%) r=0.953 Failure~rne;11~
1200.1 .......................
i!
10o~
40
Faiturer.~e', l lgOs i'i !

Evaluated AE event
iil
i:
600]
,," ii
:
11i
~ % )-o~-~'%...,,,-~"
2j Empio. __L- ......." 0t ~,oo~

0
50s
Evaluated AE event :" 30s Emp ricalAEevent ~" . .:

(a5%) " !
ii
0 200
200 400 600 800 1000 1200 Elapsed time (sec)
400
600 800 t000 1200 Elapsed time (sec)
FIG. 11--Results of rate process theory applied to AE activity. method of least squares. In this study, let Vbe time, rate process theory is applied to AE activity.
N = CV ~ exp(bV)
(11)
Results o f Application--Figure 11 shows predicted results estimated by the rate process
theory at the tilting level of 50, 60, 70 and 80%. In Fig.l 1, solid lines represent empirical AE
FIG. 12--Cbmparison between rate process results and graphical analysis.
170
data, and dotted lines exhibit estimated curves by Eq t t, where forecasting failure time is determined as the time when the estimated number of AE is identified to total AE number of 1063, acquired empirically. Figure 12 summarizes the comparison between the rate process results and the graphical analysis. In the case of 50% AE events, 950 sec is estimated as the final failure time, and the difference between that and real failure time of t 190 see is 240 sec. In the case of 60%, however, the difference becomes as small as 50 sec. By the graphical analysis, on the other hand, the time which gives us results tess than 50 sec difference is after a tilting level of 88%. rlqaeretbre, it is lbund that the application of the rate process theory to AE activity gives us more promising results than the techniques based on graphical analysis.
Conclusions
1. For effective and inexpensive monitoring of A E data. a wave-guide of"Unplasticized Polyvinyt Chloride (PVC) Pipes" is developed. The pipe is filled with water and AE sensors are suspended inside. 2. It is found that the proposed Ib-value is sensitive to the slope conditions and applicable to failure prediction. 3. The curve-fitting technique by graphical analysis is available for AE activity. 4. In the case that the rate process analysis is applied to AE activity, the prediction procedure is flaker improved, and gives us more promising results than the techniques based on graphical analysis.
References
[1]
Shiotani, T., Sakaino, N., Ohtsu, M. and Shigeishi, M., "Damage Diagnosis of Concrete-Piles after Earthquakes by Acoustic Emission," Proceedings Fourth Far East Conference on Nondestructive Testing, KSNT, Oct. 1997, pp. 579-588. Koemer, R. M., Lord, A. E. and McCable, W. M., "Acoustic Emission Monitoring of Soil Stability," Journal qf the GeotechnicalEngineering Division, Proceedings of ASCE, Vot. 104,No. GT5, May 1978, pp. 571-582. Chichibu, A., Jo, K., Nakamura, M., Goto, T. and Kamata, M , "Acoustic Emission Characteristics of Unstable Slopes," Journal of Acoustic Emission, 1989, pp. 107-112. Kawakami, J., Hattori, H. and Nakao, K., "On the Monitoring Method for Landslide ActMty Using Acoustic Emission," Journal of Japan Landslide Society, 30-2, 1993, pp. 17-24, (in Japanese).
[2]
[3]
[4]
SHIOTANI AND OH'I'SU ON SLOPE FAILURE
171
[5]
Nakajima, I., Negishi, M., Ujihira, M. and Tanabe, T., "Application of the Acoustic Emission Monitoring Rod to Land Slide Measurement," Proceedingsof 5th Conference ofAcoustic Emission/MicroseismicActivity in Geologic Structures and Materials, June 1991, pp. 1-15. Lord, A. E., Fisk, C. L. and Koemer, R. M., "Utilization of Steel Rods as AE Waveguides,"Journal of the Geotechnicat Engineering Division, Proceeding of the ASCE, Vol. 108, No.GT2, Feb. 1982, pp. 300-305. Hardy, H.R. Jr. and Taioli, E, "Mechanical Waveguides for Use in AE/MS Geotechnical Applications," Progressin Acoustic EmissionIV, JSNDI, Jan. 1988, pp. 292-301. Shiotani, T., Fujii, K., Aoki, T. and Amou, K., "Evaluation of Progressive Failure Using AE Sources and Improved b-value on Slope Model Tests," Progress in Acoustic Emission VII, JSNDI, Jan. 1994, pp. 529-534. Mogi, K., "Magnitude Frequency Relation for Elastic Shocks Accompanying Fractures of Various Materials and Some Related Problems in Earthquakes," Bulletin of Eartkquake Research Institute, 40, 1962, pp. 831-853. Sholz, H., "The Frequency-MagnitudeRelation of Microfracturing in Rock and Its Relation to Earthquakes," Bulletin of Seismological Society ofAmerica, Vol. 58, No. 1, 1968, pp. 399-415. Shiotani, T., Aoki, T. and Ohtsu, M., "Prediction of Slope Failure Based on AE Activity," The First US-JapanSymposium on Advances in NDT Proceedings Book, ASNT, June 1996, pp. 239-244. Saito, M. and Uezawa, H, Failure of Soil Due to Creep," Proceedings of Fifth International Conference on Soil Mechanics and Foundations of Engineering, 1, 1961, pp. 315-318. Matsuoka, H. and Sugiyama, Y., "Failure Mechanism and Effective Reinforcement of Granular Soil Slope," Proceedings of the International Symposium on Earth Reinforcement, JGS, Nov. 1996, pp. 803-808.
[6]
[7]
[8]
[9]
[10]
[~l]
[121
[131
172
[14]
Ohtsu,M., "Rate ProcessAnalysisof AE Activityin UniaxialCompressionTest of Core Sample,"Progress in Acoustic Emission V, JSNDI, 1990, pp. 311-316.
AE Sources: Research Topics
Mitsuhiro Shigeishi t and Masayasu Ohtsu 2 IDENTIFICATION OF AE SOURCES BY USING SiGMA-2D MOMENT TENSOR ANALYSIS REFERENCE: Shigeishi, M. and Ohtsu, M., "Identification of AE Sources by Using SiGMA-2D Moment Tensor Analysis," Acoustic Emission: Standards and Technology Update, ASTM STP 1353, S. J. Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999. Abstract: Acoustic Emission (AE) is a phenomenon caused by the emission and propagation of elastic waves generated from micro-cracking. A procedure named "SIGMA" (simplified Green's function for moment tensor analysis) to identify the crack kinematics from AE waveforms has been developed recently. Using this procedure, the crack location, crack orientation and crack type can be determined. Computer software for the SIGMA procedure has been developed and applied in practice. To apply the SIGMA procedure for AE source inversion in a thin plate and other two-dimensional (2D) models, the SiGMA-2D procedure is proposed in this paper. The most distinctive feature of this procedure is projection of 3D moment tensor components onto a plane. For confirmation of SiGMA-2D solutions in practical AE waveforms, the in-plane uniaxial compression tests of plates, which are made by mortar having an internal throughthickness slit, are carried out. The results confirm the applicability of the SiGMA-2D procedure for elucidating the crack mechanisms in 2D models. Further, estimation of crack volume is proposed. For this purpose, the AE sensors are quantitatively calibrated using the Davies-bar technique. A laser opto-interferometer is used to measure the oscillation at the bar end. These results are applied to determination of micro-crack volume. Keyword: acoustic emission, waveform analysis, source inversion, moment tensor, eigenvalue analysis, micro-crack, SIGMA, Davies-bar, sensor calibration, crack volume
The phenomenon, acoustic emission (AE), is a propagation of elastic waves generated by releasing internal energy, as micro-fracturing in an elastic material. Cracking of a solid, which generates two surfaces by conversion of internal energy to surface energy, is mentioned as the typical cause of AE generation. At the same time, a part of the released energy is consumed as propagation of dynamic motion in the elastic material, which is called elastic waves (AE waves). Generally, AE techniques based on the pattern recognition of AE signals and AE parameters such as AE energy and AE count have been used to evaluate the physical property and the quality of materials. However, these techniques bring only qualitative results. To make the cracking and fracturing mechanisms in concrete clear, quantitative information about micro-crack motion at AE source is indispensable. i Associate Professor, Department of Civil Engineering and Architecture, Kumamoto University, Kumamoto 860-8555, Japan. 2 Professor, Department of Civil Engineering and Architecture, Kumamoto University, Kumamoto 860-8555, Japan.
175
Copyright9
www.astm.org
176
Recently, a procedure, which is named "SIGMA", for determination o f AE source kinematics, such as AE source location, directions of micro-crack surface and motion and micro-crack type, has been developed. This procedure has been applied to many AE experiments in concrete engineering. To execute the SIGMA analysis, the procedure requires two parameters, which are arrival time of the primary P-wave portion and its amplitude. For each data set o f AE waveforms, these two parameters must be detected at six or more observation points simultaneously. This simplified procedure provides easy operation and quick results. Furthermore, the SIGMA procedure could be applied to quantitative evaluation of microcrack volumes at AE sources. In the present paper, the procedure is extended and is applied to a test o f cement-mortar plate under uniaxial compression.
Moment Tensor Analysis for Two-Dimensional Problems Summary of Moment Tensor Analysis
Referring to the generalized theory o f AE based on the elastodynamic field in the boundary element method (BEM) [1], the amplitude A (x), o f the primary P-wave portion, which is observed at point x, is simply represented A(x)= where d(x) = output from the AE sensor at observation point x (Fig. 1) p = density of the material vp = velocity of the P-wave r = distance between the source and the sensor s = sensor-sensitivity directions r = direction of wave incidence from the source Re(s,, r,) = reflection coefficient associated with s and r m~O') = moment tensor at AE source y DF = area of the micro-crack surface 1 _1Re(s,, ri)7,7 qm ~ ( x ' ) ' D F 4npv~ 3 r
(i)
Figure 1 - d E observation by a sensor placed on a boundary surface.
SHIGEISHI AND OHTSU ON SiGMA-2D MOMENT TENSOR
177
The moment tensor mt,q can be represented

m pq = ~,blknk 8 m + ~tbl pn q + p.bl qn p
(2)
where 2, ,u = Lame's constants 8 = Kronecker's delta symbol bl = Burgers vector n = normal vector to crack surface
b =bl
Figure 2 - Dislocation model f o r a micro-crack.

n &
el
9
::'
P,":!
:'
,'[ ......
e3
."
Figure 3 - Crack orientations and eigenvectors o f moment tensor. When an elastic wave due to one AE event is detected by more than six sensors arrayed properly, the source location [2] and the amplitudes of the first motion are known. Hence, the independent six components of the moment tensor can be determined by solving the simultaneous equations ofEq. 1 at each observation point.
178
Information about crack kinematics can be obtained from the eigenvalue analysis of the moment tensor. Using Poisson's ratio v, the eigenvalues E, and eigenvectors e, are determined from Eq. 2
mt, q[e ,
e2
e31=
E2
[e,
e2
e,]
0
1-2v = 0 0 where
E3
[Ikn*I 0 +
2v lknk 1-2v 0
(3)
z,-, _1]Ll/+nl
1-2v
l+___nn
It
I1-.I
'-"1
I,n, =(E, + E , - 2 E 2 ) / ( E , - E , ) ; (E, > e 2 > E3) Note that all vectors are normalized in the calculation. Making reference to Fig. 3, original magnitudes of the vectors g, can be reproduced from the following equations g, = .42 + 21,n k .e, g3 = ~ 'e3 (4) (5)
Thus, the direction of the normal n to the crack surface and the direction 1 of the crack motion can be recovered. A quantitative classification of the crack types into shear mode, tensile mode and mixed mode is developed. The eigenvalues of the moment tensor are decomposed into a shear component; a deviatric compensated linear vector dipole (CLVD) component and a hydrostatic component [3]. Generally, AE sources consist of a mixed mode with tensile and shear components. Setting the ratio of shear component as X, that of CLVD component as Y and that of hydrostatic component as Z (Fig. 4), all eigenvalues are uniquely decomposed as follows X = e2 - e3
eI
(6) (7) (8)
y = 2 (e, - 2e 2 + e 3) 3 eI Z = e, + e 2 + e 3 3 e~
179
Intermediate
Minimum
--.5Y
X Y Z
$
Hydrostatics (Z)
Shear (X)
CLVD (Y)
Figure 4 - Eigenvatue decomposition of a moment tensor. In the case that the moment tensor components are determined from detected AE waveforms, inherent information on locations, types and directions of cracks is determined by moment tensor analysis of AE sources. This procedure for moment tensor analysis for AE is named "SIGMA analysis" and program code for computer calculation has already been developed and distributed. The experimental verification of this procedure in concrete structures has already been reported by several researchers [4-7].
SIGMA Application to Two-Dimensional Problems

For only the two-dimensional (2D) problems, a procedure for identification of AE source was researched by Ouyang et al [8]. However, SIGMA analysis can be treated similarly in the 2D models. In the 2D case, the moment tensor is projected on to x~- x 2 plane. Therefore, the eigen vectors of a moment tensor in respect with x3 direction become zero, as/3 = n3 0. The moment tensor components in Eq. 2 are represented by
=
mm =
3~l~n, + 2~tl~n~ sym.
~t(t~n +12nl) 2 ~lkn k + 2~t12n2
~ ]
(9)
~,lknk
where 1 n k = I~ n 1+ l2 n2 k
180
In the case that all AE sensors (observation points) are located and directed on the same plane, the component m33 of the moment tensor cannot be obtained from the detected AE waveforms because the out of the plane displacement cannot be measured. Therefore, the estimation of the m33 component is performed in the following manner based on Eq. 9
m33 = ~.(l,n, + 12n2)= (~./2(~. + kt)Xm,, + m22 ) = v(m,, + m22 )
(10)
In the same way as for the three-dimensional (3D) problem, crack types can be classified by decomposition of eigenvalues, and crack orientations are also determined. This procedure is named "SiGMA-2D analysis", and some numerical simulations have already demonstrated the applicability of SiGMA-2D [9]. Evaluation of Micro-crack Volume
Micro-crack Estimation by SIGMA Analysis

Expanding on the SIGMA analysis of AE, a quantitative method for evaluation of micro-crack volumes is developed in fracturing of a material. Note that all components are normalized in the calculation. Hence, original magnitudes of the moment tensor must be reproduced for evaluation of the micro-crack volume. It is considered that the original moment tensor m;q is the product of the maximum component ]mpql in the original moment tensor and the normalized moment tensor mm', or mpq = Impql mpq'. Then Eq. 1 is modified as
Re(s,, r,)~tp)lqmpq By relation ~, =
Ar
, =
l lmeql'DF3 4rcpvp
(ll)
2lay~ (1-2 v), Eq. 2 can be rewritten as mpq =bla(1-~2vl, n,8 ~ +lpnq +lqnpl
(12)
The direction ! and n of the crack kinematics are known from Eqs. 4 and 5. Then, the maximum component of Eq. 12, max(m), can be derived by substituting the unit vectors, ! and n, into Eq. 12,
[m~[ = blamax(m)
Furthermore, substituting Eq. 13 into the Eq. 1 1, the equation for micro-crack volume b.DF can be determined by Ar 1 lamax(m)(b. 4npvp 3
(13)
Re(s,,ri)~,yqm'pq
DF)
(14)
If the sensitivities of sensors are equalized, the calculated moment tensor consists of relative values. To evaluate the magnitude of the micro-crack quantitatively, the sensor outputs should absolutely be calibrated.
SHIGEISHIAND OHTSUON SiGMA-2DMOMENTTENSOR

Sensor Calibration Using Davies-Bar Technique
To determine the micro-crack volume by the procedure explained above, it is indispensable that the AE sensor measures an absolute quantity for displacement. Hence, the dynamic characteristics of the AE sensor are examined. Generally, a simple physical phenomenon based on a known law must be utilized for characterizing sensors because it is essential to have a clear knowledge on an input signal into the sensor. Further, it is ideal that the method is extremely precise and provides easy-operation. A simple method for characterizing AE sensors using Davies-bar as a propagating media of elastic waves was presented [10]. Figure 5 shows the principle of this method. The sensor response in the frequency domain corresponding to the input signal as displacement can be characterized. When a projectile strikes the end of a bar, a compression pulse is generated as an elastic wave. This wave propagates along the bar as far as the bar deforms elastically.
181
Projectile StrainGauge 9~ " DaviesBar \-~
I=
.] iLas'e';'i'nie~erom'et'er ]
" q ~ i i i l ] " ............................... [AE
Sensor
,['Charge [Amplifier Amplifier I ['"
I Oomu,er
rans,ent emo t
........................
Figure 5 - Block diagram of the AE sensor characterization. The equation of the one-dimensional elastic wave is represented as ~2u = C 2 a2u ~t 2 C3Z2 C = x/Ex/~
(15)
(16)
Here, u is the displacement of the particle of the bar in the axial direction, t is time. z is spatial coordinate parallel to the axis of the bar. C is the velocity of the longitudinal elastic wave in the bar. E is Young's modulus of the bar. p is the density of the bar. The wave reflects at the other end of the bar where AE sensor has been attached on, and propagates in the opposite direction as a tensile pulse. At the instance o f reflection, the wave produces the displacement given as
act)= 2cf,/t) t
(17)
Here, d(t) is the displacement and ~(t) is the strain of the incident pulse at the end of the bar. It is assumed that the cross-sectional area and the acoustic impedance of the bar are larger than those of the sensor, and that the influence of the presence of the sensor on the wave reflection is negligible. When the strain pulse 6,,(t) is measured by the strain gauges glued to the bar at the distance L apart from the end, the displacement at the
182
sensor position is derived from Eq. 9. The transfer function G(jco)of the sensor, describing the relationship between the output signal am(t)and the displacement, is expressed by
G0co)=
2CL[G,(t _ L ) ]
(18)
Here, L[ ] is the Laplace transform operator, j is the imaginary unit and co is the angular frequency. The gain characteristic go(co) and the phase characteristic ~a (co) of the sensor are given as follows.
ga(o)=lGd(jO ]
r = arg[Ga (jo)]
(19) (20)
Here, the waveforms of strain-gauge output and the AE sensor signals are recorded by digitizing into N samples in length at At sampling rate. The spectrums, F/and FA are given by Fast Fourier Transformation (FFT) of the strain-gauge output and the AE sensor signal, respectively. From the spectrum of the gauge outputs as the velocity, the displacement allowing the time function to the AE sensor is represented as
1 e = ~_~ F E
LT~J
(21)
The sensor sensitivity S can be given by
s=FA=
FA = 27tf F__ A _EA
(22)
1L_ G 2~f
&
In practice, it is necessary that the response of the whole system including Daviesbar with the sensor should be clarified. A laser interferometer could be useful for this purpose.
Experiment for Cement Mortar Plates
According to the SIGMA procedure and sensor calibration, evaluation of microcrack volume was applied to the practical AE waveforms, which were recorded in a compression test of a plate specimen made of cement-mortar.
AE Sensor Calibration Using CementMortar Bar

For evaluation of micro-crack volume, six wide-band AE sensors were made using PZT piezoelectric elements, characterized in advance. Then, they were used in AE waveform sampling during a compression test. A Davies-bar made of cement-mortar of the same mix proportion as the specimen and was 1 350 mm in length and 20 mm diameter. Two strain gauges, which were made by alloy leaf of copper (Cu) and nickel (Ni) and 10 mm long, were bonded 400 mm from the end. Elastic waves were generated and propagated by the strike of a hammerhead,
183
which was made of carbon steel with the same diameter as the Davies-bar. The amplification of the signal from the strain gauges was defined so that the recorded 1 volt was equivalent to 5 000/.t strain. It is noticed that the length of the strain gauge is long enough to ignore the size of the sand, which is qualified for JIS Concrete Materials (R 5201-1997), for the mortar. The average size of the sand grain was 200/an. The measured velocity of the longitudinal wave was 3 543 rn/sec. These results lead to the fact that the strain gauge could detect frequency components up to 350 kHz theoretically. The sensors were bonded on to the end face of bar using silicon grease. Silicon grease was used to turn air out between the sensor and the rod surface. This procedure was taken in the experiment of a mortar plate again. Thus, the characteristics of the whole AE measurement system could be calibrated. Typical waveforms of the gauge output and the AE sensor output recorded in the Davies-bar test are shown in Figs. 6 and 7. The spectrums of the each waveform are also shown in Figs. 8 and 9. Figure 10 shows the characteristic curve o f this AE sensor by Eq. 22. It shows that the resonant frequency of this sensor is about 60 kHz. The sensor sensitivities calculated up to 60 kHz are summarized in Table 1. TABLE 1 - Sensor sensitivities. Sensor # 1 2 3 4 5 6 Sensitivity, V/mm 6.32 X 6.32 X 3.56 X 3.56X 3.98 X 11.24 X 104 104 104 104 104 104
>
0.5 0
9 -0.5
-1 ! I I I
0.001 Elaosed Time (see)
0.002
Figure 6 - Output of strain gauge.
184
1 >, 0.5 0 9 -0.5

_]
i
'
0.001 Elapsed Time (sec)
0.002
Figure 7 - Output of AE sensor.
'
-,
100
200
300
Frequency (kHz) Figure 8 - Spectrum of signal from strain gauge.
'
'
'
100
200
300
Frequency (kHz) Figure 9 - Spectrum of signalfrom AE sensor.
185
~ ~
~
4 2
0 0 100 200 300
Frequency (kHz) Figure 10 - Sensor characteristics curve.
Uniaxial Compression Test of Cement Mortar Plate

To evaluate micro-crack volumes, a cement-mortar plate was employed for inplane uniaxial compressive loading tests. The sketch of the experiment is shown in Fig. 11. Mixture proportion was arranged so that the mass ratios of cement, sand and water were 1 : 2 : 0.65 the same as the Davies-bar used at the sensor calibration. The plate specimen dimensions was 150 mm x 150 mm x 15 mm and contained a through-thickness slit o f I mm wide and 30 mm long at the center o f the specimen. The inclination angle o f the slit was 45 degrees. The configuration of the specimen and sensor locations is shown in Fig. 12. Six calibrated AE sensors on both sides of the specimen detected AE waveforms. Total gain o f the measurement is 40 dB. One AE waveform is digitized into 1 024 words in length at 1MHz sampling frequency.
Sensor #1 Sensor #2 Sensor #3 ~
Sensor #4 Sensor #5 Sensor #6 Internal Slit
Figure 11 -Overview of in-plane uniaxial compression test of plate.
186
150
!
~-
I
,
!
~
1~35,
Figure 12 - Plate specimen and sensor arrangement. Applying the "SiGMA-2D" analysis (which is a version for two-dimensional AE problem of the SIGMA) AE sources were identified. The result is shown in Fig. 13. The arrow symbol (<--~)indicates the AE sources classified into a tensile type and the direction of the arrow symbol represents the crack opening direction. The cross symbol (x) indicates shear-type AE sources and their two directions are o f crack normal and crack motion, The observed final failure surface o f the specimen is shown by dotted lines in Fig. 13. Evaluation o f micro-crack volume was performed for AE sources, with the result that their volumes were between from 0.33 mm 3 to 44.00 m m . Estimated volumes o f micro-crack sampled randomly from the result are shown in Table 2. However, the reliability of these volumes for micro-cracks have been not authenticated yet because a practical means for observation of the interior (micro-cracks) has not been found. It could be considered that these estimated volumes are index values including relative magnitudes of AE sources.
,,,
x,,,
Figure 13 - Source locations and crack motions.
187
Table 2 - Results of crack volume estimation. Shear type Hit # 16 50 60 62 70 023 024 034 048 050 358 360 367 374 375 Volume, mm 3 4.59260 8.35497 0.71634 1.02402 4.26770 5.35833 4.73927 5.36791 0.94326 1.85148 1.75165 3.55262 6.83452 2.71419 1.64583 Hit # 81 146 165 194 467 1 086 1 154 1 162 1 182 1 204 2 400 2 518 2 523 2 542 2 575 Tensile type Volume, mm 3 1.26234 0.34080 0.73336 0.43135 2.19672 0.41629 2.32983 0.86267 0.67694 0.99309 0.48238 0.39802 2.71605 1.83941 0.41912
Conclusion
A quantitative method for evaluation of micro-crack volumes has been developed using acoustic emission (AE) waveforms, detected during fracturing of a material, extending the SIGMA procedure for an AE moment tensor analysis. For application of this method to practice, characteristics of AE sensors in the frequency domain are examined using a Davies-bar technique. To examine the applicability of the method to evaluation of micro-crack volumes, a cement-mortar plate was employed for in-plane uniaxial compressive loading tests. The estimated volume could be referred to as relative for AE source magnitude. To improve the precision of the sensor calibration technique, the study is in progress.
References
[1] Ohtsu, M. and Ono, K., "A Generalized Theory of Acoustic Emission and Green's Functions in a Half Space," Journal of Acoustic Emission, Acoustic Emission Group, Vol. 3, 1984, pp. 124-133. Ohtsu, M., "Simplified Moment Tensor Analysis and Unified Decomposition of Acoustic Emission Source: Application to in Situ Hydrofracturing Test," Journal of Geophysical Research, The American Geophysical Union, Vol. 96, 1987, pp. 6211-6221. Knopoff, L. and Randall, M. J., "The Compensated Linear Vector Dipole: A Possible Mechanism for Deep Earthquakes," Journal of Geophysical Research, The American Geophysical Union, Vol. 75, 1970, pp. 4975-4963. Ohtsu, M., Shigeishi, M. and Iwase, H., "AE Observation in the Pull-out Process of Shallow Hook Anchors," Proceedings of the Japan Society of Civil Engineers, The Japan Society of Civil Engineers, No. 408, 1989, pp. 177-186.
[2]
[31
[4]
188
[5]
Yuyama, S., Okamoto, T., Shigeishi, M. and Ohtsu, M., "Acoustic Emission Generated in Corners of Reinforced Concrete Rigid Frame under Cyclic Loading," Material Evaluation, The American Society for Nondestructive Testing Inc., Vol. 53, 1995, pp. 409-412. Yuyama, S., Okamoto, T., Shigeishi, M. and Ohtsu, M., "Quantitative Evaluation and Visualization of Cracking Process in Reinforced Concrete by a Moment Tensor Analysis of Acoustic Emission," Material Evaluation, The American Society for Nondestructive Testing Inc., Vol. 53, 1995, pp. 751-756. Grosse, C., Reinhardt, H. and Dahm, T., "Localization and Classification of Fracture Types in Concrete with Quantitative Acoustic Emission Measurement Techniques," NDT & E International, Elsevier Science, Vol. 30, 1997, pp. 223231. Ouyang, C., Landis, E. and Shar, S. P., "Damage Assessment in Concrete Using Quantitative Acoustic Emission," Journal of Engineering Mechanics, Vol. 117, 1991, pp. 2681-2898. Shigeishi, M. and Ohtsu, M., "A SIGMA Analysis of the 2-Dimensional PMMA Model," Progress in Acoustic Emission, Vol. 6, The Japan Society for NonDestructive Inspection, 1992, pp. 211-217. Ueda, K. and Umeda, A., "A Study on the Characterization of Accelerometers Using Davies' Bar Technique," Transactions of the Japan Society of Mechanical Engineers, The Japan Society of Mechanical Engineers, Vol. 57, 1991, pp. 143147.
[6]
[7]
[8]
[9l
[10]
Transportation Applications, Standards, and Methodology
John M. Carlyle, I Harvey L. Bodine, 2 Steven S. Henley, 2 Robert L. Dawes, 2 Robert Demeski, 3 and Eric v. K. Hill3 P R A C T I C A L AE M E T H O D O L O G Y F O R USE ON A I R C R A F T
REFERENCE: Carlyle, J. M., Bodine, H. L., Henley, S. S., Dawes, R. L., Demeski, R., and Hill, E. v. K., "Practical AE Methodology for Use on Aircraft," Acoustic Emission: Standards and Technology Update, ASTM STP 1353, S. J. Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999. ABSTRACT: The first production use of practical acoustic emission (AE) on aircraft was on the F-111, where more than 240 U.S. and Australian aircraft have been successfully monitored during cold proof testing since 1987. A key factor was the design of an instrument which automatically configured itself at power-on, so that aircraft testing could take place in a highly efficient and reliable manner. The second production use of AE on aircraft was on the VC-10, where an entire fleet of 22 aircraft was monitored 40 times during pneumatic proof pressurization. The monitoring of this large transport required anywhere from 282 to 313 narrowband AE sensors per aircraft. In both aircraft types, numerous significant defects were discovered through AE-based nondestructive testing. Finally, we present on-going in-flight AE research. This new research uses digital waveform processing of wideband AE signals and offers the potential of adding new and complementary capabilities to classical, narrowband AE. KEYWORDS: acoustic emission, aging aircraft, processing, global defect detection, proof testing nondestructive testing, signal
Introduction The average age of aircraft in both military and civilian service is increasing. As a consequence of this aging process, high cycle fatigue from pressurization and aerodynamic forces causes very small defects to appear at widespread locations throughout the airframe. This multi-site damage (MSD) is not easily amenable to being detected and mapped by conventional nondestructive inspection. Furthermore, the presence of MSD makes the evaluation of the serviceability of the airframe extremely
1 Chief Scientist, Carlyle Consulting, 1009 Buckingham Way, Yardley, PA 19067 2 Martingale Research Corp., 1485 Richardson Drive, Suite 110, Richardson, TX 75080 3 Embry-Riddle Aeronautical University, Aerospace Engineering Department, 6100 Clyde Morris Blvd., Daytona Beach, FL 32114-3900
191
Copyright9
www.astm.org
192
difficult. A classic example of this problem is the Aloha Airlines accident, in which a section of the upper lobe separated from the airframe during flight at 7.6 km (25,000 ft). What is required is a nondestructive evaluation technique that can detect, map and quantify the extent of MSD. Lasers, eddy current, thermographic and ultrasonic methods have been proposed and studied for mapping MSD, but they appear to be fundamentally flawed in that they require visible or tactile access to the entire airframe structure. Wheel wells, front and rear pressure bulkheads, and the bulkheads at the front and rear of the wing torque boxes are typically filled with shafts, hoses, tubes, wiring, electronic equipment, air-packs and other items that make it impossible to see or touch the pressure structure in these areas. MSD detection here requires remote sensing of flaws. A nondestructive evaluation method that has shown promise, but which is not yet widely used for general aircraft application, is acoustic emission. Acoustic emission is essentially seismology for a structure. As in seismology, flaws are detected remotely from their actual source position using AE. This inherent ability completely eliminates the access problem which affects all other nondestructive methods. The promise of AE has been demonstrated in various successful applications on aircraft, including the work of Hutton on the Macchi MB-326 (crack growth detected in flight) [I], McBride on the Northrop CF-5 (cracks detected well before eddy current could confirm them) [2], and Carlyle on the General Dynamics F - I l l and the Vickers VC-10 (entire airframes monitored globally for crack growth) [3, 4, 5, 6, 7]. Despite these successes, AE is still not widely employed on aircraft. Part of the reason for this is a negative impression of the technology in some quarters. In the U.S. Air Force AE gained a bad reputation during an in-flight KC-135 wing structure inspection project. The basic problem was due to poor system design - the signal processing techniques utilized did not properly eliminate noise, and so generated a false alarm that was shown during flight to a general who was piloting the aircraft. For the U.S. Navy a bad impression of AE was given during full-scale testing of a F-14 bulkhead. The basic problem was that the entire bulkhead was not monitored, only a small portion which was thought most likely to fail. The bulkhead catastrophically failed in an unexpected area in front of an admiral, completely without warning from the AE system because the crack was outside of the monitored area. Both of these so-called problems with AE, a type I error (false detection) in the first case and a type II error (missed detection) in the second, were due solely to implementation failures and were not due to any fundamental failure of AE technology. Nevertheless, poor impressions were made at very high management levels, which unfortunately persist to this day. We believe that AE can definitely contribute to aircraft safety, and that dwelling upon several poorly implemented monitoring projects is a mistake. As proof, we offer the following success stories.
Success Stories AE-based nondestructive testing has been applied successfully to aircraft in a production environment, not just in experimental situations. As such, it has been
CARLYLE ET AL. ON METHODOLOGY FOR AIRCRAFT
193
proven to be a cost-effective and powerful tool for locating damage and failures in structures where other technologies were either not useable or not practical. Two of the more notable production successes of AE-based NDT are the General Dynamics F-111 fighter-bomber aircraft and the Vickers VC-10 transport aircraft.
F- 111 Production AE-Based Nondestructive Testing
The AE testing of the F - I l l m a r k e d the first production use of AE on inservice aircraft. All U.S. and Australian F-111A's and EF-11 l's undergoing cold proof testing since May 18, 1987 (about 240 aircraft) have been monitored with AE. The F-111 fighter/bomber is capable of Mach 2.5 at better than 18.2 km (60,000 ft). First delivered in the 1960s, its most unique feature is its variable sweep wing. The wing is supported by a D6AC steel wing carry-through box, which has the job of holding the 22.8 m (75 ft) long, 19.2 m (63 ft) wide, 5.2 m (17 ft) high, 45,300 kg (100,000 lb) aircraft aloft. Although D6AC steel has high strength and toughness, it also has a small critical crack size. Failures of a wing carry-through box in 1968 during ground based fatigue testing, and a wing pivot fitting in 1969 which resulted in the loss of an aircraft, caused the Air Force to institute a thorough nondestructive inspection program for the F111 fleet [3]. The structure (a portion is shown in Figure 1) is quite complicated, making it a challenge to find small defects. Innovative use is therefore made of the ductile to brittle transition temperature, a metallurgical condition in which small flaws can be made to propagate catastrophically. The entire aircraft is cooled using liquid nitrogen to -40 ~ C (-40 ~ F) inside of a 24.4 m (80 ft) by 18.3 m (60 ft) by 9.1 m (30 ft) chamber, and loaded to +7.3 g and -3.0 g with theowing sweep in the 0.14 rad FIG. 1 - F - I l l structure withAEsensors. (26) and 0.31 rad (56 ~ positions. Load is applied to the aircraft through fixtures that attach to the wing stores positions, the nose gear, the main landing gear, the arresting hook, and the tail. The wing tip deflects about 1.2 m (48 in.) under maximum positive load. The test program was designed to enhance the safety and longevity of the F-111, since any failure that occurs during the testing could potentially occur in flight. The major purpose of the cold proof testing is to test the integrity of the steel components of the F-111; failure would result in visible damage to the aircraft. The nonferrous components are also highly stressed, but their failure may not cause visible
194
damage. Because the proof testing is conducted late in the re-work cycle, it is important to ascertain quickly if non-visible structural damage has occurred. AE monitoring was implemented on the F-111, because AE offered a high sensitivity for locating secondary structural damage as the cold proof test was being conducted, and because AE analysis is capable of a high degree of automation. Discussions on the stress analysis and historical failure sites of the F-111 during cold proof testing were held with Air Force structures personnel. This resulted in a priority listing of the areas which needed to be monitored. They were (in descending order of importance) the wing carry-through box, the tunnel structure, the 770 bulkhead, the wings, the main gear well, the 560 bulkhead to 770 bulkhead structure, the forward lower fuselage structure, and the forward upper fuselage structure. The next concern was the number and location of sensors needed to obtain the most effective location in the areas to be monitored, bearing in mind that the contract called for _+_0.3 m (12 in.) accuracy. Normally optimum sensor location can be calculated from the acoustic attenuation of the structure, which is a single number expressed in dB per unit distance. However, because of the complexity of the F - I l l structure the attenuation was not unique. It not only varied from place to place, it also had different values for various directions from a single point. Thus the number of sensors required and their placement had to be determined experimentally instead of analytically. In the end 28 narrowband 300 kHz sensors were used. Sensor spacing varied from 0.5 m (20 in.) to 6.7 m (265 in.), and averaged 2.4 m (93 in.) over the aircraft. The AE monitoring system used for the F-111 project was a specially modified commercial instrument [4]. Because three dimensional location of the AE events was necessary in the F-I 11, additional features had to be developed. These additional features included color graphics hardware to depict the aircraft in two views; color graphics software to draw the aircraft, calculate the location, rank the severity, and plot the AE events; main data acquisition program modifications to pass data to the color graphics software; and initialization routines to automatically configure the entire system at poweron so that aircraft testing could take place in a highly efficient and reliable manner. The operator interface was designed to present the information in a timely and intelligent fashion with minimum input requirements. The primary presentation is a color graphics CRT, which has two separate displays. The main color screen, shown in Figure 2, depicts the top and side views of the F-111. Superimposed upon each of these views are the locatable AE events, colored green, yellow or red according to severity, as well as the sensor locations which show as small black dots. AE locations are displayed essentially in real-time, i.e., they appear on the color display within a second of being detected on the F-I 11. All AE results were located to + 0.15 m (6 in.) on a scaled diagram of the F-111. The second color screen is a numerical listing of the attributes of each locatable AE event, with each line colored green, yellow or red according to severity. This display aids the operator in determining how many events occurred at a particular spot, since on the aircraft display AE events can superimpose, with yellow taking priority over green and red taking priority over everything. The numerical listing shows the fuselage station, water line and butt line position (in inches) for each Iocatable AE event, as well as the load (in percent of limit) at which each event occurred. Another use for the
195
numerical display is in determining the exact severity of the AE event, instead of relying on the green, yellow and red color coding.
YELLOW J" .-'F'~ 7 f
RED j/.-/'.
YELLOW ~/" r
/"
,/
J 2
YELLOW
RED
~"\.
,\ \,
' "~...
\ ~
NOTE: COLOR REMOVEDFOR REPRODUCTION
""~...__~
FIG. 2 - Side and top ofF-111, showing results of AE monitoring. Of particular interest in Figure 2 are the cluster of yellow AE events in the tunnel area, the yellow AE event on the starboard wing, and the red AE event in the left front comer of the wing carry-through box. The yellow events in the tunnel area were caused by the formation of a crack in the forward tunnel truss during cold proof testing. This is an arch shaped piece of aluminum; its aft counterpart is shown in Figure 1. The yellow events on the starboard wing (two AE events happened at the same site) are believed to be due to the formation of a disbond in the wing skin to wing pivot support assembly joint. The red event is due to the failure of a large bolt which helps hold on the cover of wing carry-through box (the wing carry-through box also serves as a fuel tank). The failure of the bolt could not be verified immediately, as fuel tank sealant held it firmly in position. However, ultrasonic inspection of bolt lengths revealed that one bolt was shorter than its installed length. Similar bolts on the rear left edge of the wing carry-through box may be seen in Figure 1. Noise discrimination was accomplished using several AE signal attributes. It was found during the project that AE signals caused by actual airframe damage had characteristics which included high amplitude and high energy in the 300 kHz frequency range. Loading noise and normal airframe reactions, although they were sometimes loud enough for a human to hear, did not have high amplitudes and energies in the ultrasonic range. This fact was exploited by building a filter for amplitude and energy with three different thresholds. Signals exceeding both the first amplitude threshold and the first energy threshold were colored green; these were termed "moderate" signals. Those signals which exceeded both the second amplitude threshold and the second energy threshold were colored yellow; they were termed "strong" and triggered a low pitched audible alarm. Signals which exceeded both the third amplitude threshold and the third
196
ACOUSTIC EMISSION:STANDARDS AND TECHNOLOGY
energy threshold were colored red; they were termed "severe" and triggered a high pitched audible alarm. It should be noted that three other AE signal characteristics are measured by the AE instrument. They were not used to categorize the severity of the AE events, however, because they were so susceptible to dispersion of the AE wave form. Various system parameters were required for proper operation of the data acquisition computers of the AE instrument; these included gain, detection threshold, three detection time out values, sampling interval and dual alarm thresholds. In addition channel neighbor relationships and wave propagation times were needed for the location algorithms. Initial values for these parameters were derived during the sensor coverage experiments, and the values were refined during three cold proof tests. All of this data was recorded in an automatically executed computer program, so that when the system powered on graphs were set up, severity classification levels defined, amplifier gains and detection thresholds set, alarms adjusted, etc., without any intervention from the operator. The goal was to automate the system to the point where minimal operator training was needed. This approach also enhanced quality control, since the cold proof testing procedure, programs and parameters could be verified against factory recorded check values through computer command. As a final quality control measure, a written check list was developed so that no item would be inadvertently forgotten during testing.
Vickers VC- 10 Pressurization Proof Testing

The second successful implementation of production-type AE testing was done on the Vickers VC-10. The entire fleet of the U.K. Royal Air Force VC-10 transports and tankers has been tested several times (a total of 40 tests on the 22 aircraft fleet). The goal of the AE monitoring on the VC-10 aircraft was to provide a real-time capability to the RAF for detecting degradation of structural integrity in VC-10 aircraft during 82.7 kPa (12 psi) proof pressurization testing (1.33 times the normal in-flight limit). Specifically, it was desired to detect the presence of growing cracks (or other significant flaws) if they occurred, in order to prevent serious damage to the aircraft. In several acoustic monitoring trials conducted for the RAF it had been demonstrated that the requisite real-time sensitivity needed to detect structure threatening, growing defects existed; and that there existed a real-time method of monitoring the AE output to detect any growing defects as they grew [5]. Given these crucial capabilities, AE personnel had the authority to advise the pressurization team to rapidly reduce pressure in the aircraft if growing cracks (or other significant flaws) were detected, offering assurance to the RAF that the VC-10 aircraft would not be damaged during proof pressure testing. The entire pressure hull of the 48.5 m (159 ft) long, 44.5 m (146 ft) wide, 12.2 m (40 ft) high, 136,000 kg (300,000 lb) VC-10 aircraft was monitored. This included the entire lower belly fore and aft of the wing, the nose-crown area, the tail section around the stub wing and vertical stabilizer, the entire crown section, the front pressure bulkhead, the cockpit structure, the Doppler bay, the nose landing gear well, the forward lower cargo door, the main cargo door, the aft bulkhead of the forward cargo bay, the roof of the slat bay, the roof of both main landing gear wells, the forward bulkhead of the rear cargo bay,
197
the aft lower cargo door, and the aft pressure bulkhead. Depending upon which model of the three types of VC-10 aircraft was being tested, the AE monitoring was accomplished with anywhere from 282 to 313 narrowband 150 kHz sensors per aircraft. Figure 3 shows the instrumented nose of a VC-10. Pressurization of the fuselage was accomplished using two air compressors feeding at a rate of up to 5,270 cm3s "1 (50 ft3s "1) through a specially modified overwing emergency door. The entire pressure cycle used to load the VC-10 consisted of a climb at 3.4 kPa (0.5 psi) per minute to 41.3 kPa (6 psi), a 5 minute hold, a F I G . 3 - AE monitoring on a Royal Air Force VC-IO. descent to 27.4 kPa (4 psi), a 2 minute hold, a climb to 62.0 kPa (9 psi), a 5 minute hold, a descent to 48.2 kPa (7 psi), a 2 minute hold, a climb to 68.9 kPa (10 psi), a 5 minute hold, a descent to 55.1 kPa (8 psi), a 2 minute hold, a climb to 75.8 kPa (11 psi), a 5 minute hold, a descent to 62.0 kPa (9 psi), a 2 minute hold, a climb to 82.7 kPa (12 psi), a 5 minute hold, and finally a descent to sea level [6]. The AE monitoring system provided the RAF with a real-time capability for detecting degradation of structural integrity in VC-10 aircraft during proof pressure testing. This was provided, first, by using the channel activity lights on the front panels of the AE instruments, which gave the operators an immediate indication (within microseconds) that a sensor was receiving signals; and second, by using graphs of signal amplitude and signal arrival rates per channel which showed (within seconds) the characteristics of the signals that were being received. Because AE is the direct result of stress, if a defect continued to energetically emit sound during a pressure hold period it was an obvious indication that harmful damage was occurring. Conversely, if the defect stopped emitting or its emission characteristics changed drastically when the pressure was held constant it was an indication that the defect was benign. To establish that very small and/or growing cracks could be detected in the structure of the VC-10 British Aerospace fastened "dog-bone" shaped tensile specimens containing fatigue cracks to the outer surface of a test aircraft. The specimens were attached to the aircraft using epoxy and rivets in such a fashion that they spanned the space between adjacent stringer members. Thus, when the aircraft was pressurized to 59.3 kPa (8.6 psi) the fatigue cracks were subjected to tensile forces. The dimensions of the specimens were chosen such that the pressure loading would be sufficient to cause the fatigue cracks to extend. In this manner, known cracks were made physically and
198
acoustically "part" of the airframe so that the sensitivity of AE could be readily checked. This would be done by showing that the standard sensor arrays proposed for use on the VC-10 could locate the growing cracks and by showing that AE data from the cracks was present during load hold periods. UC-IBXUI~B8 No~le awl C,~k~it T 1 ~ 4 / ~ ~.3~ . ~ ?
=me
LEADBREAKS
~ Lie-
~LEAK
NOISE
CRACKI~
e-IN"
IHI
1~1
FIG. 4a - Simulated acoustic signals on samples mounted at two locations.
FIG. 4b - Acoustic data obtained during pressurization. Diffuse patch is from a small leak source.
Figure 4 demonstrates the ability to locate the pre-cracked specimens during pressurization of a VC-10. The "a" part of the figure shows how a simulated AE signal can be located prior to pressurization. The purpose was to show that the standard sensor spacing of 1.8 m (72 in.) for the VC-10 aircraft was proper, and also to direct the eye to the expected location of actual crack signals. The "b" part of the figure shows locations obtained during the actual pressurization of the VC-10; the locations at the site of the specimens are due to cracking during actual pressure loading. Figure 4a shows the location of two specimens due to simulated AE signals, one at fuselage station 237.5, stringer 4.5 right and the other at station 253, stringer 4.5 right. Figure 4b shows the AE activity from the cracking of these specimens during pressurization, plus a few signals from a small leak source. It is clear that the cracking of the two tensile specimens was detectable with AE through the use of the post-test source location program. For AE to be successful in preventing serious damage to a VC-10 during proof pressurization, however, it was necessary to determine that cracking was occurring while the aircraft was actually being loaded. This was done by using channel activity indicators and a series of plots that were updated in real-time. During the test the operator would keep graphs of cumulative AE hits versus channels on the computer screen, and would scan the channel activity indicators. Any channel that showed activity would be investigated for "leak-like" or "crack-like" behavior, using additional plots and specific quantitative criteria. Active investigation (during the test) of any suspicious channels also would be carried out on the aircraft itself by mechanics. An idea of how the interpretation process worked can be had by referring to Figure 5. The "a" part of the figure shows both the amplitude of the signals being received as a function of time, as well as an overlay of the pressure within the aircraft. Notice the four large amplitude bursts received before the pressure hold, and the marked drop off in amplitude during the load hold. The "b" part of the figure shows individual channel
199
activity as a function of time, with the pressure overlaid again. Notice that the activity on some channels remains constant throughout the pressure hold period. The hold period activity on channel 1 was caused by an air leak, while the pressure hold activity on channels 65, 66 and 67 is due to movement of the cockpit windows - notice that this activity ceases as the pressure decays. The cracking of the tensile specimen at station 237.5, stringer 3.5 right, was detected by channels 28, 29 and 35. It can be seen that their activity in Figure 5b corresponds to the amplitude bursts seen in Figure 5a, when it is probable that the actual cracking occurred.
uc-ze
leeA -
xulee
.
No.a
~na
cocwpit
7.,24/9e
z3:ls:x?
M~_
P _ :
~.-
H~"-I
A r~ ....................
T ....
U -
-ii
,:,~--"--"-~: : ::---:I"--
. . . . .
" "
"
:
D,2-.-T--~::'.-..~.-;.:..-).:.:.__.-:}:-.-_I-.] E _ - ~.'--~-~: ~--.-..:!~.:: ~ :-{--:"
" ~ L~L_- ....... ..........
~ :. .-'.
F::-I
:-i"l
F I G . 5a - Amplitude (in dB) versus time (in seconds), plus pressure versus time.
F I G . 5b - Channel number versus time (in seconds).
The detection of cracking in real-time was not as difficult during the actual test as might be thought from the preceding example, both because the operator had many more AE graphs available on the computer, and because all of the AE plots showed dynamic changes rather dramatically. The cracking of the tensile specimen can readily be seen when looking at a dynamic replay of amplitude and hits versus channel, though it is admittedly somewhat obscure in the static "snapshot" graphs shown in Figure 5. Indeed, it was sufficiently clear in real-time that a R A F oversight committee concluded that the occurrence of cracking was not only detectable while the aircraft was actually being loaded, but that there existed a relatively straight forward real-time method of differentiating harmful cracking from benign defects, such as leaks, frets and the movement of composite window structures. One important aspect of the AE testing of the VC-10 that should be noted was that the RAF considered AE to be only an "electronic safety net," which was there only to help the proof pressurization to be conducted in safety. Officially, the R A F had no interest in locating AE sources which did not require calling a halt to the proof pressurization testing. However, a little ad hoc effort was spent in trying to determine the source of some location clusters that were seen in the post-test analysis (which consisted of location plots over the entire aircraft, filtered by pressure, amplitude and energy criteria). Figure 6a shows an internal view of the VC-10 fuselage in an area where a location cluster was observed in one aircraft. This cluster had not shown up until pressures in excess of 9 psi were attained. Figure 6b shows a close up of a site in this overall area. It can be seen that the intercostal has a crack in it - located right at the site of the AE cluster. Several other location clusters were investigated in other aircraft; for
200
example in one a problem with the rear pressure bulkhead was found, while in another skin cracks were found coming from the main deck cargo door [7].
FIG. 6a - Overall Ji4selage structure showing position of cracked intercostal. ()n~,oing Research
F I G . 6b - Detail r~cracked intercostal.
In addition to past successfnl AE applications, other a p p r o a c h e s are being examined that take a d v a n t a g e of recent advances in the fields of p a t t e r n recognition to automate the process of crack detection a n d recognition. Martingale Research Corporation is addressing the issues of automatic detection and classification by applying advanced signal processing and pattern recognition techniques, including neural networks, to actual AE waveforms. This effort is sponsored by the National Science Foundation as a Phase II SBIR project. In an earlier Phase I project, it was demonstrated that detection and classification results of up to 93% correct detection and classification for crack growth could be produced using data from a single wideband sensor. Note that these results were produced at various sensor spacings from the originating acoustic events, thus demonstrating the relative insensitivity of the approach to the effects of dispersion. The Phase I project used a laboratory test article (see Figure 7), constructed to simulate an aircraft pressure hull, which was tested to destruction using many pressurization cycles [8, 9]. Although the results from the Phase I project were extremely impressive, the critical proof-ofconcept is to extend the technique to AE data from real-world testing. F I G . 7 - The laboratory test article was a riveted That is the goal of the current cylindrical plate structure that simulated aspects o f Phase II research project. In Phase an aircraft pressure hull.
201
II, the automated pattern recognition techniques proven in Phase I are being applied to the detection and classification of AE resulting from crack growth using actual in-flight data. This data is being collected using a commercial AE data acquisition system, and uses digitized waveform data from wide bandwidth sensors. The research project is being conducted with subcontract support from EmbryRiddle Aeronautical University (ERAU), Daytona Beach, F L ERAU is supplying the aircraft and collecting the in-flight data for analysis by Martingale Research. Professor Dr. Eric v. K. Hill is the subcontract lead for this effort. Figure 8 shows the target aircraft for the Phase II project.
FIG. 8 - In-flight AE data has been collected from a Cessna Crusader. The Cessna Crusader was selected as the target aircraft for in-flight AE data collection. The portion of the aircraft selected for data acquisition was the vertical stabilizer area (see Figure 9). The tail section of this aircraft is relatively flexible, and during flight maneuvers, such as Dutch rolls, enough tail motion is present to cause AE events in the instrumented portion of the structure. Special test articles, designed to be mounted in the tail section and act as a passive element, were fabricated for insertion into the aircraft. These test articles are built of 7075-T6 aluminum and are highly flexible. The article is mounted to the main tail structure using bolts and rivets, all of which helps to cause noise and create clutter in the data. The test article is notched to allow the quick initiation of fatigue cracks, and proof qualification of the test article design was done by cycling to destruction a test specimen in an MTS machine. The special test structure that is being used for AE data generation (shown in Figure 10) meets the constraints of flight safety, type certification, and data collection. The test article is mounted just behind the front spar of the vertical stabilizer. The structure does not compromise the aircraft's structural integrity in any way, since it acts as a doubler for the existing spar. Because of motion in the structure, it fatigues during flight and produces AE data for subsequent processing. The installation did not require the certification of the aircraft as an experimental aircraft, although it was of course done
202
with the proper F A A oversight and approvals. The test structure, AE sensors, and cabling were installed in the aircraft during May 1997. Data collection was also performed during May 1997 to get actual in-flight data, including data from various in-flight maneuvers.
FIG. 9 ('h;se-w~ q/ the Cessna Crusa~h'r tail section, where the Jatigue test article is inserted.
F I G . 10 - Photo qf test article showing sensor placement and stress notch.
For an even more stringent test, another aircraft is also being used to collect inflight AE data. This aircraft, a Piper Cherokee, is shown in Figure 1 I. This aircraft has experienced cracking in the engine cowling. A total of four sensors, shown in Figure 12, have been used for the collection of data. As before, these are wideband sensors, and the data is collected and digitized using a commercial AE system.
FIG. I1 7he Piper Cherokee has e_u~erienced cracking in the engine cowling area.
FIG. 12 - Reverse view (?/ the engine cowling, showing the AE sensor placement.
The characteristics of this data collection environment are much more stringent than the Cessna Crusader, because of the vibration and noise from the engine located just beneath the cowling. In both aircraft, the data are collected in flight, downloaded at the
203
end of flight, and processed off-line where a variety of specialized routines are evaluated for their relative efficacy with respect to detection and classification. Processing of the digital waveform data (see Figure 13) involves various specialized routines, and includes time domain and frequency domain feature extraction. Figure 14 shows, for example, a simple I-D wavelet transform of the acoustic event shown in Figure 13. Note the multi-resolution aspects of the wavelet transform, which produces a signature quite different from that of a Fourier transform. In Figure 14 the coefficients from 512-1023, 256-511, 128-255, 64-127, etc. provide information at different time-frequency scales because of the simultaneous time-space localization of the wavelet transform. This property of the wavelet transform process offers interesting benefits when processing AE events. The wavelet basis function used for this example is a Daubechies-4 mother wavelet which is a highly-localized basis function. However, we are also exploring other representations for AE event detection and classification.
File FT 107007, Event 38 8 2 HHZ
375025001250-
-1250
-2500--3750
0 I 200 I 400 I 600 I 800
Se~ople
Nttn~oer
FIG. 13 - Acoustic event captured from in-flight testing.

1024-P~ ~aveleg Transform FIle FT 107007, Eveng 38
7S00
I 0
I ZOO
I 400 Wavelet
1 600 Coefficient
800
I 1000
FIG. 14 - 1024-point wavelet transform of the acoustic event in Figure 13.
204
Training, test, and validation data sets are built from the processed data, then applied to the neural network-based detector and classifier. Results to date are quite encouraging and validate the Phase I results. Good AE data has been collected from both aircraft, and the processed data reacts as did the Phase I data when run through the detection and classification software.
Conclusions Acoustic emission provides proven benefits to the nondestructive testing industry. When properly applied, AE-based NDT can isolate failures in real-time and can
locate cracks even when hidden by other structures or added equipment. We have shown in this paper how AE has been applied successfully to the testing of the F-111 and VC-l0 aircraft, both of which were difficult applications that simply could not be satisfied by other NDT technologies. The use of AE-based NDT in a production environment demonstrates that this technology has truly arrived. Moreover, new research results involving digital waveform processing for "wideband AE" offer the potential of adding new, but complementary capabilities to "classical AE" done with narrowband sensors. The proven benefits of wide sensor spacing and high sensitivity of narrowband sensors can be combined with the automatic detection and classification demonstrated with wideband sensors. This offers the potential of new AE systems that will blend automated detection, classification, and location of failures as part of an integrated intelligent NDT system that can be used in operational as well as maintenance applications.
Acknowledgments
The work on the F-111 and the VC-10 aircraft was performed when Dr. Carlyle was with Physical Acoustics Corporation. The work on the Cessna Crusader and the Piper Cherokee was performed with the support of the National Science Foundation under Grant Number 9503017. All opinions are those of the authors alone and not necessarily those of the sponsoring organizations.
References
[ 1] Hutton, P.H., Skorpik, J.R., "Develop the Application of a Digital Memory Acoustic Emission System to Aircraft Flaw Monitoring", Battelle - Pacific Northwest Laboratory Report No. PNL-2873/UC-37, ARPA Contract No. 3476, Code 7DL0, Dec. t978. [2] McBride, S.L., Deziel, G., "Acoustic Emission Monitoring of the CF116 (CF5) FullScale Durability and Damage Tolerance Test", 1992 USAF Structural Integrity Program Conference, WL-TR-93-4080, pp. 496-506, 1993.
205
[3] Carlyle, J.M., "Research on Acoustic Emission Testing of the F-111 Aircraft", Final Report, P.O. 1207011, General Dynamics Corporation, Fort Worth, TX, September 1987. [4] Carlyle, J.M., "AE Testing the F-111", NDT International, Vol. 22, No. 2, pp. 67-73, April 1989. [5] Carlyle, J.M., "Acoustic Emission Testing of XV108, a Vickers VC-10-CI", Final Report, P.O. AS-824015, British Aerospace (Commercial Aircraft) Ltd., Airlines Division, Woodford, England, May 1990. [6] Carlyle, J.M., "VC-10-C1 Acoustic Emission Testing Procedure", Phase Report, UKDPO Contract G5759, U.K. Ministry of Defence, London, England, November 1991. [7] Carlyle, J.M., "VC-10 AE Review", Final Report, Contract ASF/20660L, Defence Research Agency, Farnborough, Hampshire, England, September 1996. [8] Bodine, H., Dawes, R., Henley, S., Hill, E. v. K., "Automatic Detection and Classification of Cracks in Complex Structures - Applying the Parametric Avalanche to Acoustic Emissions", American Society for Nondestructive Testing, Proceedings of the ASNT Spring Conference, 1995 [9] Bodine, H., Dawes, R., Henley, S., Hill, E. v. K., "Detecting Crack Growth in Metal Structures using Temporal Processing", Artificial Neural Networks in Engineering (ANNIE), 1995
Compressed Gas Applications and Standards
Philip R. Blackburn 1
PERIODIC AE RE-TESTS of SEAMLESS STEEL GAS CYLINDERS
REFERENCE: Blackburn, P. R. "Periodic AE Re-tests of Seamless Steel Gas Cylinders," Acoustic Emission Standards and Technology Update, ASTM STP 1353, S. J. Vahaviolos Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999.
ABSTRACT: Sometimes industrial gas distribution involves hauling pressurized gas
over the highways in seamless, forged, steel cylinders which are mounted on truck trailers. To ensure public safety, Federal regulations require periodic cylinder examinations. In 1983 the U.S. Department of Transportation (DOT) began to allow use of acoustic emission (AE) examinations in lieu of hydrostatic tests for this purpose. AE data are recorded in the linear location mode while cylinders are filled with gas. Examples and explanation of excellent location accuracy are provided. Significant AE sources are subjected to secondary inspection and flaw depth is determined. If flaw depth exceeds a prescribed limit the cylinder is removed from service. Study of emission from specific flaws over two and three subsequent periodic re-tests shows that, generally, flaws are benign. KEYWORDS: acoustic emission, secondary inspection, alloy steel, seamless pressure vessels, industrial gases
One distribution mode for industrial gases (albeit a minor one) involves hauling pressurized gas over highways in assemblies of seamless, forged, steel cylinders which are mounted on truck trailers (i.e. "tube trailers"). Perhaps 4000 such trailers and skids (i.e. both jumbo tube trailers with 3AAX and 3T cylinders2 and trailers with smaller 3A and
1 Consulting Engineer, P.O. Box 761, Buffalo, NY 14213 2 U.S. Department of Transportation cylinder specifications
209
Copyright9
www.astm.org
210
3AA cylinders) are in use in North America 3. (In addition to the above-mentioned tube trailers there is a fleet of about 100 rail cars with DOT 107A cylinders which the U S Bureau of Mines uses to transport helium.) To ensure public safety, federal regulations require periodic cylinder examinations, typically at five year intervals. In 1983 the U.S. Department of Transportation (DOT) first allowed use of acoustic emission (AE) examinations for this purpose. Forged, seamless, alloy steel cylinders are used on tube trailers and on rail cars. These are routinely examined by AE test. Some dimensions and service pressures are listed in Table 1. The AE examination, which is approved by the U.S. Department of Transportation, involves recording data in the linear location mode while cylinders are filled with gas. Sources which produce a prescribed number of AE events are subjected to secondary inspection (e.g. shear wave angle beam inspection) to verify flaw presence and to determine flaw depth. If measured depth exceeds a prescribed limit the cylinder is removed from service. (This limit is prescribed based on estimated fracture toughness of the cylinder material, linear elastic fracture mechanics analysis of the specific cylinder geometry, and estimates of fatigue crack propagation through specific cylinder materials and geometries.) Usually, AE examination and secondary local inspection are accomplished without disassembly of cylinder bundles. Usually, there is enough space between cylinders for passage of the ultrasonic search unit. Usually, all of the cylinders which are mounted on a trailer are examined simultaneously while they are filled with gas to 110% of normal fill pressure. Two sensors are mounted on each cylinder and AE data are acquired with the signal processor set up in a linear location mode. This arrangement is depicted in Figure 1. The AE method does accomplish inspection of 100% of the cylinder; precise locations are computed for AE sources which lie between the sensors; AE sources which lie in a cylinder end are lumped together into one location (that being referred to as an "end of the cylinder"). This AE examination was described by Blackburn & Rana [1]; it represented a significant improvement over the established retest procedure. Another description of the method is found in Standard Test Method for Examination of Seamless. Gas-Filled. Pressure Vessels Usin~ Acoustic Emission, ASTM E 1419-96. The prior, established, retest procedure involves disassembling cylinder bundles, filling cylinders with water, individually pressurizing them to 5/3 times normal service pressure while they are submerged in a water jacket, measuring volumetric expansion of the cylinder, and removing from service any cylinders which exhibit large expansion or which leak.
sW.E. Angus, Weldship Corporation, Bethlehem, PA - private communication
BLACKBURN ON SEAMLESS STEEL GAS CYLINDERS
211
DOT Specification Cylinders, Typical dimensions and service pressure
212
ACOUSTIC EMICClON RE.RESTING OF JUMBO TUBE TRAILERS
AE SENSOR
PLANT COMPRESSOR
-(
MULTI CHANNEL SIGNAL PROC|$$OR
~o * * i r L T I i b i I
FLOPPYD15C DATA STORAGE
04
| II~ Ill !a~l egal' .o,o

~OCATICN
Test Schematic FIGURE 1
213
Improvements associated with use of AE examinations (in lieu of hydrostatic tests) on cylinders in tube trailer service are: Economy - a hydrostatic test typically costs $20,0004 whereas an AE examination typically costs $7,000. (A purveyor of industrial gases with a fleet of 200 tube trailers might save about 2.6 million dollars in plant cost during five years operation.) More information about cylinder condition - flaw location (axial location) is determined with the AE examination and circumferential location and flaw depth are measured with subsequent ultrasonic examinations. Product contamination is reduced - cylinders are pressurized with nitrogen or with actual product. Cylinders are not contaminated with water, which is forced into discontinuities located on the internal surface by the high pressure used in hydrostatic tests, and which cannot be completely removed. Equipment utilization can improve - rolling stock need not be lost from service for weeks while it is pulled to a retest facility, disassembled, hydro tested, re-assembled and returned. AE examinations can be accomplished in the production environment while cylinders are filled with product. Also, this AE test method accomplishes examination of the entire cylinder. (It is a 100% examination.) AE sources in cylinder walls are identified as well as sources in cylinder ends. Nowadays, fourteen years after initial use, AE re-examinations are widely used in industrial gas distribution operations; perhaps 90% of the trailers with jumbo tubes (i.e. 3AAX and 3T cylinders), in the US, are routinely re-examined with the acoustic emission method. THE NATURE OF FLAWS IN FORGED SEAMLESS STEEL CYLINDERS Gas cylinders used in North America are made from chrome-moly steel, mostly. Seamless pipe is formed by piercing a heated billet on a mandrel (e.g. the Mannesmann process). Pipe is formed to size as it is forced over lubricated mandrels. A series of four or five separate forming steps may be used; pipe sections are re-heated (to about 790 ~ between some of these forming steps. Flaws originate at folds or depressions on billet surfaces. These discontinuities are stretched and moved within the pipe during forming operations, Mandrel lubricant is forced into flaws; lubricant burns when the hot pipe is ejected from the mandrel. Combustion leaves solid carbonaceous scale (mill scale) within a flaw volume. A grinding operation is performed to clean the inside wall and improve concentricity between inside and outside diameters. Pipe ends are heated and then hammer forged to produce closures at the cylinder ends. Visual inspections locate flaws on the inside and outside surfaces. These are removed with hand held grinding wheels. 41998 US dollars
214
There is ready access to the cylinder exterior; hence, few flaws remain on exterior walls. However, access to the inside surface, for visual inspection and for grinding, is more difficult; and, if flaws exist they are more apt to be on the inside surface. Automated ultrasonic shear wave angle beam inspection is conducted; any flaws with depth greater than five percent of wall thickness are located 5. These are removed. Cylinders are heat treated to achieve specified strength. Finished cylinders undergo hydrostatic test; here water is forced into any flaws on the inside surface. Subsequently, residual water from the hydrostatic test causes corrosion within the flaw volume while the tube is in service. During AE examinations corrosion and mill scale within flaws do produce measurable emission and benign flaws are located based on emission from these secondary sources, even when flaws do not grow in service.
AE EXAMINATIONS OF CYLINDERS IN SERVICE When an emission source is in the end of a cylinder (i.e outboard of the space between the two AE sensors) the AE signal processor will not compute a precise source location and the cylinder end is treated as one location. If five or more AE events are recorded from one axial location then a secondary (ultrasonic) inspection is performed.
Exemptions to the Code of Federal Regulations which have been issued by the U S Department of Transportation to allow use of AE examinations in lieu of hydrostatic retests require that secondary inspections (e.g. ultrasonic angle beam) be performed at locations which produce five or more AE events (i.e. five or more events within an eight inch axial distance). This "five event limit" is specified for smaller cylinders (DOT 3A and 3AA), for large cylinders (DOT 3AAX and 3T) and for very large cylinders (DOT 107A). Actually flaws in thick walled cylinders tend to be deeper and more voluminous and they can grow to greater length. Hence, in thick walled cylinders crack growth and mill and corrosion scale can be expected to produce more AE events. Experience teaches that the "five event limit" is appropriate for the cylinders with thin walls. (3A and 3AA cylinders have about 0.7 cm wall thickness.) And, a larger number of events is physically appropriate for larger cylinders. However, the "five event limit"does represent practice which is conservative and which is easy to implement and to regulate.
AE system threshold setting A scenario for structural degradation of cylinders during service is initiation of fatigue crack growth at the deepest part of a flaw (a lap or fold) located in the tube wall. The largest stress is in the hoop direction on the large diameter portion of cylinders; maximum stress occurs when cylinders are filled with product. Such fatigue cracks would have longitudinal orientation and would grow when a cylinder is periodically pressurized during service. Estimates of crack growth in such cylinders appear in the open literature
lq.
SFelbaum, J. CP Industries Inc. McKeesport PA, November 1997 - private communication
215
AE examinations are performed at 110% of the maximum pressure which occurs during service. The AE system must be set-up so as to produce adequate sensitivity so as to detect emission from flaw growth which might occur during an AE examination. Signal processor threshold of 32 dB (where 0 dB equals 1 microvolt at the preamplifier input) is used. One must consider peak amplitudes associated with typical sources of AE at flaws and attenuation of such emission as they travel through a cylinder. Typical sources of AE in cylinders are rust patches, fracture of, and interference with, mill scale and corrosion products when a flaw is subjected to strain, contact of fatigue crack surfaces while the cylinder is strained, and crack propagation through the parent metal. Data from cylinder retests, and from laboratory work with pre-cracked compact tension specimens, were reported by Blackburn [2]. It was shown that growth of a fatigue crack through parent metal in 3AAX and 3T steels (Le. chrome-molybdenum steel, quenched and tempered) produces, typically, peak amplitude at the source as high as 60 dB. Other typical sources (crack surface contact, mill scale fracture and rubbing and spalling of rust scale) all produce emission with even higher amplitude than emission from crack growth. Table 2 shows ranges of peak amplitudes (corrected back to the source) which are associated with these emission sources. Consider attenuation of acoustic emission in a cylinder. Figure 2. shows a typical attenuation curve; this is from measurements on a 3AAX cylinder with 1040 centimetre overall length. Sensors are usually mounted about 30 cm from the very end of cylinders. Inspection of Figure 2 indicates that emission from a source which is located between a "first hit sensor" and the very end of a cylinder would suffer about 8 dB attenuation; and, emission from a source which is located between the two sensors on a cylinder could suffer as much as about 8 dB attenuation before it reached the "first hit sensor" and as much as perhaps 16 dB before it hit the far sensor. Hence use of a 32 dB signal processor threshold setting provides assurance that emission from crack growth through parent metal will be detected and that emission source will be located. Moreover, emission from secondary sources discussed above, with their characteristically larger peak amplitudes, would accompany emission from actual crack growth and provide further assurance that crack growth through parent metal will be detected.
ALLOWABLE FLAW DEPTH When flaw depth exceeds a prescribed limit a cylinder is removed from service. The depth limit is prescribed based on estimated fracture toughness of the cylinder material, linear elastic fracture mechanics analysis of the specific cylinder geometry, and estimates of fatigue crack propagation through the specific cylinder geometry [1]. Allowable flaw depths for some cylinder specifications are shown in Table 1.
216
TABLE 2
Sources of acoustic emission at flaws
(DOT 3AAX and 3T steels)
AE Source
Peak Amplitude tat source in dB~ (From field test data) (From CT specimen data)
mill scale external corrosion
50 to 110 66 to 78
mechanical contact (at crack surfaces) crack growth (in parent metal)
19 to 66
19 to 60
217
-5
-10 dB
O O
-15
O O O
~..
O
-20
measured with 150 kHz sensor
-25
O
-30 0 200 400 600 800 1000

DISTANCE (cm) Attenuation in 3AAX cylinders
Figure 2
218
SECONDARY
INSPECTION
- FLAW
DEPTH
MEASUREMENT
Routinely, after an AE test, local ultrasonic examination with a distance amplitude technique is used to measure flaw depth, The method is described in Standard Practice for Ultrasonic Examination of H e a w Steel Formngs. ASTM Standard A 388/A 388M-94,
- v
Secondary inspections are made while cylinders are at full pressure, when the cylinder is strained and the flaw surfaces are separated. Fatigue cracks which initiate on the inside surface of cylinders (i.e, flaws which we most often encounter with these cylinders) exhibit considerable space between crack surfaces. On the other hand, fatigue cracks which initiate on the outside surface exhibit little space between crack surfaces (these cracks are very tight.) Hence we should expect better depth measurements with cracks which initiate in flaws located on the inside surface of cylinders. Indeed, inside surface flaws are most often encountered for reasons discussed in the above paragraphs above under the "NATURE OF FLAWS" heading. Inside or outside surface location of flaws can be readily determined by a skilled operator using a distance - amplitude ultrasonic inspection technique. Ultrasonic calibrations are performed on rings with machined notches which are located on the inside surface. Distance - amplitude curves are drawn on the display from the ultrasonic instrument. Flaws located on outside surfaces produce amplitude peaks which are displaced on the time axis relative to the calibration curve; these are readily identified as being located "on the outside surface." When a distance - amplitude technique is used depth measurements may exceed actual flaw depth because laps in cylinder walls meet the outside surface at some angle. Reflections of ultrasonic beams which strike the side of a flaw which forms an acute angle with the cylinder surface have larger amplitude than a reflection from a notch which is perpendicular with the surface (e.g. a calibration notch). This situation is described by Yokono [3]. Use of a reflected ultrasonic tip diffraction technique should provide more accurate flaw depth measurements. A comparison of six different flaw sizing techniques by Mansour & McGaughey [4] indicates that a tip diffraction method should provide more accurate indication of flaw depth than a distance - amplitude technique. I f a decision about removal from service must be made, a cylinder owner should want to consider additional inspection to verify flaw depth. Jumbo trailer tubes ( DOT specification 3AAX or 3T cylinders) cost about $10,000 each; investment in thorough secondary inspection (e.g. use of more than one secondary inspection method) can be well justified before a cylinder is actually removed from service.
219
CONDITION OF FLAWS WHICH HAVE BEEN DETECTED AE data from periodic retests of one tube trailer fleet, over a fifteen year period, were examined to establish any trend in measured emission from specific flaws during service6. (e.g. Do depths of known flaws increase during service? Do known flaws produce more emission during a subsequent periodic AE retest?) Based on these data, it was determined that known flaws produced fewer AE events during subsequent retests. Figure 3 shows AE event count from specific flaws, averaged, and plotted versus AE test number (AE tests were for the most part five year apart). The trend is "AE event count decreases with time in service." This is evidence that the flaws which were detected and measured were benign. They were not growing in service; and, emission which were measured were produced by mill scale and corrosion within laps. Elimination of water, which was introduced during periodic hydrostatic tests, had eliminated a corrosion agent; hence, emission from benign flaws diminished during subsequent re-tests. AXIAL LOCATION OF AE SOURCES Ordinary linear location algorithms are used. A typical result is illustrated in Figure 4. Here the AE location distribution and a depth profile (from subsequent LIT inspection) associated with a 15 centimetre long flaw in a DOT 3T cylinder are plotted; the abscissa for these plots is the longitudinal axis of the flaw. Inspection of the figure shows that AE events are clustered in the deepest regions of the flaw (i.e. regions which should contain more mill scale and/or corrosion). Location measurements are set-up by breaking pencil leads to produce elastic waves; elapsed time between threshold crossings at the two sensors on a cylinder is measured. When the method was used initially, state of the art AE signal processor software permitted entry of elapsed times for each cylinder into the AE test set-up. With that capability, excellent location results like those shown in Figure 4 are attained. (This flaw was 90 degrees away from the axis between sensors and it was located at 3% of the distance between the AE sensors. Measurement of such a location represents a reasonable challenge to the linear location technique; but it worked very well indeed.) Nowadays, state of the art AE signal processor software does not always permit entry of the actual measured wavespeed from each cylinder. A typical number of cylinders in one test is twelve. Wavespeeds are different in each cylinder because plate waves develop when pencil leads are broken during location calibrations. Speed of plate waves depend strongly on cylinder wall thickness. There is significant difference in wall thickness due to manufacturing tolerances. Use of incorrect wavespeeds can cause location inaccuracy. Actually, any lack of accuracy which is associated with such an AE system deficiency can be accommodated by performing local ultrasonic inspection of emission sites over a larger distance along the cylinder longitudinal axis. It is instructive to consider elastic wave propagation in more general terms. In the 6Lee, Andrew, Amko Service Company, July 1995 - private communication
220
A C O U S T I C EMISSION: S T A N D A R D S AND T E C H N O L O G Y
~! . . . . . . . . . . . . . . . . . . . .
t:3 O
~O
eG)
iii (!)
1
<
0 1 2
AE Test No.
Trend in AE event count, from individual flaws, with time (note: subsequent re-tests follow former retests by five years)
Figure 3
BLACKBURN
ON SEAMLESS
STEEL GAS CYLINDERS
221
.030
~: .020 u Z" 010-
I'J
>
7
'
'
' 6 INCHES
'
'
'j'
,,
;
III,, I I
12 13
,,
114
1~) 1'1 AXIAL LOCATION, INCHES
Depth Profile and AE Location Distribution (from a 15 centimetre long flaw in a DOT 3T cylinder) FIGURE 4
222
ideal case of an infinite olate with stress waves which travel at constant speed in all directions, the locus of points which satisfy an observed hit sequence and measured time difference will comprise a hyperbola (which can be drawn on the surface of the plate). In this case, if an emission source is displaced from the axis between sensors there will be an error in the computed location (said error being related to the shape of the hyperbola)[5]. Experience with location measurements in DOT specification cylinders (e.g. the off-axis example which is shown in Figure 4) suggests we measure the passage of waves which move uniformly along the cylinder axis. Linear location measurements are accurate even when sources are displaced from the axis on which the sensors lie. It seems that we locate AE sources in these cylinders by measuring the passage of Lamb waves. It was pointed out that 3T and 3AAX cylinders represent a favorable case for the measurement source location with such waves 7. If we use the AE sensor resonant frequency (i.e. 150 kHz) together with the wall thicknesses indicated in Table 1 we can see that the product of frequency and thickness is about 2 mm- MHZ. Examination of dispersion curves for plate waves in steel (see Krautkramer [6]) show that at 2 mm-MHZ there are three separate wave modes (i.e. so, ao and a0 which have the same group velocity. If we assume that wave dispersion is similar in DOT specification cylinders then we can perceive a favorable situation for location measurement. Zero order bending waves produced by an out-of-plane source (e.g. pencil lead break) would travel at (or at about) the same speed as zero order symmetric mode waves produced by an in-plane source (e.g. crack growth or mill scale fracture within a lap). Accuracy of linear location measurements should be excellent in this case; and, indeed, it is. For 3AAX and 3T cylinders, wall thickness variation is about plus-minus 15% (based on manufacturing specifications). A survey o f cylinder re-test data shows that measured time between hits at sensor pairs (when the source is outboard of the sensor set) is typically within a range of plus 19% and minus 9%. This difference in time between hits (i.e. difference in wave speed) is attributed to variation in wall thickness from cylinder to cylinder. In order to realize best practice with this method one should use data acquisition software which can accept and use separate values for wave speed or elapsed time between hits on the sensor pair for each individual cylinder which is examined. Experience with location measurement, as described above, indicates that in these cylinders we measure source location based on arrival o f plate waves. Speed o f plate waves depend upon cylinder wall thickness. Location measurements are accurate even when the AE source is displaced (i.e. in a circumferential direction) from the sensor axis.
7pollock, A.A., Physical Acoustics Corporation, August 1986 - private communication
223
CONCLUSION To date, in DOT specification cylinders (Note: DOT cylinders are examples of very conservative structural design), no flaws which are actively growing in service have been foundf Experience with this application for AE testing teaches the following lessons: Secondary sources of emission (e.g. mill scale, flaw surface contact) enable one to locate flaws. Secondary sources of emission produce larger peak amplitudes than actual crack growth through parent metal; but, it is possible to detect crack growth when a proper AE system threshold is used.
REFERENCES:
[ 1] Blackburn, P,R. & Rana, M.D. "Acoustic'Emission Testing and Structural Evaluation of Seamless Steel Tubes in Compressed Gas Service "Transactions of the ASME, Journal of Pressure Vessel Technology, VoL 108 May 1986 [2] Blackburn, P.R. "'AcousticEmissionfrom Fatigue Cracks in Chrome - Molybdenum Steel Cylinders" Journal of Acoustic Emission, Vol. 7, March 1988 [3 ] Yokono, Y. et al "Influence of Inclination Angle of Notch- Type Defect on Echo Amplitude m Ultrasonic Testing" Proceedings of 13th World Conference on Nondestructive Testing, Sao Paulo, Brazil, October 1992 [4] Mansour T. & McGaughey W. "UltrcL~onicTesting Applications m Welding", Section 16, Volume 7, Nondestructive Testing Handbook, Second Edition, P. Mclntire Editor, American Society for Nondestructive Testing, 1991 [5] Baron, J.A. & Ying, SP. "Acoustic Emission Source Location, " Nondestructive Testing Handbook, Volume Five, Second Edition, Miller R.K. & Mclntire P. (Editors) American Society for Nondestructive Testing, Columbus, Ohio 1987 [6] Krautkrarner J. & H. "Ultrasonic Testing of Materials", page 576 - Diagram 9, 4th Edition, Springer-Verlag New York, 1990
Roy D. Fultineer, Jr. I and James R. Mitchell2 FIELD DATA ON TESTING OF NATURAL GAS VEHICLE (NGV) CONTAINERS USING PROPOSED ASTM STANDARD TEST M E T H O D FOR EXAMINATION OF GAS-FILLED FILAMENT-WOUND PRESSURE VESSELS USING ACOUSTIC EMISSION (ASTM E070403-95/1)
REFERENCE: Fultineer, Jr., R. D. and Mitchell, J. R., "Field Data on Testing of Natural Gas Vehicle (NGV) Containers Using Proposed ASTM Standard Test Method for Examination and Gas-Filled Filament-Wound Pressure Vessels Using Acoustic Emission (ASTM E070403-95/1)," Acoustic Emission: Standards and Technology Update, ASTM STP 1353, S. J. Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999. ABSTRACT: There are many composite wrapped pressure vessels in service. These containers are most widely used for gas storage in natural gas vehicles (NGV). A standard has been developed for the testing of these vessels by the subcommittee ASTM E07.04.03 Acoustic Emission (AE) applications. The AE test method is supported by both field test data and laboratory destructive testing. The test method describes a global volumetric testing technique which is offered as an alternative to the current practice of visual inspection. KEYWORDS: acoustic emission, compressed natural gas containers, pressure vessels, composites, burst testing, nondestructive testing
Introduction
A new standard titled Standard Test Method for Examination of Gas-Filled Filament-Wound Pressure Vessels Using Acoustic Emission (ASTM E070403-95/1) has been developed by the subcommittee on Acoustic Emission (AE) applications. The basis of this document is founded on over 20 years of industrial experience in the use of AE techniques for evaluating composite pressure vessels, tanks and numerous other components fabricated from composite materials. The test method is supported by test
i Engineer, Spencer Testing Services, Inc, P.O. Box 429, Spencer, WV 25276. 2 Accounts Executive, New England Region, Physical Acoustics Corporation, 711 Cow Hill Rd., Mystic, CT 06355.
224
Copyright9
www.astm.org
FULTINEER AND MITCHELL ON NATURAL GAS VEHICLE
225
results from several independent testing agencies who have agreed to follow the standard test method and report the results to the ASTM subcommittee. Reported here is the input from one testing agency which has accumulated data from over 3 000 in-service containers and 14 destructive tests.
Background
The theory which supports the use of AE for measuring mechanical properties is derived from the observation that a predictable pattern of acoustic emission is produced when a composite material is subjected to stress. Specifically, there is very little emission at low stress and an abruptly higher level of emission at higher stress [1]. The transition, called the "Knee in the AE vs Load curve," is characteristic of the particular material being examined and typically occurs between 40% and 70% of ultimate load (Figure 1). Numerous codes and standard test methods utilize this observation. They are as follows: Standard Practice for Acoustic Emission Examination of Man Rated Aerial Devices (ASTM F 914) Standard Practice for Acoustic Emission Examination of Fiberglass Reinforced Plastic Resin (FRP) Tanks/Vessels (ASTM E 1067) Standard Practice for Acoustic Emission Examination of Reinforced Thermosetting Resin Pipe (RTRP) (ASTM E 1118) Acoustic Emission Examination of Fiber Reinforced Plastic Pressure Vessels (ASME Section V Article II) Design and Testing of Composite Pressure Vessels (ASME Section X, RPT-1) CARP Recommmended Practice for Acoustic Emission Testing of Pressurized Highway Tankers Made of Fiber Reinforced Plastics with Balsa Cores (ASNT) In essence, a composite structure is considered to be safe for continued service if loading to service stress produces low emission (below the KNEE) and questionable if the emission level is high (above the KNEE). ASTM E070403-95/1 specifically addresses the use of AE to evaluate composite wrapped pressure vessels. The most wide use of these components is for compressed natural gas (CNG) storage in natural gas vehicles (NGV). Many thousand NGV's are in service throughout the US. Their use has been heavily promoted by the natural gas providers and auto manufacturers as a means to lower vehicle emission. A typical NGV fuel container operates at a maximum fill pressure of 3 000 to 4 200 psi (20.68 to 28.95 MPa). The risk of a catastrophic failure of the gas container is a major safety concern. The US DOT-RSPA is mandated to regulate all gas containers that are transported on public roads. The historical record indicates that regulatory accountability and periodic inspection can lower the risk to a satisfactory level. AE and other NDT methods are in wide use under regulatory approval (e.g. ASTM Standard Test Method for Examination of Seamless, Gas-Filled, Pressure Vessels Using Acoustic Emission (ASTM E 1419) and the Compressed Gas Association (CGA) Methods for Acoustic Emission Recertification of Seamless Steel Compressed Gas Tubes (CGA C-18)). Unfortunately, from a safety
226
viewpoint, NGV fuel containers fall outside of DOT-RSPA authority and instead fall under DOT-NHTSA which cannot mandate a national inspection policy for in-service containers. In-service inspection is currently accomplished on a voluntary basis following ANSI/AGA Basic Requirements for Compressed NGV Fuel Containers (ANSI/AGA NGV-2) which recommends a visual inspection at 36 month intervals or alternatively, nondestructive testing with specific inspection details described in CGA Guidelines for Reinspection of NGV-2 Type Fuel Containers (CGA C-6.4). ASTM E070403-95/1 provides a volumetric nondestructive alternative to visual inspection. The authors' opinion is that visual inspection is too subjective and arbitrary to be a stand alone method. The AE test method offers additional advantages over visual inspection: a) AE is sensitive to the internal condition of the NGV container that may not be detected with visual inspection, b) Examination can be carried out during normal filling, c) Container removal is not required, d) Gas is not released into the environment, e) Sensitivity to leaks at connections.
BURST TEST CONTAINER #14 Cumulative AE Counts Versus % of Maximum Fill Pressure Maximum Fill Pressure -__3000psi (20.68 MPa 100 90: Burst Pressure - 7800 80:
70:
60: "~ 40~ 30120~

i i L i i i i--, , i i i i i i
0 10 30 50 70 90 110 130 150 170 % of Maximun Fill Pressure
Figure 1 - Illustration of the "Knee".
227
Procedure for In-Service Inspection of NGV Containers

The testing procedure used for the evaluation of the NGV containers is the same procedure found in ASTM E070403-95/1. First, all accessible exterior surfaces of the vessel are visually examined. The second step calls for the inspector to isolate the vessel to prevent contact if possible. The next step includes connecting the fill hose and the elimination of any leaks at the connections. The AE sensors are then attached in accordance with ASTM Standard Guide for Mounting Piezoelectric Acoustic Emission Sensors (E 650). In step five, the examiner adjusts signal processor setup conditions. A performance check at each sensor should be performed at this time to verify that the peak amplitude is greater than a specified value. Pressurization of the vessel will then start. A fast-fill or slow-fill pressure schedule may be used. Plots of AE activity versus time or pressure should be monitored during the pressurization of the vessel in case any unusual responses occur. The next step is to store all data on mass storage media. The data is then compared to the acceptance criteria and the results recorded. At this time, vessels which have produced excessive emission may be reexamined without delay. Finally, a system performance check at each sensor should be carried out.
Acceptance Criteria
The nonmandatory appendix found in the ASTM E070403-95/1 includes acceptance criteria and instrumentation settings adopted from ASTM E 1067 and other related standards. The acceptance criteria are based on the most widely accepted AE parameters, i.e. counts, signal strength, and peak amplitude as a function of time and pressure. Evaluation of the acceptance criteria leads to 3 possible outcomes: 1) The emission level is low indicating that the container is not over stressed at service pressure. The vast majority (99.0%) of containers tested fall into this category and are put back into service. The recommended retest period, 3 years, is consistent with ANSI/AGA NGV-2. 2) The emission level is moderate (0.9%) indicating that the container may have a slightly lower burst pressure than when new, but is safe for continued service. The recommended retest period for these containers is 1 year. 3) The emission level is high (0.1%) indicating that the containers is over stressed at service pressure and should be removed from service.
Destructive Testing
Test Setup
The first step in the test setup was to construct an explosion chamber. The chamber was constructed with 12 in (30.38 cm) thick fiber reinforced concrete walls, a 4 in (10.16 cm) thick concrete fiber reinforced floor with a drainage system, and a solid oak cover. The container to be tested was filled with water and a high strength, flexible supply
228
line was connected. The supply line was connected to a steel manifold that contained a pressure relief valve and a pressure transducer. The container was then laid down in the chamber. The water pressure was supplied by a air driven water pump which was connected to the steel manifold with another section of the supply line. A 150 kHz AE sensor was connected to each end of the container, The pressure transducer and the AE sensors were connected directly to the data acquisition system. This allowed the operator to monitor the pressure and AE activity during the pressurization of the container. A schematic of the test setup can be seen in Figure 2.
DATA ACQUISITION SYSTEM
PRESSURE TRANSDUCER STEEL ..... "..... P R E S S U R E roA)qlFOLI~ ~ R E - 'IE r~ L-F ~'~. . . . . . . . ...... VALVE ~ ~
- ~2~__ . . "-
Ll n
~:~
"', "~ :_~-'-'-7._-__z_
"
",._
. . . . . . . . . . . . . . . . .
WATER SUPPLY
Figure 2 - Schematic of test setup.
Pressurization of Containers
Pressurization of NGV containers for purposes of AE testing can be either fast-fill or slow-fill. For the purpose o f this study, we will be considering fast-fill pressurization. Each container will be pressurized up to its maximum fill pressure with pressure holds at 50, 75, 87.5, and 100%. AE activity collected during these load holds is compared to acceptance criteria at the conclusion of the test.
FULTINEER AND MITCHELLON NATURAL GAS VEHICLE
229
Results of Destructive Tests

Fourteen type II NGV containers were AE tested according to ASTM E07040395/1. The containers were then destructively tested to burst pressure. This burst pressure was then recorded for each container and used for comparison with AE test recommendations. Figures 3, 4, and 5 are plots of the cumulative AE counts versus percentage of maximum fill pressure for the containers tested. Figure 3 contains the data for containers 1 through 5. Containers 1 and 5 showed high levels of AE activity indicating serious structural problems. Following the guidelines found in ASTM E070403-95/1, these containers would be removed from service. Results of the destructive test of containers 1 and 5 confirmed that the burst pressures were low, 5 800 psi (34.47 MPa) and 5 664 psi (36.79 MPa), respectively. Containers 1 and 5 were properly identified by ASTM E070403-95/1 as being unsafe for continued service. There were also two containers with moderate levels of AE activity which called for a one year retest. Containers 2 and 4 were the two containers in question and they had burst pressures of 7 400 psi (51.02 MPa) and 8 228 psi (56.73 MPa). The final container, container 3 on this graph, was given a three year retest certification. It produced low levels of AE activity, and was considered to have little or no loss of structural integrity. The AE indicated acceptability of this was confirmed with a high burst pressure of 8 100 psi (55.85 MPa). Figure 4 has a plot of cumulative AE counts versus percentage of maximum fill pressure for five more containers. These containers are labeled containers 6 through 10. All of these containers were recommended for one year retests due to the presence of moderate AE activity. These containers had burst pressures of 7 208 psi (49.70 Mpa), 6 640 psi (45.78 MPa), 7 080 psi (48.81 MPa), 7 880 psi (54.33 MPa), and 7 080 psi (48.81 Mpa), respectively. These containers should be reevaluated in one year and the results compared to the previous years results to see if there have been any changes. Figure 5 shows the data for containers 11 through 14. One of these containers (container 11) was recommended for a one year retest. It had a burst pressure of 7 000 psi (48.26 MPa), and produced moderate amounts of AE activity. The other containers (containers 12, 13, and 14) represented on this graph were given a recommendation for a reevaluation in three years. The burst pressure of these containers were 7 840 psi (54.06 MPa), 8 200 psi (56.54 MPa), and 7 800 psi (53.77 MPa), respectively. Figure 6 illustrates a composite correlation plot of the cumulative AE counts during the AE test versus the final burst pressure. The trend can be seen that the containers with lower burst pressures had higher levels of AE activity, and containers with higher burst pressures had lower levels of AE activity. The containers with high levels of AE activity were considered to have low burst pressures and would have been removed from service. The remaining containers were considered to have moderate or lower levels of AE activity and were allowed to remain in service for one or three years.
230
PRESSURE TESTING OF NGV CONTAINERS Cumulative AE Counts versus % of Maximum Fill Pressure Maximum Fill Pressure - 3000 psi (20.68 MPa 8O 70 i.. 60
.~ ~ 40
~3C
IG 0 ~ 0
f
." +
20 40 60 80 100 % of Maximum Fill Pressure #1 .Container #2 o Container #3 ~ #4 , Container #5
Figure 3 - Containers I through 5.
PRESSURE TESTING OF NGV CONTAINERS Cumulative AE Counts versus % of Maximum Fill Pressure 80 __Ma_ m u m Fill Pressure - 3000 psi (20.68 MPa' ~ 70
4 ~
~
... 60
50 20
10 0
.~o~40
-~3o
20 40 60 80 100 % of Maximum Fill Pressure o Container #6 9 Container #7 o Container #8 Container #9 ~ Container # 10
Figure 4 - Containers 6 through 10.
231
PRESSURE TESTING OF NGV CONTAINERS Cumulative AE Counts versus % of Maximum Fill Pressure 80 Maximum Fill Pressure - 3000 psi (20,68 MPa
70
~ .60
15o
9 ~40 ~ '* ~ 3 0 10
0 9 ~......2~0 ' ,~0 | 60 80 100 % of Maximum Fill Pressure o Container # 11 9 Container # 12 o Container # 13 9Container # 14
Figure 5 - Containers 11 through 14,
Cum. AE Counts at 100% Maximum Fill Pressure versus Burst Pressure (% of Maximum Fill Pressure) 80f Maximum Fill Pressure - 3000 psi (20.68 MPa
,.6
10t
~176
0 150'1-70' 190 2 i 0 ' 289 250 270 290 Burst pressure (% of Maximum Fill Pressure)
Figure 6 - Containers 1 through 14.
23 2
Field Testing
During the testing o f approximately 3 000 containers, we discovered six suspect containers at four test sites. Figure 7 is a plot & d a t a obtained from containers at Test Site #1. This test site as well as all other test sites discussed in this paper will be labeled by number such as Test Site #1. Figure 7 shows a plot o f AE counts versus % of maximum fill pressure for six containers at Test Site #1. The curve labeled one year retest represents a container which did not meet the criteria for a three year retest. With 12 605 AE counts, this container was recommended for a one year retest. The other containers illustrated on this figure had very little AE activity and passed the three year retest criteria. The data for Test Site #2 can be seen in Figure 8 At this test site, two suspect containers were identified and recommended for a one year retest in accordance with ASTM E070403-95/1. One o f these containers had l0 688 counts and the other container had 12 631 counts as can be seen on Figure 8. The remaining four containers all met the criteria for a three year retest. At Test Site #3, two suspect containers were identified. One o f these containers was recommended for a one year retest when the number o f counts reached 17 397 aiter the final hold. With that number of counts, the container did not meet ASTM E070403-95/1 criteria for a three year retest. The curve for this container was labeled one year retest in Figure 9 and showed a jump in AE counts between 75% and 100% o f its maximum fill pressure. The second o f these suspect containers was recommended to be removed from service when the number o f AE counts exceeded 50 000 before the container was even pressurized to maximum fill pressure. The container represented by the curve in Figure 9 labeled remove from service showed a large jump in AE counts between 50% and 87% o f its maximum fill pressure. Pressurizing o f this container was stopped at 87% of the maximum fill pressure because o f the large amount o f AE being observed. This container showed evidence o f the "knee," discussed earlier, in the counts versus pressure curve which would indicate a loss in structural integrity and a loss in ultimate strength. The container removed from service at Test Site #3 was taken to the manufacturer for further evaluation. The container was retested with AE during this reevaluation and almost identical data was received. Figure 10 shows a plot for both the field data as well as the retest data. It is obvious from this graph that the curves are close to identical. This shows that this method is repeatable as well as effective in removing any possible suspect containers. The container recommended to be retested in one year showed a jump in the AE counts received, however, the "knee" has not yet been observed. The remaining four containers represented were typical containers that passed the three year retest criteria. The final test site discussed in this paper is Test Site #4 where one suspect container was identified. This suspect container was removed from service. The reason the data on Figure 11 shows only 12 167 counts is because at the end o f the third hold there was a dramatic jump in AE counts, This rapid increase in counts is attributed to the beginning o f exponential count growth. When this occurred, the AE monitor stopped the test for safety purposes and recommended the container be removed from service,
233
Test Site # 1 Cumulative AE Counts Versus % of Maximum Fill Pressure 14Maximum Fill Pressure, 3000 psi (20.68 MPa)
% of Maximum Fill Pressure ~One Year Retest 9Three Year Retest oThree Year Retest ~Three YearRetest ~Three YearRetest ~Three YearRetest
Figure 7 - Field." Test Site #1.
Test Site #2 Cumulative AE Counts Versus % of Maximum Fill Pressure 14Maximum Fill Pressure 3000 psi (20.68 MPa)
-
9 ~
il
nl
v O
/ - ~
i ~ o, 9
20 40 60 8'0 100 % of Fill Maximum Pressure o Three Year Retest * ]~hre~.Year_Retest ~ Three Year Retest Three Year Retest ~ One Year Retest One Year Retest
Figure 8 - Field: Test Site #2.
234
ACOUSTIC EMISSION: STANDARDS TECHNOLOGY AND
Test Site #3 Cumulative AE Counts Versus % of Maximum Fill Pressure 6a Maximum Fill Pressure - 3000 psi (20.68 MPa~
! ,~
9 g 3ot.
9 m ["" "" 20[
/
'
/
IOF OU 6 -
,J"
20 , ~ - 1 - 60 80 ,,,0 % of Maximum Fill Pressure ~'Three Year Retest" Three Year Retest~ Three Year Retest Three Year Retest ~One Year Retest ~Remove From Service
Test and Retest of Suspect Container Cumulative AE Counts Versus % of Maximum Fill Pressure !613 Maximum Fill Pressure - 3000 psi (20.68 M P a )
!40F ~20F !OOF

~80F 60F L40F L2@~00~ 80~ 60~-
p////
40F
o ~ 20 40 60
~:
80 160'1~0'
% of Maximum Fill Pressure o Field Test *Retest
Figure 10 - Retest of suspect container.
235
Test Site #4 Cumulative AE Counts versus % of Maximum Fill Pressure 13 Maximum Fill Pressure - 3000 psi (28.68 MPa)
20 40 - 60 -80 - 100 % of Maximum Fill Pressure o Three Year Retest +Three Year Retest ~Three Year Retest Three Year Retest xRemove From Service ~Three Year Retest
(~
Conclusions
1)
2) 3)
4) 5)
AE inspection in conjunction with a visual inspection provides a more objective evaluation of the containers structural condition than does visual inspection alone. This is particularly evident if the visual inspection does not include the area under the straps or if gradual chemical attack is taking place in inaccessible areas. AE inspection o f containers using the proposed standard showed that AE was capable o f detecting containers with low burst pressures. Burst tests have shown that the AE activity is inversely proportional to the burst pressure. This, along with phenomenon o f the "knee in the AE versus pressure curve," has shown that the methodology o f A S T M E070403-95/1 is quite effective. Field test data shows that containers exposed to ukraviolet rays, chemical attack, and various types of mechanical damage produce more acoustic activity than containers that are protected. No container examined with the AE test method has failed in service.
References
[1]
Mitchell, J. R., "Standard Test Method to Quantify the Knee in the Acoustic Emission vs Load Curve as a Material Parameter for Composites," Proceedings
of the 3rd International Symposium on Acoustic Emission From Composite Materials, AECM-3/ASNT, Paris, France, July 17-21, 1989.
Ainul Akhtar~and David Kung 2 ACOUSTIC EMISSION TESTING OF STEEL-LINED FlIP HOOP-WRAPPED NGV CYLINDERS
REFERENCE: Akhtar, A., and Kung, D., "Acoustic Emission Testing of Steel-Lined FRP Hoop-Wrapped NGV Cylinders," Acoustic Emission: Standards and Technology Update, ASTM STP 1353, S. J. Vahaviolos, Ed., American Society for Testing and Materials, West Conshohocken, PA, 1999. ABSTRACT: Acoustic emissions (AE) were examined from ten steel-lined cylinders, hoop-wrapped with continuous glass fiber reinforced plastic (FRP), after 4 years in natural gas vehicle (NGV) service. A test pressure of 27.6 MPa (4000 psi) was used which is 7% above the standard specified factor of 1.25 of the nominal service pressure. Hydraulic pressure cycling was applied in the range of 2.4 - 24.1 MPa (350 - 3500 psi) to simulate refueling in the laboratory. Burst pressure was measured for vessels containing no detectable flaw, with flaws in the FRP and cracks in the metallic liner. It was found that visual examination has shortcomings for the inspection of hoop-wrapped cylinders. AE data obtained through incremental pressurization to 27.6 MPa (4000 psi), holding pressure for 5 minutes, and unloading, do not provide useful information on structural integrity. Relative emission during two consecutive pressurization cycles holds promise for the nondestructive evaluation of metal lined FRP hoop-wrapped cylinders.
KEYWORDS: acoustic emission testing, visual inspection, continuous fiber reinforced plastic, metal lined FRP hoop-wrapped cylinders, natural gas vehicle, structural integrity
There are existing specifications for the acoustic emission testing of pressure vessels made of metallic materials and those made of fiber reinforced plastic (FRP). A relevant document is the American Society for Mechanical Engineers (ASME), Boiler and Pressure Vessel Code, Section V, Nondestructive Examination, Article 12, "Acoustic Emission Examination of Metallic Pressure Vessels." Test methods for vessels made of FRP were developed in 1982 and revised in 1987 by the Society of the Plastics Industry.
JDirector, Materials Engineering, Powertech labs Inc., Surrey, V3W 7R7, Canada. Adjunct Professor, Metals & Materials Engineering, University of B.C., Vancouver, Canada. ~Senior Materials Technologist, Powertech Labs Inc., 12388-88th Avenue, Surrey, V3W 7R7, Canada.
Copyright9
236 www.astm.org
AKHTAR AND KUNG ON NGV CYLINDERS
237
That document, "Recommended Practice for Acoustic Emission Testing of Fiberglass Reinforced Plastic Resin (RP) Tanks/Vessels," commonly referred to as the CARP Recommended Practice, is used widely as is the ASME Boiler and Pressure Vessel Code, Section V; Article 11, "Acoustic Emission Examination of Fiber Reinforced Plastic Vessels," and Section X, Article RT-6, "Acceptance Test Procedures for Class II Vessels." Cylinders used for the storage of compressed fuel on natural gas vehicles (NGV) are designed to minimize the mass of the vessel. As a result, the metal-lined hoopwrapped NGV cylinders operate in the regime of low cycle fatigue, a situation different from that of vessels designed to the ASME Boiler and Pressure Vessel Code. The American National Standards Institute/American Gas Association (ANSI/AGA) document "NGV-2 - Basic Requirements for Compressed Natural Gas Vehicle (NGV) Fuel Containers" provides the details of the various designs currently in use. Typically, a metal lined FRP hoop-wrapped cylinder with a nominal service pressure of 20.7 MPa (3000 psi) would have a burst pressure of 51.7 MPa (7500 psi). The metallic liner (without the FRP) is required to withstand a pressure of 1.25 times the nominal service pressure. Pressure rise of up to 1.25 times the nominal service pressure may occur in service in extreme cases from heat of compression due to fueling. A process called autofrettage is used in the final fabrication stage. The cylinder is pressurized to produce a slight plastic deformation of the liner. Upon depressurization, the liner goes into compression while the FRP remains under tension. Although autofrettage enhances the fatigue life of the cylinder, the complexity of strain interaction between the metallic liner and the FRP makes the metal-lined FRP hoop-wrapped cylinder different from ASME designed vessels. Acoustic emission testing is attractive for it holds the promise of in-situ inspection without the cylinder being removed from the vehicle. Removal of the cylinder for retesting and re-installation are added costs, hence unattractive. The object of this paper is to explore the possibility of acoustic emission testing of steel lined glass FRP hoop-wrapped cylinders.
Materials and Methodology

Vessels used in the present investigation (commonly referred to as Type-2) had an overall length of approximately 1118 mm (44 in.), outer diameter of about 356 mm (14 in.) and an internal volume of around 81.82 liters (4992 cu. in.). The steel liner was cylindrical over a length of about 787 mm (31 in,) with one near hemispherical closed end and a threaded 19 mm (0.75 in.) opening at the other. The cylindrical metallic wall had a thickness of 6.3 mm (0.25 in.) and an internal diameter of 330 mm (13 in.) A modified version of AISI 4130 steel had been used in the quenched and tempered condition to a hardness of Rockwell C-24. The cylindrical portion of the vessel was hoop-wrapped with an FRP thickness of approximately 6.7 mm (0.26 in.). Continuous E-glass fibers, making up about 60% of the composite, were embedded in a curable polymer matrix. Following wrapping and curing, the cylinder had undergone autofrettage at a pressure of around 39.3 MPa (5700 psi). The vessels had a minimum design burst pressure of 51.7 MPa (7500 psi) and the nominal service pressure was 20.7 MPa (3000 psi). Ten cylinders were tested that had been in NGV service on fleet vehicles for
238
approximately 4 years. Seven were manufactured in the year 1991, two in 1989 and one in the year 1990. The cylinder identification, L1122-4/91 for example, includes the specific cylinder number (L1122) followed by the month and year of fabrication (April 1991 in the above example). The tests carried out consisted of visual inspection, eddy current scanning of the interior surface of the metallic liner, acoustic emission testing, pressure cycling, hydrostatic tests, burst tests, and magnetic particle inspection.
Visual Inspection and Eddy Current Scanning

Flaws on the exposed FRP surface were subjected to visual examination. Eddy current scanning was used to detect flaws on the cylinder interior surface that might have been a result of manufacturing or NGV service. The interior surface of the cylinder is subjected to the highest level of dynamic hoop stress and remains exposed to the attack of natural gas contaminants. Hence this surface is potentially the site for the nucleation of flaws on the liner. To detect such flaws, an eddy current pencil probe was inserted through the 19 mm (0.75 in.) opening at the cylinder end. The probe was mechanically manipulated to scan the entire surface for cracks having axial-radial orientation (Fig. 1), A 500 kHz probe was used with a Nortec-19-e II instrument to generate the impedance plane diagrams. Flaws 0.5 - 3.0 mm (0.02 - 0.2 in) could be detected.
I:IG. l --Photogrw~h shows a eylitzder Hndergoi~g eddr current scamfing.
Acoustic Emission Testing

The experimental set-up for acoustic emission testing is shown in Fig. 2. A Spartan-AT instrument supplied by Physical Acoustics Corporation was used for data acquisition and analysis. Two resonant frequency piezo-electric transducers of the R6 type (frequency response range 20 - 100 kHz with resonance at 60 kHz) were mounted on the metallic liner adjacent to the FRP at either end of the cylinder as shown schematically in Fig. 2. This lower frequency transducer was preferred over the commonly used R15 type (frequency range of 100 - 300 kHz with resonance at 150 kHz) due to the lower signal attenuation with distance associated with the former. The ultimate objective is not only to detect the signal (feasible with either transducer) but to quantify damage using
239
only two transducers on vessels up to 3 m (10 ft) in length. Using a pulser with a 100 dB setting, it was found that an attenuation level of up to 20 dB may be expected with the R6 transducers over a length of 1.83 m (6 ft). A higher attenuation associated with the R15 transducer (30 - 35 dB over the same distance) was considered undesirable for further signal processing. The signal from the sensor was pre-amplified with the usual 40 dB setting. An appropriate band pass filter was employed along with a threshold setting of 40 dB. The pencil lead break test was used to check the set up prior to the commencement of the experiment. The Hit Definition Time (HDT) was 1000 laS. A hydraulic pump was used to pressurize the cylinder with water. Parametric input from the pressure transducer (Diaphragm gage) was fed into one of the channels of the AE system for the simultaneous recording of pressure and acoustic data with time. Fig. 3 Containment Vessel
Pressu transdu
--
~7
y
Preamplifiers
(wave form record
t ~ 4 0 db gain
Digital Storage So
Cylind~
,.
~ carau|q )umplt AE Transdur
Parametric Input FIG. 21Schematic drawing shows the acoustic emission test set up. shows the loading, hold and unloading sequence used. Since surface abrasion is likely to have occurred during removal of the cylinders from the vehicles, transportation and handling during the experimental set up, a conditioning cycle to 24,1 MPa (3500 psi) was used to simulate refueling, Such a conditioning cycle, however, would not be needed for the in-situ testing of the cylinder which is the eventual goal. Acoustic emissions were monitored during two consecutive loading, hold, and unloading cycles to 27.6 MPa (4000 psi). For each of these two latter test cycles it took 5 minutes to reach the peak load. The load was held for 5 minutes which was followed with unloading. Each cycle took approximately 15 minutes throughout which AE data were collected.
240
6000
40
4000
A
I 27.6
3,SO0psi
24.1 ; MPa
/ /
,,~ MPa l
27.6 MPa
',
~,
30
} ;
, ',
I, h
f
2000
20 |
r
/
10
q~
X)NOITIONING
0
'~
!i
~ MONITORING
cvc,E-~
MONITORING CYCLE-2
'
~
', " 0
Time
FIG. 3--The loading, hold and unloading sequence used with acoustic emission testing. Fueling Simulation, Hydrostatic Test and Burst Test
The pressure change between two consecutive fuelings was simulated using hydraulic pressure cycling between 2.4 - 24.1 MPa (350 - 3500 psi). An electro-hydraulic control system was used with a sawtooth waveform. Pressure cycling was carried out at the rate of 2 - 3 cycles per minute. Hydrostatic testing was conducted to determine the nature of deformation of the cylinder with increasing pressure. A water jacket surrounded the cylinder during the test. The water volume expelled from the jacket was measured to determine cylinder expansion. For cylinder burst testing, the same configuration was used with the exception that the water was removed from the jacket to accommodate the sudden release of water from cylinder rupture. The initial rate of loading was 6.9 MPa/24 s (1000 psi/24 s). This rate of pressurization dropped during the rapid expansion of the cylinder immediately prior to the occurrence of rupture. To measure the burst pressure of flawed cylinders, axial flaws of known lengths were cut mechanically through the entire thickness of the FRP. Some of the cylinders tested were cut transversely into 4 - 8 sections for fluorescent magnetic particle inspection to determine the nature of flaws on the interior surface of the vessel. Significant flaws located via the magnetic particle inspection method were subjected to fracture face examination. The fracture face was revealed through cooling of the specimen using liquid nitrogen followed by impact loading.
241
Results
Visual inspection of the FRP outer surface indicated varying degrees of resin cracking on all cylinders. Gross fiber breakage was not visible on seven of the ten cylinders examined. The remaining three (L0749, L1122 and L1131) had abrasion damage to the FRP in the lower transition region (between the FRP and the liner) which had resulted in fiber breakage. Figure 4 is a photograph of the damaged region on cylinder L0749. In appearance, these damaged regions on all three cylinders were similar, being approximately 75 nun (3 inches) long axially and tapered so as to have gone through the entire thickness of the FRP at the lower transition region. According to the guidelines provided in the Compressed Gas Association (CGA, Virginia) pamphlet C-6.4, 1996 "Methods for External Visual Inspection of Natural gas Vehicle (NGV) Fuel Containers and their Installations," a flaw such as that shown in Fig. 4 would be classified as Level 3 damage, which is cause for removal of the vessel from service. Circumferential delaminations may be noted in Fig. 4. Such delaminations were present on each of the ten cylinders examined. The eddy current scan indicated an absence of flaws on the interior surface of the liner of each of the ten cylinders examined.
FIG. 4--Photograph shows abrasion damage to the FRP on cylinder L0749-10/90.

Acoustic Emission
Figure 5 shows the acoustic emission data collected for two cylinders, one with abrasion damage and the other without. The plots show non-cumulative hits plotted on the y-axis against amplitude (dB on the x-axis)
242
"r
0
40OO
3000
E
3 2OOO 0 Z 1000
20
40
60
80
100
Amplitude (db)
50O0
I4)
40OO
3ooo
-I (3 C O Z
2O00
1000
20
40
6O
80
100
Amplitude (db)
FIG. 5--Non-cumulative hits vs. amplitude during the first test cycle for an apparently undamaged cylinder (Ll126-4/91) and one with abrasion damage (Ll122-4/91).
Above the threshold setting of 40 dB, the number of hits increased with amplitude, attained a maximum in the vicinity of 42 - 43 dB and decreased thereafter in general. Qualitatively, the acoustic emission data collected for the other six undamaged cylinders and for the remaining damaged cylinders were similar to those shown in Fig. 5. The emissions recorded for each of the ten cylinders tested in the as-received condition, listed in Table 1, show that the cylinders containing no apparent fiber breakage produced emissions in the first test cycle ranging from a low of 2212 total hits to a high
243
value of 72 880. Total hits during the first test cycle for two cylinders with visually apparent fiber damage were well within the range measured for the undamaged cylinders. A higher value of 91 241 hits was recorded for one damaged vessel (L0749). A similar trend may be noted in Table 1 for the number of high amplitude hits (>_60 dB), and for the data obtained during the second AE test cycle. For each of the ten as received cylinders, the AE data were examined after dividing the test cycle into loading, hold and unloading segments. Table 2 shows for each segment the hits and the corresponding counts (shown in parenthesis) during the first 0-27.6 MPa (0-4000 psi) cycle. The hits and the corresponding counts for each of the three segments listed in Table 2 show variability similar in nature to that found with total hits and hits having amplitudes ~60 dB, the latter two listed in Table 1. TABLE l--Acoustic emission from as-received cylinders Cycle 1 Cylinder All Hits 2 212 11 311 72 880 25 362 53 854 25 074 65 174 91 241 24 958 48 803 Amplitude ~60 dB 31 505 2 497 780 1 691 1 498 5403 6 458 1 446 6 070 All Hits 706 5 277 37 356 14 468 28 589 13 687 33 131 55 835 15 516 26 791 Cycle 2 Amplitude ~60 dB 9 171 902 314 635 571 1 955 2 095 554 2 488 Cycle 2/Cycle 1 All Hits 0.32 0.47 0.51 0.57 0.53 0.54 0.51 0.61 0.62 0.55 Amplitude z60 dB 0.29 0.34 0.36 0.39 0.37 0.38 0.36 0.32 0.39 0.40
Undamaged Cylinders L0529-7/89 L0531-7/89 L1026-3/91 L1123-4/91 L1125-4/91 L1126-4/91 Ll132-4/91 L0749-10/90 L1122-4/91 L 1131-4/91
FRP Abrasion Damaged Vessels
The three damaged cylinders were pressure cycled to simulate fueling in the laboratory with interruptions for AE testing. The results are summarized in Table 3. Total emissions decreased as the number of simulated fueling cycles increased in the case of vessels L0749 and L1131. However, cylinder L1122 was different in this regard. The total emissions decreased upon applying 14 230 simulated fueling cycles (3442 hits from 24 958) but increased gradually upon subsequent pressure cycling to 3779 hits after 29 230 simulated fueling cycles.
244
TABLE 2--AE hits (the corresponding counts in paranthesis) for the various segments of
the first 0-2Z 6 MPa (0-4000 psi) cycle

Cylinder L0529-7/89 L0531-7/89 L1026-3/91 Ll123-4/91 Lll25-4/91 L1126-4/91 Ll132-4/91 L0749-10/90 L1122-4/91 L1131-4/91 Loading 1 280 6600 16000 5000 9500 5000 20000 20000 7200 12000 Hold Unloading 172 1 311 880 362 3854 3574 5174 3241 6158 5803 ( ( ( ( ( ( ( 3 076) 14753) 8874) 6253) 19587) 55345) 20925)
Undamaged Cylinders ( 9 600) 760 ( 6 000) ( 84000) 3400 ( 112000) ( 340000) 56000 (1140000) ( 60000) 20000 ( 260000) ( 160000) 40500 ( 780000) ( 200000) 16500 ( 600000) ( 580000) 40000 (1180000) FRP Abrasion Damaged Vessels (1000000) 68000 (3200000) (150000) 11600 ( 3 2 0 0 0 0 ) (560000) 31000 (1400000)
(47953) (87674) (45658)
Results of Post AE Investigation

Hydrostatic testing was carried out on the three FRP damaged cylinders following the last set of pressure cycling and AE tests. The permanent hydrostatic expansion was measured upon pressurization to and depressurizing the vessels from 34.5 MPa (5000 psi). Permanent expansion figures expressed as a percentage of total expansion (elastic and plastic) were as shown in Table 4. These figures suggest that cylinder L1122 underwent a larger permanent expansion than did the other two damaged vessels. Each of the three vessels was subsequently pressurized to successively higher pressures until expansion continued to occur without an increase in pressure. Tests were interrupted at about 32% total expansion. The highest pressure reached ( estimated burst pressure) is shown in Table 4. Table 5 shows the burst pressures of as-received cylinders and those containing through thickness FRP flaws of various lengths.Vessels without apparent fiber breakage, gave burst pressures within a narrow range of 61.9 -+2.6 MPa (8975 -+375 psi). These values are well in excess of the design burst minimum (ANSI/AGA-NGV2) figure of 51.7 MPa (7500 psi). The burst pressure decreased rapidly with about a 52 mm (2 inch) long FRP cut, but a gradual decrease occurred thereafter with the length of cut. It is noteworthy that the three cylinders that sustained in service abrasion damage (Fig. 4) had estimated burst pressures following the simulated fueling in the laboratory (Table 4) that would be equivalent to approximately a 102 mm (4 inch) axial-radial through FRP cut.
245
TABLE 3 o A E data at intervals offueling simulation Test Cycle 1 All 260 Hits dB 91241 6 458 8 404 245 4431 87 24 958 1 446 3 442 97 3 652 98 3 779 86 48 803 6070 5 870 166 3 897 104 3 125 96 Test Cycle 2 All 260 dB Hits 55 835 2095 7 768 161 3056 65 15 516 554 2 320 50 2640 49 2 618 40 26 791 2 488 4 523 116 3 170 122 2 630 84 Cycle 2/Cycle 1 All 260 dB Him 0.61 0.32 0.92 0.68 0.69 0.75 0.62 0.39 0.67 0.51 0.71 0.52 0.69 0.47 0.55 0.40 0.77 0.70 0.81 1.17 0.84 0.87
Cylinder L0749-10/90
Cycles 0* 10 749 15000 0* 14 230 19 230 29230 0* 14 230 19 230 29230
Ll122-4/91
L1131-4/91
As-received
TABLE 4--Cylinder characteristics after simulated fueling

Simulated Fueling Cycles 15 000 29 230 29 230 Permanent Expansion % 3.5 5.4 3.6 .......... Estimated Burst Pressure MPa (psi) 45.5 (6 600) 44.9 (6 500) 45.5 (6 600) Liner Flaw Dimensions length x depth mm (inch) 6.4 x 1.2 (0.25 x 0.047) 20 X 0.3 (9.79 X 0.012) 12.4 X 4.2 (0.49 X 0.165)
Cylinder
L0749-10/90 L1122-4/91 L1131-4/91
Cylinders that had in-service abrasion damage were sectioned transversely following the simulated fueling (Table 2) and the hydrostatic test noted earlier. Magnetic particle inspection of the inner surface of the liner revealed bands of superficial flaws that are commonly encountered following NGV service of steel vessels [I]. One significant crack, however, was found on each of the three cylinders examined. The location of that significant liner flaw is shown in Fig. 6 in relation to the FRP-abrasion damage for all three vessels. The dimensions of those major liner flaws are shown in Table 3. The internal diameter of the liner was slightly larger at the locations containing the liner flaw which suggested that these were the locations of the bulge, although the bulge was not visually apparent.
246
TABLE 5--Burst pressures of as-received cylinders and those with artificially induced FRP damage
Cylinder L0529-7/89 L053t-7/89 L1126-4/91 L1132-4/91 L1026-3/91 L1125-4/91 L1123-4/91
Discussion
Axial Through Thickness FRP Cut mm (inch) ... ... ... ... 51 (2) 102 (4) 152 46)
Burst Pressure MPa (psi) 63.8 (9 250) 64.5 (9 350) 59.3 (8 600) 60.7 (8 800) 49.0 (7 100) 44.7 (6 480) 41.7 46 050)
Visual inspection has been proposed recently for the periodic inspection of CNG cylinders on natural gas vehicles (CGA pamphlet C6.4-1996). The evidence gathered in the present work suggests that visual examination has significant shortcomings for the periodic inspection of metal lined FRP hoop-wrapped cylinders on two accounts. Figure 7a shows the liner flaw in cylinder L1131 following 29 230 simulated fueling cycles, 8 AE test cycles, and after hydraulic pressurization to 45.5 MPa (6600 psi). The liner flaw had grown through roughly 67% of the liner thickness. Figure 7b is a schematic drawing of the transverse section of the cylinder through that liner flaw. Along the radius of the cylinder containing the liner flaw, a crack had initiated in the FRP (Fig. 7b). The crack in the FRP, however, was not on the outer surface of the cylinder. Hence, it could not be detected visually. The FRP crack had initiated at the interface between the metallic liner and the FRP (Fig. 7c) and had propagated about 20% (1.3 mm or 0.05 in.) through the thickness of the FRP (Fig. 7d). As shown in Fig. 7, cracks well in excess of the tolerance limit of 0.254 mm (O.010 in.) depth, the latter considered acceptable for cuts, scratches and gouges mentioned in the CGA-C6.4 pamphlet, may remain in the FRP of metal lined hoop-wrapped cylinders, without being detected through visual inspection. Secondly, the inspection criteria (CGA-C6.4) are such that vessels that are fit for usage must be removed from NGV service. As seen from Table 4, cylinders L1122 and L1131, in spite of FRP damage in NGV service, had a remaining life each in excess of 29 230 cycles and had burst pressures equal to or in excess of 44.9 MPa (6500 psi). Such vessels are suitable for NGV service from a fitness for purpose standpoint and yet they must he removed according to CGA-C6.4. Therefore, there exists a need to use alternative methods for the periodic inspection of CNG cylinders in NGV service.
247
.................................
1 ~ ...................................
i
9 L079-10/90 II j i I
I 1
L1131-4/91
i
I
.............................
i-"Damage
.....
'--~5~
1
FIG. 6---Orientation of liner flaws in relation to FRP abrasion damage. This illustration is a two dimensional representation of the curved cylinder surface obtained through making an imaginary axial cut and flattening. Acoustic Emission Testing
Acoustic emission remains an attractive option because of its potential for in-situ testing, i.e. without removing the cylinder from the vehicle. However, that potential may be realized only through demonstrated correlation between acoustic emission and structural integrity of the vessel. The total number of hits measured in the present work during the first test cycle (Table 1) are not helpful in that regard. The as-received cylinders subjected to burst tests (Table 3) gave values in the range of 61.9 2.6 MPa (8975 375 psi). This scatter in the measured values of burst pressure is small being yet the associated acoustic emission hits ranged from 2212 (L0529) to 65 174 (L1132). More importantly, this wide range of hits did not show a systematic variation with burst pressure within the narrow span of burst pressures measured. This lack of correlation between burst pressure and total acoustic emission noted above, is also observed if one uses loading, hold and unloading segments for the analysis of hits or counts as seen from Table 2. An analysis of "knee pressure" has been proposed by Mitchell and Newhouse [2] based on their work with one type of all composite NGV cylinders. These last mentioned authors have claimed that a lower "knee pressure" implies a lower burst pressure. Some of the data obtained in the present work are shown in Fig. 8 as a plot of cumulative acoustic hits vs pressure during the first test cycle. A
248
cursory examination of Fig. 8 shows that for a given criterion to define "knee pressure" (such as 1000 hits) one obtains a wide range of knee pressures even though the cylinders have a narrow range of scatter in their burst pressure. One arrives at similar conclusions as regards felicity ratio and emissions during pressure hold at 27.6 MPa (4000 psi).
The Sources of Acoustic Emission

Reasons for emissions at low pressures (Fig. 8) and the variability in the quantity of emission from vessels having similar structural integrity may be considered. It is noteworthy that the steel liner and the FRP both produce AE in hoop-wrapped cylinders. Moreover, emissions may occur at their interface due to the cracking of the coating applied sometimes on the liner for corrosion protection prior to the application of the FRP. Remnants of such a coating are seen in Fig. 7c. The AE sources in the FRP are: matrix cracking which may occur in varying degrees in NGV service due to ultraviolet exposure, etc., delaminations which occur in the transverse plane resulting from the axial expansion of the vessel (Fig. 4) and fiber breakage. Of these, the first two have no direct bearing on structural integrity. Although fiber breakage occurs when the structural integrity of the vessel has been compromised (Fig. 7), the occurrence of fiber breakage is not necessarily an indication of structural integrity loss. In an idealized hoop-wrapped cylinder, the fiber axis will lie along the circumference of a circle. However, in practice the roving of fibers used for the fabrication of cylinders contains many misaligned and twisted fibers. Such fibers will fracture at relatively low pressures or even in the absence of cylinder pressurization with time, since the FRP remains under tension due to the autofrettage treatment. An attempt has been made with some success by Walker et al. [3] through neural networks and amplitude distribution of acoustic events to predict the burst pressure of all composite cylinders in their as fabricated state. Matrix cracking and fiber breakage were considered relevant sources by those authors, while debonding was considered not relevant. However, the situation appears to be different with hoop-wrapped cylinders removed from NGV service. As shown in Fig. 5, the amplitude distribution remains similar for vessels with and without gross fiber damage. The explanation for that similarity possibly lies in the varibility of transverse delamination (Fig. 4) observed with NGV cylinders removed from service. Amplitude distribution is therefore unlikely to be of value when applied to cylinders removed from NGV service. Another source of emission is the oxide scale left on the interior surface of the steel due to fabrication heat treatment. In their search for a correlation between the quantity of emission and structural integrity of all-steel NGV cylinders, Akhtar et al. [4] eliminated the oxide scale associated emissions and established a quantitative relationship between crack depth and AE. In hoop-wrapped vessels, the autofrettage treatment forces the steel liner into compression while the vessel remains at low pressures. The cracked and debonded oxide scale interfaces will produce AE as a result even at low pressures.
249
FIG. 7--Cylinder L1131-4/91 after 29 230 simulated fuelling cycles, AE tests, and pressurization to 45.5 MPa (6600 psi). (a) Fracture face of the liner flaw. (b) Schematic transverse section of the cylinder through the liner flaw. (c) Underside of FRP showing the crack and two delaminations (vertical). (d) Transverse section shows FRP crack originated from the FRP inner surface.
250
Those sources in the liner, the FRP and their interface may explain emissions at low pressures (Fig. 8) and the variability in the quantity of emission observed in the present work with vessels having similar burst pressures (Tables 1, 2 and 5). An approach to overcome those difficulties in the AE testing of hoop-wrapped cylinders may be to examine relative emissions during two consecutive test cycles (Tables 1 and 3). The ratio of high amplitude hits (>_60 dB) shows a trend towards higher values when the vessel undergoes degradation through simulated fueling (Table 3). A similar approach is recommended in ASTM E 1888-97, "Standard Test Method for Acoustic Emission Testing of Pressurized Containers Made of Fiberglass Reinforced Plastic with Balsa Wood Cores".
Pressure (psi)
0 12 000 1000
1 ....
20O0
1
3000
4000
t-i---
.............. L. . . . . . . . . . . . . . . .
10000
8000
Cylinder Burst MPa (psi)

# L0531 64.5 (9 350)
. D
8000J
; L0529
63.8 (9 250)
"r"
9 Ll132 60.7 (8 800) 9 L1126 59.3(8600) . Ll122 FiberDarnaged
0oo
10
20
30
Pressure ( M P a )
FIG. 8--Cumulative hits vs. pressure for as-received cylinders during the first cycle.
251
Conclusion 1. Visual inspection has shortcomings; it is not capable of detecting certain types of flaws in the FRP and it rejects cylinders which are fit for service with a wide margin of safety as observed in the present study. Hence, there exists a need for other nondestructive test methods for the periodic inspection of NGV cylinders in service. Acoustic emission data obtained through incremental pressurization to 27.6 MPa (4000 psi) holding pressure for 5 minutes and unloading, do not provide useful information for structural integrity assessment. Relative emissions during two consecutive test cycles hold promise for the periodic inspection of hoop-wrapped NGV cylinders.
2.
3.
Acknowledgment
A portion of the work reported here was carried out under Gas Research Institute (GRI-Chicago) sponsorship. The authors are grateful to Steve Takagishi and Marco Liem, formerly of GRI, for their patience and support. The opinions expressed in this paper are those of the authors and not of Powertech Labs. References
[1]
Akhtar, A. and Kung, D., "An Assessment of All-Steel Cylinders Currently in NGV Service for the Storage of Compressed Natural Gas Fuels on Vehicles," Report NGV200-3.19, Gas Technology Canada, Toronto, Ontario, 1996. Mitchell, J.R. and Newhouse, N., "Techniques for Using Acoustic Emission to Produce Smart Tanks for Natural Gas Vehicles." Paper presented at the Fifth International Symposium on Acoustic Emission from Composite Materials (AECM 5), Sundvall, Sweden, 1995. Walker, J.L., Russell, S.S., Workman, G.L., and Hill, E.V.K., "Neural Network/ Acoustic Emission Burst Pressure Prediction for Impact Damaged Composite Pressure Vessels," Materials Evaluation, Vol. 55, No. 8, August 1997, pp. 903907. Akhtar, A., Wong, J.Y., Bhuyan, G.S., Webster, C.T., Kung, D., Gambone, L., Neufeld, N., and Brezden, W.J., "Acoustic Emission Testing of Steel Cylinders for the Storage of Natural Gas on Vehicles," Nondestructive Testing and Evaluation International, Vol. 25, No. 3, March 1992, pp. 115-125.
[21
[3]
[41
252
DISCUSSION
L.R. Gambonel--Comment 1--The conclusion that "Visual inspection has some shortcomings; it is not capable of detecting certain types of flaws in the FRP...." is misleading and without basis given the data provided in the paper. The authors have failed to include pertinent design information regarding the intended service life of natural gas vehicle (NGV) cylinders. Specifically, the ANSI/NGV2-1992 standard for NGV cylinders requires the designs to provide 13 000 pressure cycles to service pressure and 5 000 pressure cycles to 1.25 times service pressure. The July 1997 draft revision to the ANSI/NGV2 standard requires that cylinders intended for a 15 year design life provide only 11 250 pressure cycles to 1.25 times service pressure. These performance test requirements are based on actual pressure cycling service conditions experienced by cylinders in NGV service. The cracks in the composite wrap of the three hoop-wrapped cylinders were artificially created in the laboratory by applying test conditions that imposed an excessive number of pressure cycles (fatigue), and overpressurization cycles, several of which far exceeded the allowable design stress for the glass fibers. For example, the cylinders L1122 and L1131 were subjected to the following: (i) First used in NGV service for 4 years (possibly some 1 000 pressure cycles incurred). (ii) Pressure cycled in the laboratory up to 29 230 cycles to 3 500 psi (equivalent to an additional 39 years of NGV service). (iii) Overpressurized in the laboratory 8 cycles to 4 000 psi for AE measurements (not done in NGV service). (iv) Overpressurized in the laboratory to 5 000 psi for a hydrostatic test (not done in NGV service - there is no requirement for periodic retesting). (v) Overpressurized in the laboratory to some 6 500 psi in an attempt to predict the eventual burst pressure (certainly not done in NGV service). As a result, the damage observed in the FRP wrap would have been caused by stress rupture. Neither the fatigue conditions (excessive pressure cycles), nor the stress rupture damage could occur to the FRP wrap in service. Tens of thousands of hoop-wrapped cylinders have been used in NGV service since 1982. Since that time there has never been an incident or failure associated with "hidden" damage. A review by Powertech Labs in 1997 of the condition of hoopwrapped cylinders removed from NGV service ["Condition Assessment of Glass Fiber Hoop-Wrapped Cylinders Used in NGV Service" - Gas Research Institute Report
i Engineer, Materials TechnologiesUnit, PowertechLabs Inc. 12388- 88th Avenue, Surrey, B.C. Canada. V3W 7R7.
253
GRI-97/0052] did not provide any evidence of "hidden" damage in the FRP wrap. There has never been any reported instance of any metal or metal-lined composite NGV cylinder falling by fatigue associated with pressure cycling in service. Of the only Type 2 (hoop-wrapped) cylinders that have failed in service, both were associated with extensive external damage to the composite wrap that would have been readily detected by visual inspection. ["Cylinder Safety Revisited" by W. Liss - Gas Research Institute Natural Gas Fuels, November 1996]. Comment 2--The authors claim in their discussion that the visually apparent damage on cylinders L1122 and L1131 would be sufficient to warrant their removal from NGV service in accordance with CGA C-6.4; however, both cylinders still had considerable pressure cycle life remaining. As a result, the authors concluded "...there exists a need for other nondestructive test methods for the periodic inspection of NGV cylinders...". The fact that the visual inspection criteria in CGA C-6.4 is conservative for this particular hoop-wrapped cylinder design does not provide evidence that some other inspection method is required. It is a result that must be expected of inspection criteria erring on the side of conservatism. For other cylinder designs the inspection criteria will be less conservative. The purpose of the inspection criteria in the CGA C-6.4 document is as follows: (i) To prevent essentially any damage to the composite wrap (other than scratches from routine handling) from remaining in service. (ii) To have a single set of inspection criteria applicable to all cylinder types, and not confuse inspection staff by trying to establish different criteria for each different cylinder design and cylinder size. Comment 3---The conclusion that AE emissions generated during two consecutive pressure cycles "...holds promise for the periodic inspection of hoop-wrapped NGV cylinders" cannot be justified since the data has been generated from cylinders subjected to excessive pressure cycling and overpressurization conditions that would not occur in service.
A. Akhtar and D. Kung (authors' closure)--The authors appreciate this opportunity to clarify issues surrounding visual inspection. Mr. Gambone has made two assumptions of which one is false and the other speculative at best. His translation of 29 230 laboratory hydraulic pressure cycles into 39 years of NGV service presumably means that a vehicle cylinder may be fueled to a maximum of 750 times a year. While the authors agree that 750 cycles is a reasonable assumption, the implication of the critic that a fueling cycle in NGV service is equivalent to a hydraulic pressure cycle is incorrect. The synergistic action of the dynamic stresses between refuelings and the corrosive contaminants present in natural gas, principally H2S, CO2 and H20, causes accelerated degradation of the cylinder in NGV service which is only beginning to be documented. For example, in a recent study (Ref I of the paper) approximately 350 all-steel cylinders were removed for ultrasonic scanning after they had been in NGV service for up to 13 years. Fifteen among them were pressure cycled in the laboratory for their remaining life assessment. It was concluded that the incubation period (the fatigue regime leading to the nucleation but prior to the growth of cracks) is reduced by a factor of 5 in NGV service when compared
254
with that obtained through hydraulic pressure cycling. Such information is still lacking for the crack growth regime. However, if one applies that equivalence of 5 laboratory hydraulic pressure cycles to one NGV service pressure cycle, the 29 230 laboratory hydraulic pressure cycles reported in the paper (Table 3) would not translate into 39 years as done by the critic but to a further service life of 5.8 years, placing the cylinder well within its intended service life of 15 years. The statements concerning ANSI/NGV2 document (comment 1) would have one believe that 11 250 cycles were applied over a period of 15 years using natural gas for fueling to a peak pressure of 1.25 times the service pressure through each cycle. Neither a reference to this effect is provided in the said ANSI/NGV2 document nor is it likely that such precise information has been or would be generated. What is certain is that the figures used in the ANSI/NGV2 document (and in other standards as well), which are set with the state of the knowledge at the time of formulation of the standard, will undergo revision as new information (such as that contained in Ref I of the paper) becomes available. In his systematic enumeration of the cylinder treatment used to create the FRP flaw, Mr. Gambone has overlooked the important fact that Type 2 cylinders made from fiber rovings (all the cylinders examined in the paper were made in this manner) are subjected to an autofrettage pressure of approximately 40 MPa (5 800 psi) as apart of the fabrication process. However, the evidence suggests that neither the autofrettage treatment nor the laboratory pressure cycling (including the subsequent overpressurization) caused stress rupture damage. Had there been stress rupture damage, it would be widespread. The damage seen to the underside of the FRP (Fig. 7 in the paper) was confined to the region immediately above the flaw in the steel (which was only part way through the metallic liner). The authors believe that this FRP flaw, which cannot be detected through visual inspection, to have been a result of the liner flaw. The final overpressurization cycle to 44.9 MPa (6 500 psi) did not produce a detectable change in the liner flaw depth. There was only a slight increase in the axial length of the flaw as seen in Fig. 7a. It is conceivable that a smaller FRP flaw would have resulted had there been no overpressurization. The salient point conveyed in the paper, however, is that the creation of a liner flaw over the next 5.8 years of NGV service would produce FRP damage that cannot be detected through visual inspection. The second assumption made by Mr. Gambone that there has never been any incident of failure associated with "hidden" damage is speculative at best. With reference to the two Type 2 (hoop-wrapped) cylinder failures in NGV service, he has stated that both were associated with extensive external damage to the composite wrap that would have been readily detected by visual inspection. When a Type 2 cylinder ruptures in a catastrophic manner under gas pressure as opposed to hydraulic pressure (the former being the case with the two incidents under discussion), the composite material is obliterated adjacent to the region of the FRP flaw which might have caused the failure. Thus the relevant material not being available, post failure analysis is carried out on the adjacent regions of the FRP which have not been obliterated. Hence, the existence of FRP surface flaws at these adjacent regions, revealed through post failure analysis, does not preclude the possibility that the failure occurred as a result of FRP flaws that were not on the surface. The evidence from materials examined in the adjacent regions being
255
circumstantial, the conclusion regarding the nature of the FRP flaw that caused the failure may at best be considered speculative. A reference has been made by Mr. Gambone to the project carried out at Powertech Labs on behalf of Gas Research Institute (GRI Chicago) for the evaluation of Type 2 cylinders removed after they had been in NGV service. That work was done by the present authors (A. Akhtar and D. Kung). A conclusion of that investigation, transmitted to GRI, was that visual inspection has shortcomings. The GRI did not wish to see such a conclusion in its report. Upon request from GRI, an alternative interpretation was provided by L.R. Gambone, C.T. Webster and J.Y. Wong of Powertech Labs Inc. who concluded that visual inspection is an acceptable periodic inspection method. The latter interpretation was accepted by GRI. The statement made by Mr. Gambone that the review by Powertech Labs of hoop-wrapped cylinders removed from NGV service did not provide any evidence of "hidden" damage in the FRP is correct to the extent that the draft report submitted by the present authors to the GRI and to Mr. Gambone et al for their reinterpretation did not contain the evidence shown in Figure 7 of this paper. That information was gathered later. A single rationally based acceptance criterion when applied to a number of cylinder designs may reject cylinders of one of those designs with a wider margin than that dictated by fitness for purpose for that specific design. However, this is not the case with the visual inspection of NGV cylinders. That so called "margin" is such that 2 out of 3 cylinders shown in Figure 6 of the paper (L079-10/90 and L1133-4/91) would have life limitation occurring at locations far removed from the band which the visual inspection has identified as being FRP flawed. In other words, not only is the acceptance criterion not rationally based, the visual inspection method itself is not rationally based as far as the NGV cylinders are concerned. If one adds to the shortcomings of visual inspection identified in the present work the fact that certain types of significant impact damage on carbon fiber wrapped vessels can not be detected visually2'3, visual inspection becomes unsuitable indeed for the inspection of NGV cylinders.
2 Christoforou, A.P., and Swanson, S.R., "Strength Loss in Composite Cylinders Under Impact", Trans ASME, JEMT, Vol. 110, April 1988, pp 180-184 3 Kaczmarek, H., and Maison, S., "Comparative Ultrasonic Analysis of Damage in CFRP Under Static Indentation and Low Velocity Impact", Composites ,Scienceand Technology, Vol. 51, 1994, pp 11-26
STP 1353-EB/Oct. 1999
Auihor Index
A Akhtar, A., 236

B
Blackburn, P. R., 209 Bodine, H. L., 191 C Calva, M. V., 79 Carlyle, J. M., 67, 191 Chow, D. L., 3 Curtis, C. E., 3
D
Markelov, A. P., 125 Miller, R. K., 67 Mitchell, J. R., 224 Murthy, C. R. L., 141
N
Nagaraja Rao, G. M., 141 Nesvijski, E. G., 41 Nikulin, S. A., 125 O Ohtsu, M., 25, 156, 175 Okamoto, T., 25
P
Dawes, R. L., 191 De Aguiar, P. R., 107 Demeski, R., 191
Finkel, P., 67 Fowler, T. J., 50 Fultineer, R. D., Jr., 224 G Gandy, T. J., 79
H
Patterson, T. S., 79 Pollock, A. A., 67 Puckett, J. A., 50

R
Hamilton, H. R., lIl, 50 Hatano, H., 93 Henley, S. S., 191 Hill, E. v. K., 191
K
Rajachar, R. M., 3 Raju, N. M., 141
Khanzhin, V. G., 125 Kishi, T., 25 Kohn, D. H., 3 Kung, D., 236 Kurianova, E. Y., 125 Copyright9 by ASTM International 257 www.astm.org
Shigeishi, M., 25, 175 Shiotani, T., 156 Shtremel, M. A., 125
258 ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY T Tafuri, A. N., 67 Ternowchek, S. J., 79 Y W Watts, D. J., 67 Webster, J., 107 Yezzi, J. J., Jr., 67 Yuyama, S., 25 Weissman, N. A., 3 Willett, P., 107
STP 1353-EB/Oct. 1999
Subject Index
A Aging dock, 25 Aircraft, aging, 191 Amplitude, 3
B F
Defect detection, 191 Deformation, 125

E
Eigenvalue analysis, 175 Embrittlement, 79
Bone, microdamage, 3 Burn, workpiece, 107 Burst testing, 224, 236 B-value, 156 C Calibration procedure standard, 93 Cold proof testing, 191 Composite wrapped pressure vessels, 224, 236 Concrete beams, structural integrity, 25 Concrete, reinforced beams, structural integrity, 25 structures, characterization, 41 Corrosion, 79 reinforced concrete beams, 25 Cortical bone, fatigue, 3 Crack characterization, rocks, 141 Crack determination, 79 Crack lengths, 3 Crack measurement, 125 Crack mechanisms, twodimensional model, 175 Crack resistance, 41 Crack, shear, 25 Crack volume estimation, 175 Curve-fitting techniques, 156 Cyclic loading test, 25
D
Fatigue, aging aircraft, 191 Fatigue damage, traffic signal poles, 50 Fatigue testing, bone, 3 Flaw depth, 209 Fracture, 125 brittle, traffic signal structures, 50 rock, 141 G Gas containers, natural, 224, 236 Gas cylinders, seamless steel, 209 Graphical analysis, 156 Green's function, simplified, for moment tensor analysis, 175 Grinding, 107
lnconel, workpiece burn detection, 107 Iron alloys, material quality monitoring, 125
K
Kaiser effect, 25
Damage zone dimensions, 3 Davies-bar technique, 175 259
Japanese Societ~r for NonDestructive Inspection calibration procedure standard, 93
260 ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY natural gas, filament wound, 224 natural gas, hoop wrapped, steel lined, 236 Process safety management, 79 Proof testing, 191
R
Laser opto-interferometer, bar end oscillation measurement, 175 Leak detection, buried pipelines, 67 Linear location mode, 209 Loading unloading, 25 l_x)cation methods, buried pipelines, 67 la)ngitudinal wave, 93 M Material quality monitoring, 125 Mechanical integrity testing, 79 Metals gas cylinders, seamless steel, inspection, 209 material quality monitoring, 125 steel-lined hoop wrapped, 236 workpiece burn detection, 107 Microscopy, confocal, 3 Mineralogy, rock fault formation, 141 Moment tensor analysis, 175 N Natural gas vehicle, 224, 236 Niobium, material quality monitoring, 125
P
Rate process analysis, 156 Rayleigh wave, 93 Reciprocity method, 93 Reference standards, laboratory, pipeline leaks, 67 Rocks, crack characterization, 141 S Screening, in-service, traffic signal poles, 50 Sensor calibration, 175 Shear cracking, 25 Shear wave analysis, 79 SiGMA-two-dimensional procedure, 175 Signal difference location technique, 67 Signal enhancement, 67 Signalprocessing, 41, 107, 191 Slope failure prediction, 156 Standards characterization, concrete structures, 41 laboratory reference, 67 proposed, gas-filled filament-wound pressure vessels, 224 proposed, reinforced concrete structural integrity, 25 transducer calibration, reciprocity method, 93 Steels alloy, pressure vessel testing, 209 bearing, workpiece burn detection, 107
Parametric plot, 141 Peak-amplitude distributions, 156 Pipelines, buried, leak detection, 67 Plastic, continuous fiberreinforced, 236 Pneumatic proof pressurization, 191 Pressure processing, 125 Pressure vessels gas distribution, seamless steel, testing, 209 inspection, ultrasonics, 79
INDEX 261 dual phase, material quality monitoring, 125 gas cylinders, steel-lined, 236 Stiffness, 41 Storage tank, mechanical integrity, 79 Strength, concrete structures, 41 Stress effects, rock, 141 Superconductors, 125 T Tin alloys, material quality monitoring, 125 Titanium alloy superconductors, 125 Traffic signal poles, 50 Transducer calibration, 93 U Ultrasonic imaging, 141 Ultrasonic testmg, 79 U.S. Department of Transportation, seamless pressure vessel inspection, 209
Z
V Volumetric strain, 141 Volumetric testing technique, 224
Wave attenuation, 156 Waveforms analysis, 141, 175 b-value, 156 digital, 191 Welds, 50 Wind-induced vibrations, 50 Workpiece burn, 107
Zirconium-tin-niobium-iron alloys, 125
D !
ILl
rU
O~ !

Acoustic Emission - Standards and Technology Update

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Acoustic Emission - Standards and Technology Update

Uploaded by

Copyright:

Available Formats

S T P 1353

Acoustic Emission: Standards and Technology Update

Sotirios J. Vahaviolos, editor

ASTM Stock #: STP1353

Library of Congress Cataloging-in-Publication Data

Peer Review Policy

Printed in Philadelphia,PA October 1999

INTmGRrrY AND LEAK

Acoustic Emission and Ultrasonic Testing for Mechanical Integrity--s. J. TERNOWCHEK,

AE SENSORS, STANDAm)S, AND QUANTITATIVEAE

Calibration of Acoustic Emission Transducers by a Reciprocity Method--H. HATANO

DIVERSE INDUS1RIAL APPLICATIONS

Prediction of Slope Failure Based on AE Activity--T. SHIOTANIANDM. OHTSU

Identification of AE Sources by Using SiGMA-2D Moment Tensor Analysism

TRANSPORTATION APPLICATIONS, STANDARDS, AND METHODOLOGy Practical AE

for Use on AircraftmJ.

M. CARLYLE, H. L. BODINE, S. S. HFa'qLEY,

R. L. DAWES, R. DEMESKI, AND E. v. K. HILL

COMPRESSED GAS APPLICATIONS AND STANDARDS

236 257 259

Author Index Subject Index

ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY

Sotirios J. Vahavidos, Ph.D.

Physical Acoustics Corporation Princeton Junction,NJ 08550 SymposiumChairmanand Editor

USE O F A C O U S T I C E M I S S I O N T O C H A R A C T E R I Z E F O C A L AND D I F F U S E M I C R O D A M A G E IN BONE

K E Y W O R D S : acoustic emission, bone, microdamage, confocal microscopy

ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY

RAJACHAR ET AL, ON MtCRODAMAGE IN BONE

Cortical Bone Specimens

ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY

Mature Bovine Bone(femur) DiaphysealLongitudinalCut End Milled Paralleleplped I

Flow chart of experimental design.

RAJACHAR ET AL. ON MICRODAMAGE IN BONE

Acoustic Emission Analysis

ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY

~1,. .nn,. nn, , , . .

FIGURE 2 - Location distribution histogram (LDH) of AE events generated in V-

RAJACHAR ET AL. ON MICRODAMAGE IN BONE

ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY

RAJACHAR ET AL. ON MICRODAMAGE IN BONE

ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY

RAJACHAR ET AL. ON MICRODAMAGE IN BONE

ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY

RAJACHAR ET AL. ON MICRODAMAGE IN BONE

ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY

RAJACHAR ET AL. ON MICRODAMAGE IN BONE

ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY

-1.0 -2.0 -3.0 -4.0 O.OE+O0 Time (seconds)

6.0 4.0 2.0

RAJACHAR ET AL. ON MICRODAMAGE IN BONE

Summary and Conclusions

ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY

RAJACHAR ET AL. ON MICRODAMAGE IN BONE

ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY

Case Study 1: RC Beams with a Single Reinforcing Bar

YUYAMA ET AL. ON REINFORCED CONCRETE BEAMS

/:~.:i-:"..:-..t . ". ;.L::.~ DEFORM E D:BAR :

FIG.l-Configuration o f the specimen with a single reinforcing bar.

FIG.2-Relationship between number o f AE hits and the applied load.

ACOUSTIC EMISSION: STANDARDS AND TECHNOLOGY

Case Study 2: Repaired RC Beams

L CH.,6 CH.,5 (6)

1_2ooJ2oo.I 2001_2oo.l.2oo 12oo_1